SOUTH ASIAN DIALOGUE ON POVERTY AND HEALTH HELD IN BANGALORE 15TH - 18TH NOVEMBER 1999
Item
- Title
- SOUTH ASIAN DIALOGUE ON POVERTY AND HEALTH HELD IN BANGALORE 15TH - 18TH NOVEMBER 1999
- extracted text
-
Lew H - ©o.)
RF_COM_H_70_A_PART_1_SUDHA
©
MEASURES OF POVERTY SUITED TO
| DESIGN AND EVALUATE POVERTY
ALLEVIATION PROGRAMMES'
By
T .Krishna Kumar"
I
INTRODUCTION
Measurement of poverty has. assumed a greater prominence in
recent years as a result of increasing use of poverty measures in
designing and evaluating poverty alleviation programmes (Kanbur
(1985), Thorbecke (1986), Kakwani and Subba Rao (1990), Ravallion
(1991' and Keen (1992)). There are mainly four measures of poverty
than are in popular use.
These are (i) the head count ratio (H),
-which is widely used in India for the comparison of incidence of
poverty across regions and ever, time; (ii) the income gap ratio
(I), used by the U.S. Social Security Administration (Batchelder
'(1971)); (iii) Sen's index of poverty ( R ) (Sen (197*5), and (iv)
Foster, Greer, and Thorbecke (FGT) index <Pa) (Foster, Greer, and
Thorbecke (1984), Foster (1984), Foster and Shorrocks (1991)).
Inspite of the defects of the head count ratio as a measure of
? poverty, as pointed out by Sen (1976), this index is continuing to
: be used to measure not only the incidence of poverty but also the
> effectiveness of poverty alleviation programmes (Kakwani and Subba
Rao (1990), Minnas, Jain, and Tendulkar (1991)).
Minnas, Jain and
Tendulkar use rate of change of head count ratio to rank the Indian
states with respect to poverty alleviation efforts.
It is
...intuitively
clear
that
poverty
alleviation
depends
on
t-he
-propensity of households below the povety line to cross the poverty
line, and that households closer to the poverty line have a greater
■propensity to cross the poverty line.
It must be expected that
with a skewed income distribution the larger the initial head count
ratio the smaller could be the rate of change in head count ratio.
Thus, using the rate of change of head count ratio to evaluate the
performance of poverty alleviation is biased against those states
prepared for presentation
at
the 29th
Annual
Paper
Conference of the Indian Econometric Society to be held at l.l.T.
: Kan pu r, Mar c h 20-22, 1993.
'Quantitative Analysis Unit, Institute tor Social ano economic
m. Change,
Nagarbhavi,
Bangalore-560072.
The author
thanks K S
; . K r i 5 h n a. s w a m y , Anil d Gore, N ?. rtshnaji, and T V S R a m m o r. a n Rao rar
■. the i r helpful comments on a.n earlier draft.
The author a . s o tna n xs
l-.'tG, Nagara j u for his able assistance in processing the paper on PC.
or regions that are poorer to start with.1
It is therefore.
necessary to choose a better measure of poverty that is suited for
evaluating poverty alleviation programmes.
The plan of the paper is as follows.
In the next section
Sen's axiomatic scheme is restated in terms of the cumulative
income distribution function.
Section I 1 1 proposes a new and
simple measure, of poverty. Section IV suggests how one can use the
new index or poverty and conditional logit or probit analysis to
determine empirically the factors that influence the incomes of
households in different poverty categories.
Using the results of
such analyses, and the relative importance the policy makers attach
to various poverty categories, one can devise suitable poverty
alleviation
programmes.
evaluation
of
poverty
alleviation
programmes is also facilitate: by the new measure, as it takes into
cue account the relative importance that the policy makers attach
to various poverty■categories .
II
A RESTATEMENT OF SEN'S AXIOMATIC APPROACH
Any measure of poverty must' satisfy certain criteria which we
normal ly associate with the- words - poor, poverty, degree of
poverty, poverty alleviation etc.
It is natural to expect that any
measure of poverty must depend on four things - (i) identification
of the poor, (ii) identification of different degrees of poverty
among the poor, (iii) the distribution of households according to
different degrees of poverty, and (iv) the relative weightages the
policy makers attach to different degrees of poverty.
Let s’ be the poverty line - i.e,, a. definite level of per
capita household income such that any household whose per capita
household income fails below it is treats as a Door household
(Dandekar and Rath (1971) ,
Let y denote the per capita household
income.
Let us assume that the distribution of per capita
household
income
can
be
represented
(i.e.,
smoothed
or
approximated) by a continuous relative frequency distribution, or
density,
f (y ),
Let F(y> denote the cumulative distribution
function: i.e.,
1 It is the use of "his measure by G.Parthasarathy to
evaluate the poverty alleviation in Andhra Pradesh that drew my
attention to this prooiem. G. Parthasarathy presented the findings
of his research in a seminar at our Institute in May 1992.
L Dandekar and Rath (1971) attempt to give a scientific basis
for the ooverty line they proposed.
This scientific basis was
T-s=tioned by Eukhatme (1975).
From purely a policy perspective
wha t we need is an administratively chosen target group called the
; Po.-,'.
The choice of target group may be based on certain criteria,
I op- such criterion being persons or households below a certain
iJncoaa. often called the poverty line.
F(y).=ff(x)dx
o
The following axioms reflect the common sense notions of poverty.
Monotonicity Axiom ; Given other things, an increase in per capita
income of a household below the poverty line must decrease the
poverty.
Transfer Axiom: Given other things, a pure transfer of income from
a household below the poverty line to any other household which is
richer must increase poverty.
Transfer Sensitivity Axiom: Given other things, if a transfer t>0
of income takes place from a poor household with per capita income
y to another ooor household with percauita income y + d (d>0) then
,.
.
. ’
.
'
'
5
tne magnitude or increase in poverty decreases as y increases,
The poverty index must depend on z and the distribution of
income - represented by the density f(y>, Let the poverty index be
represented by:
P = [ b (y) f(y) ay
(2.1)
o
where h(y) is called a deprivation function and it satisfies the
cone iilions: h(y) > 0; h'(y) < O; h’(y) >0
z
?<z}
p = f h(y) f(y) = 1 h{F~2(F(y')')}dF
0
0
= Mr-HP(y))} = h(y) >o
aF
dF2
dF(y)
f(y)
<°
(2.2)
<2-3>
Consider the mono tomicity axiom. Suppose a household's Incom
is changed from y to y + d
Then f(y) is decreased by 1/n (wher
n is the total number of households) whereas f(y + A) increases b
I/n.
Suppose that the household is poor, i.e,, y < z.
AP = {h(y+A)-h(y)} — < 0
J2
(2.4)
gxnce h(y) ts 3. d ec r 95 3 n g runction or y* Hence poverty decrea.ses.
flonc t otn i c i ty t'neretore implies that poverty decreases with an
increase in income to any poor household,
all
other things
remaining the same.
Let us now consider the transfer axiom.
Let a household with
income y t ransr er an income A to another household with income y'
(y < y'< z'-
It can be easily verified that
if(y') = -l^n, and J/fy'+A? = 1/n
&f(y~A)
=
1/n;
Af(y)
AP = — {h(y-A) - h(y) } - ±{£(yy) - h(y7+A) } > 0
J3
This
= -1/n;
(2.5)
is so because h' (y ) < 0-and h’ (y)>0
Hence,
under transfer axiom,
we see that poverty
increases.
Now consider the transfer sensitivity axiom.
Let
y(, y7 and y3 denote three- levels
income such that y^y^ty^a.
of
per
capita household
Let there be a transfer of income A from a household with income y^
to another with income y
Then &P is given by
(AP )
= ^ [ h(yt-b.} - MyJ] - ± [h(y2) - h(y2+A)] >0
(2.6)
y2.7i
Similarly if there is a transfer of income A
income
to another with income yj
from a household with
(AP ) = -i [ A(y2-A) - h(y2)] - A [h(y3) - h(y3+A)] >0
Zj. 7i
(2.7)
Since h’{y) > 0 it follows that
(AP )
<
(AP )
Zj.Zj
Zj.Zi
This is what the transfer sensitivity axiom says.
if we do not wish to distinguish poor by different degrees of
Poverty then we arrive at a simple poverty index of the form (2.1)
.-•his satisfies the' no mo tonicity axiom.
h the deprivation function h(y) = 1
(2.8)
a
As we do nc . wish to distinguish between different degrees of
povertry
the
question
of
satisfying
transfer
and
transfer
sensitivity axiom does not arise.
- - - e w i s n *. o distinguish the poor by their degree of poverty
the heao count ratio will not be useful.
As pointed out by Sen the
heao count r 3 c c does not satisfy the transfer axiom.
it does not
satisry the transfer sensitivity axiom also.
Sen (1976) developed
an i n o e x which satisfies the transfer axiom.
d = s>in (1, y/z )
(1-d)
wax (0,
(2,9)
(z~y)/z>
(2. 10)
JD on
d measures the degree of relative affluence of poor households.
The poor who are at the poverty line are relatively most affluent
among tne poor and have d = 1.
Al 1 households who are above the
poverty line are assigned a value 1 for d.
g measures the poverty
t measures, put in percentage, the percentage shortfall of.
household's percapita income from the poverty line.
For
house.no.ds above the poverty line the poverty gap g is assigned a
value of zero.
Sen defines a poverty measure which depends on z,g, and. the
distribution of z.
It is a special case of a general do v e r tv
index :
Ps
=
Ps{z,g,F)
(2.11)
J h(y) f (y} dy
0
where h(y) depends on g
=
Foster, Greer, and Thorbecke (1984.)
take a monotone non
decreasing transformation of g, viz ga
(a > 0),
Their index can
be written as:
Ps = Pa (z, ga, f}
\
= / (g(y)]a^(y) dy
(2.12)
c
where h (y) = [j(y)]“
It can be v e r i
>0 for « > 1.
ed that h'(y) < O for all values of
a
0 and h’(y)
I 1 1 AN ADDITIVELY SEPARABLE DECOMPOSABLE MEASURE OF POVERTY
Suppose that there are different degrees of poverty defined in
terms of proportion of households whose income falls in different
ranges below the poverty line.
One way of doing this is to assume
5
tha ‘
n z poverty ma
zk and zk be
1
0 < zi
one
□ O';
1 in
below the
k-1
:k
zk-2 Deina
between diff
those whose
? households whose
siated with these k 1 n
households who be! ong
p2.
k-
zl> z2’............. .. = pc. v arty line
wh*
mar
could be some
per cent or zk
4
Jne can now du
n
the poor. The poorest of the
ween 0 ano Zj
mg
mes f a 1 1 bet'
Zn
S1
k
ai
of poverty Fk,
............. -pk-
p
one
def :
p (P
?v
’pk>
(1)
dp
dp,
W 1
cP,
0
aP
dPJ
for
tnes
i mp 1
that
the
by the
One may cons
i n oa
foilow ing add
one
P
61 P1
such
61
fin P
index :
Sk Pk
6k >
B2
0
and
S
1 R
6k
1
y one povert
measure becomes the head
count ratio H (f ■ r k=l, F=H)
roup.
Secondly this
income transfers within an income
does not take ir to account the income gaps of different income
take the later into account one can modify the above
groups.
when
P
61
with
P1
P2
>
+
sk ^k pk
> fik
fi2
1
,where
distinguish
n
act
wn
abso i u te
1a w i th a
ty
6
(
z,+zt .
= 2-—< J 1
'
J
O’-
K| It can be easily checked
5?
ax i cms by noting that:
(3.5)
that
measure satisfies
this
> o
~ =
the
three
(3.6)
and
= 3^
M
(3,7)
It may be noted that the contribution to income gap of households
in the igroup equals n g, pp
The above measure assumes that
in the over-all poverty index the marginal rate of substitution
rh
*h
between tne income gap of
i
and
j
income groups is a
constant.
(
CP
1 a (^Pz))
f
f unct ion of
can define a
P
cP
3
Si
1
P1 + B2 g?2 P2
simp!
' I
f
'
J
measured in terms of another monotone non-decreasing
-lative income gap g
viz, g3, where a > 0 then one
lew poverty index
with 6
$
(3.8)
62
modification of F.G.T.
+
6k
6k >
0
>
Bk
X B;
1
pover ty index.0
It is perhaps even more meaningful to co
generalized version of FGT poverty i ndex
with Sj > fi2
Pk
> 0,
i de r
iSjj
1.
> 0
E0'
dit■ us no
(3.10)
P.
reasonable to assume that income elasticity of consumption
groups justifying using different as.
index also satisfies the basic axioms:
dp
dP:
cP
for
1 > 0
cP
cP.-,
——
ir a
-
a;
*
[3 ; > 0
(3.11)
(3.12)
are simple
FGT
account
k c!
-■
s
11 u 11 o n
income
IV DESIGNING AND EVALUATING POVERTY ALLEVIATION PROGRAMMES
;ures given in the previous section distinguish
poverty and give higher wei ghtages to higher
One may wish to distinguish different type of
schemes - one that enables people just below
cross the poverty line must be distinguished
the poorest of the poor to move closer to
the pov er ty
-
line.
different types
poverty alleviation
understand cl
the factors that
belong to different poverty categories.
factors that determine why households
groups is to employ an n-chor.omous
logit mode I .
>£•
this kind of interview approach to
[fare trade-offs (with reference to
he head of a University between teaching,
ce activities of a University) see Kumar
K r i shnaswamy rightly cautions me in
policy subjectivity.
In order that the
simply
jjective the
each of the k
pov e r t y
or
8
Under
this
approach
we can
hypothesize
that
there are
| different factors such as education,
occupation,
ownership of
certain assets, access to credit, caste, etc which determine the
probability that a household belongs to a particular poverty group.
The er -ec t of
these variables may be different for different
poverty groups.
This can be postulated by
the fcllowing model:
........................
for 1 = 1,2,3,.......... k-1
where
Hossain and Sen '.1992.' distinguish between extreme or hardcore
poverty and genera: poverty, and identify the differences between
the two groups in terms of lend ownership and land tenure.
This
information enables us to design proper land reforms to alleviate
pover ty.
Thus, identifying different degrees of poverty, estimating the
number of households in each one of various categories of poverty,
distinguishing the characteristics of households in each of those
categories are the basic empirical research agenda for designing
poverty alleviation programmes.
Once this is done the next step is to see if the poverty
alleviation programmes actually chosen and
implemented
reduce
poverty.
In
this evaluation care must be taken to see that
reduction of poverty among the poorest of the poor has a greater
emphas is.
in general a reduction of poverty of poorer households
must nave a greater emphasis tnan that of the less poor households.
This is achieved through the poverty indices suggested in Section
Ill.
It may also be noted that the poverty indices suggested in
Section III are also eminently useful in evaluating the impact of
the Structural Adjustment Programme (SAP) on the rural poor and
various disadvantaged groups.
This SAP is basically a collection
of those policies that are regarded as market-friend 1y.
it is
genrally suspected tr.at these so-called market friendly policies
are, using Orwellian tn rase, more friendly to thenon-poor and less
friendly to the poor.
Some of the recent studies undertaken by
UNCTAD,
11.0,
and
«-.-lc Bank
have
shown
that
the structural
adjustment programmes increased poverty and benefitted only a small
minori ty.
REFERENCES
A. B. ,
(1971),
3. no
P ov
The Economics
of Poverty,
New
York,
(1981), "Measurement of Deprivation and
he Proportion Spent on Food: An Expiora
‘lopment, Vol.9, No.4, 1981, pp.337No.193.
1
(1971), Poverty
Economy.
in
1 Economic Rover
in Econometrics,
India,
Poona,
ey o-
.51
OUD
"A
Class
Vo 1.=2 (3) ,
P ov e ty
Asian Development Review,
ikwani,
N.C.
Econometrics,
(1980),
"On a Cl.
Vol.43, pp 437-446.
of
Poverty
or
PP-
Vol. 10,
Measures
K.Su’pba Rao (1990),
"Rural Poverty and Its
India", Economic and Political Meekly, March
,
(1986)
"Budgetary Rul'es for Poverty
Alleviation", Department of Economics, Universit
and Uoodrow Ui Ison School, Princeton University.
11.
12.
Keen,
Vol .
M. ( 1992),
102, J anua
Kumar ., T.K . , and
Pune t i ons
ano
M. i n h a s
L. .-
PO7=
ner.es
r or
Rove
The Economic Journal
67-79.
vid,(1972) 'Determination of
Allocation
of
Resour
of Business and Economic
ties I Association.
and S.D.Tendulkar ( 1991 ) ,
the 1980s:Evidence
167.3- 1682.
Meekly, July 6-13,
P ub I i
On the Coverage
Alleviation",
Journal
57-79.
ty:An Ordinal Approach
No.2, pp. 219-231.
to
READER IN ECONOMICS. GaHALE
"
Or POLITICS 4, ECONOMOS. PONE 4»0u .
The R.B.R.R. Kale Memorial Lecture, 1981, was delivered by Pro
fessor V. M. Dandekar, on Sth June 1981 at the Institute. The subject
of the lecture was “On Measurement of Poverty”.
Professor Dandekar joined the Gokhale Institute of Politics and
Economics in 1945. He was the Director of the Institute during 1966-68
and again from 1970 to 1980. He is also the Honorary Director of the
Indian School of Political Economy at Lonavala, which was conceived
and registered under his initiative in 1969-70. He has worked as member
or chairman of many committees and commissions set up by the State
and Union Governments as well as by international organizations. He was
the Chairman of the National Sample Survey Organization from 1970 to
1980. He has contributed several thought-provoking articles to professional
journals. Some of his important books are: Report on the Poona Schedules
of the National Sample Survey (G.I.P.E. Pub. No. 26) 1953; Use of
Food Surpluses for Economic Development (G.I.P.E. Pub. No. 33), 1956;
Working of the Bombay Tenancy Act-1948 (Report of Investigation)
(G.I.P.E. Pub. No. 35), 1957 (jointly with his colleagues at the Gokhale
Institute) and Poverty in India (jointly with N. Rath).
Measurement
of
Poverty
V. M. DANDEKAR
On his retirement on 5th January 1981 from the services of the
Institute, he was conferred the status of Professor Emeritus.
R. R. Kale. Memorial Lecture, 1981
Gokhale institute of Politics and Economics,
Pune 411 on I
Orient Longman Limited
Bombay Calcutta Madras New Delhi Bangalore Hyderabad Patna
n
o
3
I
)
vj
6
?
© Gokr.sle Inmiute of Politics and Ec^Knics
GN MEASUREMENT OF POVERTY
Pune 411 004
Shri Venkatraman, Professor Nilakantha Rath, Colleagues in the
Gokhale Institute, and Friends :
Price: Rs. 5.00
PRINTBD IS INDIA
“ Spicer College Press, Aundh Road, Ganeshkhind, Pane - 411 007■ ,„d
a
aid published by N’ihkanlhn Path ,, a,, P
•:nd cited
Kash at the Gokhale Institute of Politics and Economies
Pune 411 004.
’
As vou are aware, I retired from the service of the Gokhale Institute
of Politics and Economics just about five months ago on January 5.
Almost the following day. Professor Rath asked me whether I would
agree to deliver this year's Kale Memorial Lecture and quoted precedence
that I had made a similar request to Professor Gadgil soon after he
retired. I had therefore no escape. Unfortunately, before the end of the
week, I was involved in a road accident and since then, almost todate, I
have remained immobilised with a broken leg. I therefore felt uncertain
whether I would deliver this year’s Memorial Lecture. Professor Rath
was kind enough to extend the time to its limit and I am glad that today
I am able to be present here and, in the form of a Lecture, pay my
respects to the memorv of Late Rao Bahadur Raoji Ramachandra Kale,
the founder of the Gokhale Institute of Politics and Economics which
it was my. privilege to serve for the past thirty-five years and which after
my retirement, by conferring on me the status of Professor Emeritus, has
provided me a life-long seat to read and write and to learn and teach.
1 propose to take this opportunity to return to a subject I dealt with
ten years ago. In January 1971, Professor Rath and myself published, in
ihc form of two rather long articles in the Economic and Political Weekly,
a small study titled Poverty in India. The main purpose of the study was
to offer a rough estimate of the extent of prevailing poverty, to examine
policy alternatives, to emphasize that a massive transfer of incomes was
essential if immediate relief was to be provided to the poor, co indicate
the size of such needed transfer and to suggest a modus operand! by
means of which such a transfer might be effected. Naturally, the study
attracted considerable attention but mostly in its first part namely the
estimate of the extent of prevailing poverty. The data and methods em
ployed by us for the purpo.-e have been critically examined and, by
applying more sophisticated methods, somewhat different and presumably
improved estimates of rhe extent of poverty have been put forward. I
have rather studiously avoided joining this debate, first, because rhe re
vised estimates of the extent of proverty are often only marginally differ
ent from those given in our study; second, because 1 believe that the
quality of data on which these estimates are based does not justify much
greater sophistication in analysis; and third, because I thought that a
debate on the estimate of the extent of poverty might detract attention
from the policy needed to provide immediate relief to the poor.
3
But, more recently, our estimate of the extent of poverty has come
under more fundamental attack from two very senior and distinguished
academicians namely Professor V. K. R. V. Rao and Dr. P. V. Sukhatme.
(Rao, 1977. Sukhatme, 1977, 1978. The point of Professor V. K. R. V.
Rao's criticism is the following : "A method commonly employed for the
measurement of poverty is to take the nutritional norm in terms of daily
calorie intake by consumer unit, and the cut-off point by the expenditure
class which has an average daily calorie intake per consumer unit nearest
to the norm and then treat half of the population lying in this expenditure
class and the entire population in the lower expenditure classes as the
poor. This is the method employed by Dandckar and Rath (1971) in
their famous brochure 'Poverty in India’. And it has been followed bv
more studies using a more or less similar method.. . But) the methodology
. .. does not appear to be correct. While the proportion of under-nutrifional poverty undoubted!'' declines with increasing income, we find the
-paradoxical result that the pcor as denned also include the not-poor and
that the not-poor include the poor.” (Rao, 1977 . Professor V. K. R. V.
Rao seems to believe that we get what he calls the ‘paradoxical result’
that “the poor as defined also include the not-poor and that the not-poor
include the poor” because of our basing the estimates of poverty on the
sole criterion of calorie intake. Hence he proposes : ‘‘Povertv has to be
identified with deficiency in the total level of living. And total level of
living includes net only energy requirements but also balanced diet needed
for health, and the other components of basic needs essential for human
existence at a tolerable level. This exercise on the connotation and iden
tification of poverty has still to be undertaken in India.” (Rao, 1977.'.
The gist of Dr. Sukhatme’s criticism is as follows: “A person who cannot
afford a diet which meets his minimum energy needs for a healthy active
life is both poor and undernourished... However, the figure for calorie
requirement which the authors use is not the minimum below which a
person can be said to be undernourished, but is the average requirement
of a healthy active population of the ‘reference' tvpe. Consequendv, the
dimensions of poverty are exaggerated.” .Sukhatme, 1977., In fact, Dr.
Sukhatme argues that the ‘correct' estimate of poverty is only about onehalf the estimate put forward in our study.
Professor V. K. R. V. Rao and Dr. P. V. Sukhatme both had raised
these points in their presentations before the meeting on Problems in the
Measurement of Poverty during the 41st Session of the International
Statistical Institute held at Delhi in December 1977. In the discussion
that followed, I had made a brief reply. I propose now' to elaborate. I
shall begin with a few preliminary points.
The commonly used measure of the extent of poverty is the pro
portion of the population living below a certain level of living called the
poverty line. One may, if one desires, supplement this measure of the
extent of poverty bv another designed to measure what may be called the
intensity of poverty namely how many people live how much below the
poverty line. Tirus, basic to a measure of the extent as also the intensity
of poverty is a certain level of living defined as the poverty line. The
measures of povertv thus depend upon how we determine the poverty
line. That is what the debate is about.
The simplest method of determining the poverty line is to agree upon
a certain level of income or consumer expenditure as necessary to meet
the minimum needs of life. This was how a distinguished group, of which
Professor V. K. R. V. Rao was a member, set up by the Government of
India in July 1962 decided that a per capita monthly expenditure of
Rs. 20 at 1960-61 prices should be considered the national minimum or
the poverty line. The group did not reveal the basis of this determination
and hence the particular definition of poverty line appears rather arbit
rary. As we shall presently see, a certain element of arbitrariness is inevita
ble. However, the poverty line such as the one determined by the Group
is unsatisfactory for other reasons. A per capita monthly consumer expen
diture of Rs. 20 without a reference to the prices of a number of essential
commodities docs not make clear the implied level of living; and even
when prices arc indicated, such as the 1960-61 prices in the present case,
they fail to make the implied level of living sufficiently concrete and
explicit. Hence, while the poverty line may be defined as a certain level
of consumer expenditure, it is desirable and useful that the connotation
of the level of living associated with the particular expenditure level is
made explicit. This requires a criterion which, while it connotes different
levels of living, is also highly associated with levels of consumer expen
diture particularly in the range in which the poverty line lies.
There arc several such criteria. Probably the simplest is the propor
tion of expenditure spent on certain essential items of consumption such
as food. As is well known, poor living is characterised by a laige propor
tion of the total consumer expenditure taken up by items essential for
sheer physical existence and survival, such as food. One may therefore
define poverty line as that level of consumer expenditure at which a certain
proportion such as 80 per cent or 85 per cent of total expenditure is
devoted to food, or food and fuel, or food, fuel and light. In our study,
Proftssor Rath and myself did not use this criterion to define the poverty
line nor am I aware of anyone else having used it. But on second thought,
I sec great merit in it. The criterion is sufficiently concrete and vivid.
5
Besides, in view of the quality of available data, I expect that the poverty
line thus defined will be statistically more stable and reliable than if
based on some other criteria.
Other criteria connoting different levels of living are based on pre
scribed physical norms in respect of one or more items of consumption
considered important, essential or crucial. If one must choose one single
item, the choice inevitably falls 0:1 the energy intake in terms of calorie
value of food. In our study, Professor Rath and myself had taken the
norm to be 2250 calories per day per capita and defined the poverty
line as that expenditure level at which the average caloric intake met
this norm. But, in deference to the recommendation of the group
appointed by the Government 01 India referred to earlier, we had raised
the poverty line for the rural population slightly from Rs. 170 to Rs. 180
per annum at 1060-61 prices. I mention this only incidentally. It is
not relevant to the present discussion.
If one is willing to prescribe physical norms for more than one items
but still remain confined to food, one may prescribe a normative balanced
or minimum diet. The FAO has worked out normative diets for groups of
countries and one is available tor the South East Asia. For India, various
recommendations have been made by the Nutrition Advisory Committee.
Dr. P. V. Sukhatme 1965 has worked out two food baskets correspond
ing to a minimum and a medium concept. His minimum basket per day
per person consists of 0.403 kg. of cereals, 0.104 kg. of pulses, 0.201 kg. of
milk, 0.137 kg. of fruits and vegetables, and certain quantities of starchy
roots, sugar, oils and fats, and meat, fish and eggs. The Report of the
Second Pay Commission - Central Government Emplovees (1957-59)
gives a minimum diet reportedly constructed or. the advice of Dr.
Patwardhan. The diet per day per adult unit consists of 0.425 kg. of
cereals, 0.113 kg. of pulses, 0.113 kg. of milk, 0.170 kg. of fruit and
vegetables, and certain quantities of sugar, oils and fats but no meat,
fish or eggs. Thu -, while Dr. Sukhatme’s minimum diet includes certain
quantities of meat, fish and eggs. Dr. Parwardhan's minimum diet does
not include any neat, fish or eggs. There are also other differences
between the two recommended diets.
Given a ‘ba! meed’ or ‘minimum’ diet, the poverty line may be
defined as that c tpenditure level at which the households seem to have
the specified diet. In practice, it is not easy to be satisfied item by item
on this point. For instance, households may have more, milk and less or
none of meat, fish and eggs. Obviously it is neither possible nor meaning
ful to insist that households have precisely the prescribed diet, item by
item. Hence, the method usually adapted is to find the cost of the specified
diet and to take that level of total expenditure at which the households
seem to spend on the items of the specified diet an amount equal to its
worked out cost, not item by item but on all items put together. Thus,
suppose the cost, of the specified diet works out to Re. 1 per person per
dav, then the poverty line is that expenditure level at which the house
holds spend Re. 1 per person per day on items included in the specified
diet but not necessarily conforming to its prescribed composition. This is
what Bardhan and Rudra do. Professor V. K. R. V. Rao refers to their
studies approvingly. He says: “The balanced diet approach is..............
preferable to the caloric intake approach. And this is what writers like
Bardhan (1974), Rudra (1974) and others have done, unlike Dandekar
and Rath (1971) who have only used the calorie intake criterion.”
(Rao, 1977). Professor V. K. R. V. Rao prefers the balanced diet
approach because he believes that it takes into account the nutritional
quality of the calorie intake. In fact, as we see, in the application of the
balanced diet approach to determine the poverty line, the composition of
the balanced diet is abandoned and only its aggregate cost is retained.
All that it ensures is that the households on the poverty line spend on
their diets enough to give them a balanced diet; it is by no means certain
that they in fact buy balanced diets. I do not see that this is an improve
ment over the method which ensures that the households on the poverty
line have the specified caloric intake particularly if we see that a diet
adequate in calorics is almost always also a balanced diet.
Professor V. K. R. V. Rao, though he prefers the balanced diet to
caloric intake as a criterion for determining the poverty lute, is not
entirely satisfied with the balanced diet approach either. As I already
noted, he says: “Poverty has to be identified with deficiency in the
total level of living. And total level of living includes not only energy
requirements but also balanced diet needed for health, and the other
basic needs essential for human existence at a tolerable level.” And he
notes: “This exercise on the connotation and identification of poverty
has still to be undertaken in India.” In this Professor V. K. R. V. Rao
is of course right. But I am not aware if similar exercise has been done
in any other country. 1 sec great difficulty in defining a tolerable level
of human existence and identifying its basic needs namelv its material
content in terms of specified quantities of probably a hundred different
commodities. Moreover, even if this exercise is done and essentials of
human existence at a tolerable level are identified, judging by what
happens to the balanced diet when used to define the poverty line, it
seems inevitable that all the details of the tolerable human existence will
be abandoned and only its total cost will be used to define the poverty
line.
6
Thus, several criteria may be used to define the poverty line. I have
mentioned four such criteria, namely, i the proportion of expenditure
taken up by specified• essential items such as food; (ii) caloric value of
food; (iii) cost of a balanced diet; and finally, iv) cost of essentials
of tolerable human existence. The point to note is that, whichever criterion
we may choose, it is used to determine ar. expenditure level which meets
that criterion; and ultimately, it is the expenditure level so determined
and not the chosen criterion which defines the poverty line
But, alternatively, we might use the chosen criterion to define poverty
directly that is without the intermediation of the total expenditure. Thus,
we might identify the poverty line by a certain specified proportion such
as 80 per cent of the total expenditure spent on certain specified items
such as food and define poor as all those who spend more than 80 per
cent of the total expenditure on food. We may do this if we choose. But
we should recognise that this definition of poverty is different from the
one given by an expenditure lower than the one at which households
on an average spend a certain proponion of their expenditure on the
specified items. The two definitions arc related because households at
lower levels of total expenditure are seen to spend on an average higher
proportion of their total expenditure on the specified items. But the
two definitions arc not identical.
Similarly, taking me second criterion namely the calorie value of
food, we may identify the poverty line directly by a certain minimum
calorie intake and define poor as all these whose caloric intake is less than
the prescribed minimum. We may do this if we choose. But we should
note that this definition of poverty is different from the one given by an
expenditure lower than the one at which on an average the calorie intake
meets the prescribed minimum. The two definitions are related because
the average calorie intake of households at lower expenditure levels is
generally lower. But the two definitions are not identical.
In fact, the two definitions define two different, though related.
phenomena and it will help avoid confusion if we refer to them bv their
respective names, namely, poverty and under-nutrition. When a popula
tion is classified on the basis of a certain income or expenditure howsoever
determined, provided it is sufficiently low, we are defining poverty and
sorting out poor and not-poor so defined. On the other hand, if we
classify a population bv its energy intake, we arc trying to identify under
nutrition. Want of adequate income howsoever defined is poverty;
deficiency of energy appropriately defined is under-nutrition. The two
are related in the sense that statistically they go together. But the two
are not identical: in fact they arc two different phenomena.
In our study on poverty, Professor Ra± and myself have been
talking about poverty and not under-nutrition. We did not emphasise
this. Indeed, I should confess, in a couple of places we were guilty of
loose wording and treating povertv and under-nutrition as identical. For
instance, referring to rural population living on per capita annual ex
penditure of less than Rs. 170 at 1960-61 prices, or the urban population
living on per capita annual expenditure less than Rs. 271 at 1960-61
prices, we said that it lived on diets inadequate even in respect of calories.
This was clearly an error and we wish to stand corrected. The correct
statement would be that this population lived on such levels of consumer
expenditure that, judged by average standards of household management,
it could not provide for itself diet adequate even in respect of calories.
There would certainly be some households among the poor, defined by
a certain expenditure level, who with better household management and
better priorities of expenditure did provide for themselves diets adequate
at least in respect of calories. The contrary is also true. There would be
some households who were not poor by the same definition but who
nevertheless, by mismanagement of their households and wrong priorities
of expenditure, failed to provide for themselves diets adequate even in
respect of calories. Poor households do not cease co be poor simply
because they manage their households better than the average; and notpoor households do not cease to be not-poor because the management
of their households is below the average.
Table 1: National Sample Survey : 26th Round (1971-72) : Rural
Households
Monthly
expenditure
per con
sumer unit
Rs.
Number
of house
holds
Above
2300
Average
calorie in
take per
day per
consumer
uni:
1493
Number of households
with calorie intake per
day per consumer unit
Below
2300
0-15
444
404
40
15-21
120"
921
286
1957
21-24
313
466
347
2287
24-2R
1174
518
656
2431
28-34
1748
492
1256
2734
34-43
2028
300
1728
3127
4-3-55
1655
123
1532
3513
1319
53
1266
4016
75-100
598
10
588
4574
100--
482
11
471
6181
11468
3298
8170
2724
All classes
Direct statistical evidence on this point became available when the
consumer expenditure data from the 26th Round .1971-72) of the
National Sample Survey was published showing households cross-classified
by monthly expenditure per consumer unit and calorie intake per day per
consumer unit.
In his paper read before the 41st Session of the International Sta
tistical Institute held in Delhi in December 1977, Professor V. K. R. V.
Rao invited attention to these data Table I and commented as follows:
‘‘Previous writers on the subject have used the calorie intake norms to
determine the cut-off expenditure class for determining the magnitude
ot poverty. Taking a daily intake of less than 2300 calories for determining
the cut-off expenditure class, the relevant figure in respect of the N.S.S.
data for 1971-72 is the monthly expenditure class of Rs. 21-24. On this
basis, the number of households representing the poor among the consumer
units would be 2057 out of a total of 11468 or 17.9 per cent. But this
number includes 673 {it should be 499 households... whose daily intake
is above 2300 calories.’’ On the other hand, among the households with
expenditure above the cut-off point, there are 1740 households whose
daily intake is bei-w 2300 calories. Ir. fact, he points out, the number
of households with daily intake of below 2300 calories irrespective of the
expenditure class is 3298 compared to .he number 2057 living below the
cut-off expenditure defining poverty.
Professor V. K. R. V. Rao is dismayed by these results. He says:
"Clearly the methodology followed hitherto for estimating the magnitude
of rural poverty does not appear to be correct. While the proportion
of under-nutritional poverty undoubtedly declines with increasing income,
we find the paradoxical result that the poor as defined also include the
not-poor and that the not-poor include the poor.” In fact, there is nothing
wrong with the methodology. Professor V. K. R. V. Rao gets into what
he cads paradoxic?., results because he uses the terms 'poor* and ‘not-poor’
tn two different s -rises without seeing the difference. Indeed, he comes
close to seeing the difference when he uses two different terms, namelv,
‘poverty’ and ‘ur. Ur-nutritional poverty' but he misses the distinction.
\\ hat Professor \ K. R. \ . Rao calls ‘under-nutritional poverty’ is what
I earlier suggested we should call simply ‘under-nutrition’. Then all that
I rofessor . K. R \ . Rao discovers is that some poor are not under
nourished while some not-poor arc under-nourished. There is nothing
paradoxical in th>‘‘ result. Indeed, because poverty and under-nutrition
arc two different, .hough related, phenomena, a cross-classification of the
households bv thi wo criteria gives a fourfold classification :
Poor
Not-poor
Total
1558
1740
3298
Not-undcr-nourishcd
499
7671
8170
Total
2057
9411
11468
Under-nourished
Thus, with nutritional norm of 2300 calories as Professor V. K. R. V.
Rao has chosen, out of a sample of 11468 households, 2057 were poor,
3298 were under-nourished and only 1558 were both poor and under
nourished.
Professor V. K. R. V. Rao does not sec this. Instead he seems to
believe that he gets what he calls the paradoxical’ results because our
definition of poverty is based on the sole criterion of caloric intake. Hence
he advises the use of a broader base for defining poverty such as the
balanced diet or in fact all basic needs essential for a tolerable human
existence. In this, Professor V. K. R. V. Rao is again not right. He may
of course use the balanced diet or a comprehensive portfolio of basic
needs as the basis for defining the poverty line. But he must hold to one
definition of poverty. If he fails to do that, he will again get the
paradoxical results he is worried about. Whichever criterion we might
use to define the poor, if poor and not-poor arc to be mutually exclusive
classes, we must use one single definition of poverty.
Before I leave Professor V. K. R. V. Rao’s critique of our estimates
of poverty, I wish to make two incidental comments. The first concerns
Professor V. K. R. V. Rao's choice of the calorie norm and the estimate
of the rural poor he derives on that basis. On the basis of the data from
the 16th Round (1960-61) of the National Sample Survey and calorie
intake norm of 2250 per person per day, Professor Rath and myself
estimated that about one-third (33.12 per cent) of the rural population
lived below the cut-off expenditure of Rs. 170 per capita per annum.
Now on the basis of the data from 26th Round ( 1971-72) of the National
Sample Survey and a caloric intake norm of 2300, Professor V. K. R. V.
Rao estimates that only 17.9 per cent of the households live below the
cut-off expenditure. One might wonder whether the large drop from
33.12 per cent to 17.9 per cent in the magnitude of poverty is in fact
what happened between 1960-61 and 1971-72. Unfortunately, this is
not the case. The difference between 33.12 and 17.9 is almost entirely
due to two methodological inaccuracies in Professor V. K. R. V. Rao’s
presentation. First, he chooses the daily caloric intake norm of 2300 with-
10
out being sufficiently clear whether it is per person or per consumer
unit. The data from the 16th Round 1960-61 : of the National Sample
Survey is on per capita that is per person basis. The data from the 26th
Round 1971-72; of the National Sample Survey is on per consumer
unit basis. Hence, the norm of 2300 calories per day chosen by Professor
V. K. R. V. Rao is per consumer unit and is too low compared to the
norm of 2250 calories per person per day we had chosen. There arc
about 0.8 consumer units per person. Hence the norm of 2250 per person
per day is equivalent to 2812.5 calorics per day per consumer unit. Bv
the latest standards developed by FAO WHO, the caloric norm for
India is 2780 per day per consumer unit at the retail level. Hence, the
caloric norm chosen by Professor X'. K. R. V. Rao is too low and, in the
absence of an explanation, is very misleading. Second, wc had expressed
the magnitude of poverty as 33.12 per cent of population; Professor
X’. K. R. X’. Rao expresses it as 17.9 per cent of households. As is well
known, the poorer households’ have a larger size. Hence, expressing povertv
as a per cent of the households clearly understates its magnitude. If Pro
fessor X". K. R. X'. Rao had chosen a caloric intake comparable to the one
we had chosen such as the one given by the FAO/WHO and had expressed
the poverty as the proportion of the population living below the cut-off
expenditure level. Itis estimate of the magnitude of poverty could not be
so different. There is a large lay audience interested in the estimates of
the magnitude of poverty but net familiar with some of these techni
calities. It will help avoid confusion if all research workers, particularlv
the eminently senior ones, use standard methodology or explain when
they deviate from it.
The second incidental comment I wish to make concerns a very ele
mentary mathematical or statistical error appearing in Professor X7. K. R. V.
Rao's above-mentioned paper presented to the International Statistical
Institute. Professor X'. K. R. V. Rao says: “The data given earlier . . .
show that the range of intake varies from an average intake of 1493 calories
in the monthly expenditure class Rs. 0-15 to one of 6181 calorics in the
monthly expenditure class of Rs. 100 and above. In fact, all expenditure
classes above a monthly Egurc of Rs. 55 show an average intake per con
sumer unit of above 3500 calories. One docs not know how to explain
this phenomenon in nutritional terms. The N.S.S. data do not also sup
port the normally held thesis that the relation between caloric intake
and monthly expenditure increases rapidly to start with and graduallv
thereafter till it stabilises itself at a given maximum level.” Professor
X’. K. R. X . Rao proceeds to demonstate his point on the basis of the
N.S.S. 26th Round (1971-72) data ‘Table 2).
11
Table 2: National Sample Survey : 26th Round (1971-72) : Rural
Households
Monthly
expenditure
per con
sumer unit
Rs.
Mid-point
of monthly
expendi
ture class
Average
calorie in
take per
day per
consumer
unit
Percent
increase
over the
preceding
class
(1)
(2)
(3)
(4)
Rate of
increase in
calorie in
take per
rupee in
crease in
monthly ex
penditure
(□)
0-15
7.5
1493
—
—
15-21
18.0
1957
13.1
44.2
21-24
22.5
2287
16.9*
73.3
24-28
26.0
2431
6.3
32.0
28-34
31.0
2734
12.5
60.6
34-43
38.5
3127
14.4
52.4
43-55
49.0
3513
12.3
55-75
65.0
4016
14.3
75-100
87.5
4574
13.9
24.3
100-
(150.0)
6181
35.1
(22.2)
36.8
.
31.4
*Tne printed figure is U.2\which is either a numerical or a printing error.
From the average caloric intake per day per consumer unit for each
monthly expenditure class, Professor V. K. R. V. Rao works out the
percentage increase in the calorie intake of a class over the preceding
class (Col. 4). It is obvious that the percentage increase over the preced
ing class shows no sign of declining. Indeed, in the last class of Rs. 100-r,
there is a sharp increase. On this, Professor V. K. R. V. Rao remarks:
“The continuity of the rise in calorie intake of each expenditure class
over the preceding class culminating in a rise over the immediately
preceding next highest class is certainly an unusual phenomenon in
nutritional behaviour and raises doubt about the credibility of the N.S.S.
data ...” It is obvious that Professor V. K. R. V. Rao is making an
elementary mathematical-statistical error to argue his point. The per
centage increase in the average calorie intake of each class over the
preceding class is not relevant to the point which Professor V. K. R. X'.
Rao desires to make. The appropriate rate to calculate is the rate of
increase in calorie intake per rupee increase in the monthly expenditure.
For this purpose it is necessary to know the average total monthly expen
12
13
diture for each class. This is not given in the published data. Hence,
I have taken the mid-point of rhe expenditure class as an approximation
of the average expenditure and calculated the rate of increase in caloric
intake per rupee increase in monthly expenditure (col. 5). For the last
open-ended class of Rs. !00“, I have taken the average expenditure to
be Rs. 150, which I suppose is conservative and calculated the correspond
ing rate of increase in caloric intake. It will be seen that the rate of
increase declines continuously after monthly expenditure of Rs. 34-. In
fact, if we neglect the exceptionally low figure in the expenditure class
Rs. 24-28, it might appear that the rate of increase declines almost
continuously after the monthly expenditure cf Rs. 24. The data therefore
do not support Professor V. K. R. V. Rao’s contention.
in applying this criterion to estimate the extent of poverty they have
misused the meaning of requirement. In particular, they have mistaken
the average energy need of an individual for the minimum need ignoring
the fact that energy needs vary between and within individuals even of
the same age-sex groups.’
This does not mean that everything is all right with the N.S.S. data.
The rate of increase in caloric intake per rupee increase in monthlv
expenditure declines but not rapidly enough so that the caloric intake
might reach its maximum at a very high level. The high caloric intake
in the higher expenditure classes in the N.S.S. data arc known to be
over-estimates and at least one reason for it is explicitly stated in - the
N.S.S. reports. 1 he 2-.S.O. docs not count guests and domestic and farm
servants, even when eating in the households, as its members; but it fails
to exclude their cct-sumption from the household consumption. This
affects the per consume, unit caloric intake particularly in the higher
expenditure classes m rural areas. There are also other reporting errors
such as inclusion oi pou.iry and cattle feed in household consumption.
Cases where very large per consumer unit caloric consumption is re
ported, certainly need careful scrutiny.
Let me now turn to the critical attack which Dr. P. V. Sukhatmc
has launched on our estimate of poverty. This attack is on a completely
difft rent plane unfamiliar to and unsuspected by economists including
ourselves. Hence, in ord r to make clear the essential point of bis criti
cism, I propose to quo -: him rather at length. Dr. Sukhatmc says:
‘■Analysis of data confirms that as income increases, the energy intake
increases, rapidly to start with and gradually thereafter indicating that
an appreciable number cf people remain undernourished for want of
adequate income . . E oi.omists were quick io see in this analysis that
poverty was the principal cause of the large and widespread incidence
of malnutrition and began to use the minimum energy requirement as
the criterion for estimating the extent of poverty. A person who cannot
afford a diet which meets his minimum energy needs for a healthy
active life is certainly bath poor and malnourished. Thus Dandekar and
Rath were among the first economists to apply this criterion. However,
Dr. Sukhatmc proceeds to explain: “As is well known the figure for
average energv need is based on measurement of energy intake in healthy
active subjects of the ‘reference’ type. Consequently, one expects that if
the population is hcalthv and active, about half cf the individuals will
have intakes less than the average and half more than the average. \\ hen,
therefore, Dandekar and Rath find that 40 per cent of rhe rural and 50
per cent of the urban India live on a level of total expenditure below the
level corresponding to the average energy need for the country', it simply
means that they find a situation which is in line with what one expects
to find in a healthy active population of the ‘reference’ type.”
Dr. Sukhatmc then gives a telling example of the kind of inference
which he believes Professor Rath and myself are involved in. He says:
"We have in fact an example of such assessment carried out in Great
Britain by the late Sir Arthur Bowley. He found that 50 per cent of the
population in U.K. were below the average need for U.K. and he con
cluded that some 50 per cent of the population must be under-nourished.
I need hardlv add that, notwithstanding the eminence of Sir Arthur
Bowlcv, the conclusion was rejected bv the Government of L.K.” Dr.
Sukhatmc proceeds: "This was 40 years ago when the concept cf require
ment had hardly developed to a point to grasp its full implications. But
to adopt the same method today that was rejected as inapplicable in the
U.K. decades ago. is to ignore the knowledge we have gained in under
standing the concept of physiological requirement.” Impliedly, Dr.
Sukhatmc suggests that Dandekar and Rath are adopting the same method
which Sir Arthur Bowley used forty years ago and which was rejected as
not valid.
I have not been able to check the reference to Sir Arthur Bowley’s
work. But 1 feel quite uncertain that Sir Arthur Bowley in fact did what
Dr. Sukhatmc alleges he did; because, notwithstanding his eminence, Dr.
Sukhatmc clearly fails to sec the difference between one procedure and
another^xhe procedure which Sir Arthur Bowley allegedly followed and
the procedure which Dandekar and Rath follow. If we had followed the
same procedure as Sir Arthur Bowley allegedly did, we would have classi
fied the households by their daily per person caloric intake and declared
all those households whose caloric intake fell below the norm to be poor.
We did not do this and Dr. Sukhatmc know; this well. Even by Dr.
15
14
Sukhatme's account, Sir Arthur Bowley was trying to estimate the extent
of under-nutrition; we are estimating the extent of poverty. Like Professor
V. K. R. V. Rao, Dr. Sukhatme fails to see the difference between poverty
and under-nutrition.
For instance. Dr. Sukhatme says that a person who cannot afford a
diet which meets his minimum energy needs for a healthy active life is
certainlv both poor and malnourished. Note that Dr. Sukhatme says that
the person is both poor and malnourished. It means that he recognises
that povertv and under-nutrition arc two different phenomena but he is
unable to see the distinction clearly. Hence he says that when a person
cannot ajord the needed diet, he is both poor and under-nourished. This
is the source of confusion. When we say that a household cannot afford
the requisite diet, we mean that as judged by the average consumer be
haviour and household management at that expenditure level. The house
hold is therefore poor. But, as I have already explained, a poor household
need not necessarily be under-nourished; by better priorities and better
management of expenditure, it may in fact cat an adequate diet. Simi
larly, and for the opposite reason, a not-poor household may in fact suffer
from under-nutrition. I wish to emphasise and plead with Dr. Sukhatme
to understand that, all through our little study on Poverty in India, Pro
fessor Rath and myself have been discussing poverty and not under
nutrition. The two are related because as Dr. Sukhatme points out "as
income incrcrw.-, the energy intake increases, rapidly to start with and
gradually thereafter indicating that an appreciable number of people
remain under-nourished for want of adequate income.” But the two are
not identical; in fact, they are two different phenomena.
Consider again Dr. Sukhatme’s statement : “When . . . Dandekar
and Rath find that 40 ncr cent of the rural population and 50 per cent
of the urban India live on a level of total expenditure below the level cor
responding to tl.e average energy need for the country, it simply means
that they find
situation which is in line with what one expects to
find in a h-..!lhy active population of the ‘reference’ tvpe.”
He argues thi- c; the ground that, in a healthy active population, one
expects to find that about half the individuals will have energy intakes less
than the aver,. 'nd half more than the average; and he docs not sec the
difference. Dr. Sukhatme docs not see the difference between half the
population ha. ing calorie intake less than the average and half the popu
lation living I - ’ow the total expenditure level corresponding to the average
energy need. I a healthy active population one would expect to find
about half the population to live on caloric intake below the average. But,
in such a pep.I.-J ion, should wc also expect about half the population
to live on total expenditure less than the expenditure corresponding to the
average caloric intake? I ask, is this also an essential attribute of a healthy
active population? Dr. Sukhatme is confused.
But let us return to the main ground of Dr. Sukhatme's criticism of
our estimates of the magnitude of poterty. It is his explicit recognition of
the existence of inter and intra individual variation in energy needs. Inter
individual variation means that the energy intake of even normally healthy
and active individuals with similar body weight and occupation varies
implying that some individuals are more efficient machines than others.
Intra-individual variation means that energy intake of an individual en
gaged in similar activity and maintaining body weight varies from day to
day. L am not familiar with the literature on this subject. But I have no
difficulty in. admitting such variation on a priori grounds. It means that the
daily, weekly, or monthly energy intake of different individuals wiil show
variation even if all of them arc normally healthy and active.
A statistical measure of such variation is called the standard deviation.
As for its magnitude, Dr. Sukhatme admits that the available data arc
scantv. The most recent evidence, and the one on which Dr. Sukhatme
seems to rely most, is from a study reported by Edholm and his associates
in 1970. The study was carried out on army cadets engaged tn verysimilar activities and covered three weeks in a ten-weck period. I regret
to say that in reporting the results of this studv, Dr. Sukhatme does
not meet the norms of accuracy expected of a scientist of his standing and
reputation. Nowhere wc are told how many cadets were involved; probably
they were no more than 20 or may be 30. There are wide differences
in the results as Dr. Sukhatme reports them in his Lal Bahadur Shastri
Memorial Lecture delivered tn January- 1977 and in his article in August
1, 1978 issue of the Economic and Political Weekly. His own assessment
regarding the magnitude of the standard deviation based on these results
also seems to suffer from considerable intra-Sukhatmc variation. For
instance, in his Lal Bahadur Shastri Memorial Lecture (January 1977),
he says that “the requirement of an individual will vary around the mean
value with a standard deviation of approximately 400 calorics.” In his
paper contributed to the 41st Session of the International Statistical
Institute (December 1977), he says that “the standard deviation of an
individual’s requirement can be niaced at approximately 300 calories.”
In his paper in the Economic and Political Weekly August 1978), he
says that “requirement of an healthy active adult will vary around the
mean values with standard deviation of approximately 375 calorics.” As
much as I know, this is his last pronouncement. Hence I thought I should
proceed on this basis. Cnfonunately, there is also intra-paper variation.
In the same paper, in the same paragraph, Dr. Sukhatme says that “the
16
coefficient of variation of requirement can be placed at approximately
15 per cent.” This means that the standard deviation may be placed at
15 per cent of the average requirement which Dr. Sukhatme lakes to be
2750 calories. This puts the standard deviation at 412.5 calories. As I
shall presently show, to be consistent with what Dr. Sukhatme does later
in the same paragraph, we shall have to place the standard deviation of
an individual’s requirement at 450 calorics. All through, I presume that
the individual is an adult.
The National Sample Survey does not report data on the consump
tion or energy intake of individuals but of households which are groups
of individuals or groups of consumer units. In the 26th Round (1971-72)
of the National Sample Survey, the sample of 11,468 rural- households
consisted of 49,198 consumer units. Thus the average size of a household
is 4.29 consumer units which, for convenience of numerical exposition,
we may take to be 4.0 as Dr. Sukhatme does. The mean or average per
consumer unit energy requirement of households is of course the same as
the average requirement of individuals namely 2750 calories per day per
consumer unit. But, because the households arc groups of individuals,
the variation in their per consumer unit energy requirements is smaller
than the same among individuals. By a simple rule, if the standard devi
ation of individual requirements is s, the standard deviation of per con
sumer unit requirement of households of an average size of 4.0 consumer
units will be approximately s/2, the divider 2 being the square root of 4,
namelv the size of the households. Thus, taking the standard deviation
of individual requirements equal to 450 calorics, the standard deviation
of household requirements per consumer unit will be 450/2 = 225 calo
rics. To sum up: As a consequence of inter and intra individual variation
in energy requirement, we should expect that the per day per consumer
unit energy requirements of households will vary; or, in statistical par
lance, will be distributed with mean or average requirement of 2750 calo
ries and a standard deviation of 225 calorics.
The data are too meagre to judge the shape or form of this, distri
bution. In the absence of specific knowledge, it is customary in statistical
practice to assume that the distribution is ‘normal’ which, please note,
has no normative connotation. The normal distribution is a particular
statistical distribution found to hold good in many diverse situations and
particularly satisfactory in the case of biometric data such as the energy
requirements. Hence, in what follows, I shall assume that the per day per
consumer unit energy requirements of households are normally distributed
with mean 2750 calorics and standard deviation 225 calories. For purposes
of illustration, a numerical distribution of 100,000 housesholds distributed
in this manner is given in Table 3.
17
Table o: Normal distribctiox of households by per day per con
sumer unit calorie requirement
m = 2750; s = 225)
Energy requirement in calories
per day per consumer unit
less than (m—3s) i .e. less than 2075
Number of households out
of a total of 100,000
(m—3s) to (m—2s) i.e. 2075 to 2300
135
2,140
(m—2s) to (m—s) i.e. 2300 to 2525
(m—s) to (m) i.e. 2525 to 2750
34,135
(m) to im+s) i.e. 2750 to 2975
34,135
(m+s) to (m+2s) i.e. 2975 to 3200
(m+2s) to (m+3s)1 i.e. 3200 to 3425
more than (m+3s) i.e. more than 3425
13,590
13,590
2,140
135
100,000
Thus, in a population of 100,000 households of healthy and active
members with an average per day per consumer unit energy requirement
of 2750 calories, wc may expect 135 households with energy requirement
less than 2075 calorics, 135 + 2140 = 2275 that is 2.275 per cent house
holds with energy requirement less than 2300 calories, 135+2140+13590
= 15,865 that is 15.865 per cent households with energy requirement less
than 2525 calorics per dav per consumer unit. There will be equal num
ber of households on the other side with energy requirements more than
3425, more than 3200, and more than 2975 calories per day per con
sumer unit respectively. Even the remaining middle 34,135 + 34.135 =
68,270 that is 68.27 per cent of the households will have energy require
ments ranging between 2525 and 2975 calories per day per consumer
unit. Such is the consequence of the existence of inter and intra indi
vidual variation in energy requirement.
From this, Dr.' Sukhatme draws the following conclusions: “When
rhe observed intake for any day or period is therefore less than the aver
age requirement,... it cannot be taken to imply that a man is under
nourished, as Dandckar and Rath do, unless his intake is so low as to be
below the lower limit of confidence interval for the chosen level of signi
ficance. . . It follows that in any observed intake distribution on nutrition
unit basis with a nutrition unit having the same daily requirement as the
reference adult, namely m, the proportion of the population, below
(m—2s) will determine the incidence of undemutrition and poverty.”
19
18
I should once again remind that Dandekar and Rath do not say
that an observed intake less than average requirement implies under
nutrition. In fact, Dandekar and Rath do not compare the obsetved
intake with the average requirement except to identify the expen
diture level where the observed average intake is equal to the
average requirement. Thereafter, Dandekar and Rath compare only the
total expenditure with the cut-off expenditure level so identified and say
that all households with expenditure less than the cut-off expenditure are
poor. Dandekar and Rath do not talk of under-nutrition. As I have al
ready argued. Dr. Sukhatme is not able to sort out the two phenomena
of under-nutrition and poverty'.
But let me proceed. First, it will be useful to clarify in what sense
m—2s) = 2300 calories is the 'lower limit of confidence interval for the
chosen level of significance’. This is part of the logic of statistical infer
ence which commonly takes the form of what is called a ‘test of signifi
cance’ or ‘a test of a hypothesis’. The procedure is to set up a hypothesis
called the null hypothesis and examine given evidence in the light of this
hypothesis. It is highly unlikely that the evidence is in complete accord
with the hypothesis; some amount of deviation is common. The question
is then asked as to how frequently, that is with what probability, the
deviation as large as or larger than the observed one would appear if the
hypothesis were true. If this probability is very low, rather than holding
that the hypothesis is true but a rather improbable evidence has turned
up, the evidence is taken to contradict the hypothesis and on that ground
the hypothesis is rejected. How low must this probability be in order that
the null hypothesis is rejected, is a matter of choice. The chosen level of
probability is called the ‘critical level of significance’. It is customary
to work with either 5 per cent or 1 per cent level of significance. It means
that the null hypothesis is rejected only if evidence turns up so unfavour
able that the probability of such or more unfavourable evidence turning
up, if the null hypothesis were in fact tine, is less than 5 per cent of less
than I per cent depending upon the chosen level of significance.
In the present case, the null hypothesis is that there is no under
nutrition. It means that we shall not accept the existence of under-nutri
tion unless the evidence is overwhelming. If we meet a household whose
energv intake is below the average requirement, we shall suppose that its
particular rquircment must also be below the average or in fact below its
actual intake; it eats less because it needs less. We take cognisance of
its low intake as possible evidence of under-nutrition only if the intake is
so low that the probability of the requirement being lower still is very
low. Dr. Sukhatme suggests that we may consider a household to be under
nourished' only if its intake is below 2300 calories per day per consumer
unit. As we have seen, on the basis of a normal distribution of require
ments of households with mean 2750 and standard deviation 225 calories,
the probability of a household having its requirement less than 2300
calories is 2.275 per cent. Dr. Sukhatme considers this to be suffici
ently low to admit the household to be under-nourished. It means that
his chosen level of significance is 2.275 per cent.
With lower probability level of significance, the lower limit of ‘no
under-nutrition’ will also be lower. For instance, if we choose 1 per cent
rather than 2.275 per cent level of significance, the lower limit of ‘no
under-nutrition’ will be 2225 calorics instead of 2300 calories: naturally,
our estimate of the magnitude of under-nutrition will also be lower. Thus,
if one accepts Dr. Sukhatme’s logic, the estimate of under-nutrition de
pends crucially upon the choice of the level of significance. Hence, Dr.
Sukhatme should have explained and justified the choice of the particular
level of significance. He does not do it. Indeed, he dees not even mention
the level of significance he has chosen. He merely refers to it as the
'chosen level of significance’ as though it was chosen bv God.
Tabled: National Sample Survey : 26th Round (1971-72): Rural
Households. All expenditure classes.
Caloric
intake
per day
per
consumer
unit
Number
of sample
house
holds
Number
of
consumer
units
per house
hold
Number
of house
holds
with calo
rie requi
rement
as in
(1)
Revised
number
of house
holds
with calo
rie requi
rement
as in
(1)
Estimated
number
of deficit& excess
incake
house
holds
(1)
(2)
(3)
(4)
(5)
(6)
Upto 1500
651
—
—
651
1501-1700
455
576
5.01
4.77
—
—
—
455
—
1701-1900
4.75
1901-2100
752
4.71
22
5
576
/Art
2101-2300
854
4.79
2301-2500
947
4.51
239
1267
56
300
793
647
2501-2700
882
4.55
125
1234
4.31
3198
5214
“57
2701-3000
1234
3001-3500
1774
4.13
1523
360
•4008.**
1414
3501-4000
4000 & above
1174
3.92
5
I
1173
2159
3.42
—•
—
2159
All groups
11468
4.29
11468
2714*
(4716)***
balanced-intake; ** deficit-intake; ***>
excess-intake
'20
21
Let me proceed and apply Dr. Sukliatme's advice to the distribution
of households and corresponding consumer units by their caloric intake
per day per consumer unit .is given by the 26th Round (1971-72) of
the National Sample Survey Table 4 .
a group of 100,000 households, the existence of 2275 households with
energy intake less than 2300 calories offers no evidence of under-nutrition
and hence they cannot be judged under-nourished. This is the difference
between the two situations; in one, we are judging a single household
with energy intake less than 2300 calories; in the other, wc have a large
number of households of which a small number have their energy intake
less than 2300 calories. The logic of statistical inference appropriate to
the first situation is not appropriate to the second.
According to these data, out of a total 11,468 sample households,
3298 households, which is 28.76 per cent, had energy intake less than
2300 calories per day per consumer unit. Even on Dr. Sukhatme’s reckon
ing, these must be considered under-nourished. Moreover, the households
with lower energy intake are seen to have larger number of consumer
units. Hence, in terms of consumer units, the proportion of under-nourish
ed is even larger; it is 32.21 per cent. This is the estimate of under
nutrition Dr. Sukhatme should accept because it is based on his own
advice.
But. as I snail presently demonstrate, the procedure is not valid. Let
us see its implications. As 1 have already mentioned, Dr. Sukhatme accuses
us of having "mistaken the average energy need of an individual for the
minimum need ’ treating everybody with energy intake below the average,
say 2750 calories, to be under-nourished. As I have explained and empha
sised, we did not do this; this is not our procedure. But, now when Dr.
Sukhatme says that all those and only those whose energy intake is less than
2300 calories should be counted as under-nourished, does he realise that
what he calls the minimum’, howsoever defined, is considered the require
ment for everybody? Once we recognise the existence cf inter and intra in
dividual variation in requirement, each individual on each occasion has a
particular requirement and whether he is under-nourished must be deter
mined bv comparing his observed intake with his particular requirement.
Like the individual intakes cannot be judged by the average, they cannot
also be judged by any single figure called the ‘minimum’ howsoever
defined. Dr. Sukhatme does not see this because, as I shall presently argue,
he is applying the logic of test of significance appropriate to one situation
to another where it is not appropriate.
As I explained, if we meet a household with caloric intake less than
2300 calories per day per consumer unit, wc would admit it as under
nourished because the probability of its requirement being less than 2300
calories is 2.275 per cent which wc consider too low. This accords with
the logic of statistical test of significance. But suppose that we have a
group of 100,000 households and among them 2275 households arc seen
to have calorie intakes less than 2300 calories per day per consumer unit.
should we consider them as under-nourished? Dr. Sukhatme would and he
is wrong. In a group of 100,000 households wc expect about 2275 house
holds to have energy requirements less than 2300 calorics; therefore, in
Now. suppose that in a group of 100,000 households, 5000 house
holds have energy intake less than 2300 calories. Should we regard them
all as under-nourished? No, not ail the 5000 are under-nourished; but
some of them arc. In a group of 100,000 households, the existence of
2275 households with energy intake less than 2300 calories would not
constitute evidence of under-nutrition. But any larger number of house
holds such as 5000 with intake less than 2300 calorics, would constitute
evidence of existence of under-nutriticn. In such a situation, our estimate
of under-nutrition should be not 5000 but only the excess (5000 — 2275 ;
= 2725 households.
Let me apply this principle to the distribution of households by the
caloric intake given by the 26th Round '1971-72) of the National Sample
Survey (Table 4). The total sample of rural households is 11.168 house
holds. Wc should first. find the distribution of these households by their
energy requirements on the basis of a normal distribution with mean
2750 and standard deviation 225 calorics 'Col. 4). If we compare the
frequencies of energy intake with frequencies of energy requirement in
all caloric classes, it will appear that the intake frequencies exceed the
requirement frequencies in all caloric classes upto 2300 calorics and in
all classes above 3000 calories. In the middle classes from 2301 to 3000
calorics the intake frequencies are short of the requirement frequencies.
The excess intake frequencies in the lower calorie classes is evidence that
among the sample of 11,468 households, some households have intake
short of requirement. The excess intake frequencies in the higher calorie
classes is evidence that some households have intake in excess of require
ment: or the intake is over-stated. The deficit in the middle calorie classes
is the consequence of the excess in the lower and higher clades.
To sum up: A comparison of the frequency distribution of households
by energy intake with their frequency distribution by energy requirement
shows that in the sample of 11,468 households, there must be some
households with intake short of requirement; and some households with
intake in excess of requirement. It is only for the remaining households
that intake matches requirement. The problem, as I see it. is how to
22
23
estimate the number of households in the three sub-groups, namely, the
deficit-intake group, the excess-intake group, and the balanccd-mtake
group. I am not aware if a solution has been suggested to this or similar
problem either in statistical theory or practice. Hence. I shall propose
a rough and ready solution leaving a more sophisticated resolution of the
problem to more competent hands.
I suggest that this should be the procedure to estimate under-nutri
tion from an observed distribution of individuals or of households by
their energy intake, once we recognise that the individual energy require
ment is subject to inter and intra individual variation. Dr. Sukhatme’s
assertion that, in such a distribution, the proportion of population below
the mean-minus-twicc-standard-dcviation limit estimates the incidence of
under-nutrition is not correct.
1 propose that the number of households with balanced-intake should
be so determined that when distributed by energy requirement (Col. 5),
the freouencies in the middle calorie classes do not exceed tire actual in
take frequencies in those classes. In the present case, a little arithmetic
shows that, on this basis, out of the total of 11,468 households only about
2714 households may have balanced-intake. When these are distributed by
their energy requirement (Col. 5), the excess intake frequencies in the
lower calorie classes provide the estimate of tire deficit-intake households;
in the present case it is 4008 (Col. 6). The excess intake frequencies
in the higher caloric classes provide the estimate of excess-intake house
holds; in the present case, it is 4746 (Col. 6). Thus, the sample of
11.468 households seems to be composed of three sub-groups; 4008 deficit
intake households; 4746 excess-intake households; and 2714 balanced-in
take households.
Surprisingly, Dr. Sukhatme does not follow his own, though wrong,
advice and apply his procedure to the distribution of households by their
per consumer unit calorie intake available from the 26th Round 197172) of the National Sample Survey (Table 4). He recognises that the
individual energy requirement is subject to inter and intra individual varia
tion; but then he completely misses the point. He makes no use whatever
of the distribution of the households by the per consumer unit calorie
intake. Instead, not seeing the difference, he uses the distribution of house
holds by the per consumer unit expenditure, as we do, but uses 2300
calories, and not 2750 calorics, to determine the cut-off expenditure level.
It will be noticed that all the deficit-intake households do not have
intake less than 2300 calories; there are deficit-intake households with
energy intake in the range of 2301-2500 calories as also 2501-2700
calories. Nevertheless, they must all be considered as under-nourished
in the sense that their intake is smaller than their requirement. On
this basis, 4008 out of 11,468 that is 34.95 per cent households arc under
nourished. As before, if we convert the households into consumer units,
the proportion of under-nourished will be larger; it is 33.70 per cent.
I should emphasise that this estimate of under-nutrition does not depend
upon any arbitrarily defined minimum or lower limit such as 2300 calorics.
It depends only upon the distribution of the households by their energy
requirement and hence essentially on the standard deviation of this dis
tribution. In all the above numerical illustration, I have taken the standard
deviation of the individual requirements at 450 calorics or the standard
deviation of the household requirements at 225 calories so as to remain
consistent with (m—2s) = (2750 — 2 X 225) = 2300 calories. I should
mention that this is higher than any value of the standard deviation
indicated by Dr. Sukhatme. With lower standard deviation, the estimate
of the magnitude of under-nutrition will be higher than the above estimate
of 38.70 per cent.
Thus, the difference between the procedure adopted by us and the
one Dr. Sukhatme proposes rests on the level of energy intake one should
choose in order to identify the poverty line or the cut-off expenditure
level. Applying our procedure to the data from the 26th Round (197172' of the National Sample Survey Table 5), we would take 2750 calories
to be average requirement and, seeing that the average intake in the expen
diture class Rs. 28-34 is 2734 calories, identify the cut-off expenditure
level somewhere in the middle of this expenditure class and, estimate that
about 46.4 per cent of the consumer units live on expenditures below this
level and hence are poor. Dr. Sukhatme says that this procedure is wrong;
that it ignores the inter and intra individual variation in energy require
ment; that it treats energy requirement as a fixed quantum such as 2750
calories. In fact, as we have seen, the energy requirement of households
is distributed with a mean 2750 calorics and a standard deviation 225
calories. Hence, Dr. Sukhatme says, that rhe actual energy requirement
of households in an expenditure class could be as low as (2750 — 2 X
225) = 2300 calories, being the lower limit of the confidence interval at
the cho-tn level of significance: and therefore, the cut-off level of expen
diture should be identified as that expenditure at which the actual energy
intake equals, not 2750 calorics, but 2300 calories. On this basis, the cut
off expenditure level is located somewhere in the expenditure range Rs. 2124 and about 20 per cent of the consumer units arc seen to live on expen
ditures below this level. Dr. Sukhatme concludes that only about 20 per
cent of the rural population (measured in consumer units) is poor and not
46.4 per cent as we would estimate it. Dr. Sukhatme says that Dandekar
and Rath have greatly exaggerated poverty .
Table 5: Nation al Sample Survey ■ 26th Rocsd ;1971-72): Rural
House:rIOLDS
Monthly expen
diture per con
sumer unit
Rs.
Number of
households
.Average num
ber of consumer
units per house
hold
Average calorie
intake per day
per consumer
unit
4.99
4.74
1493
1957
2287
0-15
4^1
15-21
1207
91.2 J.
313
1174
1748
2028
4.78
4.51
4.44
43-55
55-75
75-100
100-
1655
’3iG
4.08
3.70
598
3.31
482
2.84
Ail classes
11468
4.29
24-28
28-34
34-43
4.20
2431
2734
3127
3513
4016
4574
6181
I am afraid that Dr. Sukhatme is making a serious statistical error
.As I have said, I readily admit the existence of inter and intra individual
variation in energy requirement of individuals. I also accept the mag
nitude of this variation as Dr. Sukhatme has indicated it; namely, a
standard deviation of 450 calories for energy requirement of individuals
and 225 calories for energy requirement of households. But I ask, what
is the standard deviation of the energy requirement of a group of
households - because now we are concerned with groups of households
in different expenditure classes. For identifying the cut-off expediturc
level, we examine the actual average intake of the group of households
in each expenditure class and we must compare it with the average
energy requirement of a group of the same size. Because the energy'
requirement of individual households is variable, the average requirement
of a group of households is also variable. This is agreed. But how variable
it is? In other words, what is its standard deviation? Not the same as
the standard deviation of energy’ requirement of individual households.
Dr. Sukhatme has overlooked this and has continued to work with a
standard deviation of 225 calories even when he is considering the average
energy requirement of groups of households in different expenditure
classes. That is the basis of his determining the cut-off expenditure level
at the average intake of .2750 — 2 X 225) = 2300 calories. The error
is obvious.
Dr. Sukhatme should know that if the standard deviation of the
energy requirement of a household is 225 calories, the standard deviation
of the average requirement of a group of 1748 households is not 225
calories but 225/41.8 = 5.4 calorics. 41.8 being the square root of 17.48.
Hence, with the same level of significance as chosen by Dr. Sukhatme,
the average energy need of the .group of 1748 households in the expendi
ture class of Rs. 28-34, cannot be placed below (2750 — 2 X 5.4)
= 2739.2 calorics and one cannot be far wrong by placing it at 2750
calories. I confess that Professor Rath and myself did not take explicit
cognisance of inter and intra individual variation in energy requirement
of individuals but I submit that its neglect at the level of groups of
several hundred households does not vitiate our estimate of poverty. On
the basis of the 26th Round (1971-72) of the National Sample Sur
vey, about 46.4 per cent of the rural population was poor and not merely
20 per cent as Dr. Sukhatme would estimate it.
Before concluding, I may add a final word of explanation. Dr.
Sukhatmc’s critique of our estimate of poverty is full of much nutritiontheoretic discussion and some, though rather scanty, data to support it.
It may be said that I have not taken sufficient note of this. This is true;
but, instead, I have accepted its principal finding namely the existence of
inter and intra individual variation in energy requirement. The existence
of inter-individual variation can be readily established and explained; it
simplv means that some individuals arc more efficient biological machines
than others. But the intra-indiv idual variation, namely variation in energy
requirement of an individual from day to day or week to week, is some
what intriguing and there is currently considerable speculation regarding its
causation. An explanation, apparently widely accepted, is that the human
body has an essential homeostatic or regulatory' mechanism which ad
justs to varying intakes of food without drawing on or adding to reserves,
while at the same time enabling the individual to continue activity in
volving a given quantum of energy expenditure. If this is correct, it means
that the energy requirement of an individual varies mainly in response to
variation in his energy intake. Thus, rural poor suffer from want, intermittantly and during certain parts of the year. The homeostatic mecha
nism ensures that they tide over the difficult period without imparing
their activity and efficiency. I trust that it is not implied that the homeo
static mechanism eliminates their poverty. I also suppose that the homeo
static mechanism functions within limits and that energy intake beyond
these limits over a sustained period results in changes in bodily function
and activity. I suspect that these limits of homeostatic mechanism are dif-
26
fercnt for different individuals. Such, briefly, seems to be the explanation
for the existence of intra-individual variation in energy requirement.
Relevant to our purpose, namely measurement of poverty, is the fact of
existence of inter and intra-individual variation of energy requirement. I have
accepted the phenomenon and its magnitude as Dr. Sukhatme has indi
cated. It seems to me that the existence of such variation by itself would
not materially affect our estimate of poverty in India unless supported by
improper use of statistical methods and some plain blundering.
References :
Rao, V. K. R. V. (1977). “Nutritional norms by calorie intake and
measurement of poverty’’. Bulletin of the International Statistical
Institute. Proceedings of the 41st Session. Vol. XLVII-Book 1 —
Invited Papers, pp. 645-654.
Sukhatme, P. V. (1977). “Measurement of poverty based on nutritional
needs”. Bulletin of the International Statistical Institute, Vol.
XLVII-Book 4 — Contributed Papers, pp. 553-556.
Sukhatme, P. Vy-yl978)—“Asessment of adequacy-of diets at different
income levels”. Economic and' Political Weekly. Special Number,
August 1978, pp. 1373-1384. Unless otherwise stated, all quota
tions are from this article.
-■.'i '
follows.
The
"What is a Poverty Index?
au>-
t races
tion
methodological
of
jjSth
□ ea 1 s
4?3>
developments
an axiomatic approach by Sen
s ome
with
detail
policy
and
it. is
some
recently
w i th
integrated
i a st.
Final 1y,
statistical
r ew
presents
.^section,
i
as
-■ f:«gii,are compa:
^poverty
i
estimation
index
v a r i a o i! 11 y
ji'SHAT IS A POVERTY
1<
INDEX ?
quest ion
what a poverty
?. i'.jhiess we state <
I -T '
i-poverty connotes
ly what
meant
with either an absol
it
'npmic deo
.<■
t • '■71»oduc i ng
a 5ideal
we i 1 - be i n g
a
norm,
deprivation,
or
in which case
or
a
the
at i on i mp I 1
al
individual ' s
An
hav e
stat
-dnf ron t ed
of
on comparing an indiv i dua1 ' s
‘S,jai led an abs o 1
;■
can not be answered
i 1 1 - be i n
■1. sco horn i c
< ri;?cpnom i c
i
ooo r
i ar a state
i ■"
.jjdejprivat i on
I'vhich case
consumption
i nd i v i d ua1
notion
or
is
ordering on
or
the notion of
with.
is
of
to
■. i-.-tates by an
than the norm
than
one can say whether a
norm
' =‘son
onom i c
be redef ined
The
’Srnis
an
is thu
that the norm used
-ermi n i ng deprive ‘ ion has to be soecific to each individual.
Driest who vows to I i v e
s-QPle. this norm should be
POVgr£y and
.. W h.
■'r®sst
-
poverty voluntarily.
a■
severs
c o m m u n i t y,
such that persons
to arrange
..An a group can
indiv i dua i s
ex pe
ot
SOME CONCEPTUAL AND STATISTICAL ISSUES ON
POVERTY
3Y
T Krishna Kumar*
’
important conceptual issues in the measurement of povSr^ .
remained undiscussed for too long. Uha t was needed was a great^ ■
degree of vertical integration between the statistical measure ' w
of poverty on the one hand and welfare economics on the other’ .®:-
A, 3. Atkinson < 198?}1.
I
INTRODUCTION
It was nearly ten years ago that
i
overheard a conversations
between two professional colleagues, both concurring with the vies,-®.
that
continuing
a third colleague was
’poverty",
I
took
little attention
to
poverty
topic
recently
in
measure of
which
the
their
seriously
and
I
attended
a
until
using
was
to measure regional
poverty
over-workssf
conversation
studies
author
on an
to wqrk
seminar®
ratio as
head-count
differences
paid H
in
ooverty. ■ !
could not avoid taking interest in the topic as I felt somehow that 5:
two specific aspects of poverty had not been adequately ex p 1 ored.
this
allegedly
"over-worked’
topic.
These
aspects
two
are:
(i) *•',
synthesizing the poverty measurement with applied welfare economics
and
public
policy,
and
(ii)
statistical
issues
relating
measurement of poverty index based on sample survey data on incoEe.J.
and
expenditure
distributions.
methodology pertaining
to
This
review
poverty measurement
of
econometrics’
is therefore to-’M”
regarded as a review by a late comer to the field with emphasis
the two specific issues just cited that have a great potential *°r |
future
research.
Several
important
contributions
to
the
poverb
literature such as empirical analysis of poverty indices and theM
variation are excluded in this review in order to focus on issue5'-'|
relating to methodology and in order to maintain a moderate lehgbyl
for the paper.
Quantitative Analysis Unit,
institute for Social
Economic Change
Nagarabhavi, Bangalore - 560 072,
India. ..Tfry
author thanks V. Sitaramam, Anil Gore, N. S. ly n g a r , V. h •
and Vinod Vyasuiu for discussions and comments on various issUif
X
that are raised hare. The author also thanks 3. K. Kai I
•• $
Nagaraju, and >1. S. Vasuki for their assistance in pr oces s i no
paper.
or partial ordering and persons in different groups have different
orderings.
partial
Within each homogenous group one can determine
the number and/or proportion of
persons who are deprived.
be as many such deprivation indices,
will
persons who are deprived,
There
viz.,
the proportion of
as there are groups.
The question then
arises as to how one can combine the group deprivation indices into
Such a conversion of group
a single community deprivation index.
deprivation
indices
index
deprivation
single
a
to
requires
a
partial ordering of economic well-being of different groups - viz.,
welfare function.
a social
•t
was
assumed
ao.ove,
inaivioual's
economic
state
not
deprived.
is
can
that
convenience,
represented
one
But
deprived.
for
just
categories
two
in
introduce
also
each
partial
a
ordering of different economic states of an individual relative to
Then
norm.
relative
an absolute
or
a
conditions
on
the
individual's
deprivation
can
be
represented
certain
under
an
preferences
a
by
regularity
individual's
deprivation
function
if the
demonstrated by Debreu (1959). Using an analogous reasoning,
partial
social
regularity
ordering
group
of
there
conditions
deprivations
community
a
exists
satisfy
as
certain
deprivation
f unc t i on.
A
group poverty
index
that
group.
deprivation
defined
..as
for
a
be
can
defined
The community
consistent"
"subgroup
constituent group poverty
indices.
as
the
mean
level
of
poverty “ index can be
aggregate
the
of
The meaning of
various
the term sub
group consistency and the problem of aggregation and decomposition
of poverty indices are dealt with in detail
HI EARLY
As
DEVELOPMENTS IN MEASUREMENT OF POVERTY
indicated
in
the previous section measurement
depends on a norm with respect
individual
in Section III.
is compared.
to which
of poverty
the economic state of an
The standard approach which goes at least
as far back as Rowntree (1901) is to define a poverty line in terms
a
minimum
necessities
of
P“rce tage of
level . of
ife
income
and use
the
.• e people have an
needed
to
purchase
income distribution
income
the
basic
to see what
less than such a poverty
ratio
This measure is called Head Count
line.
contributors
who
used
approach
this
Some other
(H).
Bewley
are
and
Burnett-
(1365) and Atkinson (1970b).
3utt(1915), Townsend (1954), Weisbrod
Another measure which also depends on the poverty
line and
income
distribution is the poverty-gap used by Batchelder
(1371).
it is
the aggregate income short-fall from the poverty line of all
persons
whose
are
incomes
the
below
poverty
modified and normalised version - viz.,
of all the poor,
line.
A
those
slightly
per person percentage gap
I,
called the income-gap ratio,
is in popular use.
Suppose there are n individuals with incomes yi , y: , .......yn
arranged in an increasing order and let z be the poverty line.
q be
the
maximum index
number of the poor.
that yi
i such
£
The head-count ratio H
H = q/n
Let gbe the income gap defined as g;
percentage gap of all
—
=
---- /.Jr =
c-zi^ ~
the poor,
— >,(-----)
z
=
I
q denotes
the
is given by;
..................................................... (3.1)
- z - y, ,
Then the per oerson
is given by ;
Gifq
I f one def i nes the mean income of
Then,
z,
Let
.................
z
................................
the poor as m then
=
m
Z£G
=
I
It
is clear
...................
that
the head count
•................... (3.2 a)
ratio gives only
a count of
the
poor and it gives no account of how poor the coor are relative to
the poverty i me. The income-gap ratio tel Is now poor the poor are,
as a group,
.n relation to the poverty
indices give any
poor.
it
inequality o:
is
idea of
desirable
how
the
that
line.
income
.he
But neither of these
is distributed among
poverty
index
income distribution among the poor. For
reflect
the
the
this one can
4
uSe
an
(1912).
<3
.
Lorenz
by
(3.3)
that
conclude
may
the
thing
best
poverty is to present the triplet H,
do
to
I, and G,
for
I
tel Is us by how much the mean
short
of
the
poor
falls
inequality or
poverty
line,
G
and
income or
tells
the
us
The first item measures
the income among the poor.
the extent of poverty,
measuring
where H tells us how
many persons are poor.
the
Gini
<j
1 = 1 Jal
One
and
(1905)
is given by:
index
This
introduced
index
inequality
the second measures severity of ooverty and
The weights a policy
the third measures the incidence of poverty.
maker attaches to these three components must be commensurate with
the
policy
maker's
poverty - extent,
concern
these three
for
policy
dimensions of
severity and incidence.
There was, and there has been, a considerable debate on how to
determine a poverty line.
This debate is quite significant because
this choice
the
poverty
Here
there are three different points
the
poverty
indices
line
depend
need
not
on
have
a
of
that one may note.
scientific
line.
poverty
the
basis.
First,
It
can
be
criteria.1
chosen
administratively
Second,
one can possibly consider a ranking of people according to
the
degree
of
using
deprivation
in
certain
such
a
objective
way
that
invariant to the choice of the povertv line.
made
by
Atkinson
(1987).
Third,
there
may
the
ranking
is
This suggestion was
not
be
a
unique
deterministic minimum requirement on the basis of which a poverty
line can be determined.
This
last point
is closely related
to a
criticism made by Sukhatme (1978) of a method which uses an average
requirement in choosing the poverty
line.
1
Usually there is a hierarchy of needs that are to be
Uhich items of this hierarchical need
satisfied in a given order.
structure should enter into the minimum needs that are to be met
may vary from society to society.
Similarly the desired levels of
consumption also may vary from society to society.
Hence, the
choice of poverty line may be different for different societies.
and Rath
(1971)
a nutritional
used
norm and
then
used the minimum total expenditure needed to meet that norm as the
poverty
For
line.
this
they
the
used
daily
average
energy
requirements of an individual of a given age. sex, body weight., and
physical
activity.
instead
use
an
minimum daily
a
out
pointed
(1973)
requirement
than
rather
must
one
that
average
an
He had aiso established that the energy requirements
requirement.
of
Sukhatme
a
follow
would
individual
first
order
autoregressive
Sukhatme did not take
stochastic process with a constant variance.
his argument to its logical end.
If one were to deal with a human
deprivation based on nutritional
inadequacy then what is important
is
survival
the
risk
the
indiv i dua 1
as
Such
a
risk
of
nutritional
deficiency.
nutritional
deficiency on a given day.
hWtory
nutritional
of
survival
A
a
consequence
of
not
on
depends
Instead,
it depends on a
part
-this
history
is
contained in the individual's body weight and state of health,
and
deficiency.
of
J
the rest is contained in the cumulative nutritional deficiency over
This is an important point to note as it leads to
the recent past.
certain statistical issues that will be dealt with later in Section
VI.
if
intake
both
energy
are
energy
the actual
In particular
random
the
intake and
nutritional
the required
deficiency
is
determined by the probability that actual energy intake falls below
the requirement.
The survival
risk due to nutritional
deficiency
is then a function of the probability that the actual energy intake
falls short of
IV
the requirements over a period.
■; .>
LATER DEVELOPMENTS IN MEASUREMENT OF POVERTY
Sen
(1974)
used
normative poverty
approach
(Sen
axiomatic
an
index along
lines of
the
for
developing
a
his earlier axiomatic
for arriving at normative measures of
{19731).
income
inequality
He extended this work by developing a. new ordinal
approach
to
the
measurement
mentioned
in
the
previous
derived
approach
of
(Sen
poverty
section
that
from three indices H, I and G,
a
(1976)).
poverty
index
It
was
can
be
Sen introduced three axioms
that a poverty index must satisfy and from them he deduced that the
only
poverty
inde
that
satisfies
those
three axioms
is of
the
form:
6
p
H{I+{1-I)G)
(4. 1)
three other major contributors besides Sen who made
significant contributions' to the development of
are
These
poddar
Atkinson
(1976),
and
(1970a,
1970b,
Kakwani
( 1980a, 1980b)),
Kakwani
1987),
indices
poverty
(Kakwani
and
and
Greer and Thorbeeke (1984), Foster and Shorrocks (1991), and Foster
Some
(1984 ) ) .
of
these
modified form and
below
deveIopments
in- a
some detail as thev lead us to some useful and
I
be discussed
ivation and
V and V I .
poverty
measured
po v e r
in
the income
individual.
F (y ) is
the cumulative distribution
funct ion
i ncome
o 1 1 ow i ng Sen (1976) and Kakwani (1980a)
pov e r ty
be expected to satisfy the following three
o n o r. o n ’
: Given
income of
an
below the poverty
line must decrease the
Transf er Axiom: Given other
a pure transfer
income from
a person below the poverty
line to any other person
with a higher
income must increase the poverty.
s. ■ ■.
a transfer
Sensitivity Axiom: Given other things,
t>0
of income takes place from a poor household with per capita income
poor
ome
the magnitude of
increase in poverty dec Teas es as
From
the
ear 1 ier
individual's
discussion
depr ivation
either
absolute
poverty
level:
or
"d"
relative
+ d (d>0) then
v increases.
it
is
clear
ex pres sed
as
a
in
Section
can
be
dev iat ion
II
of
his
that
an
function
of
from
the
income
d = g(z-y) .............. ................. ......................( 4, 2a )
d = { ( z -y ) / z
. ........................................(4. 2b)
7
4
n
— neral terms the deprivation can be expressed as a function oj
1
Sa|
y and z:
d = d ( y, z ).......... .. ..................................... ' 4' 2c j
T
Atkinson (1987) has shown that most of the poverty indices proposed
-1
in the literature can be represented by a poverty index defined a3' " '
foilows :
Z
............................................... (4.3)
fd(y,z}f(y)dy
=
o
From Figure 2
paper,
in Atkinson's
a
non-increasing
of
the same
those cases the deprivation function
paper it is clear that in all
is
1
from Table
and
function
of
y and
convex
the
to
origin,
or
strictly
convex
tc
can be shown
that Atkinson's
class of
poverty
indices satisfy t'r e three axioms
listed above if
the
11
origin.
the deprivation function is a decreasing function of y and strictly
(Kumar
convex
(1992)).2
either due to analytical
It is often desirable,
into different
population
for
index
policy
for
and
convenience
each
problem of
group.
purposes,
groups
This
of
gives
to
divide
and
persons
rise
to
necessity or
the
the
entire
define
poverty
basic conceptual
consistent aggregation or decomposition.
tackle
this
axiom
termed
issue Foster and Shorrocks
"subgroup
consistency"
(1991)
which
in order
introduced another
only
means
concept of monotonicity applies to groups of persons,
poverty index of al 1
same and
poverty
if
groups except one group,
the poverty
index must
of
group
i
increases
to
viz.,
group i,
then
that
the
if the
remains the
the aggregate
increase.
Suppose that we partition the people into a finite number of
mutually
......G,;.
exclusive
and
collectively
exhaustive
These may refer to different regions,
groups
G1 ,
Gi ,
different income
2 After I had written my paper I had come across a paper by
Keen in which he cites the result that convexity of the deprivation ■
function corresponds to the transfer principle (see Keen (1992)
p.70).
8 '
K
different
groups,
groups
ethnic
Assume
etc..
that
w,
is
the
proportion of househo1 ds■who belong to group i and f, (y) and P4 (y)
refer
the
to
conditional
function
distribution
with
associated
group
of’
density
and
income
distribution,
Let
i.
function associated with group
d-. (y,z, )
i,
conditional
denote
cumulative
respectively.
the
deprivation
we can then define a. sub-group
as follows:
poverty index p,
■
spy
■
JI
..................... .................. (4.4)
i d,(y, zj f£(y) dy
c
3
p-
i Bp? ■ ■
I
■
The interesting question to pose is: how can one combine the
into an .aggregate poverty
index so that
aggregate poverty index satisfies the three axioms
listed earlier
sub-group poverty indices
and also the sub-group consistency axiom of
11991).
roster and Shorrocks
Let the aggregate poverty index P be given by:
p
(4.5)
[d(y, z) f (y) dy
=
"o
where z and d are yet unknown and fly)
what
is given by
k
E^fjy)
i=i
Jf(y)
=
If
one
notes
is
needed
is
that
a
d(y,z)
condition
(4.6)
is
consumption deprivation
for
consumption.over individuals that will
consistent
aggregation
then
of
give rise to the following
consistency relation:3
3 Here it is being assumed that d1y,zt ) = 0 for y > z, .
For
conditions on consistent aggregation of consumption expenditure the
reader may refer to Muelbauer (1975).
,q
i
22
d(y,z)
U n *9 r .9 z -
If
(4.7)
{y, zj
3. x ■. z i ?
the above condi
(4.6)
is satisfied then employing
and 4,7;
one can write:
= T3aX
3
d (y, z) f (y) dy =
f
P -
J2 wjdi (y, zj fi(y)dy
i“l
0
0
(4.9)
7 , ot D
index
poverty
A
....(4.8)
that
satisfies
called
is
(4.9)
additively)
(an
decomposable poverty index.
Two points that must be noted here are
that
deprivation of
the aesn
is
d-_ (y,z. )
group may have its own poverty line.
i and
group
that each
These two points are related
to the issues raised by Sukhatme and these are quite important for
some statistical
(1937),
V
synthesis
a
is
above
issues
raised
in Section VI.
ideas
and
of
roster and Shorrocks
The analysis
in Atkinson
results contained
and Kumar
(1991),
given
(1992).
POVERTY MEASUREMENT AND APPLIED WELFARE ECONOMICS
its
By
very
nature measurement
of
was
poverty
motivated by
is through such measurement that one
welfare economics because it
can identify the need for welfare measures and also aevise suitable
for
methods
needs
However,
different
poverty
measurement,
In
economics.
methodological
depended
function.
Recent
Thorbecke
(1989).
hence
direct
on
based
literature
certain
and
on
The
of
Keen
were
(1992)).
followed
applied
in
welfare
poverty
measurement
a
deprivation
and
targeting
(Kanbur
(1987),
that used such poverty
indices
poverty
through
needs
the
and
other
distribution
and Keen (1992))
on
approaches
the approaches
income
transfers.
(1987),
targeting,
focused on changing
income
function
needs
particular
invariably
were
(Kanbur
targeting
incidence
arbitrary
axiomatic
of
choice
of
considerations
deprivation
has
placed
pOverty-hased
compa-tment
different
altogether
theory
based
welfare
In the latter type of analysis one would normally take
economics.
social welfare as a function of
3nd this
an
microeconomic
other
from
into
economics
welfare
ieve.
of
individual consumption,
levels of
consumption as a. function of
individual
income
and prices confronting the individual.
A synthesis between these two seemingly different approaches
is possible by noting that consumption deprivation can be taken as
the deprivation function in the poverty index.
Bhanoji Rao (1981)
had already drawn the attention to the relation between consumption
deprivation
poverty.
and
Quite
independently
the
without
and
knowledge of Bhanoji Rao's paper, Kumar, Sitaramam, and Gore 11992)
had developed
index and consumption by
link between poverty
this
suggesting that the deprivation function used in the poverty index
can be the consumption deprivation,
this term being defined
shortfall of mean consumption from a bliss point.
to
the
that
commodities
consumption
deprivation
difference
between
at
the
which
curves,
expenditure
of
income
bliss
level
of
consumption
strictly
defined
income,
concave,
it
is
the
Thus,
the
are
relation
between
functions
the
that
the
and
the
Since for necessities the
increasing
follows
this
be
would
level
depict
and
then
any
consumption expenditure on necessities.
Engel
if one restricts
necessities
the
constitute
as the
consumption
of
deprivation
income
and
function
so
a decreasing function of
income and strictly, convex.4
poverty
consumption
index
based,
on
deprivation
of
necessities satisfies the three axioms suggested by Sen (1976), and
Kakwani
(1980a).
aggregation of
get
group
By assuming
that
consumption hold,
decomposable
the conditions
by
aggregate
this procedure,
poverty
index
roster-Shorrocks 1s axiom of. subgroup consistency.
of
curves, are
Engel
writes
the
aggregate
functions
consumption
consumer demand
of
for consistent
own and
deprivation
cross
functions
functions for
one can also
that
satisfies
The parameters
prices.
in
full
If
one
using
necessities they will
be
4 In
he consumption basket of persons who are near the
officially defined poverty line of India such essential items or
necessities constituted in 1983-84 more than 75 per cent of the
total consumption expenditure.
See Technical note on the Seventh
r i v e Year Plan, p. 10.
or ices
of
f unci ons
income.
and
Gne
can
devise
then
policies
through which deprivation can be reduced employing price and income
policies
Such
policies.
both
affect
the
level
deprivation and the resulting income distribution,
the major components of the poverty
consumption
of
these two being
index.
Such a synthetic approach to applied welfare economics can be
illustrated by considering the following enlarged definition of the
poverty index
p =
52 v< [dAy, zitPi)
re p.;
p,-) tfy
............................... (5.1)
5. vector of prices ancountereo by persons in group i,
Let
f« 3 iy?Pio ) denote
the distribution of
in group
income
i
prior to introduction of public policy measu res, some of which are
directly meant to be poverty alleviation programs while others are
policies
that
p-.
f ■. i
do
an
have
impact
on
J denote the distribution of
coverty
the
Let
level.
income of group i after the
introduction of public policy measures.
The policy measures are in
general
aimed
policies,
incomes,
pure income transfer policies,
at
asset
redistribution
Rationing and public distribution etc.
price
vectors
possible
without
and
with
new
that there are direct and
and price policies such as
Pi □
and p. ,
policy
represent
interventions.
i ndirect effects of
poverty
levels
can
be
alleviation
schemes
but
they can
be adversely affected
one
consider
The
policies.
public
Hence
favourably
must
factor
changing
affected
the
by
It
the
is
programs.
poverty
the
by
other
welfare
over-all
impacts of ail public policies put together. Such an evaluation is
eminently carried out by describing the functioning' of the economy
as a temporary equilibrium model with price controls and rationing.
One
can
types
factor
consider
of
households
inputs
Similarly,
sector
to
extended
an
as
the
input-output
different
other
endogenous
sectors
and
model
with
sectors
receive
different
that
factor
supply
incomes.
in this set-up one must include government as a separate
that
provides
government
services
to
various
sectors..
including different household sectors
taxes and services of
and
it receives as
inputs
factors.3
The welfare economic prob 1 eras can then be posed as f o i lows
■
rti
si
wiaj di (y, Zi, Pia) fia (y, pia) dy
Q
k
2
(5.2)
sub j e
J
(y, zv pi0) fi(7 (y, pia) dy
(5.3)
0
(5.4)
(5.5a)
(5.5b)
yio
Pia
(5.6)
Pi0 + D^d-^ *P<n
v a r o us
sectors
the
B
A
supply;
are
n
x
n
and capital coefficients, e
DA is
a
x
n diagonal
matrix
of
net imports
that
of
the
speed of market adjustments between demand and supply. the diagonal
e1 emen ts
of
controlled,
DA
can
be
zero
r, „ denotes the
for
those
endowment of
whose
of the
no doubt a very difficult task to allocate between
ors the public goods that the government produces. One
approach is to treat public goods as a factor of production that
into all sectors at the same levels.
i3
the
person while
r. ,
additions
to
factor
services
while
is
final
endowments
is
through
additions
are
due
other
Thus
in equation
(5.5a)
t, 4
is a
transfer
an asset
equation
(5.6)
.
policies
while
of
to
asset
is a pure
t?4
Some of
markets
the
factor
for
distribution
transfer
income
income equivalent),
(its
in
in price which is
that component of change
is
p, <
factors.
endowment of
set through government's price control.' In the above optimization
problem the government policies regarding choice of elements of the
and its activity
price vector p, .
(certain components of x
levels
vector denote the activities of the government) are instruments of
e is exogenous and all other variables are to be determined
policy,
endogenous 1y.
Thorbecke
presented
(1989)
this kind
of model
to assess the
impact of programs targeted for poverty alleviation.
His approach
also included both the direct and
employing the
Accounting
Social
the
impacts and
including
cut
(SAM).
Matrix
es t ima ting
indirect
of
income-squivalence
the post-policy
impacts
his
approach
involved
and
indirect
direct
such
incomes
in a Foster-Greer-
Thorbecke index of poverty
(Foster,
Greer,
and Thorbecke
What
a
direct
approach
is
policy
suggested
impacts,
computable
both
general
are
levels
here
used
is
more
and
direct
indirect,
equilibrium
model.
to
consumption
derive
The
are
in
the
which
through a
traced
resulting
(1984)).
consumption
and
deprivation
the
consumption deprivations are used to measure poverty.
VI
STATISTICAL ISSUES ON MEASUREMENT OF POVERTY
There
issues arise
except
measurement
distribution
estimation
of
rare
which
statistical
First,
since
the poverty
income some
similar
occasions,
quite often
problems confronting
a
Second,
by
sample
surveys.
population
poverty
index
provided
whet’ er
of
census
employing
problems
statistical
the
to
made
following the critical
deprivation,
in
probability distributions.
in
is
are
ways
poverty.
the density of
measurement
associated with
have,
distinct
in measurement of
index depends on
poverty
three
mainly
are
as we do not
incomes
information
income
on
This
gives
a
sample.
using
poverty
rise
to
Third,
contribution of Sukhatme one can interpret
it
is
a
nutritional
deprivation
or
a
as not being deterministic but as being
consumption deprivation,
Then even if we have a census of
stochastic.
incomes the poverty
index must oe based on a stochastic deprivation.
the
of
Some
statistical
measurement of inequality.
income
and
inequality
methodological
with
the
Since there is a close relation between
poverty
relating
issues
arose
first
issues
is
it
to
to
cover
the
inequality
also.
One
useful
income
early application of properties of density functions was by Levine
and Singer
the
income
Using
an
inequa!ity
exponential
conclusion
tax
does
They consider the closed-form expression for
(1370).
the
income
distribution
for
incomes
terms
that under such an
not
change
income
depends
on
function.
reach
they
the
income distribution a proportional
inequality
but
if
lump sum
a
tax
is
tax then the effect of that on income
imposed after a proportional
inequality
density
of
in
the
tax
proportional
also.
The
usual
definition of Lorenz curve is in terms of two equations both having
a common
Stewart
parameter
(1369).
(decile or
fractile)
(1971)
Gastwirth
as
presents
in Kendall
given
the equation of
and
Lorenz
curve as a single function of the percentile employing the inverse
function of the cumulative distribution function.
provides
an elegant
analytic
scheme
for
Such an approach
studying
the effects
income transfers, through taxation, on income inequality.
P
C
revolution
integration
this
and
the
availability
technique
has
great
of
for
software
With the
numerical
applications
potential
of
for
studies on impacts of taxation on income inequality and welfare.
Another
study
that
has
a
significant
potential
for
future
research on measurement of poverty is by Singh and Maddala (1976).
They noted that the two distributions that are quite often used in
income distribution studies,
viz,
the Pareto and
tne
were not quite suited for the graduation of incomes.
log normal,
These authors
capitalized on the well-known result that the larger the income the
more is the probability of having more income.
This phenomenon is
just the opposite of the phenomenon of probability of survival with
aging.
The authors argued that upto a point,
of income,
i.e., at lower levels
the failure rate or hazard rate defined by f(y)/{l-F(y>}
is increasing and. then it is decreasing.
And they also noted that
this property is shared by the log normal
distribution of incomes.
J Q r r i • -- •
■ fc
This
or
approach
Singh
and
to
Maddala
probability
relate
'
distributions in economics to those in reliability theory of Barlow
(19651 has a great potential
and Froschan
poverty measurement.
>•
for further research on
This is because the deprivation function used
I
rate or hazard
®
indices must be related to failure
in the poverty
rate that measures the risk of survival.
return to this
We shall
tooic when we discuss the stochastic nature of deprivation.
I
There were quite a few studies that addressed to the question’
of dominance of
income distribution over another
one
the dominating distribution
to an unambiguously mors
rise
giving
in terms of
income
distribution
than
the
Dasgupta, Sen.
and Starrett
(1973),
Rotschild and Stiglitz (1973),
Stromberg
(1988)),
a
issue
equal
and
Kanbur
and
(1991)
examined
Lorenz
curve
similar
Bhatty
of
entirely
being
other
and
(1974)
Moothathu
another
one
for
conditions
deriving
above
(1970a).
(Atkinson
Lorenz
curve.
Monotonicity axiom plays a central role in measurement of poverty.
axiom
This
imposes
conditions
certain
on
cumulative
the
distribution function as shown by Atkinson (1987) and Kumar (1992).
Spencer and Fisher (1992)
point out that this condition is same as
the condition of stochastic ordering or majorization.
is
a need
to
in detail
further explore
oncepts
of
stochastic
arshalI
and
Olkin
dominance
and
the applicability of
the
developed
by
majorization
problems
(1979)
to
utmost
significance
the
there
Hence,
of
measuring
and
comparing poverty.
problem
The
poverty
ano
of
alleviation programs
time
as
a
is
multivariate
to
treat
stochastic
devising
sustainable
income over
individuals
in
process
and
to
derive
conditions under which one stochastic process dominates another in
terms
of
Stromberg
reducing
(1988)
ths
statist icaI
Marsha I 1
theory
and Olkin
this topic whicl
poverty
examined
which a conditional
I
this
index
over
a
problem for
period.
a. particular case
transitional density was used.
of
stochastic
(1979)
dominance
Kanbur
and
and
in
Employing the
majorization
of
some more research can be undertaken on
has great policy relevance.
Most of th? applied and empirical work on poverty measurement
that one comes across is of the "economic statistics" variety such
3S
construction
and
comparison
economic
of
realising or
ignoring
Several
time and space and made
authors had computed poverty indices over
comparisons act
indices.
the fact
indices
that such
were constructed on the basis of sample income distributions.
They
are therefore only estimates of the underlying population poverty
indices.
(1360,1964)
Iyengar
Lorenz
ratio
under
the
derived
a
the
consistent
estimator
of
that
underlying
income
assumption
the
distribution was log normal and that this sample distribution was
available
and
Similarly Haiti and Pal
in grouped data form.
Poddar
and
Kakwani
developed
(1976)
efficient
(1983),
methods
of
estimating the population Gini coefficient using grouped frequency
it is very rare to find instances where a
However,
distributions.
comparison of estimates of Gini coefficient over time or space is
done
against
the
standard
can
be
made
statement
poverty
indices
over
errors
regarding
space and
of
Similar
estimates.
such
the
comparison
of
time.
There
a
is
estimates
need
to
of
judge
whether the observed differences in estimates of poverty over time
and space or groups of persons are significant in relation to their
standard errors which reflect the inherent variation attributable
to mere random variation.
One
important statistical
problem that
needs immediate research attention is the problem of estimation of
population
poverty
index
employing
grouped
frequency
sample
distribution of incomes with a deterministic deprivation function.
In this case the population poverty
index can be written as:
z
j d(y, z) f(y) dy
o
P -
..(6.1)
based on grouped sample frequency distribution,
Its estimate,
can
be written as
z
[d(y, z) ^(y) dy
(6.2)
o
where f„ (y)
is an estimated sample density
density function can be estimated either
function.
The sample
parametrically,
assuming
a specific functional form for the density with unknown parameters
non-parametrical1y
or
Kernel
one
If
estimators.
any
employing
one
interprets
mean of deprivation given an income of
of
the
d(y,z )
as
non-parametric
conditional
the
y then the poverty index p
of (6.1) has the interpretation of unconditional mean deprivation.
One can then treat (6.1) as being quite similar to a non-parametric
employ
and
regression
suggested
methods
of estimators P of
(1364)
ancj
one needs to know the properties
More importantly,
Uatson (1364).
Nadaraya
by
(6.2) unaer various types of density estimators
fn (y ).
if
a
is
approach
parametric
functional form for f(y)
assuming
followed
then one can derive an estimator
in (6,1)
of
P based on appropriate sufficient statistics of
of
f{y)
estimator.
If
oneself
assuring
thereby
methods
analytic
the parameters
consistent
of
and
of
the
efficient
drawing
for
intractable
become
on the sampling distribution
inferences
specific
a
poverty estimator
given by (6.2) one can employ bootstraps technique of Efron (1382)
to generate replicated samples and generate sampling distributions
of
the poverty estimators.
If non-parametric approach is adopted
for estimating f„ (y) then the problems of inference under that set
up are more challenging.
One may
refer to Prakasa Rao
in situations of non-parametric functional
inference
(1983)
for
estimation.
There are some attempts in the literature for ’correcting’ or
the estimate
improving
from
grouped
sample
coefficient which was
Gini
of
calculated
Gastwirth and Glauberman
frequency.
(1976)
proposed an interpolation formula employing Hermite interpolation.
This
rather
than
(1931)
traditional
procedure
and
trapezoidal
the
of
trapezoidal
method
curve-fitting
estimating
compared
concluded
of
nature
efficiently
Suryanarayana
the
is
approach
and
Gini
the
apply
method and by the Kakwani
and Poddar
estimates
by
the
that
the
that
the
underestimate
Gastwirth's
method
to
estimates of the Gini coefficient.
assume a
parametric
form
to
coefficient.
coefficients estimated by
the Gini
the
Gini
provided
index
and
estimate was sensitive to the method of estimation.
to
smoothing
the
the
NSS
But
income
data
It is possible
to
improve
the
it is more interesting to
density
function
and
to
an
obtain
estimator
efficient
sufficient statistics of
Gini
the
of
the parameters of
Suryanarayana and Geetha
coefficient
on
Greer,
and
the density.
estimate Foster,
(1992)
based
Thorbecke's poverty index using grouped sample income distribution
and
specifying
a
while
authors,
do
index
normal
log
is
this
realizing
poverty
population
parameter
two
not
a
estimator and its sampling distribution.
express
the Foster,
Thorbecke
Greer,
for
a
interesting
of
series
the
of
the
properties
of
this
Suryanarayana and Geetha
index
a
as
the
function of
This approach opens up the
moments of the log normal distribution.
doors
estimate
sample
examine
The
distribution.
statistical
problems.
of
Some
these may be outl’ined here.
First, any deprivation function d(y,z) can be approximated by
a sufficiently high degree polynomial
general
any
the
for a given z,
Hence
index can thus be expressed as a. function of
poverty
the moments of
in y,
even if a given
Second,
income distribution.
deprivation function can not be easily approximated by polynomial
of a
low degree,
ask
if
since the poverty
there is any monotonic
index
one can
is only ordinal,
increasing
function of
can be approximated by a low degree polynomial.
that
d 4 y, z )
since the
Third,
statistical
literature provides
the moments,
one can take such estimates of the moments and derive
efficient
estimators
take
maximum
the
poverty
index
can
the
of
estimators
for
a function
the
Such
an
of
the
maximum
estimator
of
likelihood
instance one
can
the
moments.
Then
of
such consistent
For
be estimated, as
estimators.
function
index.
poverty
likelihood
for estimating
efficient methods
poverty
index
estimators
itself
consistent and efficient under certain very general
Many
practical
problems
poverty
of
which
a
is
will
be
conditions.
alleviation
can
be
converted into policies aimed at modifying the income distribution
Since the poverty index can be expressed
so as to reduce poverty.
in
terms
of
the
of
moments
the
income
distribution
the
policy
problem reduces to that of choosing public policies that affect the
foments
of
the
income distributions
in
suitable ways.
generalize
the concept
of
regression which
°onditional
expectation
or
conditional
mean.
is
an
One can
expression
Similarly
for
one can
19
representing conditional
and
conditional
rd
moment
Thes e
etc.
moments
variance
of
income
distributions can be expressed as functions of
as pub 1ic
forces.
policies aimed at poverty alleviation and
other market
He thus have a hierarchy of
Ely
’
X) = h (X ; S)
k
(8.2a)
(X ; y)
(6.2b)
these hierarchy
regressions one
etc.
r r om
the coe
ients of
can dev ise suitable policies such as altering the values of certain
components or
X which are under
contro L.
ossible that some policies increase not only the mean of the
ncome but also
while some policies change only the mean
not change
This type of knowledge will
help
po I i
poverty
def i ned
section
IV
deterministic
assumed
a
is paid
operational procedures for deriving the
deprivation f unction.
It
was
cons umpt i on
depr i vat i on can be taken as the deprivation function.
The question
then arises as to whether consumption deprivation function can be
regarded as dete:
be
regarded
as
Consumption deprivation in fact shouldstochastic
since
the
expenditure for a given amount of
income
known.
of
The
usual
interpretation
the
amount
of
consumption
is not deterministically
consumption
deprivation
• It may be noted that the same way yt is a sample information
on the mean (yi + yj - (y, +y4 )/2)2 is the sample information on the
variance.
Hence the above expression can be calculated for every
pair of observations and those calculated values can be used as
values assumed by a dependent variable for the regression of the
variance. The mean values of the corresponding pairs of values f°r
x can be taken as the values of the independent variables.
suggestion of this author Suryanarayana (1986?
Ratio
sally the partial derivatives of the Lorer
normal
with respect to the location and scale parameters of a log
income distribution (Suryanarayana (1986) pp. 234-237). The type of
Pr
hierarchical regressions given above together with such partiaj
derivatives provide the necessary information to devise policies'*'0
reduce the income inequality.
At
function is that it is the mean value of consumption deprivation at
The real problem with poverty is not the
a given level of income.
mean level of consumption deprivation, but it is the variability of
the consumption deprivation.
The
persons are more
income
lower
susceptible for deprivation as the spread of actual consumption is
so wide due to high variance that it can go below the consumption
more frequently.
requirements
Even
the variability is
if
same at
different income levels the probability that a person's consumption
falls below
the minimum required consumption is more for a lower
income person than for a higher income person.
This is because the
mean deprivation is a decreasing function of incomes at all
It is this variability in the consumption deprivation,
of income.
and
levels
too
that
the
possibility
differential
of
at
variability
that causes a major problem for the poor.
different income levels,
The existing literature on poverty indices does not take into
account
this
of
aspect
variance
deprivation
in
measure of poverty can be derined as an
useful
level.
A
more
(E,V) measure that
is frequently used in portfolio choice:
P
=
z
z
fE[d(y, z) ] f (y) dy+c^V[d(y/ z) ] f (y) dy............. (6.3)
o
o
(wi th 0 > 0)
.The problem raised by Sukhatme
Sukhatme
treated z
to be
is a special
case of this problem.
random (variability of
intra-individuaI
energy requirements).
In general both actual consumption and the
requirements may
variability
show
giving
rise
to
variability
in
d(y,z).
•••
In fact the above suggestion to replace Eid(y,z)l by E{d(y,z)l
•:y-
* o V({d(y,z)}
generalise
is a standard approach of measuring risk.
this
procedure and
risk in place of d (y3z).
approach of
z)
'Barlow
. ;
introduce alternative
reliability theory.
Boschan
From
a
of
and
Based on the distribution of
rate or
failure
rate
function
use
that
function
perspective
the
sustainability
(1965))
policy
measures
Another promising approach seems to be
one can determine a hazard
and
One can
instead
of
of
a
21
p’cxv srty 3.1 Isvistion pr0gr3.cn dspends on reducing tkie vuln&r^bi .> i t y
* of
people
poor
the
sustainable poverty
towards
consumption
a policy
alleviation as
Hence
deprivation.
devising a measure of poverty which captures
objective
calls
- :f
fOP
such vulnerability.
(6.3) provides such an alternative measure of poverty.
Ecuation
The
of
problems
statistical
inference
cited
above
with
a
deterministic or known deprivation function need to be modified to
The population
function.
the stochastic deprivation
incorporate
poverty index may now be defined as :
z
P ~ t E{d(y, z)} f(y) dy
o
this case the sample estimate may oe written as:
z
$ * ^d(y,z) £n(y) dy
.........(6.5)
o
One needs to examine the sampling properties of this estimator.
the
absence
of
distribution of
the
poverty
employing
any
analytic
can
index
given
for
numerical
sample
integration,
derive
sampling
the
estimates
for
which
d(y,z)
and
appropriate
f-(y)
computer
The bootstrap technique developed by Efron
be employed
to
poverty estimator given by
suitable modification of
i moor tant
classification.
to
this estimator one can compute this estimator of
software is available.
^F9S2'>
approach
In
generate sampling
(6.5).
(6.5)
if
distribution of
the
.Similarly one should
consider
the population poverty
index is
statistical
prob1em
is
that
of ■
If one knows that different groups of persons must
have different poverty
lines,
the determination of which
is quite
difficult, and uses for convenience a uniform poverty line for all,
what will be the extent of the errors in classification of the poor $
in different groups?
22 '
:■
<■
World Bank Reprint Series: Number Seventy-seven
Conn H
T. N. Srinivasan
Poverty:
Some Measurement Problems
Reprinted from conference proceedings of the 41st Session
of the International Statistical Institute, December 1977, New Delhi, India
■fit
... jL..—
.si. —
-
.n
9R4S
POVERTY : SOME MEASUREMENT PROBLEMS*
T. N. SRINIVASAN
Development Research Center, World Bank, U.S.A.
and
Indian Statistical Institute, India
1.
Introduction
The recognition that an important, if not the overriding, objective of
economic development in less developed countries is to provide at lea,st a
minimum level of living for their entire population, is not new. As far hack
as 1902, the then chief of the Perspective Planning Division of the Planning
Commission in India, the late Pitambar Pant, produced a paper exploring the
implications for planning for providing a minimum level of living to the Indian
population by the end of Fifteen year time horizon (Pant, 1974). An opera
tional definition of poverty was used in this paper : a household was defined
as poor if it cannot afford to buy (at the going market prices) a normatively
set bundle of basic goods, particularly food1. At 1960-61 prices, this bundle
cost Rs. 20 per capita per month. The pioneering approach of this paper was
not reflected in the Indian Plans until a Minimum Needs programme was
proposed in the Approach p iper of the Indian Fifth Five-Year Plan (1972).
The purpose of the present paper is not to go into the reasons for the failure
of Indian Planning to pursue their earlier perceptions or to those that led
some of the International organizations and aid donors to embrace what
has como to be known as the Basic Human Needs approach to development,
as if it is a fundamentally new discovery. Nor will some of the policy issues
be discussed except in so far as they raise some problems of measurement
(Srinivasan, 1977). The aim of this paper is to raise some conceptual and
measurement problems relating to poverty with an aim towards formulation
of policies for poverty alleviation2.
Before entering into any discussion of measurement of poverty, it is
important to draw attention to a problem which has both political as woll as
* The views expressed are personal and do not necessarily reflect those of the World Bank
and tho Indian Statistical Institute.
1 This bundle did not include education and health care and services that wore sought to bo
provided through tho public budget.
2 Only tho so-called “size distribution” of income or consumption among persons or house
holds will bo considered in this paper. Functional (i.o. as between wages and asset income;,
regional and other such distributions of total income will not bo discussed.
9R4S-1
2
T. X. SRINIVASAN
philosophical dimensions. It can be argued that one should be concerned
not with the observed inequalities in the income or wealth distributons but
whether the institutional structure provides equality of opportunity in. a
broad sense. Tn other words, what one should be concerned with is equality
of “access”, be it to educational facilities, medical facilities, job opportunities
and not necessarily with equality of “success”, i.e., what different individuals
make of the opportunities as long as the system is “fair”. This is not to
say that those who fail in a “fair” system should not be aided through transfers
from those who succeed. However, this would not call for institutional
change. If this view is accepted, one would want to measure the degree to
which equality of “access” and fairness of the operation of the system is
observed rather than the extent of inequality in the end results. Though
this problem arises in the context of almost all issues to be discussed in the
following, no further reference to it will be made for the sake of brevity.
At the outset, it is worthwhile to distinguish between absolute and relative
measures of poverty. The most widely used absolute measure of the extent
of poverty (at a point of time in some nation, state or region) is the proportion
of the relevant population whose level of living falls below some poverty line.
Often the level of living is identified with the per capita consumption expendi
ture (monthly or annual) of the household which a person is a member and the
poverty line is the cost at relevant prices of a normatively defined basket of
consumption. Perhaps as frequently, the level of living and poverty lines
are defined in terms of per capita income of a household rather than its con
sumption. Less frequently, the level of living of a household is identified
with its per capita calorie intake (monthly or annual) and the poverty line
is the calorie requirement based on nutritional considerations.
The familiar measures, namely, the Gini coefficient of the distribution
of persons (or households) according to income or consumption expenditure
(per capita or total) is a relative measure in the sense it reflects the relative
■ positions of different persons or households (Pyatt, 1977). So are other
' measures based on such a distribution, such as the coefficient of variation,
variance of logarithm of income or consumption, the share in total income
or consumption accruing to the bottom and top deciles, quintile or other
percentile of the population. The relative measures are better described as
summary measures of the inequality rather than as poverty measures. There
is by now a growing literature on the economics of inequality measures, i.n.,
whether a jiarticular inequality measure can be used as a social welfare indicator
in the sense that the ranking of alternative income distributions according
to some well-defined social welfare function is the same as that produced by
POVERTY : SOME MEASUREMENT PROBLEMS
3
tho inequality measure, as well as the statistical estimation problems (Sen,
1973 and Kakwani, 1977), While this literature is impressive and useful, tho
link between the positive aspects such as measurement (more appropriately,
description) and the normative aspects of social and economic policy towards
poverty abatement, has not received as much attention as it deserves. This
is not to suggest that effects of some policies, such as, for instance, of the
proposal for a relative income tax at the lower ends of the income profile
m addition to the usual progressively aspects of it are not discussed exten
sively. But it is fair to say that this discussion is not usually in the context
cf pove' ty abatement and economic development of a less developed country.
The paper is organized as follows : Tn section 2, the problem of inter
preting the data (cross-sections as well as time series) on income distribution
is reviewed. In Section 3, the problems associated with defining an absolute
norm called poverty line as well as estimating the proportion of a population
having a standard of living lower than the poverty line are discussed. In
Section 4, problems associated with estimating the impact of specific policy
measures, such as investment in education, health care, etc., are discussed.
Section 5, offers some concluding remarks.
2.
Income distribution trends and patterns
Kuznets (1963) in his historical study of the development of some of
the rich countries of Western Europe and North America put forward the
hypothesis that as development proceeds, the inequality in the distribution of
income first increases before it decreases. Ahluwalia (1976) tested the Kuznets
hypothesis by relating the share in total income of the top 20%, middle 40%,
lowest 60% (as well as 40% and 20%) of the population to the logrithm of
per capita gnp in constant 1970 US dollars of 60 (developed and developing
countries in the form of a quadratic regression. The relationship was
estimated separately for the entire sample of 60 countries and a sub
sample of 40 developing countries. His results show that as per capita
gnp rises, the share of income accruing to the poor—say the bottom 40%—
first falls, reaches a minimum and then rises. The estimated per capita gnp
at the “turning point”, at which the share begins to rise again, is S 468 if the
entire sample is used and 8 371 if only the developing countries are considered.
Tho income share of the bottom 40% falls from an average value of 17% at
a per capita gnp level of § 100 to 11% at the turning point, and rises to 15%
at a level of $ 2.000. If this cross-section result is interpreted as if the increases
in income were taking place over time, it confirms the Kuznets hypothesis,
4
T. N. SRINIVASAN
There arc a number of well-known difficulties with interpreting and
projecting the cross-sectional relationship over time. One problem is, of
course, the implicit assumption that the parameters of a cross-sectional rela
tionship used for temporal projections are stable over time. Tn some situa
tions whore the data base is richer in that a time series of cross sections arc
available, various statistical techniques including that of variance components
models, the random coefficients regression models, Kalman Filter models,
etc., can be used to recover efficiently the information contained in the data
(Maddala, 1971; Swamy, 1970 and Cooper, 1973). The well-known and oftrepeated issues relating to whether gnp is an appropriate indicator of tho
level of development or the less stressed, but perhaps more important problem,
namely, that gnp estimates for different countries (or for different time
points for the same country) are subject to differing conceptual and measure
ment errors need not detain us. Given the policy emphasis of the present
paper, two issues are particularly relevant. If the cross-sectional relation
ship is accepted as a valid tool of projection over time, is there an implicit
denial in such acceptance of any role of economic policy in changing income
distribution ’ A literal interpretation of the statistical assumption under
lying the model, namely, that the residuals (i.e. the departure of an observa
tion for a country from the expected value over countries with the same
gnp) are random and have the same probability distribution as gnp varies,
would suggest this to be the case. If. however, one assumes that the residual
represents both random factors as well as any policy differences between
countries with the same gnp, the statistical assumptions on the distribution
of residuals have to be strengthened to state that residuals include “country
effects” that ,-re uncorrelated with other explanatory variables such as gnp
of the country. In the former case, one would be interpreting the cross
section relationship and the projections from it as representing some sort of
“iron law” of development. In the latter case, at best one obtains projections
for each country that corresponds to an unchanging country effect or policy
mix with no clue rs to how policy changes may affect the projectionsThere is another important problem in using relationships and descrip
tions related to a point in time. It is not always clear whether and to what
extent, the relationships represent a long run structural aspect of an economy.
This can be illustrated by the following. Suppose we are comparing the
income distribution among persons in two different economics at a point in
time and let us assume for the sake of argument that they happen to be
the same and the common distribution is characterized by extreme inequality.
Suppose, further, that in one of the economies the observed distribution
POVERTY : SOME MEASUREMENT PROBLEMS
5
represents only the relationship between income and age of person, in the
sense, that every individual in this economy will have the same income profile
over his life time with low incomes when young and old, and higher incomes
during the middle years. For contrast, let us assume that in the second
economy, there is no mobility of individuals across income profiles in the sense
that a person born poor .(rich) stays poor (rich) and dies poor (rich). It is
obvious that the observed income inequality in the first economy, is irrelevant
from a policy angle, since the life-time income profile of all persons are identical
whereas the same level of inequality in the second economy is far more serious
from a policy point of view. There have been some attempts to apply a
Markovian model with a transition probability matrix describing the
probabilites of an individual belonging to one income group in one period
staying in the same or moving to another income group in the next period.
On a version of this approach, see Champernowne (1973).
The above illustration, while extreme, in itself, raises a number of measure
ment issues, [t has been suggested that the “life cycle” aspect of the income
profile over time of an individual is better reflected, if instead of comparing
the distribution of actual income, one compared the distribution of “perma
nent” incomes. The trouble is that there is no completely satisfactory proxy
for the unobserved and unobservable permanent income studying the dis
tribution of material and financial assets is not adequate for this purpose.
Those who believe the permanent income hypothesis of Milton Friedman
(1957) to be an adequate description of behavior, would suggest that the dis
tribution of consumption is the relevant distribution to study when the aim
is to examine the distribution of levels of living. The latter is not affected
much by transitory phenomena, in the sense that the variance of the transi
tory components of consumption are considerably less than that of income.
There is also the additional advantage that consumption is less subject to
conceptual anti measurement error than income. The fact that in most
household surveys in which data on both income and consumption expenditure
were collected show negative saving by households with low per capita income is
consistent with the permanent income hypothesis. At any rate, it is fairly
clear that the inequality of observed incomes overstates the inequality in the
levels of living appropriately defined.
It may be worthwhile to point out here that data on consumption arc
often collected through household surveys in which the household’s consump
tion in the previous week or month (called the reference period) is recorded.
The design of the survey usually is such that the enquiry is suitably spread
tlu’oughout the year and the sample size is set appropriately so that the
6
T. N. SRINIVASAN
annual average per capita consumption is estimated with very little seasoned bias
and a sampling error not exceeding a desired value. But in so far as any
inequality measure is an increasing function of the variance of the distribution
as obtained from the survey, an element of upward bias arises duo to
seasonality. However, surveys which draw their samples of households from
a listing of houses .may miss some of the really urban poor who live in slums
not included in any house list. Equally, the non-response from the extremely
rich is likely to be higher. As such, the survey based measure may also have
a downward bias. But, if these biases are constant over-time, estimates
of trends in inequality arc not affected. Further, if as is often done, the
average per capita consumption of selected deciles or quintiles of the popula
tion arc also estimated, the problem of inadequate sample size as well as biases
also arise.
In the discussion so far, it was tacitly assumed that somehow or other
a distribution of individuals in the relevant population according to their
income or consumption has been obtained. Of course, this distribution is a
derived construct : the primary data areavailable in the form of distribution of
(i) individuals according to their earnings, (ii) income tax assessees according to
income as defined for taxation, (iii) households according to their income or
consumption and soon. In deriving the distribution of individuals, two alternative
approaches are available : attribute to each individual who is a member of
the. household (or a dependent of an earner or asssssee), the total consumption
(or earnings or income) respectively of the household (earner and assesses)
oil alternatively one can attribute to each individual the per capita income
(earnings or consumption). It is important to note that the two alterntivc
descriptions of the same data may yield different policy inferences : again
to take, an extreme illustration, the distribution of persons according to total
household consumption may just reflect the dfferencas in household size
so that no matter what the household size happens to be, its per capita con
sumption is the same. Admittedly this is extreme. On the other hand,
household size per se may not be revealing. For instance, it has often been
observed that single member households fall in both tails of the distribution :
rich, single and the eligible at the upper tail and the poor, old and unemploy
able at the lower tail.
The above discussion, suggests that it could be misleading to make policy
inferences without carefully analyzing the age-sex composition of the house
holds, the educational attainment of its members, its material asset holding,
etc. Again to illustrate: if it so happens that the households that fall in the
lower tail of the relevant distribution are characterized by their high dependency
POVERTY : SOME MEASUREMENT PROBLEMS
7
ratios and possibly being headed by females with dependent children
or consist of the elderly or the disabled, the relevant policy option would be
one of the income transfers to them. On the other hand, if the pool- house
holds arc characterized by their poor asset holding (say, land) and/or by their
members being illiterate or poorly educated, the policy intervention could
also include asset redistribution, better and greater access to education, etc.
Indeed these considerations have led the National Sample Survey in India to
collect data on the “weaker” sections of the society, employment and un
employment, asset and liabilities of rural housheolds, etc. These data have
not yet been fully analyzed taking into account possible interrelations among
these variables3. The analysis has to go beyond an attempt to “decompose”,
for instance, a. chosen measure of inequality into contributions of various
“factors” and their interactions, almost by analogy with analysis of variance.
While the analysis of variance has behind it a statistical model, the decom
positions of measures of inequality often have no explanatory model or theory
behind them so that they are no better than ex post accounting. As such,
not much can be learnt about the process of generation of inequalities (Rao,
1969; Pyatt. 1976 and Mangahas, 1975).
ft is also worth mentioning that, even though the household surveys
often can provide data on asset (human and materia]) holdings, the information
on the “quality” or “productivity” of these assets is usually not collected
or may even be impossible to collect. Again, as an illustration, the Indian
data show that small farmers tend to have a larger proportion of their land
irrigated. But the concept of irrigation used in Indian data, is to classify
any land that received at least, one irrigation during the crop season, how
ever inadequate it might have been, as irrigated. Clearly from a productivity
point of view, irrigation quality (intensity) as well as quantity is important.
The “quality” of skills and education (i.e. human capital) is even more difficult
to assess. There is a further problem that an identical asset may yield different
returns to different individuals, because the access to and/or the terms at
which other inputs are acquired for making productive use of the asset may
differ between individuals.
If the cost of fertilizer or irrigation water or credit are not the same for
two farmers holding land of identical inherent fertility but in different amounts
(as is often the case, the large farmer gets inputs at a lower cost) then the
3 Tho Income Distribution Division of tho Dovolopmont Research Center nt the World
Bunk is currently engaged in analyzing household survey data from a number of South Asian
anti Latin American countries in a research project in collaboration with ESCAP and ECLA.
Tho World Employment Programme of tho ILO has also generated a number of studios.
8
T. N. SRINIVASAN
size differential in their holdings will understate their income differential.
Obviously, if other contractual aspects of cultivation, such as the forms of
tenancy or sharecropping, are also brought in, the picture becomes oven more
complicated. Moro generally, in understanding the proximate causes of
poverty, one will necessarily have to prove into the nature of the markets
or other institutional arrangements by which resources get allocated and their
returns determined.
3.
Definition and measurement of a “Poverty Line”
In defining the “poor”, one could adopt either a relative concept, such
as that any one is deemed to be poor if that person is a member of a household
that happens to fall in the bottom 40% of the relevant income or expen
diture distribution or an absolute concept that a person is poor if the income
or consumption of the household is below a normatively defined poverty
line. While it is clear that in understanding poverty, the absolute concept
is more useful, its usefulness is intimately connected with the definition
(and measurement) of the poverty line.
As stated earlier, there have been two main approaches in the literature
in defining a poverty line at a point in time; the first is to define it as the cost
of a bundle of “basic goods” at the relevant prices (MuelTbauer, 1974)-’ and
the second is to define it as a nutritional norm, such as the “required” intake
of calories and proteins.
A number of measurement problems arise in using either definition. The
composition of the bundle of “basic goods” is not independent of the socio
cultural characteristics of the population and the structure of the economy,
or even the climate of the country. Even if we assume that “basic goods”
are defined narrowly to include consumption of food, clothing, shelter, and
services, such as education and health, the fact that some of this consumption
is provided to households through the public budget becomes relevant. The
Indian norm of poverty mentioned earlier, specifically excluded publicly
provided services and defined the poverty line as the mimimal per capita
household expenditure that was expected to cover the cost at base year
(1960-61) prices of a basket of basic goods in the market. Implicit in this
4 Clearly in defining the bundle, the size and age-sex composition of the household has to
bo taken into account. If wo rule out economies of scale in consumption, then only the ratio
of a household’s income (or expenditure) to the cost of its basic needs bundle is relevant, those
with a ratio loss than 1 being defined as poor. There are a number of articles on the methodology
us Wall as empirical estimates of consumer unit scales to adjust for age-sox composition of a house
hold.
POVERTY : SOME MEASUREMENT PROBLEMS
9
is the belief that the excluded items of consumption veil somehow be provided
to the required extent to all sections of the population. Whether, in fact,
this has happened is an issue on which the evidence is fragmentary—there
are few studies on the distribution of benefits from public expenditures on
health, education, water supply and sanitation, etc.6 Be that as it may, there
is a further problem that tho prices for the same set of commodities vary
depending on the location of the transaction (rural, urban, metropolitan, etc.),
time of year, and, more importantly, on the economic status of the buyer.
Thus, it is often the case that the poor pay more and get an inferior variety
or brand of tho same commodity that tho rich get cheaper. These issues
become extremely important if one wore to make comparisons (over time and
or space) of the proportion of the poor in a population (Srinivasan and Bardhan,
1974). Frequently^ however, they are ignored and inferences as to trends
in poverty are drawn on tho basis of as few as two observations widely
separated in time (Griffin and Khan, 1976). There is some evidence, at least
for India, that the fluctuations over time in tho proportion of rural popula
tion below tho poverty lino are substantial; while there is no time trend in
the proportion, tho fluctuations appear to bo associated with those in agri
cultural Output por head (Ahluwalia, 1977 and Son, 1974). It turns out that
Ahluwalia’s results are not affected if the so-called Sen index of poverty is
used rather than the proportion of poor in the population for comparisons
over time-
A nutritional requirement based poverty lure is fraught with oven more
problems of a conceptual nature. This is particularly the case when the
poor are defined as persons who are members of households whose per capita
calorie intake falls below some average norm set for tho population as a whole.
Even if ono wore to ignore tho intra-household variations (in actual intakes)
so that the per capita intake represents the actual intake of each member
of a household, the fact that tho calorie norms vary from person to person
(and, in fact, from day to day for the same person), can result in erroneous
estimates of the population of tho poor1. To take two extreme examples :
if every person, in fact, consumes as much calories as his “norm”, by defini
tion there are no poor persons in this society; yet, by identifying tho propor
tion of persons whoso intake is less than tho average norm for tho population
as tho poverty proportion, wo obtain an overestimate of the extent of poverty.
At another extreme, if everyone consumed the same number of calories but
“ Besides tho question of incidence of public expenditure on these items as well as the means
to finance it, there is also tho question whether tho level of expenditure is “optimal” in some
well-defined sense.
9R4S-2
10
T. N. SRINIVASAN
tho individual requirements varied, the estimated proportion of the poor (i.o.
those with a calorie deficiency) in the population could bo either 100% or 0%
depending on whether the identical consumption is less or greater than tho
moan requirement, while the true proportion, of course, will be somewhere
in between. Sometimes, the single number used to characterize the distribu
tion of calorie requirements of a given population may include a substantial
safety margin above the mean requirements, thus adding to the possibility of
over-estimation of the proportion of poor. Some of these problems arc
discussed by Sukhatme (1977), who shows, that, if instead of using a poverty
lino based simply on average calorie requirements, allowance is made for
variations in individual requirements, the estimated incidence of poverty in
India is brought down from about 50% to about 25% in urban areas, and
from about 40% to about 15% in rural areas.6 Indeed, as he points out,
from a policy point of view, the fact that these proportions are not as large
as was once believed, raises the hope that the problem of poverty amelioration
need not be intractable.
In some studios, it has been observed that calorie intake based estimates
of the proportion of malnourished can be substantially different from tho
proportion obtained of when malnourishment is defined in terms of body
weight in relation to some suitable standard. All this suggests, that perhaps
the most reliable method of estimating the extent of malnourishment is
anthropometric measurement and only recently this has come to be extensivolj'
adopted in poor countries. Even if the best estimate of tho malnourished
is available, there is still the problem of its utility for policy purposes.
Firstly, the evidence on the deleterious effects of mild malnourishment is
scanty since most studies concentrate on the severely malnourished. Secondly,
and more seriously, the problem of malnutrition may have other explanations
besides inadequate calorie intake, such as disease and unsatisfactory food
habits. Thirdly, certain sections of tho population, such as children, parti
cularly female children, expectant and nursing females, etc., may bo parti
cularly susceptible to malnourishment because they may not got their duo
share of tho household food consumption—the experience with direct nutrition
intervention programs directed towards these groups in India has not been
6 At tho request of FAO, tho National Sample Survey in India has tabulated tho caloric
and protoin content (por consumer unit por day) of tho food consumed according to por capita
total consumption expenditure of the household. These data wore collected as part of a nation
wide survey during July 1971-Juno 1972. Sukhatmo*s analysis relates to earlier data. It is
not clear whether this elaborate tabulation by itself adds very much to our understanding of
poverty or malnourishment.
11
POVERTY : SOME MEASUREMENT PROBLEMS
spectacularly successful.’ Finally, to dispel malnutrition may require raising
incomes, making available appropriate foods, reducing the incidence of
diseases and changing food habits. Further, provision of pure drinking water
may rise the nutritional and health status of tho population without any
additional food intake, simply by reducing the incidence of gastrointestinal
diseases Better education may again improve nutrition by enabling indivi
duals to improve the nutrition content of their food basket by proper choice
of foods, etc. The complementarity as well as substitution possibilities in all
these suggests that it would be misleading, as is often done, to estimate
separately the proportion of population subject to calorie deficiency, to diseases
of various kind, lack of protected water supply and sewerage, etc., the invest
ment and other costs of eliminating these deficiencies and adding up these
proportions and costs to arrive at the extent of poverty and the cost cf its
elimination. Even at a national level, this procedure is dubious. But some
hardy souls attempt global estimates. For examples of such flights of fancy,
(see McHale and McHale, 1977). Another extreme approach is to define
a hierarchy of “wants” or “needs”, say food, clothing, shelter, etc., in
that order. Then anyone who has inadequate food will be termed poor
even if ho is well clothed and housed. If he is well fed, then he is termed poor
only if he has inadequate clothing etc. The inadequacy of this approach
is apparent.
4.
Health, education and proverty abatement
It was mentioned above that there are complex interrelations among
nutrition, health and hygiene and education. Attention was also drawn to
the fact that poor households are often characterized by high dependency
ratios, in particular, having a larger proportion of non-earning children.
Tho relationships between fertility and infant mortality on the one hand,
and between those two and nutrition, health and education, particularly
female education on the other, arc relevant in understanding tho dynamics
of poverty. Space doos not permit a detailed discussion of tho issues involved.
It can be stated without fear of contradiction, that while a number of hypo
theses are available in tho literature on the nature of these rela tions, very few
of them have boon tested statistically with adequate and reliable da.ta from
developing countries (Cassen, 1976; Rosonzweig and Evenson, 1977).
11t has boon suggested that, one reason why direct intervention programs, such as mid-day
moal at school do not show any effect (i.e. no significant difference between tho nutritional status
of children subject lo the program compared to control groups) is that poor families often send
their children to school without feeding thorn at home. Thus the school meal acts as a substitute
and not as a supplement to homo food.
and
[J1EntM10N )
UNH
."
■'
12
T. N. SRINIVASAN
Nevertheless, there are signs that many of the countries in which the majority
of the world’s poor live seem to have passed the peak in the growth rate of
their population (McNamara, 1977). The decline in brith rates is more
pronounced and perceptible in some countries, such as Sii Lanka and regions
of India, such as Keiala, where longstanding public policies towards better
education (particularly of women), nutrition and health care seem to have
reinforced each other in bringing about the decline (U.N., 1975). This
interesting study reaches some rather far reaching conclusions that not many
would draw from the same body of data.
Turning now specifically to education, the available data do suggest
that poor households have disproportionately higher concentration of adults
and children -with little or no formal education. It has also been frequently
said that poor countries invest in higher education (rather than in primary
education, adult and functional literacy programs) more than they should,
in some sense. Both the data on the poor and the alleged misdirection of
educational expenditure reflect the notion that investment in education will
certainly bring high private returns to the educated while the returns to the
society from expenditure on education, at least some forms of it, may not bo
that high. An extreme version of this difference between private and social
returns to education arises in the screening theory of education, in which
education simply acts as a signal to the potential employers of the inherent
ability of the individual who has acquired the education, and, as such, educa
tion has no social productivity (Arrow, 1973). The individual has an incen
tive in getting educated, thereby revealing his superior inherent ability and
receiving its reward; for the employer, it serves the useful purpose of screen
ing out those with low inlierent abilty. But since the individual’s produc
tivity depends in this extreme case only on his or her inherent ability, and
not on the particular job assigned to him or her, there is no
social benefit from the individual’s acquiring education. In sharp contrast,
the human capital view of education treats educational investment on par
with investment in physical capita,! and both have social as well as private
returns though they may differ. Needless to say that both the screening
theory and the human capital theory describe some valid facets of education,
but neither provides the only and exclusive rationale for an individual’s
acquiring education. However, it is obvious that, if expansion of educa
tional opportunities is one of the poverty abatement policies, it would bo
useful to measure the returns to education.
There have been a number of studies in which returns (social and private)
to education have been estimated by essentially comparing the earnings of
POVERTY : SOME MEASUREMENT PROBLEMS
13
those with a specified level of education (more precisely schooling) to those
with the next lower level of education (Chamberlain and Grilliches, 1975 ;
B'a’ig Layard and Woodhall, 1969). An interesting methodological problem
of “self-selection” has only recently surfaced in this connection : suppose,
only those who believe that they can benefit from getting educated go to
school and if their beliefs in fact come true, any method of estimation of
benefit that does not allow for this self-selection may be misleading
(Maddala, 1977).
5.
Conclusion
Before concluding, a brief reference may be made to tire literature on
quantitative models of economic development, both to the planning and
simulation models that have explicitly oriented towards policies for poverty
abatement and reduction of income inequalities (Maddala, 1977: Adelman and
Robinson, 1977; Bacha and Taylor, 1977; Ahluwalia, etal, 1977; and Adelman
et al , 1977). These models attempt to incorporate some of the demographic
variables and their interrelationships with economic variables. The simulations
from these models show surprisingly little redistribution effect from almost any
policy short of changing the human and physical asset distribution. Only
when a number of policies are simultaneously applied, some perceptible redistri
bution effect emerges. These disappointing results point to the need for more
empirical work in understanding and quantifying the complex economic,
demographic and socio-cultural causes of poverty so that a better analytical
framework for policymaking becomes available. In devising such a framework,
the political and institutional factors clearly have to be brought in. To
conclude :
1)
It is essential to distinguish between inequality (of an income distribu
tion) and poverty, both from measurement and policy points of view.
2)
There are a large number of conceptual problems in making policy
use of cross-sectional relationships (whether the cross-sections relate
to countries or to households within a country).
3)
Two alternative methods of defining an absolute poverty lino have been
proposed in the literature : the first defines it in terms of income or
expenditure needed to ensure that a household can afford to purchase
in the market a basket of “basic goods”. The second defines it in
terms of the required intake of “calorics”. Though both measures
involve a number of conceptual and operational problems, perhaps
the expenditure based definition is preferable.
14
T. N. SRINIVASAN
4)
Even if an adequate definition of poverty is available, the inter
relationships between policies to abate poverty are not solidly estab
lished. Direct interventionist policies raise further problems, not only
regarding their feasibility and efficiency, but also from a much
broader point of view, that they sometimes invade into what are
essentially private decisions, such as the allocation of food among
family members, tho decisions on family size, on migration from rural
to urban areas, etc. It can be argued that the aim of policy should
bo to create the decision environment through appropriate incentives
(and non-coercive disincentives) in which individuals decide of their
own choice, to move in the direction that the society would want
them to move.
5)
The lack of knowledge about interrelationships and complementarities
is most evident in the areas of fertility, health, nutrition and education.
f>) Given the severity of the measurement problem, any umdimensisional
description of the nature and extent of poverty should be avo'ded.
6.
Acknowledgments
I wish to thank M. S. Ahluwalia. John Duloy, Bagicha Minhas and
Graham Pyatt for their comments.
References
Adelman, I. and Robinson, Sherman (1977). Income distribution policy in developing countries :
--1 case study of Korea. Stanford University Press (tn press).
Adelman, I., Hopkins, M. J. D., Robinson, S., Rodgers, G. B. and Wery, R. (1977). A Com
parison of two models for income distribution planning, (mimco.) Paper presented at the
Workshop on Analysis of Distributional issues in Development Planning at Bollagio, Italy.
Ahluwalia, M. S. (197G). Inequality, Poverty and Development. Journal of Development
Economics, 3, 307-342.
Ahluwalia, M. S. (1977). Rural Poverty and Agricultural Growth in India, (mimoo) Paper
presented at tho workshop on Analysis of Distribution issues in Development Planning
at Bollagio, Italy.
Ahluwalia, M. S., Lysy, Frank and Pyatt, Graham. (1977). A price endogenous model of
Malaysia: Some static experiments, (mimco) paper presented at tho workshop on Analysis
of Distributional issues in Development Planning at Bollagio, Italy.
Arrow, J. K. (1973) Higher education as a filter. Journal of Public Economics.
Bacha, E. and Taylor, Lance. (1977). Brazilian Income distribution in tho 1960’s : Facts,
results arid tho controversy (mimeo). Paper presented at tho workshop on Analysis of
distributional issues in development planning at Bollagio, Italy.
Cassen, Robert (1976). Population and development : A survey. World Development, 4.
Chamberlain, G. and Grilltches Z. (1975). Unobservables with a variance component struc
ture : Ability, schooling, and the economic success of Brothers, international Economic
Review.
Ohampernowne, D. G. (1973). The distribution of income. Cambridge University Press. London.
POVERTY : SOME MEASUREMENT PROBLEMS
15
Cooper, J. P. (1973). Timo varying regression coefficients : A mixed estimation approach and
operational limitation of tho general Markov structure. Annals of Economic and Social
Measurement,
Friedman, Miltox. (1957). .4 theory of the consumption function. Princeton Univ. Press,
Princeton.
Griffin, K. and Khan, A. R. (1970). Poverty and Landlessness in Rural Asia, (rr.imeo). I.L.O.,
Geneva.
Kakwani, N. C. (1977). Income distribution Mot-nods and applications. World Bank, Washing
ton, D. C. (Forthcoming).
Kuznets, Simon. (1933). Quantitative aspects of economic Growth of Nations, Part II. Econo
mic Development and Cultural Change. II.
Lay.ard, M. R. and Woodhall, M. (1969). The Causes of Graduate Unemployment in India.
Penguin, London.
M addala, G. S. (1971). Tho use of Variance components Models in Pooling cross-section and
Time Series data. Econonometrica.
Maddala, G. S. (1977). Self-selectivity Problems in econometric models (miroeo.) Univ, of florida
at Gainesville.
Mangahas, M. (1975). Income Inequality in tho Philippines: A decomposition analysis, world
Employment Programme working Paper, L.T.L.O., Genova.
Morale, Johx and Men.ale, Magda, C. (1977). Basic human needs : A framework for action.
U. N. Environment Programo, University of Houston, Toxas.
McNamara, Robert S. (1977). An address to the Massachusetts Institute of Technology on the
Population Problem. Word Bank, Washington D.C.
Muellbauer, John (1971). Prices and Inequality: the United Kingdom Experience. Economic
Journal.
Pant, Pitambar (1974). Perspectives of Development, India 1960-61 to 1975-76 : Implications
of planning for a Minimum level of living, In Poverty and Income Distribution in India
eds. T. N. Srinivasan and P. K. Bardhan. Statistical Publishing Society, Calcutta.
Planning Commission (1972). An approach to fifth five year Plan. Government of India. Now
Delhi.
Pyatt, G. (1976). On the interpretation of disaggregation of the gini coefficients. Economic
Journal.
Pyatt, G. (1977). Distribution of income or wealth : On international comparisons of inequality.
American Economic Review.
Rao, V. M. (1969). Two decompositions of the concentration ratio JRSS, 132/1.
Rosenzweig, Mark R. and Evensox, Robert. (1977). Fertility, schooling and the economic
contribution of children in rural Lidia. Econometrica.
Sex, A. K. (1973). On economic inequality. Oxford, London.
Sen, A. K. (1974). Poverty, Inequality and Unemployment: Some conceptual issues. Li Poverty
and income distribution in India, eds. T. N. Srinivasan and P. K. Bardhan. Statistical Pub
lishing Society, Calcutta.
Srinivasan, T. N. and Bardhan, P. K. (1974). eds. Poverty and income distribution in India.
Statistical Publishing Society, Calcutta.
Srinivasan, T. N. (1977). Development policies and levels of living of the poor (mimeo.). World
Bank, Washington D. C.
Sukhatme P. V. (1977) Nutrition and Poverty, Lal Bahadur Shastri Memorial Lecture. Indian
Agricultural Research Institute, Now Delhi.
Swamy, P. K. V. B. (1970). Efficient inference in a Random Coefficient Model. Econometrica.
United Nations (1975). Poverty, unemployment and development policy : A case study of selected
issues with reference to Kerala. Now York.
16
T. N. SRINIVASAN
Abstract
The paper raises some conceptual and measurement problems relating to poverty with
an aim towards formulation of policies for poverty alleviation. After drawing a distinction
between relative measures of the inequality associated with an income distribution and absolute
measures of poverty such as the proportion of population having a level of living below a norma
tively defined poverty line, the paper discusses measurement problems associated with both.
Problems of interpreting and using estimated cross-sectional and time series relationships for
policy formulation arc discussed. Attention is drawn in particular to the conceptual and
measurement problems associated with poverty linos based on nutritional norms. It is pointed
out that a serious gap exists in the literature from a policy point of view. This gap is the lack
of firm knowledge about interrelationships and complimentarities in the areas of fertility, health,
nutrition and poverty.
R6suxn6
Ce document aborde certains problemes conceptuch et de mesurc posds par la pauvretd,
on vue de la formulation de politiquos susceptibles d’attenuer co fleau social. Apies avoir fait
la distinction ontre la mesure relative de Yinegalitd associee & uno cortaino repartition du rovonu
ot la mosuvo absolue de la pauvretd, donn^e par exemplo par le pourcentago do la population ayant
un niveau do vio inferiour a la norme definie commo seuil de pauvretd, 1’auteur examine los difficultes do mesure qui apparaissent dans les deux cas. Il etudio notamment los problemes quo
soulevent l’intorpr6tation et 1’utilisation, aux fins de formulation de politiquos, des relations
ostimeos sur la base do donnons transversales et de series chronologiques ; 1’accont cst mis on
particulior sur los problemes concoptuols et de mesure que pose la definition des souils do pouvretd
& partir do normes nutritionnolles. Com me le souligne cetto etudo, la litteraturo spdcialisdo
presents en effect, du point de vue do 1’elaboration dos politiquos, uno gravo lacime : 1’inccrtitudo
dos oonnaissancos actuellos touchant aux relations de reciprocity et de compl6montarite entro
fdcondito, sante, nutrition ot pauvretd.
THE WORLD BANK
Hei dquarters:
18 S H Street, N.W.
\ Washington, D.C. 20433, U.S.A.
European Office:
66, avenue d'lena
75116 Paris, France
Tokyo Office:
Kokusai Building,
1-1 Marunouchi 3-chome
Chiyoda-ku, Tokyo 100, Japan
The full range of World Bank publications, both ft.ee and for sale, is
described in the World Bank Catalog of Publications, and of the continuing
research program of the World Bank, in World Bank Research Program:
Abstracts of Current Studies. The most recent edition of each is available
without charge from:
PUBLICATIONS UNIT
THE WORLD BANK
1818 H STREET, N.W.
WASHINGTON, D.C. 20433
U.S.A.
World Bank Reprint Series: Number Seventy-six
Co
T. N. Srinivasan
Development, Poverty,
and Basic Hranan Needs:
Some Issues
Reprinted from Food Research Institute Studies,
vol XVI, no. 2(1977), pp. 11-28
T. N. Srinivasan*
DEVELOPMENT, POVERTY, AND BASIC
HUMAN NEEDS: SOME ISSUESt
From the earliest days when development planning was at
tempted in many of the developing countries, raising the standard of living of the
poorest sections of the population to an acceptable level has been one of the major
goals, explicitly stated as such in the development plans in some countries and
implicit in others. However, over the nearly three decades of experience, the
perceptions of the strategies to be pursued in trying to achieve this goal have
changed.1 The early development plans aimed at accelerating the rate of growth
of real national income, focusing essentially on the process of capital accumula
tion and its allocation. The need for raising domestic savings as well as supple
menting it by external capital flow was emphasized. The debates were on the
sectoral allocation of investment, such as between agriculture and manufacturing
industry, choice of technology, and import substitution versus export promo
tion. The question of how the benefits of growth in national income were shared
by different socioeconomic groups in the society was infrequently raised. One
reason for this neglect was, of course, the belief that even the poorest will benefit
from growth, more so since institutional changes that were promoted at the same
time, such as some land reform and an increasing role of the public sector, were
supposed to facilitate this. Perhaps the main reason was that in the framework of a
mixed economy that excluded any revolutionary restructuring of production and
exchange relations, excessive emphasis on redistribution at an early stage in the
growth process was thought to retard growth and hence the long-run feasibility of
sustaining any appreciable increase in the levels of living of the poor.
The conviction that sustained and rapid growth is the desirable route toward a
better life for the poor countries as well as the poor in these countries was shared
by the major aid donors. Once the poor countries reached the stage of sustainable
and sustained growth, that is, the "take-off stage in the terminology of the
• The author is Special Adviser at the Development Research Center of the World Bank and
Research Professor at the Indian Statistical Institute, New Delhi
t I wish to thank, without implicating them, Messrs. M. S. Ahluwalia, B. Balassa, C. L. G.
Bell, H. B. Chenery, J. Duloy, G. Lamb, C. Lluch, P. Streeten, and a referee for their valuable
comments on earlier drafts. I also thank Ms. Rachael Weaving for her editorial assistance. The views
expressed are personal and not necessarily those of the World Bank and the Indian Statistical
Institute.
1 See Minhas (18) for a wider discussion of these issues.
Food Research Institute Studies, XVI, 2, 1977
T. N. SRINIVASAN
times, they would increasingly look like the mature economies of the West.
Furthermore, the late start of these countries would enable them to take advan
tage of modern technology and, with aid, to shorten considerably the period
needed to reach the take-off as compared to the historical experience of the mature
economies. Aid was viewed as helping this process of modernization without
revolutionary change.
In the early 1960s. at least in one major developing country, India, doubts
began to be raised whether in fact the poor had benefited from the growth in
national income achieved in the 1950s.2 Further, by the middle and late 1960s,
there was growing disenchantment with foreign aid in some of the major
developed countries for various reasons. But it did not reflect any significant
rethinking of development problems. In fact, the Pearson Commission reporting
to the President of the World Bank in 1969, held that during the first two
decades of developmental efforts, the less developed countries grew faster than the
industrialized countries did at a corresponding stage in their development. It was
also argued that if only the industrialized countries would fight off their aid
weariness and actively augment the flow of aid, the less developed countries
would succeed in lifting themselves out of the depths of underdevelopment in
reasonable time (9).
Events subsequent to the report of the Pearson Commission showed that aid
weariness did not disappear. The Commission's target for aid flows was not only
not achieved but the ratio of aid flow to gross national product (GNP) began to
decline for some of rhe major aid donors. By the early 1970s, concern about
environmental pollution led to a questioning of both the feasibility and the
desirability of further growth of GNP in many developed countries, including
the United States. This period also marked the realization that the problems of
the poor in developed countries were far harder to solve through public welfare
policy than had been believed earlier. These domestic concerns led to some
change in the understanding of development issues as well. Many aid donors
explicitly shifted the emphasis in aid policies to the problems of the poor.
The concern about the distributional aspects of growth was reflected in appeals
by the International Labour Office (ILO) and others to make the creation of
productive employment opportunities, rather than aggregate income growth, a
primary objective of policy. "A fundamental redirection of development
strategy was called for consisting of a rural strategy that "focuses on increasing
the productivity of the small farmer and the self-employed through better access
to land, water, credit, markets and other facilities" and an urban strategy of
restructuring] rhe modern sector to make it more responsive to the opportu
nity cost of labor and capital . . . [and] policies designed to reach the selfemployed and to make small-scale producers more efficient" (8. pp. xvii-xviii).
Subsequent emphasis in the World Bank on integrated rural development
Prime Minister Nehru of India was one of the earliest to voice doubts about the impact of such
strategies on rhe poor The Committee on Distribution of Income and Levels of Living was
appointed by rhe Government ol India in 1960 to inquire into the changes in levels of living during
the First and Second Plans, to study the trends in distribution of income and wealth, and in
particular to ascertain the extent to which the operation of the economic system has resulted in
concentration of wealth and means of production (i i, p, i)
DEVELOPMENT. POVERTY AND BASIC HUMAN NEEDS;
is
strategies and the choice of urban projects for Bank support reflected this
perception.
The apprehension that even the suggested shift in emphasis toward employ
ment goals may not be enough to tackle the problem of poverty within a
reasonable time led rhe 11.0 to go a step further. The declaration of principles and
program of action adopted by the Tripartite World Conference on Employment
organized by the 1LO proposed that strategies and national development plans
and policies should include explicitly, as a priority objective, the promotion of
employment and the satisfaction of basic needs of each country's population. It
further specified that basic needs should be understood to include certain
minimum requirements of a family for private consumption, such as adequate
food, shelter, and clothing as well as certain household equipment and furniture,
as well as certain essential services, such as safe drinking water, sanitation, public
transport, and health, educational and cultural facilities (z6).
It is of interest to note that just as redistributional concerns were first expressed
in India, employment generation in addition to income growth was included as
an objective as early as 1956 in India's Second Five-Year Plan. Further, the mam
ideas of the basic needs approach to the problem of the poor can be traced to the
paper by rhe late Pitambar Pant of the Indian Planning Commission (zj). The
author explicitly posed the problem of poverty alleviation in terms of providing at
least a minimum level of living for the entire population. This minimum needs
basket included essential items of consumption such as food, fuel ;ind light.
clothing, and shelter, as well as services such as health, sanitation, safe drinking,
water, and education to be provided through the government budget. The author
recognized that some sec tions of the population might not benefit from develop
ment that creates productive employment opportunities because of the high
dependency ratios in their households. These groups were to be provided their
minimum level of living through income transfers. The problem was also posed of
determining a rate of income growth that would nor be so high as to be infeasible,
but high enough to enable the minimum needs to be met. In one sense this article
is a precursor of Chenery (8) with its emphasis on the income-earning
capacity of the majority of the population, but it also goes beyond in explicitly
focusing on basic or minimum needs. However, it had little influence on Indian
policy until 1972, when the approach paper on the Fifth Five-Year Plan included
a minimum needs program as part of the plan that was quietly dropped in the
draft as well as the final version of the plan.3
In the remainder of the paper, it is proposed to examine the extent to which the
development performance of some of the major developing countries would
corroborate the premises underlying the "new” perceptions on development,
evaluate the basic needs approach, and finally raise some policy issues. It will
draw on both analytical studies and on actual experience in various countries,
including the papers presented and the discussions at a workshop organized by
the World Bank on Analysis of Distributional Issues in Development Planning
and held in Bellagio, Italy, during April 22-27, *977 A complete list of papers
is given in the Appendix.
' In the early 1973s there was extensive discussion among Indian economists on quantifying
the extent of poverty as well as on the evaluation of policies pursued toward poverty abatement. See
Srinivasan and Bardhan (23).
T N. SRINIVASAN
14
POVERTY AND INCOME DISTRIBUTION: CROSS-SECTIONAL
EVIDENCE AND TIME TRENDS
The extent of poverty in a country or region can be measured using either
absolute or relative indicators. Perhaps the most widely used absolute measure of
poverty is the proportion of the population below some poverty line. The familiar
measures of income inequality, such as the Gini coefficient, variance of the
logarithm of individual incomes, coefficient of variation, share in income or
consumption accruing to the bottom and top deciles or quintiles, are relative
measures, in the sense that they reflect the relative positions of different indivi
duals or groups of individuals in respect of their income or consumption. It is
perhaps better to describe them as summary measures of the inequality of income
distribution rather than as poverty measures. There are a number of conceptual
and measurement problems associated with both absolute and relative measures,
none of which will be discussed in this paper. (See 22.1
Cross-sectional Evidence
Kuznets, from his historical study of the development of some of the presently
developed countries, hypothesized that income inequality first increases and then
decreases as development proceeds. In testing this hypothesis, Ahluwalia related
the share in income of various income classes to the logarithm of per capita GNP
in constant 1970 U.S. dollars of 60 (developed and developing) countries, in the
form of a quadratic regression.4 The relationship was estimated separately for the
entire sample and a sub-sample of 40 developing countries. His results showed
that as per capita GNP rises, the share of income accruing to the poor—say the
bottom 40 percent of the population—first falls, reaches a minimum, and then
rises. The estimated per capita GNP at the "turning point,” at which the share
begins to rise again is S468 if the entire sample is used and S371 if only the
developing countries are considered. Their percent share in income falls from an
average value of 17 at a per capita GNP level of S1 oo to 1 1 at the turning point,
and rises to 15 at a level of Sa.ooo. This cross-sectional result appears to confirm
the Kuznets hypothesis.
There are well known difficulties with interpreting and projecting a crosssectional relationship over time. The cross-sectional curve essentiallv represents
an average relationship. The deviation of an individual countrv observation from
the estimated curve could be viewed as rhe effect of the policies being followed as
well as other relevant specific features of that country. Two types of projections
can be made from the curve: in one, starting from any level of per capita GNP,
one projects the per capita income for a future year and from the curve reads off the
share of the bottom 40 percent. Making projections in this wav. one is really
comparing the expected income (hypothetical average) share of the bottom 40
percent in countries which have the initial level of per capita GNP to the expected
share in countries where income has reached the projected value. This type of
projection is clearly not country-specific. In the second type of projection, one
starts from the given initial income level and the initial share ot the bottom 40
percent, then one adds rhe change in the share as estimated from the curve to the
4
The income classes were: top 20 percent, middle 40 percent. lowest 60 percent. 42 percent.
and 2? percent or the population. Sec Ahluwalia 12, pp. u”-42’
DEVELOPMENT, POVERTY, AND BASIC HUMAN NEEDS
15
initial share to obtain the share associated with the projected terminal income. In
this exercise, some allowance is made for the country’s specific initial cir
cumstances. Projections of either type, ifthey mean anything ar all, indicate what
might happen if incomes changed but the distributional and other policy envi
ronment did not change significantly. It would be wrong, therefore, to interpret
rhe curve and the projections from it as representing some sort of "iron law" of
development. It is, however, possible to make some limited and stylized policy
simulations based on the curve.
Time Trends in Poverty
Turning now from cross-sectional studies to time trends in poverty, Griffin
and Khan concluded that "development of rhe type experienced by the majority
of Third World countries in rhe last quarter century has meant, for very large
numbers of people, increased impoverishment” (74). This conclusion seems to be
somewhat hasty since continuous and comparable time series data on the size
distribution of income or expenditure are not available for most of the developing
countries. At best, data relating to two or three points in time are usually
available, but even so these data sets are rarely comparable, because of changes in
concepts and coverage. India is a major exception in that annual data on a
comparable basis are available on consumption expenditure by households sepa
rately for the rural and urban area of each of the major states, as well as for the
country as a whole. The data for India do not confirm the Griffin-Khan conclu
sion.
Ahluwalia used the Indian data to show, first, that in rural areas, where nearly
80 percent of India's population lives, the proportion of the population below a
normatively defined poverty line fluctuated substantially over the period 1956/57
to r 973/74, falling initially from over 50 percent in the mid-1950s to around 40
percent in rhe early 1960s, and then reaching a peak of nearly 57 percent in
1967/68 before declining (3). Second, there was no evidence of a significant trend
in these fluctuations over rime. Third, though the fluctuations in the incidence of
poverty in individual stares largely follow the all-India pattern, showing no clear
trend, there is a statistically significant positive time trend (that is. the propor
tion of people below the poverty line steadily increased) in three states, and a
negative trend in one state. The state of Punjab is among those showing no clear
trend—surprisingly, since real income and agricultural output grew faster here
than in most other states. Fourth, for India as a whole there is a significant inverse
relationship between the incidence of rural poverty and real agricultural income
per head. This is also true in seven individual states (again with rhe surprising
exception of Punjab), covering two-thirds of rhe rural population in India. Fifth,
the relative inequality, as measured by Gini coefficients, of the distribution of per
capita household consumption noticeably declined in eight states out of thirteen.
while the remaining five showed no trend. Thus Ahluwalia found no significant
increase in the proportion of poor in India, and relative inequality appeared to
have decreased. Indirect evidence, such as the rising expectation of life at birth,
and declining mortality (including infant mortality) and fertility rates would
seem to corroborate the conclusion that there could not have been a serious decline
in the living standards of the poor.
16
T. N. SRINIVASAN
It might be argued that the fact that nearly half the Indian population falls
below the poverty line in spite of over 2 5 years of development planning is in itself
A proof of the failure of the development strategy pursued by India. It is beyond the
' 'i scope of the present paper to evaluate India's development efforts. It suffices to
Sr \C Whatever its other results, Indian planning did not result in growth rates of
real national income as high or as steady as those in some other developing
countries.
2. Even the moderate growth achieved has decelerated since the late 1960s—a
period that was also marked by a steep drop in the external resource inflow, the
onset and strengthening of inflation, and a substantial drop in the growth of real
public investment, as well as what some believe to be a slowing in the growth of
food output, despite the Green Revolution. Not surprisingly, in the late 1960s
the incidence of poverty rose some after falling from the middle 1950s onwards.
3. The economic policies pursued, such as the bias against export activities, use
of administrative controls as a resource allocation mechanism and, above all, the
failure to enforce legislated institutional changes, such as land reform, meant that
both growth and income distribution performance were worse than they could
otherwise have been. India is clearly not an example where a successful growth
strategy failed to help the poorer sections of the society.Based on data for six years in the 1960s, instead of only two as used by Griffin
and Khan, and three different poverty lines corresponding to a calorie intake of
95, 92, and 90 percent, respectively, of the requirement, Naseem showed that
the proportion of the rural population of Pakistan below the first poverty line
decreased initially and then increased, while there is no trend if the other two
poverty lines are used (19). However, it should be pointed out that there are
conceptual difficulties with using a calorie requirement based poverty line. These
are discussed later.
Contrasted with India's unspectacular growth record, Brazil has had an average
annual growth of real income exceeding 6.5 percent since 1950. But data on the
size distribution of income are available only for the years 1960 and 1970 from the
demographic census data and for the year 1972 from a special household survey.
1 he census data do not include income in kind, direct tax payments, and
unrealized capital gains. Some of the controversy on the Brazilian income
distribution performance arises out of differing interpretations and adjustments
made to the census data. Other data of varying quality and quantity are available
relating to average wages, occupational wages, factor shares, and distribution of
earnmgs in the formal urban sector. While it is hazardous to base statements
about time trends on only two or three observations. Fishlow reported a consensus
among scholars that relative inequality increased in the decade of the 1960s,
though not enough to force down rhe ,h...,
/-c
,,
, .
ul™n tin absolute incomes of the poor <9)However, some would areue to the conrrare
, reported increase
'■outran on .1.
rhe 1basis of■ rhe
in infant mortality rates m some areas A Ise ,1, ,..
1
,.
„
,
- '
,
Ab0- 'he regional economic disparities in
reasons,■ the Norrlvesr
■ s,..1 relatively
1 ■ 1 poor,
.brazil where,
, , tor. structural
. ,
1',ortiuast continues
robe
nave to be kept m mind. Arnone those
1
•
■
-T ■
*who
w no do not deny
that
income
distribution
For further details see Bhagwaci and Desai is 1 ni.
and Bardhan Q;' '* nna^wau and Srinivasan (6) and Srinivasan
DEVELOPMENT. POVERTY. AND BASIC HUMAN NEEDS:
x-,
deteriorated in a relative sense in the 1960s. there are two prevalent schools of
thought. One focuses on the unequalizing effects on incomes of wage increases for
skilled labor, demand for which increasingly exceeds supply as aggregate growth
accelerates. The other stresses the importance of the post-1964 wage squeeze,
which slowed down inHation somewhat, but with rapid growth also allowed
shares of profits and top remunerations to increase. Bacha and Taylor are “agnos
tic but predisposed toward the wage squeeze explanations" (4). Fishlow rejects
the hypothesis of unequalizing inevitable effects of growth" arguing that the
Brazilian experience seems to have been one in which the strains of growth have
been amplified rather than counteracted by policy" and that a policy of increased
governmental transfers that are linked to educational investment in rural areas is
not only feasible but will help offset some of the concentration in incomes (9).
Korea and Taiwan provide examples ofcountries where real income growth was
both rapid (with annual growth rates exceeding 7 percent since the early 1960s)
and apparently shared—and increasingly—by those at the bottom of the income
distribution. Reliable data are available for Taiwan on a continuous basis only
from the mid-1960s. Somewhat less reliable data for two years in the 1950s show
that Gini coefficients then were comparable to those in other less developed
countries. The Gini coefficient for all households in Taiwan remained virtually
constant at about 0.33 between 1964 and 1968, and then declined by over 1 1
percent between 1968 and 1972. Both Korea and Taiwan were Japanese colonies
and went through major structural changes following the Second World War,
redistributing land and other productive assets not destroyed during the war.
Both have followed similar policies in respect of foreign trade. Ranis argued that
(20)
the dominant cause of the relatively favorable income distribution perfor
mance in Taiwan . . . [was] the massive shift of rural households from
agricultural to non-agricultural activity in the presence of a dynamic ag
riculture and in the absence of massive rural/urban migration . . . Taiwan
continued to invest heavily in its decentralized infrastructure . . . encour
aged rural industry directly via rural electrification grid, the maintenance of
equality in power and fuel rates as between rural and urban locations, and
the establishment of rural-industrial estates, bonded factories and process
ing zones located with an eye to rural location and mobility .
. [suc
ceeded] in the maintenance of a surprisingly high labor share in urban
industrial cum service activities [through] a labor-intensive output mix and
technologies, intimately tied up with the relative mildness of import
substitution combined with [subsequent] thorough liberalization efforts.
In the Korean case also the share of agriculture in employment declined
substantially while the share of the manufacturing sector increased with remarka
bly little capital deepening as reported by Rao (2/). Within manufacturing
substantial growth of labor-intensive export activities has occurred. Rural-urban
income differentials were narrowed through a variety of policy instruments,
including intervention in the determination of grain prices. For the economy as a
whole the average real wage has increased by about 7 percent per year during
1963-75. While the regime has not encouraged union activity, and strikes have
iS
T. N. SRINIVASAN
been illegal, there is apparently government pressure to improve the earnings
level of low wage groups. Emphasis on primary education and a successful
campaign of adult education have reduced illiteracy to negligible levels. This has
helped the industrialization process by making available well-trained and traina
ble workers.
The brief review of data suggests that it is simplistic to argue that in spite of
growth poverty has increased in developing countries.6 It does clearly indicate,
however, that other policies that were (or equally were not) being followed along
with the emphasis on aggregate income growth had a lot to do with success or
failure in achieving growth as well as equity. Also, it is not easy to isolate the
effect of changes in ' initial conditions" on which policies must be brought to
bear such as those wrought by social revolution (China, Cuba) or by war (Korea,
Taiwan)—either in bringing about equity initially or in increasing the chances of
success of equity-promoting policies subsequently.
There is no denying, however, that the extent of absolute poverty in the less
developed countries as a whole, and in some of the poorer, larger, and more
populous among them, is indeed staggering. It is this, rather than any claims of
its logical evolution in development theory with emphasis first on income
growth, then on employment, and finally on basic needs, which necessitates a
careful analysis of the proposed basic needs strategy.7
BASIC NEEDS APPROACH: DISTINGUISHING FEATURES
The approach to the problem of poverty contained in Chenery (8) and even in
rhe Minimum Needs Program of the Indian Fifth Plan in essence is a strategy of
insuring that adequate real purchasing power is placed in the hands of the poor.
Except for the socially provided services, the consumption pattern is to be
determined by private market decisions. The Indian approach is to determine the
necessary level of purchasing power on the basis of the cost of the minimum
consumption needs basket, again excluding socially provided services. Except for
those groups among the poor who cannot take advantage of employment and
income generation opportunities and who are to be provided transfer income, the
poor are to be reached through the income generation process.
The basic needs approach, on the other hand, does not rely solely on income
generation or transfers, and places primary emphasis on the production and
delivery to the intended groups of the basic needs basket through "supply
management" and a “delivery system."8 In a system in which production and
consumption decisions are primarily mediated through the market, the failure of
Needless to say, the responsibility for rhe particular use (critical or otherwise) of the data and
the interpretations derived lie with rhe authors of the papers presented at the workshop.
There is also a cynical interpretation of the conversion of some developed countries to the
basic needs approach By linking aid to performance of developing countries in providing the basic
needs of their population (and given the inherent difficulty ofa successful basic needs strategy), such
developed countries can de facto reduce their aid commitment while still appearing to be concerned
about poverty! Further, insofar as basic needs goods are primarily agricultural, emphasis on basic
needs will have the added consequence of slowing down industrialization of developing countries
and hence growth oi choir demand for non-renewable resources
8 While there are by now a number of arricles and pamphlets on rhe basic needs approach, the
clearest and most balanced account is given in Streeten (24).
DEVELOPMENT. POVERTY. AND BASIC HUMAN NEEDS'
>9
the poor to get their basic needs presumably reflects not only the unequal initial
distribution of real purchasing power, but also market imperfections and failures
In such a context redistribution of purchasing power alone may be insufficient to
insure that the poor receive their basic needs: market interventions may be
necessary on a continuous basis.
It can be argued that by selective direct interventions in the production and
distribution processes (rather than through creating purchasing power in the
hands of those who need it and expecting them to consume the basic needs
basket), the basic needs approach may lead to the provision of basic needs to
people at much lower levels of aggregate income per head than would otherwise
be possible. Further, it is possible that in tackling nutritional deficiencies and
similar problems, the income route may be inefficient, if only because ignorance
ot nutritional principles or preference for less nutritious foods as incomes rise. The
key issue then becomes one of delivery: is it possible to insure that the poor get a
nutritionally adequate diet in a cost-effective way? A similar issue arises in the
provision of health services and water supply. If the government or the ruling
elite’s preferences result in the public budget being used for arms, airports, big
urban hospitals, and higher education, instead of rural clinics, water supply,
roads, and primary education, the key question becomes how a shift in these
preferences might be brought about. In analyzing the delivery issue its political
implications are inescapable: the question really is "what sort of feasible produc
tion organization, institutional reform, and interventions are required in this or
that particular country to provide basic needs on a sustainable basis?" Some of
these issues will be taken up in the policy discussion below. But first it is
necessary to discuss the quantification of the basic needs bundle.
Measurement of Baste Needs
Even if the basic needs program were to be focused simply on eliminating
inadequacies in the caloric content of food consumption, quantification would not
be easy.9 A person's calorie "requirement” depends on his age, sex, and normal
activity, so there is a distribution of calorie requirements fora given population at
a point in time.10 Where a single number is used to characterize this distribution,
it is very often calculated so as to incorporate a substantial safety margin, in the
sense that if every member of the population took in this specified number of
calories per day, the actual calorie requirement of, say, ar least 95 percent of the
population would be met. While the true measure of rhe population with
deficiency in calorie consumption is that part of the population whose members
actually consume less calories than their individual requirements, the usual
estimates are based on a comparison, for the population as a whole, of the actual
consumption with the single number estimate of the calorie requirement for the
population as a whole.11 Estimates of a population's calorie deficit made this way
9 Some years ago alarms were raised about protein deficiency in the diets of the poor and the dire
consequences thereof It was later established that protein deficiency is part of a broader problem of
inadequate food and energy intake. Protein deficiency by itself without other deficiencies is
infrequently observed.
10 A person’s requirement can vary from day to day. See Sukhatme (25).
*1 Household survey-based data give total or per capita calorie consumption of each household
thus averaging out intra-household variations. An additional element of bias arises out of this
situation.
20
T. N. SRINIVASAN
can exceed or fall short of the true value: to consider an extreme example, if
everyone actually consumed the same number of calories but the individual
requirements varied, the estimated proportion of the population with calorie
deficiency could be either 100 or o percent, depending upon whether the
identical actual consumption was less or greater than the norm. The true propor
tion, of course, will be somewhere in between. Sukhatme showed that if, instead
of using a poverty line based simply on average calorie requirements, allowance is
made for variations in individual calorie requirements, the estimated incidence of
poverty is brought down from about 50 to about 25 percent in urban areas, and
from about 40 to about 15 percent in rural areas (2^). The literature on
malnutrition has very little to say on the effects of mild malnutrition.12 This
aspect becomes relevant if for lack of resources or other reasons it is not feasible to
meet the calorie requirement of the entire population fully.
Another, perhaps more serious, problem is that nutritional intervention alone,
better water supply, or curative medicine may have little effect on the mortality
or morbidity of a poor country, though if they are combined their impact can be
. considerable. This suggests that the quantification of basic needs, if feasible at all,
will have to be in terms of a bundle of things together rather than specific
I requirements independently derived of the elements constituting such a bundle.
I This issue is discussed further in the policy section below.
It is sometimes asserted that the great strength of the basic needs approach is
that by focusing directly on nutrition, health care, and so forth, it will have a very
favorable impact on fertility, infant mortality, or labor productivity. But these
effects may be present regardless of whether the poor live better because they have
higher incomes, or because their basket of basic needs is delivered to them. This
' being the case, the issue is really whether a basic needs approach will succeed in
| raising the quality of life of the poor more effectively than other types of policies.
/; Again this depends on the policy framework.
It is clear that it makes little sense to attempt to define universal standards of
basic needs, and that efforts at global modelling along this direction or at global
cost estimates for meeting basic needs are futile.13 The measurement problem
may lead some to conclude that it would be a sheer waste of time to wait for rhe
results of research on resolving these, and that the need of the hour is to push
governments to a commitment to seeing that basic needs are satisfied on a
sustainable basis over a reasonable period of time. Many would argue on the other
hand that while the cliche that "the best need not to be enemy of the good" has
some measure of validity, the issue is one of arriving at an appropriate set of
policies.
Policy Framework of Basic Needs
Approaches to development problems other than basic needs implicitly or
explicitly face the issue of temporal trade-offs in the sense that raising the incomes
12 The proportion of the population that is malnourished, measured as those "consuming less
calories" than their requirement, can be different from the proportion obtained if malnourishment
is defined in terms of body weight in relation to some standard
There arc a few hardy souls who are undeterred by conceptual or data problems. For flights of
fancy in global estimation, see McHale and McHale (r7) and Burki and Voorhowe (7).
DEVELOPMENT, POVERTY, AND BASIC HUMAN NEEDS.
of the poor through redistribution now, if pushed, makes it difficult to sustain
these incomes if the redistribution cuts too much into savings for growth. The
literature on basic needs has not adequately discussed the issues of trade-offs
among different basic needs (food, shelter, or health) at one time and over time,
that is, satisfaction of a basic need now, versus more of this good or another in the
future. Given that resources are scarce, increasing the supply of one set ofgoods
involves the sacrifice of one or more other goods, if the system is productively
efficient initially. Either basic needs involves the production of a different
consumption basket with the same resources that were earlier devoted toward
consumption, tn which case investment (in the aggregate though not in composi
tion) is not affected, or, even if resources have to be diverted away from
investment activities to producing basic needs, the productivity-raising aspects
of basic needs will be sufficient to offset the loss in future production possibilities
that would otherwise have occurred. Whether the trade-off problem is serious or
not then becomes an empirical issue. Both Sri Lanka and Tanzania have followed a
basic needs-type strategy apparently at the expense of growth, and thus of future
capacity to provide basic needs, and it is possible that both will find it difficult to
continue this strategy.14
The distinction between countries in which a moderate redistribution of
current income flows would be adequate to meet the basic needs of the entire
population (for example, Brazil, Mexico) and countries where even the most
radical redistribution feasible will still leave a large section of the population with
deficiencies in their consumption of basic needs (Bangladesh, India, Pakistan) is
clearly important from a policy angle. The temporal trade-off mentioned earlier is
far more serious in the latter group of countries. In the other group, policies to
close the basic needs gap need not necessarily involve major structural change.
The extreme fuzziness in the basic needs literature on policy aspects can be
illustrated with reference to nutrition programs. As mentioned earlier, it is
believed that raising the incomes of the poor need not eliminate their nutritional
deficiency since, left to themselves, the poor may spend their additional incomes
either on foods of lower nutritional content than those consumed at low levels—
for example, replacing coarse with finer cereals, such as rice and wheat—or on
non-food items. On the other hand, direct nutritional-supplementing programs
oriented toward target groups, such as pregnant or lactating mothers or school
children, have often run into substitution problems: for example, if pregnant
women treat the special foods merely as substitutes for what they would have
eaten at home, or where the supplementary food provided in a school program is
offset by reducing the food intake ofchildren at home. This may not be altogether
bad, in the sense that the household as a whole benefits either because others eat
more than they would otherwise have, or because some resources are released for
consumption of other items. But the primary objective, namely to reach the
target groups, is not achieved. Indirect evidence for this phenomenon has come
from several studies which show only insignificant differences between the
average weight gain or health situation of intended beneficiaries of special feeding
programs and the weight or health observed in control groups. Besides being
14 There is some doubt whether Tanzanian income inequality has;iWifncrtTrsminmed
by some.
XV'usR""
0ocu^'onl a
22
T. N. SRINIVASAN
ineffective, such programs are very costly ways of doing wrong things. The
Applied Nutrition Programs that have been part of the Five-Year Plans in India
for over a decade have run into similar problems.
It was mentioned earlier that there is a significant complementarity among
health, sanitation, water supply, and nutrition programs. In the absence of a safe
water supply and control over communicable diseases, efforts at improving the
nutritional status of the population may be ineffective and costly. And in the
absence of adequate nutrition, resistance to diseases will be lower and the cost of
curative health programs will be higher. A critical mjnimum effort, may be %
needed mall these directions simultaneously ifeach is to have any effect at all! The
impact of the spread of education on raising the productivity of investment in all
these areas in general and in nutrition programs in particular may be significant
' but is as yet not fully researched.
Another important issue in providing many of the basic needs is the question of
appropriate technology and delivery systems. A water supply scheme which
would be prohibitively costly if it used the urban technology of advanced
countries may become feasible if local initiatives and resources are used in
con|unction with a technology which does not necessarily involve individual
house connection to water mains. Thus the sociopolitical institutional framework
in which the basic needs programs are to be implemented may be the overwhelm
ing determinant of their feasibility and effectiveness.
The importance of the content and the system of delivery of health care cannot
be underestimated. Reference has already been made to the inadequate budget
share and the ill-conceived delivery system (modern large urban hospitals) of
health care in many of the developing countries. However, there are countries
(even if the People's Republic of China is left out), such as Sri Lanka, or regions
within a country, such as Kerala in India, where the conception and delivery of
health care have differed markedly from that elsewhere in developing countries.
Critical evaluation of these efforts in a comparative framework would yield
valuable policy lessons, as would a similar analysis of policies toward education
and literacy It is no coincidence that the contrast in the educational policies of
Korea and Taiwan on rhe one hand and of Brazil on the other is remarkable.
Apart from the vagueness of a basic needs approach in respect of crucial policy
aspects, there is an inherent contradiction in the position adopted by some basic
needs proponents. It blames the existing sociopolitical framework with its vested
interests for preventing the poor from sharing in the fruits of development, while
at the same time these institutional bottlenecks are assumed to be somehow less
relevant for a basic needs strategy.
Distributional Policies: Models, Facts, and Politics
There are by now a number of computable development models incorporating
redistributional considerations in one form or the other. The consumption vectors
in some of rhe early exercises in the 1960s on input-output based consistency
models for India were derived using estimated Engel curves and alternative
Lorenz ratios of the distribution of consumption expenditure. An updated version
of this type of exercise was part of the Fifth Five Year Plan (72). These exercises
were of limited use since the policies that were to bring about the reduction in
DEVELOPMENT. POVERTY. AND BASIC HUMAN NEEDS.
2i
inequality of consumption expenditure were not part of the model By contrast
the models for Brazil, Korea, Malaysia, and the Philippines attempt to link
income generation to factor endowments through factor prices in a general
equ.hbr.um framework in which most factor and commodity prices are deter
mined endogenously. Some of these models, for instance, the one for the Philip
pines, include a demographic subsystem as well.
1 he conclusions that emerge from the policy simulations of the Korea and
Philippines models are:
i 1 he size distribution of income remains exceedingly stable even in the face
of substantial policy interventions.
2. Further, the relative degree of poverty and wealth as a whole is much less
affected than the location (rural or urban) or poor and wealthy.
3- Limiting population growth leads to a deteriorating income distribution
and increasing poverty in the short and medium term; beneficial effects emerge
only in the very long term Rural-urban migration is the most important
demographic variable in improving income distribution, up to a 25-year horizon.
4. Appropriate trade strategy will help increase the absolute incomes, as well as
the share, of the poor.
5- Agricultural terms of trade are the most important policy instrument for
improving the lot of the poor.
7. Above all, only a massive, wide-ranging, balanced and continued attack on
poverty and maldistribution of income has much chance of succeeding; lesser
modifications to existing strategies will fail. Successful planning of income
distribution can be devised using an integrated array of policies, without chang
ing the fundamental rules of the economy (z).
It appears that the specification of these models explained some of their
counter-intuitive results. The result from the Korea model that any increase in
agricultural productivity would be absorbed through a reduction in the agricul
tural terms of trade, and hence would increase inequality, stemmed directly from
rhe model's peculiar lack of attention to foreign demand for agricultural output;
considerations of comparative advantage and access to the world markets were
modelled inadequately in respect to the agricultural sector. For small open
economies, such as Korea, this does not make sense. The result that limiting
population growth had negative effects on income distribution, except in the very
long run. seems to arise from the effect on the agricultural terms of trade of the
shift in demand for agricultural products, relative to their supply, which is caused
by lower population growth. This suggests that rhe agricultural sector had been
inappropriately modelled. The remarkable insensitivity of "size" distributions to
policy changes may be due to the fact that these models basically describe the
functional income distribution, the size distribution being derived more or less
mechanically from it. The flexibility of money wages and prices built into the
model, along with some payment flows fixed in nominal terms, meant that
substantial changes in exogenous demand had limited effects on the overall level
of economic activity and the distribution of income.
The models for Brazil reviewed in the paper by Bacha and Taylor focus on the
savings-investment equilibrium (4). As exogenous demands for investment,
exports, or government consumption shift, relative prices and employment (and
hence income flows) adjust to maintain the macro-economic balance, while rhe
24
T. N. SR/X/VASAX
exact mechanisms of adjustment vary from model to model. As investment
demand increases with growth, income distribution becomes more concentrated,
to the benefit of high saving groups While the model results are by no means
conclusive, Bacha and Taylor believe that their models are no worse, and perhaps
better, than other explanations of Brazilian income inequality
The above discussion suggests that general equilibrium neo-classical models
with fairly smooth price adjustments may not be satisfactory tools for analyzing
dynamic processes arising out of discrete jumps, rather than gradual changes,
from an initial position of disequilibrium. Indeed most of these models are
essentially comparative static in character, and in the dynamic models the
postulated dynamic adjustment mechanisms, such as the determination of aggre
gate investment and its sectoral allocation, are often simplistic. It is conceivable
that structural changes (in ownership of assets, labor force participation, skill
acquisition, and demographic characteristics) are the dominant forces in effecting
any significant changes in income distributions and available models are in
adequate to portray the process of structural change.
Country Experiences
In evaluating the policies pursued in countries that have experienced im
provement in their income distribution (Korea, Taiwan) and deterioration
(Brazil, Colombia, the Philippines) two divergent characterizations are possible.
One could attribute rhe “success" of Korea and Taiwan more to the change in
their “initial” conditions brought about by the violence of war, occupation by
non-native regimes (mainland Chinese in Taiwan) and the interest of rhe domi
nant military allies of the regime (the United States in both Taiwan and Korea)
than to their subsequent economic policies. In this view, the success of Korea and
Taiwan was not replicable in other countries with different initial conditions. The
other characterization, while not denying rhe importance of initial conditions,
would suggest that the mix of economic policies actually pursued did matter a
great deal. Taiwan, while initially following import substitution policies in
consumer goods, soon enough switched to a policy of encouraging labor-intensive
industrial consumer goods. On the other hand, Colombia and the Philippines
maintained an import-substitution strategy, extending it to more capitalintensive intermediate and capital-goods industries. The location of industrial
estates in rural communities enabled Taiwan to pursue a balanced agricultural
and non-agricultural rural growth. Korean growth involved very little capital
deepening in the aggregate and its emphasis on labor-intensive manufactured
exports was similar to that of Taiwan. This enabled Korea to absorb a growth in
the labor force of more than 3 percent per annum during 1963-75 and to reduce
unemployment significantly. Both in Taiwan and Korea, real wages rose sig
nificantly once the labor surplus phase was over. Rural-urban wage differentials
were not allowed to deteriorate, and in fact were improved by government
interventions in the determination of agricultures terms of trade. The contrast
with the policies pursued in the other group of countries, particularly in Brazil,
could nor have been greater. The success ofTaiwan and Korea, it could be argued,
strengthened the case for a feasible reformist strategy in other countries.
DEVELOPMENT. POVERTY. AND BASIC HUMAN NEEDS.
*5
Political Framework and Distributional Policies
Ac the risk of sounding naive and ignorant politically, a few remarks on rhe
politics of income distribution will be offered. It is obvious that except in the
unlikely situation where everyone benefits from a policy change, the gain to rhepoor in a redistribution policy has to be at the expense (at best relatively and at
worst absolutely) of rhe non-poor. If a reformist strategy oriented toward redis
tribution is to be successfully implemented either the regime has to be sufficiently
authoritarian to be able to impose it, or, in a liberal and plural framework, those in
power have to be able to count on or able to mobilize the support of a broad coalition
of necessarily different interest groups that nonetheless advocate the reforms pro
posed for their own reasons. Some would argue that historically speaking major
redistributions of wealth have resulted only after a war or occupation by a foreign
power or they have been imposed by an authoritarian regime or a violent
revolution. However, these events are either exogenous or unlikely to be deliber
ately promoted by governments in power. Reformist, rather than radical.
strategies are likely to be the only feasible options for improving distribution in
most countries. While it is of great interest and importance to understand howviable, progressive coalitions of interest groups could be formed in different
situations, this process is not easily modelled nor can it be orchestrated from
outside. The brighter side of the picture is the fact that the technological and
institutional (in a broad sense) conditions under which a reformist strategy could
be pursued in the third quarter of the twentieth century are not the same as those
that existed in countries where redistribution took place after violence of one sort
or another. For instance, the technology of the so-called Green Revolution is, if
anything, scale-neutral. New high-yielding varieties raise output per acre as
compared with traditional varieties, even if no additional inputs of fertilizer or
water arc- used. To the extent that this is valid, lack of thoroughgoing land
reform is less serious a bottleneck than it would otherwise have been. To say this is
not to minimize the need for land reform.
SUMMARY AND CONCLUSIONS
Before offering some concluding remarks, it is useful to summarize briefly the
above discussion. The cross-country data seem to support the hypothesis of
Kuznets that as development proceeds, income inequality worsens first before it
improves. But clearly this is not an iron law of development as is evident from the
variations in the performance of different countries. There is no strong evidence to
suggest that the problem of absolute poverty in developing countries has wors
ened despite growth in GNP in the last three decades. In fact, the evidence is
mixed: poverty has been significantly reduced in some but not in others of the
group of fast growing countries. Similar mixed evidence was obtained with
respect to slow growing countries as well. A shift of development policy, to the
provision of basic needs to target groups through selective interventions in the
production and distribution processes, in spite of its appeal on the surface,
appears to be based on an inadequate understanding of the conceptual and
measurement problems in quantifying basic needs and on an almost naive belief
26
T. N. SRINIVASAN
that the very same institutional bottlenecks that prevented the benefits of growth
from reaching the poor to any significant extent, would somehow be absent if the
policy is the provision of basic needs to the poor. It appears that the only sensible
approach is to emphasize growth as in the past, but supplementing (rather than
supplanting) the growth strategy with policies toward better distribution of
benefits of growth and experimentation with alternative approaches and delivery
systems for providing food, education, health, water supply, and sanitation to the
poor. The question then is whether such a "reformist," as contrasted with a
"revolutionary,” strategy depends for its adoption and success more on favorable
initial conditions, which in the past have been brought about by exogenous
events such as war or occupation, than on the economic policies pursued during
the course of development. No firm answer seems to emerge from the analysis of
the development policies and performance of the developing countries since the
Second World War.
It would be a serious error to conclude that the growth performance of the
developing countries is insignificant and that there has been no improvement in
the levels of living of the poor. The growth rates achieved by these countries since
1950 are impressive compared to their own past record and to the record of
presently developed countries when they were at their initial stage of develop
ment. Nor have the poor been completely left out of the development process.
Indicators such as expectation of life at birth, mortality rates (in particular, infant
mortality), and school enrollment rates do suggest that some improvement has
taken place in the levels of living of the poor. Undoubtedly growth achievements
have fallen short of expectations. But it would be tragic if the serious misun
derstanding of the performance of past development strategy leads to the adop
tion of development policies based on ill-defined concepts such as basic needs, to
the detriment of growth. A development strategy cannot be fully articulated on
the basis of the need to provide a limited set of goods and services to a part of the
population. Nor can any success achieved in the provision of basic needs be
sustained in the future without growth. Instead of turning into a blind alley,
students of development should devote their efforts to the difficult task of
understanding rhe sociopolitical characteristics of the development process. Such
an understanding is the first step on the road toward developing a framework for
defining and evaluating alternative development strategies for poverty allevia
tion.
CITATIONS
1 I. Adelman et al., "A Comparison of Two Models for Income Distribution
Planning," paper presented at rhe World Bank Workshop on Analysis of Dis
tributional Issues in Development Planning, Bellagio, Italy, April 22-27, 19772 M. S. Ahluwalia, “Inequality, Poverty, and Development, Journal of
Development Economics, 3, 1976.
3
, "Rural Poverty and Agricultural Growth in India, paper pre
sented at the World Bank Workshop on Analysis of Distributional Issues in
Development Planning, Bellagio, Italy, April 22-27, >9774 E. L. Bacha and L. Taylor, "Brazilian Income Distribution in the 1960s.
'Facts,' Model Results and the Controversy," paper presented at the World Ban
DEVELOPMENT, POVERTY. AND BASIC HUMAN NEEDS:
Workshop on Analysis of Distributional Issues in Development Planning, Bel
lagio, Italy, April 22-27, >977•
5 J. Bhagwati and P. Desai, India: Planning for Industrialisation , Oxford
University Press, London, 1970.
6 J. Bhagwati and T. N. Srinivasan, Foreign Trade Regimes and Economic
Development: India, Columbia University Press, New York, 1975.
7 S. J. Burki and J. J. C. Voorhowe, "Global Estimates for Meeting Basic
Needs, Background Paper," Policy Planning and Program Review Department,
World Bank, August 1977.
8 Hollis Chenery et al., Redistribution with Growth, Oxford University Press,
London, 1974.
9 Commission on International Development, Partners in Development,
Praeger, New York, 1969.
so A. Fishlow, "Brazilian Income Distribution: Does Trickle-Down Really
Work?" paper presented at the World Bank Workshop on Analysis of Distribu
tional Issues in Development Planning, Bellagio, Italy, April 22-27, [97711 Government of Ind ia, The Report ofthe Committee on Distribution ofI ncorne and
Levels of Living: Part I, Planning Commission, New Delhi, February 1964.
12 ------------ , A Technical Note on the Approach to the Fifth Five Year Plan ofIndia.
1974-79, Planning Commission, New Delhi, April 1973.
1 3 ------------ , Perspective of Development 1961-1976: I mplications of Planning for a
Minimum Level of Living, Planning Commission, New Delhi, 1962; reprinted in
(23).
14 K. Griffin and A. R. Khan, "Poverty and the Third World: Ugly Facts and
Fancy Models," paper presented at the World Bank Workshop on Analysis of
Distributional Issues in Developing Planning, Bellagio, Italy, April 22-27,
'97715 International Labour Office, "Poverty and Landlessness in Rural Asia,"
Geneva, 1976, mimeograph.
16
, Meeting Basic Needs, Geneva, 1977.
17 John McHale and Magda Cordell McHale, Basic Human Needs: A
Framework for Action, Center for Integrative Studies, University of Houston,
Houston, April 1977.
18 B. S. Minhas, "The Current Development Debate," Institute of Develop
ment Studies, Sussex, February 1977, mimeograph.
19 S. M. Naseem, "Rural Poverty and Landlessness in Pakistan, Dimensions
and Trends,” International Labour Office, Geneva, April 1977.
20 G. Ranis, "Equity with Growth in Taiwan: How 'Special' Is the 'Special
Case’?" paper presented at the World Bank Workshop on Analysis of Distribu
tional Issues in Development Planning, Bellagio, Italy, April 22-27, >97721 D. C. Rao, "Economic Growth and Equity in Korea," paper presented at
the World Bank Workshop on Analysis of Distributional Issues in Development
Planning, Bellagio, Italy, April 22-27, 1977.
22 T. N. Srinivasan, “Poverty: Some Measurement Problems," paper pre
sented at the 41st Session of the International Statistical Institute, New Delhi,
December 5-15, 1977.
23 T. N. Srinivasan and P. K. Bardhan, editors. Poverty and Income Distribn-
28
T. X. SRIXIVASAX
tion in India, Calcutta Statistical Publishing Society, Calcutta, 1974.
24 Paul Strecten. ' The Distinctive Features of a Basic Needs Approach to
Development," Program and Policy Review Department, World Bank, August
1977, mimeograph.
25 P. V. Sukhatme. “Nutrition and Poverty." Ninth Lal Bahadur Shastri
Memorial Lecture, Indian Agriculture Institute, New Delhi, 1977.
APPENDIX
PAPERS OF THE WORLD BANK WORKSHOP ON ANALYSIS OF
DISTRIBUTIONAL ISSUES IN DEVELOPMENT PLANNING, BELLAGIO,
ITALY, APRIL 22-27, 1977
15.
16.
17.
I. Adelman. M. J. Hopkins, S. Robinson, G. B. Rodgers, R. Wery, "A
Comparison of Two Models for Income Distribution Planning"
M. S. Ahluwalia, “Rural Poverty and Agricultural Growth in India"
M. S. Ahluwalia and J. Duloy, "Poverty Alleviation and Growth Pessi
mism: A Reexamination of Cross-Country Evidence"
E. L. Bacha, "The Kuznets Curve and Beyond: Growth and Changes in
Inequalities”
E. L. Bacha. and L. Taylor. "Brazilian Income Distribution in the 1960s:
Facts, Model Results and the Controversy”
C. L. G. Bell, “A Simple Dualistic Economy in a Comparative Statics
Setting”
M. Bruno, "Distributional Issues in Development Planning: Some Reflec
tions on the State of the Art"
H. B. Chenery and N. Carter, "International Aspects of Poverty and
Growth"
A. Fishlow, "Brazilian Income Distribution: Does Trickle-Down Really
Work?"
K. Griffin and A. R. Khan, “Poverty in the Third World: Ugly Facts and
Fancy Models"
G. Lamb, "Distributive Politics in Tanzania"
D. Lehmann, "The Death of Land Reform"
, "The Politics of Armaggedon: Chile 1970-73"
M. Lipton, "The Technology, the System and the Poor: The Case of the
New Cereal Varieties"
C. Lluch. "On Simple Macroeconomic Models"
G. Pyatt. "Labor Markets and the Efficiency of Labor"
G. Ranis, "Equity with Growth in Taiwan: How Special is the Special
18.
19.
20.
21.
22.
Case?"
D. C. Rao, "Economic Growth and Equity in Korea"
G. B. Rodgers, "Demography and Distribution"
F. Stewart, “Inequality. Technology and Payments Systems”
P. Streeten, “Basic Needs: An Issues Paper”
L. Taylor and F. Lysy, “Vanishing Short-Run Income Redistributions:
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
Keynesian Clues about Model Surprises”
THE WORLD BANK
Headquarters:
1818 H Street, N.W.
Washington, D.C. 20433, U.S.A.
European Office:
66, avenue d'lena
75116 Paris, France
Tokyo Office:
Kokusai Building,
1-1 Marunouchi 3-chome
z~hivoda-ku, Tokyo 100, Japan
The full range of World Bank publications, both free and for sale, is
described in the World Bank Catalog of Publications, and of the continuing
research program of the World Bank, in World Bank Research Program:
Abstracts of Current Studies. The most recent edition of each is available
without charge from:
PUBLICATIONS UNIT
THE WORLD BANK
1818 H STREET. NW.
WASHINGTON, D.C. 20433
U.S.A.
CHAPTER
11
The Structural Nature of Poverty in India
F. Franco S. J.
Introduction :
Poverty in India has been a much-discussed phenomenon;
a great deal has also been done to reduce its severity and dimen
sions. At the same time, one is left with the feeling that there
remains, in many minds, the unstated belief that poverty, like
malaria, is a disease contracted by individuals at birth, due to
some external, unavoidable natural causes. Like malaria-era
dicating programmes, anti-poverty programmes tend to conceive
the poor (and poverty) as a target group which can be sprayed
with benefits so that every year a number of them can cross a
certain magical line (the poverty line) and be cured of poverty,
as one is cured of malaria.1 This unconscious belief assumes
that poverty is mainly an individual phenomenon capable of
being quantified and analysed by counting heads.
This paper makes a deliberate attempt to state the oppo
site: poverty is not only or even mainly the problem of some
individuals — however numerous they may be — who are
deprived of the necessities of life, but a necessary corollary of a
society which is organised and structured in a particular manner.
The logical theme developed in the paper runs as follows:
economic activities or practices play a decisive role in shaping
the life of society. Economic growth may not, necessarily, be
equated with growth in the welfare of people, but economic
development is an important determinant of the latter. A tradi
tion of economic theory, which tends to be constantly relegated,
by interested parties, to the archives of history, attributes to the
economic surplus generated by society a prominant role in the
analysis and understanding of socio-economic change. The
size and composition of this surplus determines, to a great extent,
the pattern of future development. Effective control over the
38
39
distribution of the surplus results in control over the pattern of
development and change; it, therefore, follows that the crucial
factor in the analysis of societal change is the analysis of the
means by which people achieve control over the surplus. It is,
ultimately, the effective ownership of various assets that entitles
individuals and groups to control the distribution of the surplus.
This effective ownership must be understood not in terms of legal
rights alone, but rather in terms of existing power relations: an
unequal distribution of assets is concomitant with an unequal
distribution of power. Asymmetrical power relations deter
mine a social structure that in turn maintains and reproduces an
unequal and unjust pattern of asset-distribution. The circle is
closed and the movement repeats itself. Poverty is not only a
situation of deprivation for individuals or groups, but a set of un
equal power-relations existing within a particular society, affec
ting all its members. These power-relations determine a con
crete situation in which a number of people are unable to control
their own lives and destinies. Poverty is not a line or a number,
but the result of structural relations of dominance. An analysis
of poverty must, therefore, explain the unequal distribution of
assets and, finally, identify the groups that control the distribu
tion of the surplus and the distribution of power.
Two further clarifications are in order; first, there is a
certain bias in the paper towards economic facts; it must be con
stantly borne in mind that we understand the term "economic"
as always including relations of power among people; at the same
time, limitations of space have prevented us from treating other
semi-autonomous social forces and processes like caste, in
greater detail. Second, we have neither dealt with the policy
implications emerging from this analysis nor have we attempted
a critical review of anti-poverty programmes.
Pedagogical considerations have been responsible for alter
ing the logical order of the argument in the exposition that follows.
The first section poses the problem of the coexistence of growth
and poverty which must finally be explained by structural causes.
The second section analyses the unequal distribution of assets in
India using the conceptual terminology developed by Amartya
Sen; it should be clear that the use of Sen's conceptual frame
work is subordinated to the understanding of the centrality of
the surplus. This approach creates a definite boundary bet
ween Sen's analysis and the argument developed in this paper.
The third section contains the main theoretical assumptions and
explains the conditions that generate poverty; it is logically prior
to the other two sections. The last section examines the nature
and behaviour of the groups controlling the surplus and follows
generally the ground already explored by other political econo
mists like P. Bardhan and R. Sau.
I.
COEXISTENCE OF GROWTH AND POVERTY
The past thirty years have witnessed a dramatic transforma
tion of the Indian economy, which is reflected in the growth
rates experienced by some key indicators. During the period
1951-1985, national income and per capita income have grown
at 3.5 and 1 .4 per cent per annum, respectively.2
Net domes
tic product at 1970-71 prices, has more than doubled in the last
25 years; from Rs. 24,360 crores in 1960-61 to Rs. 57,286
crores in 1984-85. The growth measured by these indicators
has been reflected in a number of areas.
(i) Creation of a broad and varied industrial infrastructure:
for instance, crude oil production which was non-existent
in 1950-51 has reached the level of 29 million tonnes in
1984-85; fertiliser consumption which was almost non
existent in 1950-51 has reached the figure of 8.4 million
tonnes in 1984-85; this amounts to a four-fold increase
over the last ten years.3
(ii) Outlays during the various plans have been largely
financed from internal resources, creating no problems of
debt-servicing similar to those experienced by many deve
loping countries today. Development has largely been
self-sustained.
(iii) A three-fold increase in food-grain production during
the past years has made the country self-sufficient with a
surplus stock, at the end of 1985, amounting to more than
23.73 million tonnes, which is approximately 15.8 per
cent of the annual production. India's dependence on
40
food imports has ended. This has made possible the
creation of a public distribution system and the use of
foodgrains for the generation of employment in the rural
sector.
(iv) A phenomenal increase in banking facilities especially
as regards the quantum, form, and terms of credit to the
agricultural and industrial sectors: deposits have increa
sed from Rs. 1,160 crores in 1956, to Rs. 84,719 crores in
1986; the number of branches of commercial banks has in
creased, in the same period, from 5,078 to 51,385?
The past thirty years have definitely seen more growth than
the previous twenty centuries, or to put it differently, growth of
this magnitude in India is a post-independence phenomenon.
A number of observations support this statement.
(i) In 1950, income per head in India was Rs.466(at 197071 prices ) and this low level of income could hardly have
grown from a much lower base.
(ii) If an average Indian in the year zero A.D., had yearly
enjoyed Rs. 230 worth of goods (at 1970-71 prices),
growth at only 0.5 per cent per year would, by 1950, have
brought him to an amount of Rs. 31 .48 lakhs per year.
(iii) Historical reconstructions suggest that income per
person in India stagnated between 1600 and 1900, and fell
between 1900 and 1950. The colonial "drain", however
it may be measured, during the British rule, must be held
responsible for this stagnation. Mass outflows of revenue
were directed to the metropolis. S. Habib and U. Patnaik
have estimated that during the crucial decade of 17951805, a quarter to a third of the UK's domestic capital
formation was financed by the Indian tribute.5
(iv) The comparison, in terms of area cultivated, yield per
hectare and agricultural output, between the pre and post
independence rates of growth, shows that the exponential
rates increased from 0.4 to 0.8 per cent, from negative to
1 .4 per cent and from 0.4 to 2.6 per cent, respectively?
Though substantial, these rates of growth do not compare
favourably with the performance of other countries. Agricul
tural output in China and India has trebled in twenty-five years,
but China started with much higher levels of output than India.
This means that China's increase in agricultural output has to be
attributed to considerable increases in yield per hectare. Per
capita steel production In India is about 16 kgs, while the world
figure is estimated at 161 kgs in the period 1973-83. A glance
at Table 1 shows that India's industrial growth rate has been
comparatively low.
Recent studies on industrial production in India agree that
after the mid-sixties, there has been a significant decline in the
growth rate.7 At the same time, further increases in agricultural
output will have to depend more and more on raising the yields
of various crops, since further area extension is not feasible.
In short, the growth experienced by the post-Independence
Indian economy has been unparalleled in its own history, but com
pared to the growth of other developing countries, it has been a
slow and painful process. The past thirty years have also wit
nessed the concomitant phenomenon of mass poverty, that is, no
mere pockets in backward areas, but a staggering number of
people living below the poverty line throughout the country.
The poverty line is usually defined in terms of the minimum
number of calories (around 2400) required for a human being to
live and do ordinary work, and hence this measure of poverty
does not include other subsistence requirements like housing,
clothing, health and education. In this sense, it is a real mini
mum.
Estimates of the number of people living below the poverty
line have been presented in Table 2. In spite of the controversy
raised by Dr Sukhatme and others,8 V.M. Dandekar, in a recent
paper, presents two sets of estimates of the rural population
below the poverty line (see Table 3).
An analysis of Tables 2 and 3 shows that:
(i) the number of those below the poverty line has gone on
increasing (at least until the mid-1970s), while the
economy has been growing;
42
(ii) the great majority of those below the poverty line live
in rural areas; in other words, approximately 80 per cent of
the rural population lives below the poverty line. They
are landless agricultural labourers, marginal farmers, rural
craftsmen etc;
(iii) there has been a slow decline in the percentage of
people below the poverty line (from 49.5 to 44.4) probably
due to the effect of the anti-poverty programmes sponsored
by the Government. As some have pointed out, it is very
likely that the situation of those at the bottom has deterio
rated further.
Our attempt so far has been to measure the extent of poverty
following an economic yardstick. This approach is one-sided
and partial because it does not take into consideration social,
political and psychological aspects of poverty. The mass of the
poor in India is largely composed of those who are at the same
time socially discriminated against (Adivasis, Harijans, backward
castes) and have no political voice of their own. Poverty is not
only a state of material deprivation but of human and psycholo
gical as well. When a section of society has been for years
branded, in word and deed, as useless and inferior, the result is
the acceptance of these labels by the poor. When self-respect
and self-esteem break down, when one lives without a future,
when taking risks endangers survival itself, then material depriva
tion results in the dehumanisation of man.
The paradox is that economic growth and mass poverty
seem to thrive in a strange symbiosis, disproving the theory —
defended in the 1960s — that once growth has been achieved,
its effects will automatically trickle down to the poor. Experi
ence has proved beyond doubt that one-third of the population
has not tasted the benefits of growth. Growth is a necessary
but not - by itself - a sufficient condition for the development of
all, it does not lead to a more humane life for the majority of the
population. It is in this sense, that we distinguish between
growth and development. The question remains: why does
this happen ?
43
II.
ENTITLEMENT APPROACH
Amartya Sen has put forward a novel explanation of poverty
on the basis of the entitlement approach. In simple terms, his
theory means that poverty depends on two main factors: first,
a person's ownership bundle, i.e., the number and the type of
assets he possesses, and second, his exchange entitlement, i.e.,
what he can command in the market with these assets. This
depends on the legal, social and political characteristics of society
as well as on his own position in it. To give an example, an
extreme situation of death due to deprivation can occur because
food is available but the person does not own anything with
which food can be bought (exchanged); or a person may have
money to buy food but food is so scarce, its price so high, that it
can hardly be bought with the existing assets. It follows that
to avoid poverty (a situation of acute deprivation), everybody
should have a sufficient number of assets and, the exchange
entitlement should be equivalent to the owned assets.
The thesis presented in this paper is that the socio-economic
structure of India has deliberately kept the ownership bundle of
the majority down to the barest minimum and has made the
exchange entitlement as unfavourable as possible to them.
Poverty, therefore, is a man-made phenomenon, in the sense
that it is the result of the self-interest of the powerful few who
have established a type of socio-economic structure which
necessarily creates and perpetuates poverty.
In what follows we attempt to show the unequal distribu
tion of ownership-bundles (assets) in India and the pattern of
unfavourable exchange entitlements.
Unequal Distribution of Income
An individual's income is a flow of money and services per
period of time and it depends on the quantity and quality of the
assets (stocks) he possesses and utilises. Land gives rise to
income in the form of rent and profits; shares and debentures
generate income in the form of dividend; acquired skills like the
ability to manage an enterprise, to maintain accounts, or to
handle a computer give rise to incomes in the form of salaries.
44
For the great majority in India their physical capacity to do manual
work, their labour-power, generates income in the form of wages.
An unequal distribution of assets over the entire population will
result in unequal income flows accruing to various people. We
start from a brief analysis of income to move into a study of the
asset position.
The share, in personal income, of the bottom 20 per cent
has fallen from 9 to 4 per cent in the rural areas and from 7 to 4
per cent in the urban areas (see Table 4). This is confirmed by
the All India Household Survey of Income conducted by the
NCAER in 1962 and 1967-68. The study shows that the share
of the bottom 50 per cent declined from 22.5 to 19 per cent
while, the share of the top 10 per cent increased from 32.8 to
36.1 per cent of the total disposable income. The Reserve
Bank of India (RBI) figures show that in rural areas the bottom
20 per cent of the population has only 9 per cent of the aggre
gate income, while the top 5 percent has 17 per cent of the total
income. In urban areas, the position is almost the same except
that the few rich are richer and the many poor are poorer as com
pared to the situation of the rural population.
The same picture emerges from a study conducted in
1971-72: 54 per cent of the rural population in the lowest
expenditure class (below Rs. 18 per month) share 30.8 per cent
of the total disposable income, while in the urban areas, 35 per
cent of the population in this expenditure class shares only 13
per cent of the total disposable income.9 This seems to confirm
the trend that income distribution is relatively more unequal in
the urban than in the rural sector.
Income distribution has lately been measured by private
consumption expenditure. Table 5 shows the distribution of
private consumption expenditure from the Sixth Plan draft. The
pattern of unequal income (private consumption) distribution
has not changed significantly. The bottom 40 per cent of the
population in rural areas accounted for 21 .6 per cent of total
private consumption, while the top 10 per cent accounted for
25.6 per cent. The situation is worse in urban areas.
In contrast, I. J. Ahluwalia claims that "the available infor
mation on the distribution of income suggests no evidence of
45
increasing inequalities over time.”'® As U. Patnaik has rightly
pointed out, Ahluwalia’s analysis is based on NSS data regarding
consumption expenditure. It is well known, as Ahluwalia
admits,11 that an analysis of income distribution based on con
sumption expenditure data, is vitiated on two grounds; first, the
data are generally biased because of wide-spread under-reporting
by the higher income groups, and second, because either nominal
consumption is recorded, ignoring the effect of inflation on the
poor, or differential deflators are not used to capture the differen
tial prices faced by various classes. It must also be remembered
that there may be, in general, a tendency in high-income groups
to save more, while low-income groups will probably have to
dissave. This means that "income distribution at any point of
time is bound to be more skewed than consumption distri
bution."12
The problem of income distribution can also be analysed
from a sectoral perspective. The distribution of the per capita
National Domestic Product (NDP), and the number of workers
in the non-agricultural and agricultural sectors is given in Tables
6 and 7.
V. M. Dandekar remarks that in the non-agricultural sector,
the per capita NDP in 1970-71 was already 4.2 times greater
than in the agricultural and that by 1980-81 the gap had widened
to 5.7 times. This unequal distribution of the NDP has clearly
benefited the top 10 per cent of the population at the expense
of the rest. Dandekar relates the behaviour of the per capita
NDP at 1970-71 prices to the disturbing fact that the per capita
consumption of foodgrains has hardly increased over the last
thirty years.
The reason is that in 1954-58 the per capita NDP in the
non-agricultural sector was already around 60 per cent higher
than that in the agricultural sector. Further increases in the
per capita NDP in the non-agricultural sector would not have
resulted in the rise of per capita consumption of foodgrains, as
its demand had already been satisfied. On the other hand,
the per capita NDP in the agricultural sector has remained un
changed and this may explain the inability of the rural poor to
increase the per capita consumption of foodgrains. Dandekar
46
47
concludes that "in spite of increased production of foodgrains
and availability of imports, if needed, the per capita consumption
of foodgrains in the economy has hardly increased over a period
of thirty years. In other words, in spite of an almost 66 per cent
increase in the per capita NDP over thirty-four years, the problem
of hard poverty remains almost untouched. An increase of 66
per cent in the per capita NDP over thirty-four years is not great.
What is worse is that these small gains have remained confined
to a small section of the population, may be 20 per cent, may
be 25 per cent, may be at most 30 per cent. The remaining
70 to 80 per cent of the population has stayed where it was
thirty years ago."13
2. The RBI Debt and Investment Survey showed that the top
one-eighth ranked by asset value, possessed 53.3 per cent
of the total assets. U. Patnaik believes that the top one-eighth
of the rural households account for at least 75 per cent of the
total marketable surplus.14
This analysis confirms our hypothesis that the condition
of mass poverty cannot be explained away in terms of food
scarcity; it must rather be explained by the unequal distribution
of income. We turn now to examine the distribution of assets.
Unequal Distribution of Rural Assets
The distribution of assets in rural India is highly skewed in
favour of a few. The following observations support this
statement.
1 . Operational holdings and operated area-. There has
been a marked increase in the number of small and marginal
(below 2 hectares) operational holdings from 49.63 million
in 1970-71 to 66.6 million in 1980-81. They constituted
74.5 per cent of the total holdings in 1980-81, but operated only
26.3 per cent of the total operated area (see Tables 8 and 9).
On an average, 1 .1 million marginal holdings come into existence
every year. The annual rate of marginalisation has been around
4 per cent, significantly higher than the 1 .9 percent annual rate
of growth of the rural population. The conclusion seems to
be inescapable: marginalisation is definitely due to immiseration.
Holdings above 10 hectares have decreased marginally from 2.77
million in 1970-71, to 2.15 million in 1980-81. They consti
tuted 2.4 per cent of the total holdings in the same year, but
operated as much as 22.8 per cent (37.13 million hectares)
of the total operated area.
3. It has been argued that land reforms have made an impact
on the unequal distribution of rural assets. In a recent contribution
D. Bandyopadhyay points out that as a result of ceiling laws,
2.97 million hectares were declared surplus and only 1.15
million hectares had been distributed by December 1982. Of
the undistributed area, 0.66 million hectares are under litigation
and 0.35 million are reported to be unfit for cultivation. He
suggests that if the ceiling for unirrigated land were to be reduced
to 12 hectares per family, and other minor changes introduced,
the estimate of surplus land would be at least 3 to 4 million
hectares. He adds that during the decade of the seventies,
12.93 million hectares from large holdings "had been consci
ously and wilfully dispersed obviously with a view to avoiding
ceiling laws."15
4. Eviction of tenants and indebtedness have had a very
negative influence on the rural poor. The number of landless
agricultural labourers has increased from 31 million in 1964-65
to 55 million in 1981. The number of those who have only
labour power as an asset has almost doubled.
5. The life of the rural poor depends, to a great extent, on the
use of common property resources (CPR) like village pastures,
forests, wastelands, village ponds etc. A recent study shows
that between 84 and 100 per cent of the poor households gather
fuel, fodder and fibre items from the CPR. During the past
15 years, the size of the CPR has been declining due to physical
loss of resources (submersion of grazing land on account of
dam building), deterioration in productivity and gradual transfer
of such land into private hands. The distribution of part of it
to the poor has resulted in a high rate (63 to 91 per cent) of
transfer to the rich, as well as in its under utilization because
of the lack of complementary inputs from the poor. This phen
omenon has been particularly glaring in tribal areas.
48
49
6. It has been extensively demonstrated that the technical
revolution in agriculture has only succeeded with a few crops,
grown in small pockets and it has largely benefited the rich farmers.
In an interesting study of one of the richest talukas (Borsad)
of Gujarat, B. Singh shows that where the growth of agriculture
is highest the incidence of poverty is most severe.17 Analysing
the impact of technological change on Indian agriculture, C. H.
Hanumantha Rao concludes that "technical changes have contri
buted to widening the disparities in income between different
regions, between small and large farms and between landowners
on the one hand and landless labourers and tenants onthe other."18
He also points out that technical inputs (irrigation, HYV, tractorization, fertilizers) have been intensively used in big holdings,
and though output and employment have increased, the relevant
question is to examine whether a much more extensive utilisation
of the same resources would not have raised output and employ
ment much more. Many today are of the opinion that the answer
is definitely in the affirmative.
on these assets. Even among the owners of industrial assets
the degree of inequality is very high.
7. New cultivating methods necessitate a high dose of invest
ment and this is totally beyond the means of medium and small
farmers. Credit facilities distributed through the Primary Agri
cultural Credit Societies (PACS) benefit large farmers: only
36.3 percent of the small farmers borrowed from PACS, as
compared to 51 .3 per cent of the large farmers.19 The fact that
the small farmers do not avail themselves of credit facilities
affects their economic position negatively in a number of ways.
For example, they are unable to take advantage of the Crop
credit Insurance Scheme since granting of insurance is linked
to the credit distribution that takes place through the PACS.20
Inequality in the Distribution of Industrial Assets
Industrial assets are legal titles to ownership of capital, in
the form of shares, debentures, equities, etc. They entitle their
owners to a flow of income in the form of dividends. If we take
the population of the country as a whole, it is relatively easy to
see that the ownership of these assets is heavily concentrated in
a very small urban section. The bulk of the rural masses, the
salaried employees and the industrial work-force have no claim
Several studies in the past have stressed the concentration
of industrial power in the hands of a few. R. K. Nigam and
N. C. Chowdhury, in their study entitled "Corporate Sector in
India", revealed that in 1957-58, 88 per cent of the jointstock
companies in the private sector had a paid-up capital of less
than Rs. 5 lakhs and they accounted for only 15 per cent of the
total paid-up capital. As against this, 0.4 per cent of the total
number of companies at the top of the pyramid, accounted for 34
per cent of the total paid-up capital, and one per cent of the com
panies accounted for 47 per cent of the total paid-up capital.
The study conducted by the RBI in 1975 for 1650 non-financial,
non-public limited companies, relating to the year 1973-74,
shows that 22.4 per cent of the companies at the top of the
pyramid with a paid-up capital of Rs. 1 crore and above, accoun
ted for 74.8 per cent of the total paid-up capital of all companies.
These companies also accounted for 65.4 per cent of the value
of production, and 71 .5 per cent of the gross profits of all com
panies.21 It is worth mentioning that ten years earlier, they had
accounted for only one-third of the total paid-up capital.
The Annual Survey of Industries (1981-82) shows that
the highest group by size of capital (Rs. 25 lakhs and above)
included only 5.3 per cent of the total number of factories, but
they commanded 92.5 per cent of the total fixed capital, they
manufactured 73.6 per cent of the total gross output and they
contributed 81 .2 per cent of the total net value added.
The growth rate of the twenty big houses in India shows
that during the 1963-64 to 1966-67 period, the annual growth
rate in assets was of the order of 21 .6 per cent; it declined in the
next five-year period (1966-67 to 1972-73) to 6.6 per cent.
due mainly to the anti-monopoly legislation passed by the
Government and to the hesitation on the part of the Ministry
officials in granting licences. A reversal in the licensing policy
after 1972 resulted in a new spurt of growth; in the nextthreeyear period (1972-73 to 1975-76), the growth rate increased
to 12.5 per cent.22
50
51
It must be emphasised that this concentration of power has
been achieved with the help of the public sector financial insti
tutions. The Dutt Committee on Industrial Licensing found
that about 56 per cent of the total assistance provided by finan
cial institutions (IFCI, IDBI etc) had gone to the large industrial
houses. Birla obtained 25 per cent of the total assistance
granted to 20 large houses; 70 per cent of LIC term-loans and 62
per cent of the State Bank of India loans went to the large indus
trial houses. It is also important to remember that the family
control of these big houses requires a minimum ownership of
total assets. For example, ownership of 0.36 per cent of the
total assets enables the Tata family to control more than 1,600
crores. In the case of Birla, the family owns 0.18 per cent of
the total assets and controls about Rs. 1,500 cores.23 All this
indicates that government and financial institutions provide the
bulk of share capital; therefore these families can exercise control
over the affairs of the companies without much risk.
across the board.25 There is a general consensus that indus
trial productivity has experienced a serious setback. In other
words, public funds in the form of financial assistance to the
powerful few have not even been productively utilised.
An understanding of the structural nature of poverty
requires that the unequal distribution of industrial assets be seen
in the light of the following two facts:
(i) Critics of the maze of bureaucratic procedures which
seem to be the greatest impediment to the growth of the indus
trial sector in India, should remember that the general public has
contributed considerably to their formation and growth, because
of the extensive financial help given to the private sector by the
State-owned financial institutions and banks. We must re
member that around 70 per cent of the tax revenue received by
the Government comes from indirect taxes which affect the
common man. If one considers also that real interest rates
charged on these loans have been quite low, one is led to the
conclusion that the private industrial sector has been able to
syphon off a considerable amount of resources from the most
vulnerable sections of society.
(ii) Growth rates in the industrial sector have experienced
a significant decline after the mid-sixties and this phenomenon
has become the object of debate recently.24 The stagnation of
the industrial sector is confirmed by the rising trend in the capital
output ratios not only in the manufacturing sector but also
Inequality of Human Capital
The term 'human capital' refers to the quality of human
labour-power, to the acquired skills of a person and his state of
health. It is primarily determined by the availability of health
facilities, and educational opportunities. Present data tend to
suggest that the distribution of these assets is highly unequal in
India, resulting in an unequal distribution of human capital assets.
As regards health. Table 10 provides a comparison between
India and other developing countries. Infant mortality in India
is more than double that in China. This is due to the com
paratively high incidence of death among children between one
and four years of age. Life-expectancy for men is higher than
for women. This contrasts with the normal trend in more healthy
environments where women have a higher life-expectancy than
men. This limited analysis shows that health conditions in India
are still far from satisfactory and that they are specially unfavour
able to women and children. It is reasonable to assume that the
mass of the rural poor are the worst affected.
Table 11 confirms the existence of a glaring discrimination,
in the distribution of heatlh services, between urban and rural
areas Eighty per cent of the total number of doctors practise in
urban areas, while only 20 per cent are established in rural areas.
Of the total number of hospitals, 85 per cent are established in
cities, and only 15 per cent are located in rural areas. Nutri
tional Standards reveal a paradoxical picture (see Table 12).
Given the present production levels of foodgrains, vegetable, fat
and sugar, Dandekar calculates that the per capita availability
of calories is 2054 per day and this is only 10.69 per cent short
of the recommended requirement of 2300 calories per day. He
adds, "we may therefore say that in 1985 India's population had
a diet which, on an average, was adequate in respect of calories.
But in view of the known inequality in the distribution of pur
chasing power in the population, the conclusion is inescapable^
Com yi 'i’it -O
05635
y
52
that at least half the population lives on diet inadequate even in
respect of calories. There is enough evidence to show that a
large majority of those with consumption of calories lower than
the recommended, have purchasing power lower than the
average. Hence, the low consumption of calories in their case
cannot be attributed to what is called inter-personal variation in
calorie requirements. The low calorie consumption in most
cases must be regarded as a consequence of low purchasing
power and hence a sign of poverty"26
On the basis of NSS data for 1971-72, Dr. V.K.R.V. Rao
shows that "first, the nutritional inequality as between the
bottom one-third and the top one-third of the population is much
higher in the case of the rural sector as compared to the urban
sector in respect of total calorie intake and also the calorie intake
from each of five food groups; second, nutritional inequality is
much higher in the case of both the rural and other urban sectors
in respect of items other than cereals and pulses reflecting the
effect of purchasing power inequality on the consumption of
those other food items."27
The availability of safe drinking water in the rural areas is
highly inadequate; only one out of every ten villages has safe
drinking water.28
Educational standards in the country, as measured by the number
of illiterate, are still very low. According to the 1981 census,
63.8 per cent of the population is still illiterate. For women the
situation is worse, since 75.2 per cent are illiterate. Reasons
for the dismal performance of the educational policies in the
country are not difficult to find.
(i) During the period 1950-65, the annual rate of growth
of primary schools was 5.76, of middle schools 30.46 and of
higher education 40. As late as 1975-78, the rate of growth
of higher education was three times greater than that of primary
education. In spite of policy statements to the contrary, the
educational approach has had a clear urban bias.
(ii) Total Government expenditure on education as a
percentage of GNP has been very low. It increased very margi
nally from 1 .3 per cent in 1950-51 to 3.4 per cent in 1984-85.
53
An international comparison of Government expenditure, as
percentage of the GNP, in 1982 between various countries, shows
that India spent 1.9 per cent of the GNP on education and 20.2
per cent on defence; Mexico spent 13.1 per cent on education
and 1.6 per cent on defence; the figures for Sri Lanka were 7.4
and 1 .4 per cent, for Egypt 9.2 and 12.7 per cent and for Brazil
4.6 and 4.3 per cent respectively.29
(iii) The allocation of even these meagre resources has
been biased, against the poorer sections. The per capita
annual expenditure on primary education during the Sixth Plan,
works out to Rs. 0.33 on primary education, Rs. 0.50 on secondary
and Rs. 3.00 on higher non-technical education. To understand
the implication of this bias against primary education, we must
analyse the next problem.
(iv) The most acute problem affecting the availability of
educational services to the poor is the high drop-out rate at the
primary and secondary levels of education. This is due to eco
nomic reasons, to the almost non-existence of proper buildings
and facilities and to the dismal performance of teachers in rural
areas. At the end of 1977-78, about 20 per cent of the children
who had reached school-going age did not enter schools at all;
and of those (80 per cent) wo did enter nearly half dropped out
by class V and nearly three-fourths dropped out by class VII;
only about 15 per cent reached class XII, and less than one per
cent got the first degree. The main beneficiaries of the entire
educational system are just 1 per cent of the total population
of children in the country.
It is in the light of these figures that one must question
the difference in expenditure between primary and higher non
technical education discussed earlier. The failure of the educa
tional strategy to produce a more egalitarian distribution of human
capital assets is confirmed by the simultaneous growth of specia
lised institutions of learning which have attained international
standards.
Professional and technical institutions in India
have turned out one of the largest supplies of technical manpower
in the world. Unfortunately, the entrance to these institutions
remains mainly confined, in spite of the reservation policy, to
those students from the high-income groups who can afford
54
55
to pay exhorbitant tuition-fees for extra classes at the crucial
examination levels (Standards X and XII).
the picture. The other side is equally dismal because their ex
change-entitlements, the number of goods they can get in ex
change for their assets, is being constantly reduced.
Inequality in Employment Opportunities
For the millions in India who possess labour-power, that is,
the capacity to work as the only asset which can be exchanged
for food and other necessities, total or partial unemployment
is one of the main causes of poverty. The distribution of
work opportunities is unequal between urban and rural areas,
between man and woman, between the organised and unorga
nised sectors and finally between the affluent and middle-class
groups of city-dwellers.
Unemployment figures are very unreliable. The Draft
of the Sixth Five Year Plan points out that the "most inclusive
and significant indicator of the magnitude of unemployment"
is the calculation of unemployed days rather than unemplo
yed persons.30 This person-day unemployment has been
placed at 130 million days per week in 1973 which is equivalent
to 18.4 million persons being unemployed on a typical day.
For 1978 the estimate works out to 19.5 million, and for 1983,
to 21 .8 million.
In both rural and urban areas, the most affected are the
poorest households. The same Draft remarks that in 1972-73,
households with a per capita monthly expenditure of less than
Rs. 11 (i.e., well below the poverty line) had unemployment
rates of about 22 per cent and 29 per cent in rural and urban
areas, respectively; the rate for households with a per capita
monthly expenditure of Rs. 100 and above, was 2 percent in
rural and 5 per cent in urban India. The planners conclude:
"thus for the poor, the curse of unemployment and the curse
of poverty coexist though unemployment is not the only cause
of their poverty"31.
Unfavourable Exchange Entitlement for the Poor
There are about 250 million people in India who live below
the poverty line because they do not possess a sufficient number
of assets. Tragic as this fact is, it only represents one side of
The majority of the poor in India still depend on agriculture.
The terms of trade between agriculture and industry can be taken
roughly as one of the indicators to measure the erosion of the
exchange entitlement of the poor.
An anlysis of Table 13 reveals the following:
(i) The ratio of output prices to input prices has been
declining: output prices have failed to rise as fast as input prices.
The agricultural sector has to pay more for the inputs required
than it gets by selling its output.
(ii) The crude barter terms of trade, defined as the ratio of
agricultural commodity prices to manufacture prices has been
steadily declining after 1974-75.
(iii) U. Patnaik shows that up to the end of the 1970s, the
trend in the terms of trade (taking 1960-61 as the base year) was
in favour of agriculture. This may explain the investments in
agriculture which took place during those years and also the
flow of funds from the urban to the rural sector. When the base
is taken as 1970-71, the trend declines by about 13 per cent for
1975-78, and it has remained more or less the same subse
quently.32
(iv) In spite of the bias against agriculture inherent in the
Laspayere Index used by Rath, a recent study on his work, in
which the terms of trade are analysed with reference to particular
crops in each state, concludes that "the terms of trade between
product—supply and input—demand, (taking 1970-71 = 100)
reveal that from 1970-71 to 1980-81, the terms of trade have
moved against the cultivators in 14 states for ten crops." If we
exclude sugarcane and a few other crops, we must conclude that
there is "overwhelming evidence of the deterioration in the terms
of trade against the cultivators during the 70s."33
(v) Taking into consideration the annual growth rates of
marketed surplus, and the annual deterioration in the terms of
56
57
trade, "the compound average annual loss to farmers is 5.92
per cent ... This loss is transferred to the non-agricultural sector
through the price mechanism ... it is a disguised effective tax
rate."34 One could argue that the unfavourable terms of trade
affect mainly the rich cultivators. Though there is a lot of truth
in this statement, it is difficult to see how in such a situation the
exchange entitlement of small and poor farmers producing mainly
for self consumption, as well as that of the mass of landless
labourers, could have improved.
The price index for food (see Table 14) has increased almost
five times .in twenty years and money wages received by agri
cultural labourers in India have certainly not risen proportion
ately. U. Patnaik contends that while "money wages have
risen during 1963-64 to 1974-75, when adjusted by the con
sumer price index for agricultural labourers, real wages are seen
to decline drastically in the case of female workers. Since days
employed annually have also declined, the real earnings have
fallen even more than have real wage-rates."38 This situation,
she also argues, is reflected in the sharp rise that has taken place
in real indebtedness.
(vi) It is not only that the assets of the poor are insigni
ficant, but even when they are deposited in the rural branches of
scheduled banks, they are not matched by a corresponding flow
of credit. On the basis of Rs. 100 deposited in rural areas, only
Rs. 55.1 is granted as credit, while in urban areas, the proportion
is 100:74:6.35 These figures indicate the extent to which the
meagre assets of rural India are transferred to urban centres.
The exchange entitlement of the poor has been affected
directly by the combination of inflation, indirect taxation, and by
the lack of effective implementation of minimum-wage laws for
agricultural labourers in most areas of the country. Because of
these reasons the real wage rate has fallen.
Money wages of agricultural labourers have been prover
bially low, and implementation of laws regarding minimum wages
has been
disregarded almost everywhere.
India Today
(30-3-1988) carried out a survey regarding the implementation
of the Minimum Wages Act by State ministers and Congress(l)
leaders in Madhya Pradesh, and discovered that they pay less
than the Act stipulates and that most workers are not even aware
of the existence of these laws.
Inflation rates in India have not been, by and large, exces
sively high: 7.7 per cent from 1973 to 1983.36 Nevertheless,
the purchasing power of the rupee has declined from 96.2 paise
in 1961 to 15.7 paise in March 1986.37 This erosion in pur
chasing power cannot be effectively absorbed by the majority of
the poor who do not benefit from in-built mechanisms (like DA)
applying only to the salaried classes.
Conclusion
Following A. Sen's conceptual understanding of poverty
it is easy to perceive its nature and magnitude in India. Thehighly unequal distribution of all types of assets, including human
capital assets and the lack of opportunity to use labour-power
leaves at least one-third of the population without effective
means of satisfying their basic needs. At the same time, the
scarce assets of the poor encounter systematically unfavourable
conditions for exchange. The net result is that a large number
of people live just above or below the poverty line and a few keep
on accumulating assets. It is in this sense that poverty is a man
made phenomenon.
This understanding of poverty has an "economic" bias.
This must be clearly stated, because poverty is neither a mere
economic situation of deprivation nor a question of numbers and
percentages.
The economic aspect must be seen as an
important but partial component of a process taking place in in
dividuals. This process is one of dehumanization. The mass
of the poor are not only economically disadvantaged but a large
section of them is also socially discriminated against. Tribals,
Harijans and other members of the backward castes form a great
proportion of those living around or below the poverty line. The
combined effect of socio-economic deprivation is severe on the
kind of life they are able to lead. Fatalism, a poor self-image
low risk-taking capacity, fomented divisions and rivalries along
ethnic, religious and caste lines... all these cannot but have a
profound psychological impact. This is the total effect that
58
59
must be understood, if we are to look at the phenomenon of
poverty as a process which makes people more dependent, more
voiceless, more exploited and ultimately incapable of controlling
their lives and destinies.
This brief description of economic life is clearly too simple
to capture all the complexities involved in the reproduction of
social life. At the same time, it gives us a fairly clear picture of
the working of the economy because it underlines the crucial role
played by the economic surplus in determining the pace and
pattern of growth. The latter does not only depend on the quan
titative and qualitative nature of the surplus, but also on the
manner of its disposal.
The picture we have drawn of poverty as a process, under
lines its structural nature. Given the present set of socio-eco
nomic relations among sections of society, poverty tends to per
petuate itself, to be created anew. In other words, the socio
economic structure is such that the conditions for the reproduc
tion of poverty are inherent in the system. We turn now to
analyse this phenomenon.
111.
REPRODUCING THE CONDITIONS OF POVERTY
The Importance of the Surplus:
The reproduction of any society depends primarily on the
set of productive activities which maintain its life. Any produc
tive activity can be defined as a process in which inputs are trans
formed into outputs (material aspect), simultaneously genera
ting an understanding of itself among those persons participating
in it (conceptual aspect). Productive practice, therefore, com
prises both physical and mental activity; it is a combination of
content with meaning, of physical output and conceptual under
standing. For the time being, we deal only with the material
aspect of productive practices.39
Material production can be fruitfully conceived as a circular
process in which inputs are transformed into outputs, which in
turn are used as new inputs for a new productive activity. The
continuation and maintenance of a productive activity in time'
require that the output be large enough to replace the used in
puts, so that at least the same activity can be undertaken again
on the same scale. Economic surplus can be defined as the
output net off all goods and services necessary for the replace
ment and maintenance of all inputs including labour. If the
economic surplus or net output is nil, the productive activity or
set of all productive activities will merely reproduce itself time
after time. When the net output is positive, when an economic
surplus is generated, the conditions for growth and development
have been established.
Relations between Production and Distribution
In real life the process of producing the surplus is almost
indistinguishable from the process of its distribution among
various sections of the population. Text-books in economics
separate these two processes and an attempt is made to explain
the appropriation of the surplus by the same laws that are sup
posed to regulate production. Traditional theory maintains
that the distribution of the surplus generally depends on the pro
ductivity of factors of production, that is, on their effective con
tribution to the process of production. This latter interpretation
cannot be defended even as correct economic analysis.40 From
a holistic point of view, it is unfair to analyse the distribution of
the surplus as if it were a purely technical problem. In what
follows, we propose to analyse the processes of production and
distribution of the surplus.
Production is not only a technical activity transforming in
puts into outputs. It is a fundamental activity always undertaken
within a specific set of relations existing among the groups of
persons engaged in it. This set of relations is ultimately defined
in terms of people's positions in the structure of power that con
trols the process of production. The same set of relations, con
ditions the nature of the production process and is also the result
of that process. Productive inputs and outputs necessarily in
clude social components which are not generally mentioned in
most economic text-books. To put it differently, a productive
activity not only reproduces the economic life of society, on the
same, enlarged or diminished scale, but also a set of power rela
tions (domination and subordination) under which the persons
involved in this process operate.
60
61
We have mentioned before, that a productive practice is a
meaningful physical activity. The understanding of the activity
grows from the way in which it is actually performed and, at the
same time, this understanding sets the boundaries which limit
the development of this activity in the future. It is, therefore,
important that we devote some time to analysing the way in
which concepts describing production are generally used. One
can speak of production in terms of inputs or go beyond that and
call them capital and labour. The consequences and implica
tions of this terminological shift (from inputs to capital and
labour) must be clearly realised. The use of the term "capital"
presupposes the deliberate attempt to understand it only as
means of production (tools, implements, machinery), while in
fact the concept of "capital" also includes specific social relations
of power and control between those who own the means of
production and those who possess only their capacity to work.
In the same way, the term "labour" seems to refer only to the
expenditure of human energy during the process of production,
hiding the fact that it also refers to a specific type of labour,
namely wage-labour. The latter must be understood as a con
tractual relationship between the owners of capital and the own
ers of labour-power. It is natural, therefore, that the concomi
tant understanding of productive practices be influenced by
the positions of the different actors in the process of production.
Those involved in productive practices will conceptualise them
only in accordance with the relative positions they occupy in
the structure of power that controls production.
The distribution of the economic surplus among various
groups or classes does not depend on their contribution to the
productive process, but on the effective ownership of productive
assets. Effective control of the shares in a company or of a
piece of land, entitles individuals and groups to a share of the
surplus in the form of dividends, rent or profits. Assuming, for
the sake of simplicity, a socially determined basket of commodi
ties necessary for the maintenance of all those engaged in pro
ductive activities, any output over and above this basket, accrues
to people because of their special claims on the surplus, based
on their ownership bundle.
The analysis of production and distribution leads to the
following conclusions:
(i) The amount and quality of the ownership-bundle
determines the relative share to be appropriated from the surplus.
Since the future pattern of growth for the whole society depends
on those who can dispose of the surplus, the former is in the
hands of the few who have been able to accumulate assets in
the past.
When we understand production in this way, we are led
also to the analysis of the unequal power structure in which it
takes place. The objective position within the power structure
is determined by the effective ownership of assets (ownership
bundle) and more particularly by ownership of and control over
the means of production. In short, effective control of productive
assets results necessarily in control over the whole production
process. 'This control enables a few people to determine finally
what is to be produced (the final product-mix) and how (techn
ology) the combination of inputs and outputs will be effected.
Production, therefore, is determined by the prior distribution of
productive assets in society.
(ii) Production and distribution are two sides of the same
coin. Concentration of economic power through a dispropor
tionate appropriation of the surplus leads to effective control
over productive activities, while the pattern of production tends
to reinforce the unequal distribution of assets. In a market
economy, the type and quantity of commodities to be produced
as well as the technology to be used in their production, appears
to be determined by commodity and factor prices-by the market.
On the surface, this is true, but at a deeper level of reflection and
analysis, market demand for commodities and factors is not
independent of the way in which the surplus has been distributed.
Clarification of this point requires an explanation regarding the
disposal of the surplus.
The economic surplus can be used for consumption (more
precisely extra-consumption or luxury consumption) or for
investment. In the former case, it provides extra purchasing
power to those who appropriate it. In a market economy, the
demand for luxury goods will increase and market researches will
immediately sense this opportunity and attempt to direct produc
tive potential into these areas. Since the market operates, at least
62
63
in theory, on the democratic principle that only votes (rupees)
count, those who can control a sizeable quantity of votes (rupees)
will finally get what they want. Notice how the same vicious
circle operates: control over the surplus creates effective demand
for luxury goods, and this, in turn, attracts new investments in
the sectors producing these goods. It is no exaggeration to
say that the pattern of demand influenced by income and asset
distribution will determine the sectors into which investment
will flow.
We can also look at the disposal of the surplus from the
point of view of investment. First, the preferences of the upper
income groups will determine the division of the surplus between
present consumption and savings. Second, expected returns
from various productive sectors, will influence the areas in which
part of the savings are to be invested, and the type of technology
to be used. The contention of this paper is that the expected
returns on investment have been higher in the capital-intensive
sectors; hence, investible funds have been channelled into these
sectors. A detailed analysis of the factors contributing to this
process falls outside the scope of this paper. Suffice it to say
that many industrial houses have obtained relatively cheap and
abundant credit from the financial institutions; indeed, when
the nominal rates of interest are adjusted for inflation, it turns
out that the real rates of interest paid by large investors, have been
very low. This has favoured investment in the more capitalintensive sectors and in the more capital-intensive methods
of production. As a consequence, expansion of employment
in this sector becomes very costly since it requires more refined
skills and a much greater capital per man ratio. Wages for
those who find employment in this sector have risen, thereby
depressing the demand for labour further.
controlled by those who have access to liquid assets; in many
cases these persons are different from the owners of means
of production. In the rural sector, trading practices are generally
linked to money-lending and productive practices.
Though productive practices constitute the basis of society,
they require another set of practices to hold them together: such
practices may be called integrative. Thye are of two kinds:
exchange and political practices.
Exchange and trading practices are conducted by a sizeable
proportion of the population in India. They are responsible
for bringing buyers and sellers together. The interaction between
them takes place through a network of interconnected markets
At a higher level, the totality of economic practices is held
together by those engaged in 'political' practices; administrative,
judicial and legislative. Society is always characterised by
groups possessing varied and, at times contradictory interests,
which generally correspond to the various practices. It is the
responsibility of 'political' practices to achieve the smooth inte
gration of all productive practices. Political activity proper —
the legislative and executive branches — decides the ultimate
common goal of society so that all its members may act together,
through various practices, towards it. Ultimate power and
authority always rest with those engaged in the sub-set of poli
tical practices, which we have called proper. The judiciary acts
as mediator of conflicts and the administration (bureaucracy)
implements the goals decided by the political authority.
The main characteristic of those engaged in political prac
tices is the availability of coercive power (police, army) to
ensure the execution of its decisions. This end may also be
achieved through the mass communication media.
The goal of the integrative practices is to hold the whole
structure together; hence they are a vital source of power. If
control over the surplus results in the power to regulate economic
growth as well as to recreate the power relations existing in
the economic sphere, control over integrative practices ensures
the maintenance of these power relations over the whole of
society. It is easy to see why political power is avidly sought.
Structure of Poverty Reproduction
We are now in a position to explain the existence of poverty
in India as a structural phenomenon. Figure 1 serves as a brief
summary of the argument which will be developed presently.
As we have shown, poverty is the result of the unequal
distribution of ownership bundles as well as exchange
entitlement biased in favour of the rich. This unjust structure
INTEGRATIVE PRACTICES
65
necessarily results in the control of productive practices by a few
who, on the basis of their ownership titles, can, at the same
time, appropriate large portions of the economic surplus; in
short, economic power is concentrated in the hands of those
controlling the surplus.
As we move down, to analyse the mechanism through
which the surplus is disposed of, we notice that the demand
for luxury consumption increases and this in turn, calls for invest
ment in the sector producing luxury goods. This tendency
reinforces and complements investment in the capital-intensive
sector.
Demand factors are not the only ones responsible
for this gradual shift from labour-to capital-intensive methods
of production.
A number of effects follow from this diversion of invest
ment resources to the capital-intensive sector:
(i) The productivity of the great majority decreases since
the ratio of investment per capita falls alarmingly in the rural
and informal sectors.
(ii) The income of these people may fail to increase pro
portionately or fall, and hence effective demand becomes narrow
based. It tends to be concentrated in those who can spend on
luxuries and reduces the capacity of the majority to purchase
the available surplus of food-grains.
i
(iii) Because of the labour-displacing nature of these pro
ductive activities, employment opportunities may tend to fall; at
the same time, the gap between new employment opportunities
generated in the high-investment sector and the rising supply
of labour (due to demographic pressures) increases, with the
result that there is a low expansion of employment opportunities.
In India, black money forms an important source of surplus
utilisation. An official estimate calculates that, for 1983-84,
the total amount of black income generated was around 30 per
cent of the Gross National Product (GNP) calculated at factor
costs, ora sum of Rs. 36,786 crores.41 This large flow of income
is used to increase luxury consumption and to speculate in urban
67
66
property. Black money also plays an important role in esta
blishing smooth linkages between integrative practices and
productive activities. For instance, it facilitates the cheap and
easy availability of finance, the procurement of licences, access
to cheap raw-materials, and these may be used to pressurise the
authorities so that infrastructure facilities like electricity, water,
and transportation are made available quickly on required sites.
In this way, black money influences integrative pratices to
facilitate the type of investment required by those controlling
the surplus.
At the end of the whole process, the unequal distribution
of resources is maintained or hardly altered. As the circular
flow of economic activity is completed, the conditions of poverty
are created anew.
The right-hand side of the figure describes the role that
integrative practices p|ay in keeping the system going. Trading
practices ensure that the market mechanism allocates resources
in a manner which does not alter economic power relations. It
is true that, at times, control over these practices may create
semi-independent power centres in various parts of rural
India. In the long run, however, they must be linked to centres
of productive power to be able to survive.
Political practices tend, in general, to sanction the legitimacy
of this process; but this does not exhaust the role that the State
plays as the ultimate controller of these practices. It is to this
new aspect of the State that we must now turn our attention.
IV.
STRUGGLES FOR POWER
Control over the whole gamut of integrative practices rests
ultimately with the State. Some theories of the State describe
it as a mere tool in the hands of those who control the economic
surplus. This explanation is too simplistic and does not take
into consideration the active role of the Indian State and the
differences among the various groups that control the surplus.
This final section deals with these problems.
Pivotal Role of the State
In post-independence India, the State was conceived as
the guardian of a set of integrative practices which were designed
to usher in a socialist society. The State also took on the res
ponsibility of planning all productive activities and itself engaged
in a number of productive practices which were supposed to
form the core of economic development. Appropriate fiscal
and monetary policies were to help equalize the disposal of the
surplus; thus, the State would act as a powerful force to redress
the imbalance inherent in the socio-economic structure.
On the one hand, there is today a general feeling that the
importance of State planning has declined. On the other hand,
the role of the State as an economic agent engaged in productive
practices, has increased. The State owns 60 per cent of all
productive capital, it runs 8 of the top 10 industrial units, it dir
ectly employs two-thirds of all the workers in the organised
sector, it holds—through nationalised financial institutions —
more than 25 per cent of paid up capital in the joint-stock sector
and regulates private investment through a complex bureau
cratic machinery. Besides its control over productive practi
ces, the State regulates the allocation of large amounts of credit
to various sectors, it has the monopoly of railway and air-trans
port, intervenes decisively in trading practices through a number
of regulatory devices (price and quantity controls), it has set
up an extensive distribution system for essential commodities
and controls foreign trade in commodities and capital.
The increasing involvement of the state in productive
practices has resulted in a greater proportion of the surplus being
directly appropriated by it. This might have helped to offset
the inherent inequalities of the system and to allocate resources
in such a way that a massive assault on poverty could be mounted.
This possibility has not been realised. The reasons for this
failure must be analysed in some detail.
The Main Competing Groups
Many political economists feel that the theoretical mode
based on a two-class (dominant and dominated) analysis of
68
Indian society, has become irrelevant because it assumes a homo
geneity and commonality of interests in the ruling class. The
real nature of the dominant class in India is quite the opposite.
P. Bardhan has argued that this class is composed of the two
top deciles of the population, and characterised by the existence
of competing interests and rivalries among its members.
According to him, these groups can be classified in three cate
gories: the industrial class, the rich farmers and the professional
class.
The industrial class comprises some of the top business
families and a number of entrepreneurs who are in the process
of being co-opted because they share a similar perception of
interests. As a class they are not a post-independence pheno
menon; they had already established their claims during the
Independence movement. During the last thirty years they have
received generous support from the State in a number of ways:
credit facilities provided by the State-owned financial institu
tions, a developed infrastructure of essential services created
by the State, and most important of all, the assurance of a shel
tered domestic market. Commenting on the Government's
latest policy of nationalising "sick mills". Bardhan remarks that
"the Government has acted as a risk-absorber of the last resort
and as a charitable hospital where the private sector can dump
its sick units."42 Other recent developments need to be men
tioned:
(i) Unlike many other developing countries, the linkages
of this class in India with foreign capital are minimal at the level
of equity participation. In recent years the number of technical
collaborations with foreign firms has been increasing—an indi
cation that this route of penetration of foreign capital into the
industrial sector is becoming more popular.
(ii) The growth of the small-scale sector has been poor.
Many enterprises are begun only to take advantage of govern
ment subsidies and others are linked to the major houses
through sub-contracting.
(iii) Caste alliances have played an important part in the
development of this class in the past; at present, these alliances
69
help articulate and link the
predominantly mercantile
sections of this class with the big industrial houses.
The composition of the class of rich farmers has undergone
a profound change after Independence. In most states, the
traditional zamindars and absentee landlords have been replaced
by a group of enterprising middle castes, like the Patels in Gujarat.
During the last thirty years they have received a number of benefits
from the government: price support, laxity in the implementa
tion of tenancy reforms, subsidised inputs like power, seeds and
fertilizers and institutional credit through their control over the
Primary Credit Societies and cooperatives. At the national level,
they still lack organisational power, but they exercise a great
measure of control over the political machinery at the level of the
legislative assemblies. Bardhan notes a few more characteris
tics of this class:
(i) in the economic sphere, they have managed to diver
sify their investment into trading, money-lending and small
processing units in an attempt to minimise the risks involved in
farming;
(ii) caste plays a paradoxical role: it strengthens the
cohesion of the class at the micro-level, but impedes other alli
ances laterally and vertically;
(iii) in the past, this class has been quite successful in
securing the help of many small cultivators and marginal farmers
whose interests it does not really represent.
The class of professionals includes the bureaucracy (civil
and military), professional groups like doctors, lawyers and edu
cators and many salaried employees. Bardhan's enumeration of
the members of this class is unsatisfactory. All the same, for
our present purpose it may be sufficient to stress the common
factors that give a semblance of cohesion to this class. Its
members share one common characteristic: they have been
the main beneficiaries of the present educational system and for
some middle castes education has been the most important
vehicle of social mobility. Their consciousness as a group has
increased in recent years. They have felt the urgent need to
70
71
protect their monopoly over educational services from the on
slaught of all kinds of reservations.
granting subsidies to the agricultural sector and at the same
time the situation of the poor must not be allowed to show rapid
deterioration.
It must satisfy the demands of the bureaucracy
by frequently granting salary-hikes, which increases non-plan
expenditure.
Conflicts at the top and the future role of the State
As we have stated above, the most important characteristic
of the class controlling the disposal of the surplus is its lack of
homogeneity and the existence of conflicts within it. Let us
briefly examine some of them.
The class of rich farmers represents, to a certain extent, the
interests of rural agricultural India. The most vocal elements
from this class believe that the industrial sector has been the main
beneficiary of the development process at the expense of the
farming community. We have earlier presented some evidence
to suggest that there has been, in the last decade, a deterioration
in the terms of trade, against the agricultural sector. On the
other hand, the more articulate voices of the industrial class argue
that rich farmers have benefited from a number of subsidies and
credit facilities and have contributed very little to Government
revenue because they have been exempted from direct taxation.
The class of professionals represented mainly by the civil
and military bureaucracy has opposed the liberalisation of con-,
trols in the private sector. The conflict between the bureaucracy
and the industrial class is often expressed as a controversy regard
ing the efficiency of the public sector v/s-a-v/s the private sector.
A dismantling of all regulatory mechanisms would definitely cur
tail the extensive power of the bureaucracy. New avenues for
the expansion of the State's productive involvement always
appear as a guarantee that the job opportunities of the present
bureaucracy will not be curtailed but expanded. Various
methods are being used by the bureaucracy today to divide the
corporate sector: some efforts operate from the central admini
stration, some operate from within industrial units which are par
tially controlled by the capital in the hands of the State.
The net result of these conflicting interests within the
dominant class can be briefly summed up as follows:
(i) The State seems to be caught in a sort of prisoner's
dilemma: it must promote industrial growth and keep on
(ii) The State, in the past, has had, to take up an accom
modating and compromising stand which is manifested in the
lack of clear and bold policy directives, in the inability to imple
ment fully those policy prescriptions which were clear, and finally
in the adoption of the "subsidy"- or "concession-policy" to
tackle daily problems; this has led in the past to ad-hocism, that
is, to an approach which constantly sacrifices long-term for short
term goals.
(iii) This approach has seriously affected the rate of
growth of agriculture and industry. We have said before that the
present growth rate in agriculture seems to have reached a thre
shold and that the industrial sector, has been in the grip of a fatal
stagnation. In other words, even from a pure growth perspec
tive, this situation is untenable. Under these circumstances
the economy will never realise the potential rate of growth or
which it is capable. With the help of the game theory, R. Sat
shows how in a situation of conflicting interests among the
member groups of the dominant class, the equilibrium position
which is achieved by the system always represents a lower than
optimal growth rate. This presupposes that coalitions among
the contending partners are not possible.43
This scenario provides us with some clues about the future
role of the State. First, there is a movement towards a greater
centralisation and control over integrative practices. Second,
there is a marked change in the nature and role of the State:
from arbitrator of conflicting interests to corporate State designed
on the lines of a transnational corporation, relying much more on
market forces. Third, there is a greater stress on the power of
the State v/s-i-vZs the power of intermediate groups and indi
viduals. This power relies heavily on technology rather than
on economic considerations and on military rather than politica:
means.
72
73
Rajni Kothari has commented extensively on these
changes.44 His analysis presupposes a greater integration
within the dominant class—what he calls the 'classes' (upper
and middle) — than present trends seem to justify. Accommo
dative politics is not yet dead.45
TABLE 2:
Year
1960-61
TABLE 1:
1967-70
Industrial Growth Rates (%)
1.
India
4.3
2.
Bangladesh
8.1
3.
China
8.4
4.
Egypt
10.6
5.
Indonesia
8.6
6.
Republic of Korea
11.2
7.
Pakistan
7.2
8.
Sri Lanka
4.8
Source:
World Development Report, 1985.
1977-78
1980-85
Estimates of Poverty
Author
Rural
180
P. D. Ohja
P. K. Bardhan
Dandekar & Rath (43.2%)
Urban
Total
24
204
(28.8%)
(42.5%)
49
270.5
P. K. Bardhan
B. S. Minhas
Dandekar & Rath (53.4%)
(50.0%)
(41 .0%)
238.6
55.2
293.8
(47.9%)
(40.7%)
(46.3%)
VI Plan
259.6
57.2
316.8
(Revised)
(50.7%)
(40.0%)
(48.4%)
P. D. Ohja
VI Plan
221 .5
Source: Tables given by R. Datt and K.P.M. Sundaram, Indian
Economy, New Delhi : S. Chand and Co., 1978, p. 298.
75
74
TABLE 3:
TABLE 4:
Percentage Below Poverty Line
. Estimate of the
Reserve Bank
of India
•—
Estimate of
Iyengar and
Mukherjee
46
1953-54X0
1956-57
1952-53 to
1956-57 '
53.9
54.9
Rural Urban
Rural Urban
1977-78
51 .9
49.5
Top 5%
17.0
26.0
14.0
17.5
—
31 .0
1983
48.9
44.4
Top 10%
25.0
37.0
34.0
25.0
33.6
42.4
Top 50%
69.0
75.0
—
—
79.3
83.0
Bottom 20%
9.0
7.0
7.5
8.5
4.0
4.0
Years
/VSS Data
Revised'
1960-61
32.1
— ■“
1971-72
46
1972-73
—
*
Personal Income Distribution
Based on a different estimate of weights for food and
manufactured food articles, Dandekar believes that "the
estimates for 77-78 and 83 appear reasonable and in line
with the estimates of 46 per cent for 1971 -72, but the esti
mate for 72-73 appears to be on the higher side; it seems
unlikely that the proportion of the rural population increased
almost by 9 percent points in one year."
Source: V.M. Dandekar, "Agriculture, Employment and po
verty", E.P.W., Sept. 20-27, 1986.
Fractile
Group
Source:
Estimate of
N.C.A.E.R.
Rural Urban
Report of the Committee (Mahalanobis Committee)
on Distribution of Income and Levels of Living, p.14.
Poverty Debate in India: A Minority View
N SJodha
Rural socio-economic change is often inadequately captured by social science research in the field. This happens
partly due to perceptions of the researchers and partly due to inadequacies of research tools and approaches.
This paper illustrates the situation by presenting evidence on incidence of rural poverty in two villages of Rajasthan
as examined through different approaches, during 1963-66 and 1982-84. Households that have become poorer
by conventional measurement of income in fact appear better off when seen through different qualitative indicators
of their economic well-being. The paper suggests the needfor supplementing conventional measurements of income
by qualitative indicators of change to arrive at a realistic understanding of rural socio-economic change.
The Fallacy:
—The first step is to measure whatever can
be easily measured: This is ok as far as
it goes.
—The second step is to disregard that which
cannot be measured or give it an arbitrary
quantitative value: This is artificial and
misleading.
—The third step is to presume that what can
not be measured easily is not very impor
tant: This is blindness.
—The fourth step is to say that what cannot
be easily measured really does not exist:
This is suicide.
A Smith; Super Money
I
Introduction
THE paper represents a minority view in
the context of mainstream situation where,
following the seminal work by.Dandekar
and Rath [1971], researchers are competing
with each other in proving higher and higher
incidence of poverty in India with the
passage of time. Secondly, the data used
here, though rich in depth, covers a small
sample. According to this paper part of the
problem lies in the methods and distance
with which we are accustomed to look at the
field realities. Consequently, contemporary
field-oriented social science research is often
marked by contradictions and inconsisten
cies between the results obtained by macroapd micro-level studies, between the observ
ed or experienced realities and the results
generated by field research and between
observed developments and the ones in
dicated by field research. This problem is
acute in studies of rural change. Factors
underlying such changes are too detailed and
at times too complex to be captured by stan
dard and simplistic methods. Hence, incon
sistencies between resillts from different field
studies can be largely attributed to methodo
logy. This paper discusses methodological
aspects of rural economics, and suggests the
need for supplementing standard techniques
with methodological approaches appro
priate to the field. A case study covering two
villages in Rajasthan is presented to illustrate
the gap between incidence of poverty when
assessed through two different approaches.
II
Methodological Gaps
Methodological deficiencies of field
studies in social sciences are often associated
Economic r.- Poli:
Specie! Nuinl
with: (i) the concepts and categories used for
titative equivalence between such categories
identification of rural realities; (ii) the yard is often difficult.1
sticks and norms employed for assessment
(iii) Difference in the degree of precior measurement of rural realities; and
sion/vagueness attached to the quantitative
(iii) the ‘communication gaps’ between information by the respondent while giving
researcher and respondent while using (i)
the response and by the investigator while
and (ii) mentioned above.
recording it. For instance, a farmer often
The concepts and categories used to iden
reports quantitative information in terms of
tify and classify rural realities are often too
range of units or hyphenated terms (e g,
restrictive to encompass the details of petty
10-12 mandays spent on weeding a plot),
but collectively significant components of
while the investigator seeks and, using his
rural characterisation. Appendix A presents
best judgment, records it in precise terms.
a few examples. They indicate the possibili The analyst often goes a step further and
ty of disregarding variables and their interac subjects these data to sophisticated quan
tions, while using the formal/standard con titative techniques which are sensitive to
cepts for identification of different facets of
variations as small as a fraction of an hour.
rural household economy.
Appendix B illustrates some possible
What applies to the choice of concepts
‘communication gaps’. Their extent depends
and categories also applies to choice of
on the difference in the background and
norms and yardsticks. The limited coverage
working environment of the respondent and
of rural realities, owing to the use of restric
the researcher, the relative degree of
tive categories, is further reduced by using
seriousness with which investigations are
standard yardsticks to measure them. The
taken up by the two, and the ability of in
factors that do not lend themselves for easy
vestigator to establish precise equivalence
between' the respondent’s report and the
assessment/measurement through these
yardsticks are often bypassed while measur
researcher’s intended record or actual record.
ing and quantifying different variables.
Depending on the degree of ‘communica
tion gaps’, the results of different field
These limitations are now increasingly
studies of the same phenomenon in the same
recognised and the need for supplementing
area/community may differ. Other things
the formal concepts and norms by quali
being equal, the possibility of such gaps
tative approaches is emphasised [see Streetcn
being wider is greater in the macro-level
1974, McCloskey 1983, Sen 1983, Chambers
studies than those of micro-level- studies.
1986].
There are greater opportunities for partici
Realising the gap between what is reported
pant observation as well as prolonged and
through formal field studies and reality
more intimate contact between the respon
researchers have attempted to dilute or widen
dent and researcher that help narrow down
the standard categories and yardsticks.
the ‘communication gaps’.2
Despite that, inconsistencies between results
It is not difficult to imagine the distor
from different field studies persist. They exist
tions (under-reporting/over-reporting),
because of several factors, which we may call
generated by these gaps in values of different
‘communication gaps’. These gaps take place
variables recorded through household
in three forms:
(i) Difference in the connotation of the surveys. Table 1 illustrates the point on the
same concepts as they are understood by the
basis of data culled from different studies
respondent and researcher. For instance, the
with which the author has been associated^
The data reported in Table 1 relate to the
connotation of ‘manday’ or ‘man-hour’ of
cases where the extensive approach to data
labour input as understood, estimated, and
gathering was supplemented by subsequent
reported by a ‘not so time-conscious’ farmer,
detailed purpose-specific, intensive investiga
may be different from the investigator’s
tion, following the first stage screening of
understanding.
(ii) Qualitative difference in the yardsticks data. Although the number of observations
and norms used by researcher and respon
in most cases is small, they do help illustrate
the point.
dent for measurement of variables. For in
stance, a farmer reports use of farmyard
Important implications of the methodo
manure in terms of cart-loads, the resear
logical gaps include generation of incon
cher attempts to understand and record it
sistencies of the result from different studies
in terms of quintals. Establishing quan
on the same subject and the possibility of
.abcr i>88
misleading the whole approach of future
research as well as future policies relating to
specific subjects.
One way to reduce the gaps is to supple
ment the researchers’ approach by the
respondents’ approach of looking towards
the issues being studied. One may profitably
look at categories and norms used by the
respondent for identification and assessment
of variables affecting him or her.
In the following section this approach has
been attempted. For the purpose of illustra
tion we have taken one of the most debated
themes of the day, viz, the change in in
cidence of poverty in rural areas.
Ill
Approach and Data
For studying the change in incidence of
poverty we have.data for two periods of time
covering a sample of farmers3 from two
villages in the arid zone of western
Rajasthan. There are several criteria—e g,
change in household income, consumption
level, extent of employment, etc, used by
social scientists to assess the change in pover
ty levels. We have data on net household or
per capita income collected by using the con
ventional concepts and yardsticks to measure
income. Additionally, we have included the
categories or concepts which farmers/
villagers themselves use for assessing
changes in their own economic status. These
indicators of their economic status or pover
ty levels not only help in assessment of
change but they also facilitate the understan
ding of the process of change. They tend to
capture existing situation as it operates
rather than capture its formally quantifiable
proxies. Through use of these norms or in
dicators, it is easier to reduce the above ‘com
munication gaps’ and capture the past
despite long periods of recall. Their major
limitation is that they involve a more
‘investigation-intensive’ approach to field
research, because they require researchers to
approach the respondents’ level of thinking
and the issues studied. They put greater em
phasis on participant observation.
The choice of these ‘unconventional’ in
dicators of change in the present study
emerged from anecdotal information col
lected during resurveying the study villages
in 1978. They were initially studied (through
prolonged stay there, i e, over 20 days in
every month for three years ending 1965-66).
The anecdotes suggested the possibility of
substantial change in the economic status of.
households considered poor during 1963-66.
The anecdotes were used for developing
specific questions and a list of variables,
which in the village context, were considered
as real indicators of change in the people’s
economic status over time. These indicators
guided the participant observation as well
as collection of quantitative information
from 95 selected households. Those house
holds belonged to two villages, one each in
Nagaur and Jodhpur districts of Rajasthan.
They constituted a part of a larger sample
of households studied in 1963-66. Informa
tion about and from these 95 households
was collected in instalments during 1977-78,
1982-83, and 1983-84 during field work for
other projects.4 Additionally, details about
the whole village situation were collected.
Broad information on major changes (since
1963-66) observed in the study villages was
also gathered from chokala (clusters of
neighbouring) villages. The information in
dicated that study villages were not atypical
in terms of these changes. The sub-sample
of 95 households consisted of farm house
holds only. It included 35 small' and
marginal farm households, i e, who owned
less than 4.5 hectares of arid land.
Farmers’ Perception of Change
Income data of the sample households
were collected using the standard concepts
employed by farm management studies in
India. Net income data covered the follow
ing sources: crop production, animal hus
bandry, labour/bullock hire, remittances,
rental, petty trading, and property income.
Net income figures were arrived at by deduc
ting paid out and imputed costs.of all in
Table 1: Differences in Values of Selected Variables According to Method of Data Gathering
Value as Per the
Difference in Values Reasons for Difference: Items
Choice of M cthods (b) Absolute A& X100( Bypassed by (A) and Captured
by (B)
A-B
Method(A) Method(B)
A
Variable
Number and Type
of Observation
Unit of
Measurement
Average income
78 house
holds (hh)
C villages)
19 hh
23 plots,
(1 village)
12 hh
44 workers
(2 villages)
32 hh
(4 villages)
Rs/hh
6814
7564
-750
-11.0
Rs/ha
291
334
-43
-14.8
Hours/day
6.75
9.58
-2.53
-41.9
Rs/person (for
3 week one in
each season
68
79
-11
-16.2
73
105
-32
-43.8
67
120
-53
-79.1
648
822
-174
-26.9
-21.27 Accretionery process of capital
formation. [Jodha 1967].
-144 Cost of borrowing beyond
interest rate [AERC 1971],
Gross returns
Per worker/day
engagement in
farm activities
Value of food
consumption
Use level of
tractor
Extent of
land tenancy
12 tractors
Hours/week
(2 villages)
Total leased in/out ’ha
land. 86 hh
(6 villages)
26 hh (2 villages) Rs/hh
Cost of food
borrowed during
drought year
Capital investment 78 farms hh
Rs/ha
382
471
-81
Cost of credit
23 borrowers
from institutions (5 villages)
Cost as per cent of
principal
9
22
-13
Income from casual, routine acti
vities based on com mon property
resources [Jodha 1986].
Casual harvest of minor crops
for self-provisioning, etc [Jodha
et al 1978]. '
Petty and routine farm activities
[Jodha et al, 1978, Ryan et al,
1984].
Food items from common pro
perty resources/petty self
provisioning arrangements [Jodha
1986].
Most part of tractor hiring
[Jodha 1974J.
Tenancy status of plots initially
concealed [Jodha 1981].
Costs due to interlocked factor
markets [Jodha 1977[.
Notes-, (a) Based on data/information for selected cases, from the studies referred in the last column.
(b) Method ‘A’ indicates the conventional extensive approach to data gathering through one or two shot surveys using structured ques
tionnaires. Method ‘B’ involves prolonged and intensive interviews and in most cases participant observations besides what is indicated.
under ‘A’.
(c) These reasons can be related to items mentioned under appendices A and B.
2422
Economic and Political Weekly Special Number November 1908
puts from gross income. Being so standar Table 2. However, our further analysis
poorer during 1982-84 compared to 20 years
dised and so often used, these concepts hard follows a different approach. As a first step
ago. If one goes by the poverty line, i e, per
ly need further elaboration. However, it may
we compare for each household the annual
capita income of Rs 180 per year, the pro
be added that the main purpose of collec per capita income (i e, constant at 1964-66
portion of households below it has increased
ting income information during the base prices) during the two periods (i e, 1964-66
from 17 per cent in 1964-66 to 23 per cent
period was to relate it to the process of and 1982-84). The proportion of households
during 1982-84. But the latter does not in
capital formation by the sample households.
shlowing more than 5 per cent decline in
clude all the households that constituted the
The income data during the second period
group under poverty line during the base
their per capita annual income is considered
were collected to ascertain the extent of con as indicator of increased incidence of pover period. In other words some households
tribution of common property resources
who were below the poverty line in 1964-66
ty. The increased incidence of poverty thus
(pasture, forests, etc) towards household
revealed is compared with the changes in the have risen above it during 1982-84.6
income.
economic status of the people revealed by
Table 3: Indicators of Declining
qualitative indicators of change as perceiv Indispensibility of Patron’s (Rich People’s)
The terms in which villagers narrated
change in their own economic status are un ed by the villagers.
Support/Mercy/Patronage for
conventional, and they require explanation.
Employment, Income and Sustenance of
The analysis of the income data (Table 2)
Poor Households’
They are classified under the following five showed that the average per capita annual
major groups:
income of the sample households was Rs 162 Indicators
Per Cent of
(i) reduced reliance of the poor on tradi during 1964-66. This increased to Rs 1,050
Households
tional patrons, landlords, and resourceful
at current prices during 1982-84. However,
during
people for sustenance, employment, and
when the income was deflated and express
1963-66
1982-84
income;
ed in terms of constant prices (1964-66
(ii) reduced dependence on low pay-off prices),5 the figure came down to Rs 175. Households with one/more
jobs/options;
members working as
To arrive at average annual income figures
attached/semi-attached
(iii) improved mobility and liquidity for the base period, the year 1963-64 was not
labour
37
7
position;
included, because it was a severe drought
Households
residing on
(iv) shifts in consumption patterns/prac- year. The household by household com
patron
’
s
land/yard
31
0
tices; and
parison of per capita income during the two
Households resorting to
(v) acquisition of consumer durables.
periods (including by pooling the data of
off-season borrowing of
households which had split since the ben
Information on most of the above items
foodgrain from patrons
77
26
chmark period), indicated that for 38 per Households taking seed
was available from benchmark data on
resource endowment, production, market cent of the households, the per capita an
loan from patrons
34
9
nual income had declined by more than 5 Households marketing
ing, and consumption activities of the sam
per cent of the base period income. For 47
ple households. For the resurvey period, it
farm produce only
per cent of households income increased by
was purposely collected to see the change.
through patrons
23
86
more than 5per cent. The remaining house Households taking loan
The indicators of change perceived by the
holds, where per capita annual income
from others besides
villagers can be grouped under categories
47
changed only within + 5 per cent, have been
patrons
13
which are more familiar to economists and
treated as the cases where per capita incomes
used in their professional communication.
remained constant during the reference Note', a Details in this and the following four
They are: (a)’indicators of enlarging op
tables relate only to 35 households
periods.
portunity sets or increasing number of
whose per capita annual income (at
choices (e g, in the matter of employment,
According to the above figures, 38 per cent
constant prices) had declined during
borrowing, marketing, etc);
of the sample households have become
1982-84 compared to 1964-66.
(b) indicators of-consumption activities
with high income elasticities (e g, travel,
Table 2: Details of Income Position of Sample Households at Two Points of Time
slack season purchases, length of materni
(Per capita annual net income in Ps)3
ty feeding of women, etc);
(c) indicators of investment in lumpy con
Details
•
_____Average'Situation during
sumer durables (e g, pucca structures of
1964-66b<
1982-84
houses, compounds to houses, etc).
” '''At Current At Constant
Prices s
Prices0
1050.;jvj:;
175
43V
33
11
23
—
—
—
_
47
—
38
_
15
CHANGES IN INCIDENCE OF
Poverty
The incidence of poverty in the ongoing
debate on the subject in India is judged with
reference to poverty line and the changes
over time in the proportions of population
below the poverty line. Though conceived in
terms of per capita conusmption expenditure
the poverty line is indicated by a figure of
monthly per capita income, such as Rs 15
for rural areas at 1960-61 prices, according
to Dandekar and Rath [1971] and Rs 65 at
1977-78 prices, according to the Indian Planning Commission (1981). Per Capita annual
income of Rs 180 (at 1964-66 prices) may
be considered as a poverty line comparable
to the one suggested by Dandekar and Rath
[1971]. The proportion of sample households
m study villages falling below this income
level during the two periods is indicated in
Average per capita annual income (Rs)
162
Contribution of different-sources of income (per cent)
—Crop farming
48
—Animal husbandry
27
—Labour/bullock hire
14
—Others (rent, remittance, etc)
21
Proportion of households with per capita annual
income (at constant prices)5
—less than Rs 180 (i e, poverty line) (per cent)
17
—showing increase of more than 5 per cent
_
over the period (per cent)
—showing decline of more than 5 per cent over
the period (per cent)
—showing positive or negative change up to
5 per cent over the period (constant
_
income) (per cent)
23
Notes: a Data relates to 95 sample households from two villages one each from Jodhpur and
Nagaur districts in Rajasthan.
b 1963-64 being a severe drought year its income figures are not considered.
c At 1964-66 prices.
Economic and Political Weekly Special Number November 1988
2423
qualitative Indicators
If one goes by the qualitative indicators
of poverty or absence of it as mentioned
earlier, a completely opposite picture seems
to appear. Tables 3 to 7 illustrate the
phenomenon. These tables give details about
only those (35 of 95) households whose per
capita annual income has declined by more
Table 4: Indicators of Reduced Dependence
on Low Pay-off (Inferior) Jobs in Case of
Poor Households
Indicators
Per Cent of
Households
during
1963-66 1982-84
Households engaged ina
—food gathering
—fuel gathering
—fodder gathering
Households having members
engaged in part-time
petty jobsb
Households with members
seasonally out-migrating
for job
Households withdrawing
their children from school
during crop .eason for
work help, earning etc
100
100
100
20
63
23
100
23
34
17
11
6
Note-, a Only items like wild fruits during sum
mer season, and fuel/foddcr during
post-harvest period are considered. In
these cases supply is not a constraint
to reduce peoples’ dependence on
them.
b Jobs like helping in fencing, etc, for
getting one meal as wage.
Table 5: Indicators of Improved Mobility
and Liquidity Position of Poor Households
Indicators
Per Cent of
Households
during
1963-66 1982-84
Households selling over 80
per cent of their marketed
produce during post
100
harvest period
Households retaining up to
25 per cent of surplus for
0
sale up to next rain
Households purchasing key
0
provisions in bulk
Households relying on
day-to-day petty purchases
100
.of key provisions3
Households making cash
purchases during slack
6
season festivals, etc
Households possessing ready
cash up to Rs 200 or more at
home' during slack season
0
Households having members
who travel by paid transport
more than twice a year to
17
outside the district
46
6
6
51
51
26
78
Note-. a Provisions like chilly, onion, gur, oil,
etc.
2424
than 5 per cent during the period under
review. Furthermore, these tables present the
extent of change in terms of proportion of
households whose situation as per the abovementioned indicators has changed during
1982-84 compared to the base period.
Table 3 indicates the extent of decline in
the reliance on patronage and the support
of the rich (patrons) for the employment and
sustenance of the poor households, i e, the
households that have become poorer since
1964-66 (Table 2). Some of the indicators,
such as the practice of attached labour, seed
loan in Und (at exorbitant interest rate),
marketing produce only through patrons,
depending solely on patrons for credit, and
residence on patron’s land necessitating
supply of unpaid and unaccounted labour
services to the patrons, have inherent an ele
ment of exploitation of the poor. The poor
people’s ability to dispense with these prac
tices is the surest indicator of their improved
economic status. Despite several socio
economic reform measures such as anti
bonded labour laws, etc, the poor people
continue to accept these exploitative ar
rangements by patrons. They tend to give up
these arrangements only when they become
economically more independent.7
The inferior or low pay-off jobs (including
food gathering from the fast-declining com
mon property resources) are usually taken
up by the poor in the villages [Jodha 1986].
The recourse to such jobs declines as one im
proves his or her economic condition.
Table 4 indicates that the group of
households that have become poorer in
1982-84, as per the formal income criteria,
had relied more on these inferior options
during the base period when they were
relatively rich. Now, despite increase in their
poverty (i e, reduced per capita income) their
preference for inferior jobs has declined, as
indicated by proportion of households under
relevant categories under Table 4.
Several indicators in Table 5 reveal that
general liquidity of the group of households
that have become poorer is better now (i e,
in 1982-84) than it was during the base
period, when income-wise they were relative
ly rich. Their ability to make pruchase of
provisions in bulk by paying for it in a single
instalment, cash purchases during summer
season festivals, and keeping significant
amount of cash in hand during the slack
season are definite signs of improvement
notwithstanding the decline in their formally
recorded income position.8
The consumption pattern, particularly in
terms of inclusion of items which poor peo
ple rarely use, is another indicator of
substantial change in the economic condi
tion of these people. Now there is a much
higher proportion of the concerned group
of households (Table 6) who frequently con
sume better quality food items,9 offer bet
ter maternity diet to women for a longer
period, and where women and children
regularly wear shoes. The only item where
the situation seems to have deteriorated is
the proportion of households regularly using
Table 6: Indicators of Shifts in
Consumption Pattern of Poor Households
Indicators
Per Cent of
Households
during
I 963-66 1982-84
Households occasionally
consuming green vegetables
during non-crop season
0
Households consuming
curries mainly made from
cereals3
100
"Households using milk/
milk products regularly
34
Households consuming
sugar regularly
0
Households consuming rice on
non-festive occasions also
0
Households with adults
skipping third meal in the
day during the summer
(scarcity period)
86
Households where women
and children wear shoes
regularly
0
Households where maternity
feeding to women provided
up to a month or more
6
100
14
6
20
14
20
86
23
Note: a As per the local saying one who can
not afford vegetables, etc, eats cereals
with the help of poor quality curry
made of cereals only.
milk and milk products. This is, in fact, a
side-effect of improved milk marketing
facilities in the villages. The sale of milk has
helped raise the share of livestock income
in total income (Table 2), but has also reduc
ed the opportunities for self-consumption of
milk and milk products.10
The situation regarding the changes in the
possession of consumer durables seems more
impressive (Table 7). Pucca structures of
houses, provision of doors and gates, com
pound walls, separate quarters for humans
and livestock in the house, and better faci
lities for women are important indicators of
positive change in the economic status of the
people. The higher proportion of the house
holds possessing these items in. 1982-84 com
pared to the base period indicate a substan
tial improvement in their economic position.
The detailed explanation of these changes
falls outside the scope of this paper.
However, it may be mentioned that a com
bination of factors has led to the improved
condition of the households in the study
villages. Occurrence of these factors observ
ed in several villages of the districts of
Nagaur, Jodhpur, Pali, and Sikar in western.
Rajasthan would suggest that the changes
refelcted through study villages may extend
to wider areas of the region.
The possible factors responsible for im
proved economic conditions of sample
households include the following:
(i) A continuous spell of good rain years
during 1974 to 1978.
(ii) Possibility of double cropping in sandy
loam soils without change in rainfall or ir
rigation due to raya (a minor oilseed) crop
.Economic and Political Weekly Special Number
No\cmbe. 198C
for post-rainy season, brought by seasonal
holding discouraging ownership of unpro
having become poorer over time. They may
migrants to Punjab in the early 1970s and
ductive animals.
include some households who were rich
its spread in the dry region without any
(v) Reduced incidence of guini-worm enough and a fall of 5 per cent in their in
research and extension effort. The net
among adult workers in the recent years
come did not make them much poorer.
returns from this crop are higher than the
which often incapacitates them during the
However, the data for households grouped
main rainy season crop like pearl millet.
crop season. This happened due to Drought
according to level of decline in income were
(iii) Coverage of larger area by moisture Prone Area Programme (DPAP) provision
also examined. The emerging number of
conserving practice of bunding, which in
of piped groundwater supply for drinking,
observations in each group became too small
association with timely ploughing through
replacing traditional practice of using pond
to be meaningfully reported. However, the
tractors helped in adoption of hybrid pearl
water.
inferences from retabulation, which could
millet like BJ4.
(vi) Off-season employment under rural help to satisfy the above objection may be
(iv) Facility of milk marketing which works programme/DPAP and regular offmentioned:
generated regular cash income and also in
farm jobs to some people.
(a) The proportion of households show
duced changes in the composition of animal
(vii) Institutional reforms helping people ing qualitative improvement in economic
in getting lands including house sites and
conditions as per the above indicators were
Table 7: Indicators of Change in Asset
reduction in indebtedness.
not very different in the case of sub-groups
Position of Poor Households
(viii) Gains to poor as a byproduct of fac of high and low income households, which
tionalism
among
the
rural
rich,
where
each
suffered decline in their per capita annual
Indicators
Percentage of
faction tried to woo the poor for their
income as per Table 2.
Households
support.
during
(b) Even the 23 per cent of sample house
Table
8
summarises
the
changes
in
the
situa
holds who were below' poverty line (i e, per
1963-66 1982-84
tion with reference to some of the factors
capita annual income of Rs 180 at constant
Households having houses with
mentioned. The data relate to the 95 sam prices (Table 2), had a fairly large propor
—fully pucca structure
0
14
ple households.
tion of households that showed improve
—partly pucca structure
9
52
ment in their economic status as per the
—only kutcha structure
91
34
Towards Reconciliation
qualitative indicators discussed above.
—gate with doors
6
43
—compound wall/fence
(i) The first inference from the perusal of (c) There was a small number of house
13
52
holds in the group that neither faced decline
—separate provision of
information under Table 2 on the one hand
in their per capita annual income nor slip
stay for humans and
and Tables 3 to 7 on the other, is that the
ped below the poverty line and yet did not
animals
52
6
extent of increased incidence of poverty
show improvement in terms of qualitative
—private place (bath room,
reflected by Table 2 is not borne by the
indicators.
■
■
etc) for women
0
23
qualitative indicators of change under the
Thus the main explanation may lie in the
Households possessing:
remaining tables.
—quilts of cotton
6
20
use of specific approaches. to assess and
Part of the explanation could be that we
94
—quilts of old rags
80
record economic change affecting the rural
have considered all households, whose per
—radio
0
7
households. Furthermore, change in econo
capita
annual
income
has
declined
by
more
—bicycle
0
3
mic status revealed by qualitative indicators
than 5 per cent of base period income, as
is an outcome of gradual change over a
period of time. Difference of per capita net
Table 8: Possible Factors Underlying Qualitative Improvement in Condition of
income at two points of time may not cap-,
Sample Households Since 1963-66a
ture this change. The measurement of in
Average (Per Year)
Remarks.
Units of
Factors
come at one point in time captures only the
Observation
Situation during
current transitory component of income.
1982-84
1963-66
The permanent components (accumulated '
transitory components) of income in the past
Percentoftotal
Extent of raycP crop
are not captured . This reinforces the need
cropped area
26
0
for revising the research aproach to under
38
0
Extent of hybrid pearl millet
stand the dynamics of rural change, and to
4
1
Extent of irrigation
cover permanent components, of income
7
68
Mostly by hire
Extent of tractor cultivation
besides the transitory components, each of
134
Cumulative
43
No of plots
Extent of bunding
which may not move in the same direction.
totals
(ii) The reported case study is to small in
36
Households selling milk
No
5
its coverage to encourage any generalisation
2
6
Cows and
No
Unproductive animals per
of results. However, this does indicate the
buffaloes only
productive animal
7
29
need for complementing formal concepts
No
Off-farm regular jobs
4
People affected by guini-worm':
and norms by more informal categories and
No
58
27
5
methods to capture a greater extent of reality
No
Litigation cases
3.2
Non-workers per worker
No
3.9
through social science research in the field.
Households benefitting from:
It also underscores the importance of par
Got land
0
18
No
i Institutional reforms
ticipant observations and in-depth micro
animals, debt
level investigations in field studies
reduction etc,
(iii) Intensive and qualitative information
since 1986.
gathering may prove costly. Hence, this ap
ii Factionalism among
proach can be used for generating relevant
No
0
29
the rich
indicators (proxies) that can form part of the
large-scale, formal data-gathering projects.
Notes: a Data relate to 95 sample households only.
Furthermore, the insights received through
b Raya, a high value small oilseed post-rainy season crop, has spread without any research
such intensive investigations can help in the
or extension effort in the region.
better interpretation of results from exten
c Guini-worm disease caused by drinking water from ponds. Piped water supply under
DDAP scheme helped reduce it.
sive studies.
Eccr.omic
Special Numbc;
Novcmbc* 1988
2425
(iv) A factor which can enhance the com debated ‘Quality of Life Index’ as against
plementarity of the macro-level and high:*
gross domestic product, etc, as a better in
quantitative studies on the one hand and in
dicator of a nation’s economic well-being.
tensive and micro-level research on the ether
is close links of principal researchers with
Notes
the field situation.
1 In ICRISAT’s village level studies, the
(v) This case study indicates the need for
measurement problems have been handled
by physical weighing or measuring of the
a fresh look at the conceptualisations
quantities reported in volumes. Such con
underlying the measurement of the level and
versions were done on random basis to
change in rural poverty. The complementary’
evolve equivalence between two categories.
use of quantitative and qualitative concepts
See Binswanger and Jodha [1977]. can help improve our understanding of the
2 Various types of measurement errors
dynamics of poverty.
emanating from aforementioned factors will
(vi) To the extent that the incidence of
influence the results depending on the type
poverty can be partly inferred from obser
of analysis. For instance, if a variable is
vance of poverty indicators, the next pro
measured with a random error, that will not
blem relates to the possibility of measure
affect the estimate of its mean and regres
ment of these indicators for comparative
sion estimate if it is the dependent variable
studies. To the extent a pan of the indicators
in a multivariate regression. But it still will
of change discussed in this paper may be
bias towards zero its coefficient if it is us
area or community-specific, their use for
ed as a right-side variable in a regression.
inter-community comparisons will be
On the other hand, systematic measurement
limited. Thought may be given to evolution
error may cause more or less problems,
depending on the mode of analysis and
of some indices on the pattern of currently
Appendix A
Examples of Concepts/Categories and Yardsticks/Norms Used by Social Science
Researchers to Identify and Measure Variables Comprising Rural Realities and
Facets of Reality Likely to be Bypassed by Them
Concepts and Norms
Aspects Covered
Facets Bypassed
Household income Cash and kind inflows (in Ignores time context and transaction partner
cluding imputed values of context of income generating activity;
major non-traded items). disregards flow of low value self-provisioning
activities with significant collective contribu
tion to sustenance of the people.
Farm production
Production from all farm Series of intermediate activities (often con
enterprises.
sidered as consumption activities), which
facilitate the final output from farm enter
prises in self-provisioning societies.
Food consumption Volume and quality of for Ignores seasonally varying streams of self
basket
mally recorded food items. provisioning items/services.
Household resource Only privately owned land, Ignores households’ collective access to comendowment
labour and capital resources, mon property resources; access to power and
influence too.
Factor/product
Competitive, impersonal
Ignores distortions, imperfections, etc, due to
market
interactive process of
factors like influence, power,., affinities and
framework.
inequities.
s Farm size grouping Based on owned or operated Ignores totality of asset position including
landholdings (often standar- household’s access to common property
dised for productivity and resources, its workforce which determines
irrigation).
^households’ ultimate potential to harness land
resources and environment for sustenance.
Labour input
Labour as standard unit
Disregards heterogeneity of labour of same
expressed in terms of man- age/sex in terms of differences in stamina and
hours or mandays, etc, (Dif- productivity; ignores differences in intensity
ferentiation based on age and of effort of a self-employed worker and hired
worker. (In appropriate imputation of value
sex not withstanding).
of the labour of self-employed worker is done
on the basis of wage rate of hired or attached
labourer).
Capital formation Acquisition of assets.
Ignores accretionery process, and petty accre
tions which are important collectively.
Depreciation of
Book-keeping-value based Ignores continued usability and recyclibility.
assets
reduction in the worth of the
asset.
Efficiency/
Quantity and value of final Ignores totality of the operation of the system
productivity norm produce of an activity (based directed to satisfaction of multiple objectives
on market criteria)
rather than single criterion.
Economic and Political Weekly Special Number November 1988
nature of the error. Systematic measurement
error will bias the mean but may not bias
the regression. This may be added that
systematic mismeasurement over time
should not lead to the obfuscation of
changes in the variable that is being
mismeasured. If mismeasurement errors
themselves do not change over time, valid
conclusion on dynamics can still be drawn.
Our sample does not include a landless
household. There were hardly any landless
households (except traders, etc) in the study
villages. In fact, landlessness of the type
observed in high population areas can hard
ly survive in the arid lands. Furthermore,
a number of sample households of benchmark period had split over time. For the
purpose of comparison at two points of
time data of such households were pooled
to reconstitute original households.
4 Data during 1963-66 were collected as a part
of the field work for the author’s PhD thesis
[Jodha 1967] and land transformation
studies of Central Arid Zone Research In
stitute (CAZRI). The data for subsequent
periods were gathered while collecting in
formation for ICRISAT’s research projects
on Farmers’ Group Action for Watershed
Based Resource Development in 1977-78,
and Role of Common Property Resources
in Farming Systems in 1982-84 [Jodha
1986].
5 Income during 1982-84 was deflated by the
extent of increase in gold price in the
villages. The logic of using the change in
gold price as index of inflation is that one
tola (10g) of gold fetched the same quanti
ty of bajra (pearl millet)—the staple
foodgrain of the'people—in 1982-84 as it
fetched in 1964-66. However, gold price per
tola has increased by about six times since
then. For further details on this approach
see Jodha [1985]. The calculations based on
changes in consumer price index for
agricultural workers in Rajasthan during the
period under review also indicated the price
change of similar magnitude (i e, 5.7 times).
6 In terms of rainfall and crops 1963-64 was
a complete drought year. As per the an
na wari system of crop assessment 1964-65
had bumper crops, while 1965-66 had
average crops. The year 1982-83 had above
average crops while crops were" below
average during 1983-84. On an average crop
wise the period 1964-66 was slightly better
than 1982-84. This influenced the income
positions of the sample households to some1
extent. Of 95 sample households, 35 had
less per capita income during 1982-84 com
pared to 1964-66. A part of it could be due
to life cycle related factors such as increased
number of members especially dependents
in the households. However, due to a variety
of factors 22 of the 35 households had in
come below poverty line during'-1982-84.
This included seven households who were
already below poverty line and 15 house
holds who were above it during 1964-66.
There were 11 households who moved above
poverty line during the same period.
7 Reduced reliance on patronage of rich (and
on inferior) options such as Common Pro
perty Resource (CPR)-activities, could be
both supply-determined and demand-
3
2427
Appendix B
Examples of ‘Communication Gaps’ (Under Three Categories)
(1) Possible differences in connotation of same concept as understood- by respondent
and researcher.
Concept
Researcher
Food consumption Total food
Connotation As Per:
Respondent
Major food items excluding petty self
provisioning
Final produce excluding items harvested
during the intra-season period.
Total work time often more than 8-10 hours.
Produce
Total
Manday
Formal work hour
8-10 hours, etc
Hired + exchanged
Only hired.
Involuntary unemployment. Disguised unemployment treated as full
unemployment.
Hired labour
Unemployment
(2) Possible gaps in yardsticks guiding respondent’s quantitative responses and researcher’s recor
ding of responses which may make it difficult to establish perfect equivalence between the reported
and recorded quantities.
Item
Length/area
Researcher
Modern units (metre,
Traditional—foot-lengths, steps, arm-lengths,
hectares, inches, etc
finger widths.
Modern measures such as Cart-loads, bag fulls, volume based measures
kilograms, quintals, litres, (barrels, etc).
etc,
Modern measures, quintals, Self-sufficiency periods of subsistenceetc
requirement, e g, total production equal to 6
months of requirements, etc. .
Precise—days, hour, etc
Vague in terms of proportion of a day or a
week, etc, i e, half-a-day, 3/< of a day, etc.
Weight/volumes
Production
Time
(3) Degree of precision/vagueness associated with responses as they are given and recorded.
Reporting by Respondent
Item
Recording by Researcher
Labour input
Grain yield
Ranged units, e g, 5-7 hour, 10-12 days, etc.
Exact days/hours
Exact quantities in modern Range: e g, 5-6 bags or 50-55 quintals, etc.
measures/units (quintals/
kgs, etc)
Range in terms of proportion. Vi to Vi of bag,
Exact quantities
etc.
r
w
Input use/output
sold
to undertake this investigation related to the
determined options. However, in our study
liquidity position of the rural poor. During
we have included only demand induced
my 1978 revisit to one of the villages, I was
cases. For instance, the patrons now given
talking to a villager whose room I rented
up by th? concerned poor households were
during my early (1963-66) stay in the village.
still (at the time of resurvey), in the same
A woman labourer arrived there to collect
business of offering facilities like site for
her wages for the work she did for my ex
living, crisis period food and money supply,
landlord. To avoid her, he pleaded non
etc. However, they didn’t have many of rele
availability of change and called her two
vant customers to work as attached workers.
The poor who left their patrons now have
days later. The woman promptly untied a
knot in her lugari (sari), took out change
their own house site and united facilities of
and said “you need change for how muchcredit, marketing, etc, from others, in
Rs 100? Rs 50?”. Contrast this with the
cluding from co-operatives. The factiona
situation during 1963-66, when, if by
lism between rural rich indirectly favouring
mistake I failed to carry change, there was
the poor, on the one hand, and some public
nobody in the village who could offer me
programmes, on the other, seem tp have
change for Rs 100, and I had to visit the
helped the poor in getting rid or exploitative
district place/neighbouring town to get
patronage (Table 8).
change for Rs 100.
• In the case of dependence on CPRs, only
9 Of the 35 households more than 20 used to
those activities have been considered where
offer tea made with jaggery during my fre
supply was not a constraining factor. They
quent visits to their houses for data during
included collection of wild fruits (ker
1963-66. During revisits I found all of them
sangari, etc) during summer season and
fuel/fodder accumulation during the period
using sugar instead ofjaggery for the same
soon after the harvest of crops.
purpose.
8 The very first anecdote which provoked me 10 The cash nexus induces farmers to part with
2428
practically all of their milk supplies, leav
ing little milk for self-provisioning or for
sharing (buttermilk, etc) with others in the
villages. Cases were observed where
households producing as much as 10 litter
of milk a day brought milk from the tea
shop to prepare tea for the visitors (in
cluding myself).
References
Binswanger, HP, and Jodha, NS 1978, Manual
of Instructions for Economic Investigators
in ICRISAT's Village Level Studies, Vol II,
ICRISAT Economics Programme Village
Level Studies Series, Patencheru, AP, India.
Chambers, R 1986, Poverty in India: concepts,
Research and Reality—An Exploration, In-,
stitute of Development Studies, University
of Sussex, Brighton, UK.
Dandekar, V and Rath, N 1971, Poverty in India".
Dimensions and Trends, Economic and
Political Weekly, 6(1).
Jodha, N S 1967, ‘Capital Formation in Arid
Agriculture: A Study of Resource Conser
vation and Reclamation^Measures Applied
to Arid Agriculture’, unpublished Ph D,
thesis, Jodhpur, Rajasthan: University of
Jodhpur.
Jodha, NS 1977, ‘Effectiveness of Farmers’ Ad
justment to Risk} Economic and Political
Weekly, 13(25): A49-A62.
Jodha, N S 1985, ‘Market Forces and Erosion .
of Common Property Resources’, 263-278,
in Agricultural Markets in Semi-Arid
Tropics: Proceedings of the International
Workshop, October 24-28, 1983, ICRISAT
Centre, Patancheru, AP, India.
Jodha N S, 1986, ‘Common Property
Resources apd Rural Poor in Dry Regions
of India’, Economic and Political Weekly,
21(27): 1169-1181.
McCloskey D M 1983, ‘The Rhetoric of
Economics’, Journal of Economics
Literature Vol XXI: 481-517.
Planning Commission, 1981, Sixth Five-Year
Plan 1980-85, New Delhi, Government of
India.
Sen, A K 1982, Choice Welfare and Measure
ment, Oxford: Blackwell.
Streeten, P P 1974, ‘Social Science Research on
Development: Some Problems in the use of
Transfer of an Intellectual Technology’,
Journal of Economic Literature
p 1290-1300.
Economic and Political Weekly
Available from:
M/s Gangavaram Agencies,
94/1, 5th Cross Road,
Malleswaram,
Bangalore - 560 003.
M/s EBS News Agency,
1601 Sector 22 B,
Chandigarh - 160 022.
M/s Subhash News Agency,
22nd Outside Hall Gate,
Amritsar, Punjab.
Economic and Political Weekly Special Number November 1988
Got^-Hl -90-10
Dimensions of Rural Poverty: An
Inter-Regional Profile
©
L R Jain
K Sundaram
S D Tendulkar
This study considers six inter-related dimensions of poverty and seeks to (a) provide region-level estimates for
all the six variables, (b) measure and analyse inter-regional disparities in the indicators of poverty; (c) examine
the spatial distribution of regions by reference to the quartile-wise ranking along each of the dimensions ofpoverty;
(d) identify spatial patterns of contiguity by mapping; and (e) analyse the inter-relations among the chosen indicators
by means of bivariate quartile-wise cross-tabulation.
THIS paper presents an inter-regional pro 27th round. The survey covered 72,270 rural
file of rural poverty in the early seventies in
households on a stratified sampling basis.
several dimensions.
(The survey results have been published in
Regions considered in this study are those
Sarvekshana, Vol VI, nos 3-4, January-April
delineated by the National Sample Survey
1983.)
Organisation (NSSO) for their twenty
The rural areas have been divided into 62
seventh round nation-wide survey of con
regions covering 21 states on the basis of
sumer expenditure. In this survey, NSSO
agro-climatic homogeneity. The present
divided the country into 62 agro-climatically
study is confined to 56 out of 62 regions and
homogeneous regions each consisting of two
excludes the following six regions: two
or more districts with typically more than
regions of Manipur and single-region states
one region per state. Our study is confined
of Meghalaya, Himachal Pradesh, Tripura
to 56 out of these 62 regions, (see Section II
and Delhi. Other union territories were not
for the discussion of the data base and
covered by the survey. The exclusion of the
related issues).
six regions is a carry-over from our earlier
In this study, we consider six inter-related
study [5] which sought to explain the inter
dimensions of poverty, namely (1) the
regional variations in rural poverty. In that
average private consumer expenditure per
study and in its extension [3] by the present
capita per month (AVCE); (2) the Gini co
authors, the omission of these six regions
efficient summarising the size distribution
was dictated by the absence of information
of AVCE (denoted by GINIC); (3) the pro
on one important explanatory variable. For
portion of the rural population below the
each region within a state, region code,
poverty line (POVT); (4) the poverty-gap
description of the region and composition
ratio (PGR); (5) The Gini coefficient (GP)
(in terms of districts/tehsils) of the region
for the size distribution of consumer expen
are given in Appendix Table I and shown in
diture within the set of the poor; and (6) the
Map A-I.
Sen Index (SI) that captures the intensity
In computing the alternative measures of
dimension of poverty by combining POVT,
poverty for each of the 56 regions, we first
PGR and GP.
derive
state-specific poverty lines. We start
The paper seeks to
a) provide the region-level estimates for with the widely used all-India rural poverty
line of Rs 15 per capita per month at 1960-61
all the six variables listed above;
b) measure and analyse inter-regional prices. This is first adjusted for inter-state
price-differentials for the base ye*r. This ad
disparities in the indicators of poverty
justment was made on the basis of the Fisher
(Section 111);
c) examine the spatial distribution of price index for each state compared to all
India as base and applicable to 40-60 per
regions by reference to the quartile-wise
cent fractile group of the rural population.
ranking along each of rhe dimensions of
poverty (Section IV);
This set of price relatives was derived by
d) identify spatial patterns of contiguity Chatterjee and Bhattacharya [1] for the year
1963-64 and the same is assumed to apply
by mapping (Section V); and
e) analyse the inter-relations among the to 1960-61 as well. The resultant poverty lines
at 1960-61 prices are adjusted for price
chosen indicators by means of bivariate
changes between 1960-61 and 1972-73 by
quartile-wise cross-tabulation (Section VI).
reference to state-specific consumer price
Concluding observations appear in
indices for agricultural labourers. Each state
Section VII.
specific poverty line at 1972-73 prices so
derived is assumed to apply to all the regions
II
within that state.
Data Base and Related Issues
As regards the computational procedures,
The study is based on the data relating to
we have essentially followed the interpola
the National Sample Survey (NSS for short)
tion methods suggested by Kakwani [2] to
on Consumer Expenditure carried out bet
derive the different indicators of poverty
ween July 1972 and June 1973 during the
used in this per (For details; see [3]).
Ec^r.?”ric c".d Political Weekly Special Nu.ab./ November 1988
III
Regional Dimensions of Poverty:
A Summary View
In this section we present and analyse the
estimates of six indicators highlighting dif
ferent dimensions of poverty for the rural
areas of 56 NSS regions covering a little
more than 97 per cent of the all-India rural
population. The six indicators used in this
analysis are:
a) POVT or the proportion of the rural
population below the poverty line or what
is called the head-count measure of poverty;
b) AVCE or the average private consumer
expenditure per capita per month as a
measure of the average level of living in the
region;
c) GINIC or the Gini coefficient sum
marising the size-distribution of AVCE in
dicative of the extent of relative inequality
in levels of living within a region;
d) the Sen-Index (SI) capturing the ,
intensity-dimension of poverty that brings
within its ambit not only the head-count
measure of poverty but also;
e) the Poverty-Gap Ratio (PGR) which
indicates the gap between the poverty-level
per capita expenditure and the average per
capita expenditure for the subset of the poor
population relative to the poverty level per
capita expenditure; and
f) the Gini-coeuicient summarising the
relative inequality in the (truncated) size
distribution of per capita consumer expen
diture within the set of the poor (or GP).
It may be noted that AVCE and GINIC,
besides being important in their own right
as indicators of rural poverty are the two
proximate variables that have been shown
to explain in a substantial measure the
inter-regional variations in POVT and SI
(see [3]).
We begin our analysis by considering
statistics which summarise the variations
across 56 regions in respect of each of the
six indicators listed (see Table 1). The
detailed estimates of the six indicators for
each of the 56 regions are presented in the
Appendix Thble II.) The summary statistics
relate to the minimum and the maximum
values defining the range of variation, the
value of the unweighted mean of the 56
2395
observations x, the all-India average which
is the weighted mean across 62 regions x* and
the unweighted coefficients of variation
across 56 regions defined alternatively by
reference to x and x*.
The following points may be noted:
I) Comparing the unweighted mean
values (defined over 56 regions) and the all
India averages, one finds that the all-India
POVT is lower (if only slightly). This is so,
despite the fact that the all-India value of
AVCE is lower and that of GINIC higher
than the corresponding unweighted mean
values across 56 regions. Also, the all-India
Sen-Index SI is lower even though the all
India values of two of Si’s component
arguments, PGR and GP—respectively the
Poverty Gap ratio and the Gini-coefficient
of consumption among the poor—are higher
than their unweighted counterparts relating
to 56 regions.
2) Except in the case of GP and PGR, the
inter-regional coefficients of variation (CV
for short) of the indicators are higher when
defined with respect to their all-India
average values x* than when defined with
respect to x.
3) Whether we consider CV’s with respect
to x or X*, inter-regional disparity as
measured by CV is the least in respect of
GINIC which summarises the relative in
equality within the region. The inter-regional
disparity is the highest in respect of SenIndex, SI.
4) The coefficient of variation of GINIC
is lower than that of GP, though the mean
value of the latter is only about a half of
the mean value of the former. To put it dif
ferently, even though within a region the
inequality in the size distribution of per
capita consumption expenditure among
those below the poverty line is much lower
than that relating to consumption of the
total population, the relative variability
across regions is higher is the former case.
quartile. A possible exception especially at
V
the lower enji is the Sen-Index.
Patterns
of
Contiguity
When we divide the percentage given in
line (c) by that given in line' (b) we get the
. A visual picture of the quartile-wise ranks
Quartile-wise Analysis
head-count ratio in that quartile relative to
for the 56 regions in respect of each of the
the head-count ratio for al! the 56 regions.
six ranking variables is presented in the maps
To facilitate further analysis, we have
If we multiply this quartile specific relative
(1*through 6) that follow. While the maps
grouped 56 regions into 4 quartiles of 14
are self-explanatory, we offer brief com
each when the regions are ranked in ascen head-count ratio by the all-India head-count
ments on the important patterns of conti
ratio given in Thble 1 (column 6), we can
ding order according to each one of the six
approximate the weighted average head guity, revealed by these maps, especially as
indicators. Table 2 gives for each indictor:
count ratio in each quartile. The quartile
they relate to the lowest and the highest
a) the upper-terminal value of the specific head-count ratios are given in Ihble
quartiles.
indicator;
3. Quartile specific head-count ratios can be
Let us consider the map representing the
quartile-wise distribution of regions ranked
b) the percentage of the total rural seen to progressively decline when the
regions are ranked according to the size of in terms of the average per capita private
population in the quartile; and
consumption expenditure or AVCE. Of the
AVCE. A progressive rise in the quartile
c) the percentage of the total rural poor specific head-count ratio can ^e observed
14 regions in the lowest quartile, 8 are con
population in the quartile.
when the regions are ranked according to the
tiguous regions located in the states of West
When we combine the upper-terminal value
Sen-Index and its constituent elements
Bengal (2), Orissa (2), Andhra Pradesh (1),
for the first quartile with the minimum value
(Table 3, lines 3 to 6). GINIC is the only •Madhya Pradesh (2) and Bihar (1). In
of the variable given in Table 1 (column 3),
ranking variable for which the head-count
addition, all the three regions of Tamil Nadu
we get the full range of the variable accor
fall in the lowest quartile of AVCE. In fact,
ratio does not exhibit monotonic relation
ding to quartiles.
ship across quartiles. In fact, the quartile ! but for the presence of the inland southern
For all the indicators considered, the
specific head-count ratio varies within nar , region of Andhra Pradesh (which falls in the
regions are densely concentrated over a
row bound between 43 and 50 per cent when : third quartile of AVCE) breaking the con
narrow range in the second and the third
tiguity between inland northern region of
the regions are ranked according to GINIC.
2396 .
.Economic and Political Wec/Jj
Spec:;.! Number bHvnrnber 1ORR
Andhra Pradesh and the coastal northern
region of Tamil Nadu, the contiguous low
income bdt would hare stretched even longer
from the eastern plains of West Bengal to
Kanyakumari in Tamil Nadu.
When we shift our focus to the other end
of the spectrum, we again find a contiguous
stretch of 8 regions in the north-west belt of
Haryana (2), Punjab (2), Rajasthan (2) and
parts of Madhya Pradesh (2) in the highest
quartile of AVCE. Further, the contiguity of
these eight regions with three contiguous
regions of Gujarat is broken by the presence
of the dry areas region of Gujarat which is
located in the third quartile when ranked by
AVCE.
We turn now to the two maps (2 and 3)
depicting the quartile- wise ranking of
regions in terms of the two alternative
measures of poverty, namely, the head-count
ratio (POVT) and the Sen-Index (SI). As
regards POVT, we find that all the 14 regions
in the highest quartile form a contiguous
east-west belt spread all the way from West
Bengal to Rajasthan encompassing three
regions of West Bengal, two regions of
Orissa, one region of Bihar, two regions
of Madhya Pradesh, four regions of
Maharashtra and one region each of Gujarat
and Rajasthan. The picture is almost iden
tical when the regions are ranked in terms
of Sen-Index with two variations. In the east,
the central plains region rather than the
Himalayan region of West Bengal lies in the
fourth quartile of SI. More significantly, the
substitution in the fourth quartile of SI of
the inland northern region of Maharashtra
by the inland central region of the same
state, results in breaking the MaharashtraGujarat contiguity observed in the case of
the fourth quartile on the basis of POVT.
At the other end of the scale, we have in
the last quartile of SI a long contiguous belt
Table 2: Summary of Information Regarding the Quartiles of Selected Variables Across
56 Regions
Ranking Variables
for Quartiles
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
(c)
AVCE
GINIC
POVT
SI
.
PGR
q4
(2)
(3)
(4)
(5).
Rs 38.38
24.92
31.88
0.2583
25.74
23.74
0.3777
13.81
7.42
0.1121
16.77
10.05
0.1023
15.27
10.44
0.2265
22.95
16.18
Rs 43.43
27.01
28.98
0.2862
24.29
25.88
0.4693
37.34
32.84
0.1708
29.89
25.90
0.1333
31.80
27.07
0.2700
24.26
20.86
Rs 46.60
31.49
28.51
0.3039
31.19
30.12
0.5939
28.87
31.28
0.2428
31.78
33.78
0.1498
26.55
30.14
0.3051
28.99
31.62
Rs 76.24
16.62
10.60
0.3855
18.82
20.23
0.8500
20.02
28.43
0.4300
21.60
30.24
0.1807
26.42
32.32
0.3982
23.84
31.31
(1)
GP
Quartiles of Regions
Q:
Qs
~or-
Notes: (L)Each quartile consists of 14 regions ranked according to the variable given in column (1),
(2) a: Upper terminal value, (3) b: Percentage of total rural population in the quartile,
(4) c: Percentage of total rural poor population in the quartile and (5) Totals correspon
ding to b and c relate to 56 regions.
• •• •
Table 3: Quartile Specific Head-Count Ratio According to the Quartiles of Selected
Indicators Across 56 Regions
SI.
No.
Ranking Variable _
1
Head-Count Ratio in Quartile
2
3
4
(1)
(2)
(3)
(4)
(5)
(6)
(1)
(2)
(3)
(4)
(5)
(6)
AVCE
GINIC
POVT
SI
GP
PGR
60.13
43.35
25.26
28.17
32.13
33.14
50.43
50.08
41.34
40.23
40.01
40.41
42.55
45.39
50.92
• 49.96
53.36
51.26
29.98
47.49
66.74
65.80
57.50
61.73
Source: Calculated from Tables 1 and 2.
Table 1: Summary Statistics of the Variables
SI
No
Variables
(1)
(2)
(3)
(4)
(1) Headcount Ratio (Per Cent) (POVT)
11.04
(162)
28.19
(322)
0.1864
(341)
0.0327
(162)
0.1514
(131)
0.0714
(131)
(2) Average (per capita) consumer expenditure
(Rs. per month) (AVCE)
(3) Gini coefficient for entire population (Ratio) (GINIC)
(4) Sen-Index (Ratio) (SI)
(5) Poverty-Gap Ratio (Ratio) (PGR)
(6) Gini coefficient among the poor (Ratio) (GP)
Minimum
Maximum
Mean
CV(1)
Per Cent
CV(2)
Per Cent
56
Regions
(5)
All India
(6)
(7)
(8)
85.02
(322)
47.33
47.03
36.23
36.46
76.24
(122)
0.3855
(513)
0.4223
(322)
0.3982
(323)
0.1807
(323)
45.24
43.98
22.76
23.59
0.2839
0.3056
14.30
15.10
0.1810
0.1787
51.36
52.04
0.2683
0.2792
21.32
20.86
0.1286
0.1399
22.03
21.83
.
Notes: 1 Mean value of the variable for 56 regions in column (5) is an unweighted average of observations for the 56 regions.
2 Mean value of the variable for All India in column (6) is derived from the all-India tables covering all the 62 regions.
3 CV(1) in column (7) gives the unweighted coefficient of variation across 56 regions with respect to the unweighted mean for 56 regions
given in column (5).
4 CV(2) in column (8) gives the unweighted coefficient of variation across 56 regions with respect to the all-India mean values given in
column (6).
5 Numbers within brackets in columns (3) and (4) refer to the code numbers of the regions having minimum and maximum values of
the variable, respectively.
Economic a°d Political Weekly Special Numoer November 1988 .
2397
in the north and north-west. We have in this
belt two of the three regions of Jammu and
Kashmir, two regions of Rajasthan and the
whole of Punjab and Haryana each com
prising of two regions. A ranking by POVT
extends this contiguous belt further south
by taking-in the northern region of Madhya
Pradesh.
At this point, it would be useful to com
pare the rankings in terms of AVCE on the
one hand and that in terms of the two
measures of poverty on the other. Given the
inverse relationship between AVCE and the
two measures of poverty, one would expect
that in general the regions falling in the
lowest quartile when ranked in terms of
AVCE will be located in the highest quar
tile when ranked in terms of either POVT
or SI. This is partly confirmed by our maps.
In the eastern segment, seven out of the eight
regions falling in the lowest quartile in the
AVCE ranking are also located in the highest
quartile when ranked in terms of POVT.
Similarly, in the north-west, the four regions
of Punjab and Haryana and 2 regions of
Rajasthan located in the highest quartile in
terms of AVCE are in the lowest quartile
when the regions are ranked in terms of
POVT or SI. However, it is noteworthy that,
while the extreme poverty belt turns south
to take in parts of Andhra Pradesh and the
whole of Tamil Nadu when we consider
AVCE, it extends further west to bring in the
regions of Maharashtra when we consider
either POVT or SI.
Moving away from extreme values, we
find in the second quartile on the POVT
ranking two sets of five contiguous regions.
The first of these covers the entire state of
Uttar Pradesh in the north. The second set
consists of all the three regions of Andhra
Pradesh and two regions of Karnataka in the
south.
When we consider the ranking in terms
of relative inequality in private consumption
expenditure or GINIC we find that eight of
the fourteen regions in the highest quartile
form one contiguous belt—essentially cover
ing the dry regions of Rajasthan, Madhya
Pradesh and Maharashtra.
I\vo more comments are necessary on the
foregoing regional patterns.
The wide east-west belt in the highest
quartile of POVT contains four of the six
regions of Maharashtra which were severely
affected by drought in 1972-73 so that the
wide geographical spread in the west could
be year-specific.
One region from Rajasthan, namely
western Rajasthan is located in the highest
quartile of AVCE and in the lowest quartile
of POVT and SI. This is something of a sur
prise in that this happens to be one of the
driest regions of the country with low levels
of agricultural productivity (see [4]).
VI
Associations among Poverty
Indicators Viewed through
Quartile-wise Cross-Tabulation
In this section, we present and analyse cer
tain broad associations on the basis of the
Economic and Political Weekly Special Number
November 1988
cross-tabulation of results when 56 regions.
are grouped into 4 quartiles of 14 regions
each according to alternative ranking criteria
taken two at a time. This takes us beyond
a summary measure of association like the
product-moment correlation coefficient or
the Spearman’s Rank Correlation coefficient
and provides spatial content to such associa
tions. The broad order of association bet
ween any two given variables would be
reflected in the number of entries along the
diagonal in the 4x4 cross-tabulations in
volving the quartile groups of regions. In
each case, the diagonal will have to be ap
propriately defined taking into account the
expected direction of association between the
two variables under consideration.
For example, when the two variables, say,
the head-count measure of poverty (H) and
the Sen-Index (S) are positively associated,
the diagonal elements will be defined as the
set of pairs of (Q,(H), Q,(S), Q2(H), Q (S),
Q3(H), Q3(S) and Q4(H), Q4(S» where
Q,(x) is the ith quartile (i = 1, 4) for
variable x (x=H, S). When the variables are
inversely associated as for example, head
count measure (H) and per capita consump
tion expenditure (C), the diagonal elements
will be defined as the set of pairs of (QjfH),
0,(0, (Q2(H), Q2(C), Q3(H). Q3(Q, Q2,(C);
and Q4(H), Q,(C)). Once the diagonalis ap
propriately defined, the off-diagonal
elements would indicate the regions which
do not conform to the expected association.
In understanding such ‘non-conforming’
regions, we turn to quartile-wise ranking of
those regions in the dimension(s) of the
2399
x : Cumulative percentage of the population having PCT E X or less
y: Cumulative percentage of total consumer expenditure accruing population with PCTE x or less
appropriate auxiliary variables). When a given
‘non-conforming’ region ‘defies’ such
analysis, we call it an ‘outlier’.
The following points may be noted about
our method of analysis. The quartile-wise
grouping of regions adopted in this exercise
ignores the level of the variable fofa given
region a well as the intra-quartile rank of
that region. In principle, it is possible to
define finer fractile-groups for the cross
tabulation. In the limit, one would end-up
considering pairs of ranks of a given region
with reference to the ranking variables. A
summary measure of rank-order correlation
coefficient is based on such pairs of ranks.
However, the quartile-wise grouping, besides
providing a spatial content, helps one to
meaningfully bring in the auxiliary variables
in a manageable fashion.
In each cell in a given two-way tabulation,
besides specifying the region-codes belong
ing to that cell, we provide two other in
dicators of the relative importance of the
regions in that cell. These indicators are:
(a) percentage share in total rural popula
tion (for 56 regions) located in the regions
figuring in that cell; and (b) percentage share
in total rural poor population (for 56
regions) located in the regions figuring in
that cell.
The following discussion is organised in
two parts. The first part examines the rela
tionship between the head-count measure of
poverty (POVT for short) on the one hand
and the a>.‘rage (per capita) consumer
expenditure (AVCE) and the Gini coefficient
(GINIC) summarising the size distribution
of AVCE. The second part focuses on the
inter-relationships between the Sen-Index of
Table 4: Quartile-wise Cross-Tabulation between Head-Count Ratio (POVT)
and Average Per Capita Consumer Expenditure (AVCE)
Across 56 Regions
Qi
Cd
Q
d.
o
Q,
<?z
U
; —
412,422
132,421,423
a
b
—
■ — ■
4.89
4.71
5.04
5.30
311,511,221,
222,322,323,
113,332,334
20.02
28.43
411,341,534,
211
10.43
9.07
312,431,321,
333
11.45
13.27
521,523;525,
526,331
5.03
6.56 •
413,531,224,
212,213,214,
215
20.78
18.07
313,514,432,
522,114
—
9.84
9.86
—
512
223,524
—
1.24
0.99
2.54
2.85
—
Region-code
342
a
b .
0.10
0.08
Region-code
131,532
a
b
0.87
0.58
Region-code
513,515,161,
162,133,533,
225,121,122,
111,112
16.62
10.60
T-
•a
O'
Qi
Qa
Region-code
00
K
•u
oo
Quartiles According to POVT
. Qz
Qz
a
b
—
—
Notes'. (1) Region-codes and description of regions arc given in Appendix Table 1.
(2) a. Percentage of total rural population in the cell.
(3) b. Percentage of total rural poor population in the cell.
(4) Totals corresponding to a and b relate to 56 regions.
(5) Upper terminal values of the quartiles according to the various classifying variable
are given in Table 2.
poverty (SI) and its component elements,
namely, POVT, poverty-gap ratio (PGR) and
the Gini coefficient of the size distribution
of AVCE among the set of the household;
(GP).
As a convention in the rest of the paper,
we refer to the combination of the ith quar
tile according to x and the jth quartile
according to y as the combination (i, j) in
x-y cross-tabulation. For example, the com
bination of the 2nd quartile according to
AVCE and the 1st quartile according to
POVT is referred to as the combination (2, 1)
in the AVCE-POVT cross- tab.
In our earlier study (see [3] and [5] we
found that AVCE had a statistically signifi
cant negative effect on POVT and was the
dominant variable explaining the inter
regional variations in POVT among 56
regions. G1NIC while also being statistically
significant in its partial positive impact on
POVT was relatively less important than
AVCE in explaining the inter-regional varia
tions in POVT. This is also obvious from the
product-moment correlation coefficients of
-0.8259 between POVT and AVCE and
0.2776 between POVT and GIN1C. Conse
quently, we use POVT-AVCE as the main
two-way classification with POVT-GINIC
being used as an auxiliary classification to
isolate the pure outliers.
Given the expected negative relationship
between POVT and AVCE, we define the
diagonal elements in the reverse direction.
Consequently, 1st, 2nd, 3rd and 4th quar
tiles of AVCE are to be paired with respec
tively 4th, 3rd, 2nd and 1st quartiles of
POVT. These diagonal pairs consisting of
31 regions together account for 68.87 per
cent of the rural population and 70.37 per
cent of the rural poor population (see
Table 4 for details). There is a strong
negative association between AVCE and
POVT especially at the lower and the upper
ends of the AVCE scale. This is brought out
by the following empirical regularities. There
are 14 regions in the lowest quartile of AVCE
out of which 9 are located in the highest
quartile of POVT. Similarly, out of the 14
regions in the highest quartile of AVCE, 11
are found in the lowest quartile of POVT.
In principle one could raise an auxiliary
question: whether the strong negative
association between AVCE and POVT is due
to a combination of a strong negative
association between AVCE and GINIC and
a positive association between GINIC and
POVT. For this purpose, we examine
whether the regions located along the
diagonal (as defined) of AVCE-POVT cross
tab are also found in the same location in
the AVCE-GINIC cross-tab. A comparison
of Tables 4 and 5 indicates that only 2 out
of 9 regions in (1, 4) combination and 2 out
of 11 regions in (4, 1) combination in the
AVCE-POVT and AVCE-GINIC cross-tabs
are the same. This confirms the earlier fin
ding (see [5]) that POVT is governed more
by the level of AVCE than by the relative in
equality in AVCE around that level.
Let us now consider the regions which do
not fall along the inverse diagonal in the
AVCE-POVT cross-tab. We may distinguish
two types of off-diagonal elements. First, we
have those regions which do better in terms
of having a lower head-count ratio than is
indicated by their quartile-wise location
along the AVCE-scale. Let us call them leftof-diagonal elements in view of their loca
tion in the AVCE-POVT cross-tab. Similarly,
we have the fight-of- diagonal elements in
the AVCE-POVT cross-tab which do worse
in terms of having a higher head-count ratio
than is suggested by their quartile-wise loca
tion along the AVCE- scale Using the earlier
notation of (i, j) in the x-y cross-tab denoting
the ith quartile of x combined with jth quar
tile of y, we can distinguish the various cells
in the AVCE-POVT cross-tab as follows:
a) Diagonal-elements: (1,4), (2,3), (3,2) and
(4,1)
b) Left-of-diagonal elements: (1,1), (1,2),
(1,3), (2,1), (2,2), (3,1)
c) Right-of-diagonal elements: (4,2), (4,3),
(4,4), (3,3), (3, 4), (2,4)
■ Let us exami ne the category (b) of the leftof-diagonal elements. It would be plausible
to postulate a positive relationship between
GINIC and POVT for a given level of
AVCE. This would suggest that a region
Table 5: Quartile.wise Cross Tabulation between the Head-Count Ratio (POVT) and Gini
Coefficient for Entire Population (GINIC) Across 56 Regions
Q.
Region-code
Q,
a
$
2
S’ q2
p
S -
<u
■=
S
O'
~
Qj
Qi
4
»
Region-code
342,515,131,
133
1.78
0.94
162,533
Quartiles According to POVT
Qz
Q3
411,341,512,
422,214
17.58
15.08
412,531,534,
211,215
7.51
7.10
212,213
a
b
Region-code
2.27
1.25
161,532,121,
111,112
6.95
3.57
513,225,122
9.05
7.85
413,224
a
b
2.81
1.66
3.20
2.81
a
b
Region-code
Qi
514,132,522
221,331
3.18
3.50
312,421,423,
3.20
4.22
511,222,523,
322
4.34
-6.16
311,521,323,
332
7.70
11.14
525,526,113,
334
4.78
6.91
10.17
11.37
313,431,321
7.49
7.56
432,223,524,
114,333
8.03
8.85
Note: Same as for Table 4.
Table 6: Quartile wise Cross-Tabulation between the Gini Coefficient for Entire
Population (GINIC) and Average Per Capita Consumer Expenditure (AVCE) Across
56 Regions
Qi
Q,
w
<
o
GO
c
•o Q2
oQ
U
C3
O
Qj
Qi
Quartiles According to GINIC
Qj
Qz
Qi
113,334
Region-code
132,221,422
412,511,222,
421,423,322
311,323,332
a
b
Region-code
4.60
5.28
411,341,342,
331
10.95
12.71
312,534,523,
211
6.67
9.82
431,521,321
2.70
4.07
525,526,333
a
b
Region-code
8.61
7.18
514,131,522,
214
9.68
10.69
531,215
4.82
5.57
313,532,212,
213
4.35
5.54
413,432,224,
114
a
b
Region-code
10.25
9.51
512,515,133
1.39
1.23
162,533
13.43
11.46
161,121,111,
112
6.42
6.11
513,223,225,
524,122
a
b
2.73
1.77
2.27
1.25
6.27
3.07
5.35
4.51
Note: Same as for Table 4.
Economic and Political Wc->:!y Special Number November 1988
. 2403
which happens to be in a left-of-diagonal cell
for a given quartile of AVCE is so located
because ranking along the GINIC-scale
places it in the same or lower-quartile com
pared to its location according to POVTquartile. To illustrate, consider a region
located in (1.3) cell in the AVCE-POVT
cross-tab. Our foregoing argument would
imply that it is so located because its rank
ing along the GINIC-scale places it in the
3rd or lower quartile. An exactly parallel
argument would apply to the right-ofdiagonal elements. In this case, a region
located in (4,2) cell in the AVCE-POVT
cross-tab would, on the above argument, be
so located because its ranking along the
GINIC-scale places it in the 2nd or higher
quartile. This reasoning with the help of
Table 6 helps sort out 11 out of 12 left-ofdiagonal regions and 8 out of 13 right-ofdiagonal regions in the AVCE-POVT cross
tab in Table 4.
We are now left with the following offdiagonal regions:
A) Left-of-diagonal regions: 532: Karnataka:
Inland Eastern.
B) Right-of-diagonal: 512: Gujarat: Plains;
514: Gujarat: Dry Areas; 522: Maharashtra:
Western; 523: Maharashtra: Northern;
331: West Bengal: Himalayan.
These may be termed as pure outliers in the
AVCE-POVT cross-tab. In characterising
these regions as pure outliers we are
highlighting the fact that they do better
(category A) or worse (category B) in terms
of their .quartile-wise POVT ranking than
indicated by their quartile-wise rank in terms
of AVCE despite being adversely (category
A) or favourably (category B) placed in
terms of their quartile-wise ranking along
the GINIC-scale.
The existence of these outliers leads us to
a closer examination of the analytical links
between POVT and GINIC. Given the
poverty line and AVCE, head-count ratio
(POVT) is determined by the shape of the
Lorenz curve in the neighbourhood of and
below the poverty line. Notice that GINIC
summarises the Lorenz curve over the entire
range of per capita household consumer
expenditure including those above the
poverty line. This gives rise to the possibility
that the same GINIC, depending on the
shape of the Lorenz Curve may yet yield dif
ferent values of POVT for the same value
of AVCE and the poverty line. In fact, it is
possible to construct Lorenz Curve such that
for given values of AVCE and poverty line
a lower level of POVT is associated with a
higher level of GINIC and vice versa. These
two possibilities are illustrated in Diagrams
1 and 2.
The above discussion offers one possible
interpretation of the cases of ‘pure outliers’
identified earlier by reference to their
quartile-wise ranking in terms of POVT
relative to their ranking in terms of AVCE
and GINIC. This would be that in these
2404
Table 7: Quartile wise Cross-Tabulation between the Sen Index (SI) and the Haed-Count
Ratio (POVT) across 56 Regions
Quartiles According to SI
Q,
Qa
Qi
Region-code
a
b
Region-code
a
b
Region-code
4.74
3.95
—
a
b
'Region-code
—
—
—
0.87
0.89
—
a
. b
—
—
—
—
Q,
J
a.
f Q1
§
O
<
g
1.78
1.32
411,413,531,
224,422,211,
212,213,214,
215
27.24
23.69
514,132
3
° Qa
q4
Qr
—
_
—
—
412,534
_
—
_
5.3
5.20
312,313.431,
432,223.522,
321.114,421,
423
24.50
26.18
523,331
—
—
532,225
342,515,513,
161,162,131,
133,533,121,
122,111,112
12.03
6.10
341,512
1.92
2.40
524,333
3.50
4.21
311,511,221,222,
521,525,526,
323,322,113, .
332,334 . .
18.10
.26.03
Note-. Same as for Table 4.
Table 8: Quartile-wise Cross-Tabulation between the Sen-Index (SI) and Poverty Gap Ratio
(PGR) Across 56 Regions
Quartiles According to SI
Qt
Qg
422,214
a
b
Region-code
341,342,512,
515,161,162,
131,133,533,
121,122,111
13.85
8.15
513,112
a
b
Region-code
2.92
1.90
—
17.59
15.06
413,224
—
3.75
3.90
—
312.534,432, 221,526
223,522,523,
321,114,421,423
a
b
Region-code
—
—
—
3.20
2.81
—
22.74
24.71
313,431
Region-code
Q.
O
no
CO
c:
Qg
o
<
3
o
Qa
Qj
—
—
a
b
— ■
9.10
8.03
—
411,514,132,
412,331
531,532,225,
211,212,213,215
?4
Qr
—
—
5.29
5.17
• r-
—
_
3.05
4.10
311,511,222,
521,524,525,
322,323,113,
332,333,334
18.55
26.14
Note: Same as for Table 4.
.Economic and Political Weekly Special Number November 198£
cases the precise shape of the Lorenz curve
around and below the poverty line is not ef
fectively captured by the value of GIN1C.
The immediately preceding discussion
enables us to understand the absence of any
systematic pattern in the POVT-GINIC
cross-tab given in Table 6. Further, if we view
the Gini coeffiaenl among the poor (GP)
as a closer proxy to the characterisation of
the shape of the Lorenz Curve around and
below the poverty line, then one would
expect a much stronger association between
POVT and GP than between POVT and
GINIC. This is confirmed by the fact that
the product-moment correlation between
POVT and GP is 0.7914 as against 0.1776
between POVT and GINIC.
Next we propose to assess the extent of
association between the Sen-Index (SI) on
the one hand and its component elements,
namely, the head-count ratio (POVT), the
poverty-gap ratio (PGR) and the Ginicoefficient summarising the size distribution
of AVCE among the subset of the poor
(GP). The informational requirements can
be seen to increase progressively as we move
from POVT to PGR to GP to SI. The cross
tabs between POVT-SI, PGR-SI and GP-SI
are presented in Tables 7, 8 and 9.
Our correlation analysis has shown that
the product-moment correlation is the
highest at 0.9719 between POVT and SI,
followed by 0.9527 between PGR and SI and
the least at 0.8717 between and GP and SI.
This pattern of association gets broadly
reflected in our two-way tabulations in terms
of the predominance of the diagonal
elements with the declining importance of
the number of regions as well as the share
in total rural and total rural poor popula
tion of the diagonal elements when we move
from POVT-SI cross-tab to PGR-SI cross
tab and further to GP-SI cross-tab. This
should be obvious from the following Table
10 (see lines 1, 2 (a) and 2(b) across column?
(3) to 5).
Secondly, it would be useful to note the
number of regions which are in the same
diagonal quartile when ranked by SI and all
of in component elements, POVT, PGR and
GP. From the foregoing summary table, it
should be clear that POVT-SI cross-tab pro
vides us with the most comprehensive list of
the diagonal elements. We find that 26
regions in all comprising 8 out of 12 regions
in (1,1) combination, 6 out 10 in (2,2), 5 out
of 10 in (3,3) and 7 out of 12 in (4,4) com
bination are common in all the three cross
tabs POVT-SI, PGR-SI and GP-SI.
Thirdly, the foregoing summary Table 10
indicates that the percentage share of the
regions along the upper half of the diagonal
is higher in respect of the rural poor popula
tion. This is much more so in (4,4) combina
tion than (3,3) combination (compare lines
4(a) and 4(b) in all the cross-tabs POVT-SI,
PGR-SI and GP-SI. This would clearly result
as a direct consequence of the ranking
‘criterion with respect to POVT but that it
repeats itself in the other cross-tabs in vary
ing degrees basically reflects the correspon
ding association between POVT on the other
hand and the other measures, namely, PGR,
GP and SI on the other.* In other words,
those regions with higher than median-level
head-count ratio (POVT) also tend to have
higher than median level of each of the three
indicators namely income-gap-ratio (PGR),
Gini-coefficient among the poor (GP) and
the Sen-Index (SI) which incorporates the
intensity dimension of poverty.
A conclusion of significant empirical im
portance emerges from the foregoing discus
sion, namely, a very close association (in
terms of ranking) between the informa
tionally least demanding and hence possibly
most-robust but crude head-count measure
(POVT) and the informationally most
demanding and conceptually most sophi
sticated Sen-Index (SI).
It can be argued that our conclusion
strictly applies to the broad quartile-wise
ranking that we have adopted. However, a
logical extension of the quartile-wise rank
ing consists of ranking each region accor
ding to POVT and SI and working out a
summary measure of the Spearman’s rank
correlation coefficient. The detailed rank
order of each region according to POVT and
SI appear in Appendix Table II columns (8)
and (9). The rank correlation coefficient bet
ween POVT and SI based on these rank
orders works out to 0.9749, which inciden
tally, is not very different from the product
moment correlation coefficient of 0.9719
between POVT and SI noted earlier. We can
thus safely conclude that if we are interested
Table 9: Quartile-wise Cross-Tabulation between the Sen-Index (SI) and Gini Coefficient
AMONG THE POOR (GP) ACROSS 56 REGIONS
Q,
341,342,512, . 514,132,422,
515,161,131,
215
133,121,122,111
Region-code
Qi
p
u
a
b
Region-code
</>
3
O'
q3
11.58
6.90
513,162,533,
112
a
b
Region-code
5.19
3.15
—
,3.69
3.54
411,531,225,
211,212,213
214
22.32
19.05
532,224
a
b
Region-code
. —
■. —
—
2.32
2.00
413
—
—
1.56
1.31
c'
oo
Quartiles According to SI
Qi
Qa
■
Q<
a
b
Qi
—
. —
— .
—
223,522 •
—
—
3.62
3.95
412,312,431,
432,114,421,
423,331
18.73
20.43
313,534,523,
321
9.43
9.40
526
0.67
0.92
511,221,222,
113
5.50
7.71
311,521,524,
525,322,323,
332,333,334
15.43
21.61
Note: Same as for Table 4.
Table 10: Summary Table Regarding SI and its Constituent Elements
SI
No
(1)
Particulars of Item
(2)
Cross-Tabulation between
POVT-SI
PGR-SI
GP-SI
(3)
(1) No of regions along the diagonal
44
(2) Percentage share of regions along the diagonal in
(a) Total rural population
81.97
(b) Total rural poor population
82.00
(3) No of regions along (3,3) diagonal
10 .
(4) Percentage share of regions along (3,3) diagonal in
(a) Total rural population
24.50
(b) Total rural poor population
26.18
(5) No of regions along (4,4) diagonal
12
(6) Percentage share of regions along (4,41 diagonal in
(a) Total rural population
18.10
(b) Total rural poor population
26.03
(4)
(5)
42
35
72.73
74.06
10
68.06
67.99
8
22.74
24.71
12
18.73
20.43
9
18.55
26.14
15.43
21.61
Source: Tables 7, 8 and 9.
Economic and Political Weekly Special Number
November 1^88,
2405
in the ranking of regions according to the
intensity of poverty as measured by SI, we
can approximate it fairly closely by ranking
the regions according to the head-count ratio
(POVT).
VII
size of the average consumer expenditure per
capita (AVCE) and rises progressively when
the regions are ranked according to the SenIndex (SI) and its constituents. GINIC is the
only ranking variable for which the head
count ratio does not exhibit the monotonic
relationship across quartiles.
type, we have five regions (viz, Gujarat
plains, Gujarat dry areas, Maharashtra
western, Maharashtra northern and West
Bengal Himalayan) which do worse in terms
of their quartile-wise POVT ranking than is
indicated by their quartile-wise rank in terms
of AVCE despite being favourably placed
along the GINIC dimension.
(5) A mapping of regions in extreme poverty
(8) We also find a much stronger associa
judged by reference to the lowest quartile of
tion between POVT and GP than between
AVCE reveals a long belt extending from
POVT and GINIC. This may be interpreted
the eastern plains of West Bengal to
in terms of GP being a closer proxy to "the
Kanyakumari
in
Tamil
Nadu
with
only
one
(1) Among the dimensions of poverty con
region of Andhra Pradesh breaking the con characterisation of the Lorenz curve around
sidered in this study, the (unweighted) co
and below the poverty line.
tiguity
in this east-south belt. However, when
efficient of variation across 56 regions was
judged by reference to the highest quartile
(9) We observe a very close association in
the highest for the Sen-Index (SI) and the
of the head-count ratio (POVT), we observe
the ranking of regions according to POVT
lowest for the Gini coefficient (GINIC) sum
an
east-west
belt
spread
all
the
way
from
and according to SI so that if we are in
marising the size distribution of per capita
West Bengal to Rajasthan and taking in terested in the ranking of regions according
consumer expenditure.
regions of Orissa, Bihar, Madhya Pradesh,
to the intensity of poverty wc can appro
(2) Even though within a region, the relative Maharashtra and Gujarat.
ximate it fairly closely by a ranking in terms
inequality in the size distribution of per
of POVT. This finding holds not only in
capita consumer expenditure among those
(6) The head-count ratio (POVT) has been terms of the broad quartile-wise ranking, but
below the poverty line (GP) is much lower
found to be much more closely associated
also in terms of a detailed ranking leading
than that relating to the size distribution
with the level of region specific average con to a very high Spearman’s rank correlation
among the entire rural population (GINIC),
sumer expenditure per capita (AVCE) than
coefficient.
the relative variability across 56 regions is
with the relative inequality in the region
[We are grateful to Bacchi Ram for preparing
much higher for GP than for GINIC.
specific size distribution of AVCE as
the maps and diagrams.]
.measured by GINIC.
(3) For all the dimensions of poverty (with
the exception of the Sen-Index), the regions
References
(7) We located two types of ‘outlier’ regions.
are densely concentrated over a relatively
In the first type, we have one region (viz,
[1]
Chatterjee,
G
S and N Bhattacharya (1974):
narrow range of the variable in the second
Karnataka inland eastern) which does bet
‘Between States Variation in Consumer
and the third quartiles.
ter in terms of its quartile-wise POVT rank
Prices and Per Capita Household Con
(4) Quartile-specific average head-count ing than is indicated by its quartile-wise rank
sumption in Rural India’, Sankhya, Series
ratio progressively declines across quartiles
in terms of AVCE despite being adversely
C, Vol 36, Parts 2 and 4, pp 337-368..
when the regions are ranked according to the
placed in terms of GINIC. In the second - [2] Kakwani, N C (1976): ‘On the Estimation
Conclusions
. The major findings of this paper may now
be recapitulated.
Appendix Table 1: Agricultural Regions within Each State
SI Region Description
No Code of the Region
(0)
(2)
(I)
411
412
SI Region Description
No Code of the Region
Composition (Districts/Tehsils)
of the Region
(0)
(3)
Andhra Pradesh (State Code 41)
Coastal
Srikakulam, Visakhapatnam, East
Godavari, West Godavari, Krishna,
Ongole, Guntur and Nellore.
Inland, Northern Adilabad, Karimnagar, Nizamabad,
511
10
512
Nedak, Warangal, Khainmam,
3
413
Nalgonda, Hyderabad-and
Mahbubnagar.
Inland, Southern Kurnool, Anantapur, Cuddapah and
Chittoor.
4
341
Plains
5
342
Hills
6
311
Southern
7
312
Northern
8
313
Central
.Assam (State Code 34)
Lakhimpur, Sibsagar, Darrang,
Nowgong, Kamrup, Goalpara and
Cachar,
Mikfr Hills, North Cochar Hills and
Mizo Hills.
Bihar (Stale Code 31)
Santal Parganas, Dhanbad,
Hazaribagh, Palamau, Ranchi and
Singhbhum.
Purnea, Saharsa, Darbhanga,
Muzzaffarpur, Champaran and
Saran.
Bhagalpur, Monghyr, Patna, Gaya
and Shahabad.
(3)
(2)
(1)
9
Composition (Districts/Tehsils)
of the Region
Gujarat (State Code 51)
Banas Kantha (tehsil Danta), Sabar
Eastern
Kantha (tehsil Khed Brahma,
Vijayanagar, Bhiloda and Meghraj),
Panch Mahals (tehsils Limkheda,
Dohad, Jhalod and Santrampur),
Vadodara (tehsils Naswadi,
Tilakwada, Chhota Udaipur and
Jambugam), Bharuch (tehsils
Ankleshwar, Valia, Jhagadhia,
Dodipadar, Sagbara and Nandej),
Surat (tehsils Mahuva, Vyara, Valod,
Songadh, Mandvi, Uchhal, Nizar,
Mangrol, Palsana and Bardoli),
Dangs (tehsil Dangs) and Valsad
(tehsils Dharampur, Chikhli,
Bansada, Umbergaon, Valsad and
Pardi).
Plains, Northern Sabar Kantha (tehsils Prantij,
Modasa, Nimatnagar, Malpur,
Bayad and Idar), Mahesana (tehsils
Sidhpur, Patan, Mahesana, Kheralu,
Visnagar, Vijapur, Kadi and Kalol),
Ahmedabad (all tehsils),
Gandhinagar (all tehsils) and Kheda
(all tehsils).
(Contd)
24J6
Economic anu Political Weekly
/
Special Number November 1968
/ .......
•
V’ ’
•
■
!
poverty and Planning
Amitabh Kundu
The proposal to dispense with the use of the poverty index in funds
disbursal, on the ground of difficulties :r its computation, amounts in a
way to delinking poverty from planning. Simplification and
standardisation of the methodology of measuring poverty should therefore
receive the urgent attention of experts if ad hoc political pressures and
vested interests are not to take over under cover of the New Economic
Policy.
ELIMINATION or reduction of poverty has
been an avowed objective in Indian planning,
it least since the early 60s. The plan
jocuments have often noted the poverty
trends in the country and tried to assess the
performance in certain programmes and
sc^Ktes through thedecline in the percentage
>!^P.pIe below the poverty line. Concern
for poor has figured explicitly in the manifesto
of various political parties as also in the
policy discussions. Despite all these,
measurement of poverty has remained the
prerogative of a few experts, the issues
becoming more and more complex over time
among, them.
Issues in Measurement of Poverty
The EPW Research Foundation (1993)
reeds to be complimented for documenting
:he studies relating to measurement and
(partly) explanation of poverty in a
thronological and comprehensive manner. It
has also highlighted the major issues
leliberated upon by the Expert Group (EG)
on poverty, set up by the Planning
Commission (1993). Discussion on some of
these and related issues in the context of the
studies on measurement of calorie deficiency
certain nutritional norms would be
'■:sW for a better understanding of the
toverty debate in the country.
The hiatus between understanding the
■oncept of poverty at the theoretical level
rid at the level of implementing the antiloverty programmes has resulted in myths
W misconceptions regardingthe methodology
if measurement. The most common myth is
hat poverty is linked directly with
Malnutrition in such a manner that the poor
an be identified as people whose calorie
•■'lake falls short of certain normatively
etermined level. It can easily be demonstrated
hat it is not so and can be argued that it
Mould not be so. Scholars have pointed out
hat the actual consumption basket and,
•herefore, the calorie intake is a matter of
■hoice of the individual or the household
'hich is outside the domain of public policy
hlinhas, Kansal and Jain 1992and Dandekar
®81], The individuals settling for a lower
Gnomic and Political Weekly
level of calories by choice, despite their
having the adequate purchasing power,
should not be considered as poor The
people, who for the sake of having a larger
amount of non-food items or expensive
calories through milk, meat, etc, voluntarily
consume less calories than the normative
level, should not be subjects of concern for
the government. Poverty, thus, gets
delinked from actual deficit of calories or
malnutrition as it is supposed to reflect the
lack ofcapacity to buy the minimum amount
of calories.
The second popular misconception is
that poverty relates to certain exogenously
determined level of income or consumption
expenditure and has no connection
whatsoever with normative calorie intake
or the concepts of malnourishment,
undernourishment and hunger. This
position is also erroneous since it is not
possible to determine poverty without
referring to a consumption basket and
expenditure level, worked out using certain
dietary requirements. Any other method of
working out the normative level of
consumption expenditure is likely to be
fraught with greater subjectivity and risk.
The methodology for estimating poverty
in India adopted by the PlanningCommission,
as also by individual researchers, strikes a
balance between the two approaches noted
above. The first step in this exercise is to
identify certain normative level of calorie
requirement for an averageindividual, based
on age/sex composition, nature of work,
environment, etc, of the concerned population.
TheCentral Government Employee's Second
Pay Commission (1957-59), for example,
had taken the norm of 2,700 calories and
55 grams of protein as an adequate diet per
day for an average adult engaged in
moderate activity. Dandekar and Rath
(1971) had assumed the calorie input of
2,250 units per person per day as adequate
both for rural and urban areas, although
they did not use it formally to work out
the poverty line expenditure. Based on the
norms recommended by the Nutrition
Expert Group of ICMR (1968) and the age,
sex and workforce structure, estimated for
June 25, 1994
the year 1972-73 (by using the data from
the National Sample Survey (NSS) and the
Expert committee on Population), the Task
Force set up by the Planning Commission
worked out the nutritional requirements
as 2,435 (or 2,400) calories per capita pel
day in rural areas and 2,095 (2,100) calories
per capita per day in urban areas [see
Planning Commission 1979]. It may be
mentioned here that the Joint Expert
Committee set up by FAO, WHO and UNU
had placed the standard requirement at
2,700 calorics [World Health Organisation
1985].
The second step in the measurement of
poverty is identification of a consumption
basket including both food and non-food
items that would ensure the required amount
of calories. The value of this basket would
then give the minimum consumption
expenditure below which people would be
classified as poor. The conversion of the
stipulated norm of calories into an
expenditure level, using a medically
presenbed consumption basket, has been
considered inappropriate. It is argued that
the consumption habits of the population
should be taken as data, beyond the scope
of alteration, at least in the short mn. The
Planning Commission, therefore, converted
the calories into consumption expenditure
by taking the actual consumption pattern
of the population from the NSS. The poverty
line expenditure is, thus, partly normative
and partly behavioural.
In a well researched paper, Minhas (1991)
has proposed a new behaviour-based approach
in place of the traditional methodology using
calorie norms determined by “nutrition
experts". He shows that self-perceived level
of adequacy of food, at which households
start giving free meals to others (net of
receiving such meals), would provide a better
cut-off point for identifying the malnourished
people in India. The limitations of collective
behaviour in fulfilling given normative
requirements are, however, too well known
to merit an elaborate discussion here. One
may only argue that leaving the provision
of basic necessities of life totally to
individuals's choice would imply tremendous
confidence in the optimality of human
behaviour in a market place. This would
erode the basic logic of any public
intervention. The dominant view among the
professionals in the field is that the minimum
level of caloric requirement should be
determined by nutritional experts based on
biological and medical considerations. The
actual consumption pattern must, however.
be left to people’s choice.
It is important to note that the poverty level
expenditure is adequate to purchase not only
the minimum calorie requirements but also
the associated non-food items for a person
with average consumption pattern (of his/her
1 sao
t.
group) in 1973-74. On the basis of the poverty in the paper by EPW Research Foundation
line determined by the methodology discussed (1993) and consequently, this is avoided
above, the poor would then be identified as here. One is happy to note that the expert
the people who.usingtheaverageconsumption group (EG) set up by the Planning
preferences, are not in a position to buy the Commission has tried to resolve some of
minimum level of calories and the associated the controversies by accepting the norm of
level of non-food items. This implies that an 2,400 calories for rural and 2,100 calories
individual, incurring more than the poverty for urban areas with reference to the
line’expenditure but having a consumption consumption pattern in 1973-74. This
pattern different from the average of his/her implies that the standardised commodity
group would not be treated as poor-, even basket corresponding to the group of people
if his/her caloric intake is less than the around the poverty line, ensuring the
stipulated amount of calorics in the year
normative requirement.
The linkage between poverty and 1973-74 at the national level, is to be
nutrition gets blurred further when poverty updated by using the price indices for other
is estimated for other years by updating the years. The EG argues that in view of easy
base year poverty line by a price index. The availability and transparency, the consumer
non-poor in any other year would be able price index for agricultural labourers should
to buy the minimum calories only if they be used for updating the rural poverty line.
hold the average consumption preference For urban areas, a simple average of the
(of the corresponding population group) in consumer price index for industrial
the base year. It is, therefore, possible that workers and that of urban non-manual
some persons above the poverty line, with workers is the recommended index. The
the average consumption pattern of their EG also overrules the necessity to adjust
group in the year, say 1987-88, would have the NSS data to bring the aggregative
calorie intake less than the normative figure at par with the CSO estimate at the
requirement In this context it is interesting national level. It, thus, accepts the criticism
to note that the percentage of people below [Minhas and Kansal 1989] that the pro
the poverty line has declined from 57.6 to rata adjustment of the NSS data, as done
32.4 in rural areas and from 49.9 to 20.7 by the Planning Commission, results in
during 1977-87, as per the official estimates gross underestimation of poverty, since
of the Planning Commission. Paradoxically, the items figuring importantly in the
however, the percentage of people consum consumption basket of the poor are not
ing calorics less than the norm has increased underenumerated in NSS.
from 57.8 to 65.8 and frorr 49,3 to 56.8 in
Another important issue in the current
rural and urban areas respectively.
poverty discussion is the use of state specific
There is no reason why the consumption poverty lines for obtaining state level
pattern or the expenditure level in the base estimates of poverty. Planning Commission
year should be considered sacrosanct for all has been estimating the number of persons
future years for measuring poverty. In that below the poverty line in different states
case, th? consumption basket determined by using the same poverty line—one for
using medical or nutritional norms can also the rural and another for urban areas. Given
be taken for computing the base year figure the differences in consumption pattern and
for consumption expenditure, viz, the poverty price levels, the estimates could be
line. The above method of updating the erroneous. Minhas, Jain and Tendulkar
poverty line for estimating the number of the (1991) improved upon the figures by
poor is, therefore, questionable as it docs not building up state specific poverty lines
take.into consideration the changes in taking the all-India figure as hundred. They,
consumption habit of the population in thus, made suitable adjustments for the
determining the normative food basket, differences in the cost of'living (of middle
making the poverty-nutrition linkage a bit category households) but did not take the
more hazy. This point has been discussed differences in the consumption basket of
below in greater detail, using the recent the relevant households across the states
empirical data.
into consideration. The EG also has
recommended construction of state specific
Methodological Controversies
poverty lines, using the commodity basket
of (he middle level consumers at the national
There has been interesting and useful level, to reflect the differences in the cost
debate on (a) the question of alternate price of living in different stales.
indices used for updating the poverty line
It may be mentioned that the all-India
[Minhas, Jain and Tendulkar 1991a and figure of the number of people below the
1991b] and (b) on adjusting the NSS poverty line, obtained by aggregating the
consumption data for different fractiles to state level figures, is higher than the direct
bring the aggregative figure at par with that estimate by using the uniform poverty line
of Central Statistical Organisation [Minhas [Minhas, Jain and Tendulkar 1991], This
and Kansal 1989 and Kundu 1993]. An implies that the increase in the number of
excellent overview of the debate is available poor in states with a higher cost of living
1570
I summ.
is larger than the decrease in the number in
norm c
states with a lower cost of living
not rcl
Scholars have argued that besides making
require
the necessary adjustments for the
[Sukha
differences in price levels, it would be
is taker
important to take the variation in
dii-tribu
consumption preferences into consideration
average
in determining the slate specific poverty
populat
lines. Serious differences exist on this issue
since a
even among the members of the EG. It is
occurs
easy to sec that the states where people use
(1989) ;
cheaper sources of calorie (like bajra or
metabol
jowar) or spend a larger percentage of
standarc
expenditure on foodgrains, the stipulated
lower iir
nutritional norm would be obtained at a
Three, i
lower level of consumption expenditure. Use
requiren
of an identical consumption basket may,
$ocio-cu
therefore, be taken as a factor, overestimating
can be
poverty in these states. It is, however, well
[Plannin
known that most of these slates are in the
that by c
lower rungs of economic development.
body anc
Consequently, recommending a lower
human i
poverty line for them would imply a reverse
discrimination in the methodology for ^|rger an
Wffe,
furth
estimating poverty. This might result in lower
calories
plan allocation and lower benefits to them
happens
under the ami-poverty programmes, which
various
would not be desirable. This would be viewed
Minhas t<
as penalising a stale for having certain kind
that “the
of consumption preferences. Also, a
Indian po
recommendation that the minimum calorie
differently
requirement (and, therefore, the poverty line)
culture an*
must vary from state to state and over time
For instar
due'to the differences in environmental
;eem to n
conditions and changes in the nature of
sse at a
economic activities, would make the
aergy ini
compulation of poverty far more difficult.
‘•here cc
To keep the methodology simple and
considerec
acceptable to the body of planners and
good’ livj
administrators in the country, it would be
seals outs
necessary to maintain the calorie norm
iem to o
uniform across the states and over the years.
ptual con;
The EG has supported this viewpoint. It
i-’ousehold;
argues that the observed all-India consump
rould no
tion pattern of the 20 to 30 per cent population
ftpenditurc
around the poverty line in 1973-74 “should
revea
constitute the state specific weighting
T®orielev
diagram... The implicit reasoning underlying
Wr consul
the procedure is that any consumer with
goods
(real) income equal to the poverty line will
^s. It can
be able to buy a normatively fixed bundle
•tt above th
which is common to 611 consumers and
consume
invariant over time”.
•hough hl
'*abovetl
Questioning the Calorie Base
■>on with
terms) i
Despite these bold attempts to resolve the
'1 above
controversies pertaining to the measurement
' delibera’
of poverty and make it an analytical tool for
^es, ihc
planned intervention and resource allocation.
'■^questii
the EG has failed to resolve the issue of
.pts that be
poverty-nutrition linkage in any satisfactory
^on". vi
manner. The thrust of criticism on poverty
J55hold, c<
measurement in recent years has been on the
j J8h free
calorie base for determining the poverty hne.
,P°int fm
viz. the critical minimum consumption
'tentpers
expenditure. The arguments may he
Economic and Political Weekly
^°mic an
1
I summarised as follows: One. the calorie more relevant exercise than measuring
The analysis of the changing consumption
’ norm considered in the poverty studies does poverty, using fixed norms. Six, the direct pattern in recent years, thus, indicates that
taking I not reflect the minimum but the average question on hunger asked by NSS gives a measuring poverty by taking a fixed level
requirement of a healthy active person figure of underfed population much less than of consumption expenditure in real terms,
r the
without considering its changing nutrition
did be ■ [Sukhatme 1977]. Two, if the poverty line (hat estimated through poverty calculations
]$ take’ - only as an average of a probability
The above criticisms clearly indicate that coni' nt would grossly underestimate poverty.
jn in
distribution and a certain deviation from the the exercise of working out the minimum Just because people having adequate
eration
average is considered allowable, the poverty calorie requirement, undertaken in the early purchasing power have voluntarily chosen
•overly
population would come down significantly, 60s or 70s. has not been conclusive In view lesser calories than the normative require
s issue
since a large concentration of population of the unresolved controversy, the EG has ment docs not imply that they should be
3. Il is
occurs around the poverty line. Sukhatme recommended that the poverty figures should considered as non-poor Demonstration
pie use
(1989) argues that the average of the basal not be used for the purpose of allocation of effects, particularly in rural, backward and
ajra or
metabolic requirement (BMR) minus two planned funds and other important policy tribal areas, have created demand for a
age of
standard deviations should constitute the decisions. This is an extremely unfortunate number of non-food items [Misra and Kundu
pulated
1980]. This is not a case of an aberration of
lower limit to which all individuals can adjust. proposition. The most important objective
Three, no single estimation of minimum for measuring poverty will be lost if it ceases an individual but of group behaviour,
•ire. Use
requirement is possible for it depends on to be an important instrument of planned reflecting the dynamics of social change.
cl may,
Our concern for the nutritional norm and
socio-culiural norms and personal habits that intervention or resource allocation.
imating
it is important that the per capita respect for the changing social behaviour
can be adjusted if necessary. Sukhatme
er. well
Planning Commission 1993] has suggested consumption of cereals has gone down from must prompt us to recalculate the poverty
: in the
15.4 kgs to 14.0 kgs per month during 1970- line at regular intervals so that it ensures
ihat by controlling the relationship between
xxiy and mind, it is possible to create better 89 in rural areas, as per the NSS data. The the minimum level ofcalories. In other words,
reverse iuman machines capable of producing corresponding figures for urban areas arc there is a case for strengthening the poverty
11.4 and 11 0 respectively. The average nutrition linkage to be able to capture all the
arger amount of energy with less calories.
jgy for
people (having the average consumption
in lower He, further, argues that the absorption of calorie intake per consumer unit in rural
to them calories in the body of a poor person areas has also declined from 2,858 to 2,784 pattern of their group) with absolute physical
5, which happens to be low in India due to the during 1973-83. The corresponding figures deprivation within the category of poor
However, in case there is a general
e viewed .arious ailments he/she suffers from. for urban areas arc 2,634 and 2,574. The
,ain kind Minhas loo has demonstrated empirically, information from National Nutrition consensus among nutrition experts that the
Also, a hat “the prevailing genetic pool of the Monitoring Bureau also confirm this caloric norm and the corresponding
-n calorie Indian population seems to be interacting declining trend Barring “green leafy consumption expenditure need to be lowered
Cityline) differently with different environments of vegetables" and "sugar and jaggery”, the down, it would be important to build up a
poverty trend using a revised poverty line
>\er time ailrure and food histories of different regions. average per capita consumption has declined
mmental For instance, the abstemious people of Gujarat for all commodities during 1975-89. A similar for all the earlier years. This may not be very
-eem to maintain their energy balance with trend is noted for nutrients, except in case difficult using the data available with the
lature of
:ase at a much low-er threshold level of of vitamin A [Radhakrishna and Ravi 1992]. NSS. Politically, however, it may be a
take the
This decline during a period recording modest questionable proposition, as»this may be
difficult. nergy intakes than the people of Punjab
• here copious food consumption is increases in total as well as food expenditure seen as an attempt to underplay the magnitude
■iple and
:onsidered as the primary component of must be viewed with concern, specially of poverty. Nonetheless, if that is the only
■ners and
Jood’ living”. Four, many poor take a few because the average consumption figures for way by which there can be a greater
would be
seals outside (net of the meals offered by the food items as also nutrients are acceptability of poverty figures—allowing
■ric norm
em to others) which w-ould push their . significantly less than the ‘Recommended these to be used as inputs in the planning
the years.
ktual consumption level, at least for some Dietary Intake’. Radhakrishna and Ravi process—this must be done. People shifting
ki'-useholds, above the poverty line. This (1992) show that there has been only a from food to non-food items or opting for
'bould not be captured in the NSS
marginal increase in the intake of cereals and moreexpensivecalories (eggs, milk and high
lopulation
Ppenditure data. Five, the data for the recent nutrients for the bottom 30 percent population quality cereals)-should not necessarily be
4 "should
-irs reveal that many persons are opting for between 1972-73 and 1987-88, although their taken as an indication of reduction of
•weighting
‘^lorie level less than the norm by changing real expenditure has gone up substantially. poverty. Distortions in the consumption
underlying
^ir consumption basket in favour of non- The authors explain that “the slower increase habits brought about through marketing
>imer with
Ai goods and for more expepsive calorie in nutrient intake despite an increase in per propaganda, forcing people to buy more
y line will
It can be shown lhat an average person capita income was, however, not due to low of non-food items and expensive calories
ed bundle
'■ <t above the (updated) poverty line in 1987- income and price elasticities of food. In fact, should be viewed with concern, if thereby
imers and
s consumed less calories than the norm, it was found that both the elasticities were they come down to a nutritional level,
^ough he/she had the purchasing power quite high. In analysing the changes in food detrimental to their health. The case of an
* above the norm. Furthermore, an average consumption levels, they found that while individual behaving differently from the
USE
^n with the same purchasing power (in the changes in consumer preferences went general pattern must be distinguished from
terms) in 1973-74, indeed, would have against the consumption ofcereals, economic the changes in the general consumption
resolve th*
above the calorie norm. Since people growth and relative prices contributed to pattern over time. It would, therefore, seem
;asurcment
'*• deliberately choosing a lower level of increases in their consumption. Since the reasonable to calculate the po vertyline afresh
cal tool ft* ^ries, the normative basis of the ‘norm’
adverse impact of taste change on cereal periodically from the latest consumer
■ allocation'
^'•bequestionedempirically. Minhas (1991) intake was more pronounced than the expenditure data available from the NSS,
he issue of
•'4cs that behaviour based “calorie threshold favourable impact of economic growth and using the methodology suggested by the
ratisfactoO'
^rion”, viz, the calorie level at which the relative prices, the increase in per capita Taskforce of the Planning Commission in
on poverty
/beholds cease to be net gainers (of calories) income did not result in increase of cereal
1979. In other words, the poverty line should
been on th*
7ugh free meals, can be taken as the cut- intake. Income distribution was found to have be placed at a level so as to ensure the
. ,vcrtyl>nC’
;Point for identifying the nutritionally no significant role in explaining the changes minimum calorie intake to the average
>-,■ umptio0
l!cnt persons. This, he suggests, is much in per capita consumption of cereals”.
population by taking into consideration his/
s may &
•her in
,^.vrv
25. I*9*
r
■^orruc and Political Weekly
June 25, 1994
.bV' library
1571
her changing consumption habits.
Associating poverty with a distribution
and not with an expenditure level would
make the concept extremely complex and
subjective. The possibility of producing better
human machines, as suggested by Sukhatme,
can certainly be explored but the Planning
Commission would find it difficult to adopt
it as a national policy. The need for water
supply, sanitation and health facilities cannot
be over-emphasised but that would have to
be complementary to the provision of
minimum nutrition.
Many would have reservation in treating
the people as being above the poverty line
owing to the meals received from others, due
to their degrading effect on human dignity.
As per the NSS data, the number of free
meals received (net of the meals offered)
is high in households belonging to the
lowest fractile in consumption distribution,
as may be expected. However, for (hose
around the line, the number of meals
received and offered tend to cancel out.
This factor, therefore, would not alter the
poverty calculations.
It is interesting that the ‘calorie threshold’
for the country as a whole, as worked out
by Minhas (1991), is 2,446 per day per
person, w'hich is higher than what has been
taken by the Planning Commission as the
minimum requirement, in working out the
poverty line for 1973-74. His suggestion
regarding treating the thresholds of Gujarat,
Karnataka and Himachal Pradesh—the states
reporting the lowest figures—as the norm,
“to which people from other stales could
(hypothetically) adjust”, would be interpreted
as an attempt to understate nutritional
deficiency. Finally, the dismissal of the
poverty figure due to a much lower estimate
of persons reporting not having “two square
meals a day” would be erroneous. This is
because many would not respond to such a
direct question correctly, due to their socio
cultural inhibitions. The enumerator’s
hesitation in asking such a question and
that of the respondent in announcing
inadequacy of food intake for his/her family
members would greatly undermine the
usability of the data.
Challenge of Measuring Poverty
Presently, (he professional input in the
planning process, particularly in important
matters such as resource allocation, etc, is
extremely limited. The proposal of not
using the poverty index in fund disbursal,
on the ground of difficulties in its
measurement, should throw a challenge to
the academics and p»ctitioners in the field.
This, in a way, would amount to delinking
poverty from planning, possibly under the
impact of the New Economic Policy. For
minimising the influence of ad hoc political
pressures and vested interests, it would be
1572
important for (he professionals to resolve
(heir controversies. Requiring the poverty
index to reflect the complexity of calorie
energy linkage through a set of distributional
parameters or making it a multi-variate
concept by bringing in a large number of
indicators relating to quality of life, etc,
would amount to doing a disservice to. the
concept. This will simply weaken the case
for using poverty as an index for slate
intervention. Simplification and standardi
sation of the methodology for measuring
poverty would be the call of the day. unless
we want to throw the baby with the bath
waler.
1983 Data’. Journal of Income and Wealth,
January.
Minhas, B S. L R Jain and S D Tendulkar
(1991a): ‘Declining Incidence of Poverty in
the 1980s Evidence versus Artifacts’,
Economic and Political Weekly, July 6-13,
—(1991b): ‘Rural and Urban Cost of Living:
1983 to 1987-88, Statewise and All-India’,
Journal ofIndian School ofPolitical Economy,
Vol 3. No 3
Minhas, B S, S M Kansal and L R Jain (1992):
‘Incidence of Poverty in Different States in
India’ in , G K Kadekodi and G V S N Murty
(eds). Poverty in India: Data Base Issues,
Vikas Publishing House. New Delhi.
Misra. G K and A Kundu (1980): Regional
Planning at the Micro Level, Indian Institute
of Public Administration. New Delhi.
Planning Commission (1979)*: Report of the
St>
Th
the
ina
cot
TH1
draf
References
mod
the 1
Dandckar, V M (1981): Measurement of Poverty,
Taskforce on Projections of Minimum Needs
of as
Kale Memorial Lecture. Gokhale Institute of
and Effective Consumption Demand, January.
for t.
Politics and Economics. Pure.
—(1993): Report of the Expert Group on
Dandckar, V M and N Rath (1971): Poverty in
UPC
Estimation ofProportion and Number ofPear,
India—J and II’, Economic and Political
Government of India, New Delhi.
for j
Weekly, January 2 and 9.
Radhakrishna. R and K V Ravi (1993): Nutrition
Euro
EPW Research Foundation (1993): ‘Poverty
Programmes in India: Review and
shorl
Levels in India Norms, EstimatesandTrends’,
Assessments, Centre for Economic and Social
prepa
Economic and Political Weekly. August 21.
Studies, Hyderabad.
const
Sukhatme.. PV (1977): ‘Measurement of Poven^
Kundu, A (1993)- In the Name ofthe Urban Poor:
and U
Access to Amenities, Sage Publications. New
Based on Nutritional Needs’, Bulletin of the
view’
Delhi
International Statistical Institute, Vol XLV11,
in cet
Minhas. B S (1991): ‘On Estimating the
Book 4
litigat
Inadequacy of Energy Intakes: Revealed Food
—(1989): ‘Nutritional Adaptation and
Consumption versus Nutritional Norms’, The
It m
Variability’, European Journal of Clinical
Journal of Development Studies, Vol 28. No 1.
Nutrition. Vol 43.
that In
Minhas, B Sand SM Kansal (1989): ‘Comparison
World Health Organisation (1985): Energy and
wheth
of the NSS and the CSO Estimates of Private
Protein Requirements. Report of a Joint FAO/
of I. ■
Consumption: Some Observations Based on
WHO/UNU Expert Consultation, Geneva.
plan-, v
to do tl1
forced
all of
toon e<
atten.pt
for itself
Havni
now, wii
RELIGIOUS
fight as
i iW
philosop
>o plant
fTo-rura
India is:
V.D. Chopra (Ed.)
■irmei.:
*ay to t
Top commentators, scholars, political
ais gerr
4is draf
analysts dwell upon the sequence of
IK g(
religious revivalism In Pakistan, Afghani
’I the
stan, Bangladesh, Sri Lanka, Central Asia
Pilose rl
and India, and Its Impact on the life and
firing f
polity of the world at large.
•Mia's :i
-trinsi-,
Contributors
Mustri
Asghar All Engineer
D.R. Goyal O Kallm Bahadur
pal is to
Inayatullah Baloch
S.R. Chakravarty and others.
Gelino
hdustri
ISBN 81-212-048 l-X
pp 325 Hard Cover Rs. 350 Paperback Rs. 150
‘‘ S«,y>nolo/:
’“sin di
BOOKS THAT MATTER
FUNDAMENTALISM
IN ASIA
\ -'•mg. ■
■SUPOV
Ph,: 3261060,3282060 Fax : (91J-11-3285914
Economic and Political Weekly
June 25. 1994
Economic Reforms, Employment and Poverty
Trends and Options
Abhrjit Sen
If poverty reduction is to be a serious part of the agenda of economic reforms, the reforms will hav£ to have an
explicitly redistributive content. This will require cuts in subsidies to the rich and also higher taxes to maintain and
increase the expenditure relevantfor the poor. In addition, the old issues of land distribution and provision of universal
primary education and health must be put back on the agenda.
But, more than anything else, it must be recognised that a reforms strategy which aims to withdraw the state from
investment, liberalise finance and thus divert finances from the state to the private sector, liberalise agricultural
trade and thus enrich the rich at the direct cost of the poor and seeks to control inflation and balance of payments
problems through deflation and devaluation is at its root a fundamentally inequitous adventure.
THIS paper is concerned with the possible
impact of the economic reforms undertaken
by the government of India in the 1990s on
the nature and incidence of poverty in India.
The point of departure is the observation
from NSS data that poverty, which had not
showed any time trend at all till the mid1970s, declined significantly between the
mid-1970s and the end-1980 but appears to
have increased again in the 1990s. In other
words, poverty appears to have declined
only in the decade and a half beginning the
mid-1970s during which there was an
explosion in public expenditure leading up
to the fiscal crisis which, among otherthings,
precipitated the economic reforms in 1991.
This suggests that there might be a much
stronger link between public expenditure
and poverty reduction than is usually
appreciated, and this in turn has the
implication that the reforms process may
actually impinge adversely on the poor if its
focus continues to be on the reduction of
public expenditure.
For this reason it is important to identify
the direct and indirect effects of public
expenditure, and of other aspects of the
economic reform policies, on poverty
alleviation. This paper is a very preliminary
effort in this direction. In the first section,
a brief outline of the trends in and structure
of poverty in India is presented, with a view
to identifying the important characteristics
of the poor population. This allows for an
estimation of thelikely effects of such policies
on the materia! condition of the poor and
those close to the poverty line. Since poverty
is found to be closely related to employment
and occupational characteristics, a discussion
on past employment trends and poverty trends
is included in the next section. In the
following section, there is a more detailed
considerationof the recent trends in poverty;
and this is followed by a section which deals
specifically with the statistical determinants
of poverty as well as the relationship between
this and economic growth. Finally, the last
Economic and Political Weekly
section sets out some brief conclusions in
terms ofdifferentpolicy options foreconomic
reform which make poverty reduction an
explicit objective.
//
11
I
I
Long-term Trends and
Profile of Poverty
The Economic Survey 1995-96 has clai med
that “the percentage of India’s population
below the poverty line has declined from
25.94 per cent in 1987-88 to 18.96 per cent
in 1993-94". This claim is basedon estimates
made by the Planning Commission using a
methodology whereby the consumption
distribution obtained from the National
Sample Survey (NSS) are applied to total
estimates of consumption expenditure as
obtained from the Central Statistical
Organisation’s (CSO) compilation of
National Accounts. On this basis, the rural
poverty ratio declined from 28.37 per cent
in 1987-88 to 21.68 per cent in 1993-94
while the urban poverty ratio fell from 16.82
per cent to 11.55 per cent. These figures,
which have been used to claim that there has
been no increase in poverty following the
economic reforms initiated in 1991, have in
turn been challenged by independent
analysts.' The criticism takes two forms.
First, that even using the Planning
Commission method, poverty in 1993-94
was higher than in 1990-91 just before
reforms began and so the comparison with
1987-88, a drought year, gives a misleading
trend. Second, and much more importantly,
that the Planning Commission method is
itself flawed as was pointed out in 1993 by
the high-level Expert Group on Estimation
of Proportion and Number of Poor. Using
the methodology suggested by this expert
group, not only are the poverty figures
much higher, these show that there is no
real trend decline in poverty since around
1986, that poverty increased massively
between 1989-90 and 1992, and that although
Special Number September 1996
poverty fell in 1993-94 this was still higher
than in the immediate pre-reform years
1989-90 or 1990-91.
Some of the issues which arise from these
different estimates are discussed in a later
section. Here, we need to outline the long
term trends in poverty, and for this we present
in Table 1, estimates from a third source
altogether - that compiled by the Poverty
and Human Resources Division of the World
Bank,2 also using NSS data. This source
gives a long series from 1951 onwards, and
the main message which emerges is
important. This is that there was no long
term time trend in poverty from 1950-51 to
1973-74 but that there was thereafter a
sharp decline in poverty till 1986-87. After
1986-1(7, the decline continued at a slower
pace till 1989-90 when it was reversed, with
a. particularly sharp increase in poverty in
1992. Poverty declined again in 1993-94 so
that rural poverty in 1993-94 although higher
than in 1989-90 or 1990-91 just before the
reforms, was at about the same level as in
1986-87. Urban poverty, which had not
increased particularly in 1992, was, however,
lower in 1993-94 than in any pre-reform
year.
These trends are important for a number
of reasons. First, the trend in rural poverty
shows a very close similarity with trends in
agricultural wages. Estimates of real
agricultural wages from a number of sources
also show stagnation till the mid-1970s with
sharp increases thereafter till the end- 1980s
when there is a slow-down again. Second,
the period of declining poverty (mid-1970
to end-1980) was relatively short, and one
which was marked by increasing government
expenditure leading to severe fiscal
imbalances by 1990. Third, that this period
of declining poverty was infact one when
rural poverty declined faster than urban
poverty. Fourth, that rural poverty stopped
falling, and indeed increased, as soon as
fiscal stabilisation was attempted after 1991,
and during this latest period the gap between
2459
rural and urban poverty has again tended to
increase. These trends require explanation
and analysis, and this is the main focus of
this paper. In the remaining part of this
section, we provide a brief outline of the
profile of Indian poverty?
The most comprehensive data on the
structure of poverty remains the information
that can be gleaned from the NSS large
sample survey of 1987-88, since details of
the more recent large sample survey
conducted in 1993-94 are not yet available.
It seems reasonable to assume that in broad
contours the picture that emerges for 198788 remains valid for the early 1990s.
Some of the evidence on the structure of
poverty in India in 1987-88 is provided in
Table 2. The first and most obvious point
to be made relates to the dominantly rural
nature of the poor population. The poor in
rural areas constituted around three-fourths
of the total poor population. This has to be
juxtaposed with the fact that subsequently
urban poverty has declined at a faster rate,
so that poverty has become even more rural
in nature. Within the rural areas, there is also
evidence of greater regional concentration
of poverty, with some backward regions
displaying a very high incidence of poverty
as discussed below.
In the rural areas at an all-India level, the
worst off economic group is that of rural
labour, both agricultural and nonagricultural. This is true both in terms of
depth of poverty and its severity in terms
of distance of average incomes from the
actual poverty line. Within this broad
category of rural labour, casual labour on
non-permanent contracts is the most
susceptible to absolute poverty. There is no
discernible difference in poverty ratios
between agricultural and non-agricultural
casual labourers, which is not surprising,
since the casual labour populations tends to
move between agricultural and nonagricultural occupations as they become
available. The self-employed rural
households, whether agricultural or nonagricultural, tend to experience much lower
levels of economic deprivation than other
rural groups.
Female-headed rural households recorded
a higher than average incidence of poverty,
both in terms of prevalence and severity.
Those rural households classified as poor
tended to have higher than average
representation of adult females and lower
than average representation of adult males.
Also, poor households in general tended to
have higher dependency ratios, so that
children dominated in the number of poor
persons, and were over-represented in the
poor population relative to the total
population. Also, there are definite social
dimensions to material deprivation, with the
category of scheduled castes and scheduled
2460
tribes recording higher extent and severity
of poverty than the general rural population.
In fact, scheduled tribe groups are even
worse off than scheduled castes on average,
and tend to be the most economically destitute
of all the rural population.
The urban areas present a slightly different
picture. Firstly, the poor are more
economically and socially heterogenous.
Thus, the most important occupational groups
among the poor urban population are those
employed in casual labour, as well as a
section of the self-employed. The selfemployed category is highly heterogenous
in urban areas, comprising both highly paid
professional occupations as well as informal
sector low paying activities. The latter
constitutes among the poorest of the urban
population, along with workers employed in
insecure casual contracts. Clearly, the
irregular andinsecurenatureofsuchincomes,
which are also typically low, is the major
source of poverty in urban households.
Scheduled castes and tribes were less
significant among the poor in urban areas
than in rural ones, and there was no real
evidence of regional disparities in urban
poverty. However, the problem of poverty
among female-headed households was far
more serious in the urban areas. Despite this,
the dependency ratio among poor urban
households was slightly lower than among
their urban counterparts.
In terms of regional concentration of
poverty, only two states - Bihar and Uttar
Pradesh - together accounted for 34 per cent
of the total poor population in 1987-88. In
Bihar in particular, there was a large over
representation of poor people, and there is
no reason to believe that this has altered
dramatically. Another six states - Andhra
Pradesh, Madhya Pradesh, Maharashtra,
Orissa, Tamil Nadu and West Bengal accounted for a further 43 per cent of the
poor. For rural poverty in particular, there
was over-representation of the poor in
Madhya Pradesh, Maharashtra, Orissa and
Tamil Nadu. In the states ofGujarat, Rajasthan
and Orissa, scheduled castes and tribes
accounted for more than half of the poor well above their share in total population.
Scheduled tribes, especially in these states,
Table 1: Poverty Estimates 1951-94
NSS
Round
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
27
28
32
38
42
43
44
45
46
47
48
50
H
Rural
PG
SPG
H
Urban
PG
SPG
47.37
43.87
48.21
54.13
61.29
64.24
51.83
48.34
58.86
62.11
55.16
53.26
50.89
45.40
47.20
48.53
53.66
57.60
64.30
63.67
59.00
57.61
54.84
55.36
55.72
50.60
45.31
38.81
39.60
39.06
34.30
36.43
37.42
43.47
38.74
16.05
14.64
16.29
19.03
21.95
25.04
18.44
15.65
19.45
21.69
19.01
17.74
15.29
13.60
13.60
13.88
16.08
17.97
22.01
21.80
18.96
18.24
16.55
17.35
17.18
15.03
12.65
10.01
9.70
9.50
7.80
8.64
8.29
10.88
9.41
7.53
6.71
7.56
9 12
10.26
12.50
8.80
6.71
8.50
10.01
8.78
7.88
6.13
5.53
5.31
5.49
6.60
7.60
10.01
9.85
8.17
7.73
6.80
7.33
7.13
6.06
4.84
3.70
3.40
3.29
2.58
2.93
2.68
3.81
3.27
35.46
36.71
40.14
42.77
49.92
46.19
43.92
43.15
51.45
48.88
47.75
44.76
49.17
44.65
43.55
44.83
48.78
52.90
52.24
52.91
49.29
47.16
44.98
45.67
47.96
40.50
35.65
34.29
35.65
36.60
33.40
32.76
33.23
33.73
30.03
11.14
10.91
13.25
13.83
17.24
15.76
14.65
13.34
18.16
16.31
15.95
13.75
15.83
13.84
13.79
13.29
15.24
16.82
16.8!
16.93
15.54
14.32
13.35
13.46
13.60
11.69
9.52
9.10
9.31
9.54
8.51
8.51
8.24
8.82
7.62
4.82
4.41
5.96
6.29
7.74
7.02
6.40
5.41
8.51
7.25
7.00
5.87
6.75
5.83
6.05
5.17
6.38
6.98
7.19
7.22
6.54
5.86
5.35
5.26
5.22
4.53
3.56
3.40
3.25
3.29
3.04
3.12
2.90
3.19
2.76
Period
Aug 51-Nov 51
Apr 52-Sep 52
Dcc52-Mar53
May 53-Sep 53
Oct 53-Mar 54
Jul 54-Mar 55
May 55-Nov 55
Dec 55-May 56
Aug 56-Feb 57
Mar 57-Aug 57
Sep 57-May 58
Jul 58-Jun 59
Jul 59-Jun 60
Jul 60-Aug 61
Sep 61-Jul 62
Feb 63-Jan 64
Jul 64-Jun 65
Jul 65-Jun 66
Jul 66-Jun 67
Jul 67-Jun 68
Jul 68-Jun 69
Jul69-Jun7O" \x
Jul 70-JunJBOct 72-Spp 73
Oct 73-Jun 74
Jul 77-Jun 78
Jan 83-Dec 83
Jul 86-Jun 87
Jul 87-Jun 88
Jul 88-Jun 89
Jul 89-Jun 90
Jul9O-Jun91
Jul91-Dec91
Jan 92-Dec 92
Jul 93-Jun 94
Noles'. H: head count ratio of poverty; PG: poverty gap ratio; SPG: squared poverty gap.
Source: B Ozler, G Dutt and M Ravallion, 'A Database on Poverty and Growth in India', The World
Bank, January 1996, for estimates up to the 48th round; For 50th round, NSS data has been used
to calculate the estimates using exactly the same methodology as in the rest of the series.
Economic and Political Weekly
Special Number September 1996
were found to be among the most absolutely
deprived and destitute of all Indians.
There is a close relationship between the
extent ofpoverty and patterns of employment
and real wages. In the rural areas in particular
- as argued in the section below/- two factors
are critical (in addition to food prices) in
explaining the incidence of poverty both
over time and across states and regions : the
behaviour of employment including the
degree of diversification away from purely
agricultural employment, and movements in
real wages. For this reason, trends in the
growth and pattern of employment are very
i mportant indicators of the extent and severity
of poverty.
II
Trends after Mid-1970s
The essential point which emerges from
the previous section is that a sustained decline
in poverty is observable only after 1973-74,
and that this process was over by 1990-91.
In fact, the decline was almost certainly a
phenomenon which began after 1975-76.
The years 1972-75 were difficult years with
high inflation was low growth, and the
drop in poverty in 1977-78 as compared
to 1973-74 is to a large extent attributable
to thecomparison of an excellent agricultural
year with a year when output was below
trend. For this reason it is more reasonable
to date the beginning of the poverty decline
to around 1977-78.
As is well known by now, during’rthe
period 1977-91 (and particularly during the
1980s) the Indian economy underwent a
consumption-led boom, spurred on by increa
sing revenue deficits of the government, and
financed in large part by high deficits on the
external current account. This is the boom
which went bust in 1991, laying the basis
for ‘reforms’. But since this boom and bust
cycle is paralleled fairly closely by what
happened to rural poverty, it is worth
recounting some of its more important
features. First, the boom was possible at all
because, with increased access to external
debt and with agricultural growth higher
than the long-term average, the Indian
economy was much better placed on the
supplyside, with both of her two traditional
supply constraints greatly eased.
Second, during this boom it was the
organised sectors of the economy which
grew fastest in terms of incomes and output,
but this growth did not lead to much increase
in organised sector employment. The rate of
growth of organised sector employment
decelerated, significantly, and the 1980s
growth of such employment was, at 1.5 per
cent per annum, much less than the rate of
populationgrowth. Within this.employment
in the private organised sector was the most
sluggish, averaging a growth rate of only
Economic and Political Weekly
0.2 per cent per annum, and there Was slow
growth also of employment provided by the
Central government and its industrial
undertakings. In fact, whatever employment
growth occurred in the organised sector was
provided mainly by state governments and
Certain quasi-government organisations, for
example the nationalised banks! Moreover,
during this period there was also a sharp
drop in labour absorption by the agricultural
sector, and agricultural employment also
grew at a rale substantially below the rate
of population growth, and below rates of
growth achieved in the past at times of lower
output growth. Thus, the rapid growth of
output in agriculture and in the organised
private sector failed to translate itself into
higher direct employment in these important
sectors.
Nonetheless, and this is the third important
point, this decade was characterised by rising
real wages and a fairly sharp drop in both
the incidence and the severity of poverty,
particularly in rural India. According to
calculations made for this paper, using data
from the National Sample Survey and
following the methodology recommended
by the Expert Group on Estimation of
Propotion and Number of Poor, there was
a steady decline in the head count measure
of poverty for the rural population from 56.4
in 1973-74 to 53.1 per cent in 1977-78 to
45.6 in 1983, 38.3 in 1986-87 and to 37.9
per cent in 1989-90/* The urban poverty
ratio similarly fell steadily from 49.2 per
cent to 32.4 per cent during the same period.
This meant that the incidence of poverty
which had fluctuated, positively with
inflation and negatively with agricultural
output, with if anything a positivcunderlying
trend up to the mid-1970s began to decline
thereafter. The most important reason for
this was the fact that real wages of unskilled
labour increased significantly in both urban
and rural areas. Several alternative sources
of data are available for agricultural wages.
and all of these suggest that real agricultural
wages increased by around 50 percent during
the decade - an increase almost double the
increasein labour productivity in agriculture
during this period.
These observations indicate that there were
important changes in the nature of inter
sectoral and other linkages in the economy.
One important point is that, although
agriculture continues to be the largest
employer of the workforce and productivity
increases here are of major weight in the
economy, the rest of the Indian economy
appears to have become progressively less
dependent on the behaviour of agricultural
output during the 1980s. This is evident
from the fact that the period of relative
stagnation in agricultural output 1983-87
was nevertheless marked by high growth
rates in non-agri culture, and, more generally,
econometricevidencesuggests that the earlier
dependence of aggregate economic growth
on the behaviour of the monsoon seems to
have diminished. There are three major
reasons for this. First, the sharp decline in
the employment elasticity of output of the
organised sectors of the economy meant that
increased output in industry and services
today involves a much lower concomitant
increase in the demand for wage goods.
Table 2: Profile or Poverty in India 1987-88
Groups
Sclf-emp agriculture
Self-employed nonagriculture
All self-employed
Agricultural labour
Other labour
Others
Scheduled castes
Scheduled tribes
Female-headed HH
All households
Population
Share
Rural
Per Cent
Poor
44.3
38.3
37.9
12.5
56.7
27.1
8.1
7.9
18.4
10.5
39.0
38.5
62.7
48.7
26.4
56.1
62.7
47.0
44.9
10.8
48.7
41.8
-9.2
4.5
22.9
14.7
38.8
12.1
43.7
5.5
11.7
3.8
100.0
100.0
100.0
Per Cent of Population
Total Poor
Share
Urban
Per Cent
Poor
41.5
68.1
25.9 ■
32.6
533-448.3
43.4
.36.5
Per Cent of
Total Poor
43.0
25.9
•27.3
4.7
17.0
5.0
100.0
Table 3: Composition or Rural Employment (NSS Usual Status Data)
Primary
1977-78 (July-June)
1983 (Jan-Dec)
1987-88 (July-June)
1989-90 (July-June)
1990-91 (July-June) 1991 (July-Dec)
1992 (Jan-Dec) ’
80.6
77.5
74.5
71.7
71.0
74.9
75.7
Special Number September 1996
Males__________
Secondary Tertiary
8.8
10.0
12.1
12.1
12.1
11.2
10.4
10.5
12.2
13.4
16.2
16.9
13.9
13.9
.______ Females
Primary Secondary
88.1
87.5
84.7
81.4
84.9
86.3
86.2
6.7
■ 7.4
10.0
... 12.4
8.1
7.9
7.8
Tertiary
5.1
4.8
5.3
6.1
7.0
5.8
6.0
2461
Secondly, the share of the traditional agro from world prices so .that there were fewer rural poverty fall during a period when
based industries fell sharply so that speculative pressures, and the PDS, along employment in both agriculture and the
agricultural raw materials played a less with the government’s free market organised sector was growing slower than
significant role as industrial inputs than operations, worked to some extent to keep population? That this development, rather
earlier. The boom sectors of the 19SOs - the prices of essential foodgrains under than the somewhat faster growth of
chemicals, consumer durables and high- control. ■
yuciwtr:
agricultural output during the 1980s, was the
tech services - had very little linkage to
In fact, from the point of view of poverty, major driving force behind rising wages and
agriculture. Thirdly, the combination of an an important trend during this period was declining poverty becomes clearer when it
easier import situation and an enhancement that while agricultural prices as a whole is noted that while agricultural growth was
of government operations meant that increased faster than the general price level, regionally diverse (with agricultural Output
government policy instruments were more cereals prices increased slower, so that it was per capita decreasing in many states), the
effective in insulating the non-agricultural possible for real wages to rise without rapid growth of rural non-agricultural
sector from the effects of monsoon increasing product wages correspondingly. employment was a phenomenon which
fluctuations.
This was an important contributory, factor occurred in almost every state in the country,
This last, i e, government policy, operated behind thedecline in poverty which occurred and almost every state recorded rising rural
on both the supply and demand sides. Given during the period. As will be discussed later, real wages and falling rural poverty between
its higher foodgrai n stocks and easier access these equations appear to have changed in
1977-78 and 1989-90. In other words, there
to foreign debt, the government could better the post-reform period. .
is need to modify the conventional view
ensure agricultural supplies to non
Moreover, there was another important among Indian economists that the main
agri culture during periodsof low agricultural development, concerning linkages in the factors determining rural poverty are
output by running down its stocks and by economy. This was the rapid growth of non- agricultural productivity and the rate of
resorting to higher imports of other agricultural employment in rural areas. After inflation. Although both these continue to
agricultural commodities. And, the demand a long period during which agriculture's be very important, the growth of rural non
consequences on non-agriculture of lower share in the labour force had remained agriculture has emerged as an additional
agricultural incomes during such periods constant, there seems to have been a change crucial link from the mid-1970s onwards.
were also better mitigated because, at such somewhere in the mid-1970s when this share
What then explained this growth of rural
times, the government stepped up its revenue began to decline. Since the urban population non-agricultural employment? The Indian
expenditure in rural areas, by expanding has grown faster than total population, this literature on the subject has been dominated
employment programmes and by generating is of course related to some extent with by two debates. First, whether the growth
more self-employment opportunities either . urbanisation. But it is important to note that of rural non-agricultural employment is a
directly through its own rural.development during the 1980s, the pace of urbanisation positive development at all, or is it simply
schemes and/or by instructing banks to extend was in Tact less than in any decade since a reflection of the fact that agricultural
more, credit. Thus; although there were independence. For this reason it may be said employment has been sluggish and certain
features in the nature of organised sector that the really important development was non-agricultural activities have emerged as
growth which tended to weaken hgriculture- thegrowth of the rural non-agricultural sector.
'residual sectors’. Second, to the extent that
nonagricullure linkages, the extent of this
According to NSS surveys, the share of the growth of rural non-agricultural
weakening depended considerably bn a agricultural workers among all male rural employment is not a ‘residual’, is it driven
particular type of government involvement. workers declined steadily from 80.6 per by developments in agriculture or are the
Because of this, the continuing importance cent in 1977-78 to 77.5 per cent in 1983 growth impulses external ? The idea that
of agriculture cannot be wished away easily, to 74.5 per cent in 1987-88 to 71.7 per non-agricultural employment is ‘residual’ is
since a fall in agricultural output can still cent in 1989-90. For rural females this share now somewhat discredited because not only
have severe negative implications for the dropped from 88.1 per cent in 1977-78 to are average wages seen to be higher in such
economy, both in terms of output and 87.5 per cent in 1983 to 84.7 per cent in employment than in agriculture, but, more
inflation.
1987-88 to 81.4 per cent in 1989-90. The importantly, becauseagricultural wages have
Indeed, what is striking about the true significance of this shift is probably increased as non-agricultural employment
experience of the 1980s is that despite the better understood in incremental terms: these has grown suggesting that what is involved
declining dependence of non-agricultural
figures imply that non-agriculture absorbed is a pull factor which tightens the agricultural
sectors on the performance of agriculture, about 70 per cent of the total increase in labour market. Nonetheless, NSS data show
the prices of agricultural goods rose faster the rural work-force between 1977-78 and that the actual picture is more complicated
than the general price level. This meant a 1989-90. And this rapid growth of rural non- and suggests that ‘distress’ movement into
reversal of the earlier terms of trade agricultural employment provides the main non-agricullure has continued to be important
movement against agriculture, and this also explanation for what would otherwise be a for a significant section of rural workers, as
went against the international trend of a puzzle: how did agricultural wages rise and well as that the main dynamii?§burceof rural
worldwide movement of terms of trade
against agriculture. That this increase in
Table 4: Composition of Urban Employment (NSS Usual Status Data)
agricultural prices did not have unbearable
Males
Females
inflationary implication was partly because
Regular
Casual
SelfSelfCasual
Regular
of the tendency described above of weakened
Employment
Employment
inter-sectoral linkages, and partly because
of the way the food procurement and public
1977-78 (July-June) • . 40.4
• 46.4
13.2
49.5
24.9
25.4.
1983 (Jan-Dec)
15.4
distribution systems functioned. Government
40.9
43.7
45.8
28.4
25.8
1987-88
(July-June)
43.7
14.6
41.7
47.1
25.4.
27.5
procurement of foodgrains was more than
1989-90
(July-June)
16.4
42.3
.
41.3
29.2
48.6
22.2,
adequate and government stocks generally
44.2
15.1
49.0
. 25.9
’ 2ifj£ .
sufficient, the procurement prices were 1990-91 (July-June) ■: ■ 40.7 r
39.9
17.2
42.9
1991 (July-Dec)
47.0
28.0
typically close to the market prices and
: 39.4
19.4
1992 (Jan-Dec) ■
42.5
28.8
■ 2**
'
domestic food prices were also not too far
2462
Economic:ahd Political Weekly
Special Number September 1996
employment generation over this period has
been the external agency of the state rather
than forces internal to the rural economy.5
There arc several planks to this argument.
Within agriculture, all the available evidence
points to the decreasing ability of agriculture
to absorb more labour, as the overall crude
elasticities of employment to output are seen
to be low in other major states and on an
all-lndiabasis. However, there are substantial
variations across states, with the
agriculturally less advanced regions showing
much higher elasticities than the developed
states like Punjab and Haryana. Since some
of the less advanced stales (such as West
Bengal and Bihar) actually showed the
highest rates of output growth over the period,
there was less of a dampening effect on the
overall elasticity as well as a pointer to the
importance of regional spread of agricultural
growth for employment generation.
Moreover, an interesting observation relates
to the flow of person-day employment in
agriculture, which, after 1977-78, was seen
to be increasing more than stock measures
of usual or weekly status workers. In a very
rough and approximate sense, this suggests
that the supply of agricultural labour (as
measured by the stock of agricultural
workers) was actually increasing slower than
the demand for agricultural labour measured
in person-days. Simultaneously there appear
to have been contractual changes under way
in agriculture, with a greater emphasis on
casual contracts.
The natural question consequent upon such
a finding is what caused the slow growth
in the stock measures of workers in
agriculture. Here it was found that pull factors
out of agriculture were significant. The
relationship between agricultural prosperity
and the growth of non-agricultural
opportunities was found to be weak and non
linear, being significant only in states such
as Punjab and Haryana where not only have
agricultural incomes crossed a threshold but
where further increases in agricultural output
are accompanied by labour displacement
rather than absorption. Outside this limited
region, the pull is provided mainly by external
stimuli. In certain regions, forexamplealong
the Bombay-New Delhi and the BombayBangalore highways, there is clear evidence
that industrial development, and the growth
of services linked to this,have made deep
inroads into rural society creating
opportunities not only in the tertiary sector
but also in small-scale industry. In addition,
in the hinterland of industrially or
commercially developed regions, there is
growing incidence of workers who live in
rural areas but commute to urban areas - a
tendency which has been enhanced by the
fact that the organised sector has tended to
prefer casual workers to regular employees,
and because rising urban rents and falling
Economic and Political Weekly
transport costs have influenced workers’
choice of residence. However, given the
limited geographical spread of such direct
links to modem industry and commerce, in
most areas the pivotal role in the expansion
of rural non-agricultural employment appears
to have been played by the expansion of
government expenditure.
As noted earlier, the 1980s were a period
when, along with a rapid increase in all sorts
of subsidies and transfers to households
from government, there was a very large
increase in revenue (as opposed to capital)
expenditure on agriculture by stale and
central governments, and this was also a
period when the expenditure on Rural
Devclopmeni expanded manifold. More
generally, throughout the period political
developments tended to give rural interests
greater power and they were able to command
an improvement in the historically low share
of government expenditure benefiting rural
areas. Although this improvement in share
should not be exaggerated, an indication
may be had from the fact thSt nearly 60 per
cent of all new government jobs created
accrued to rural areas during the decade.
Moreover, NSS data suggest that, despite a
low average contribution of only around 5
per cent of total rural employment, the
government’s contribution was around a fifth
when it comes to either total rural nonagricultural employment in 1987-88 or the
increments in total rural employment between
1977-78 and 1987-88. Moreover, in 1987-88,
about 60 per cent of the regular nonagricultural employees in rural areas were
employed by the government which created
almost 80 per cent of the increments in such
regular jobs during the decade covered.
Thus, given the magnitude of what is now
commonly accepted to have been a profligate
growth of government expenditure, the total
quantum ofincreased flow of public resources
into rural areas must have been significant.
Besides the large growth in agricultural
subsidies already mentioned, this flow of
resources took two predominant forms. There
was, first, a fairly large expansion of ‘rural
development’ schemes with an explicit
redistributive concern.6 This included not
only the various rural employment and
IRDP programmes but also a plethora of
special schemes for a variety of identifiable
‘target’ groups. Motivated by the realisation
that income growth by itself would not
‘trickle down’ in adequate amounts, these
programmes were however less than entirely
successful. They spawned a large
bureaucracy and they became a focal point
for the politics of ‘distributive coalitions’
Yet, though the intended beneficiaries often
got short-changed because of such leakages,
these programmes represented a fairly
massive net transfer to rural areas. Thesecond
avenue by which resources flowed from
government to rural areas was through the
greater accessibility of the rural elites to the
government’s normal gravy train. In part
this was a result of greater mobility due to
better infrastructure, but to a large extent it
was also the outcome of the fact that with
governments changing frequently
(particularly at the state level) more new
favours, not just jobs, but also various types
of agencies and contracts, had to be
distributed more often and the rural areas
got a greater share in such electorally
motivated largesse than they get at other
limes. The resulting flow of resources and
the resulting generation of rural demand led
to growing opportunities for diversification
of the self-employed from agriculture to
non-agriculture.
To a very large extent, the direct access
to government permanent employment and
also to many of the other resources was
confined to the better-off and more powerful
groups in rural society, to whom such
incomes were more lucrative than agriculture.
Associated with this was a large and
significant increase in the proportion of the
15 to 29 age cohort which continued in
education rather than join the work force.
In part this must have been a result of the
expansion of educational facilities as part
of the general expansion of government in
rural and semi-urban areas, but to a sub
Table 5: Changes in Employment 1989-90 to 1992 (NSS Usual Status Unadjusted)
(Million persons)
Self-cmp agriculture
Rcgular-emp agriculture
Casual-emp agriculture
Self-emp secondary
Regular emp secondary
Casual-emp secondary
Self-emp tertiary
Regular-emp tertiary
Casual-emp tertiary
Unemployed
Total workforce
Total population
Special Number September 1996
Rural
1992
1989-90
Urban
1992
1989-90
122.2
5.6
70.5
17.7
3.3
11.2
18.9
10.6
4.1
2.8
266.9
602.7
4.6
0.3
2.8
6.8
7.3
4.6
14.5
15.4
3.1
2.3
61.7
176.3
132.3
3.8
74.9
11.3
4.1
9.9
16.9
9.1
3.6
2.9
268 8
608.9
5.3
. 0.3
3.3
7.0
8.8
6.9
15.6
16.0
4.2
3.3
70.7
200.9
Total
1992
1989-90
126.8
5.9
73.3
24.5
10.6
15.8
33.4
26.0
7.2
5.1
328.6
779.0
137.6
4 1
78.2
18.3
12.9
16.8
32.5
25.1
7.8
6.2
339.5
809.8
2463
stantial extent this must represent also a
motivational change (to acquire necessary
qualifications for a regular non-agricultural
job) among the youth in the relatively welloff sections of rural society. There was thus
a movement out of agricultural work at the
margin by workers and potential workers
from such better-off rural groups, which meant
that sections of the relatively rich vacated
agriculture either to obtain regular employ
ment, mainly in the service sector, or to take
up non-agricultural self-employment.
This increased the ability of members of
the less well-off rural households to find
agricultural work, and also created a demand
forcertaintypesof rural services and industry.
The relative tightening in the agricultural
labour market which resulted, helps to
account for the increase in real wages
observed from the late 1970s. However,
although such increases in employment and
wages did improve the condition of the
poorest rural workers, their employment
diversification into non-agriculture
continued to have many characteristics of
a 'distress’ process, given the overall
tendency of labour use in agriculture.
Dictated by the need to ensure economic
survival, they increasingly entered into casual
work not only in agriculture but also in non
agriculture. The main sectors providing this
type of non-agricultural employment were
secondary sectors like construction, mining,
and small-scale manufacturing, and there is
evidence that over time the incidence of
poverty among those employed in some of
these sectors became larger than in
agriculture. Moreover, the agency of the
state was important in terms of the
diversification of opportunities for the rural
poor. Thus, 22.3 per cent of all casual labour
days spent on non-agricultural activity in
1987-88 were on public works programmes
■ of the government, this percentage having
increased from 17.7 in 1977-78 and 14.9
in 1983. And, although thereislittlecvidencc
of any increase in non-agricultural self
employment among the bottom 40 per cent
of the rural population (such increase was
largely among relatively richer households),
income generation scheme such as the 1RDP,
must also have had some effect.
This is of course an extremely schematic
presentation of what is a much more
multifarious and regionally diverse scenario,
and there were variations in the pattern across
states and over time. However, the fact that
the developments described above occurred.
in every state, irrespective of the rate of
growth in agriculture or organised industry,
does imply the increased importance of
external stimuli to rural employment and,
in particular, the crucial role of the state.
More importantly, these trends mean that the
rural labour demand is no longer determined
only by what is happening within the
Economic and Political Weekly
Thus, in the early 1990s, there was a
agricultural sector, butisdetermined crucially
also by macro-economic processes and reversal of several of the public policies
policies which do not at first appear to have which contributed to more employment and
any direct link with rural well-being.
less poverty in the rural areas in the earlier
Moreover, because much of the govern decade. It should, therefore, not be entirely
ment spending involved is project funded, surprising that rural non-agricultural
because most of the private enterprises employment appears to have declined fairly
involved are small and jack staying power, sharply as soon as the stabilisation and
and because most of the wage employment structural adjustment policies were put into
thus created are casual, the vulnerability of place in 1991. According to NSS survey
the rural non-agricultural sector to overall data, the non-agricultural proportion among
public expenditure cuts and to restrictive rural male workers was 28.3 per cent in
monetary policy is almost certainly greater
1989-90 and 29 per cent in 1990-91, before
than for its urban counterpart. This has very the reforms, and this fell to 25.1 per cent
important ramifications in thecurrent macro- in July-December 1991 and 24.3 per cent
economic context, in which the reform in 1992. For rural female workers, the
measures have particular implications for corresponding figures were 18.6,15.1, 13.7
patterns of government expenditure as well and 13.8 per cent. This represents a decline
as on internal and external trade.
of somewhere between 9 and 11 million in
It is important to note in this context that the number of workers in rural non
the pattern of structural adjustment and agriculture, or a drop of 13-15 percent in
government economic strategy since 1991
the first 18 months of the initiation of the
has been one which has involved acontinucd reform process.’
This fall occurred almost all over India,
stagnation in employment generation in the
organised sector, both public and private. with only Karnataka and Madhya Pradesh
being significant exceptions. In terms of
Moreover, this strategy involved:
(1) actual declines in Central government sectors, this decline in employment was
revenue expenditure on rural develop divided roughly equally between manu
ment (including agricultural programmes facturing, construction and community and
and rural employment and anti-poverty other services, along with a smaller drop in
schemes), as well as on the fertiliser transport; while mining, electricity, trade
subsidy, in the budgets of 1991-92 and and financial services were immune among
1992-93. Some of these cuts were how the self-employed and casual workers that
ever reversed subsequently in 1993-94. the decline was greatest, with regular
(2) declines in public infrastructural and employment being largely maintained, except
energy investments which affect the rural for some drop among regular male employees
in the tertiary sector. Thus, the pattern of
areas.
(3) reduced transfers to state governments the decline in rural non-agricultural work
which have been facing a major financial suggests that it occurred not because of any
crunch and have therefore been forced large-scale retrenchment of regular
to cut back their own spending, parti employees by the government or the
cularly on social expenditure such as on organised private sector, but because of a
education and on health and sanitation. cut back in activity in the unorganised sector
(4) reduced spread and rising prices of the and, possibly, some retrenchment of casual
workers by the organised sector. In this
public distribution system for food.
(5) financial liberalisation measures which context, it is significant that, according to
have effectively reduced the avail ability the NSS, this drop in rural non-agricultural
of credit, especially to small borrowers employment was riot accompanied by a
corresponding drop of such employment
particularly agriculturists.
Table 6: Tendulkar-Jain Estimates of Poverty
1970-71
1972-73
1973-74
1977-78
1983
1986-87*
1987-88
1988-89*
1989-90*
1990-91*
1991 (July-Dec)’
1992*
H
Urban
PG
SPG
H
Rural
PG
SPG
45.89
47.00
49.20
42.98
38.33
35.39
36.52
36.98
. 32.41
32.43
. 32.02
33.87
13.39
13.57
13.88
12.16
9.95
9.48
9.34
9.61
.8.03
8.03
.7.90
8.43
5.32
5.32
5.31
4.81
3.66
3.54
3.38
3.49
2.84
'2.88
2.84
2.97
57.33
57.21
56.17
54’47
49.02
45.21
’ 44.88
42.23
37.94
36.55
42.06 .
. 48.07
17.57
17.93
16.75
16.59
13.86
12.21
11.26
10.20
8.80
8.81
10.02
12.59
7.31
7.54
6.72
6.88
5.45
4.60
4.04
3.54
2.95
3.03
3.39
4.58
Nates: * Denotes small sample; Poverty treasures arc same as in Table 1
Special Number September 1996
2465
among urban workers. Even among rural
workers this did not lead to any large increase
in open unemployment or to any large fall
in the work participation rate. Rather, the
self-employed and the casual workers dis
placed from non-agriculture appear to have
reverted back to agriculture, leading to
disguised (rather than open) unemploy ment.
However, as a result, real agricultural valueadded per agricultural worker dropped
significantly, by over 8 per cent, even if
comparison is restricted to the years 1989-90
and 1992-93 when monsoon conditions were
very similar. Unfortunately, later data
(particularly from the 1993-94 NSS large
sample) is not yet available to verify whether
this reversal of trend has continued, but
clearly the early post-reform impact was
adverse.
In the urban areas, regular employment
has continued to stagnate, especially in the
organised sector. During 1991 -95, the growth
Bite of employment in the organised sector
nalved from its already low growth rate
during the 1.980s to only 0.8 per cent per
annum, mainly because of a massive slowing
down of employment growth in the public
sector. The increases in employment that are
discernible are essentially in casual
employment, and this is evident for the
secondary and tertiary sectors according to
both usual and weekly status definitions.
However, these increases in employment are
still below the estimated increases in urban
population over this period. The continued
process of casualisation of work in urban
areas has to be seen in relation to two other
recent tendencies which are highlighted by
several micro-level studies. First, there is the
growth of subcontracting in manufacturing,
which increasingly integrates formal and
informal sector productive activities, and
allows for a substantial part of the production
to be undertaken by very small informal and
Unorganised units at the bottom of the
production chain. These i mply that a growing
part of manufacturing production is
undertaken by units in which there is no
formal protection of any sort to labour.
Secondhand related to the first tendency,
there is evidence of some ‘feminisation’ of
employment, that is the growing share of
female employment to the total, particularly
in export-oriented activities, and with wages
and working conditions that arc typically
inferior to those of male counterpart workers.
It is evident that these processes will have
direct and indirect links to the spread of
poverty, throughout India but especially in
rural areas. These links, and the more general
relation between economic growth and
poverty, are considered below.'
in
Post-reform Trend in Poverty
In earlier sections it has been observed
that there was a declining trend in poverty
after the mid-1970s but that this trend was
reversed in the 1990s. However, while the
earlier declining trend is officially accepted,
the reversal during the 1990s is not. As
discussed earlier, the main difference between
the official view and those of independent
observers arises because till now the official
estimate is based not on the NSS data
directly but on adjusted figures obtained
by blowing up the NSS consumption
estimates for every decile group by a common
adjustment factor equal to the ratio between
the CSO estimate of private consumption
and the corresponding NSS estimate. Because
this adjustment factor has increased sharply
in recent years, the official estimate has
diverged increasingly from any estimate
based directly on NSS data. The Expert
Group which recently went into this matter
concluded quitecategorically that the practice
of ‘adjusting’ NSS data was arbitrary and
was likely to give wrong results because as
against the implicit assumption in the official
method that any underestimation of
consumption is distributed uniformly over
the entire population, it is better to assume
that the underestimation is only for those
who arc non-poor.
In fact, several alternative scries which
use the unadjusted NSS figures are available.
In addition to the World Bank series given
earlier, a scries calculated by Tendulkar and
Jain is available for the period 1970-92?
This uses the same reference poverty lines
at 1973-74 prices as the official and World
Bank series, but using a different deflator
they obtain an even larger increase in poverty
between 1990-91 and 1992 (Table 6). In
addition, it has been possible to obtain
measures of rural poverty for All India and
the major states, based on the Expert Group
Method using NSS data covering the years
from 1972-73 to 1993-949 (Table 7). Unlike
the other series this is not obtained from the
national-level NSS data but is obtained by
applying state-specific poverty lines to state
level NSS data. It must be noted that the
figures given here for 1993-94 are
preliminary , being based only on partial data
(not yet officially released) from the 50th
round of the NSS. Moreover, the data for
1986-87 and for 1.989-90 to 1992 are based
on the so-called ‘thin’ surveys by the NSS
involving a much lower sample size than the
other survey points. With only three survey
points available for the post-reform period,
and given the above qualifications for
whatever data is available, any conclusion
about post-reform trends must necessarily
be rather tentative.
Nonetheless, using the mutually compar
able thin samples alone, it is evident that
poverty increased sharply during the first 18
months of the reform period (i e, the second
half of 1991 and 1992), particularly.in the
rural areas. The partial data relating to
1993-94 suggests, however, that this upward
trend was reversed thereafter. Taken together,
these data suggest that there was a very large
increase in rural poverty in the first 18
months of reform but that this trend has been
moderated thereafter. Rural poverty in
1993-94 continued to be higher than in
Table 7: Estimates of Rural Headcount Poverty by the Expert Group Method
Andhra Pradesh
Assam
Bihar
Gujarat
Haryana
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
All India
2466
1973-74
1977-78
1983
1986-87
1987-88
1989-90
1990-91
1992
1993-94
48.4
52.7
63.0
46.4
34.2
55.1
59.2
62.7
57.7
67.3
28.2
44.8
57.4
56.5
73.2
56.4
38.1
59.8
63.3
41.8
27 7
48.2
51.5
62.5
64.0
72.4
16.4
35.9
57.7
47.6
68.3
53.1
26.5
42.6
64.4
29.8
20 6
36.3
39.0
48.9
45.2
67.5
13.2
33.5
54.0
46.5
63.1
45.6
14.6
39.7
50.1
30.3
19.5
36.6
33.5
47.8
44.6
55.2
13.0
29.2
41.2
36.6
47.3
38.3
20.9
39.4
52.6
28.7
16.2
32.8
29.1
41.9
40.8
57.6
12.6
33.2
45.8
41.1
48.3
39.1
19.5
35.2
52.4
14.8
13.3
45.4
34.4
39.5
34.8 •
52.9
3.2
26.1
38.4
30.5
37.2
34.4
22.1
33.7
46.3
21.6
19.5
34.9
30.3
42.4
35.9
36.5
9.3
25.9
37.5
34.8
49.5
35.0
27.4
51.7
61.1
33.7
17.7
45.5
26.0
47.9
53.6
49.0
10.2
31.7
44.3
47.9
44.0
44.0
16.0
45 0
58.0
22.2
28.7
28.2
25.9
40.8
38.6
49.9
12.5
27.5
32.6
42 6
40.3
37.5
•
Economic and Political Weekly
Special Number September 1996
1989-90 and 1990-91 but was less than in
1987-88. Urban poverty, on the other hand,
appears not to have increased much during
the first 18 months of the reform period and,
i ndeed, appears to have declined significantly
in 1993-94. Nonetheless, there were about
30 million more people in poverty in the
latter year than before the reforms began.10
Thus, the post-reform trends in poverty
do not suggest cither an unambiguous
improvementor an unambiguous worsening.
They do suggest, however, that the initial
impact of the stabilisation/structural
adjustment package was adverse, that this
impinged particularly on the rural sector,
with less impact on the urban sector, and that
there was some general reversal of the adverse
trend subsequently. Nonetheless, it is
important to note that the state-wise figures
show that, as far as rural poverty is concerned,
in most states the poverty ratios in 1993-94
were significantly larger than in the
immediate pre-reform period. This is
particularly true of the two largest Indian
states, Uttar Pradesh and Bihar, and also of
the hitherto successful ‘green-revolution’
states of Haryana and Punjab. The exceptions
are the Southern states of Andhra Pradesh,
Karnataka, Kerala and Tamil Nadu in all of
which the poverty ratio in 1993-94 was
lower than in the immediate pre-reform
period. However, it must be noted that in
these states, and in Maharashtra and Gujarat,
the year 1993-94 was exceptional in that the
food prices actually fell in absolute terms
as against rapid increases in both preceding
and following years. For this reason, the
calculated poverty ratios for 1993-94 are
likely to be somewhat lower than the
underlying trend.
More importantly, these trends in poverty
need to be viewed in the general context,
discussed earlier, that the stabilisation and
stuctural adjustment policies carried out so
farjn India involved a fairly sharp contraction
in fiscal and monetary policy in 1991-92 and
1992-93. followed by a return to high fiscal
deficits from 1993-94 onwards. The revival
of growth after an initial period of stagnation
also 'followed a pattern broadly coincident
with that ofthe government’s fiscal stance,
so that both the initial worsening of the
poverty situation and the subsequent
improvement seem to be broadly in line with
the overall growth performance of the
economy. Yet, there are a few surprises, the
most important of which is that although the
reform measures did not directly involve
much changes in agriculture, it was rural
poverty which appears to have been more
sensitively affected by the post-reform
developments.
This has led some analysts, for example
Tendulkar and Jain (along with many other
economists who are generally in support of
the reform process), to argue that, although
Economic and Political Weekly
poverty did increase during the first 18
months of reform, the reforms were not the
principal cause for this incrcasc.This
argument rests on the observation that “the
rural sector in general and agriculture in
particular were not the major focus of
structural adjustmentand were only indirectly
affected by fiscal compression" and on the
assumption that the reforms “would have
adversely affected primarily the urban
organised sector with second-order impact
on the urban informal sector and weaker
effect on the rural sector”. Because, in fact,
it was rural poverty which increased sharply,
they attribute this increase not to reforms per
se but to weather and to the higher post
reforms inflation for which, moreover, they
hold the reforms only indirectly responsible.
However, although the reform process has
indeed neglected agriculture, there arc two
major difficulties with the argument that,
therefore, it could not have increased rural
poverty.
First, although there was a 2 per cent drop
in agricultural production in 1991-92
compared to 1990-91, and although the
inflation rate did increase sharply
(particularly for foodgrains), these adverse
factors were simply not large enough to
explain the very large increase in the
incidence of rural poverty. As discussed in
the next section any econometric model fitted
to the data prior to reforms linking the
incidence of rural poverty only to some
measure of agricultural production or
productivity and to theinflationeffect, breaks
down as soon as the post-reform data is
included. And, indeed, all such models are
outperformed by models incorporating
relative food price, rural non-agricultural
employment . and some measure of
commercialisation, in addition to agricultural
production. The latter not only fit the prereform data better, but when fitted to prereform data accurately track the post-reform
increase, in poverty, unlike models not
including government expenditure and/or
rural non-agricultural employment which
predict much lower poverty increase than
that which took place actually.
The important point to note is that this
phenomenon of rural non-agricultural
employment, which Tendulkar and Jain
ignore, was almost certainly the major factor
which drove rural real wages up, and caused
poverty to decline, during the 1980s. And,
in turn, this was based largely on increasing
government expenditure and on the
availability of cheap credit to the small-scale
sectors. As has been pointed out in a previous
section, the onset of stabilisation and
structural adjustment appears to have led to
a rather quick and large decline in such rural
non-agricultural employment, pushing
millions of self-employed and casual rural
non-agricultural workers back to agriculture,
thus reducing per-worker incomes in
agriculture. It appears, therefore, that just as
the expansion of the 1980s involved a rapid
increase in rural non-agricultural employ
ment without any concommitant increase in
organised sectoremployment.thestagnation
in the first 18 months of reform saw a cutback
in rural non-agricultural employment with
not much effect on either organised sector
or urban employment. This suggests that,
contrary to popular opinion, the employment
multipliers associated with the government's
fiscal stance are larger for rural non
agriculture than for the urban or organised
sectors. If this is accepted, the trend in the
magnitude of employment decline and
poverty increase are not surprising,
particularly-because, as noted earlier, the
contractionary tendencies generally
impinged much more adversely on smaller
enterprises than on the corporate sector."
In addition, it is obvious that the effect
on poverty of the rather small decline in
agricultural output in 1991-92 could have
been mitigated if rural employment policies
had been used effectively, as they were
during 1987-88 a year of much larger decline
in agricultural and foodgrains output than
Table 8: Basic Regression Results: All-India Data
(Dependent Variable Is Log of Headcount Poverty Ratio)
Constant
Rural
11.34
9.56
8.83
1.37
4.68
Urban
11.16
8.57
8.24
4.61
5.23
Per Capita
Agriculture
Income
Per Capita
Non-Agri
Income
-0.72(1.2)
-0.68(1.1)
-0.45(1.9)
-0.49(2.2)
-0.45 (2.4)
-0.36 (2.1)
-0.78(1.2)
-2.38 (4.0)
-1.62(3.3)
-2.15(4.3)
LU (1.5) ,
3.20(4.8)
-0.76 (9.4)
1.73 (3.1)
2.65 (4.6)
-0.39 (2.7)
-0.75 (2.0)
-0.69 (2.0)
-0.58 (2.4)
-0.59 (2.7)
-0.58 (2.8)
-6.32 (3.0)
1.35(1.8)
-0.07 (0.1)
0.19(0.4)
-1.63 (2.3)
0.31 (0.4)
-0.26 (0.5)
Commerce
Public
Dev
Expenditure
Relative .
Price of
Cereals
1.45 (9.8)
0.80 (3.0)
0.73
0.77.
0.96
0.96
0.97
. 0.70 (4.8)
0.60 (2.7)
0.86
0.88
0.94
0.95
0.95
-0.34 (4.0)
-0.08 (1,5)
R Bar
Squared
.
Note: T values in parenthesis.
Special Number September 1996
2467
1991 -92. Indeed, a state-wise analysis shows explicitly with the idea of removing trade in agricultural products have a large and
that between 1989-90 and 1992 rural poverty discriminationagainstagriculturalgoodsand. quick impact on rural poverty.
increased in every state except Kerala, i e, therefore, with a central tenet of the
it increased even in those states (Haryana, liberalisation argument. If this had been
IV
Karnataka, Madhya Pradesh, Orissa, Punjab, followed through fully, food prices would
Determinants of Poverty
Tamil Nadu. Uttar Pradeshand West Bengal) have risen much more than they actually did:
.i*-;
where the 1991-92 foodgrains output was and farmers would have received more than
This brief review of trends in poverty
higher than in 1989-90, thus indicating a what they allegedly obtained as a result of bring us to the central issue of this paper:
rather weak link between the fall in output their political clout. In fact, prices began that is how liberalisation and structural
and the increase in poverty. Moreover, it is stabilising only when the government made adjustment may be expected to affect the
interestingto notethatthelargestpost-reform it clear that not only would some export incidence of poverty. Given the limited data
increases in poverty were registered in the restrictions continue on foodgrains but also and the somewhat conflicting empirical
two states, Gujarat and Maharashtra, which that it would continue with its earlier policy trends reviewed above, the remaining
were most enthusiastic about the reforms of importing foodgrain to stabilise domestic discussion will focus on past discussions of
process. Here, there was a fall in foodgrains prices, even if this meant making a the determinants of poverty in India, and
output in 1991-92asin 1987-88, but. unlike commercial loss. For this reason, it can be how the conventional logic needs to be
in 1987-88, the state governments neglected surmised that the inflationary problem could modified in the light of subsequent develop
rural employment and drought relief schemes have been avoided to a large extent had ments. Since the Indian literature is mainly
in 1991-92.
devaluation been accompanied at the outset on rural poverty, this discussion will also
Secondly, although inflation is clearly by an explicit policy of increasing the wedge focus largely on the rural sector.
important, Tendulkar and Jain are not entirely between world and domestic prices through
Past literature has tended to focus on two
correct when they claim that the large rise higherexport duties and a definite announce types of variables: some measure of
in inflation (to over 25 per cent for the ment that canalised imports would continue. agricultural output or productivity and some
Consumer Price Index for Agricultural That it was not, and led to an inflationary price variable. And past writings have
^Prurers) during the first year of reforms spurt which was contained only when the debated both the relative significance of
was only indirectly related to the reform -government backtracked, is one among a these variables, and, more importantly, their
process. Their argument is that inflation number of instances of how fidelity to the proper specification.13 Thus, although the
occurred because crop output fell in a year liberalisation world-view was extremely level and growth of agricultural production
when foodgrain stocks were low and the ■ costly in the short-run without being per capita of rural population is obviously
balance of payments position did not permit sustainable in the longer run.
an important variable determining levels of
large imports; and because the government
Thus.themassiveincreasein rural poverty, welfare in a predominantly agricultural rural:
succumbed to rich farmer demands to by over 60 million people, in the first 18 community, ibis also obvious that such a
increase procurement prices following the months of reform was to a very large extent relationship would bc~affected by whether
(necessary steps of) devaluation and cut in a direct result of the stabilisation-cum- agricultural growth is accompanied by
fertiliser subsidy.
structural adjustment policies. Thelaterdata, increasing inequality and whether there are
In fact, because of a record harvest in for 1993-94, which shows a moderation in other sources of rural incomes. The link
1990-91, public foodgrain stocks were high, poverty does not necesarily contradict this between poverty and prices is even more
over 21 million tonnes, when the government conclusion because, after all, public complex. For example, when Dharm Narain
embarked on its reforms at the end of June expenditure cuts were to a large extent presented regression results showing that
1991. And inflation accelerated principally restored (and so rural non-agricultural poverty was related positively with higher
because of the expectations set-off by employment might have risen somewhat) food prices, his specification waschallenged
devaluation and the impression given that and stability was returned to foodgrains because his use of the nominal food price
all discrimination against exports of markets by removing the expectation that as an explanatory variable ran counter to the
agricultural goods would be removed.12 This Indian agricultural prices were to soon reach prior.common to most economists, that what
led to an immediate speculative increase in international levels. Thus, although nothing really matters.are relative prices, and, that
private stocks, forcing the government to firm can be said about employment trends if absolute prices need to be incorporated,
®lcwn its own stocks faster, and also till the full data from the 1993-94 survey •this should be done by considering the rate
contemplate food imports. However, the is released, the decline in poverty appears. of inflation rather than the price level.
As far as the importance of agricultural
decision on such imports was postponed till only to confirm that changes in public
after the next harvest. And when this turned expenditure levels and announcem'ents output as a determinant of rural poverty is
out to be somewhat less than expected, the regarding liberalisation of international trade concerned, it is obvious that, unless the
government was faced not only with low
stocks but also with low procurement because
Table 9: All India Rural Poverty Equation
farmers withheld sales in the expectation
Our Model
Variables
Ravallidn-Dutt Model
that it would be politically and economically
I960 to 1992
1960 to 1989
1960 to 1992
I960 to 1989
di fficult for the government to justify i m ports
-0.1(0.10)
at prices well above domestic free market Constant
4.6(5.01)
3.3(2.79)
-0.6(0.67)
-0.4(3.36)
-0.1(0.56)
-0.4(4.21)
prices which in turn were higher than the Ag Productivity
703(2.69)
-0.4(2.79)
-0.7(3.53)
-0.5(3.26)
-0.3(3.29)
procurement prices. In the event, the Real Wages
0.9(6.25)
0.9(7 69)
government was forced to increase both its Relative price cereals
-0.4(3.65)
-0.5(3.97)
procurement prices massively (linking this Non-ag employment
0.9(4.92)
1.0(6.45)
with withdrawal of fertiliser subsidies) and Commercialisation •
0.0(1.86)
•
0.0(3.85)
also import wheat at prices higher than the- ■ ^ime Trend
0.5(4.74)
0.4(2.75)
increased procurement prices.
‘ Lagged Dependent
0.97
0.89
R Bar Squared
0.94
0.98
Theentire problem was thus clearly driven DW
2.37
1.28
1.63
2.27
by the fact that devaluation was linked
2468
Economic and Political Weekly
Special Number September 1996
manner in which higher agricultural output
is brought about is sharply inequalising, any
increase in agricultural output per capita
would tend to benefit most rural people. It
is precisely the fear of the possible
incqualising impact of the‘green revolution’
which had triggered off early work on this
area. But. by now, it may safely be conceded
that although relative inequalities may have
increased, the ‘green revolution' certainly
reduced the incidence of absolute rural
poverty in the regions where it was successful.
But, although proponents of the trickle down
hypothesis may have proved more correct
than the detractors in this matter, the really
striking feature of the post green revolution
period is that, nonetheless, there is a rather
weak link across states between the rale of
agricultural growth percapitaand reductions
in rural poverty.
The simple fact is that, with the green
revolution limited in geographical coverage.
most states in India did not record any
significant increase in agricultural valueadded per head of rural population during
the 1970s and 1980s, although almost all of
them recorded significant declines in poverty.
With poverty reduced even where
agricultural output did not increase, there
has thus been a reversal of the earlier
apprehension that agricultural growth could
occur without reducing poverty. But this
very disassociation between poverty
reduction and agricultural growth is a feature
which merits more attention than has been
given so far. For example, it is significant
that, while early work on the subject
invariably chose some measure of
agricultural output per capita, some recent
research finds agricultural output per hectare
to be the measureof agricultural performance
bettercorrelatedwith poverty decline. l4This
measure has the advantage, for those
persistent in viewing agricultural growth as
the main engine for the reduction of rural
poverty, thaton this basis al mostevery region
in India recorded some agricultural growth
during the period when poverty declined,
and, in most cases, this growth was also
larger than during the earlier period when
rural poverty did not decline.
Yet, given the stagnation of agricultural
output per head in most parts of rural India,
this shift in the measure used surely serves
to obfuscate matters rather than to clarify
them, especially because in actual fact the
underlying shift in Indian agriculture from
expansion of cropped area to yield increases
has been accompanied by a sharp decline
in the output elasticity of agriculture’s
demand for labour. As a result, poverty has
declined in most regions of rural India in
a context not only of stagnant agricultural
output per head of rural population but also
one where agricultural employment has
grown much slower than the growth of the
Economic and Political Weekly
rural labour force. This latter feature would
normally be expected to depress agricultural
wages and thus affect adversely the poorest
among India’s rural residents. But, in fact,
real agricultural wages increased sharply in
most parts of India between the mid-1970s
and late 1980s, and this was, in fact, one
of the main reasons why poverty declined.
Some researchers do note th? rise in wages
but chose to ‘explain’ this rise also by
reference to the increase in output per acre.
This, however, stretches credulity because
by almost every measure the rise in real
wages was at least twice as much as the
incrcasein real output per worker, and cannot,
therefore, be ascribed mainly to technical
progress in agriculture. As has been argued
already, the real explanation for the rise in
agricultural wages lies in the rapid growth
of rural non-agricultural employment and
the dynamics behind this. The Indian
literature on this has in the past toyed with
two ideas: that rural non-agriculture is itself
driven by agricultural growth through the
operation of Engel's law; and the opposite
idea that the process of commercialisation
of agriculture leads to displacement of
agricultural labour which finds distress
employment in certain 'residual' sectors of
non-agriculture. However, both these ideas
are now somewhat discredited. The link
between agricultural growth and thatof rural
non-agriculture has been found to be non
linear, and, with the exception of the limited
prime ’green revolution’ area there is little
evidence that agricultural growth has
provided the impetus to rural non-agriculture.
Also, despite considerable evidence that
commercialisation is inequalising and leads
to casualisation of the rural labour force, the
'residual' sector hypothesis stands
discredited because this docs not square
with rising real, wages. The evidence
overwhelmingly supports the thesis that the
main impetus for the growth of rural non
agriculture has come from outside the rural
areas, in considerable part from theexpansion
ofgovernment expenditure. This observation
suggests that rural incomes are no longer
derived only from agricultural production,
and that the process by which rural areas
have got integrated into the wider economy
are important. One aspect of this has been
commercialisation with its inequalising
effects but the other is that external stimulii
have provided employment opportunities
and incomes which are related not so much
with agriculture but with developments in
the wider macro-economy.
However, many economists continue to
view agricultural growth as the main solution
to poverty. One reason for this is precisely
because such an association fits neatly with
the view that in countries like India, which
protected industry in the past and therefore
require ‘structural adjustment’ today, got it
wrong not only on efficiency but also on
equity. Indeed, during the 1970s, when the
World Bank pushed its growth with
redistribution slogan, the argument was that
the pro-industry policies followed by
countries such as India hampered agricultural
growth and thus meant higher poverty than
was necessary. Even today, some World
Bank analysis tries to show that rural poverty
isunaffected (oreven worsened) by industrial
growth while agricultural growth reduces
not only rural but even urban poverty.,sThis
view continues to be a central component
of the structural adjustment policies, where^
it is argued that greater liberalisation of trad?
and industry would shift resources towards
agriculture and this would not only be more
in line with India’s comparative advantage
but would also reduce poverty much more
than earlier policies. In otherwords, although
no one will dispute that higher agricultural
output is very likely to reduce rural poverty,
it is no accident that in circles where
‘structural adjustment is seen as agood thing
there is also an almost single-minded
obsession with this causation, to the point
of excluding from consideration other
possible determinants of the incidence of
poverty.
However, given Dharm Narain’s critique
of earlier work which concentrated only on
agricultural output, these more recentanalysts
of rural poverty in India do not entirely
forget the price dimension as a possible
determinant of poverty. But, interestingly,
the focus in such recent analysis is al mo jg
entirely on how inflation is bad forthe poo"
Thus, either an inflation term is added to
agricultural output in statistical models
explaining poverty or real wages are added
as an explanatory variable to the poverty
equation and an inflation term is added to
the equation explaining real wages. The
argument is the entirely plausible one that
Table 10: Pooled Time Series and Cross Section Across States
. (Dependent Variable: Head Count Poverty Ratio)
Constant
Ag output per rural person
Per capita state domestic product
State development expenditure
Relative food price
Inflation rate
R Bar Squared
.
DW
Special Number September 1996
5.3(9.61)
-0.2(2.87)
0.1(0.89)
-0.2(5.42)
0.6(5.16)
0.1(1.42)
0.87
2.21
5.8(16.59)
-0.2(3.67)
-0.2(7.61)
0.6(5.45)
‘ 0.88
2.22
2469
rural money wages (and possibly certain
other components ot rural'income, such as
the proceeds of the previous harvest) are not
index-linked and therefore are not
immediately protected against inflation,
although these are likely to adjust jin the
longer-run :
But, although very plausible as ah
explanatory variable of short-run variations
in poverty, the choice of inflation as the
preferred price variable is again not entirely
accidental. For reformers, this leads to the
happy coincidence that while structural
adjustment is goodfor poverty because that
is likely to shift resources towards
agriculture, stabilisation is also good for
poverty because this will reduce inflation.
Most importantly, this way of incorporating
prices avoids facing the essential trade-off
that Dharm Narain was pointing to: that
between the possible beneficial effects of
higher agricultural output on poverty and
the possible losses involved if the preferred
strategy for increasing agricultural output
consists of higher agricultural prices.
This is a trade-off particularly relevant for
the structural adjustment policy as it relates
to agriculture. As has been pointed out by
many, this policy seems to have given no
serious consideration to agriculture in terms
of new programmes or investment, although
it purports to be a virtual overhaul of the
entire economy. But this is no accident either.
In theliberalisingworldview, most economic
problems can be resolved by a greater
recourse to markets and allowing the price
mechanism ‘free’ play. A similar position
governs the attitude to agricultural growth,
in that it is supposed that relative price
movements and profitability ratios will be
sufficient to ensure that supply responsive
ness in agriculture will lead to higher rates
of growth. And the critical variable here is
the ratio of agricultural to other prices in the
economy, which is sought to be increased
by reducing trade protection to industry,
through devaluation which makes non-traded
goods cheaper relative to traded goods, and
through removal of restriction on
international trade in agricultural goods
which would have the effect of increasing
the domestic relative price of most
agricultural products, including foodgrains.
In other words, the very mechanism
through which agricultural output is expected
to increase under structural adjustment
involves increasing the price of agricultural
goods, notably food, relative to all other
prices in the economy. This essential rise in
the relative price of agricultural goods is
thus not seen as a transient phenomenon like
inflation, and it would leave the rural poor
unaffected adversely only if the prices of
goods and services they sell rises in line with
the rise in price of food which makes up most
of their consumption basket. The effect would
2470
certainly be adverse if money wages are
sticky. But, even with flexible wages and
full employment, wage Workers would
invariably lose if they got a substantial part
of their income from non-agricultural
activities in relation to which agricultural
prices would require to rise. Unlike inflation
whose control may be benign (if not
accompanied by deflation), this essential
relative price implication of structural
adjustment is permanent by design and so
also is its likely adverse effect bn poverty.
It is, therefore, noteworthy, and hardly
accidental, that in the new set of poverty
models, such a relative price variable is
hardly ever included.
Moreover, deflationary policies designed
to control overall inflation during a transition
when domestic price relatives adjust to
international levels may have a
disproportionate adverse effect on the rural
poor if thelatter obtain sizeable parts of their
income from non-agricultural activity. In
Indian discussions on the subject, it is
sometimes almost assumed that income and
employment multipliers associated with
fiscal and monetary policy never spill across
municipal boundaries into rural areas.
Nothing could actually be further from the
truth. With agricultural output determined
from the supply side, and agricultural prices
made inflexible downwards by government
support operations, it may indeed be the case
that the income multipliers of a deflationary
package are borne entirely by non
agriculture. But given that most organised
sector workers still have secure employment
at pre-determined wage rates, the entire
burden of the employment multiplier falls
on the unorganised non-agricultural sector.
This sector does not respect rural-urban
boundaries and cut-backs in employment
demand here are likely to have knock-on
effects on the incomes of the rural poor:
through a combination of lower nonagricultural employment, falling real wages,
and an increase in the extent of disguised
unemployment in the agricultural sector.
That this latter effect might be important
has already been indicated by the data so
far presented on rural poverty and nonagricultural employment. Its plausibility is
enhanced because much of the government
spending which is important for rural areas
is project or programme funded and thus
more susceptible to expenditure cuts, because
most rural non-agricultural enterprises are
small and lack slaying power, and becau-c
most of the wage employment thus created
is casual.. As a result, it would be no
exaggeration to claim that the vulnerability
of the rural noh-agricultural sector to overall
public expenditure cuts and to restrictive
monetary policy is almost certainly greater
than for its urban counterpart. This has very
important ramifications in the current macroeconomic context.
Thus, there are two possible stories which
can be told about the impact of structural
adjustment andstabilisation on rural poverty.
The first is the benign one: that by increasing
agricultural output and controllinginflation,
these act to reduce poverty. Alternatively,
there is the less optimisticbutno less possible
outcome that structural adjustment acts
adversely oh the poor because ‘getting prices
right’ leads invariably to a rise in the relative
price of food, because greater reliance on
market forces spurs inequalities inherent in
the commercialisation process, and because
these adverse effects are compounded by
contracting non-agricultural employment and
falling wages in the unorganised sector if
the government wishes to contain, through
contractionary stabilisation policies, the
inflationary fall-out of adjustment.
Which of these actually transpires is an
empirical matter, and one would expect
economists to have tested for which of these
effects are more likely. Butoddly, the benign
agricultural output/ inflation story of rural
poverty seems to hold the fort without being
tested seriously against the alternative
hypothesis involving relative price changes,
commercialisation, rural non-agricultural
employment and the government’s fiscal
and monetary stance.
On the other hand, the discussion of actual
developments earlier in this paper suggests
that the simple agricultural output/ inflation
story about the determinants of rural poverty
can be quite misleading. Thus, any
explanation of falling rural poverty during
the mid-1970s and 1980s would appear to
be incomplete if it did not incorporate the
fact of increasing rural non-agricultural
Table 11: Dutt-Ravallion Cross-Section and Time Series Pooled
(Dependent Vanable: Head Count Poverty Ratio)
Time Varying Variables
Mean consumption
Ag productivity
Rate of inflation
State dev expenditure
Initial Conditions
Irrigation
Female literacy
Infant mortality
R Bar Squared
Economic and Political Weekly
-0.1(2.58)
0 6(6.57)
-0.3(5.12)
-1.1(1562)
-0.0(0.79)
0.2(2.35)
-0.1(3.33)
-0.6(3.08)
-0.4(2.59)
0.9(4.69)
0.86
-0.4(3.32)
-0.1(1.09)
0.4(3.18)
0.94
Special Number September 1996
employment and the role of government because, as suggested earlier, employment
policy behind this. Similarly, our critique of multipliers are higher for rural nonthe Tendulkar-Jain explanation of the agricultural employment. This latter
increase in poverty post-reforms is also suggestion finds some, confirmation from
essentially that they fail to go beyond the the coefficients bn the government
agricultural output/ inflation story. In the . expenditure variable, whose significance
remaining part of this section we attempt a . suggests also that such expenditure has a
statistical investigation.
much larger impact on poverty'than a general
To begin with, we regress head count increase in non-agricultural output.
measures of poverty for the period 1960-61
Surprisingly, on including an inflation
to 1993-94 at the all-India level in both term in these equations, this is found to have
urban and rural areas against per capita an insignificant effect on poverty in both
agricultural and non-agricultural incomes rural and urban areas. Replacing the inflation
(with agricultural incomes defined per head term with a relative price variable (the relative
of rural population and non-agricultural price of cereals to all commodities in the
incomes defined per capita of total wholesale price index) does however make
population). As the accompanying table 8 a difference. This variable turns out to be
shows, both income variables are highly significant in both; areas, and also
significantly negative in both rural and urban serves to reduce the significanceofthe public
areas, and in both cases the agricultural expenditure variable, which however
income variable appears more important. continues to be significant.
Next, the per capita non-agricultural income
These results need to be interpreted with
variable is split into per capita income caution because the mutual correlation
from trade and transport (an indicator of between these explanatory variables is often
commercialisation) and other non-agri high. But three points emerge quite strongly.
cultural incomes. The rationale for this is First, that agricultural incomes are important
^ftnany commentators (e g, Vaidyanathan not only for rural but also urban poverty.
1986) have argued commercialisation this Second, that non-agricultural impulses,
tends to increase rural inequalities. In these particularly public expenditure, are not only
respecified equations, agricultural incomes important but that they are. especially spin
continue to be negatively related to poverty,- the determination of rural poverty. Third;
but now there is a difference between the that, as far as the price variable is concerned,
urban and rural equations^ In urban areas, the relative price effect is if anything much
the commercialisation variable appears to more important than the effect of inflation
reduce poverty while the remaining non- perse.
agricultural incomes have a positive effect.
Taken together, these results suggest that
But exactly the opposite pattern appears in weneed to modify the view that the principal
rural areas. Next, we include a public ex determinants of poverty are agricultural
penditurevariable (development expenditure output and inflation, and that, therefore,
per capita, i e, government expenditure less both structural adjustment and stabilisation
interest payments and expenditure on defence are good. To consider this matter further, we
and administration). This variable is strongly replicated the version of this story as it
significant, reducing poverty in both rural emerges from the World Bank publication
and urban areas, with, interestingly, its Growth and Poverty in India.'1 This is more
coefficient almost double in rural as sophisticated than most other versions of the
compared to urban areas. Moreover, on story in that it is a two-equation model
inclusionofthis variable, the non-agricultural whereby the incidence of rural poverty is
income variables become insignificant in ‘explained’ by the agricultural real wage and
areas, while in rural areas the earlier the lagged and current agricultural output
n is maintained.
per net sown acre. In addition the model
These observations suggest a much more includes the lagged dependent variable and
complex relationship between growth of non- a time trend. In turn, the level of the
agricultural incomes and poverty than usually agricultural real wage is ‘explained’ in
appreciated. As expected, commercialisation another equation by the inflation rate and
does indeed seem to be associated with the earlier agricultural output variable, in
increased rural inequalities, but it seems also addition to a lagged wage term. In this
to be associated with some increased model, therefore, higher agricultural output
opportunities for the urban poor. On the reduces the incidence of poverty both directly
other hand, the expansion of other non- and through its positive effect on the wage
agricultural incomes appears to have reduced rate; and inflation works on poverty only
rural poverty while doing nothing to reduce indirectly through the wage rate.
urban poverty. This suggests that to the
Table 9 gives the coefficients of the
extent that the benefits ofsuch income growth Ravallion-Dutt poverty equation obtained
do percolate down to the poor, this spills when fitted it to the periods 1960/61-1989/90
over disproportionately to the rural sector, and 1960/61-1992 with all-India data. The
either because of rural-urban migration or first fit, which does not include the post
®
2472
reform period, is almost the same as that
reported in their original paper and appears
to be a fairly good fib However, the second
fit, that of the same model fitted to data
extended to the post-reform period shows
that the model breaks down almost
completely since the most important variable,
agricultural output per acre, turns totally
insignificant. Furthermore, the breakdown
of the model occurs essentially because the
magnitude of the actual increase in poverty
is well beyond anything that this model can
predict. In fact, when the model estimated
with data up to 1987-88 is extrapolated, it
is able to explain only a small part of the"
large actual increase in 1992, and also predicts
an increase in poverty in 1993-94 as against
an actual decline.”
This table also gives the details of an
alternative model fitted to the same data
incorporating our observations in the
preceding discussion. Here in addition to the
agricultural output and real wage variables
we included the'relative price of cereals, the
proportion of non-agricultural workers in
nira* population and the commercialisation
variable (the per capita GDP from trade and
transport). An interesting observation is that
on inclusion of these variables, the time
trend and the lagged dependent variable
used by Ravallion-Duttt turn insignificant,
suggesting that in fact the adjustment of
poverty to real factors is much faster than
thatsuggested by theRavallion-Dutt model.
This alternative model not only fits past
data much better than the Ravallion-Dutt
version, but also, unlike theirs, remains
robust when extended to the post-reform
period. In parti-cular, the massive increase
in rural poverty in 1992 is predicted with
complete accuracy by the model fitted up
to 1987-88.19
These results not only emphasise the
importance of the relative price variable and
of non-agricultural factors, they cast strong
doubts on the simple agricultural output/
inflation paradigm. This paradigm is further
compromised because it is seen that the
wage equation in the Ravallion-Dutt model
also collapses in the sense that inclusion of
alternative variables, e g, real government
expenditure per capita, renders the
agricultural output variable insignificant.
Indeed, the significance of the government
expenditure variable here confirms the pos
sible i mportance of government expenditure
for non-agricultural employment, the rural
real wage and, therefore, rural poverty. And,
indeed, including agovemment expenditure
variable along with lagging the employment
variable yields our preferred equation.
Interestingly, this equation fitted to data up
to 1987-88 accurately tracks the subsequent
movement in rural poverty, including both
the sharp upward movement in 1991 and
1992 and the reversal in 1993-94.10
Economic and Political Weekly
Special Number September 1996
Nonetheless, these observations on the
basis of All-India data, though indicative,
cannot be conclusive given that the degrees
of freedom are few and because a number
of possible explanatory variables are
mutually correlated. For this reason, the
excercise was repeated using state level data,
in the form of a pooled time-seris and cross
section analysis with data up to 1992. In this
exercise, poverty was regressed against
agricultural output per rural person, state per
capita SDP. a relative food price index
calculated by dividing the index of the food
price in the CPIAL by the SDP deflator, an
inflation index based on the SDP deflator,
and per capita real state development
expenditure. All variables except the per
capita SDP were significant, but, in addition,
the inflation term was small and just crossed
thesignificancelevel. The relative food price
variable was easily the most statistically
significant variable and it also was the most
important in terms of its impact. The next
important variable was state development
expenditure, followed by agricultural output.
These results with pooled cross-section
and time series data at the state level are in
many ways similar to results obtained
recently by Dutt and Ravallion (1995) with
more or less the same data set, but restricted
to the period up to 1989-90.21 In their model
they regress poverty measures against
agricultural output per hectare, state
development expenditure and an inflation
term, along with certain indicators of initial
conditions (e g, irrigation, infant mortality
and female literacy). All the variables had
the expected sign, and, interestingly, they
find that state development expenditure was
the most significant variable and that, unlike
agricultural output which reduced poverty
only by increasing mean consumption, state
expenditure reduced poverty by increasing
both mean income and improving income
distribution. They did not includeany relative
price variable but our experiments with the
same data suggest that this would have
swamped the inflation term had they done
so. Hence, theimportanceof stateexpenditure
and of the relative food price appears to be
fairly robust as factors explaining poverty
both acrosis time and space.
However, perhaps the most important
result of this Dutt-Ravallion exercise is that
it shows that, quite apart from the
contemporaneous effect of prices, output
and government spending, theextentto which
a particular state could reduce poverty over
time depended also on the initial conditions
with respect to physical and human
infrastructure, in terms‘of irrigation, female
literacy and infant mortality , with which that
state began. .Thus,’ of the difference of 1.8
per cent per annum between the rates of
poverty reduction in Kerala and Madhya
Pradesh, fully 1.6 per cent per annum could
Economic and Political Weekly
be attributed to the fact that Kerala began
with higher female literacy (I per cent) and
lower infant mortality (0.6 per cent). Our
own preliminary experiments with such
initial conditions confirm the long-run
importance for poverty reduction of health
and education status, though much less so
of irrigation, and suggest furthermore the
importance of initial land distribution.
Clearly, this analysis suggests that both
thcbcnignagriculturaloutput/inflation model ■
and the relative price/state expenditure/rural
non-agricullural employment models
mentioned earlier are relevant for the
determination of rural poverty. But the really
important conclusion is that, of the two, the
second is by far more important: agricultural
output and inflation do matter, but as
determinants of the incidenceof poverty, the
relative price of food and the level of
government expenditure are even more
important. In addition, the analysis points
to an important and hitherto ignored longrun synergy between efforts at improving
the health and education status of a society
and its ability to bring down poverty over
time.
V
Policy Conclusions
The results above do not lead to any very
optimistic prognosis about the effect of
structural adjustment or of further' marketist
reform’ on poverty. It is true that ceteris
paribus an increase in agricultural output
would reduce poverty and that, therefore,
there is a case for diverting more resources
to this sector. It is also true thatanyexpansion
of employment in the unorganised sectors,
say through the rapid growth of labourintensive exports, would also reduce poverty.
And it remains extremely plausible that any
policy which can moderate inflation without
leading to a cut-back in employment
opportunities would in general benefit the
poor. Nonetheless, there are trade-offs
involved in achieving each of this goals in
the structural adjustment package, and it is
precisely these trade-offs which are cause
for pessimism.
The basic thrustofthe structural adjustment
strategy is to allow greater play to market
■forces and to ensure that domestic relative
prices reflect the opportunities available in
international trade. In theory, domestic
liberalisation would cause a greater degree
of commercialisation, and liberalisation of
international trade would cause shifts in
relative prices in favour of agriculture and
exportables. Taken together, these are
expected to bring about the desirable shift
of resources towards agriculture and labourintensive exports. However, the problem
with this is that not only do these very
mechanisms, commercialisation and a rise
Special Number September 1996
in the relative price of agricultural products,
act directly to increase poverty, but also that
the magnitudes of the elasticity of poverty
to these make it extremely unlikely that the
direct loss on this account can be made up
by the indirect benefits accruing from the
better resource allocation that is expected to
result thereby. For example, in almost every
poverty equation reported above, theposmve
coefficient on the relative price variable is
twice the absolute size of the negative
coefficient on the agricultural output variable
- implying that the elasticity of food
production to the relative priceof food would
have to be greater than two if such a change
in relative price is to reduce poverty through
higher food production.
This cautions against any sudden opening
up of the foodgrains sector to international
trade; and, indeed, the caution here should
be greater than simply one of managing a
careful transition to world prices. The fact
that the relative price specification is more
important than the inflation specification
suggests that the underlying problem is
caused by more than a stickiness of money
wages in the face of price increases. If such
stickiness existed without any long-run
impact of.relative prices on poverty, the
problem would have only been a transitional
one which could be managed by keeping
inflationary pressures in check either by a
graduated movement to world prices or
through a moreeffectivestabilisation policy.
However, since poverty is extremely sensitive
to relative prices there is more than a
transitory problem with opening up
agriculture to international trade. Also, with
government expenditure important for
poverty, there is the further important trade
off between this direct effect and the indirect
effect, through inflationary pressures, of the
fiscal policy stance. Given the relative
importance of the government expenditure
and inflation variables in the poverty
equations, attempts to use contractionary
expenditure policies to deal with inflation
pressures, say as a result of a greater opening
up to international trade, could prove to be
a case of the medicine being worse than the
desease.
In any case, there appears to be
considerable evidence that the increased
government spending during 1976-90 was
among the principal reasons why India could
record rather impressive declines in poverty
during this period. However, it is also true
that the sustainability of such expenditure
increases in the future is more doubtful than
ever before. During 1976-90, real per capita
government development expenditure
increased at an annual rate of 6 per cent per
annum as against only a 3 per cent growth
i<! per capita real GDP. Real government
expenditure percapitafell 15 percent during
1990-93, but increased again by 6 per cent
2473
in 1993-94. The earlier expansion of
government expenditure had led to large
fiscal imbalances despite (he fact that taxGDP ratios had then grown quite
significantly. On the other hand, both GDP
growth rates have been lower and tax-GDP
ratios have been falling in the post-reform
period. It is therefore unlikely that the pace
of growth of government expenditure can
be sustained unless GDP grows at least 8
per cent per annum or there is a definite
policy of increasing the tax-GDP ratio
significantly.
Given this fiscal reality, and the fact that
non-agricultural GDP does not appear to
have much impact on poverty except through
its effect on the sustainability of government
expenditure, it is obvious that there will be
problems with maintaining the pace of
poverty reduction. Even if GDP growth
increased, the current fiscal priorities make
it unlikely that this would be reflected fully
in public expenditure. One possibility
discussed in this context is to alter the
composition of government expenditure so
that it is more directly focused on poverty
alleviation. But, although this is possible
and desirable, a note of caution must be
sounded on this at the outset. In our regres
sion excercises, we played around with
different components of government
expenditure, and the results suggested,
somewhat surprisingly, that it was the broader
measures of such expenditure which had a
greater poverty alleviation effect, at least
when poverty is measured by the head count
ratio, than narrower and more focused
measures such as that on agriculture and
rural development.22
There are two possible reasons for this.
First, it may well be the case that the existing
poverty alleviation programmes are not
particularly effective and that their impact
on poverty is no greater than other govern
ment expenditure. If so, there is room for
improvement in the design of expenditure
focused towards poverty alleviation. And,
7) indeed, a case can also be made out that it
f may be possible to transfer funds from such
programmes to rural capital formation
without endangering the poverty alleviation
impact. But, secondly, it also appears that
what is really at issue are much broader
multiplier effects of overall government
expenditure. Clearly much more work is
required in this area to identify ways in
which the impact of a given amount of
expenditure can lead to more poverty
alleviation, but, although there are certainly
..Tikely to be ways of achieving this, it should
not be expected that it will be possible to
cut-back overall government expenditure
without any adverse effect on poverty. The
real significance of government expenditure
appears to be that it is this which imparts
any ‘trickle-down’ characteristic to the
Economic and Political Weekly
growth process, something which appears
quite weak if only GDP growth is considered.
Nonetheless, since this effect is likely to
be greater if government expenditure is
properly targeted, it is necessary to attempt
a brief evaluation of the contribution of
government anti-poverty schemes in the
reduction of poverty. There have been
numerous evaluations of these made by the
government and by independent researchers,
and no attempt will be made here to review
this literature which attempts to measure the
effectiveness with which particular schemes
have been able to target the poor. Suffice
it is to note that such evaluations have by
and large found that asset-creation schemes,
such as the Integrated Rural Development
Scheme, have had less success in alleviating
immediate poverty than rural employment
programmes, although even the latter have
leakages and are often criticised for being
a palliative whose effectiveness at permanent
poverty reduction are rather low. However,
comparing the official figures on employment
schemes with independent data from the
NSS, four points are worth making. First,
with the NSS reporting a quantum of
employment in public works which matches
official data well, fears about large leakages
may be rather exaggerated. Secondly, the
schemes appear to be reasonably welltargeted in that they are availed of most by
casual labour households which have both
the highest poverty and the highest personday unemployment, but the regional
distribution of employment through such
programmes appears to be concentrated in
a few western Indian states, and also public
works appear to have been much more
effective in 1987-88, a drought year, than
in more normal years. Thirdly, it seems
unlikely that the effective transfer through
such schemes was much lower than the wage
cost as a result of incomes foregone by the
workers to take up .such ■employment.23.
Fourthly, possibly, because they are well
targeted, public works appear to have been
more effective in moderating the severity of
poverty rather than its head count incidence.24
Conceptually, if viewed primarily as an
anti-poverty measure, a well targeted public
works programmes should not provide
incentives for the non-poor to participate
and nor should there be impediments to
participation by anyone who is poor. At least
till 1987-88, Indian schemes seem broadly
to have passed the first test but,-except.
possibly in Maharashtra, failed the second
both because of a paucity of funds and a lack
of official commitment, except at times of
natural disasters. Since then, confusion about
the intention behind such schemes seem to
have increased. First, wages.offered have
been increased to the statutory minimum
wage rate which is often higher than locally
prevailing wages, thus making participation
Special Number September 1996
more attractive. Second, despite this, funds
available for such schemes have been cut in
real terms, causing job availability to be
even more rationed. Finally, as a result of
a misplaced importance given to the head
count poverty incidence measure there is a
feeling that these schemes have failed to
reduce poverty, and this, combined with a
general presumption that investment rather
than doles are what is really necessary, have
led many to argue for an increase in the
materials and expertise content of these
schemes, at the cost of their unskilled labour
content, so as to make them more viable
instruments of rural investment.
These reactions are confused because the
primary goal of an anti-poverty measure is
not the creation of assets and nor is its
purpose a general redistribution of income,
say by increasing the general wage level.
This is not to argue that these are not laudable
objectives but simply to point out thau
attempts to chase too many objccti ves without"
substantially increasing the budget available
risks diluting the primary goal of poverty
alleviation. Both higher wage rates and a
lower component for unskilled labour in
these schemes reduce their transfer content.
And, these objectives, by attracting richer
workers and/or by directing employment to
regions where viable investment projects,
rather than the poor, exist are also likely to
make for much less effective targeting.
Possibly, the correct approach would be to
make employment guarantee the primary
concern of such programmes, setting the
wages paid to levels where the demand for
such employment would broadly match the
funds available. Clearly, if more funds can
be directed into such programmes the wage
rate paid can be increased, and with sufficient
expenditure even the general wage rate
influenced within the employment guarantee
framework. The employment guarantee
aspect should, however, be the primaryl
concern and higher wages the secondary
concern because only this priority would
keep out the relatively rich while allowing
the .poor unimpeded access.
Secondly, the best way to dovetail
producti ve i n vestment i nto such a programme
would possibly be to give a wage subsidy
equal to the employment guarantee wage
rate for each unskilled worker working on
a class of well-defined approved investment
projects, delinking project choice from the
agency implementing the guarantee scheme
and treating the rest of the project cost and
benefit on par with any other. With the
employment guarantee scheme in place, this
need not cost the exchequer any more and
yet the linkage between poverty alleviation
and productive investment through labourintensive schemes could be decentralised.
Needless to say, this means that other project
costs would have to be met from outside the
2475
employment guarantee budget. But this is
the proper way of proceeding because while
there is a case for subsidising employment
if there is paid-underemployment at the
normatively chosen employment guarantee
wage rate, there is no case to subsidise any
particular investment more simply because
it is selected by the agency implementing
the employment guarantee scheme.
However, the main lesson from earlier
sections is that the basic thrust towards
permanent reduction of poverty must take
the form either of increasing employment
in agriculture, mainly through better
irrigation and multiple cropping, or of
increasing the stock of viable self
employment opportunities or regular jobs in
non-agriculture. It is towards these objectives
that rural investment should be encouraged
while employment guarantee provides the
framework within w'hich this can be done
without sacrificing the need to combat
Qwerty immediately. Yet, because the
"forms themselves have aspects which tend
to increase poverty, and because fiscal
considerations mean that it might be difficult
to increase both agricultural investment and
the expenditure on anti-poverty schemes,
there will be difficulties in the future.
Under these circumstances, it is clear that
1 See e g, C P Chandrasekhar and Abhijjj <jcn. ■'.$/.
‘Has Poverty Declined with Reforins?.' B
if poverty reduction is to be a serious part
Macroscan; Businessline, January 23, I995’ B
of the agenda in the reform period, the
See also Jaya Mehta, ‘Poverty figures and
reforms themselves should have an explicitly
the People of India’ in Alternative Economjc
redistributive content. This would require
Survey 1995-9(5,-published by the Delhi®
cuts in subsidies to the rich and also higher
Science Forum forJhe Alternative Survey
taxes to mai ntai n and increase the expenditure
Group.
r,'.relevant for the poor. In addition^ the old g ;2 OzlcrDutt and M Ravallion, ‘A Database
issues of land distribution and the provision .
on Poverty and Growth in India’, Poverty and.®
Human Resources Division, Policy Re$rnrrh~ B
of universal primary health and education
Development. The World Bank, January 1996.
must again be put back on the agenda. But,
oil;' This data base which is available in diskette
more than anything else, it must be recognised
form contains data up to the 48th Round ®
that a ‘reform’ strategy which aims to
(January-December 1992) pf the NSS. It also®
withdraw the state from investment, liberalise
has compilations of the poverty line for the®
50th Round (July 1993-June 1994), but does®
finance and thus divert finances from the
- not contain poverty estimates for this round .
state to the private sector, liberalise
because the NSS data was not’available. The®!
agricultural trade and thus enrich the rich
• poverty estimates for 1993-94 are ours, using®
at the direct cost of the poor, and seeks to
the World Bank poverty linei (which use the®
control inflation and BOP problems through
Planning Commission benchmarks of Rs 49/®
deflation and devaluation is at its root a
and Rs 57 at 1973-74 pnccs: for rural .andfundamentally inequitous adventure.
urban India but use somewhat different®
deflators) and the World Bank computer®
Notes
programme POVCAL for poverty estimate®
calculations with the 50th Round NSS .‘5*
[This is a slightly revised version of a paper
estimates.
delivered at the Workshop on Economic Reforms
3 This relies heavily on Tendulkar, S, DT.-I
and Poverty Reduction organised jointly by the
Sundaram and L R Jain, ‘Poverty in India .
Institute of Development Studies, Sussex and the
Lal Bahadur Shastri National Academy of
1970-71 to 1988-89’, ILO-ARTEP, 1993. '.
Administration, Mussoorie, and held in Mussoorie
4 See the next .section, and particularly
in February 1996.)
Table 7.
•; {(> ’ \
I
•• ■
:: ■■r'Jf
We've given you a series of firsts.
Because you come first and foremost
UHM
Asian Paints.
«
■lit
A series of firsts
add up to make us No. 1
......^
> •.:>:i.'j'/.vwerd
When you’re going to be first, you try a little harder. That’s why we reach out to you with the widest
,
distribution. The most competitive prices. And, of course, the largest range.
. . • It takes foresight to come out first.
.... First with a comprehensive retail network. Our products are sold in over 25,000 outlets spanning the length and breadth of the country.
First Indian paint company to set up a subsidiary abroad.
'
First with Super Decoplast - India’s No. 1 affordable Plastic Emulsion Paint
□
□
First with computers for integrated management systems. Also the first Colour Control System for 100% shade fidelity.
□
First with corrosion-resistant protective paints for Urea Prill Towers.
First with Krilo - India's first pure acrylic car finish
28 years of leadership JWasian paints
2476
Economic and Political Weekly
Special Number September 1996
Politics, Institutions, Poverty
C .O
, ]7
The Case of Karachi
S
Akbar Zaidi
Poverty alleviation is the trendy and fashionable slogan for the end of the 1990s. Projects will be defined with a
specific focus on the poor, often with the help of donor money channelled through non-governmental organisations
(NGOs). This attempt, while well meaning, will invariably be at a micro level with a narrow focus, often ignoring the
causes for the existence of poverty in the first place. Band-Aid social work of this variety will certainly improve the
living conditions of a number of beneficiaries in the project area. However, poverty is primarily a political issue, caused
and maintained by factors of a macro nature and by institutions which function in a specific, political, environment.
This paper argues that politics comes prior to poverty, as do institutions. The failure of institutions to address issues
ofpoverty and development are seen here as essentially political failures. Looking back over the last decade, it would
be difficult to find a more politicised, violent, ethnically divided, alienated city than Karachi. The paper concludes with
the assertion thatfar-reaching and substantial political and institutional reform, must comefirst in any attempt to alleviate
poverty, particularly in Karachi.
INTRODUCTION1
IN order to understand the nature and extent
of poverty, especially in the setting of a large
metropolitan city with a population of 11
million, such as Karachi, it is necessary to
understand the institutional and political
relationships which exist in that city. The
causes and reasons for persistent and recur
ring poverty may lie in the failure of political
and institutional responses in dealing with
poverty, both at a local and at a national level.
In fact, even the creation of poverty, in the
first place, may be the outcome of specific
policies followed by different levels of
institutions in response to political circum
stances and imperatives. Hence the title of
this paper, where the city of Karachi provides
a case toexaminethedialecticsof institutions
and politics, in order to assess the causes,
nature and extent of poverty in the city.
There has been a sea change in the way
governments function and in the way
institutions act on policy priorities in much
of the underdeveloped world. The military
dictatorships ofyesteryear have been replaced
by evolving forms of democracy, ranging
from the autocratic to the more liberal. Highly
centralised states, where much of policy was
formulated, have weakened their hold over
lower tiers of government, with states/
provinces and district and local level
government playing an increasing role in
public policy and in the delivery of basic
services. The nature of politics too, has
changed. With the demise of the Soviet
Union and the perceived failure of socialism
with it, politics in much oftheunderdeveloped
countries of the world has also undergone
a radical transformation [see for example
Zaman 1993; Zaidi 1994a, 1994b],
The guerrilla struggles of Latin America
and the parties closely aligned to the
Communist Party of the Soviet Union
(CPSU), subscribing to the non-capitalist
path of development, have for the most part,
3282
been transformed into social democratic and left
leaning parties. Even Marxistparties have changed
their colours, from fighting class enemies and
imperialism to becoming more expedient, chasing
multinational corporations inviting them to invest
in their countries. With the change in ideology,
politics too,.has changed. .
The dialectical relationship between politics
and institutions has given impetus to the NGO
movement. While institutions of the state and of
the private sector have existed since time
immemorial, institutions which belong to neither,
have coalesced in recent years, offering this third
alternative. The NGO movement in some countries
has replaced the role of the state in the delivery
of certain social services and has even acquired
the status, of a quasi-state in some regions
[Zaidi 1994b]. In others, the failure of public
institutions has not helped bring forth alterna
tive forms of government, organisation or
institutions, giving rise to anarchy, and at times,
even carnage.
:
. p
It is these changing perspectives and
relationships in which we focus on the issue of
poverty in Karachi. What has been the role of
national institutions of thestateincreatingpoverty
in the first place, and how have local institutions
which deliver social services, responded to the
challenges of rigid structures, inadequate revenues
and resources, and insufficient powers? Much of
Karachi’s recent history has been one dominated
by volatile and violent politics, with the most
popular political party often discriminated against
at a national level and sometimes targeted militarily.
One cannot ignore such facts in examining the
relationship of institutions with politics, when
they address issues of poverty. Then, the issue
of poverty, itself, is no less complex. How does
one identify and measure it? What sort of indicators
would be useful in determining the extent of
poverty? How relative should our measurements
be? Relative to what? and soon. While the location
of poverty and what subscribes it can be an endless
debate, this paper tries to briefly discuss the
difficulty in taring to identify and measure it
within our context.
"—
•
This paper begins with an, account
of how Karachi has grown in the last x
50years and is now destined to become
one of the most populated cities of the
21st century. We look at the city’s
history and its growth and then turn to
an evaluation of how Karachi’s
institutions function. The process of
planning plays a fundamental role in
how the city has developed, and this
role is examined in detail in the paper.
The next section then discusses the
issue of poverty in general and then in
the specific context of Karachi. Since
politics has dominated the life' of all
citizens in the city, one section of the
paper examines how institutions have
been affected by the often violent
politics and what the impact on poverty
has subsequently been. The role of less
formalised politics and institutions, in
particular the role of some NGOs in
alleviating poverty arid addressing ,
.. ?
issues of social and economic
deprivation, isalso touched upon during
the course of this paper. Finally, an
attempt is made to tie all loose ends
together, and focus on future strategies
for poverty alleviation, within the
context of political and institutional
developments in Karachi.
The Context of Karachi2 .
The population of Karachi in 1729
was 250; today it is estimated to be
around 11 million with the Karachi
metropolitan area’constituting about
4,25,000 acres, or over 3,500 sq kms
(Table 1). The growth in the size and
population of the city has been uneven
and sporadic, often influenced by
factors outside the control of either the
citizens or the administrators of the
City.
■
• Su
Karachi, when it was conquered by
the British in 1839, was a small town
Economic and Political Weekly
December 20, 1997
.A
j
I
needs of the new state. In addition, over
With a population of approximately 14,000 exporting more wheat than Bombay. As
6,00,000 refugees from India moved into the
wi,th few Sindhis living in this town. The Karachi’s population grew towards the end
city increasing its population by more than
capital of Sindh at that time was Hyderabad of the last century, some services were
161 per cent in a period of 10 years. The
where the rulers, the Mirs, lived. Hyderabad developed in the old town in the city, as well
refugees occupied all the open spaces and
was conquered by the British in 1S43, four as on the outskirts of Karachi, where water
in the city centre, the military cantonment
years after the conquest of Karachi, and it supply and drainage systems weredeveloped.
and public buildings. This migration changed
was the capture of Hyderabad, rather than The Karachi Port Trust was also established
Karachi completely. An imported language
thatof Karachi, which marked theannexation at this time.
and culture dominated the city and this in
the long run has given birth to many of
of Sindh to the British empire. Alexander
For the first four-and-a-half decades of
Karachi's and Sindh's ethnic and political
Bailie, writing in 1S90, writes that Karachi this century, Karachi’s population grew at
problems, [Hasan 1994:8].
had no importance of any sort prior to the a steady pace affected by agricultural
In 1951 when the first census of Pakistan
occupation of 1839 and in the Sindh campaign development that was taking place in Sindh
of 1843. He says that it was not even and the Punjab, and by the expansion of trade was held, Karachi’s population had risen to
mentioned in the political dispatches of the and some industrialisation that took place in 1.137 million of which 8,15,000 (as much
British. Since Hyderabad and Thatta had the city itself. Much of Karachi’s growth was as 72 per cent) were migrants from India.
been the places of residence and of political based upon migration from other parts of Of the 2,30,000 Hindus that lived in Karachi
power prior to the occupation of Sindh, these undivided India, attracting Parsecs, Bohras, prior to partition, 1,70,000 (74 per cent) had
cities gained importance. Further, they were Gujratis and other trading communities from left the city. Now, the main language spoken
both integrated with their rural hinterlands other towns in India. The port and the railway in Karachi rather than Sindhi or Baluchi, was
where surplus production from agriculture link with the restof India provided theimpetus Urdu, spoken by more than 50 per cent; only
found its way to the towns where the rich for growth [Hasan 1994]. Arif Hasan has 8.6 per cent now spoke Sindhi. The
had established their homes. While these shown that at the time of formation of Pakistan population was also overwhelmingly
two cities were relatively thriving, Karachi in 1947, Karachi had a population of around Muslim, with as much as 96 per cent of
was an outlier, of little economic or 4,50,000 with slightly over 51 percent being Karachi’s population (like the rest of west
commercial importance, prior to the advent Sindhis, followedbytheBaluch, Urdu-Hindi Pakistan) now Muslim, and less than two per
of the British [Baillie 1975].
speakers, Punjabis and Gujratis. More than cent Hindu in what was predominantly a
Karachi’s first spurt of prominence, though half of Karachi's population was Hindu and Hindu city a few years earlier [see Zaidi
1989, 1991, 1992a; and Hasan 1997].
not reflected in the increase of its population, 42 per cent Muslim; Christians and Parsees
As the economy of Pakistan began to
came about due to factors many thousands made up 3.5 and 1.1 per cent, respectively,
of miles away. The American civil war had of the city’s population. Apart from the develop in and around the capital of Pakistan,
resulted in the disruption of the cotton trade metropolitan area, Karachi districtcontained which was then Karachi, so did its economy,
from the US. To substitute for this shortfall, over 1,200 settlements and villages [Hasan size and population. With the need for
administrative and service sector staff to
there was increased demand from British
1997],
India and in particular, from Sindh, in the
The nature of Karachi was dramatically serve the capital, educated and skilled
period 1861-65. Sindhi cotton replaced altered by the advent of Pakistan. As Hasan individuals moved to Karachi from other
parts of Pakistan, as they had done from
American cotton as raw material in the textile argues,
India. The advantages of being the only port
and cotton factories in Britain. But this boom
In 1947 Karachi became the capital of the
in demand, with its impact on Karachi and
new state of Pakistan. Bureaucrats, and proximity to government meant that
government employees, semi-government industry preferred Karachi’s location and
its port, was a temporary phenomenon, as
organisations all moved to the city and new much of the large-scale industry began to
after the end of the civil war, trade in Karachi
organisations were established to meet the develop here, much of it with government
had almost been halved.
One reason why Sindh was able to
Table 1: Karachi’s Population Growth .
capitalise on the demand for cotton in Britain
, Average Annual
Population
Increase/Decrease
Per Cent
Year
to replace American cotton, was the
Growth Rate
over Last Census Survey Increase/Decrease
development of the Sindh railway in 1861
which made transport to the port much easier 1729
250 "
14,000
and faster. Due to the development of the 1838
2,77?
19.80
16,773
railways, smaller towns inside of Sindh also 1850
22,227
5,454
32.51
1853
began to expand, as market and trade grew,
56,879
. 34,652
155.90
1856
with Hyderabad and Shikarpur benefiting
56,753
,126
1872
the most. Railways were extended into the
73,560 .
16,681
1881
, . 29.32 .
Punjab in 1869 linking them with the rest 1890 .
98,000
33.22
24,440
of northern India. The population of Karachi,
136,297
38,297
1901
39.07
37.00
186,771
50,474
1911
despite a short spurt of activity between
30.70
1921
244,162
57,391
1861-65, stagnated between 1856-82 and
23.20
300,779
56,617
1931
grew by a mere 29 per cent in these 25 years.
3.70
435,887
44.90
135,108
However, growth in population and activity 1941
11.50
1,137,667
161.00
701,780
1951
was far more robust after this period.
6.05
2,044,044
.
79.70
906,377
1961
Between 1881 and the end of the century,
5.00
• 3,606,746
1,562,702
76.50
1972
Karachi’s population nearly doubled, as
4.96
5,437,984 .
50.80
1,831,238
1981
4.07
Karachi's importance as a port and as a 1986
36.80
7,443,663
2,005,679
10,250,000*
2,806,337
37.70
conduit for bringing commodities from the 1991
• 13,500,000*
31.70
3,250,000
Punjab, northern India and upper Sindh, 2001
increased. The opening up of the Suez Canal Notes: Average per cent from 1838 to 1941: 3.4 per cent; • Estimated.
in 1869 had made Karachi the nearest port Source: Hasan, Arif (1994) Profile of Three Pakistani Cities: Karachi, Paisalabad and Thatta, report
prepared for the International Institute for Environment and Development, London, p 4.
in India to Britain. By 1886, Karachi was
Economic and Political Weekly
December 20, 1997
3283
help and patronage. The trade and export
boom of the early 1950s which consequently
resulted in the emergence of an industrial
sector, was also beneficial to Karachi and
its growth, as many of the traders who later
became industrialists, all lived in this city
[Zaidi 1989, 1991, 1992a; Sayeed 1995].
In the 1960s, with a change in the way
agriculture was managed following thegreen
revolution, and with industrial and economic
growth ofas much as 17percenton average
per annum between 1959-65, and again, 9.9
percent between 1965-70, Karachi prospered
and grew as it had a disproportionate share
of industry and the productive sector located
here. ForHasan, the developmentof industrial
production and the introduction of the green
revolution meant that Karachi also benefited
from increased international and domestic
trade:
Due to this, port activities in Karachi more
than doubled during these two decades. The
rural areas of the Punjab alone produced an
average surplus of 3,600 million rupees per
year from agriculture during this period and
most of this was invested in Karachi’s
economy [Hasan 1994:9],
For all these reasons, ranging from the huge
influx of migrants from India and later from
within Pakistan, and since it was the only
port and subsequently the main beneficiary
of the type of industrial policy pursued in
the years between 1947-72, “Karachi was
perhaps the fastest growing city in the world”
[Hasan 1995:4], From 1972 when the
population of Karachi was 3.6 million, the
population grew by an average of 5 percent
till the next census in 1981, when the
population had reached 5.4 million. Since
no census has been held since, estimates
today place population of Karachi at between
11 to 12 million. About half of Karachi’s
growth in the 1970s was attributed to
migration from the rural and urban areas of
the country. However, Hasan writes, that
since 1981, while Karachi’s annual growth
has declined considerably in percentage
terms, it has increased in numbers; now most
of this growth in Karachi is due to natural
increase and not due tojmigration [Hasan
1995:4], In the 1980s, an estimated one
million inhabitants (orone-tenthof Karachi’s
population), were thought to be migrants
from Bangladesh, Afghanistan, Iran and even
from as far away as Burma.
Karachi is the largest metropolitan city of
Pakistan. It is the industrial, commercial and
trade centre of the country and has a well
developed economy which continues to show
growth rates in excess of 6 per cent per
annum. Today, Karachi houses around
8.8 per cent of Pakistan’s total population
and 24 per cent of the urban population.
The population growth rate is estimated to
be around 5.6 per cent per annum of which
3 per cent is Karachi’s natural growth and
3284
between 2-3 per cent due to migration.
Karachi provides onequarterof the country’s
federal revenues and 23.2 per cent of GDP.
Not surprisingly, since it is the industrial and
financial capital of the country, more than
half of the country’s bank deposits lie in
banks in Karachi and almost three-quarters
of all issued capital is raised in the city
[Hasan 1992; Zaidi 1991,1992; AERC 1993],
In addition, “32.9 per cent of the national
value added in manufacturing, 26.4 percent
in trade, 61.6 per cent in banking and
insurance and 37.7 per cent in services, is
generated by the city” [AERC 1993, vol 4,
P 271],
Karachi has a high per capita income $ 900 in May 1993 - almost two-and-a-half
time’s Pakistans GDP per capita. Estimates
show that more than 30 per cent of income
accrues to the relatively affluent households
in the city - those who earn as much as 10
times the country’s GDP per capita [AERC
1993, vol 4, p 3]. Other indicators also
clearly show Karachi’s dominance in the
economic and social sphere: the male literacy
rate for Karachi is 20 per cent higher than
for Pakistan’s overall male urban literacy
rate, while the female literacy rate is almost
twice as high as for the rest of the urbanised
country. Although only 8 per cent of
Pakistan’s population lives in Karachi, it
owns about 35 per cent of all television sets
in the country and almost half of the cars
registered in Pakistan, are registered in
Karachi [Zaidi 1989, 1992].
Almost all economic and social indicators
show that Karachi is still way ahead in terms
of development compared to the rest of the
country, despite the fact that other areas,
notably central Punjab, have benefited due
to changes in Pakistan’s economic, social
and political fabric [Zaidi 1989, 1992].
Nevertheless, the richest and largest
settlement in the country, which supplies a
large amount of revenue to the exchequer,
has been confronted with serious problems
which have hampered further development
and progress. While the collapse and failure
of government institutions has been allpervasive, Karachi’s fate and problems may
have been compounded by the uniqueness
of the city’s political developments. Violence
and camage dominated Karachi’s landscape
for much of the 1990s and exacerbated the
normal and typical problems associated with
that of a large and fast growing metropolis.
It is to these issues to which we now turn.
Institutions andplans: Formostofthe last
50 years, much of the development that has
taken place in Karachi, has been that
undertaken by government departments and
institutions themselves, or then as a response
to government action. Government
institutions have, for the most part, not
surprisingly, dominated in the development
process, providing housing and basic urban
infrastructure. However, as we argue in this
section, much of government interaction and
planning has been inadequate and at times,
an outright failure. In response to such failure,
the informal sector has blossomed in almost
every location of the city and in almost every
type of activity. The rise of NGOs has also
been a response to the failure of the state
to offer viable options and solutions.
Municipal government and its institutions:
Karachi division consists of five urban and
one rural district with more than 96 per cent
of Karachi’s population classified as urban.
The Karachi metropolitan area consists of
the five urban districts and each district is
an administrative unit. The Karachi Metro
politan Corporation (KMC) established in
1853, was the oldest municipal committee
in British India. The KMC is supposed to
be an elected bpdy for the entire metropolitan
area and consists of 232 councillors, one for
each ward. The councillors in turn, elect the
mayor. The five urban districts in the Karachi
metropolitan area act as somewhat auto
nomous district municipal corporations
(DMCs).
The KMC is an institution of local
government in Pakistan and before we
examine how this specific institution
functions, it is important to put the KMC
in the’broader perspective of local
government in Pakistan. This briefdigression
is made necessary precisely because local
government has had a strange and difficult
relationship with the other tiers of
government. For example, despite the
introduction of elected local bodies vide the
Local Government Ordinance of 1979 and
1980, for the last four years, since 1993, all
local bodies (except those in sparsely
populated Baluchistan) have been dismissed
by their various provincial governments.
Hence, with no elected representatives at,
perhaps, potentially the most important level
of government, local government in Pakistan
is manned by government bureaucrats.
Clearly, in assessing the role and prospect
Table 2- Planned and Unplanned Areas
in Karachi: A Comparison _
(Percent)
•’
■-
Planned Unplanned '
Permanent housing
80.0
structure
20.0
Semi-permanent
Number of persons
0.5
per room •
83.0
Water connections
98.4
Electricity connections
75.3
Gas connections
Access to sewage facilities 85.0
Solid waste management 60.0
75.9
Literacy levels
20.0
80.0
3.3
50.0
75.8
35.1
12.0
10.0
48.6
Source: Hasan, Arif and Asiya Sadiq (1994),
Mapping City Inequality: A Case of
Karachi, report prepared for the
International Institute for Environment
and Development, London.
Economic and Political Weekly
December 20, 1997
of local government at a city level in addres- sing issues of poverty and the provision of
basic services, this important fact cannot be
overlooked [see Zaidi 1996, for an extensive
evaluation of the nature of urban local
government in Pakistan).
The first significant fact about the existence
of local government in Pakistan is that it has
no constitutional provision as it is not a
formally embedded part of the constitution.
Local governments exist under the super
vision, control and even benevolence of the
various provincial governments, where pro
vincial governments have merely delegated
some of their functions and responsibilities
to local governments by the promulgation
of ordinances which specify the allocation
of residuary functions of local government.
Although local governments have existed
in the Indian subcontinent for many centuries,
the areas which constituted Pakistan in 1947
had few developed systems of local
government, and they too were confined
mainly to the Punjab. However, the little
local government that did exist was not based
on adult franchise and the agenda and budget
were under severe bureaucratic control, in
which the deputy commissioner played a
critical role in determining policy.The martial
law government of Ayub Khan (1958-69)
instituted an extensive system of elected
local government, known as the basic
democracies system, al! across the country.
With the fall of the Ayub regime, the basic
democracies system was also brought into
disrepute and disgrace. The election of the
first democratically elected government of
ZulfiqarAli Bhutto, following theseparation
of the then East Pakistan, resulted in the
formation of an altogether new system of
government in the country. However, the
proposed elections under the Peoples Local
Government Ordinance of 1975 were never
held and local government was in abeyance
for much of the time Pakistan had a
democratic government.
It was the return to martial law under
General Zia ul Haq (1977-88) when local
governments were revived under the Local
Government Ordinances of 1979 and 1980.
Elections for local councillors were held on
a non-party basis in 1979, 1983 and 1987.
However, once again, with the return of
parliamentary democracy in 1985, just as
was the case in 1971, the position of local
councillors and local government began to
be undermined. Essentially, what the history
of local government tells us’that in Pakistan,
military dictators seem to favour local govern
ment, while democratically elected govern
ments at the national level, for different
reasons, have in the past, felt threatened and
have preferred to do away with this tier
altogether.
The KMC, like al! other municipal
governments, performs a large number of
Econnmi"' and Political Weekly
functions of a compulsory and optional
nature. The KMC is supposed to be
responsible for a host of services, including
the planning, development and maintenance
of roads, for bridges, street lighting, storm
water drains, public health including
sanitation and solid waste management,
medical services, a fire fighting service, land
control, removal of encroachments, taking
care of libraries, museums and art galleries
and for social welfare. The establishment
and maintenance of educational institutions
is also meant to be a compulsory feature of
municipal government, and the KMC runs
a large number of schools and colleges in
the city. The establishment of hospitals and
dispensaries is an optional function; however,
the KMC runs a number of such institutions
throughout Karachi. The provision of clean
drinking water and sanitation and sewerage
facilities are one of the more important
responsibilities of municipal government and
in Karachi, as in other large metropolitan
cities, a separate institution known as the
Water and Sanitation Authority (WASA)
has been established (see below).
The Local Government Ordinance of 1979
for Sindh, was similar to that of the other
provinces but had specific mention of the
KMC. The KMC is specifically defined as
a corporation in the Ordinance, with its
common seal and is a full legal entity. In
an amendment to the Local Government
Ordinance in 1987, four zonal municipal
committees (ZMCs) were created as a
subdivision of KMC, where the ZMCs were
responsible for a number of basic services
at the zonal level. In .1996, through another
amendment in the Ordinance, the KMC wassplit up into five..district municipal
corporations (DMCs), where most of the
functions attributed to local municipal
government, were distributed between the
KMC and the DMCs. These DMCs along
with the KMC, are ‘local bodies’ in common
with local authorities throughout Pakistan
[Nauman 1996).
The family of public service agencies
providing services to the public in Karachi,
which have some element of representation
and democracy, such as the KMC and DMC,
is completed by the inclusion of the Karachi
Water and Sewerage Board (KWSB). The
Local Government Ordinance of 1979,
specifically defines the KWSB as a Board
within KMC and the Ordinance ensures that
KMC has powers over its actions. Hence, •
the KWSB is a subsidiary agency of the
representative KMC to the extent that the
“KWSB must be provided by KMC with any
funds required to meet its operations in the
event of short-falls in internally-generated
cash" [AREC 1993, vol 4, p 3],
Just as large municipal governments have
separate water and sanitation agencies, many
have a development authority working in the
December 20, ’097
city. The Karachi Development Authority
(KDA) is such an authority for the city of
Karachi which is primarily responsible for
the provision of shelter opportunities by
developing and selling land for residential
and commercial use. KDA also undertakes
major infrastructure projects in Karachi and
has, in the past, been the agency primarily
responsible for city planning [AERC 1993,
vol 4, p 1). “KDA is essentially a civic
agency responsible for developing
‘affordable’ land and shelter, planning for
the city and undertaking other peripheral
activities and ’deposit works’ on behalf of
the governments” [AERC 1993, vol 4, p 97],
A critical distinction between KMC and
KDA is that the latter is not answerable to
the people of Karachi and is not a
representative, elected body’. KDA is
controlled by a small governing body and
reports, and is answerable, to the ministry
of housing and town planning, government
of Sindh. The KDA “differs from other civic
agencies in both its mandate and its policy
as a provincial government agency, with its
own legislation, rather than as a local
government one” [AERC 1993, vol 4, p 97],
This distinction has important ramifications
discussed elsewhere in this paper, on how
development takes place in Karachi.
By legislation, KDA is the local body in
Karachi responsible for overall planning. To
execute this function it has formed the master
plan and environmental control department.
Apart from town planning, KDA is. also
active in the development of land and
allocation ofdevelopmentlandforresidential,
commercial, industrial and other uses. It also
oversees and enforces building controls in
the city [Hasan 1991], . •
Eighty per cent of KMC’s revenue accrues
from octroi, a feature common to all municipal
government. However, unlike most
municipal governments, a 1993 report on
KMC concluded that it has “maintained a
consistently strong financial position and
has made modest surpluses of revenue since
the mid-1980s. These have not been planned
to thelevel achieved, out havelargely resulted
from underspend due to slippage in
development projects” [AERC 1993, vol IA,
p 9], The majority of consolidated revenues
of the KMC, KWSB and KDA (between
1985-91 for which analysis has been
conducted) was, surprisingly, from own
resources or from bifurcated taxes, such as
the property tax, principally intended for
their use. These taxes and other sources of
revenue accounted for 93 per cent of own
funds in 1985 and only 2 per cent from
donors, which in 1991 had changed to 69
per cent from own resources with the donor
share increased to 21 per cent [AERC 1993,
vol 4 p 10).
External assistance, whether from donors
or from higher levels of government, has
■ ■
■: .......
''
3285
almost exclusively been used to fund capital
projects. However, because this amount is
not very large and forms a minority portion
oftotal fundingof these agencies, substantial
own funding is used for development projects.
Between 19S5-91, more than half of the
development fundingof these agencies came
from own resources. However, this trend has
changed somewhat in more recent years,
where donor funding forms a greater share
of development projects than in the past. The
high level of self-reliance in funding capital
works in the recent past, may have undergone
some change of late, and the World Bank
and the Asian Development Bank have now
become the main sources of donor funds
[AERC 1993, vol 4],
The KMC in the past has spent a large
portion of its funds on development works,
such as new roads, hospitals and flyovers.
Current expenditure is confined, principally,
to the maintenance ofroads, running hospitals
and the fire brigade. Although DMCs do not
generate revenue, they provide the first
contact of local government with the people
and are dependent for revenue from the
KMC, and much of their funds are spent on
rehabilitation, particularly of drains and
roads. While KWSB has been dependent on
the KMC in the past, it has been able to raise
revenue from the water rate to cover about
half of its expenditure. KWSB has made the
greatest use of external funding, but has
continued to require funds from the KMC.
An important difference between the KDA
and the KMC is that, KMC raises revenue
and then spends money accordingly, while
KDA is supposed to be a no-profit no-loss
agency which just breaks even. Since the
KDA is supposed to be in the business of
providing shelter, it develops land and then
is supposed to cross-subsidise the poor by
preparing land and plots for the lower income
groups. However, the way planning has taken
place in Karachi and the manner in which
the poor have actually suffered, is a theme
described at length in the following section.
Institutionalising Failure: The Planning
Process in Karachi’
A look at the planning process in Karachi
over the last 50 years, shows that, for the
most part, the formal institutions which are
meant to plan and deliver services to the
people of Karachi, particularly to the lower
income groups, have failed for a number of
reasons which seem to be repeated with each
attempt at planning. There are very basic and
fundamental mistakes which are made and
repeated in the planning process, and unless
these issues are identified and addressed, the
past trends of failure will continue to persist
well into the future.
In the section above, where we looked at
the growth and development of Karachi, we
have shown that the population of the city
328o
increased two-and-a-half fold in a mere four
years between 1947-51. The first challenge
the administration of the city faced was to
deal with that of providing immediate shelter
to the 6,00,000 refugees who had migrated
to Karachi following partition. The migrants
had to be housed, there was a need to develop
infrastructure for water and sewerage, and
there was a need to think beyond the
immediate problems, and plan and create
space for the future development of Karachi,
which also happened to be the capital of the
new country.
The Karachi Improvement Trust was
formed in 1950 for this purpose. The first
planning attempt by the government of
Pakistan was made in 1950, when the Swedish
firm Merz Rondall Vattan (MRV) was
commissioned todevelop the Greater Karachi
Plan. The plan was supposed to establish
growth corridors of the city, establish a new
capital territory towards the north of the city
and to provide housing in the form of multi
storeyed flats in different parts of the old and
new city. However, as Hasan has argued, the
housing development of the MRV Greater
Karachi Plan, "was on so small a scale that
it did not in any way affect the housing
situation in Karachi” [Hasan 1992:3]. For
the first decade after independence, from
1947-59, the government's response to the
housing crisis and its attempt at planning
"failed completely to tackle the housing crisis
that the city was faced with and in these 12
years the supply of houses lagged far behind
the demand” [Hasan 1992:3].
Although the MRV plan failed due to
some generic reasons which have been
repeated in all planning attempts and are
discussed towards the end of this section,
some very specific reasons regarding the
Greater Karachi Plan’s failure are worth
noting. For one thing, anti-government
student movements i'n the capital and
considerable political instability throughout
the 1950s, resulted in the MRV plan not
being implemented in the first place. The
administration questioned the philosophy of
the plan which had suggested that the
administrative nucleus of Karachi should
remain in the centre of the city where the
poor also lived and where the university, full
of agitating students, was located. More
critically though, there was an acute lack of
data in the early years and an adequate data
base was not established for the
implementation of the plan as the research
necessary for such an enterprise was not
carried out. Due to this seemingly simple
fact, that there was no research nor a data
base, the MRV plan was designed on the
assumption that Karachi’s population would
grow to three million in the year 2000, when
it actually reached this figure in 1969. The
population expected to reside in Karachi in
the year 2000 is almost five times that
projected in the MRV plan made in 1951
[Hasan 1992],
The next formal attempt at planning for
the city of Karachi was made in 1958 when
the Greater Karachi Resettlement Plan was
launched by the Doxiadis Associates of
Athens who had been commissioned by the
government of Pakistan to plan for develop
ment in Karachi. The Doxiadis Plan estab
lished a data base, conducted some research
about the city where it was expected to work,
and on this basis, made some far more realistic
projections than its predecessor.
The Plan objectives wereto house 1,20,000
homeless families living in the centre of
Karachi and it undertook to actually build
3,00,000 units for the poor over a 15-20 year
period. For this period, the plan projected
the need for half a million housing units. For
the remaining 2,00,000 families needing
housing, the government undertook to
develop plots with services, subsidising this
development by 30 per cent with recovery
in easy installments. Due to the political
disturbances in the city in the 1950s, the
government had decided to move the
squatters out of the city centre and resettle
them in two new townships which were to
be created 20-30 kms away from the centre.
The housing units were to be built around
industrial estates which were to be established
near the new residences of the workers and
the poor. The clearance of the inner city
slums and resettlement in other areas was
an integral part of the Greater Karachi
Resettlement Plan [Hasan 1997, 1992].
Only 10,000 units were built by 1964
when the plan, like its predecessor, was
shelved as it failed to achieve its objectives.
The reasons for this were: (i) jobs were not
generated as industrialisation did not take
place as perceived in the new areas created
for this purpose; the result was that “50 per
centofthe people who were moved to Korangi
and New Karachi [the two new settlements]
moved back to squat in the city centre, on
the fringes of the city, so as to be nearer their
places of work” [Hasan 1992:6]; (ii) the low
income and poor households for whom the
new housing units were meant, sold their
properties to speculators who sold them to
middle income households, a phenomenon
universally typical of any low income housing
project; and (iii) the scheme faced financial
problems as only 35 per cent of dues were
recovered even after 25 years of building the
units.
The abandoning of the Doxiadis Plan
halfway, had serious ramifications on the
way Karachi and its people were to develop,
and laid the foundations for developments
which continue even till today. Although
large areas of Karachi were cleared of
squatters and it became very difficult for the
poor to acquire land for building in the city
centre, the poor were forced to shift to the
Economic and Political Weekly
December 20, l’v17
fringes of the city and acquire land by illegal
subdivision. Moreover, the city was formally
divided physically, into rich and poor areas.
The Doxiadis Plan, in essence, gave birth
to the informal sector which now dominates
the shelter, and the social and economic
sectors in Karachi.
After thesccond attemptto plan for Karachi
also failed, and now that the capital of Pakistan
had been shifted to Islamabad, the
government decided to take a back-seat for
a while and actually decided not to construct
houses for the poor in the future. In 1967,
the government asked the United Nations for
assistance in dealing with the problems that
had emerged in Karachi (see next section)
and eventually, a Karachi Master Plan
Department was established in the KDA and
work on the design and development of the
Karachi Master Plan (KMP) (1974-85),
Karachi’s third plan, began.
The KMP estimated that in 1972 there
were more than 1.5 million low income
people in Karachi (42 per cent of Karachi's
population), out of which 8,00,000 lived in
squatter colonies. The projections showed
that by 1985 there would be an additional
5,90,000 new households in Karachi, out of
which 2,50,000 households would be from
the low income group. Now that the
emergence of katchi abadis had become an
integral part of the landscape of Karachi, the
KMP also had a component called theKalchi
Abadi Improvement and Regularisation
Programme, which was meant to upgrade
katchi abadis by providing urban services
to them and by regularising such settlements.
However, as Hasan shows, the katchi abadi
programme failed miserably and managed
to regularise about 18,000 out of an
approximate 2,23,000 houses after having
spent many hundred million rupees [Hasan
1992:16-17].
For the most part, the housing programme
under the KMP developed some plots and
the required infrastructure, but credit,
technical assistance and other social sector
facilities did not materialise. Moreover, the
cost of development and development
charges were inaffordable to the low income
households despite the subsidies that existed.
Yet again, housing meant for the poor was
taken over by the middle class. Less than
halfway into its term, the Karachi Master
Plan 1974-85 was abandoned. The fourth
large-scale planning attempt for Karachi by
the government of Pakistan is the Karachi
Development Plan (KDP) 2000. The
objectives of KDP 2000 are ambitious and
similar to those in the past, where the
government hopes to ‘improve the overall
conditions of human settlements in the
country, including the living conditions of
the poorest’. More specifically, the UNDPfunded plan is meant to ‘improve the
efficiency of the delivery of urban services
Econor,'f' and Po>'’i'”>l Weekly
and shelter to the inhabitants of Karachi’
and planning should be of a much more
‘realistic’ nature. The fate of KDP 2000 has
been similar to that of the plans in the past,
except that this last one has not even gotten
off the ground. Hasan has argued that
the plan did not take into consideration the
informal development lobbies that had
become important providers of services in
Karachi during the 1980s. These lobbies,
who had by then become important interest
groups were neither consulted nor did they
participate in the plan formulation.
Furthermore, it was assumed that the state
planning and development institutions had
the necessary organisational culture and
skills to implement the KDP 2000. In
addition, the steering committee of the plan,
of which the chairman was the chief minister
of Sindh, was unable to even convene to
approve the plan. As such, the plan has no
legal standing and many of its recom
mendations are being violated. The non
approval of the plan and the non-convening
of the steering committee, simply indicates
thatKarachi’splanninghasnotbeen apriority
with the four governments that have been
in power since the plan document was
completed in 1990 (Hasan 1997:11],
Today, Karachi has no plan document
which the city’s organisations follow. The
story of the master plans in the past has been
one of unmitigated disaster and failure. In
fact, the severe problems caused as a
consequenceof the failed plans have plagued
Karachi, adding to even greater problems.
While these issues are discussed later in this
paper, we now try to identify the reasons
why the planning process in Karachi has
been one of such grave failure.
Causes of Failure of Planning in Karachi
The way different government agencies
function in Karachi, with respect to planning
for the city in general, and for low income
groups in particular, represents the first reason
for the continuous failure of the planning
process in Karachi.
There is a ‘lack of integrated planning and
poor levels of co-ordination between the
agencies involved in the provision of civic
services’ [AERC 1993, vol 4, p 6] and one
of ‘the chief factors impinging on the success
of co-ordination between agencies was the
continuing inter-agency disputes on inter
agency financial dealings’ [AERC 1993,
vol 4, p 7]. The extent of the involvement
of different agencies and their roles can
best be understood by the following
description:
The Karachi Master Plan is prepared by the
Master Plan and Environmental Control
Department of the KDA. However, the KDA
only deals with the physical aspects of
housing, land development and its physical
implementation. The maintenance of KDA
development is carried out by the KMC
. which is also in charge of running markets,
certain aspects of public health and
December 70. 1997
education, recreation and parks. The water
'and sewage development and management
is done by the KWSB; energy is generated
and distributed by the Karachi Electricity
Supply Corporation (KESC) and the Sui
Southern Gas Company. In addition, the
Cantonment Boards in Karachi perform the
functions of the KDA and KMC in their own
areas and the Karachi Port Trust (KPT) is
in charge of the development, management,
operation and maintenance of the port area
and its related activities. Similarly, there are
a number of actors in the transport drama.
They include the police, the Karachi
Transport Corporation (KTC), the Bus
Owners Association and the Pakistan
Railways. There is no co-ordination between
all these agencies except on paper, although
their mandates are inter-dependent. Hence
master plans do not work [Hasan n/d-a].
Despite the fact that there have been four
master plans perceived and partially
implemented in Karachi, no studies for the
causes of the little success or failure of the
previous plans have been undertaken; there
is little understanding of how development
takes place in the city'in the first place, and
nor of the actors and lobbies involved in the
planning process. The planning agencies are
incapable of executing the plan in the first
place, as Noman Ahmed argues, that ‘all the
plans of Karachi have been made under the
auspices of KDA which does not possess any
legal or administrative control on the 19
other land development agencies of the city.
Thus the capacity of KDA to execute the
planisgrosslyconstrained’ [Ahmed 1996:3].
In a review of the KDP 2000, Hasan and
his colleagues argue that many of the planners
and consultants who are involved in master
plans have been trained as conventional
planners.and do not realise that in a third
world city like Karachi,
development often takes place before
planning and that integrating this
development into the city structure calls for
unconventional and innovative strategies
that, to be effective, should understand local
level social and economic conditions; the
informal sector is playing an increasingly
important economic and political role and
no city planning and management can be
successful without taking this factor into
consideration; urbanisation has caused a
major social and economic revolution and
the effective institutionalisation of this
revolution can take place only through the
involvementand accommodation of pressure
groups, communities, NGOs and the informal
sector; for institutions to be effective there
has to be a bond of trust between the various
actors (especially between government
institutions and low income groups) in the
urban development drama [Hasan et al
1991:8-9].
In any large city there are vested interests
and powerful lobbies of different types,
which include government, transporters
and formal and informal sector developers.
. 3287
Before the planning process begins, it' is
essential that detailed discussions are held
with representatives of various organisations
and interest groups including transporters.
industrialists, traders, professionals, shop
keepers’ associations, and representatives of
NGOs. One of the critical reasons for the
failure ofplanning in Karachi, andin Pakistan,
has been the fact that groups of vested interests
or concerned organisations are not consulted
sufficiently enough to have a bearing on the
planning process. Hence, due to the exclusion
of those whom the plan is supposed to affect,
the plan is predestined for failure. The
unrealistic assumptions made by plans rest
on the exclusion of, and lack of dialogue
with, those who are to be affected.
The failure of the KDP 2000 even taking
off the ground has been seen to be due to
technical shortcomings in the planning
process. According to one assessment
The planning process, which was always
organised in the most traditional pattern,
was faulty and inadequate. The basis of the
assumptions was drawn from sample surveys
in the absence of comprehensive views on
realities. This led to under/over estimations.
Physical data was obsolete and never up
graded. Karachi, even today, does not have
a comprehensive mapping base that is other
wise required for all kinds of planning and
development exercises. Data gathered by
the defence institutions is riot in the public
access. The property ownership records or
the alignment of jurisdictions are simply
inappropriate and obsolete [Ahmed 1996:4],
The planning experience from the past has
also shown that planning, and the housing
issues of the poor, are not just of a physical
nature, where technological matters or
logistics dominate. Planning and thedelivery
of basic services in urban areas has not taken
account of sociological factors which at times
may be of a more significant nature than
even technical issues.
Most importantly, however, an analysis of
the failure of planning in Karachi reveals
that planning is also a political matter and
ifthereisan absence for a political mandate
regarding a particular plan, the process is
doomed to failure. Due to a lack of political
participation and ownership, much of the
effort to plan is wasted, a feature which has
been observed with all the four plans that
have been implemented i n Karachi. Karachi’s
housing issues need to be related ‘to thec/ear
cut relationship between land, development
priorities, political power and finance
sources, that exist in Karachi. Without under
taking these relationships no workable solu
tion for the housing needs of the low income
groups, except for what exists today, is
possible’ [Hasan et al 1991:40; emphasis in
original]. While ‘traditional’ and ‘conven
tional’ political issues usually form a
cornerstone to the planning process, the more
recent phenomenon of donor dependence
3288
and international assistance is adding a new
dimension to political aspects of planning.
As far as the provision of low income
housing is concerned, attempts have failed
in the past to address the needs of this
community, as government policy has been
incompatible with the economics and
sociology of the poor. The cost of deve
lopment and/or lease has been inaffordable
for this target group and in most cases, low
income houses have been bought by members
of the middle class. The demand for housing
by the poor is immediate and they cannot
wait for the long drawn put development
process to be completed and possession to
be taken; often this process takes years. The
poor in Pakistan whether in rural areas or
in cities, have no representation in framing
policies. Most of the 'technocrats who give
physical shape to political thinking have also
been from the middle classes and have not
only a very poor understanding of the urban
poor, but look upon them with suspicion and
hostility. Thus, government policies have
invariably catered to the needs of the middle
and upper classes at the expense of the poor’
[Hasan 1992:17].
It is in response to this consistent failure
in formal/govemment attempts to provide
for housing and basic urban services, such
as land, credit, water, sanitation, transport,
employment, health and education, which
has given rise to the burgeoning informal
sector. In the next section, we examine the
issue of poverty, how poverty is created,
how it exists and how institutional failure
has been responsible for creating an entire
parallel economy and system.
So far in this paper, the discussion has
been onhow institutions inKarachi function,
and how the provision of urban services and
housing for the low income group and the
poor through these institutions, has failed.
The concept of poverty and who the urban
poor are, has been invoked only indirectly.
In this section of the paper, we try to
understand the concept of poverty and locate
the urban poor in Karachi. Only after
identifying them and locating who and where
they are, can any attempt to address their
specific issues be made.
Rather than enter a long epistemological
debate about what poverty is and how it can
be measured, this paper side-steps such
philosophical issues and tries to focus on a
set of relatively simple issues. Nevertheless,
some preliminary remarks on how one
identifies and measures poverty need to be
made. Usually, acompositemeasureorindex
is devised, which incorporates an arbitrarily
defined and numbered bundle of goods and
commodities, which are considered essential
to maintain an equally arbitrary, often
minimum, standard of life. The money
equivalent of the bundle of goods defined
by this criterion is then used to construct a
“poverty line". Those who are unable to buy
or access through other means this minimum
bundle of goods, are considered to be below
this line, are classified as “the poor”, and
are then the target of poverty alleviation
programmes.
This choice of ingredients which constitute
the bundle required to define a minimum
quality of lifeexistence, will vary from region
to region, and possibly from investigator to
investigator, making comparison difficult.
Moreover, the poverty line will need to be
constantly reassessed, both in terms of the
bundle of goods and in terms of its monetary
value, given changes that take place
increasingly at a faster rate in the economy.
Although used most frequently, the poverty
line does, often, not incorporate the cost of
human capital formation, the lack of which
in a household or region, despite higher
purchasing power, may indeed classify them
as poor. The debate over what really
constitutes “quality of life”, too; is endless,
and despite the existence of a number of
indicators which incorporate social sector
statistics, there is little agreement over “the
best” set of indicators. Nevertheless, despite
these (and many other) reservations which
are made about making comparisons, we
continue to do so. In the next two sections,
we try and assess the extent of poverty in
Karachi, and try to identify the urban poor.
As the discussion shows, this is not a
particularly easy task.
Poverty at the'macro level4,•
The World Bank’s Poverty Assessment
Report on Pakistan published in 1995 has ,
renewed interest in examining the level and
extent of the incidence of poverty in the
country. Much of the literature about the
extent of poverty in Pakistan has shown that
poverty had declined appreciably over the
1970s and the 1980s. The World Bank study
shows that there has been aconsistent decline
in consumption-based poverty from the
middle of the 1980s right up to thebeginning
of 1990-91. However, the lack of data
availability does not allow us to look at the
same sort of indicators after the 1990-91
period. Nevertheless, a vice president of the
World Bank, Shahid Javed Burki, who was
also .the advisor on finance to the prime
minister in the three month caretaker
government between November 1996 and
February 1997, has suggested on many
occasions that since 1990-91, the falling
trend in poverty in Pakistan may have been
reversed, with poverty making a formidable
return on the Pakistani scene. In a papc^ by
Asad Sayeed and Aisha Ghaus (1996) some
attempt is made to test Burki’s assertion.
Much of the research done on poverty
prior to the 1990-91 period shows impressive
results aboufthe downturn in its incidence
in Pakistan. A number of studies show ‘that
Fcor-mic and Poli-icnl WoeklV
December ?O 1997
decline in the incidence of consumption
based poverty in Pakistan has been consistent
and substantial throughout the decade of the
19S0s’ [Sayeed and Ghaus 1996: 1]. The
head count ratio of the poor in 1984-85,
which was as high as 46 per cent, declined
continuously to 37.4 per cent in 1987-88
and further still to 34 per cent in 1990-91.
Pakistan’s fall in poverty has taken place
at a time when it actually increased in
Latin America, the Middle East and North
Africa. However, the fall in poverty bet
ween 1984-91, was far greater in the faster
growing more dynamic countries like China,
Indonesia, Korea, Malaysia, Philippines and
Thailand [Sayeed and Ghaus 1996:2], And
as we will show below, herein lies the key
to the poverty question in Pakistan and in
Karachi.
Since no data of a comparable level for
a pre- and post-1991 period exist, the analysis
conducted by Sayeed and Ghaus (1996) uses
an indirect approach, where after identifying
the causes for the fall in incidence in the first
place, they look at these factors in the post199 1 period taking a macro perspective. The
reasons which are attributed to the fall in
poverty priorto the 1990s are thehigh overall
GDP growth rateand its sectoral distribution,
thedirect and indirect impacts of remittances
of workers in the Middle East, the role of
safety net's in protecting individuals and
households from falling below a particular
minimum level, and the role fiscal policy and
government expenditure has made to the
economy and to a large number of bene
ficiaries.
Sayeed and Ghaus (1996) argue that
‘growth in per capita income has a poverty
reducing effect because for any given
distribution of income, consumption of the
lowerdecilesof the population also increases.
With little change in the distribution of
income over the years and indeed very little
effort to alleviate poverty by the state, one
crucial determinant of reduction in the profile
of poverty in Pakistan has been a respectable
rate of growth in income (1996:3). Using
their analysis for the post-1990 period,
Sayeed and Ghaus show that there has been
a considerably slower rate of growth in most
employment providing labour-intensive
sectors, such as construction, transport and
communications, agriculture, wholesale and
retail trade and even the highly dynamic,
small-scale sector. Moreover, as the overall
and sectoral growth rates into the 1990s have
fallen compared to the past, there has also
been a decline in the growth of real wages
in the 1990s. Added to this is the faster
increase in prices and the rate of inflation,
particularly that of food prices in the 1990s,
which has reduced the purchasing power of
the low income population.
Remittances from the Middle East in the
1980s were amongst the most distinguishing
Economic and Political Weekly
and positive features of Pakistan’s economy.
Throughout the 1980s, they contributed an
average of 6.5 per cent of GDP and played
a very significant role in reducing poverty
and fostering economic growth and develop
ment. A significant feature of these remit
tances was that they were spread across wide
regions in Pakistan where many poorer
regions benefited from this process [ Addlcton
1992], Between 1991 -95, the share of remit
tances as a percentage of GDP was down
to 3.5 per cent [Sayeed and Ghaus 1996],
Food subsidies have provided a cushion
for the poor in the country as these
commodities have a high share in the
consumption of the poor. In the period 198091, the annual compound growth rate of
government subsidy on food items grew in
real terms by 6.4 percent. In the period 199195, there was a fall in real terms of as much
as 22.4 per cent in food subsidies meant for
the poor [Sayeed and Ghaus 1996:13]. The
poorareoften the main users and beneficiaries
of public social sector and development
initiatives and facilities. It is likely that they
make greater use of government facilities
than as a proportion of the well-to-do. Hence,
any slow down or curtailment in
government’s expenditure towards the social
sector and development projects will more
severely impact upon the poor. Development
expenditure in terms of the annual
development programme of the government
of Pakistan has fallen from 9.3 per cent in
1980-81, to around 3.5 per cent in the current
fiscal year 1996-97 [Zaidi 1995:120], Even
if government development expenditure is
not specifically targeted towards the poor,
ample evidence shows, not only from
Pakistan, but from other countries as well,
that such investmentcrowds in privatesector
investment and helps raise the overall growth
level. The current trend of cutting the fiscal
deficit worldwide has implied that as
governments cut their expenditure, overall
growth also suffers [Zaidi 1994a, 1995].
The work on poverty in Pakistan suggests
that poverty has returned to Pakistan in the
1990s, as the growth rate of the economy
has fallen, as have remittances, food subsidies
have been cut and as inflation has increased,
affecting the poorest the most. Government
curtailment of spending on development
expenditure only makes things worse. The
return of poverty to Pakistan in the 1990s
needs to be seen in the context of the impo
sition of the structural adjustment pro
grammes of the IMF and the World Bank
which were initiated in 1989 [Zaidi 1994a,
1995; Zaman 1995]. Research by Khan and
Aftab (1995) shows that “structural
adjustment conditionalities are proving to be
detrimental for the socio-economic well
being of the poor” (1995:18). While the
causality of the growing incidence of poverty
in Pakistan with the involvement of the IMF
December 20, 1997
and World Bank in Pakistan’s economy may
be debated, one thing is clear; research
suggests that the increase in poverty is closely
related to the poor performance of the
economy at an overall macro level. This fact
has important repercussions on issues
about Karachi discussed next, and more
importantly, on issues concerning poverty
reduction and poverty alleviation all over
Pakistan.
Poverty in Karachi
While the problems of identifying and
measuring poverty are numerous and varied
in a national context, they are compounded
in a specific, regional or city context,
especially when data is of poor quality, is
infrequently collected and seldom
desegregated. Hence, by drawing a poverty
line and doing a head count at the national
level based on household income and
expenditure surveys may be possible, the
task of incorporating other variables,
especially when looking at a particular city,
is further complicated. The comparison of
a prosperous city, like Karachi with a national
level poverty line may suggest that much of
Karachi is above that line, and hence, by that
singular definition, ’not poor’. On the other
hand, in a richer than average city like
Karachi, perhaps a more Karachi-specific
relative poverty line needs to be constructed
to assess theextentofrelativedifferentiation.
On the other hand, simply income, in the
context of acity, may be a very inappropriate
measure for capturing the level and extent
of poverty. Hence, the terms ‘poor’ and ‘low
income groups' in this paper in the context
of Karachi are used somewhat loosely and
are moreofadescriptivenaturc. Nevertheless,
even the somewhat less rigorous nature of
the use of these terms helps in addressing
the issue of poverty in Karachi.
One study has calculated Karachi’s per
capita income to be around two-and-a-half
times the national average at around USS
900 in May 1993 [AERC 1993. vol 1, p 3].
The study then states that only 15 per cent
of Karachi’s households live below an
arbi trary poverty line of Rs 3,000 per month.
Nevertheless, despite using this poverty line
asanindicatortoidentifytheextentofpoverty
in Karachi, the study continues that ‘there
is undoubtedly a significant amount of
poverty in the city. This isvisually manifested
in the mushroom growth of unservicedkarc/u
abadis, which now account for almost 40
per cent of the city's population’ [AREC
1993, vol 1, p 27; emphasis added]. Hence,
whileonly 15percentofKarachi'spopulation
is below a poverty line, the 50 per cent that
live in katchi abadis are also classified as
poor. Furthermore, the study argues that ’the
lack ofadequate water and sewerage facilities
is the major source of death and disease in
the city’ [AERC 1993, vol 2, p 29], implying
3289
♦
perhaps that there is yet another dimension
to identifying and measuring poverty.
While katchi abadis are used as a proxy for
urban poverty, so is the concept of the
informal, or the unregulated, private sector.
Often the contrast between planned and
unplanned urban settlements is used to
‘visually manifest’ the extent of poverty in
a city.
The unplanned areas in Karachi which
consist mainly of katchi abadis and squatter
settlements, have almost all been developed
by informal sector entrepreneurs through the
illegal subdivision and sale of state land.
Most of these settlements are in the peri
urban areas of Karachi, along natural drainage
channels that now carry Karachi’s sewerage ,
to the sea, along railway lines, and in areas
prone to flooding. Hasan has argued that
‘for any realistic planning exercise, it is
crucial to understand the role and functioning
of the informal sector’ [Hasan and Sadiq
1994:5],
Katchi abadis, when they are established,
are initially unserviced settlements, but over
a period of time, ‘through a process of
lobbying, assistance from informal
entrepreneurs and self-help, the residents
manage to build permanent houses and
acquire roads, transport, electricity, gas, water
and social sector facilities within 15 to 20
years of their creation. Sewerage remains the
abadis' major problem. As such, old katchi
,abadis are similar to planned low income
settlements’ [Hasan 1995:4], Today, with
half of Karachi’s population living in katchi
abadis which have a growth rate of 9 per
cent per annum compared to the overall
growth rate of the city at 4.8 per cent, it is
estimated that ‘if this trend continues, and
it seems it will, then by the year 2005 over
65 per cent of Karachi will be living in
unregulated and officially unplanned
settlements’ [Hasan 1995:4], Table 2
provides an indication of the extent of the
difference in characteristics between planned
and unplanned areas in Karachi.
The informal sector plays a key role in
housing and related sectors in Karachi.
Between 1970-85, it is estimated that the
informal sector ‘accounted for 33 per cent
of all residential land conversion and
development in the metropolitan area and
produced oyer 50 per cent of the housing
needs of the population’ [Hasan et al
1991:43]. Moreover, the informal sector
provides as much as 75 per cent of the total
jobs in the city, up from 48 per cent in 1974,
and today, is responsible for 60 per cent of
all housing needs, which range from
assistance for credit, construction material,
technical assistance and acquiring land in the
first place [see Hasan 1992, for how the
informal sector exists and functions]. The
reasons why the informal sector dominates
in the city of Karachi should be clear: Most
3290
poorer households have been unable to
Poverty of Politics
acquire either land for a house, credit or
technical advice from government agencies.
. If macro-economic policy is responsible
They have generally turned to the informal, for the creation, maintenance, or reduction
semi-legal or illegal land market to acquire in poverty, clearly, it is the nature of politics
a plot on which to build [Hasan 1990 b:77]. at the national level which is flrstresponsible
The observations which emerge based on for a particular set of economic policies.
the above analysis, suggest the following: Research from a numberofstudieshas looked
(i) The formal sector and the government at how economic policies are chosen and
planning machinery have failed to develop whatsort of political interests affect economic
and provide affordable and appropriate programmes [Sayeed 1995; Zaidi 1994a,
land, housing and/or credit for the lower 1994b, 1995, 1988, 1989, Zaman 1993,
income groups and the poor in Karachi;
1995]. In recent years, with the IMF and
instead the informal sector fulfills this World Bank replacingthe ministry of finance,
need.
as the agency responsible for economic
(ii) Due to poor planning at a city level, policy, questions of national sovereignty
poverty in terms of poor housing and and political control have also arisen [Zaidi
social services, has actually been created 1994a, 1994b; Zaman 1995]. Hence, clearly,
by government institutions.
the outcome of the interaction between
(iii) The macro-economy impacts heavily on
the incidence of poverty across Pakistan, different political entities and social classes
where higher GDP growth and other affects economic choices. The composition
related factors affect the livelihood and of the elite, its economic interests and the
livingstandards of the people. In Karachi, pressure enforced by all contending groups
because of its linkage with the rest of the and classes, determines the politics and
economy, especially in industry, economics of government. These choices, in
manufacturing and the exports sector, a turn, have an impact on the extent of poverty
poor performance of the overall economy in a country. Of course] factors beyond the
affects a larger number of people, thus control of nation-states and their govern
contributing to more unemployment pnd ments, also affect economic policy and hence
poverty.
poverty, but the primary responsibility must
(iv) Since the early 1990s, the IMF and World lie with national government and its choice
Bank structural adjustment programmes
of policy.
have had a major impact on the country’s
At the city level, as in the case of Karachi,
economy. Because of their austerity
drives, privatisation, cut back in subsidy, city level government and its institutions
reduction in development expenditure, must take the primary responsibility in
and a cut in the fiscal deficit have all dealing with issues of poverty. Of course,
helped to maintain, if not create, poverty. just as national government is also dependent
Clearly, these policies do not play apolicy on the international political and economic
climate, in the caseof local orcity government
alleviation role.
(v) Mapping poverty is difficult in a city like and institutions, there is an even greater
Karachi, where different sets of indicators organic link with institutions of the state at
are used to capture the extent of poverty. the national level. This is particularly pro
Bysayingthathalfofthecity ’spopulation nounced in the case of Pakistan where the
lives in katchi abadis, need not imply that role and development of local government
half the population is poor;katchi abadis, has often been sabotaged by provincial and
particularly in' Karachi, maintain a
national government [Zaidi 1996], While
standard of living well ahead of most
areas of the country and possibly even local government has suffered due to contra
better than that many households living dictions with higher tiers of the state all
in planned areas; die provision of services across Pakistan, in Karachi, its specific
in unplanned areas and katchi abadis, political circumstances have given politics
' due to the active role of the informal in the city a much different perspective [Zaidi
sector, may in fact be far better than that 1991,1992b]. It is noexaggeration to suggest
of the planned settlements.
that the people of Karachi have been ‘com
The key point that this section makes is pletely alienated from the political process
that poverty may in fact be generated by in the country and have no access to the
governmental policies, by following specific corridors of power at any level’ [Hasan nJ
programmes, and by not following others. d b). The most popular political party in the
At a local level, planning has been an city, the MQM, has been not merely hounded
institutionalised failure, which whilecertainly out politically, but has actively been perse
not reducing the level of poverty, may also cuted by all powers of the state. It is to the
have, in fact, created and maintained it. The politics, the nature of institutions in Karachi,
only response, hence, has been either the and how they impact upon Karachi, to which
inform al sector, ortheprescnceofcommunity we now turn [for a recent history of Karachi’s
groups. Both, failure, and the response that politics, see Zaidi 1991; Hasan n/d b].
The political problems of Karachi and
it generates, are, however, political issues to
how they are manifested at a local level, for
which we now turn.
Economic and Political Weekly
December 20, 1997
much of the 1990s, can be best summarised
as followed:
First, the people of Karachi are completely
alienated from the political process in the
country and have no access to the corridors
of power at any level. Many of their elected
leaders are in prison and the political party
they voted for has been, and is still being,
actively persecuted by the powers that be.
Second, the Karachi administration is corrupt
and ineffective, and at best helpless. The
civic agencies and their staff are subservient
to contractors and the land mafia, whose
activities they arc supposed to control and
regulate. Development agencies serve the
interest of corrupt politicians and their
cronies, land-grabbers and developers, at
the expense of the citizen. A battery of
overpaid national and international
consultants prepare grandiose plans for the
citys development that have no relation to
the ground situation. As a result, these plans
are either never implemented or are
abandoned midway. The law enforcing
agencies have been used for decades for
political victimisation and as such they are
accountable to no one anymore. Apart from
extorting money from helpless citizens and
holding entire ‘mohallas’ at ransom, they
not only give protection to every conceivable
criminal activity, but are actively involved
in it They incite fear and hatred and not
respect and protection. All the above
mentioned agencies at various levels are
often manned by staff that have been
appointed through ‘sifarish’ and are not
qualified for the jobs they hold. Since they
are ‘political’ appointees they cannot be
disciplined [Hasan n/d bj.
One of the problems highlighted in the
course of this assessment of institutions that
deal with urban and development issues.in
Karachi, is that the state sector does not have
much of a role to play in the planning process
and that ‘much of the city’s problems are
related to the absence of this role’ [Hasan
1997]. Moreover, it is believed that over the
last fifteen or so years, ‘the performance of
its [Karachi’s] development and civic
agencies and its formal economy have
declined considerably whereas in most Asian
mega-cities there have been substantial
improvements’ [Hasan 1997:14], Hence,
while institutions of the state may have
coercive and legislative power oyer the
citizens of Karachi, they do not play a
contributory role in their housing and basic
needs, and moreover, the little role that they
do play has been deteriorating, is insufficient,
and of poor quality. Aly Ercelawn and M
Nauman, examining the workings of the
KWSB argue that the
Steadily deteriorating performance of the
KWSB shares all the familiar odious
distinctions of municipal services in much
of the developing world. Our metropolis
suffers from an increasingly inadequate
delivery of both the volume and quality of
Economic and Political Weekly
promote authority. Resolving the pervasive
problems of the city calls instead for
managers downwardly responsive to
consumers [Ercelawn and Nauman 1996].
While the failure of institutions to address
the problems of housing and the provision
of basic services to the low income group
and the poor, can be explained by the way
planning takes placeand the way institutions
are structured, the political alienation of the
citizens of Karachi and of the poor.
specifically, has lead to the failure of
The nature of how institutions work and the institutions to deliver even basic services.
political disempowerment of the people of While politics is the key to the problem and
Karachi, is best reflected in the lack of to the solution of better city level
planning and co-ordinationbctweendifferent administration and government, individual
institutions functioning in the city. For and community level responses have tried
example, although the KMC is an elected to address immediate concerns for at least
body and elections are supposed to be held a section of the urban poor. Just as the private
every four years as part of the local bodies and informal sector has emerged in a big way
elections, the KMC, like all other elected as a response to governmental failure, so
institutions in the province of Sindh, was have a large number of community-based,
dismissed by the provincial government in non-governmental organisations or NGOs,
1992. Even if KMC functions democratically become active and have tried to improve
on the basis of adult franchise, the planning housing and living conditions in Karachi
and implementation of Karachi’s [Hasan 1990 b, 1992; Alam 1996;Khan 1994;
development is done by the Karachi OPP 1996],
Development Authority. Unlike the KMC,
Informal groups have developed at a
the KDA is a parallel agency run by community level to address some particular
technocrats and not answerable to the people. problems in the delivery of services or in
Consequently,
the provision of infrastructure. These
the people of Karachi, especially the urban ‘informal groups often develop as soon as
. poor, have no direct or indirect say in the anew settlement comes into being. Residents
manner in which their city develops. Nor. are faced with a lack of basic infrastructure
can they, through the process of electing and services which individual households
their councilors, express their concerns and cannot address by themselves. Some form
problems or through their councilors affect ofcommunity organisation is necessary either
the development process. In addition, the to address the deficiencies or to lobby
KMC council can be dismissed by the government for ..public provision’ [Hasan
Secretary of Local Bodies, who is a 1990b:80]. From the construction of open.
government civil servant, if the provincial
drains or the laying of underground sewers,
government feels, that the council has failed
to discharge its duties and obligations. The to street paving and collecting garbage,
functioning of the corporation is inspected groups have emerged to address specific
once a year by an officer appointed by the issues and have then disbanded. Some groups
I government. If, after such an inquiry, the have taken on longer-term projects and have
government feels that the corporation is not set up schools and health clinics. If the
capable of running a particular department organisation is registered with the social
or programme, then it can take over the welfare department; they may even receive
management of that department or some funds to continue their work, although,
programme itself. These laws make the much of their expenses are met through
councils very vulnerable to government donations. However, finances and a lack of
dictates and deprive the people of power and trained personnel are serious constraints
influenceoverplanning arid implementation which cause such groups to pack-up after
of urban policies [Hasan 1992:22].
showing a lot of enthusiasm but few results.
Moreover,
Apart from community-based groups,
As long as local government fails to be which are more focused towards an issue and
accountable to citizens through transparent prefer to work in one particular area as small
and participatory functioning, thecity cannot as a mohalla or neighbourhood, there are
mobilise the required resources and will larger NGOs which usually work in more
only squander whatever new human and
than one settlement and have a broader
physical assets it does acquire. Bureaucratic
administration by Commissioners can, at developmental outlook. Some welfarebest, only serve to contain rather than resolve oriented NGOs have also emerged in recent
fundamental inadequacies and inequities in years and have been successful in providing
municipal services. By administrative a wide range of services to the poor.
' Development-oriented NGOs are those
mandate and tradition, officially-appointed
bureaucrats can do little but protect and which have emerged in recent years .as a
water. Much of the city literally cats, drinks
and breathes poisons spread by deficient
sewerage collection and treatment. Sharply
discriminatory provision of basic water and
sewerage services exacerbates inequities and
fuels violent discord. Crippled by years of
political and administrative mismanagement,
incompetence and corruption in operations
and oversight, KWSB cannot but produce
negligible investible resources to address
gross deficiencies in the delivery of basic
services [Ercelawn and Nauman 1996],
December 20, 1997
3291
response to the failure, inability and
unwillingness of institutions of the state, to
address the issues of citizens, particularly,
the poor. Rather than simply carry outcapital
works as the state does, such NGOs involve
and encourage communities to take hold of
their own lives by organising and filling the
void created by the absence of the state.
“ ‘Participation’ is no longer seen as simply
involving poorer groups in implementation
but far more as formulating what should be
done, how it should be done and how li mited
resources should be used” [Hasan 1990b:83].
The Orangi Pilot Project is such an
organisation in Karachi which has had a far
reaching impact on the lives of the poor in
one large area of Karachi.
Despite the huge success of projects like
the Orangi Pilot Project and smaller groups
which have significantly altered the lives of
many thousands of the poor in Karachi,
NGOs are not the sole solution to address
the problems of the low income groups in
Karachi. NGOs cannot replace government
at a city level; they can play a useful role
either working with government or can then
be effective in areas where government has
failed to deliver services. Nor can government
be replaced by the private sector through an
indiscriminate process of privatising stateowned and state-run concerns [Ercclawn and
Nauman 1996], While it is clear that
government and its institutions have been
a failure in the context of Karachi, the knee
jerk response of privatisation or other options
is inappropriate. What is required is reform
of government itself, at the national,
provincial and local level.
While one can list a number of reforms
and changes needed in the structure of local
city level government, these well-meaning
suggestions will be merely that, unless there
is a political demand for such forms of
institutions, a political will to carry them
through, and a politicised public and their
representati ves to actually administer, control
and oversee such institutions. The desire to
overhaul the local bodies system in Karachi
will be lost on the ears of those who control
politics, unless political forces put pressure
on institutions of the state to accede to those
suggestions. Reform itself is a seriously
political matterdependentuponthedialectics
of different political forces. With these
political prerequisites, we address some of
the concerns in local government that are
critical for an improvement in the delivery
of urban basic services.
There is little denying the fact that there
is a need for a huge overhaul of the local
bodies system to make it representative,
responsible, viable and functionable. What
is required is a powerful, autonomous tier
of government in all municipalities, but
particularly in the larger metropolitan areas,
like Karachi. Unless local government is
3292
given the power to function as an effective
tier of governance, it is unlikely that it will
address any of the issue which have been
raised in this paper. A self-contained and
independent tier of government, not tied to
the strings of the provincial government, is
the first pre-requisite for better governance
at the local level.
Constitutional amendments which provide
safeguards to local bodies are essential if
local government is expected to be recognised
as a feasible option. If more than mere lip
service is paid to the potential of local
government, a change in perception must be
matched by deeds, and deeds in the form of
giving local government a formal role in the
Constitution. Constitutional cover to local
government must take place along with
greater decentralisation so that local
government can function autonomously.
Local government, especially city govern
ment the size of Karachi, must be remodeled
as a new administrative style is required for
it to be at all effective. It should have the
power to plan and implement development,
not merely look after projects planned by a
totally independent body in which local
elected government plays no role. If finances
are a constraint to the development pro
gramme of local government, then the KMC
should have the power to raise resources;
currently this power is dependent on the
well-wishes of the provincial government.
Thecurrcntdistribution of functions between
an unelected technocratic development
agency (the KDA), and an implementing
agency which is supposed to maintain project
and plans developed by KDA, the KMC,
would need to be done away with. KDA
would need to be subsumed under the KMC
(or an elected city government) so that
different agencies can work under one
umbrella, rather than at cross-purposes as is
the state at present. If city government is to
be an effective medium of governance, it
must also be party to the maintenance of law
and order in the city; currently, the numerous
agencies of the federal and provincial
government at work in a city like Karachi,
undermine all semblance of authority and
control, and precisely for this reason, power
over law and order is held by agencies which
are not city-specific.
While the suggestions for local bodies and
for reform of local government presented
above, are more of a generic nature, the
specific situation of Karachi warrantsspccific
additional measures. The political history of
Karachi over the last decade has been one
of the people of Karachi and their political
leaders and parties, against the state and its
coercive institutions. Karachi, for numerous
reasons, despite its paramount importance
in the economy of the country, has been
alienated from political power and is a
political outcast. The democratically elected
party of the city has been persecuted in times
when democracy, in some form or the other,
reigned supreme over the rest of the country.
Hence, all the problems that local bodies and
local government faced in general, were
compounded inthecaseof Karachi. Whatever
reform may be envisaged for local govern
ment in Pakistan in the future, whether of
a constitutional, financial or administrative
nature, for Karachi, political issues will have
to be at the forefront, and must be a pre
requisite for any attempt at improving
structure and performance.
Conclusions
Poverty ‘alleviation’ is the trendy and
fashionable slogan for the end of the 1990s.
A great deal of money has been, and will
continue to be, spent on identifying who the
poor are. Attempts will be made at under
standing poverty, and projects, many initiated
on the insistence of foreign donors offering
a lot of money, will commence with the goal
to eliminate or reduce poverty. Projects will
be defined with a specific focus on either
the urban or the rural poor, in some specific
location. This attempt, while well meaning,
will invariably be at a micro level with a
narrow focus, often ignoring the causes for
the existence of poverty in the first place.
Band-Aid social work of this variety will
certainly improve the living conditions of
a number of beneficiaries in the project area.
However, the limited success ofsuch schemes
may suggest, that perhaps, theissue of poverty
is somewhat larger in its existence and
mani festation, and what is required is a more,
global, macro, and most importantly, political
approach.
Poverty isprimarily apolitical issue,caused
and maintained by factors of a macro nature
and by institutions which function in a
specific, political, environment. This paper
has argued that politics comes prior to
poverty, as do institutions. The failure of
institutions to address issues of poverty and
development are seen here as essentially
political failures. At a macro-economic,
national, level the choice ofcconomic strategy
SMALL ENTERPRISES
Mean a lot in the modern world
Read ISED Publications
Contact;
The Publication Manager
Institute of Small Enterprises '
and Development
ISED House, Vennala
Cochin - 682 028
Tel: 347884/335171
Fax: 347884/345163
e-mail: ised @md2.vsnl.netin
Economic and Political Weekly
December 20, 1997
based on class alliances and political
* expediency (as well as international factors)
may have a severe bearing on the way people
live. Poor economic performance filters down
to increasing poverty for a large and growing
number of people.
Institutions controlled and run by techno
crats and bureaucrats at a national or local
(city) level, may come up with suggestions
and programmes which are infeasible because
they do not incorporate the desires, expecta
tions, or choices of the prospective bene
ficiaries. Many well conceived plans, while
technocratically well planned, fail precisely
because they are unpolitical in their outlook
and are essentially thrust upon an uninvolved
public. While elected government, especially
in Pakistan, has numerous faults and
weaknesses, with corruption being on top of
the list, the process of electioneering and
accountability, may at least involve a much
larger section of the citizenry. If the manner
in which local government is structured and
the way it works, is reconstructed in a way
which involves participation, consultation,
and dialogue, it is possible that a more
efficient system of representation at the local
, level may evolve.
As the role of international donors increases
_ in Pakistan, political decisions must be taken
regarding issues of national sovereignty and
local control over decisions. With inter
national agencies dictating the economic and
political agendas of the government at a
national level, political control over policy
has waned. This is also manifested at a
Karachi level, where the privatisation of
local government institutions 1 ike the KWS B
have provoked commentators to argue that,
it is a matter of deep concern that Islamabad
and Washington are foisting their privati
sation programme upon Karachi, without
the clear transparency and wide citizen
participation that would ensure consensus
on so vital a matter as the provision of
water and sewerage services [Ercelawn and
Nauman 1996].
Lookingback over the last decade, it would
be difficult to find a morepoliticised, violent,
ethnically divided, alienated, city than
Karachi. We conclude with the assertion,
that politics must come first in any attempt
to alleviate poverty, especially in Karachi.
Urban governance and urban institutions are
also of a political nature and the power or
effectiveness of such institutions may lie in
the balance between the different political
factors in the field. While politics in Pakistan
is itself wrought with numerous contra
dictions and does at time inspire disgust, a
process of representation, consultation,
accountability and openness, may at least
begin to address some of the problems that
exist in Karachi. Experience from many large
cities in underdeveloped countries shows
that powerful, effective, autonomous and
Lvonoir.ic and Political Weekly
- (1997): ‘Post Independence Development of
Karachi’ in Hamecda Khuhro (ed), Karachi:
Megacity of Our Times, Oxford University
Press, Karachi.
- (n/d a): ‘Karachi’s Master Plan’ (unpublished
manuscript).
Notes
(n/d b): ‘Uncomplicated Karachi’ (unpublished
manuscript).
[This paper was first presented at a workshop held
jointly by the Global Urban Research Initiative - (n/d c): ‘Squatter Settlements in Karachi’,
Mimar, 38.
of the University of Toronto, and the Centre of
Urban Studies, University of Dhaka, in Dhaka, Hasan, Arif and Asiya Sadiq (1994): *Mapping
City Inequality: A Case Study of Karachi’
Bangladesh, on May 16,1997. Since then, I have
(unpublished manuscript), October.
received extensive comments from Shahrukh
Rafi Khan and Haris Gazdar, to both of whom Hasan, Ari fet al (1991 f.Pakistan: Karachi Master
Plan 1986-2000 - Report of the Evaluation
I. owe many thanks.]
Mission. UNDP, July.
1 It is not possible to write about Karachi unless
an extensive and thorough reading and Khan, Akhtar Hamecd (1994): Orangi Pilot
Project Programs, OPP-RTI, Karachi.
understanding ofthe largeoeuvre of Arif Hasan
is made. Arif has documented Karachi’s history Khan, Shahrukh Rafi and Safiya Aftab (1996):
‘Structural Adjustment, Labour and the
for many years and is a scholar who has been
Poor in Pakistan’ (mimeo), Pakistan Insti
involved in numerous community initiatives
tute of Labour Education and Research,
in the city. Much of this paper draws upon his
' Karachi.
work and extensive use is made of his published
and unpublished material. I also acknowledge Nouman, Mohammad (1996): ‘How to Improve
Performance of Karachi Metropolitan
numerous ideas generated by him during the
Corporation’, paper presented al the seminar
course of this study.
on ‘Citizens’ Role in the Governance of
2 This section makes liberal use of Arif Hasan’s
Karachi’, November 29-30, 1996.
published and unpublished work, particularly,
Orangi Pilot Project (1996): Sixty-Seventh
Hasan 1997; Hasan 1994; Hasan 1993; Hasan
Quarterly Report,- OPP, Karachi.
1992; Hasan 1990; Hasan n/d (a); and Hasan
Sayeed, Asad (1995): ‘Political Alignments, the
n/d (b).
State and Industrial Policy in Pakistan’,
3 In addition to the references cited above, this
A Comparison of Performance in the 1960s
section also makes liberal use of Ahmed 1996.
and 1980s’ (unpublished D Phil disserta4 Part of this section summaries the arguments ■ tion), University of Cambridge.
in Sayeed and Ghaus 1996.
Sayeed, Asad and Aisha Ghaus (1996): ‘Has
Poverty Returned to Pakistan, Social Policy
References
and Development Centre, Karachi.
Addleton, Jonathan (1992): Undermining the Zaidi, S Akbar (1988): ‘How the Bourgeoisie
Centre: The Gulf Migration and Pakistan,
Views Pakistan’, Economic and Political
Oxford University Press, Karachi.
Weekly, Vol 23, No 48.
Ahmed, Noman (1996): ‘What Ails Our - (1989): ‘Regional Imbalance and the National
Planning Process?’, Shehri, Vol 1, No 18,
Question in Pakistan: Some Indications’,
Karachi.
'
Economic and Political Weekly, Vol 24, No 6.
Alam, M Shahid (1996): ‘Community - (1991): ‘Sihdhi vs Mohajin Contradiction,
Organisations and Urban Development: A
Conflict, Compromise’, Economic and
Study of Selected Urban Settlements in
Political Weekly, Vol 26, No 20.
Karachi’ (mimeo), NGO Resource Centre, - (1992a): Regional Imbalances and the National
Karachi.
Question in Pakistan, Vanguard,' Lahore.
Applied Economic Research Centre (1993): - (1992b): ‘The Roots of the Crisis, Herald,
. Metropolitan Resource Generation Study (in
Karachi, August.
seven volumes), Karachi.
-(1994a): ‘The Structural Adjustment Programme
Bailie, Alexander (1975): Kurrachee: Past,
and Pakistan: External Influence or Internal
Present and Future, Oxford Univerity Press,
Acquiescence?’, Pakistan Journal ofApplied
Karachi.
Economics,, Vol 10, Nos 1 and 2.
Ercelawn, Aly and M Nauman (1996): - (1994b): ‘Planning in the Health Sector. For
‘Restructuring Water and Sewerage Services
Whom, by Whom?’,. Social Science and
in Karachi: Citizen Consent or Coercion’,
Medicine, Vol 39, No 9.
paper presented at the seminar on Citizens’ - (1995): ‘Locating the Budget Deficit in Con
Role in the Governance. of Karachi,
text: The Case of Pakistan’, Pakistan
November 29-30, 1996.
Journal ofApplied Economics, Vol 11, Nos 1
Hasan, Arif (1990): ‘A Decade of Urban Decay’,
and 2.
Herald, Karachi, January.
- (1996): ‘Urban Local Government in Pakistan,
- (1990): ‘Community Groups and NonExpecting Too Much from Too Little’,
■. Government Organisations in the Urban Field
Economic and Political Weekly, Vol 31,
in Pakistan, Environment and Urbanization,
No 44.
Zaman,
Arshad (1993): ‘Sustainable Develop
Vol 2, No 1, April.
ment, Poverty and Policy Adjustments:
- (1992): Seven Reports on Housing, OPP-RTI,
Linkages and Levels for Change’ (mimeo),
Karachi. .
International Institute for Sustainable
- (1993): ‘The Changing Face of Karachi, Herald,
Development, Canada.
Karachi, February.
- (1994): ‘Profile of Three Pakistani Cities’ -'(1995): ‘The Government’s Present Agreement
with the IMF: Mis-Govemment or Jolly?
(unpublished manuscript), April.
Pakistan Journal of Applied Economics,
- (1995): ‘The City Behind the Scene’, the Way
Vol 11, Nos l and 2.
Ahead, Vol 2, No 4, Karachi, December.
representative government may be a partial
answer to the problems of poverty and lack
of development. Perhaps it is time to test this
model in Karachi.
De.ember 20, 1997
______________________________________ SPECIAL ARTICLES
c actual rcpoijj
ict. the copi^
’ help seckir*
'he study hig^
disadvantage
then question
a fairly similj.
rlher details of
restore, micro.
nature of help
ries would have
the book. The
ipen-cndcd and
Jed to loss of
' low frequency
. though cons).
■I preoccupation.
;ral myths For
of
scent
inothi^^eslcni
cm youth have
ve coping' and
iassive coping'
I disadvantage,
idy.
l chapter which
nal factors such
.ind, nationality,
r. The chapter
-r a multinational
adolescents, in
approach. It is
•aches (i) should
meet the needs
uki have special
or the poor and
hey exist, and
entive approach
stem theory ap-
Better Resource Management for
Poverty Alleviation
J
,
e a mu^^tional
centi^^vhich
of helping pro- ;
sciplinary work
h professionals,
icrs. In sum, the ,
ial counsellor of
ch out to adole social activist
»m with flexible
fulfil the needs
to 1993 are
ai 400 001.
June 8,
B B Vohra
Although there are many reasons for our poverty - over-population, illiteracy, disease and social backwardness.
to name only a few - one of the most important is the sub-optimal use that is being made today of our natural resources
of land and water, which constitute between them the country’s basic production apparatus and life support systems.
Situated as we are. there can be no question of our ever being able to overcome poverty unless we can make very
quick and substantial progress on this front.
An attempt has been made in this paper to outline the present disarray in the field of resource management, to
view it in its historical setting and to suggest how it may be possibly remedied before the point of no return is reached.
I
Current Scenario in Land
Management
from degradation is 95.5 mh (175 mh minus
79.5 mh).
(d) It also follows that this area of 95 5 mh
must necessarily be a part of the 142.2 mh
ACCORDING to the latest Land Use of lands that arc under agriculture. This
Statistics (LUS), the 304.9 million hectares means that nearly two-thirds of our agricul
(mh) of the country’s geographical area for tural lands are sick to some extent or another.
which information is available arc being (c) The above picture would change
somewhat if the figure of 175 mh is found
used as follows:
(mh) to include item no 2. However the broad
picture would still be that nearly two-thirds
1 Area under non-agricultura! uses
21.2
of our total land resources arc suffering from
2
Barren and uncultivable lands
19.7
3
Net area sown
142.2 degradation, of which about 50 percent have
undergone such degradation (hat they have,
4
Forest lands under good tree cover
for all purposes, ceased to be productive.
(40 per cent density and above)
38.612
5
Miscellaneous tree crops and groves 3.7
There has been a great deal of confusion
6 Forest lands under poor tree cover 29.3-’ regarding the extent of wastelands in the
7
Cultivable wastelands
15.0 country. This has been caused by, firstly, the
8 Current fallows
13 8 fact that different authorities have adopted
9
Old fallows
96
different definitions of ‘wastelands’ and,
10 Permanent pastures and grazing
secondly, by their failure to distinguish
grounds
I 1.8
between lands that arc so badly degraded
Total
304.9
that they have gone out of production and
I As per the latest statistics of the Forest
those that arc still in production although
Department.
they are suffering from degradation to some
2 Arrived at by deducting the area under good
extent or the other. It would obviously not
forests (SI No 4 above) from the total area
be appropriate to describe the latter category
under forests, i e, 67.9 mh.
of lands as ‘wastelands’ - ‘degraded lands’
An analysis of the above figures reveals would be a more scientific appellation for
that
them.
Altogether too much fuss has been made
(2) If items 1 and 2 are excluded from
consideration the total land resources of the in recent years over the determination of the
country that possess any potential for biotic exact extent of wastelands and of their exact
production are no more than 264.0 mh.
location. Since the country’s readiness to
<b) Assuming that items 6 and 10 arc more deal with this problem is still at a rudimentary
Or less bereft of vegetal cover- which would level, common sense demands that we should
be a fair assumption to make-the maximum not get lost in such essentially peripheral
mailers but start work on the amelioration
area that can be considered as ‘wastelands’
,s the sum of items 6 to 10. This comes to of sick lands in right earnest on the basis
^•5 mh, which is almost one-third of264 mh. of the knowledge that is already available
fc) This however does not mean that the in ample measure in every affected village.
It may be noticed that the land suffers
rcmaining area of 184.5 mh (264 mh minus
^•5 mh) is in good health. According to the basically from only two major ailments LtS the total extent of lands that suffer from denudation and erosion, which results in the
^gradation, to a greater or less degree, is loss of the top-soil through the action of
^5 mh. Since this figure obviously includes water and wind, and waterlogging which
Clelands, it follows that the area of lands results in the salinisalion of the soil.
that are still productive but are suffering According to available estimates, of the 175
Economic and Political Weekly
June 8, 1996
mh of degraded and wastelands that the
country possesses, around 150 mh suffer
from erosion of the top-soil and its attendant
ills (such as floods and gully formation, etc)
and around 25 mh from waterlogging and
salinisalion. Of the lands subject to erosion,
around 125 mh suffer from water erosion
and around 25 mh from wind erosion.
There is no doubt that lands subject to
erosion constitute the biggest single threat
to the country’s economy. For not only do
such lands suffer an increasing loss of
productivity because of the progressive loss
of the fertile top soil but they also contribute
to the loss of a great deal of priceless sweet
waler by way of excessive run-off along
denuded slopes. This run-off, loaded as it
is with soil, also causes a great deal of
damage - by contributing to the causation
of floods and the premature siltation of river
beds, tanks and reservoirs - before it reaches
the sea. And since a large part of the water
could, under better conditions of land
management, have been retained either as
soil moisture (so vital forrainfed agricultural
lands) or as ground water (which is the
mainstay of the country’s irrigation
infrastructure today) its loss is a major reason
for droughts. Floods and droughts arc indeed
two sides of the same coin of poor land
management and both can be moderated
very substantially by preventing excessive
soil erosion.
Our record in tackling problems of
denudation and soil erosion has unfortunately
been most unsatisfactory and wasteful of
money. Although soil conservation schemes
for agricultural lands have been on the scene
for nearly half a century, they have failed
to make any significant dent on the problem
on account of their having disregarded the
‘complete mini-watershed’ principle. As a
result, excessive run-off from denuded forest
lands that arc almost invariably situated in
the higher reaches of watersheds has caused
great damage to the terraces and bunds on
agricultural fields, particularly because such
bunds are not correctly aligned alongcontour
1397
lines but built along field boundaries. This
basic flaw is common to all the other schemes
- such as the DPAP, DDP, RVP, FRRP,
NWPRA, JRY (in part) EAS and EGS (in
part) and the IWDP - which aim essentially
at soil and water conservation and explains
why they too have not succeeded in their
objectives.
The financial loss that the country has
suffered as a result of this circumstance has
never been computed but it must be very
considerable, considering that some Rs 2,000
crore have been spent on the DPAP and DDP
alone since their inception. This is indeed
a matter for great concern.
Our record in tackling problems of
waterlogging and salinisation is equally poor.
We have just not taken any notice of these
problems. As the Eighth Plan document
admits, there has so far not been even a
systematic survey of the extent and location
of lands that are affected by this malady.
P
Current Scenario in Water
Management
Since the soil, however well-endowed it
may be, is incapable of any biotic production
in the absence of moisture, the management
of water lies at the very heart of land
management.
Although the total precipitation received
by the country as a whole is around 350
million hectare metres (mhm) per annum.
which is theoretically capable of placing its
entire land surface under around 115 cm of
water, this resource must be treated as a
scarce resource because of, firstly, its highly
uneven spread in space as well as in time
and, secondly, the steadily increasing
demands that are being made on it not only
by agriculture but also the industrial and
domestic sectors.
As in the case of all scarce resources, the
^magement of waler demands that special
Mention should be paid, firstly, to its
conservation to the maximum possible extent
and. secondly, in the present context, to its
optimal use for agricultural production. A
third requirement is that in no circumstances
whatever should this annually renewable
resource be allowed to damage the non
renewable resource of the soil which it is
meant to serve.
An overview of the water management
scenario in India reveals that our traditional
policies have proved to be seriously deficient
on all these three counts and, therefore, need
to be reviewed urgently.
As far as conservation is concerned, the
traditional policy has been to rely on the
creation of surface storages, whether big,
mediumorsmall.lt is estimated that between
1950 and 1995 we have spent over Rs 50,000
crore on such projects and created a storage
capacity of around 20 million hectare metres
(mhm). An idea of the scale of investment
1398
in this field can be obtained from the fact
that during the Eighth Plan we shall be
spending around Rs 27,500 crore on this
sector, or about Rs 5,500 crore per annum.
However, we have clearly come to a dead
end on this route as investments in surface
projects have latterly shown unmistakable
signs of becoming unproductive. Thus, an
investment of Rs 11,107 crore on major and
medium (M and M) projects during the
Seventh Plan has, according to the ministry
of water resources’ (MWR)own figures, not
only not resulted in the creation of any
additional potential but has resulted in the
loss of 0.6 mh of even the potential that
existed at the end of the Sixth Plan. Consider
ing that the Seventh Plan had envisaged the
creation of an additional potential of 4.3 mh,
the net loss of planned potential that took
place between 1985 and 1990 amounted to
4.9 mh, the replacement value of which al
current prices would be well in excess of
Rs 30,000crore. This is an extremely serious
development indeed, all the more so because
no explanation has so far been offered by
the MWR as to how it took place.
The track record of small surface projects
has, if anything, been even worse. According
to the LUS, the net area served by such
projects declined from 8.2 mh in 1961 to
6.8 mh in 1989 in spite of the fact that around
Rs 6,000 crore were invested in them during
the intervening 28 years. However, in this
.case the reason behind the debacle is known.
It lies in the premature siltation of reservoirs
which, being much smaller than those of the
M and M sector, go out of operation so much
more quickly.
In view of these developments, the time
has obviously come to reconsider our
traditional approach to the problem of
conserving water and to turn to the only
other option available to us, namely, the
storage of water mainly in the form of soil
moisture and ground water - within, rather
than on the surface of, the country’s land
mass. Not only is this option incomparably
cheaper but it also involves no high
technology and has proved to be a great
success wherever it has been adopted. It
consists basically in reducing the run-off of
water to the sea by creating biotic as well
as engineering impediments to the free flow
of water along slopes so that it may get a
better chance to percolate into the soil and
sub-soil strata. The restoration of permanent
vegetal cover, whether of grasses or trees,
on all denuded lands - as far as possible
through natural regeneration - the
construction of innumerable small weirs,
check-dams and small tanks across all
drainage lines in all micro-catchments and
the treatment of all erosion-prone agricultural
lands for the conservation of both soil and
water constitute the key elements of this
alternative strategy.
Common sensedemands that a nationwide
programme for soil and water conservation
should be given the highest possible pri0 •
for a variety of reasons. It would, in ihc p y
place, help to reduce run-off losses a M
increase the availability of water in the (q
of soil moisture and ground water or Wat
stored in countless small tanks and pOn(j
The enhancement of soil moisture would L
particularly beneficial for our rainf a
agricultural lands which account for near!
two-thirdsof the total land underculti vatio
The storage of waler in village tanks and
ponds would not only be of great benefit to
local communities, but would also help jn
the replenishment of ground water, h nLjy
be noticed here that ground waler not only
accounts for more than 50 per cent of the
total area under irrigation, but is also around
100 per cent more efficient than canal water
in terms of productivity per hectare. This is
indeed the reason why it is in such great
demand on the part of farmers who can
develop it quickly and easily with their own
resources, assisted by bank loans, wherever
necessary.
In the second place, such a programme
would help in controlling the premature
siltation of reservoirs and tanks - which in
most cases are irreplaceable - and in
moderating floods by reducing the quantity
of water and top-soil that rivers have to cany
at peak periods, and by protecting their
carrying capacity against siltation. In the
third place, the return flow of waler that
lakes place from fully charged ground water
aquifers into springs and rivers during the
lean season would also help to mitigate
droughts.
The wisdom of conserving water mainly
within the land mass and in nature's own
way rather than in man-made reservoirs
becomes apparent when we consider the
great difficulties that the MWR has
encountered in putting to actual use the
irrigation potential created by surface
storages. Thus, in the M and M sector, the
total potential created between 1950 and
1990 was 20.2 mh of which, according to
the ministry’s own claim, only 15 8 mh had
been utilised by 1990, thus revealing .aa
unutilised gap of4.4 mh. However, according
to the LUS, which command much higher
credibility, the unutilised gap in 1990 ***
as big as 9 mh.
The above analysis shows that our
traditional policies have failed significant X
from the point of view' ot both
conservation of water and putting itt0 8
use. The continued failure on both
counts is reflected by the unbelievably
level that the cost of irrigation hasre3C
- as distinguished from the cost of
creating potential which is of no
anybody till it has been actually Put l°-^
Thus, according to the LUS, the net ad<JJ _
net area that was brought under
in the M and M sector during the >
Plan was only 0.262 mh. If the tota
of Rs 11,107 crore is divided by this 1
Economic and Political Weekly
cost of act
j irrigation throug
« j985-9O works<
< j^h per hectare
The very gr
: fpfiserving wat
’
is indicate
,
(or nearly or
jyffacc) suffer f:
if treated appro
^-off losses vi»
^ure today is
precipitation that
160 mhm are lo
.. around 20 mhm
'■ around 125mhrr
; 45 mhm as gro
■ compiled by th
■ experiments cai
; tnicro-watershe(
,? suggests that a
] afforestation anc
’ may weH be abb
j losses by 25 pc
; increase the qu>
moisture and gi
Needless to say,
change the face
We have al sc
count and perm
land in two way
to our failure t
country’s total la
2nd erosion, we f
ofthetopsoil toi
! a a result of the
back in 1972, t
eroded was csti
tonnes per annu
at least double
appreciated that
in the progressi
affected by redi
because the top;
Wil strata - but;
blc damage in d
Roods, to whic
vtdncrable.
In the second
almost complete
j^oage, water
.so^’ aHowe
fertile lands t«
fording to th
^earea affected
14 mh in
registering
Assumi
Prevailed du
greeted in 199.
part of
in cana
£"^5 matter in
’'control of ere
U* average c
J^.theame
^'andsreq
border of:
com of actually bringing land under
legation through the M and M route during
19^5-90 works out to an incredible Rs 4.24
jjkh pef hectare.
The very great scope that exists for
conserving water within the Indian land
^5 is indicated by the fact that around 150
jph (or nearly one-half of the country’s land
surface) suffer from soil erosion and would,
if treated appropriately, be able to reduce
run-off losses very substantially. The broad
picture today is that, of the 350 mhm of
precipitation that we receive annually, around
160 mhm arc lost to the sea as river flows,
around 20 mhm are stored as surface water,
around 125 mhm as soil moisture and around
45 mhm as ground water. Empirical data
compiled by the ICAR - on the basis of
experiments earned out over 20 years in
micro-watersheds in all parts of India suggests that a nationwide programme of
afforestation and soil and water conservation
may well be able to reduce present run-off
losses by 25 per cent or say 40 mhm and
increase the quantity of water held as soil
moisture and ground water to that extent.
Needless to say, such a development would
change the face of the country.
We have also failed badly on the third
count and permitted water to damage the
land in two ways. In the first place, thanks
to our failure to save a large part of the
country’s total land surface from denudation
and erosion, we have allowed large quantities
of the top soil to be displaced, year after year,
as a result of the action of rain water. Way
back in 1972, the quantity of top soil so
eroded was estimated to be 6,000 million
tonnes per annum - today the loss must be
at least double this amount. It must be
appreciated that such losses not only result
in the progressive degradation of the lands
affected by reducing their fertility levels because the topsoil is the most fertile of all
soil strata - but also contribute to considera
ble damage in downstream areas by way of
floods, to which around 40 mh are still
vulnerable.
In the second place, we have, through our
almost complete neglect of the problems of
drainage, waterlogging and salinisation of
fhc soil, allowed large quantities of once
fertile lands to be lost to production.
According to the latest available estimates,
the area affected by these maladies increased
from 14 mh in 1981 to 17.6 mh in 1985,
thus registering a growth rate of 0.9 mh per
armum. Assuming that the same growth rate
has prevailed during the last decade, the area
affected in 1995 should be around 27 mh,
a large part of which is almost certainly
situated in canal commands. This is a very
serious matter indeed, considering that unlike
decontrol of erosion, which can be achieved
ai an average cost of around Rs 4,000 per
hectare, the amelioration of waterlogged and
saline lands requires much larger outlays of ’he order of Rs 30,000 to Rs 40.000 per
Economic and Political Weekly
hectare-as drainage, especially underground restored to health, of producing additional
drainage, is an expensive proposition.
wealth - whether in the shape of crops or
In view of the above analysis, a thorough fruits or timber or fuel or even mere grasses
revamping of existing policies in water - worth on an average around Rs 10,000 per
management has become unavoidable. Such hectare, the loss that is being sustained by
a review must take particular note of the thccountry works out to around Rs 1,75,000
following considerations:
crore per annum. However, even this figure
(a) Further investment in the creation of will prove to be an underestimate if we also
fresh potential in the surface irrigation sector take into account the tremendous damage
must be suspended and all avai lable resources that is being caused to irrigation systems by
should be diverted towards the utilisation of the premature siltation of reservoirs and
the very substantial potential that has been tanks and by the damage caused by floods
and droughts.
already created but not used.
(b) The working of existing canal systems It may be mentioned in this connection
must be improved so as to increase their that while very large numbers of small
productivity in terms of yields per hectare. reservoirshave already gone out of operation
Once this is done, irrigation rates, which arc on account of premature siltation, many of
at present inordinately low, can be raised so our bigger ones arc also getting silted up at
as to prevent the loss of some Rs 3,000 crore rales that are 4 to 16 times higher than those
per annum on maintenance and operational assumed at the stage of project formulation
(it is a different matter that such siltation
costs alone.
(c) Special attention must be paid to the pro takes time to show up). As far as floods are
blem of waterlogging and salinisation which concerned, the Eighth Plan document says
that 40 mh of our land surface are floodhas received almost no attention so far.
(d) Existing flood protection policies-which prone and that the area affected annually is
aim at dealing with the symptoms of the on an average about 7.7 mh. On an average,
disease rather than its real cause, viz, poor over 1,400 lives are lost every year and the
land management - must be revised in favour damage caused to crops, homes, cattle
of policies aimed at the prevention rather and public utilities between 1953 and 1987
than the control of floods, through dykes was nearly Rs 27,000 crore. Information
regarding expenditure incurred on meeting
and bunds.
(c) The predominant position attained by drinking water requirements during droughts
ground water in the field of irrigation must is not readily available but is known to be
be recognised and the management of this substantial.
Another way of appreciating the economic
resource strengthened. The replenishment
of this resource must be assisted through consequences of poor resource management
natural means such as better land would be to put a price tag on both sweet
management as well as through artificial water and the topsoil, and to stop looking
recharge, and it must not be treated as a at these resources as if they were free and
inexhaustible gifts of nature. As far as water
source of only ‘minor irrigation’.
The point to note is that, as in the field is concerned, it is known that it has cost us
of land management, so also in the field of around Rs 50,000 crore to create a storage
water management, our existing policies are capacity of around 20 mhm. What this means
highly unsatisfactory and arc the cause of is that if a systematic and effective nationwide
very substantial losses to the government programme for the conservation of soil - and
and damage to the economy. It is also clear therefore also water - succeeds in reducing
that the key to better water management lies run-off losses by 25 per cent or by 40 mhm,
essentially in improved land management, we shall have obtained an advantage, in
through the control of denudation and soil money terms, of something like Rs 1,00,000
erosion across the length and breadth of the crore per annum.
As far as the fertile topsoil is concerned,
country.
there is really no way of pricing it, because
it is not merely a collection of chemicals and
HI
Nexus between Poverty and Poor
plant nutrients but an almost living medium
that teems with micro-organisms whose
Resource Management
variety and complexity continues to baffle
No attempt has ever been made - by an scientists. And since it takes nature hundreds
establishment that the late Sudhir Sen. of years to build an inch of the topsoil, it
eminent economist and author, used to is for all purposes a non-renewable resource.
describe as 'resource illiterate’ - to quantify However, even if we assume for it a price
in monetary terms the losses that the economy of only Rs 100 per tonne and also assume
is suffering as a result of poor resource r which is very likely - that the current rate
management. However, there can be little (^displacement of the topsoil due to erosion
doubt that these are of the most serious is around 10,000 million tonnes (mt) per
annum (up from around 6,000 ml in 1972),
proportions.
Even if it is assumed, on a most the annual loss suffered by the economy on
conservative basis, that our 175-odd mh of this account would be around Rs 1.00,000
degraded lands arc intrinsically capable, if crore'
June 8, 1996
1399
Be that as it may. it is quite clear that the
country is paying extremely heavily for its
failure to manage its natural resources
properly and that this is one of the prime
reasons for its poverty. The enormous but
entirely avoidable losses that arc taking place
as a result of deforestation, soil erosion,
excessive run-off and the continued neglect
of problems of waterlogging and salinisation
can indeed be likened to an uncontrolled
haemorrhage in a patient who is already
severely debilitated.
It may be also mentioned in this connection
that, in a predominantly agricultural country
like ours, it is not correct to make too fine
a distinction between rural poverty and urban
poverty. For the latter is merely an offshoot
of the former and is to a large extent due
to the exodus of the rural poor into urban
areas in search of employment. Urban poverty
will gel automatically reduced once this
influx first ceases and then goes into reverse
gearaseconomic conditions begin to improve
in rural areas.
^^e need of the hour is to recognise that
N^Pe are at all serious about combating
poverty - poverty that has become a national
disgrace fora country that can rightfully take
pride in its achievements in so many other
fields - there is no better strategy than to
make a frontal and determined attack on all
aspects of poor resource management. The
surplus labour available in rural areas must
be harnessed and systematically converted
into permanent productive assets through
extensive soil and water conservation and
drainage works, the lining of canals and
water channels, desiking operations in small
tanks, canals and river beds, the raising of
plantations of various kinds (including
horticulture) and other allied activities.
Better resource management must indeed
form the bedrock of all our plans for rural
development and employment and replace
the confusing medley of wasteful and unco
ordinated schemes that operate in the field
today under the aegis of many different
c^iments which work in watertight
cW^artments. so to say. And if infructuous
expenditure and disappointments are to be
avoided, it must be undertaken in a holistic
and not a narrowly sectoral manner. The
mini-watershed must be adopted as the
unit for planning as well as implementation
of all programmes of land improvement
and the resources, both financial and
human, of all departments concerned, such
as forests, horticulture, agriculture, soil
conservation, minor irrigation, drainage and
rural development, etc, must be brought
together at the field level to ensure the best
possible time and at the minimum possible
cost.
All this is easier said than done, considering
that departmental loyalties and mind-sets
are still fiercely exclusive. But hope lies in
the fact that with the coming in of panchayati
raj, all development agencies will necessarily
1400
have to yield to co-ordination al the hands
of zilla parishads, block samilies and even
gram panchayats. Hope also lies in the
emergence of increasingly active and
knowledgeable NGOs in the field of rural
development and resource management.
Above all. there are the living examples of
villages like Ralcgaon Shindi and Sukho
Majri, which have transformed their
economies dramatically by achieving the
fullest possible utilisation of all local
resources of land and water, to inspire other
village communities to adopt similar
approaches, and force government
organisations, however recalcitrant they may
be, to fall in line with the need of the hour.
IV
Wastelands Development:
A Flawed Concept
In hindsight it is clear that while the 1985
decision to give special importance to the
problem of wastelands was well-intentioned,
it was not based on a proper appreciation
of lheovcrall situation in the field of resource
management. It represented, in fact, an overly
simplistic approach to a problem of great
complexity and betrayed the preoccupation
of its authors with degraded forest lands
alone.
While announcing the setting up of the
National Wastelands Development Board
(NWDB), Rajiv Gandhi mentioned that
“continuing deforestation has brought us
face to face with a major ecological and
socio-economic crisis”. This observation was
no doubt correct, but only partially so,
because it completely ignored the part played
by non-foresl lands in bringing about the
crisis in question. What the then PM failed
to appreciate was that although degraded
forest lands represented a most serious
problem, they accounted, all said and done,
for only 30-odd mh out of the total of around
80 mh that are wastelands, and represented
an even smaller proportion of the country’s
total degraded area of 175 mh, which must
be held responsible as a whole for the mess
that we find ourselves in.
The fact that the notification setting up the
NWDB talks of the need for a “massive
programme of afforestation and tree planting"
- obviously on forest wastelands alone and makes no mention of the existence of
around 25 mh of wastelands that have been
created by waterlogging and salinisation
also shows what an inexcusably narrow
view was taken at that time of the problem
of wastelands.
The failure of the NWDB was mainly due
to the inability of the forest departments in
(he stales to work in close co-ordination with
other agencies concerned with non-foresl
wastelands in taking up cost-effective
programmes in accordance with the
‘complete mini-waiershed’ principle. The
Eighth Plan document (1992) is explicit on
this point and has observed as followsAn important reason why planning and aCii0
programmes for wastelands develop^
have tended to remain inadequate is the I,- t
of co-ordination between the FOr
Organisation which is the implcmeniin
agency in most States and other department
like Agriculture, Horticulture, sOjJ
Conservation, Minor Irrigation and RUraJ
Department (para 4.14.13).
d
The existing wastelands development
schemes generally are not based on integral.
ing the control of run-off rain water for
reducing erosion, soil and water conservatIOn
and water harvesting (para 4.14.14)
Il was no doubt the disappointing
performance of the NWDB that led the
government in 1992 to move it out of the
purview of the ministry of environment and
forests (MEF) and place it in the ministry
of rural development (MRD), where a special
new department of wastelands development
(DWD) was created to host it. However this
move, which was obviously intended to
provide the NWDB with a new image and
a new sense of purpose, lost much of its
significance when, following the MEF’s
reluctance to part with its jurisdiction on
‘forest wastelands’, it was decided to entrust
the NWDB. as well as of course the DWD,
with responsibility only for ‘non-foresl
wastelands’. Both these organisations thus
became misnomers and the formeremerged
weaker than before as a result of this
administrative reform.
The present scenario in wastelands
development is depressing in the extreme
because while there is just no talk of
wastelands that are suffering from
waterlogging and salinisation. even the
responsibility for attending to denuded
wastelands has been split between two
ministries and there is as yet no institutional
arrangement for bringing them together so
that they may follow the ‘complete mini
watershed’ approach on the ground. It is
necessary to remember in this connection
that since forest wastelands and non-lorest
wastelands are inextricably juxtaposed in
most situations, neither ofthesecan be tackled
alone in a cost-effective manner under the
existing administrative arrangements.
However this is by no means the end of
the story. For even if we were to correct this
anomaly by once again creating a unified
agency for dealing with both forest and nonforest wastelands, and even if we were to
also place responsibility for the rcclarnane0
of waterlogged and saline lands squarely 00
such an agency, the case for treat*0*
‘wastelands development’ as a sCPaf_
subject by itself would still beuntenablc^.
such an approach would necessarily
the problem of the 95-odd mh ol deg
agricultural lands further into
background. This would be a great
because contrary to popular belie
department of agriculture’s schemes
conservation of soil and waler on e
Economic and Political Weekly
, . o.
s 19#
June
•
agricultural Jan
because they
violation of the
principle.
Considering l
j
cUre, and tha
production in .
degradation-pre
further deterior;
higher priority
former. This n
worrying a littl*
a little more at
jf they are not
■ 'slowly convert'
!
There is yet a
j
cannot afford,
urgent need of
the even more u
■1
for attention,
complacent wii
depletion and c
i.
thatdonotbclor
and are believei
lands are aroun
extent and com
forests and arou
. lands.
A little thou$
as we are. it
complacent abc
asourremainin;
it is common kr
to be exploited
south and funct
the north-east z
Il is necessary t
that it would noi
on satellite imag
areas under goo
even if the den
down from 10(
a result of me
felling, it will
forest with *gO'
the rapid natural
on large open
country can als
the area under
As far as out
■
concerned, the
irrigation and a
• threat of waterk
are also often d<
;
and receive lar
fertilisers and p
long run, damn;
lands can also
• thoughtlessly <
... uses such as fa
. necessary to re
^0 per capita
Jands which st
. ,s expected to
' ,he year 2000.1
Jo keep a speci
lands and sa1
shrinkage.
;
. ^Gnomic and
j-ricultural lands are not at all doing well
In view of the above analysis, it is clear
be producti ve enough to sustain a population
[<cause they are being implemented in that the very concept of giving special
ofone billion by the end ofthis century with
higher standards ofliving than now prevail.
Eolation of the 'complete mini-watershed’ attention to wastelands needs to be discarded
Wc must have long-term plans to meet this
in favour of a broader approach that will
principlecontingency (emphasis added.)
Considering chat prevention is belter than cater to the needs of land management in
Indira Gandhi considered this matter to be
cure. and that wastelands are out of all its aspects. This means that, instead of
production in any case, the protection of a department of wastelands development, of such urgency that she asked her minister
degradation-prone agricultural lands against we should have a Department of Land of planning to examine, within a period of
further deterioration merits obviously much Resources and that instead of a so-called two months, how the proposed commission
higher priority than the amelioration of the NWDB that is concerned with the health of could be set up. However, while all the
former. This means that we should start only around 50 mh of non-forest wastelands necessary motions were gone through, the
worrying a little less about wastelands, but (but is paying no attention to problems of proposal was ultimately allowed to die a
a little more about degraded lands which, waterlogging and salinisation) we must have slow death, ostensibly on the legalistic ground
if they are not saved in time, may also get an apex body like the Central Land Use that land management is a state subject.
slowly convened into wastelands.
Commission that the government had decided The centre contented itself with a
There is yet another consideration. If we to create in 1974 but has not so far been able recommendation lolhestatesthat they should
cannot afford, any longer, to ignore the to bring into being because of a general lack take up the work of land management in
urgent need of wastelands for attention and of interest in this subject. Both these bodies their own territories through state land use
the even more urgent need of degraded lands must naturally accept responsibility for all boards. But in the absence of an apex body
for attention, can we continue to be problems relating to the country’s land al the centre to provide the necessary
complacent with regard to the dangers of resources in their totality, no matter whether leadership and backing in a new field of
depletion and deterioration faced by lands they are classified as forest lands or non activity, such boards as do exist have proved
that do not belong to either of these categories forest lands, as public lands or private lands, to be singularly ineffective.
Il is interesting to note that the National
and are believed to be in good health? Such as healthy lands or sick lands and in case
lands are around 89 mh (264 minus 175) in of the latter, whether they suffer from erosion Commission on Agriculture (1976) also gave
its full support to the proposal for the setting
extent and comprise around 39 mh of good or waterlogging.
up of a central land commission in words
forests and around 59 mh of good agricultural
which deserve to be quoted:
lands.
V
No specific agency of the government was
A little thought would show that placed
A Record of Apathy
charged till the end of the Fourth Five-Year
as we are. it would be dangerous to be
Plan with the responsibility for the proper
Our rather casual approach towards
complacent about our non-sick lands. As far
useof the land. It was, however, soon realised
as our remaining good forests are concerned, problems of resource management becomes
that such a state of affairs where this
it is common knowledge that these continue evident when we consider the manner in
important basic resource has no known
to be exploited illegally. Veerappan in the which we have thrown away some very good
custodian of its interests cannot be allowed
south and functioning plywood factories in opportunities for tackling them effectively.
to continue. Fully realising the urgency of
the north-east are proof enough of this fact. The first opportunity was presented by K M
the problem, it has recently been decided
It is necessary to mention in this connection Munshi’sclarioncall in 1952 for the greening
that the existing vacuum in policies.
that it would not be prudent to rely overmuch of the country through a massive tree-planting
organisations and programmes relating to
on satellite imagery forinformation regarding programme aimed at placing one-third of the
land and soil management should be filled
areas under good forest cover. For one thing, country’s land surface under tree cover.
on an urgent basis... At the national level,
However,
instead
of
getting
down
seriously
even if the density of a good forest comes
it is proposed to have a central land
down from 100 per cent to 40 per cent as to the work of converting this grand vision
commission which will be charged with the
into
a
solid
reality,
a
cynical
establishment
overall responsibilityforall matters relating
a result of moneycombing and selective
to the assessment and optimum management
felling, it will continue to be shown as a quietly turned it into a meaningless annual
of the country's land resources. We fully
forest with ‘good tree cover’. For another, ritual for the ceremonial planting of trees by
support these measures (emphasis added.)
VIPs. As a result, 40 years later not 33 per
the rapid natural spread of 'Prosopisjuliflora’
Four years later, in 1980, the N D Tiwari
on large open tracts in many parts of the cent but only around 13 per cent of the
country can also create the impression that country’s land surface can claim to be under Committee on the Environment revived the
proposal for the setting up of the central land
good tree cover.
the area under forests is not diminishing.
The next opportunity came in 1973 when commission but to no avail. This committee
As far as our good agricultural lands are
concerned, they are almost entirely under Indira Gandhi approved a bold suggestion recommended that while a full-fledged new
irrigation and as such are susceptible to the that the centre should assume greater department should be set up to look after
threat of waterlogging and salinisation. They responsibility for the care of the country’s the environmental problems of the country,
are also often double, or even triple, cropped total land resources and create a nodal the subject of proper land management was
and receive large applications of inorganic authority for this purpose. Ina historic minute so important that it deserved to be looked
fertilisers and pesticides which can, over the dated December 29, 1973 she observed, after by a central land commission which
should serve as a “policy-making, co
long run, damage the soil. Good agricultural inter alia, as follows:
Based on our experience of soil erosion, ordinating and monitoring agency for all
lands can also suffer depletion by being
droughts and floods and their increasing issues concerning the health and scientific
thoughtlessly diverted to non-agricultural
financial liability, a large part of which had management of our land resources”.
uses such as farm-houses for the rich. It is
to be borne by the Centre, the paperf*] However, whi Ic the former recommendation
necessary to remember in this context that
argues in favour of the creation of a central was immediately accepted and implemented
the per capita availability of agricultural
land commission. I am ifi broad sympathy before the year was out, no action was taken
lands which stood at 0.48 hectare in 1951
with its approach and feel that we can no on the latter. Clearly, there was as yet no
Is expected to go down to 0.14 hectare in
longer afford to neglect our most important
the year 2000.11 would therefore be desirable
natural resource. This is not simply an
to keep a special eye on the health of such
environmental problem but one which is [*] ‘A Charier for (he Land’ by B B Vohra.
lands and save them from damage or
basic to the future of our country. The stark
Economic and Political Weekly. March 31.
shrinkage.
question before us is whether our soil will
1973.
Economic and Political Weekly
June 8, 1996
1401
political will to place the management of our
land resources on a sound footing.
The Sixth Plan document (1980) also made
a strong plea for better land management in
terms which arc as valid today as they were
15 years ago:
The losses which (he country is hearing on
account of the continued degradation of its
land resources are of staggering dimensions
and constitute one of the important threats
to our economic progress... The country can
hope to achieve a continuous improvement
in agricultural production only if the
problems of land degradation are tackled
with the utmost vigour. Such an effort, though
gigantic by any standards is, however,
inescapable if the country’s agricultural
future is to be assured. Considering that
even after all possible steps are initiated
immediately, it will be years before results
begin to show and that further massive
damage will inevitably continue during this
^^period, there is absolutely no room for
^^complacency on thisfront (emphasis added.)
These fine sentiments were however only
in the nature of lip-service to the cause
because they were not matched by any
significantly largerallocations for better land
management during the Sixth Plan period.
Hopes for a better deal for the land were
revived once again in early 1985 when Rajiv
Gandhi warned the nation of the serious
“ecological and socio-economic crisis” it
faced and set up. along with the ill-fated
NWDB, the National Land Use and Conser
vation Board (NLUCB) with responsibilities
which were more or less in line with what
had been earlier envisaged for the proposed
central land commission. However, the
NLCUB proved to be stillborn, thanks mainly
to its curious constitution - this bloated. 32member body possessed no whole-time
members at all. Unbelievable as it may seem.
even its pan-time member-secretary was
located in a ministry different from that of
d^part-time chairman, and was therefore
Unaccountable to him in any manner.
With such a track record behind us, it is
difficult to be sanguine about the future.
However, regardless of what has happened
in the past, it is incumbent on the centre at
this critical juncture to realise the gravity of
the situation and treat land management as
the core item of an agenda for national
survival. It would also be useful to place this
subject above party politics and hold urgent
consultations with ail important political
parties on how it should be approached.
The present arrangements - under which
the agriculture department is supposed to be
responsible for only eroding agricultural
lands, the ministry of water resources for
only command area development and for the
control of floods and waterlogging, the
department of forests for only forest lands
and the rural development department for
only community and revenue lands and area
development programmes - are thoroughly
irrational and must be scrapped.
1402
A 10 to 15 year indicative plan for dealing
with all aspects of land management must
be drawn up by (he centre within the shortest
possible time. Simultaneously, the states
must be asked to draw up their own long
term plans and to implement them in a time
bound manner under the watchful eye of the
centre. State Land Use Boards must be
revamped and strengthened and a prestigious
and adequately empowered Central Land
Use Commission should be constituted to
act as a custodian and conscience-keeper of
the interests of the land, as a think-tank and
repository of reliable data, as a clearing
house for relevant information and as a
catalyst for creating public awareness of
what is at stake.
VI
Money Is Not an Important
Constraint
A superficial look at the magnitude of the
problems that face us with regard to 175
odd mh of degraded lands and wastelands
may give the impression that huge
investments will be required to implement
a time-bound programme for their
amelioration, and that lack of financial
resources may come in the way of such an
undertaking.
Such fears are however largely imaginary.
If we look at the matter a little more closely,
we find that even if we assume that our 150
odd mh of denuded and eroding lands will,
on an average, require an investment of Rs
4,000 per hectare and that 25 odd mh
waterlogged and saline lands will require Rs
30,000 per hectare, the total bill will be
around Rs 1,65,000 crore. If the programme
is spread over 15 years, it will demand an
annual outlay of around Rs 11,000 crore
According to information collected by the
DWD, the amounts that arc presently
available for schemes which have an
important component of afforestation and
soil and water conservation arc as follows:
(Rs crore per annum)
I
Ministry of Rural Areas and
Employment (Previously Rural
Development)
1250
2 Ministry of Environment and Forests 906
3 Ministry of Agriculture and
Co-operation
260
4 Planning Commission
362
5 NABARD
50
6 State Soil Conservation Departments 341
7 State Land Development Banks
1106
Total
4275
This means that the gap in resources will
be around Rs 7000 crorc per annum. How
ever, it will in fact be much smaller because
allocations for rural development arc likely
to be increased steeply in the Ninth Plan.
Experience has shown that wherever local
communities have come forward to take an
active part incontrollinggrazing,and thereby
facilitating the natural regeneration of vegetal
cover on denuded lands, and in adoptjnp
other biotic and engineering means
conserving both soil and water, departmental
costs have come down appreciably. Again
it cannot be denied that costs will also come
down if existing leakages of funds are
effectively plugged and schemes arc
implemented in a more efficient and cost.
effective manner. It is pertinent to recall
in this connection, the well-known
observation of Rajiv Gandhi that hardly 15
per cent of the enormous sums spent on rural
development programmes succeed m
benefiting the intended beneficiaries, thc
rest of the money either goes waste or into
the wrong pockets.
What also needs to be appreciated is that
thanks to our failure to plan and implement
all soil and water conservation schemes on
a strictly ‘complete mini-watershed’ basis
a good part of the investments that arc now
being made in this field prove to be infructuous. Once these deficiencies are removed
and all available resources are carefully
pooled and utilised meaningfully, the entire
position will change dramatically. Thc real
problem therefore lies not in the scarcity of
financial resources but in our present inabi
lity to utilise them to the best advantage
VII
Major Tasks Ahead and Some
Suggestions for Tackling Them
In the paragraphs that follow an attempt
has been made firstly, to outline the most
important tasks that face us and secondly.
to suggest what kind of policy and
administrative changes will need to be made
to tackle them effectively.
Task No 1: Complacency must be shed
The record shows that there has so far been
no stern political will to tackle the country’s
central problem of poverty and therefore of
poor resource management. This in turn is
due to the fact that thc people who matter
mistakenly believe that all is well on the
agricultural front because we can operate the
PDS without having to import foodgrams.
Unless this vicious circle is broken by a
sustained and effective campaign for
shedding complacency and creating greater
awareness of thc perils inherent in the
continued neglect and mismanagement of
our natural resources, there is little likelihood
of any great improvement in (he situation.
A decision should be taken to mount such
a campaign and to make thc ministry ol infor
mation and broadcasting responsible for HTask No 2: Soil and Water must be
Conserved to the Maximum Posstb
Extent
There is no question that soil erosion '
which affects around 150 mh out ot 1 ®
country’s total land area of 305 nth
constitutes thc biggest single threat to
sustainability of our agriculture, as wel
Economic and Political Weekly
q
June o.
1996
of our economy as a whole. For not only
joes it increasingly reduce the productivity
f lands subject to erosion but also results
in the loss to the sea of large quantities of
priceless sweet water, in the siltation of
reservoirs and rivers and in the aggravation
of both floods and droughts.
The technologies for conserving both soil
and water are wellknown and simple in
nature and an increasing number of villages
(like Ralegaon Shmdi) which have adopted
them have demonstrated that these can be
easily implemented by farmers themselves
with only a little outside help. Denuded
lands must be enabled to regenerate
themselves, through the control of grazing,
and their soil and moisture regimes must be
improved by biotic as well as engineering
devices, such as contour trenches, before
thay are placed under plantations whether
of fruit, fodder, fuel or timber. Simul
taneously, agricultural lands - which are
almost invariably situated in the lower
reaches of the mini-watersheds - must be
terraced and bunded along true contour lines.
Run-off losses must be reduced at every
possible point in each mini-watershed by
creating physical barriers, such as weirs,
nallah plugs and check-dams and storages,
across all drainage lines. Such impediments
not only help to conserve local resources of
rainfall to the maximum possible extent in
the form of soil moisture, ground water and
small storages but also act as silt traps, and
ensure that the water that leaves the mini
watershed is genuinely surplus to its own
requirements.
A total approach of this kind has already
brought about dramatic changes wherever
it has been tried and needs to be adopted
in all parts of the country, regardless of
whether they receive heavy rainfall, or
moderate rainfall or little rainfall. The concept
of zero or minimum soil loss, aimed at
achieving the maximum conservation of both
soil and water through biotic as well as
engineering means needs to be popularised
among all rural communities so that they
may begin to take an increasing interest in
managing their own resources. At the same
time, all the schemes which are essentially
aimed at soil and water conservation but arc
being carried out today under a variety of
tiescriptions - such as DPAP, DDP, RVP,
FPRY, NWDPRA, IWDP, JRY, EAS, GGS
- and by a number of departments, should
tic merged into a single scheme for ‘soil and
*ater conservation’ which should be squarely
tiased on the ‘complete mini-watershed’
Principle.
Such a reform will result in saving a lot
the expenditure which is at present being
meurred wastefully because in the absence
ofinter-departmental co-ordination, none of
existing schemes, whether of the ministry
E and F or of A and C or of R A and
. • being implemented according to the
Complcte mini-watershed’ principle.
^cononiic and Political Weekly
Considering that the total resources that arc
being invested in such schemes arc over Rs
4000 crore per annum this reform will result
in great financial benefit to the country.
It is suggested that a 15-year perspective
national plan for the conservation of both
soil and water and therefore for the
amelioration of all the 150 odd mh of
degraded lands and wastelands that arc
erosion-prone should be formulated and
taken up for implementation not later than
the start of the Ninth Plan. This Plan should
be only indicative in nature and should not
be imposed in any way on state governments.
It is the affected villages that should be
encouraged to draw up (heir own plans which
should then be consolidated into district and
state plans.
Responsibility for this ambitious
programme should appropriately be placed
on the ministry of R A and E which is
responsible for the alleviation of rural
poverty. The ministry must achieve the
requisite co-ordination between all the three
ministries concerned with afforestation and
soil and water conservation schemes, if
necessary, by obtaining orders of the cabinet
on this all-important point. It must also
arrange for the suitable re-orientation and
training of all existing staff in these three
ministries. It must also consider how itshould
reorganise itself for its new responsibilities,
and how the existing DWD should be
transferred to a department of land resources.
The dissolution of both the NWDB and the
NAEB is another matter which will need to
be considered urgently.
Task. No 3: Reclamation of waterlogged
and saline lands
This is a subject that has suffered great
neglect. So much so that even reliable data
regarding the extent of the damage done is
not readily available. However, as already
mentioned above, it is very likely that in 1995,
the affected area is as large as around 27 mh.
Sinch most such areas require to be
provided with drainage, preferably
underground, the cost of reclamation is very
high - somewhere in the region of Rs 30,000
to 40,000 per hectare. Perhaps it is this
circumstance which has prevented both the
department of agriculture and the ministry
of water resources from taking an active
interest in this matter.
As in the case of land subject to erosion,
a 15-ycar plan must be drawn up for
ameliorating not only the lands that have
already suffered damage but also those which
are likely to face this threat in the near future.
The responsibility for formulating and
implementing this plan must be placed
squarely on MWR because of its expertise
in executing drainage works and its
responsibility for reducing seepage losses
from unlincd canals and preventing the
improvident use of waler, both of which
contribute to waterlogging.
June 8, 1996
Task No 4: Containment of deserts
It is estimated that around 25 mh suffer
from wind erosion. These are mostly lands
situated in the Rajasthan desert, and there arc
reports that it is slowly expanding as a result
ofthc movement ofsand through wind action
The ways of controlling the spread of
deserts arc known - they lie mostly in the
putting up of wind barriers and shelter belts.
A lOto 15-year plan to enclose the Rajasthan
desert within a bell of suitable trees should
be drawn and implemented. Simultaneously
steps should be taken to reclaim desert areas
by controlling grazing so that natural
regeneration of trees and grasses may take
place.
Responsibility for this programme should
be placed on the ministry of R A and E.
Task No 5: Protection ofgood agricultural
lands
As already noticed there are only about
50 mh of agricultural lands that arc apparently
in good health today, but are vulnerable to
many serious threats. The health and physical
integrity of all such lands must be carefully
monitored and guarded as suggested above.
Responsibility for this task should be
placed on the ministry of agriculture.
Task No 6: Protection ofremaining natural
forests
The pace at which the deterioration of our
39 odd mh of good natural forests is taking
place is not generally recognised. Many of
these forests arc not classified as ‘Reserve
Forests’ because of the rights enjoyed by
local tribal populations. There are also other
legal impediments in the way of effective
action.
The Veerappan incident in the south and
the apparent ease with which the extraction
of valuable timber continues to take place
in the north-east show how serious the
problem is It is necessary to give the highest
priority to this matter and put a complete
and very early end to all unauthorised fellings
in our remaining forests, if necessary by
arming our foresters with enhanced punitive
and legal powers, as well as with weapons
wherever the situation may so require.
Responsibility in this field should be placed
on the ministry of environment and forests
Task No 7: Containment ofcoastal erosion
This is another area of great neglect which,
considering the length of our coastline, can
be the cause of great damage along
uninhabited reaches without the government
becoming aware of it. The matter needs to
be carefully studied with the use of satellite
imagery so that vulnerable areas maybe
regularly monitored, and effecti ve steps taken
in time.
Responsibility for this task should be
placed on the department ol ocean
development.
1403
Task No 8: Review offlood controlpolicies
The failureofexisting policies in this field
is apparent from the fact that although Rs
2,500 crore were spent on ‘flood control’
programmes such as the construction of
earthen embankments and dykes between
1954 and 1989, the area described as ‘flood
prone’ has nevertheless increased from
around 25 mh in 1950 to around 40 mh in
1989.
It is time to realise that the root of the
trouble lies in excessive run-off and soil
losses in denuded catchments. These place
additional demands on the water-carrying
capacity of rivers even while reducing it by
raising their beds through siltation. The real
answer to the problem therefore lies in
stepping up natural regeneration, affores
tation and soil and water conservation
programmes in catchment areas in an
effective manner. The emphasis should shift
from ‘flood control’ to ‘flood prevention’
and from the treatment of symptoms to the
treatment of the disease itself. The money
saved by curtailing infructuous expenditure
on the construction of earthen structures that
get washed away ever so often should be
diverted to the treatment of catchment areas.
Responsibility in this regard should be
placed on the ministry of waler resources
Task No 9: Review of policies in surface
water
The MWR has been traditionally
concentrating only on the construction of
surface irrigation projects as if this was an
end in itself and not merely a means to the
end of greater agricultural production. This
concept needs to be given up and replaced
by one that stresses the accountability of
MWR for its performance in terms of its
actual contribution to enhanced production.
This is a matter of great importance because
the more we succeed in the field of irrigation,
the less will be the pressure on marginal
rainfed agricultural lands which, in happier
circumstances, should be reverted from
cropping to horticulture, silviculture or
pasture production in the interests of their
own health and productivity as well as of
downstream areas.
The seriousness of the present situation
in the field of surface water management has
already been described at length in Section
II. The suggestions contained in that section
deserve to be considered urgently by the
MWR.
Task No 10: Review ofpolicies in ground
water
Ground water is bound to assume even
greater importance in the years to come,
firstly, because of the failure of surface
water projects and, secondly, because of the
extreme ease and speed with which it can
be developed in the private sector wherever
it is available at reasonable depths. However
the very attractiveness of this priceless the secretaries of the concerned depart menu
resource is turning into a threat to its health namely, agriculture and co-opcrati0n’
and sustainability. Water tables are going environment and forests, water resource/
down rapidly in many regions under rural areas and poverty alleviation and the
conditions of indiscriminate over-pumping Planning Commission. The member.
and in certain, mostly coastal, areas, aquifers secretary of the proposed commission should
be a whole-time officer of the rank Of
are getting infested with saline water.
So far inadequate importance has been secretary to the GOI, and should be assisted
given to ground water management by a by an adequately equipped secretariat.
ministry that is overly pre-occupied with the
The commission should be given a suitable
expansion of the M and M sector. However mandate which should include responsibility
it would be a tragic mistake to continue to for ensuring that the 10 tasks mentioned
neglect this resource and take it for granted above arc pursued energetically by thc
so to say, merely because it is a free gift of ministries concerned and are not allowed to
Nature. Action clearly needs to be taken in recede into the background. Il should act as
custodian and conscience keeper of the
the following three major directions:
(a) The exploitation of ground water should country’s land resources and a vigilant watchbe controlled to ensure that withdrawals do dog of its interests.
Responsibility for this task should be
not exceed sustainable limits - the CGWB
and state government water boards should placed on the ministry of R A and E.
be vested with the necessary administrative
and legal powers to achieve this end and also Task No 12: Creation ofland use authorities
at the state and district levels.
suitably strengthened.
Although state governments were advised
(b) Research in ground water should be
stepped up. We must have the most complete as far back as 1974 to set up state land use
possible knowledge of the nature and boards, the boards that have been formed
capability of each aquifer, and the source are more or less defunct. There is obvious
and exact extent of its recharge. Research need for setting up state land use boards in
in artificial recharge should be given the image of the CLUC and ensuring that
they work energetically al the district level.
particular attention for obvious reasons.
(c) In view of the growing demands on this the zilla parishad should discharge all
resource, its replenishment should be functions covering the optimal use of local
facilitated by all possible means, both natural land resource.
Responsibility in this regard will rest with
and artificial As far as the former is
concerned, the successful implementation the ministry of R A and E.
of Task No 2 above will go a long way
VIII
towards the enhancement of ground water
Last Word
resources.
Responsibility for this task has to be
When all is said and done, it must be
borne by the ministry of water resources.
remembered that I ike any other issue ofgreat
Task No 11: Creation of a central land importance better resource management is
use commission
too serious a matter to be left to be tackled
The “vacuum in policies, organisation, by government alone. This is particularly
and programmes relating to land and soil true at a time when the latter have theirhandr
management” that was noticed by the NCA full with many crises of a much more
in 1976 has unfortunately yet to be Tilled immediate nature than the insidious threat
even though more than 20 years have elapsed posed by the continuing mismanagement
since the proposal for an adequately struc of the country’s natural resources, however
tured and adequately empowered Central awesome this threat may be. One must
Land Use Commission (CLUC) was first also reckon with the fact that in the absence
mooted in 1973. There is no question that of a strong and informed public opinion
if we really mean business in the all-important on the subject, there is at present no *”
field of resource management this vacuum on the part of any political party to pressim**
should be filled without any further delay. the government on this forgotten troot.
In the circumstances, very great
The exact form the proposed commission
should take is a matter that will require ponsibility rests on concerned c'1'zens’_£
.
detailed study. However, a suggestion that can read the writing on the wall.10
(following the pattern of the Planning together to create a strong v0'u^\(/
Commission) can be safely made is that it organisation that will act as a watch aNj .
should be presided over by the minister for the nation’s interests in this ncglecte
j
R A and E (who is responsible for the Such a body should do everything
j
amelioration of rural poverty) and have three to sec that the issues that have been dis
,
or four whole-lime members of suitable rank above are kept alive, the greater a»a
who should be well-versed in matters relating is created with regard to their urge
■’
to land management and of whom one should that governments, both al the centre
j
be designated as the deputy chairman. Other the states, are constantly reminded 0
members of the proposed body should be duties in this field.
Unlike
important
intra-regi
not memb
are a mai
-1
I
TEXTILE an
sectors whicl
restraints in c
of experts ha
these import
optimal invol
in the world
years have v
arrangements
levels whief
changes in in
apparel. The p
(due to MFA
Round of Tn
Insubstantial!
and apparel fi
On the other h
blocs' may a<
textile and
countries, if t
blocs.
Unlike the i
of the 1960s
trading blocs
specific induj
one of such
©ven speciale
treatment, th
^cessions,
rigorous and
rehting to ‘ru
*’8nificantim|
Furthe
^dards to r<
Micy with
"•ndards wot
^apparel ex
**togionalis
’fc’evel of im
"^‘gn invc
’^Oology tf
. !®<his conte
praised at
U,e fo>rr
. ■ to stgnif
1^'1
8, 1^'
1404
Economic and Political Weekly
^unC
|
Atomic
Poverty and ’Income Distribution in India
K N Ninan
Against the background of global poverty and income distribution pattern, this paper analyses the trends and causal
factors behind rural poverty in India both at the national and state levels during 1957-58 to 1986-87. Adopting an alternate
model and categorisation of the time period of analysis into two phases which is empirically and theoretically justified
the paper observes that contrary' to the findings of other researchers, not only are there distinct time trends in the incidence
of rural poverty in India, but also-while these trends were positive and significant in Period I (1957-58 to 1968-69), they
were negative and significant in Period II (1969-70 to 1986-87). Also, the rate of decline in the incidence of rural poverty
in the latter period was much higher than the rate of increase in rural poverty in the preceding period. These observations
are valid for both all-India and across states, using alternate measures of poverty, i e, the head count ratio and Sen's
poverty index. The paper then attempts both a time series and cross-section analysis of the causal factors behind rural
poverty in India, especially probing into the role of agricultural growth, inflation, access to subsidised food through the
public distribution system, population pressure on environmental resources, rural consumption levels and inequality, and
infrastructure development on the incidence of rural poverty for all-India and across states.
The paper suggests that policies to accelerate agricultural growth, infrastructure development and provide better
access to subsidisedfood, along with measures to control inflation promise to be most effective in reducing the incidence
of rural poverty in India. Measures to control population growth and promote environmental conservation too ought
to be incorporated into anti-poverty alleviation strategies in India. The paper cautions against the implications of
recent policy changes in India, viz, structural adjustments, resulting in low priority to agriculture as against industry,
slashing of public expenditures on social sectors including subsidised food, etc, which are detrimental to the poor
and could reverse the negative trends in rural poverty visible after 1969-70.
I
Introduction
SUSTAINED economic growth along with
investment and public policies to improve
labour productivity and access to basic needs
are widely perceived to be most effective in
reducing poverty. The experience of the
western countries earlier and more recently
of those from east Asia bears this out. On
the other hand the experience of some of the
south Asian and African countries shows
ho w-the absence of such a growth momentum
ahd policy environment could constrain
efforts to reduce poverty. In some countries
such as Brazil and Pakistan, despite rapid
economic growth, progress in terms of social
indicators like under five mortality, primary
'enrolment rates has been dismally low.
Whereas Sri Lanka, despite slow economic
growth, ranked high in terms of social or
quality of life indicators. The countries of
sub-Saharan Africa provide the curious
combination of low economic growth and
social progress resulting in endemic poverty.
Whether there are trade-offs between growth
and poverty or inequality as suggested by
the Kuznet’s hypothesis or the Immiserisation
hypothesis are difficult to surmise in the face
of diverse evidences. While countries like
Japan, Korea recorded significant declines
in poverty following a reduction in inequities
after widespread land reforms, others like
Indonesia were able to reduce poverty with
the income distribution pattern unchanged;
in still others such as Brazil. Costa Rica a
worsening of poverty was associated with
worsening inequities. All these illustrate how
complex a subject poverty is and that no
1544
study of poverty will be complete unless it incidence of rural poverty in India. Other
takes note of local conditions and historical researchers using two point comparisons,
processes distinct to each country and region.
tabular or graphic analysis, came to diverse
It is in this context that this paper focuses conclusions. While Bardhan (1973)
on poverty in India. The size of India’s observed a rising trend in rural poverty
absolute poor (around 420 million people between 1960-61 and 1968-69, Minhas
in 1985 with annual per capita incomes (1970) noted a fall between 1956-57 and
below US $ 370), the availability of time
1967-68; Mellor and Desai (1985), Sundrum
series data on poverty and related factors for (1987) and Bhattacharya el a) (1991)
a reasonable length unique for any developing observed a zig-zag pattern in the inter
country as also the diversity of situations temporal incidence of rural poverty, rising
and experiences make India idea) for for a few years, then falling, again rising and
analysing the dynamic processes and causal falling subsequently. These contradictory
factors behind poverty. Moreover, given the findings may be attributed to the differences
similarity of problems faced by many in methodology employed, the time period
developing countries the Indian experience and choice of base and terminal years. A
may also have lessons for other developing major short-coming of most of these studies
countries to imbibe. Our focus is specifically is that while attempting to draw inferences
on rural poverty. This is because unlike about underlying lime trends in rural poverty,
many Latin American countries where they implicitly assume that the lime period
poverty is largely an urban phenomenon, in under consideration is structurally and in
India as in the rest of Asia and Africa it is . terms of the policy environment favouring
largely a rural phenomenon. The rural poor ' the poor homogeneous which is far from
constitute about 80 per cent of India’s total true. Adopting an alternate model and a
poor. Apart from analysing the trends in categorisation of the time period of analysis
rural poverty in India over time and states, into phases which is theoretically and
the paper also attempts to analyse the factors empirically justified, we come up with more
influencing rural poverty in India both at the meaningful and consistent results. These
show that contrary to the findings of other
national and state level.
Pioneering efforts in this direction have researchers, not only are there distinct time
been made by Ahluwalia (1978), Narain trends in rural poverty both al the all-lndia
[vide, Mellor and Desai 1985] and others [cf and state-level, but also while these trends
various articles in Bardhan and Srinivasan were positive and significant in one period.
1974; Srinivasan and Bardhan 1988; they were negative and significant in the
Krishnaswamy 1990J. Ahluwalia’s analysis subsequent period. The rate of decline in
for t^e period 1956-57 to 1973-74 revealed rural poverty during Period II was higher
no underlying time trend in rural poverty in than the rate of increase in rural poverty
India and for most states. He also observed during Period I, both for all-India and most
wide fluctuations in the inter-temporal states. Our study also confirms the strong
Economic and Political Weekly
June 18. 1994
negative association between agricultural
performance and the incidence of rural
poverty observed by Ahluwalia and others,
using both all-India and inter-state data, and
to that extent disproves the proposition of
a weak link between agricultural growth and
rural poverty put forward by some, notably
Rajaraman (1975) and, Griffin and Ghose
(1979). It also confirms the positive
association between inflation and rural
poverty observed by Narain. But linlike
Narain’s results, ours are based on more
rigorous tests covering not only all-India
data, the head count ratio (as the dependent
variable) and absolute prices but also inter
state data using alternate measures of
poverty and price including examining the
effects of lagged agricultural output and
price variables on rural poverty. We also
extend our horizon beyond the narrow
confines of the agricultural performance
and price variables to also empirically study
the relatively neglected issues relating to the
role of population growth and environment,
access to subsidised food through the public
distribution system (PDS), the level of
infrastructure development, inequality and
other factors on the incidence of rural
poverty. The paper also questions on
theoretical and empirical grounds the practice
common among many poverty researchers
of including a separate time trend variable
as an additional explanatory variable in
poverty functions to serve the role of a
cover-all variable for all other time-related
factors not explicitly considered in the given
model. In short, we not only seek to update
and extend further the contributions of
Ahluwalia, Narain and others in this area,
but also overcome some of their
shortcomings. As mentioned earlier, in
addition to trends in rural poverty at national
and state level we also analyse the causal
factors behind rural poverty in India. This
is at two levels—a time series analysis of
factors affecting the incidence of rural poverty
at all-India level and a cross-section analysis
of factors affecting inter-state incidence of
rural poverty at three points of time. The data
for the study are drawn from official reports
and refer to the period 1957-58 to 1986-87.
Before proceeding, it will be useful to
situate the poverty and related question of
income distribution in India in an
international perspective. Table 1 shows that
south Asia (including India) with 23 percent
of the world’s population alone accounts for
almost half of the world's poor. India’s share
itself is more than a third of the world’s poor.
In terms of some social indicators India lags
behind though in terms of life expectancy
and net primary school enrolment rates
India’s position is better. However, within
India, some regions (e g, Kerala) are on par
'vith developed countries in terms of social
Progress. India’s per capita income in 1989
*as US S 340 which is higher than that of
Bangladesh but lower than of Pakistan, Sri
I-anka and China. Using the UN's ICP
Economic and Political Weekly
estimates, with the US per capita income as
base with 100, India’s per capita income
which is 4.7 of this, is on par with that of
Bangladesh, but less than that of Pakistan,
Sri Lanka. The income distribution pattern
in India conforms to that in many.counlrieL
For instance, the share in income of the
bottom 20 per cent population in India and
Japan is over 8 per cent each whereas that
of the top 10 per cent are 27 and 22 per cents
respectively. The share of the bottom 20 per
cent population in India is higher than in the
UK, US, Brazil and several other countries.
But then the income levels in India are several
times lower compared to in Japan, the UK
US, Brazil and several other countries.
n
Trends in Rural Poverty
Most poverty studies on India rely on the
household consumer expenditure survey
data collected by the National Sample Survey
(NSS) for their analysis. These arc available
almost uninterrupted on an annual basis from
the mid-50s up to 1973-74. Subsequently
Table 1: Poverty and Income Distribution: A Global Profile
Countries/Regions
Sub-Saharan Africa
East Asia
China
South Asia
India
Eastern Europe
Middle East/North
Africa
Latin America and
Caribbean
All developing
countries
Countries
Bangladesh
India
China
Pakistan
Ghana
Sri Lanka
Indonesia
Colombia
Botswana
Malaysia
Venezuela
Brazil
UK
USA
Japan
Percentage Share
to World Total
Population Poor
(in 1985)
Proportion of Poor
Social Indicators
(Head Count Ratio
Under 5
Life
Net Primary
in Per Cent)
Mortality E;<pectancjf School
1985
2000
(Per 1000
(Years) Enrolment
Rate
(Pr ejected) Births)
(Per Cent)
16 1
25.1
18.8
46 6
37.6
0.5
47
20
20
51
55
8
43
4
3
26
25
8
196
S>6
58
172
199
:>3
50
67
69
56
57
71
56
96
93
74
81
90
5.4
31
23
148
61
75
87
6.3
19
11
75
66
92
82 2
100.0
33
18
121
62
83
94
32 8
23 2
23.0
17.1
8.3
GNP Per Capita UN’s ICP Estimates
in US S
of GDP Per
(1989)
Capita: US=100
(1989)
180
340
350
370
390
430
500
1200
1600
2160
2450
2540
14610
20910
23810
4.7
4.7
—■
8.2
—
10.5
—
19.3
—
—
—
66.1
100.0
71.5
Year
1981-82
1983
—
1984-85
1987-88
1985-86
1987
1988
1985-86
1987
1987
1983
1979
1985
1979
Share in Household Income by
Percentile Groups of Households
Top 10
Bottom 20
Per Ccn?
Per Cent
9.3
8 1
—
7.8
6.5
4.8
8.8
4.0
2.5
4.6
4.7
2.4
5.8
4.7
8.7
24.9
26.7
__
31.3
29.1
43.0
26.5
37 1
42.8
34.8
34 2
46.2
23.3
25.0
22.4
(1) The poverty line in 1985 purchasing power parity (PPP) dollars is $ 370 per capita a year
for the poor.
(2) Social indicators: Under 5 mortality rates arc for 1980-85 except for China and south Asia
where the period is 1975-80. These are the probabilities of dying before age 5, life
expectancy at birth—it is the number of years a new-born infant would live if prevailing
patterns of mortality at the time of its birth were to stay the same throughout its life; net
primary enrolment rale—the number of children aged 6 to 11 enrolled in primary school
as a percentage of the population age 6.to 11, adjusted for each country’s age structure
for primary school, the figures of the latter two indices refer to the mid-80s.
(3) The UN’s international comparison programme (ICP) has developed measures of real
GDP on an internationally comparable scale, using purchasing power parities (PPP)
instead of exchange rates as conversion factors.
(4) Income distribution: For Bangladesh, India, Ghana and Indonesia data refer to per capita
expenditure; for Pakistan—household expenditure; for Sri Lanka, Colombia. Malaysia
and Venezuela—per capita income; for remaining countries—household income.
Sources: (1) World Development Report 1990—Poverty, World Bank, Oxford University Press, New
York, 1990.
(2) World Development Report 1991—The Challenge ofDevelopment, World Bank, Oxford
University Press, New York, 1991.
Notes'.
June 18, 1994
1545
the NSS decided to collect these data on a
quinquennium basis. However, with a view
to building a time scries of data a decision
was again taken to collect such data on an
annual basis from 1986-87 based on a smaller
sample to supplement those from the
quinquennium surveys. In the absence of
time series data on household incomes, most
poverty researchers in India have relied on
the NSS consumer expenditure survey data.
Moreover, consumer expenditure bemg a
better proxy for permanent income is more
suited for such analysis than household
income data which may correspond to only
current income. Most Indian researchers on
poverty have used a poverty norm of Rs 15
monthly per capita consumption expenditure
at 1960-61 prices for rural India to estimate
the incidence of poverty in rural India [cf
Dandekar and Rath 1971; Bardhan 1973;
Ahluwalia 1978], This expenditure was
deemed to ensure a person access to a
specified minimum bundle of goods and
services. This norm was inflated for
subsequent years using the Consumer Price
Index for Agricultural Labourers (CPIAL)
with the base 1960-61 = 100, collected by
the labour bureau, government of India
which is the only available rural-specific
consumer price index. Because of variations
in commodity prices and rates of inflation
across states to derive the corresponding
state-specific poverty lines this norm was
adjusted using the state-specific consumer
price indices for rural areas with the all -1 ndi a
rural price as 100 for a given year. Details
as to how these are computed are available
in a number of studies [cf Bardhan 1973].
The state-specific CPI ALs are used to inflate
the state level poverty lines for subsequent
years. This norm has been used to arrive at
the incidence of poverty for rural India as
a whole and statewise. Of the measures used
to estimate the incidence of poverty, the
most popular are the head count ratio which
estimates the proportion of poor with
reference to the specified poverty line and
the Sen’s poverty index which is a more
sophisticated and composite poverty index
which takes note of the proportion of poor,
kthe gap between the poverty line and the
mean consumption ofthe poor and the Lorenz
ratio of consumer expenditure of the below
poverty households. Using the NSS consumer
expenditure data and the-above norms
Ahluwalia (1978) estimated the incidence of
poverty in India in terms of these two indices
for the period 1956-57 to 1973-74. These
indices have been updated up to 1986-87 for
all-India and statewise by Suryanarayana
and are readily available in a recent study
[Mahendra Dev et al 1991 ]. These estimates
supplemented by those from Ahluwalia’s
study have been used for our analysis.
Though our analysis spans a 30-year
period, as mentioned earlier, we have only
17 observations at O”r disposal for analysis,
because of gaps in 'he data cited earlier.
This implies that while our trends are based
1546
on annual observations from 1957-58 to
1973-74, save for two or three missing
observations, thereafter up to 1986-87 they
arc based on observations available al
greater point intervals. Most of the trend
fitting exercises earlier [cf Ahluwalia 1978}
implicitly assume that the period to which
the observations belong arc structurally and
in terms of the policy environment favouring
the poor homogeneous which is far from
true. It is our view that the period stretching
from 1957-58 to 1986-87 can be broadly
visualised as consisting of two phases. Period
I from 1957-58 to 1968-69 and Period II
from 1969-70 to 1986-87, the latest year for
which poverty estimates were available at
the time of writing. As is well known the
green revolution marked an important phase
in India’s agricultural development when
there was a structural break in the trend rate
of agricultural growth. Due to the bad drought
years of 1965-66 and 1966-67 and its after
effects the benefits of the green revolution
were perceptible only from 1967-68/1968-69
onwards. Even a visual examination of the
time series data reveals that the incidence
of poverty measured in terms of the head
count ratio or Sen’s poverty index except
for the late 50s generally showed an upward
tendency and reached peak levels during
1965-68 and thereafter reversed to record a
fall. The post-1969-70 period also marked
an important shift in the policy environment
towards the poor when following the split
in the ruling Congress Party Indira Gandhi
in a bid to outwit her political opponents and
•-____
fulfil her ‘Garibi Hatao’ (Banish Hunger)
slogan sought to give a pro-poor content to
her party’s programmes. Recognition
regarding the limitations of the market
mechanism in reaching the fruits of develop
ment to the poor and also political and eco
nomic compulsions, the post-1969-70 phase
witnessed a spurt in poor-centred welfare
programmes through direct institutional
interventions such as Integrated Rural
Development Programme, National Rural
Employment Programme, Food for Works
Programmes, Employment Guarantee
Scheme, Mid-day Meal Schemes for school
children, etc, focusing on improving their
asset base, employment generation and
providing access to basic needs, etc. Thus
there are strong theoretical and empirical
grounds to treat the period from 1957-58 to
1986-87 as consisting of two distinct broad
phases, Period I from 1957-58 to 1968-69
and Period II from 1969-70 to 1986-87, as
indicated earlier
Now regarding the mechanics of our
analysis. For fitting trends we have two
options. One is to fit separate trends for the
two sub-periods. But with only limited
observations available we will be left with
few degrees of freedom for econometric
analysis if we fit trends thus. Moreover, this
would also imply that our observations
pertain to two different samples. Using an
alternate methodology which overcomes
these.shortcomings we propose to estimate
the trends for rural poverty in India, using
the following model, viz*
—:
------ .—:—i-—;__•' . ■
_____ J
WORK MOTIVATION
Models for Developing Countries
edited by: RABINDRA N KANUNGO and MANUEL MENDONCA
This book greatly enhances our knowledge and understanding of the issues
involved in work motivation in developing countries and how to deal with
these in actual practice. It provides valuable insights into designing appropriate
human resource management practices which are suited to the context and
which can thereby enhance work motivation.
This exciting book combines a strong practical bias with conceptual thrust and
will be of enormous interest to those involved in organisational behavior,
human resource development, international business and cross-cultural
management.
290 pages (including 19 tables and 9 figures) a 220x140 mm
o Rs 275 (cloth) © Rs 150 (paper) o 1994
SAGE
PUBLICATIONS INDIA PRIVATE LIMITED
Post Box 4215, New Delhi 110 048
Economic and Political Weekly
June 18. 1994
g. '
wher
index
! of mo
j for C<
. period
sampl
me'hi
npP ’f
slope
two pLeast.
toe'.ti
India I
for Pei
deri ve
using
calculi
povert
Sen's
The
interest
aid the
no und
of rural
our rest
adistint
but als<
signific
sign i fin
urespeci
in terms
poverty
the inci
period •
increase
period.'
state Ics
positive
' land nee
in both ii
a rural
I -ian the
aPeriod
' -ere stat
ases. Ii
'’radesh.
egative
jtsults i
»'ertv f
■'ter 19f
-’’solute
otnpone
Facto
An obv
slat are til
■ral povc
' °ur con
'•lysis.
I picukun
•itributi
; Hie Gh
’Te thar
g, = an + V + M + a, (d-0
^hcre, g = Head count ratio or Sen’s poverty
judex; t = Time; d = Dummy variable where
j-0 for Period I; and d=l for Period II;
j t = Product of dummy and time.
The advantage of this model is m terms
?f more degrees of freedom at our disposal
for econometric analysis; inferences about
periodwise trends can be drawn from a single
sample rather than two (as per the alternate
methodology posed earlier), and more
l(pportant it enables us to see whether the
slope itself has undergone a changeover the
[v.0 periods. Linear trends using Ordinary
Least Squares (OLS) method have been used
yestimate the trends in rural poverty at all
India level and across states. The estimates
for Periods I and 11 presented in Table 2 arc
jenved from the estimated linear equations
using the above model. The trends are
calculated using two alternate measures of
poverty, viz, the head count ratio and the
Sen's poverty index, as mentioned earlier.
The results presented in Table 2 are quite
interesting. Contrary to Ahluwalia’s findings
■ ind those of other researchers who observed
pounderlying time trends in the incidence
j of rural poverty in India and for most states.
■ car results show otherwise. Not only is there
' ^distinct time trend in rural poverty in India
| bur also while this trend was positive and
!significant in Period I, it was negative and
.significant in Period II. This is true
j irrespective of whether poverty is measured
| in terms of the head count ratio or the Sen’s
fjoverty index. Further the rate of decline in
i-ie incidence of rural poverty in the latter
|period was much higher than the rate of
I increase in rural poverty in the preceding
period. These observations hold true at the
y-iate level too with most states reporting
I Positive trends in rural poverty during Period
:-indnegative trends in Period II with respect
j’^both indices as well as the rates of decline
'rural poverty in Period II being higher
’•■*n the rales of increase in rural poverty
‘‘Period I. These trends (positi ve or negative)
i';re statistically significant too in most
Interestingly three states, Andhra
’radesh. Tamil Nadu and Punjab reported
Native trends in both periods. While our
^lls indicate negative trends in rural
'■’eny for India as a whole and across states
1969-70, it may be noted that the
I'Toluie poor still constitute a significant
■ ^Ponem in India.
!
|
HI
■ ^actors Affecting Rural Poverty
<l!1 obvious question that arises is as to
j are the factors accentuating or reducing
,'■«! poverty in India and across states. This
F concern in the remaining part of our
Fa|ysis. Gi ven the importance of the
■cultural sector in the Indian economy—
• '■‘!ribuung as it does to about 40 per cent
.
GNP and providing sustenance to
j'" than two-thirds of the people—it is
L.
| ^mic and Political Weekly
obvious that the fortunes of the rural poor
in India are intrinsically linked to that of the
agricultural sector. Ahluwalia’s study cited
earlier observed a close negative association
between the incidence of rural poverty in
India and agricultural growth. Agriculture
impacts on the poor in more ways than one.
A higher agricultural output helps lower
food prices as well as improve food
availability both of which arc to the advantage
of the poor. It will not only generate
employment opportunities in the agricultural
sector but also through its linkage effects
spur growth in the non-agricultural sector
too thereby creating income earning
opportunities. Agricultural growth on the
whole will give a fillip to overall economic
development raising agricultural incomes.
However, if agricultural growth involves a
shift from labour-intensive crops and
technologies to labour saving ones this could
as well work to the detriment of the rural
poor rather than beneficial since wages from
agricultural employment constitute a major
component of the incomes of the poor.
Evidences from India, however, suggest that
on the whole the green revolution resulted
in a net increase of labour use and real wage
rales (Dantwala 1985], Some, however, feel
that in the context of the institutional and
structural constraints characteristic of most
low income countries including India the
beneficial effects of growth would be mostly
expropriated by the non-poor [cf Griffin and
Gho.se 1979]. The trickle-down effect implied
by Ahluwalia’s finding of a negative
correlation between agricultural growth and
the incidence of rural poverty was thus
Table 2: Trends in Rural Poverty in India: Statewise and for All-India, 1957-58 to 1986-87
Period I—1957-58 to 1968-69; Period 11-1969-70 to 1986-87
States
Dependent Vanable
Head Count Ratio
Sen’s Poverty Index
Constant
Constant
Time
Time
1
II
Assam
I
11
Bihar
1
11
Gujarat
1
II
Karnataka
I
II
Kerala
1
11
Madhya Pradesh 1
II
Maharashtra
1
II
Orissa
1
II
Punjab and
1
Haryana
II
I
Haryana* only
II
I
Punjab* only
II
Rajasthan
I
II
I
Tamil Nadu
II
Uttar Pradesh
I
II
West Bengal
I
II
All-India
I
II
0.50900.6949
0.2084*
0.4617**
0.47670.6785
0 36450.3075**
0.3611*
0.6590*
0.5493*
1.0170*
0 44040.7453**
0.4911*
0.6854**’
0.5932*
0.8312**
0.22940.3618-0.2160*
0.3033
0.3169*
-0.2374
0.3143*
0.5617*
0.6004*
0.7634***
0.3938*
0.5907
0.5124
0.9424*
0.4442*
0.6615**
Andhra Pradesh
-0.0053***
-0.0189*
+0 0156*
-0.0077**
+0.0140"’
-0.0071**
+0.0109’**
-0.0160*
+0.0206*
-0.0128*
+0.0090***
-0.0290’
+0.0074
-0 0139”
+0.0065
-00118**
+0.0039
-0.0131**
+0.0048
-0.0097*
+0.0118
-0.0084
-0.0157
-0.0077
+0.0028
-0.0141*
-0.0017
+0.0145*’*
+0.0125***
-0.0098**
+0.0167**
-00178"
+0.0055
-0.0135*
0.1927*
0.2645**
0.0418”*
0 1452**
0.1857*
0.3195
0 1224*
0 2481’
0 1232*
0.3005* .
0.2490*
0 4897*
0.1782*
0.3244”*
0 1819*
0.2413
0.2776*
0.4401*’
0.0790
0.1182
0.0500**
0.0966* ”
0.0200
0.071 1
0.1153*
02650*
0 2937*
0.3175
0.1526*
0.1863
0.17850.45450.15580.2823**
-0 0027
-0.0079**
+0.0058**
-0.0024*
+0 0103**
-0 0048”
+0.0061
-0.0071*
<■0.0118*
-0.0072*
+0.0040
-00155*
+0.0038
-0.0071*-*
+0.0036
-0.0039***
-0.0001
-0.0097**
-Negligible
-O.OO38***
+0.0100
-0 0036*••
+0.0180
-0.0030***
+0.0017
-0.0080*
-0.0057—
-0.0068
+0.0049
-0.0026
+0.0115**
-0.0102*
+0.0048**
-0.0069-
(1) These equations arc derived from the estimated equations using the model mentioned in
the text. The trends computed here are linear trends.
(2) *, **. ***—Statistically significant at 1.5 and 10 per cent levels of significance. In the
equations for Period II derived from the estimated equations, the significance of the
constant term is inferred on the basis of the statistical significance of the dummy variable
in the estimated equation, while that of the time trend variable is inferred on the basis of
the statistical significance of the (d.t) variable.
(3) + - Trends computed separately for Punjab and Haryana are based on data for the period
1964-65 to 1986-87, Period I—1964-65 to 1968-69 and Period II as in all other cases.
Sources: The basic data for the above analysis were taken from Mahendra Dev et al (1990) and
Ahluwalia (1978)
Notes:
June 18. 1994
1547
challenged by a number of researchers [cf
Rajaraman 1975; Griffin and Ghosc 1979].
However, these observations are based on
weak theoretical or empirical support. To
cite an instance, Rajaraman’s empirical
findings implying a weak causal link between
agricultural growth and rural poverty was
based on just 10 observations of which only
four pertain to the post-green revolution period.
Another factor influencing the incidence
of poverty is inflation. Inflation acts like a
regressive tax hitting hard the poor leading
to a deterioration in their entitlements and
real incomes [Sen 1982; see also Sen in
Mellor and Desai 1985]. Agricultural growth
itself has an in-built inflationary or
deflationary effect. A bumper harvest tends
to depress prices whereas a bad harvest tends
to push them up. Other domestic and external
factors too such as demand-supply situation.
import/expon decisions, uncertainty, etc, also
affect prices. Since food constitutes a
predominant portion of the consumption
basket of the poor it is the food prices which
cause most anxiety to the poor. Dharm
Narain’s study [vide, Mellor and Desai 1985]
highlighted the role of nominal prices in
affecting die incidence of rural poverty.
However, Narain’s analysis was restricted
only to all-India data, the head count ratio
and absolute prices. There is a need to validate
this further using data for regional
disaggregates, alternate measures of poverty
(say, the Sen’s poverty index) and prices
(relative prices loo in addition to absolute
prices), including the lagged effects of
agricultural output and price on rural
poverty. These are attempted incur analysis.
Population growth, poverty and environ
ment are closely inter-linked. Rapid popula
tion growth impacts on poverty in many
ways. It can offset the beneficial effects of
economic growth on poverty as experienced
by some of the south Asian countries.
Moreover, poverty intertwined with rapid
population growth exercises intense pressure
on scarce environmental resources resulting
in environmental degradation through
.overexploitation of fragile resources—all of
which have an adverse effect on poverty. The
role of the above factor on rural poverty too
needs to be probed into.
A factor which is believed to have worked
to the advantage of the poor in India
particularly after 1969 is the plethora of
poor-centred welfare programmes through
direct institutional interventions. There is
hardly any empirical attempt to test the
influence of these programmes on rural
poverty. Of the various institutional
interventions, provision of subsidised food
through a public distribution system (PDS)
assumes importance for the poor. However,
except in Kerala and other southern states
the programme is largely urban-oriented,
though it is gradual’y being extended to
rural areas in some of the other slates too.
The specification of this variable posed
problems tor our analysis. Except for one
1548
year (1986-87) where the NSS have furnished
data on commodilywise actual purchases
from PDS to total purchases separately for
rural and urban areas, lime scries data on
PDS arc available only in the form of PDS
offtake aggregated for the rural and urban
sectors. Rather than using a time trend
variable (for a critique of this procedure, sec
below) to account for this factor we preferred
to use the PDS variable as limited by the
data availability expressing it in the form of
PDS offtake to total net availability of
foodgrains or alternatively the fair price
shops per lakh of population except for the
cross-sectional analysis pertaining to 198687 where actual data on PDS purchases to
total purchases of foodgrains for rural areas
have been used. Here only rice and wheal
which account for bulk of the cereal purchases
through PDS have been considered.
The role of other factors, viz, the rural
consumption levels, inequality in rural
consumption (a proxy for income inequality),
and infrastructure development loo need to
be incorporated in our analysis.
It has also been customary for some
researchers [cf Ahluwalia 1978; Rao and
Mishra 1981. Narain vide Mellor and Desai
1985; and Saith vide, Bhaltacharya et al
1991] to include a time trend variable as
an additional explanatory variable in
poverty functions to serve as a cover-all
variable for all other time-related factors
influencing poverty not explicitly
considered inagiven model.This implicitly
assumes that all such time-related factors
not accounted for have a unidircction
influence on poverty which is questionabl
In fact, while some such time-relatJa
factors, for c g, rural population pressur
on agricultural lands, could be expected t*
exercise an upward-push effect on poverty
others such as PDS offtake are expected
to exercise a downward-push effect on
poverty. The inclusion of a separate tim*
trend variable in these circumstances is
questionable and could even affect the
estimates of other explanatory variables
Keeping the above factors in view and
the limitations of data, we propose to
examine the causal factors behind rural
poverty in India between 1957-58 to 198687. The analysis is at two levels—a timc
scries analysis al all-India level and a cross
section analysis of inter-state data. To test
the robustness of our results the cross
section analysis of inter-state data is
attempted at three points of time, viz, 196061, 1970-71 and 1986-87. While 1960-61
falls within Period I, 1970-71 and 198687 belong to Period II. Though some have
expressed their reseivations about crosssectional analysis [cf Srinivasan, in Mellor
and Desai 1985] others have fell that it is
important in itself as it offers important
perspectives on the varying conditions
under which poverty occurs [Mellor and
Desai 1985]. Moreover, some of the
estimation problems one faces in time
series data (e g, multicollinearity, auto
correlation) pose a less serious problem in
cross section data.
BOOKS THAT MATTER A CHRONICLE OF
WORLD EVENTS
January-March 1994
Aquarterly publication giving authentic
day-by-day account of all that Is hap
pening In India and world over—In poli
tics, public affairs, agriculture, econ
omy, Industry, sports, entertainment, fine
arts and other fields of human activity.
Indispensable reference tool for competitive examinations. Jour
nalists, authors and researchers.With an elaborate Index for easy reference;
7b be published four times In a year
ISBN 81-212-0467-8 310pp. Hard CoverRs. 375 Paperback Rs. 150
—
GYAN BOOKS.(P)J
Publishers • Distributors • Exporter.
5, Ansari Road, New Delhi-110002
3261060,3282060 Fax: (91)-11-3285914
Economic and Political Weekly
I The varu
i ,|iows:
^pendent
;
[lead .
I ‘
• -dependent
To study r I
y perform:
assure on <
^ralconsurr
.jfrastrucr.m
.■ie follow i
(i) Agrie rl
Variables ■ <
sDPAGRI -
from Agn i
of rural po
SDPPRM—
Primary Si
quarrying)
rural pop I
SDAGRI tion per he
5DFDGR tion per lie.
u) Pace Var
cations):
rDPR—Com
cultural Lat
1960-61 =
SELFDPR, Consumer
Labourers (
• RELCWPI—.
Wholesale II
> liii) Populaiici
] Resources:
RPPAL—Ru
! Agricultural
j lation per h:
1 lo take.note
logies which
; liv) Institutio
■. ?DS—Propor
Net Avaiiat
' FDSFP—Fair
lation.
; eDSr—Propo
PDS to Total
; d)Sw—Prop1
from PDS i
Rural Areas
The last two
:<t 1986-87 i
^tion ana;v!
’■) Consumpt
Wity:
NEQRC—1.<
tion.
WPCEG— A
* Capita Con
,‘J General Rcr
VVPCEBD—
j Capita Cor
I bottom Dec
•J ’*) Infrastruc
June 18. 1^
pOnomic anc
L
idirectionj)
jestionablt
mc-reht^
on pressor
expected to
on poverty,
• expect^
effect on
parate tim.
nstances
affect the
variables.
i view and
propose to
ihind rural
58 to 1986els—a time
and a cross
-ata. To test
the crosste data is
.viz, 1960le 1960-61
and 1986some have
bout cross.. in Mellor
(hat it is
important
conditions
vlellor and
r>e of the
s in timeuity, auto
problem in
The variables for our analysis arc as
,]|ows:
ypendent Variable
Head count ratio (in per cent) or
tentatively the Sen’s poverty index.
■^pendent Variables
To study the impact of agricultural growth
?r performance), prices, rural population
. c>sure on environmental resources, PDS,
.j-tl consumption levels and inequality, and
,fr.isiruclure development on rural poverty
tollowing variables are considered:
i Agricultural Output/Performance
iriables (four alternate specifications):
\DPAGR1—Real Nel Domestic Product
from Agriculture al 1960-61 prices per head
of rural population.
\f)PPRM—Real Net Domelic Product from
Primary Sector (excluding mining and
quarrying) at 1960-61 prices per head of
rural population.
iXDAGRI—Index of Agricultural Produc
tion per head of rural population.
\DFDGR—Index of Foodgrains Produc
tion per head of rural population.
li) Price Variables (three alternate specifi□tions):
rDPR—Consumer Price Index for Agri
cultural Labourers for Food Items (where
1960-61 = 100).
•IELFDPR—Relative Food to General
Consumer Price Index for Agricultural
Labourers (1960-61 = 100).
RELCWPI—-.Relative Cereal to General
Wholesale Price Index.
ni) Population Pressure on Environmental
Resources:
RPPAL—Rural Population Pressure on
Agricultural Lands expressed in lakh popu
lation per ha of gross cropped area (so as
to take.note of land augmenting techno
logies which became prominent in Period II)
’■•) Institutional Intervention (PDS):
?DS—Proportion of PDS Offtake to Total
Net Availability of Foodgrains
•“DSFP—Fair Price Shops per lakh of popu
lation.
rbSr—Proportion of Rice Purchased from
PDS to Total Rice Purchases in Rural Areas.
D$w—Proportion of Wheal Purchased
from PDS to Total Wheat Purchases in
Rural Areas.
The last two variables were available only
’•* 1986-87 and were used for the cross<clion analysis for the year 1986-87.)
9 Consumption Levels/Consumption Inei>ality:
'HQRC—Lorenz Ratio of Rural Consump
tion.
^’PCEG—Average Monthly Real per
Capita Consumer Expenditure of the
General Rural Population.
VPCEBD—Average Monthly Real Per
f-apita Consumer Expenditure of the
Attorn Decile of the Rural Population.
'9 Infrastructure Development:
INFDEV1ND—Infrastructure Development
Index as constructed by the Centre for
Monitoring Indian Economy (CMIE),
Bombay.
Not all these variables have been included
in an equation at a lime because of the
constraint of limited observations. Further,
while some variables, viz, NDPAGRI,
RELFDPR. PDS PDSFP, INEQRC were
common to both the lime scries and crosssection analyses, others, viz, NDPPRM,
INDAGRI, INDFDGR, RELCWPI,
AVPCEG and AVPCEBD were included
only in lhe ume series analysis; similarly
FDPR. RPPAL. PDSr, PDSw and
INFDEVIND figured only in the cross-section
analysis.
In addition, the agricultural output and
price variables indicated above were also
used in their lagged forms. One could take
a stand that the level of poverty in a given
year is not only determined by that year’s
agricultural performance but also that of the
previous year. A good crop not only enables
a poor household to repay past debts but also
build up reserves to meet unforeseen
eventualities. Similarly inflation too has a lagged
effect. For instance, given the low incomes
of the poor a sleep rise in prices of essentials
may force them to borrow in order to arrest
a deterioration if not maintain their
consumption standards, the reverberations
of which will be felt in subsequent years as
well. To take note of these lagged effects,
an alternate specification of the agricultural
output and price variables is introduced
which is computed thus: (t + t-1 )/2. Thus in
all wc have four sets of equations. In one
set the dependent variable is the head count
ratio, in another lhe Sen’s poverty index.
Within these two categories, again one set
of equations are without lagged variables
and another set with lagged variables.
Multiple linear regressions using OLS
technique were used to estimate the
coefficients. A few variables, viz, AVPCEG.
AVPCEBD and RELCWPI had to be
dropped from the analysis as these were
found to be highly correlated with other
explanatory variables. ThePDS variable was
also found to be strongly correlated with lhe
price variable RELFDPR and its lagged
version in the time series analysis. In such
a situation it is common among researchers
to drop one of the two collinear variables.
However, this could affect the estimates of
the retained variables [Koutsoyiannis 1977]
particularly if it is an important one. Using
a procedure akin to Frisch’s confluence
analysis suggested by Koutsoyiannis which
Table 3: Determinants of Rural Poverty in India, 1957-58 to 1986-87
Equation No
Estimated Linear Equations
R2
DW
Statistic
0.64
0.51
0.49
0.58
0.64
0.49
1.8014
1.5947
1.5947
1.7118
1.8479
1.4335
0.80
0.79
0.84
0.85
0.80
0.79
1.5247
1.5369
1.7635
1.6617
1.6809
1.5474
0.72
0.76
0.67
0.66
0.72
1.8680
2.0353
1 6658
1.5626
1.8854
0.81
0.82
0.86
0.86
0.84
0 82
0.83
0.86
1.0839
1.5979
1.6559
1.6266
1.7173
I 5921
1 6222
1.6640
Dependent Variable: Head Count Ratio (in percent):
Without Lagged Variables:
I 125.8248* - 0.4643 NDPPRM* + 0.0328 RELFDPR
2 84 5681* -243 7478 INDAGRI* + 0.1733 RELFDPR
3 70.6672** - 206.6404 INDFDGR’ + 0.2185 RELFDPR
4 113.4423” - 0.4599 NDPAGRI* + 0.0276 RELFDPR + 29.3922 INEQRC
5 118.4528** -0.4655 NDPPRM’ + 0.0515 RELFDPR + 18.5776 INEQRC
6 69.2002 - 206.6731 INDFDGR* + 0.2222 RELFDPR + 3.6017 INEQRC
With Lagged Variables7 112 9312’-413.3022 INDAGRI* +0.2685 RELFDPR***
8 98.0312* - 323.4192 INDFDGR’ + 0.2075 RELFDPR
9 180.4665* - 0.8017 NDPAGRI* - 0.5094 PDS*** + 26.3342 INEQRC
10 183.3582-0.7759 NDPPRM* - 0.4129 PDS + 15.7246 INEQRC
11 99.9550’ - 416.7543 INDAGRI* + 0.2980 RELFDPR** + 35 2317 INEQRC
12 96.7555* - 323.4634 INDFDGR + RELFDPR + 3.3385 INEQRC
Dependent Variable: Sen's Poverty Index:
Without Lagged Variables:
13 0.5730* - 0.0028 NDPAGRI* + 0.0008 RELFDPR
14 0.5659* - 0.0028 NDPPRM* + 0.0010 RELFDPR
15 0.3318” - I 5254 INDAGRI*+0.0019 RELFDPR***
16 0.2464” - 1.3053 INDFDGR* + 0.0022 RELFDGR”
17 0.5536** - 0.0028 NDPAGRI* + 0.0010 RELFDPR + 0.0500 INEQRC
With Lagged Variables:
18 0.8566*-0.0040 NDPAGRI* + 0.0004 RELFDPR
19 0.8443* - 0.0039 NDPPRM’ + 0.0003 RELFDPR
20 0.4703* - 2.3514 INDAGRI* + 0 0024 RELFDPR*
21 0.3872’ - 1.8503 INDFDGR* +0.0020 RELFDPR*
22 0 9290’-0 0043 NDPAGRI’-0.0018 PDS+ 0.0271 INEQRC
23 0.7796’ - 0.0041 NDPAGRI’ + 0.0002 RELFDPR + 0 2399 INEQRC
24 0.7932' - 0.0040 NDPPRM* + 0.0004 RELFDPR + 0.1522 INEQRC
25 0.4543’ - 2.3557 INDAGRI’ + 0.0024 RELFDPR’ + 0.0435 INEQRC
Notes- (I) For a description of the independent variables refer the text.
(2) ’,”,”*—Statistically significant at 1.5 and 10 per cent levels of significance.
(3) In the equations with lagged variables only the agricultural output and price variables are
used in their lagged forms; lhe other variables in this set of equations are not lagged.
18, 19^
'•°n.-.rrtjc and Political Weekly
June 18, 1994
1 549
enables one to test for the presence of incidence of rural poverty across states is
IV
multicollinearity and also adjudge which of negatively correlated with agricultural
Conclusions
such variables to include or drop, it was growth, PDS and the level of infrastructure
noted that inclusion of PDS resulted only development, it is positively correlated with
Contrary to the findings of othc*
in a slight improvement in the R2. Hence price, rural population pressure on researchers of there being no undcrlyjn.
in the time series analysis wherever price agricultural lands and inequality in rural lime trends in rural poverty in India,
was included as an explanatory variable, consumption.
evidence shows that there were distinct time
the PDS variable was excluded. However,
because of our interest in knowing the
Table 4: Determinants of Inter-State Incidence of Rural Poverty in India__ A Cross
nature of relationship between PDS and the
Section Analysis for 1960-61,1970-71 and 1986-87
incidence of poverty, we also fitted'a set
of equations excluding the price variable Equation
Estimated Linear Equations
bOt including PDS. Only those equations No
which gave meaningful results have been
Year 1960-61
presented. These estimated equations for
Dependent Variable: Head Count Ratio (in per cent)
the time series analysis are presented in
Without Lagged Variables
Table 3.
1
-87.6262 - 0.0639 NDPAGRI*** + 1.4284 FDFR + 4 3046 RPPAL - 0.6847 PDS 0 53
The results arequite interesting. As evident 2
-30.8533 - 0.0760 NDPAGRI** + 9.9978 INEQRC + 0.8952 FDPR +
while agricultural output variables are
2.1655 RPPAL
050
negatively correlated with the incidence of
Dependent Variable: Sen's Poverty Index
rural poverty measured in terms of cither the
Without Lagged Variables
head count ratio or the Sen's poverty index, 3
-0.8299 - 0.0004 NDPAGRI*** + 0.0107 FDPR*** + 0.0264 RPPAL 0.0050 PDS
0 59
the relative food price variable is positively
-0 4209 - 0.0005 NDPAGRI** + 0 1121 INEQRC + 0.0067 FDPR +
correlated with poverty. The coefficients are 4
0.0108 RPPAL
O 55
statistically significant in most cases. These
Year 1970-71
results are in conformity with the findings
Dependent Variable: Head Count Ratio (in per cent)
of Ahluwalia and Narain. More noteworthy
Without Lagged Variables
is that w'hile the PDS variable has the expected
5
10.8175 - 0.0555 NDPAGRI** + 111.9093 INEQRC + 0.1126 FDPR +
negative sign, the inequality in rural
2.8090 RPPAL - 0.1607 PDSFP
0.70
consumption (a proxy for income inequality) 6
0.9489 - 0.0560 NDPAGRI** + 130.1453 INEQRC + 0.2424 RELFDPR +
is positively correlated with poverty. This
3.1941 RPPAL-0.1120 PDSFP
0.69
is true of both sets of equations where we
With Lagged Variables
have used or not used lagged variables as 7
—66.1861 -0.0506 NDPAGRI** + 1.1489 RELFDPR + 4.2802 RPPALwell as where the dependent variable are
0 2267 PDSFP
0.63
4.5138 -0.0590 NDPAGRI* + 149.0907 INEQRC*** + 0.0785 FDPR +
alternatively the head count ratio and the 8
2.6643 RPPAL
0.70
Sen’s poverty index. Thus our results suggest
5.6970-0.0566 NDPAGRI** + 130.8710 INEQRC + 0945 FDPR+*
that while agricultural growth and PDS tend 9
3.1153
RPPAL
-0.0756
PDSFP
0.71
to reduce rural poverty, price and inequality
Dependent Variable: Sen's Poverty Index
in rural consumption which are positively
Without Lagged Variables
correlated with poverty tend to push it up.
10
-0.1514 - 0.0002 NDPAGRI + 0.0020 FDPR*** + 0168 RPPAL The R2s of most of the equations are not only
0.0022 PDSFP***
061
quite high but also they are much higher in
II
0.2173 - 0.0002 NDPAGRI*** + 0.6703 INEQRC + 0.0014 FDPR +
the case of the equations using lagged
0.0156 RPPAL-0013 PDSFP
0.66
variables, indicating the important role of
With Lagged Variables
dynamic factors in affecting the incidence
12
-0.1750 - 0.0003 NDPAGRI** + 0.0022 FDPR**
0.52
13
of rural poverty.
-0.2551 - 0.0003 NDPAGRI** + 0.9191 INEQRC** + 0.0012 FDPR 0.0131 RPPAL
0.66
To analyse factors affecting the inter-state
14
-0.2470- 0 0002 NDPAGRI*** + 0.7950 INEQRC + 0.0013 FDPR +
incidence of rural poverty in India we now
0.0162 RPPAL - 0.0005 PDSFP
0 66
turn to the results of our cross-section analysis
Year 1986-87
presented in Table 4. Here again while
Dependent Variable: Head Count Ratio (in per cent)
agricultural growth was negatively correlated
Without Lagged Variables
with poverty, price variable was positively
12.2324 - 0.0256 NDPAGRI** + 0.0372 FDPR
0 43
15
correlated with poverty. The PDS variable
29 8869-0.1595 INFDEVIND** + 0.0202 FDPR + 1.1758 RPPAL-0.1263 PDSr 0.41
16
was negatively correlated with poverty.
With Lagged Variables
Interestingly RPPAL and INEQRC were
41.5414 - 0.0329 NDPAGRI* + 0.0052 FDPR + 0.0219 RPPAL - 0.2959 PDSr*** 0.62
17
positively correlated with poerty indicating
18
22.0765 - 0 0322 NDPAGRI* + 0.2062 RELFDPR + 0 0565 RPPAL,
0.2950 PDSr***
°6that rural population pressure on agricultural
Dependent Variable: Sen's Poverty Index
lands as well as inequality in rural
Without Lagged Variables
consumption exercise an upward-push effect
0.0271
-0.0001 NDPAGRI**+ 0.0002 FDPR
°
19
on poverty. Infrastructure development index
0.1033 - 0.0007 INFDEVIND** + 0.0001 FDPR + 0.0054 RPPAL - 0.0006 PDSr 041
20
was negatively correlated with poverty, as
With Lagged Variables
,
it should be. Again, as earlier, these
0.15674-0.0001 NDPAGRI*+0.00002 FDPR+0.0001 RPPAL - 0.0014 PDSr***
21
observations generally hold true for the four 22
0.0651-0.0001 NDPAGRI* +0.0010 RELFDPR + 0.0003 RPPAL.
sets of equations, i e, with/without lagged
0.0013 PDSr***
variables and where the dependent variable
are alternatively the head count ratio and the Notes: (I) For a description of the independent variables refer the text.
(2) *, **, ***—Statistically significant at 1, 5 and 10 per cent levels of significance.
Sen’s poverty index. The R2s of most of the
(3) In the equations with lagged variables only the agricultural output and price vanab
equations were quite high. Thus our cross
used in their lagged forms; the other variables in this set of equations are not lagge
section analysis reveals that while the
1550
Economic and Political Weekly
June 18. I?*1
Co
n - 9'0>
SPECIAL ARTICLES
Population, Poverty siinidi Employment in India
T IN Krishnan
This paper represents a preliminary attempt to examine the successes and f'ilures of the Indian economy in
integrating population issues with development planning and what were, or would be, the consequences of rapid
population growth for the alleviation of poverty in the country.
The first section of the paper discusses the momentum of population growth since independence and examines
the prospects of reducing its growth rate in future. This section analyses the crucial role of social and human
development in influencing fertility rates and shows how little emphasis was placed on promoting these objectives
in Indian planning.
The second section is devoted to a detailed examination of the relationship between foodgrains production
and population growth. This section examines the factors that determine the inter-state differences in
foodgrains consumption and evaluates the role of the public distribution system in lessening the inequalities in
consumption.
The third and final section presents a broad analysis of a few important questions relating to labour market
adjustments in response to population growth. This section touches upon the questions of the inadequacy of the
concepts used in measuring employment and unemployment within the institutional and social structure prevail
ing in different parts of the country, the relationship between agricultural wages and agricultural productivity
and the likely pressure on employment generated by the enormous increase in labour force during the next 35 years.
FORTY-FIVE years after attaining indepen
dence, in spite of notable achievements,
unrelenting population growth, pockets of
acute poverty-and tardy growth in employ
ment in modem industry continue to plague
the Indian economy. The Indian planning
strategy was precisely designed to tackle
these very issues. India was the first coun
try in the world to recognise the perils of
unmitigated population growth and to pro
claim an official policy' to bring down the
rate of population growth. However, through
out this period, the annual population
growth rate exceeded 2 per cent. The popula
tion of India rose from 361 million in 1951
to 844 million in 1991. Why did nearly 40
years of planning fail to bring down the rate
of population growth? What went wrong
with the elaborate planning exercises that
paid particular attention to social objectives
and sectoral consistencies? Nonetheless, one
marvels at the resilience and performance
of certain segments of the Indian economy
in coping with the excessive burden impos
ed on it by populatioe growth and how it
managed to feed its bulging population and
also in absorbing millions of persons who
entered the labour force. Food output cer
tainly kept pace with population growth and,
at least in rural areas, unemployment did not
appear to have risen significantly or at all
in response to population growth. Therefore,
one would be curious to know how the
Indian economy coped with its population
increase and what were the factors that aid
ed this adjustment to demographic pres
sures. This paper represents a preliminary
attempt to examine the successes and failures
in integrating population issues with Indian
development planning and what were or
would be the consequences of rapid popula
tion growth for the alleviation of poverty in
the country. The first section discusses the
Economic and Political Weekly
momentum of population growth since in
dependence and examines the prospects of
reducing its growth rate in future. This sec
tion analyses, and provides empirical sup
port, to the crucial role of social and human
development in influencing fertility rates
and shows how little emphasis was placed
on promoting these objectives in Indian
planning.
The second section is devoted to a detailed
examination df the relationship between
foodgrains production and population
growth. Foodgrains is the main source
of calories in the Indian diet, providing as
much as two-thirds or more of the total
calorie intake among the lowest income
groups. This section examines the factors
that determine the inter-state differences in
foodgrains consumption and evaluates the
role of the public distribution system in
lessening the inequalities in consumption.
The third and final section represents a
broad analysis of a few important questions
relating to the labour market adjustments
in response to population growth. This sec
tion touches upon the questions of the in
adequacy of the concepts used in measuring
employment and unemployment within the
institutional and social structure prevailing
in different parts of the country, the rela
tionships between agricultural wages and
agricultural productivity and the likely
pressure on employment generated by the
enormous increase in labourforcc during the
next 35 years.
I
Planning and Population Growth
Forty-five years after independence, the
population of India has grown to nearly two
and a half times its size at the time of
independence. The latest population census
November 14, 1992
counted 843.9 million Indians on March 1,
1991 [Prcmi 1991.1]. During the next 35
years, it is estimated that an additional 600
million will be added to this number [United
Nations 1991:208]. India’s population is pro
jected to exceed 1,000 million by the year
2000 when it would constitute 28 per
cent of Asia’s population or nearly 17 per
cent of world population.
Il is a matter of great concern that nearly
40 years of planned development has failed
to initiate any retardation in India’s popula
tion growth. During this entire period, the
annual population growth rate averaged 2.1
per cent. Though the 1991 population cen
sus indicated that the decadal growth rate
had marginally declined from 24.66 per cent
to 23.50 per cent, it is not of much consola
tion. In every decade since 1951, the net ad
ditions to total population had been rising
steadily, from 77.68 million during 1951-61
to 160.60 million during 1981-91 [Prcmi
1991:6]. The latest United Nations popula
tion projections indicate that the net addi
tion to the 1991 population during the next
decade would be close to 215 million even
with a marginal rate of decline in growth rate
[United Nations 1991:208]. The net addition
to India’s population is expected to fall
below that of 1981-1991 only in the second.
decade of the next century (Figure 1).
Birth and Death Rates
The rate of growth of population reflects
the differences in the rates of change in birth
and death rates. Demographic transition im
plies changes in all these rates; it disturbs
the long-term static condition where both
birth and death rates are high but stable and
the corresponding population growth rate is
low. In the initial phase of transition when
birth and death rates begin to decline, the
2479
Figure 1: Decadal Additions to Population
196! to 2021
latter declines at a faster rate thus raising the
population growth rate The birth and death
rates were estimated at 44 and 26 per 1,000
population for the decade 1951-61 but
declined to 37 and 15 during 1971-81 [Premi
1991:23]. The birth rate remained unchanged
at 34.8 per 1,000 population during the
period 1977-1984. The estimates of birth and
death rates are 30.6 and 10.3 in 1989
[Registrar General, June 1991].
The all-India birth and death rates con
ceal the wide disparities between regions and
states within the country. Birth rate exceeds
39 per 1,000 population in about 20 per cent
of the total number of districts (90 districts)
in the country [The Hindu, March 21 ]. Table
I provides the state-wise estimates of birth
and death rates based on three-year moving
averages for two periods, namely, 1979-81
and 1987-90. Birth and death rates are lowest
in Kerala and Goa and highest in the
B1MARU states. The table also gives the rate
of decline in birth and death rales for each
state during the entire period. While death
rates have declined in most states by a much
higher proportion, births show much lower
rates of decline. Therefore the backward
states still suffer from higher rates of popula
tion growth.
Kerala-Goa Model
Kerala and Goa, among the states in
India, have successfully brought down the
birth rates during the past 35 years. The birth
rate in Kerala declined from 30.5 per 1,000
population during 1971-73 to 20.7 during
1987-1989. Similarly, birth rate declined from
25.4 to 17.5 in Goa for the same period
[Registrar-General, June 1991]. Therefore it
is not necessary to look elsewhere to learn
how a rapid demographic transition could
2480
be initiated within India. The case histories
of Goa and Kerala demonstrate that it is
possible to reduce the birth rate from 35 to
20 per thousand in a period of about 25
years. But this requires determined action on
a number of fronts.
The. process ol demographic transition is
a result of the interaction of a number of
diverse aspects of psychological, social,
economic, technological and institutional
factors. Essentially it involves the collective
decision of a society to shift from a large
family size'norm to a small family size norm.
The earliest explanations attributed this shift
to modernisation, urbanisation and indus
trialisation. However, Kerala and Goa are
neither industrialised nor urbanised to the
extent necessary to explain the decline in
their birth rates. While Goa’s demographic
transition remains almost unexplored, Kerala
has been studied in considerable depth [cf
Krishnan 1976, Mari Bhat, et al 1991]. A
comparison of the trends in birth rates and
the possible factors in both states seem to
indicate that similar factors have contributed
to their demographic transition.
A recent view stresses psychological
modernity as an important factor contri
buting to health and demographic transi
tions [Mechanic 1992]. The psychological
factors relate to one’s attitude to the prac
tice of family limitation, the desire underly
ing the decision on the number of children
one wishes to have, the attitudes towards
health care and the ability to handle illnesses
of children, etc. These psychological changes
may be closely related to the historical and
expected probabilities of child survival in a
given society. If parents are convinced that
the survival probabilities are high then they
may decide to have fewer number of children.
Therefore, the preconditions for a fall in
birth rates are (I) the access to the means
to reduce infant and child mortality, and (2)
measures leading to changes in psychological
and social attitudes towards health care and
family size.
Reductions in infant and child mortality
depend upon the availability and quality of
maternal and child health care, the extent
and quality of medical attention during child
birth and the coverage of protection of
children through immunisations from infec
tions and other childhood diseases. Basically
it is the access to health care, attitudes
towards dealing with illness and child care
that finally help to bring down the infant
mortality rate. Access to health care involves
three components: (a) locational access; (b)
economic access; and (c) social access. Loca
tional access is particularly important for
poorer income groups since it can increase
their opportunity cost of treatment. For in
stance, if an agricultural labourer has to seek
medical assistance in a far off town, he
might be required to forego a day’s earnings
which might be his only source of income.
In such situauons he might either hesitate
or even give up seeking medical assistance
for himself or his family unless it was an
emergency. Most states in India have paid
scant attention to this question. In most
states nearly three-fourths of the total
number of hospital beds and the attendant
medical facilities are located in urban areas
and in large towns (Table 2). The number
of beds per lakh of rural population clearly
indicate that the extent of availability of
health facilities are almost similar in both
Goa^and Kerala. Therefore, we find that
their infant mortality rates are also similar.
it is hardly necessary to spell out the
arguments in favour of economic access.
Provision of free or subsidised health care
on the basis of income is an accepted prin
ciple in most societies. In our country social
access assumes great significance because of
our past history of denial of such services
to certain castes and social groups.
Attitudinal changes to health and family
size seem to be determined by levels of
education and the spread of literacy especial
ly among women. Studies from different
countries show that the number of children
born to a woman declines as her educational
level rises. So does infant mortality rate.
Thus access to health and female education
appear to be the most'critical factors in birth
rate decline. Female education not only initi
ates changes in one’s attitude towards family
size and health of children but also enables
women to improve their status within family
and in society. Both in Kerala and Goa their
demographic transitions were preceded by
health *and educational transitions. In
Kerala, these changes were first initiated in
the princely states of Travancorc and Cochin
at the beginning of the 19th century but with
the formation of the Kerala state it was also
demonstrated that the implementation of
Economic and Political Weekly
November 14, 1992
similar policies and programmes in Malabar,
which had higher birth and mortality rates,
yielded identical results. The distribution of
beds under health care points to the superior
positions of Goa and Kerala. Rural popula
tions have access to health in these two states
and this is reflected in improved maternal
and child health care and in lower infant and
child mortality rates. An evaluation of
universal immunisation programme under
taken in 1989 indicated that the proportion
of pregnant mothers and infants immunised
were the largest in Goa and in Kerala. The
proportion of pregnant women receiving
tetanus toxoid immunisation was 97.6 in
Quilon district in Kerala, 92.3 in north Goa,
78.6 in Pune district, 28.1 in Katihar district
ot Bihar and 22.38 in Kanpur Dehat district
of Uttar Pradesh. Similarly, the proportion
of children immunised without measles was
83.6 in north Goa, 80.5 in Quilon, 71.1 in
Pune, 4.3 in Katihar and 11.4 in Kanpur
Dehat. These rates of protection of pregnant
mothers are reflected in the proportions of
neo-natal mortality attributed to nco-natal
tetanus. 55 and 70 per cent of nco-natal mor
tality were due to tetanus infection in Katihar
and Kanpur Dehat districts while there was
no death due to this factor in Goa, Kerala
and Pune [Gupta and Murali 1989:212], Ac
cess to clinics and primary health .centres is
an important factor in pregnant women
receiving ante-natal care and in the'proportion of children being immunised. Such ac
cess ensures that child births occur under the
supervision of either medical personnel or
trained birth attendants which contribute to
a significant reduction in maternal and in
fant mortality rates. For instance, in the rural
sector of Kerala 40 per cent of births took
place in medical institutions in 1978 but rose
to 83 per cent by 1988. Along with other fac
tors this too might have contributed to the
decline in- peri-natal mortality [Krishnan
1991a].
Social access was also a key factor in the
relative spread of health care in the states
of Travancore and Cochin compared to
Malabar in its early period of development
1'n the initial period, access to health and
education were denied to backward commu
nities and the then untouchable communi
ties. However, the emergence of a few
educated persons from these communities
enabled (hem to agitate and organise for the
opening up of these facilities to persons
belonging to their communities too. These
agitations and demands begun at the end of
the last century in Travancore and Cochin
finally resulted in providing access to educa
tion and health to the untouchable commu
nities [Kabir and Krishnan 1992].
The social and psychological attitudes
towards health care and population planning
are essentially shaped by education. Goa and
Kerala demonstrate that it is female educa
tion that is vital for these changes. Female
literacy rates are highest in these two states
and had been so for decades. Fertility has
been found to be negatively correlated with
Economic and Political Weekly
the level of female education- The status of
women is closely related to the level of
education which in turn initiate concomitant
changes in the age at marriage, employment
status, income levels, utilisation of health
care and in the attitude towards the practice
of family planning. Birth and infant morlality rates are negatively correlated with ail
these factors.
marriage of girls and also in reducing the
child mortality rate because better educated
mothers paid greater attention to the health
and well being of their children. We also atTable 2: Distriblthon of Beds
under Health Care System
States
Tests of the relationships
In order io test the significance of female
education and access to health on fertility
rates, we estimated regression relationships
between female literacy and infant mortalily rates and the total fertility rate. We used
total fertility rate rather than birth rate
because the former is already standardised
for age distribution For this purpose we
utilised two different sets of data—state-wise
as well as district-wise. We were able to
undertake the latter exercise because districtwise data on-child mortality, total fertility
and literacy rates are available for 1981
[Registrar General, 1989a, 1989b]. However,
due to high degree of multicollincarity
among the explanatory variables, we failed
to obtain significant results in the regression
with state level observations for 1981
Therefore, we estimated the Spearman rank
correlation coefficients with the state data
in order to gauge the strength of the relationships. The most important finding'that
emerges is the role of female literacy rate not
only in reducing directly the fertility rates
but also in indirectly influencing birth rates
through its role in raising the age at marriage and in reducing the infant or child
mortality rates. Female education (female
literacy) strongly acts to raise the age at
I
Andhra
Assam
Bihar
Goa
Gujarat
Haryana
Jammu and
Kashmir
Karnataka
Kerala
Madhya
Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Thmil Nadu
Uttar Pradesh
West Bengal
India
No of Per Cent No of
Beds Per .of Beds Beds Per
Lakh
in Rural
Lakh
Popula- Area in Populalion, 1989 1989
lion in
Rural
Area 1989
2
4
3
—
59
10.59
9
60
23.56
16
34
820
3
237
2420
105
129
11.06
22
50
7.32
5
107
80
254
4.30
9.82
56.12
na
12
193
36
130
43
115
51
86
40
85
77
9.41
9.98
16.28
41.48
8.55
21.05
12.44
15.04
17.75
4
21
8
68
6
28
6
18
18
Source Estimated from CM IE, Basic Statistics
Relating to Indian Economy, Vol 2,
September 1991.
Table 1: Percentage Change in Birth and Death Rates, 1971-89
States
1
Andhra
Assam
Bihar
Goa
Gujarat
Haryana
Himachal Pradesh
Jammu and Kashmir
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
All India
Live Birth Rates
1979-81
1987-89
2
3
Per Cent
Change
in Birth
Rate
4
31.6
32.9
38.4
17.7
35.1
36.8
31.6
31.3
28.0
26.0
37.5
28.3
31.9
29.6
37.1
28.3
39.5
32.5
33.8
27.8
32.1
36.1
17.5
29.7
34.5
30.2
31.4
28.5
20.7
36.3
28.9
31.1
28.5
34.2
23.3
37.3
28.8
32.1
-12.03
- 2.43
- 5.99
- 1.13
-15.38
- 6.25
- 4.43
0.32
1.79
-20.38
- 3.20
2.12
- 2.51
- 3.72
- 7.82
-17.67
- 5.57
-11.38
-16.85
Death Rates
1987-89
1979-81
5
6
Per Cent
Change
in Death
Rate
7
11.7
11.5
14.7
7.1
12.4
11.0
10.8
9.3
9.7
6.8
15.7
10.0
14.0
9.2
13.5
11.7
16.4
11.3
12.7
9.9
11.3
12.6
7.8
10.2
9.0
8.9
7.9
8.8
6.2
13.5
8.4
12.7
8.2
12.1
9.3
13.4
8.7
10.7
-18.18
- 1.77
-16.67
8.97
-21.57
-22.22
-21.35
-17.72
-10.23
- 9.68
-16.30
-19.05
-10.24
-12.20
-11.57
-25.81
-22.39
- 29.89
-31.54
Source: Estimated from Sample Registration Bulletin, June 1991.
November 14, 1992
2481
tempted an alternative regression relation
ship with state-level data for 1989 where we
used the crude birth rate instead of the total
fertility rate. In this equation we also includ
ed couple protection rate as another ex
planatory variable in order to estimate the
impact of the family planning programme
in reducing the birth rale. Similar problems
of multicollinearity again cropped up in this
equation and therefore we had to drop either
infant mortality rate or female literacy rate
from the equation. The results of these
estimates are given below:
Regressions with Stale Data
(I) BR = 50.38 + 0.0899(IMR)- l.585(AM)
(2.4159)
(-2.422)
+ 0.045(CP)
(0.6261)
R2 = 0.786
(2) BR = 34.23+ 0.0436<IMR) + 0.00358(CP)
(0.7237)
(0.0510)
-0.2156(FLR)
(-2.0322)
R2 = 0.760
List of Variables:
(1) BR
Crude Birth Rate
(2) TFR (unadj) = Total Fertility Rate
unadjusted
(3) TFR (adj)
= Total Fertility Rateadjusted
(4) PMF
Proportion of Married
Females
(5) CMR
Child Mortality Rate
(6) FLR
Female Literacy Rate
(7) CP
Couples Protected
(8) IMR
Infant Mortality
At the state-level analysis the contribution
oj_ couple protection rate to the birtHTate
proved to be insignificant.' The reasons for
this are explained later. But, infant mortality and literacy rates and age at marriage appcared important in explaining the inter-state
differences in fertility rates. However, when
we used all three variables in the same equa
tion, the coefficients were not significant due
to multicollinearity. There is alsq_a_high
degree of correlation betweeiTmfant rnortality and temaJe3jteracy„rates..which is ap
parent from the second regression. However,
the Spearman rank correlation coefficients
demonstrate without any doubt the interrela
tionships among all these variables.
We tested whether these relationships
were equally significant district level.
We did not face any problem of multi
collinearity at this level of disaggrega
tion because the total number of observa
tions exceeded 400 for each variable. This
is the only year for which district-wise data
for these variables are available. In these
equations we have used child mortality at
age 2 instead of infant mortality as the ex
planatory variable. These relationships ap
pear even more robust with district-wise
data. Therefore, these findings provide
strong support to the analytical and policy
linkages between fertility levels, infant mor
2482
tality and female education. The results of But, such access was out of reach of most
the regressions prove unmistakably the im of the rural populations of all states except
portance of age at marriage (for which the Kerala, Goa and possibly Punjab. Less than
proportion married women in total women 20 per cent of hospital bed facilities were
is used), infant mortality rate and female located in the rural areas. Timely medical
literacy rate in reducing the birth rate. Female attention was unavailable to deal with com
literacy rate is a proxy to represent the status plications in child births. In many situations
of women and women’s empowerment in births were supervised by untrained birth at
household and in society.
tendants. Further, poor hygiene, absence of
In the absence of the preconditions neces immunisations and lack .of proper medical
sary to achieve a decline in birth rate, one attention at complicated child births were
should not be surprised at all if birth rates factors that led to increased mortali
continued to remain high or even rose in ty. The approach to the delivery of health
some states. Nearly 45 years after indepen care in rural areas is mechanical and bureau
dence, our record on improving female lite cratic and fails to take into account the needs
racy and reducing infant mortality was in of the population.
deed dismal. According to the 1991 census,
The story is not altogether different when
there were only eight out of 25 states in it concerns the extension of education to the
India, namely, Goa, Himachal Pradesh, rural population, especially among female
Kerala, Maharashtra, Mizoram, Nagaland, children. One gets the impression that there
Tamil Nadu and Tripura, where female is a deliberate attempt to keep the popula
literacy was 50 per cent or higher. (Literacy tion ignorant and backward for the purpose
rate is applied to populations above age 7.) of political, social and economic exploitation.
In the B1MARU states, it ranged between
20 and 28 per cent. The highest female
Approach io Family Planning
literacy rates were in Goa and Kerala, 68 and
India
was the first country in the world
87 per cent respectively. When we considered
the literacy rate separately for the rural sec to declare an official population policy as
early as 1952.- However, this policy was
tor then the corresponding figures were even
much lower. They would lie below 20 per translated into a pure family planning pro
cent in the B1MARU states and below 30 per gramme and was not integrated with health
cent in other states [Registrar General and education. At the same time, the family
1989a]. But such rural-urban differentials planning programme was implemented
were non-existent in Goa and Kerala. Even through the health department. Therefore,
within the backward states there were fur it came to be dominated by a medical ap
ther wide regional and inter-district varia proach. However, the amount of resources
devoted to family planning was a paltry sum
tions in female literacy rates.
Our record in reducing infant mortality during the first 25 years of planning. During
1956-80, the total outlay on family planning
is equally disappointing. During the year
and family welfare amounted to only
1992, 2.40 million children would have died
Rs 1,020 crore Against this, the outlay
before they reached their first birthday.
Infant mortality rate was still close to or ex during the Sixth Plan alone amounted to
nearly Rs 1,500 crore and since has been rais
ceeded 100 per 1,000 live births in the rural
sector in Assam, Bihar, Gujarat, Madhya ed substantially (Planning Commission,
1985). But, the declines in birth rates are pro
Pradesh, Orissa, Rajasthan and Uttar
bably only weakly correlated with the rise
Pradesh [Registrar General, June 1991].
in expenditure In fact, during the Sixth Plan
These were also the states where birth rates
were the highest. In fact, infant mortality though expenditure rose birth rate hardly
rate in rural Uttar Pradesh, which was still showed any decline.
the highest among the states, had remained
Part of the blame for the poor perfor
unchanged between 1970 and 1984. It began mance of the family planning programme
to decline only recently. It is a matter of lies in our approach. Basically, this is to
great concern that infant mortality con specify a target for the number of couples
tinued to remain so high, in spite of the to be brought under protection during any
operation of a number of special schemes plan period which is further broken down
such as the Integrated Child Development into annual targets. For instance during the
Scheme.
Sixth Plan 17 million sterilisations were car
It is not difficult to trace the factors that ried out against a target of 24 million. The
lie behind the continued prevalence of high Seventh Five-Year Plan document stated that’
infant mortality rate. We had indicated the effective couple protection achieved by
earlier that access to maternal and child care March 1985 with the above performance is
were vital for reducing infant mortality rates. of the order of 32 per cent which means that
Table 3: Spearman Rank Correlation Matrix (State wise Data)
PMF
FLR
CMR
PMF
FLR
CMR
BR
TFR
1.00
-0.78"*
1.00
0.75"*
-0.601.00
0.57*
-0.72"
0.59
0.66"
-0.67"
0.76"*
Significance Level: • 5 per cent, " :1 per cent,
•01 per cent.
Economic and Political Weekly
November 14, 1992
the effective couple protection has been rais
ed by 10 percentage points, i e. from 22 to
32 per cent but it is still below the Sixth Plan
target of 36.6 per cent [Planning Commis
sion, 1985]. But the document fails to raise
the question why, in spite of an increase in
jouple protection rate (CPR) by 10 points,
the birth rate remained sticks around the
figure of 34? But the Eighth Five-Year Plan
document is very frank in admitting the root
cause. Il states: While the Seventh Plan
targets of achieving CPR qf_42 per cenTwas
achieved, this was not matched by ^com
mensurate decline in the birth rate, possibly
because of improper selection of the cases
[Planning Commission, 1992: 332]?———
'TVrTexamination of the age distribution of
the wives of acceptors of family planning in
dicates why the couples protected are im
properly chosen. During 1982-83, wives
above age 30 formed 64 per cent of vasec
tomy acceptors, and 53.5 per cent of those
who underwent tubectomy were also over 30
years. Thus the mean age of wives of vasec
tomy acceptors in 1982-83 was 32.1 years and
of those who underwent tubectomy 30.8
[Ministry of Health and Family Welfare
1984]. When these ages are compared with
the mean age at marriage in different states,
the reasons for the above anomaly becomes
apparent. Mean age at marriage is below 20
years in most states and it is especially low
in those states with birth rates above 35 per
1000 population. Therefore, most married
women would have been nearer the end of
their child-bearing period by the time they
received family planning protection. This is
corroborated by the poor correlations ob
tained in our regressions,between birth rates
and the per cent of eligible couples protected
in different states.
The approach to family planning also suf
fers from certain inherent limitations arising
from its heavy dependence on medical infra
structure. This dependence is due to the fact
that the programme’s major emphasis has
always been sterilisation. During the period
1967-73, nearly 85 per cent of total sterilisa
tions were vasectomies, but by 1983-84
tubectomies accounted for a similar share
[Ministry of Health and Family Welfare,
1984]. Also sterilisation accounted for 80 per
cent of all effective protection of couples.
Paucity of health care facilities and the con
sequent low proportion of child births in
medical institutions made sterilisation highly
undependable and impractical as a major in
strument to bring down birth rates in rural
areas (Table 2). Besides, sterilisation being
a terminal method, young couples were un
willing fo adopt it especially in those states
where infant mortality continued to remain
high. These factors explain why family plan
ning has failed to make any headway in most
states in India, except Goa and Kerala. For
instance, 80 per cent or more of rural child
births were under institutional care and the
infant mortality rate had already fallen to
20 or below per 1,000 live births [Bureau of
Economics and Statistics, Kerala, 1992]. The
proportion of pregnant mothers immunised
in Goa also indicates how institutional care
Economic and Political Weekly
of pregnant women and new born children
is important not only in reducing mortality
but also in providing access to family plan
ning [Gupta and Indira Murali, 1989].
9
Outlook for the Future
The United Nations latest assessment
(1990 revision) predicts that India's popula
tion would be between 1294 million and 1567
million in the year 2025 [United Nations,
1991:208]. If we accepted their medium pro
jection, then the projected population would
be 1,442 million in that year [United Na
tions, 1991:208]. The medium projection
puts India's population at 1,041 million in
the year 2000. The latest UN projections
have revised upwards their earlier population
figures for India because of the slow rate ot
decline in fertility rates. Apart from a great
deal of rhetoric there was very little realisa
tion what such increases in population im
plied for the prospects of future development
of India. WhHe we could do very little now
to change this picture insofar as it concern
ed the next decade or so, the nature of the
action that we take now would determine the
size of the ultimate stable population. The
earlier we reach a net reproduction rate of
unity, the smaller would be the size of the
ultimate stable population.
A projection exercise undertaken by
Pathak and Ram indicates that the size of
the ultimate stable population on the basis
of 1981 population would be approximately
1,507 million if the net reproduction rate of
unity was reached 30 years beyond 1981. But,
if this occurred only 40 years after 1981, the
ultimate stationary population would have
risen to 1,680 million [Pathak and Ram,
1985]. This is only for the purpose of
illustrating the consequences of lengthening
the period to reach the net reproduction rate
of unity. A delay of 10 years would add to
the population as much as between 170 and
200 million. Another projection undertaken
in 1984 indicated that India’s population
would stabilise at about 1,700 million by the
year 2,155 [Pathak and Ram 1985]. It is very
likely that this prediction might come close
to reality. Whatever action we might initiate
from now on to reduce the birth rate, it is
unlikely to reduce the size of the ultimate
stationary population below 1,700 million.
One could predict with a fair degree of con
fidence that when India’s population stabi
lised it would have certainly more than dou
ble the 1991 population. There is no doubt
that our approach to family planning has
failed to bring down the birth rates signi
ficantly. A careful analysis would show that
whatever little decline occurred in birth rates
during the past twenty years might be attri
buted to urbanisation and a marginal rise
in the effective age at marriage.
What should be the approach in the
Eighth Five Year Plan toYamily planning m
order to hasten the decline in net reproduc
tion rate? Our analysis of the performance
of family planning clearly demonstrates that
the strategy followed till now appears inade
quate to achieve this objective. The present
practice of laying down exogenously a target
number of couples to be protected through
family planning should be abandoned. In
stead, the strategy should take into account
the interrelationships between female educa
tion, infant mortality and birth rates. The
parameters of these interrelationships might
be different for different regions depending
upon the availability of health care and the
level of female literacy. While access to and
the means to limit the number of children
should be provided for everyone who desires
the same, such-provisions would come to
naught in situations of high infant mortality
and low levels of literacy.
This new approach would require the endogenisation of the demographic parameters
in the mathematical framework for plan
ning. We are now in the midst of a major
re-thinking on the adequacy and strategy of
our planning process. In the past, we had
attached overwhelming importance to plan
ning of (he commodity producing sectors at
the expense of social sectors. The develop
ment of human resources is closely related
to the improvements in social sectors. It is
now generally recognised that the quality of
life can be substantially improved with the
development of social sectors and this can
be achieved even at comparatively low in
come levels. Therefore the planning
framework in India should develop a disag
gregated socio-economic- demographic
framework incorporating female education,
female employment, utilisation of health
care, levels of infant mortality and other
social variables for the endogenous deter
mination of the number of couples to be
protected in different age-groups to achieve
different rates of population growth. The
building blocks of such a socio-economicdemographic model should be the states
Table 4: Regression Results with District Data
FLR
R2
F-Value
-0.0222
(-2.029)
0.0080
(3.3865)
0.0349 -10.3891
(5.9086) (-5.8041)
-1.0567
0.0108
(8.4414) (-2.7269)
0.31
58.66
TFR(unadj)
35.928
(32.185)
2.4485
(10.13)
0.44
104.83
TFR(adj)
5.0087 -0.00302
(19.946) ( 1.2274)
0.00695
-2.4835
(5.2303) (-6.1679)
0.32
61.99
Dependent
Constant
PMF
BR
CMR
Notes'. Figures in brackets show t-values. TFR(adj) for Sample Registration estimates may underor over-estimate the rates in some cases. The equation with TFR(unadj) provides better
results.
November 14, 1992
2483
which alone would be able to take care of
the differences in social, economic and
cultural factors prevailing within their
region. These shifts in the objectives of plan
ning would also require major changes in the
institutional structure of planning. Social
planning also implies greater emphasis on
the spatial foments of planning. This would
not be possible under the present centralised
control of the planning process. Social plan
ning would become feasible and would suc
ceed only under decentralised planning with
greater popular participation at local levels.
The endogenisation of demographic
parameters as suggested here would enable
the explicit recognition of the complemen
tarities between female education, health
development and birth rates. This would also
help to develop some norms regarding the
allocation of expenditure for family plan
ning. For instance, in a state like Uttar
Pradesh, for any given amount of expen
diture, if the decline in birth rate would be
larger if it is spent on education and health
rather than on family planning, then the
allocation of expenditure should follow ac
cordingly. In such situations, the policy must
permit a larger allocation of resources
for education and health than to family
planning.
This section was meant to raise a few
issues relating population policy to planning
and to suggest the broad contours of an
alternative integrated approach. The issue of
population growth has not yet received the
singular attention it deserves among econo
mists. The government also appears unable
to articulate and execute an effective family
planning programme because it was ignored
the interrelationships between fertility, mor
tality, education and health. The potential
effect of a large population of around 1,700
million on the standard of living and on the
demand for resources cannot be ignored.
The problems of providing adequate food,
the question of employment generation for
a growing labour force the probable enlarge
ment of population dependent on poverty
alleviation programmes, and the increasing
difficulty of raising resources to meet the
requirements of educating and keeping the
population healthy are the challenges of the
future.
The implications of a large and growing
population for India’s development can be
discerned only if we take a long-term
perspective on planning. Unfortunately,
Indian planning is increasingly becoming
myopic, and short-term expediency is often
substituted for long-term objectives. The im
pact of population growth is nol immediate
ly fell on the economy unlike the effect of,
say, a decline in foodgrains output, and
therefore there are neither political nor
economic compulsions to institute counter
measures urgently. We are now,in the pro
cess of a major redefinition of the goals,
guidelines and practices of planning. Plan
ning till now placed overwhelming impor
tance on commodity producing sectors to
the detriment of social sectors. With the
2484
liberalisation of the economy and the
transfer of administrative control of the
commodity producing sectors to the regula
tion of the market mechanism, the govern
ment and the Planning Commission would
have the time and energy to concentrate on
the development of the social sectors. The
greatest external economies lie in the
development of human resources and it is
the quality of human resources that
ultimately determines the pace and pattern
of the economic, social and cultural develop
ment of a nation. I hope that we realise the
significance of this factor and shift the
priorities of public planning to the develop
ment of social sectors and leave the rest of
the economy to function under a planned
market economy.
A large population, soon exceeding 1
billion, would affect decisively our strategy,
policy and the pace of development of our
society and economy. During the past 45
years, we have managed to raise the supply
of food pan passu with the growth in
population, but this may require much
greater effort in the future. The potential to
raise productivity in agriculture may still be
high, but it might be achieved only by ap
plying larger and larger doses of chemical
fertilisers, extension of area under irrigation
and at higher costs and prices. The applica
tion of larger doses of fertilisers regularly
might also create long-term environmental
hazards. Food-population relationships,
hence, may continue to be critical requiring
constant attention. What measures are
needed to raise food output in the future?
What are the regional dimensions of the
food- population relationship? What is the
role of public action in ensuring the
availability of food? Some of these questions
need discussion and I shall take up a few
aspects of these relationships in the next sec
tion.
A large population also implies a large
and growing labour force. Will growth in
employment keep pace with the increase in
labour force? Where will employment
generation take place? What has been the
past trends in wage rate? What is the scope
for increasing real wage rate? Some of these
questions will be taken up for discussion in
the final section.
II
Poverty, Food and Public
Distribution
Poverty is multi-dimensional, though
some facets of poverty may be more critical
than others. Illiteracy, low income, un
employment, malnutrition, frequent il
lnesses, high infant and child mortality and
lower life expectancy are attributes of
poverty. High birth rates also co-exist with
poverty though it is not high birth rates that
breed poverty but it is poverty that breeds
more children. Not only survival rates of
children are poor in an environment of
poverty but the very survival of the family
may need more children to supply labour.
Under such circumstances children are con
sidered assets and the benefits derived by the
family by having a larger number of children
outweigh their costs. It is also true that when
infant mortality is high one needs more
births for the survival of a desired number
of children. The World Fertility Survey
showed that the desired number of children
rose monotonically witfy higher infant mor
tality and lower rates of female education
[Lightbournc and Macdonald, 1982].
Therefore the mutual links between popula
tion and poverty are many and complicated.
Food is an important link between
population and poverty. In the Indian con
text, the measurement of poverty was closely
related to food intake. Historically, popula
tion as a pre-eminent issue began with a
discussion on the relation between popula
tion growth and food supply. Malthus wrote
the theory of population to demonstrate the
consequences of population growth; how
food production would fail to keep pace
with the increase in population and thus
presented a first study relating population,
food and poverty. Though we do not
subscribe to a simplistic interpretation of this
relationship, there are many aspects of the
food-population relationship which are im
portant and require deeper analysis in the
Indian context. In the Indian situation,
though aggregate food supply has kept pace
with the growth of population, we have been
invariably treading a delicate balance. We
avoided major catastrophes in the post
Independence period similar to the Bengal
famine of 1942 because we have now in place
a food management system which has pro
ved reasonably successful in dealing with the
distribution of the food in the country.
I do not propose to enter into either a
discussion on or even attempt a measure of
the extent of poverty in India. A large
number of studies on measurement of
poverty exist in India [Cf Kakwani and Rao
1990]. These studies also demonstrate that
such estimates of poverty are highly sensitive
to the assumptions underlying their esti
mates. We ail agree (hat various manifesta
tions of poverty with different degrees of
intensity exist in all parts of our country. A
few of these manifestations may even appear
in conflict with each other. For instance,
states which show high levels of calorie
intake also report high infant mortality, thus
blurring the relationships between malnutri
tion and mortality. Similarly, states which
have higher per capita income do not
necessarily have high cereals consumption.
On the other hand, Kerala state which
reported low per capita state income as well
as low calorie intake reported one of the
lowest infant mortality rates and the longest
expectation of life. All these indicate that
poverty is multi-faceted and each of these
attributes merits discussion by itself for an
understanding of the nature of or lack of
interrelationships among various micro
economic variables and to suggest ap
propriate policy measures.
Economic and Political Weekly
November 14, 1992
Figure 2: Inter-State Variation in Per Capita Cereal. Production
1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986
Year
Figure 3: Inter-State Variation in Per Capita Cereal Production
showed significant variations among dif
ferent states. The coefficient of variation in
per capita cereal production rose from 0.54
in 1970-71 to 0.84 in 1988-89. (Figure 2). We
estimated the compound rates of growth in
per capita cereals production for the period
1961 to 1989 for each state separately. These
were estimated by fitting semi-logarithmic
equations to the detrended per capita pro
duction data by taking threc-ycar moving
averages. The estimated growth rates are
given in Table 5 for three separate time
pcriods.During the post-green revolution
period, per capita cereal output really rose
only in Haryana, Punjab and Uttar Pradesh.
Per capita output more or less remained con
stant in Andhra Pradesh, Madhya Pradesh,
Maharashtra and West Bengal and declined
in all other '’ales. If we considered only the
marketed surplus the regional disparities
would have been even greater as the sur
pluses were confined to a few states only.
However inter-state disparities in con
sumption appear to have declined marginally
during the same period as most states
managed to maintain either their level of per
capita consumption of cereals or in a few
cases even showed marginal improvements.
The coefficient of variation in per capita
consumption between states marginally
declined in the 80s compared to early 70s
and was around 12 per cent compared to
over 80 per cent in per capita production
(Figure 3) [Krishnaji 1988]. It was also
interesting to note that per capita consump
tion in states like Punjab, Haryana and Uttar
Pradesh did not rise in spite of significant
increases in per capita production. On the
other hand, in some cases it even declined.
Some of these changes probably reflected
changes in the composition of the food
basket arising from substitution between dif
ferent food groups brought about by higher
income and changes in taste patterns.
Inter-State Differences in
Foodgrains Consumption
I do not also propose here to enter into
a discussion on the merits or otherwise of
the theoretical aspects of the relationships
between population and food. On the other
hand, I shall concentrate on some neglected
aspects of the functioning of the Indian
foodgrains economy and how these might
be relevant for an understanding of the rela
tionship between food intake and poverty.
The discussion, inter alia, would also
indicate the implications of population
growth on food consumption.
Nature of the Foodgrains Economy
India’s foodgrains output rose from 59.2
million tonnes in 1953 to 170.25 million ton
nes in 1989 [Ministry of Agriculture, 1990).
This is an impressive achievement but it con
ceals an important aspect of reality, namely,
that per capita availability of foodgrains has
Economic and Political Weekly
An earlier study analysed the deter
minants of inter-state levels of foodgrains
practically remained stagnant during this consumption (in terms of calorie intake) for
period. In fact, between 1971 and 1989, per the year 1961-62 and came to the conclusion
capita availability was below that of 1971 for that it was not determined by the level of per
12 out of the 19 years [Ministry of capita state income but mainly by the level
Agriculture, 1990]. The per capita availabili of per capita state foodgrains output [United
ty of cereals other than of rice and wheat, Nations 1975]. States which had higher per
in fact, declined significantly during this capita incomes had lower per capita con
period, from 44 kilograms in 1971 to 30 sumption of foodgrains than states with
kilograms in 1989 or even below in some lower per capita incomes. There was no
years [Ministry of Agriculture, 1990). The guarantee that an increase in income would
total per capita availability has remained automatically raise the level of consumption
more or less constant largely due to the of foodgrains. The foodgrains sector of the
steady rise in wheat production. These Indian economy presented a clear case of
changes in the composition of foodgrains market failure [United Nations 1975]. It was
and the shift in the regional pattern of surmised that foodgrains did not move
production.have important implications for across the country through the normal
operations of the market mechanism to even
food intake and poverty.
out inter-state differences in per capita pro
During the past 30 years, the regional duction and thus reduce the disparities in
disparity in per capita production widened consumption.
This aspect of market failure in foodgrams
because growth rates of foodgrains output
November 14, 1992
2485
is closely related to poverty and income
distribution. It was found that the Spearman
rank correlation between cereal price and per
capita cereal production v>as high and
positive implying that cereal price was higher
in those states which had lower per capita
cereal production and vice versa. The im
plicit prices derived from the NSS value and
quantity of consumption data indicated the
extent of variation in prices. For instance,
in 1973-74 the price of cereals at 1970-71
prices was Rs 1.68 in Kerala compared to
Re 0.77 in Punjab; only part of the difference
could be explained by the differences in the
proportion of wheat in total consumption
in the respective slates [NSS 1981 ]. Inter-state
movement of foodgrains was mainly geared
to meet urban demand where income levels
were high enough to afford to pay higher
prices. Foodgrains from the surplus states
failed to move to the rural areas of low per
capita production slates because prices were
not high enough to make it worthwhile for
the trade to move grain to these areas.
Though income elasticities for grain among
the poorer income groups would be high, so
too would be their price elasticities. Thus as
price of foodgrains rose, there would be
a steep decline in the demand for grains
among the poorer income groups pushing
down their consumption even further. In the
case of higher income groups both income
and price elasticities would be low- and thus
market failure would not materially affect
their level of consumption. Market failure,
however, was probably a blessing for the
poor in the high per capita output states
because part of the marketed surplus which
was not transferred from that state would
be available at a lower price for their con
sumption. Thus the poor would be able to
consume more grain in the high per capita
foodgrains output states, though the op
posite would be the situation in the case of
the poor in the low per capita production
states.
We have data now to verify whether the
relationships between food intake, per capita
income and per capita food production have
changed since 1961-62. Such data are
available from the National Sample Survey
for the years 1973-74, 1983, 1986-87 and
1988-89. Since calorie intake data are not
available for ail the years per capita quantity
of cereals consumed is used instead.
Therefore we have used per capita quantity
of cereal consumption in the estimation.
Among the determinants of consumption in
1961-62 we had found that the degree of in
equality in land distribution also was an
important factor. However, due to the non
availability of information on land distribu
tion for these latter years, we did not include
this variable in the present analysis. We
experimented with various alternative
specifications for the regressions and found
that the semi-logarithmic form piOvM^d (he
best fit. Not only were the coefficients under
this specification turned out to be most
significant but also the assumptions underly
ing it were more consistent with the theory
of consumption.
Our analysis indicated that per capita
foodgrains consumption was still positive
ly correlated with foodgrains output and
negatively correlated with per capita state
income. While it was found that the
consumption-income relationship was
somewhat weak in 1961-62, it depicted strong
negative relationships in the equations for
later years. It was surprising that these rela
tionships had not changed at all in spite of
the fact that there were no zonal restrictions
on the movement of foodgrains in the lat
ter half of the 80s. There are reasons to
believe that private trade in grains should
have been more active in the 80s because of
the virtual removal of compulsory procure
ment in most of the deficit states. During
this period all the marketed surpluses in
these states would have shifted to private
trade. Secondly, with the growth of black
money, private financial credit might have
also found a good haven in grain trade. We
also note that while in all equations there
was a negative relation between price and
quantity consumed it was not found to be
statistically significant. This lends further
support to the view that price differentials
had no influence on the quantum of inter
state movement in foodgrains. We attribute
this largely to the fact that the food
population balance continued to be the same
throughout this period.
Our analysis indicated that there had not
occurred any significant or important
change in the structure of the grain market
during the period 1961 to 1989. On this ac
count, we would hypothesise that whatever
grain movement that took place under
private trade, it would have only catered to
the requirements of the urban population.
Further, private trade was unlikely to have
moved any significant quantity of grain from
the surplus stales to the far off deficit states.
Poverty and Food Intake
The National Sample Survey data also
provide information on quantity of con
sumption of foodgrains classified according
to different expenditure classes. Based on
this information, we estimated the quantity
of.cereals consumed by the bottom 30 per
cent of the rural population in each state for
the years 1961-62, 1973-74, 1983 and
1986-87. This was estimated in two stages.
First, we estimated food demand functions
by regressing the quantity of cereals
consumed by each expenditure class and the
corresponding total expenditure. Second, we
fitted lognormal distributions to the total
expenditure data for each year and for each
state and estimated the mean expenditure
corresponding to the bottom 30 per cent of
the population. Then the quantity of cereals
consumed by the bottom 30 per cent of the
population was estimated from the first
Table 5. Rate of Growth of Per Capita Cereal Production
States
1
Andhra
Assam
Bihar
Gujarat
Haryana
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
India
1960-61 to 1988-89
Coefficient
t-Values
in Per Cent
2
3
-0.06
-1.20
-0.22
-0.50
2.80
-0.18
-1.65
-0.21
0.15
-1.00
4.30
-0.38
-0.92
2.10
0.14
0.72
-0.35
-16.96
-1.06
-1.07
10.03
0.02
-9.37
-1.00
0.34
-7.14
15.60
-1.71
-4.93
14.03
0.70
6.96
R2
4
0.004
0.92
0.04
0.04
0.83
-0.85
0.78
0.77
0.03
0.00
0.67
0.91
0.10
0.49
0.88
0.006
1960-61 to 1968-89
l-Values
Coefficient
in Per Cent
5
6
-1.67
-0.35
-1.30
0.45
na
-0.31
-0.42
-2.90
-2.80
-0.82
0.63
-1.38
-1.82
1.50
0.15
-0.73
-5.27
-0.96
-0.80
0.57
na
-0.27
-0.63
-2.00
-2.18
-0.91
0.24
-2.16
-5.11
1.55
0.19
-0.08
R2
7
0.82
0.13
0.09
0.05
na
0.01
0.06
0.40
0.44
0.12
0.009
0.43
0.81
0.29
0.006
0.06
1969-70 to 1988-89
t-Values
Coefficient
in Per Cent
8
9
0.23
-0.12
-1.84
-1.90
2.30
-1.30
-2.60
0.44
1.21
-0.91
3.60
-1.03
-1.51
2.10
-0.46
-0.46
0.71
-10.62
-3.14
-2.41
7.66
-6.30
-17.82
0.13
1.44
-3.52
18.62
-2.63
-4.69
7.88
-1.26
3.52
R2
10
0.03
0.86
0.36
0.25
0.77
0.70
0.95
0.001
0.10
0.42
0.95
0.28
0.56
0.79
0.86
0.42
Source: Bulletin on Food Statistics.
2486
Economic and Political Weekly
November 14, 1992
equation corresponding to the mean total
expenditure of the bottom 30 per cent of the
population (Table 7). These estimates thus
provide not only the quantity of cereals con
sumed but also the mean expenditure of the
bottom 30 per cent of population.
The estimated total consumer expenditure
of the bottom 30 per cent of the population
rose between 1961-62 and 1988-89 but the
rates of increase varied considerably
(Table 7). Since the bottom 30 per cent in
most of the states would be labour house
holds, the improvements in real consumer
expenditure probably also reflected some im
provements in real wage rates. However the
improvements in real consumer expenditure
did not appear to have led to an improve
ment in the quantity of grain consumption.
in fact, per capita per month quantity of
grain consumption was substantially lower
in almost all states including the grain
surplus states in 1973-74 and in 1986-87
compared to that in 1961-62. The per capita
consumption of grains derived from the Na
tional Sample Survey estimates, total as well
as the figures for the bottom 30 per cent of
the population reflected the general trends
in per capita availability estimated from pro
duction statistics. The share of grain con
sumption of the bottom 30 per cent of the
rural population more or less remained
unchanged between 1973-74 and 1986-87
though it had declined drastically from
1961-62 (Table 7). We have not examined the
factors behind this decline and such factors
may vary between states. In states like
Andhra Pradesh and Punjab these changes
possibly reflected the increasing commer
cialisation of agriculture These calculations
of consumption of grains by the bottom 30
per cent of the population further reinforce
our earlier findings that the inter-state levels
of per capita production basically determin
ed the levels of consumption. Not only that.
When foodgrains output declined the con
sumption of the bottom 30 per cent of the
population possibly declined by a larger pro
portion. These findings provide strong
evidence why measures of poverty using dif
ferent criteria could be misleading and in
dicate contradictory’ and conflicting changes
in the levels of poverty. While it is very likely
that the proportion of the poor would have
declined if consumer expenditure was taken
as a norm, it would not be so when we con
sidered a calorie intake norm as a cut-off
point for measuring poverty. In fact there
is no a priori reason to believe that poverty
under a calorie norm would have at all
declined in India between 1961 and 1989. On
the contrary, even on the assumption of a
constant proportion of the poor in the total
population, the absolute number of the poor
or hungry in the country would have certain
ly risen. The decline in the proportion of
poor as estimated from expenditure data by
the Planning Commission may thus be a
misleading finding.
The variances of the lognormal distribu
tion fitted to the total consumer expenditure
Economic and Political W'eekly
of each state would indicate the changes in
inequality in expenditure distribution. They
show that inequality in expenditure ( and
income too) rose in most states between
1973-74 and 1988-89.
from the mid-60s which was built on highyielding varieties of wheat with high levels
of fertiliser and waler inputs. While per
capita production of wheat rose, per capita
production of grains like jowar, bajra and
other dryland crops in fact declined. The
Role of Public Distribution
decline in the output of these grains not only
had regional implications but also impor
We have already mentioned that in spite tant implications for the poor in those states
of the widening disparity in per capita food where they were the major crops. These in
production, per capita consumption had ferior grains normally commanded lower ab
either remained steady or slightly improved solute prices compared to those of rice and
among the states in India. During the past wheat. The decline in their per capita
30 years we have managed to avoid major availability, therefore implied that the poor
food crises; till recently, we also succeeded would be cither required to reduce their total
in keeping the foodgrains prices within consumption of grains or be forced to spend
reasonable limits. How did we manage these more due to higher prices of the superior
results even though per capita availability grains, namely, of rice and wheat. These
had hardly risen and inter-state disparities
shifts also implied changes in relative prices
in output had risen?
within the cereals group which would have
We achieved these results by putting into
also adversely affected the consumption of
place a food management system which with
the poor.
all its limitations had enabled us to overcome
Along with the new production strategy,
to a large extent the results of market failure.
the government also established the Agri
The food management system was basically
cultural Prices Commission to determine
built on three blocks:
support prices. Initially, the support prices
(1) A production strategy which broadly
were meant to protect the farmers from
enabled food output to keep pace with
declines in grain prices in the immediate post
population growth; this strategy, however,
harvest period but over the years, with the
had resulted in greater inter-regional concentration of marketed surplus in three
disparities and in inter-crop shifts in food or four states, these support prices were
production;
gradually converted into government pro
(2) A food procurement programme with curement prices. As the dependence of the
support prices as a major element of the public distribution system on procurement
strategy; and,
increased, the farmers in the grain surplus
(3) A public distribution system which states were able to put pressure on the
moved grains from the surplus production government to raise procurement prices con
states to the deficit consumer states. This tinuously. As part of the production strategy,
transfer of grain not only reduced the inter farm inputs, especially fertilisers, were also
state disparities in per capita consumption provided to the farmers at subsidised prices.
but also provided grain through ration shops All these changes resulted in a situation
or fair price shops at subsidised prices.
where the procurement price acted as a
We already mentioned that inter-state secular push factor in rising foodgrain
disparities in per capita production had prices.
With the widening gap between the pro
increased between 1961 and 1989 but ag
gregate availability had more or less remain curement and issue prices of,cereals supplied
ed steady. The inter-state disparities had through the publicdistribution system, the
risen because the increase in output was amount of food subsidy also rose over these
largely due to the increase in wheat output years. Such subsidy not only enabled the
and wheat production was concentrated consumers to obtain part of their consump
mainly in three stajes, namely, Haryana,
tion at lower prices but also had probably
Punjab and Uttar Pradesh. This was a direct other desirable macro-economic impact,
result of the production strategy adopted namely, such as acting as a damper on the
Table 6: Regression Results
Year
Constant
Per Capita
SDP
Per Capita
Cereal
Production
Cereal
Price
R2
1973-74
411.36
(4.4373)
489.28
(5.3242)
454.10
(6.0049)
543.10
(4.6592)
-67.22’*
(-4.460)
-65.76**
(-3.939)
-63.96**
(-4.8926)
-58.71**
(-3.6609)
-27.27
37.63**
(-1.074)
(2.8270)
-24.9154
20.98*
(-0.987)
(2.342)
-10.00
24.82* *
(-0.422)
(3.933)
15.95* -29.2572
(2.489)
(-1.075)
0.80
1983
1986-87
1988-89
0.61
0.72
0.57
Significance Level: * = 5 per cent, •• - 1 per cent.
Dependent Variable: Annual Per Capita Cereal Consumption in Kilograms.
November 14, 1992
2487
rale of inflation.
The quantity of foodgrains distributed
through the public distribution system has
increased four and a half times, from 4
million tonnes in 1961 to 18 million tonnes
in 1988. However, the ratio of public
distribution to total production has remain
ed within a narrow range, around 10 to 13
per cent, between 1971 and 1989 (Table 8).
The source of grain for public distribu
tion, however, had undergone a major shift.
In 1961. nearly 80 per cent of the require
ment for public distribution was met
through PL480 import of grains but in 1980s
imports were rarely resorted to and almost
100 per cent of the requirements for public
distribution came out of domestic grain pro
duction. Interestingly, the proportion of
total foodgrains procured also had remained
constant during the period 1961 to 1989.
around 9 to 12 per cent (Table 8). However,
there were some important shifts in the
quantum of procurement in different states.
In the mid-70s, procurement was undertaken
even in deficit states, by enforcing a levy
either on millers or on large farmers. But
with the increase in grain procurement from
the northern slates, the procurement of
grains in most other states either declined
or were eliminated completely in the 80s
(Table 9).
The concentration of procurement in three
or four states also seemed to have promoted
a higher degree of commercialisation of
grain production in these states. The level
of per capita consumption of foodgrains ap
peared to have either remained stagnant or
declined in these stales while the proportion
of output procured rose continuously. This
process converted foodgrains production
from a subsistence crop to a commercial
crop. As a result, the proponion of marketed
surplus had risen in the grain surplus states
and the availability of grain in the rural areas
of those states might have declined,thus af
fecting the levels of consumption of landless
labourers.
/Xs already mentioned, the inter-state in
equality in per capita consumption of cereals
had probably declined in the 80s compared
to earlier periods. A similar change probably
had also occurred in the rural-urban ratio
of public distribution. In the 60s and early
70s, public distribution was probably urbanbiased but in the 80s a distinct change ap
pears to have taken place. Mahendra Dev
and Suryanarayana analysed the public
distribution data collected by the National
Sample Survey for the year 1986-87 and
found that the criticism that the PDS was
uiban-based was no longer correct
[Mahendra Dev and Suryanarayana 1991],
They adopted four different cinGia to
evaluate the role of public distribution,
namely, (I) rural sector’s share in total PDS
purchase, (2) relative dependence on the PDS
given by ratio of PDS purchase to total pur
chases, (3) PDS quantity purchased per
capita and (4) PDS quantity purchased per
market dependent. While the nature of bias
varied depending on the criterion used, it
was generally found that the PDS was
generally rural-biased for foodgrains.
However, this conclusion needed some
modifications when applied at the state level.
In those states where PDS was weak, the
coverage of the PDS in the rural area also
was found to be inadequate [Mahendra Dev
and Suryanarayana 1991] (Table 10).
Mahendra Dev and Suryanarayana also
examined the extent of dependence of the
poor in the rural areas on the PDS. They
found that the poor was able to secure only
less than 16 per cent of rice and wheat and
less than 5 per cent of coarse grains of their
requirements from the PDS. This situation
arose from two fads: one, procurement of
cereals other than rice and wheat were
negligible and second, the poorer income
groups consumed mainly other cereals since
their absolute prices were lower and were
unable to pay the higher prices of rice and
wheat for lheir cereal requiremenls. It was
found that the urban poor also depended on
the open market for a major share of their
needs. Another major consideration was
probably the fact that the poor did not often
have enough cash to buy their entitlement
of the ration whereas the private trader
granted them credit and naturally enabled
multiple purchases as and when needed. The
lower share of purchase of grains from the
PDS in some states might also reflect the
system of wage payment, namely, the
preponderance of payments in kind, which
used to be in terms of grains.
Foodgrain Prices and Inflation
Given the overwhelming dependence on
foodgrains and the prominence of inferior
cereals in calorie intake, the behavidur of
foodgrain prices would be critical for the
poor. First, there has been a secular decline
in the per capita production of cereals other
than of rice and wheat. This would especial
ly affect those states where the poor mainly
depended on the inferior cereals for their
consumption. Second, the prices of inferior
cereals and rice have risen relatively at a
higher rate than that of wheat. In infla
tionary situations, the prices of inferior
cereals rise by a much larger proportion than
those of rice and wheat. Fourth, the PDS
provided mostly rice and wheat and
therefore the food subsidy was confined to
these grains only. On all these accounts, dur
ing periods of inflation, the consump
tion of cereals by the poor would decline by
a larger proportion than that of other
income groups in the society. Radhakrishna
Table 7: Cereal ConSumphon and Expenditure by Bottom 30 Per Cent of the Population (Rural)
States
1
Andhra
Assam
Bihar
Gujarat
Haryana
Jammu and Kashmir
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
Ratio of Consumption
Total Amount Spent by Bottom
of Bottom 30 Per Cent to
30 Per Cent for Consumption
*
Mean Consumption_______ in Rs at 1960-61 Price
Quantity of Consumption
of Cereals by Bottom
30 Per Cent of the Population
(Monthly Per Capita)
1973-74
1986-87
1961-62
4
2
3
1961-62
5
1973-74
6
1986-87
7
1961-62
8
1973-74
9
1986-87
10
9.78
9.76
10.72
7.95
13.74
15.54
9.18
8.11
12.19
8.63
12.59
8.22
12.88
8.99
11.91
11.83
0.94
0.75
0.88
0.85
na
na
0.78
0.91
1.02
0.84
0.47
0.82
0.88
0.89
0.82
0.78
0.78
0.70
0.70
0.83
0.77
0.82
0.62
0.63
0.75
0.74
0.79
0.72
0.82
0.73
0.71
0.75
0.70
0.74
0.70
0.72
0.93
0.87
0.70
0.79
0.81
0.73
0.79
0.69
0.77
0.73
0.77
0.77
9.45
13.78
9.60
12.07
na
na
11.09
10.67
11.66
10.74
8.61
13.45
10.67
11.66
9.54
10.66
11.11
7.37
9.07
11.96
na
11.93
9.43
10.04
8.63
9.50
7.94
14.30
12.25
11.53
11.34
8.91
12.55
13.90
11.37
12.67
na
14.67
11.33
13.82
9.94
10.78
10.04
16.68
12.90
11.53
11.34
9.82
15.77
12.89
16.13
13.55
na
na
15.58
9.16
21.58
13.72
8.58
15.15
19.71
14.07
15.19
12.63
12.36
10.68
10.42
11,45
12.79
15.7
9.75
4.85
12.84
9.98
12.49
10.68
15.31
10.69
11.57
9.69
• We have used the Agricultural Labourer’s Cost of Living index to deflate the current expenditure Figures.
Source: Tables with Notes on Consumer Expenditure, NSSO Reports, various issues.
2488
Economic and Political Weekly
November 14, 1992
had shown in a raxni paper that the impact of
inflation on different income classes essentially
depended on the root cause of any given in
flation [Radhakrishna 1992]. If inflation was
mainly triggered by rise in foodgrain prices
then its impact on the poor was found to
be greater. On the other hand, if it was due
to increases «n the prices of industrial goods
then its impact on the poor would be the
minimum.
Food Subsidy and Public
Distribution
The public distribution system is to be
judged in the context of the analysis just
presented. In the absence of public distribu
tion inter-state inequalities in grain con
sumption (and also in total calorie intake)
would have been greater. Also, the inter-state
differential in foodgrain prices would have
been larger. In those states like Kerala where
public distribution system functions well,
it has helped the poor to obtain a significant
share of their grain consumption at lower
prices. Therefore, food subsidy performed
a socially desirable redistributive function.
PDS also helped to regulate open market
prices and, hence indirectly, the general level
of prices. Because of the overwhelming im
portance of foodgrains in the consumption
of the bulk of the population, stability in
foodgrain prices is vital to the stability of
our social and economic system. Therefore,
food subsidies should not be treated like
other subsidies. Food subsidy can help to
dampen the inflationary pressure, contribute
to the alleviation of poverty and reduce inter
state disparities in food consumption. All
these are desirable economic and social goals
and therefore food subsidy should be treated
as an essential item of social investment.
However, in view of '.he future size of the
population of the country and the require
ments of the public distribution system this
argument is sure to encounter increasing
resistance. Therefore, it is important that we
begin to explore alternative strategies for
foodgrains production and distribution
within our country.
turn of grain to be distributed and the
amount of food subsidy required might con
tinue to remain substantial. On the other
hand, if the food distribution was left to the
functioning of the market mechanism, as we
have indica’ed earlier, it would probably fail
to move vr.ough grains from the surplus to
the deficit dates at prices affordable by the
poor. Therefore this alternative would not
be desirable in the context of an affirmative
social policy.
The Politics of Food
The question of subsidy has become now
inextricably mixed up with the politics of
food in India. The quantum of food sub
sidy began to rise only in the 80s whic.h also
coincided with the structural changes in the
foodgrains economy. Nearly 80 per cent of
total procurement of foodgrains were from
the slates of Haryana, Punjab and Uttar
Even though the proportion of the poor Pradesh and therefore the farmers in these
measured variously either by the Planning states began to exercise monopsony power
Commission or by other scholars arc declin through political lobbying over government
ing, their absolute number might still be very procurement. This had resulted in con
high in view of the high rate of population tinuous upward revision of procurement
growth. Therefore, even if the PDS was prices which if passed on to the consumers
specifically targeted to the poor, the quan would have resulted not only in increases in
foodgrain prices but also in higher rates of
Table 8: Production, Procurement and Public Distribution of Foodgrains
inflation. As the dependence of the central
government on these states for procurement
Year
Production
Procurement
Public
Ratio of Ratio of Public
increased w-ith the grdwth of population as
(000 Tonnes) (000 Tonnes) Distribution Procurement Distribution
well as of urbanisation, the central govern
(000 Tonnes) to Production to Production
ment is bound to face increasing pressure to
(Col 3/Col 2) (Col 4/Col 2)
raise procurement prices in the future.
2
1
3
4
5
6
Another aspect of these changes in pro
curement was the virtual withdrawal of pro
1960
76672
4937
1275
0.017
0.064
curement from the large farmers in the food
1961
82018
541
3977
0.007
0.048
deficit states. This too had been done under
1962
82706
479
4365
0.006
0.053
political pressure from farmers on the
1963
80151
750
5178
0.009
0.065
respective state governments. This change
1964
80642
1430
8665
0.018
0.107
took place mainly in the 80s when with the
1965
89356
4031
10079
0.045
0.113
commercialisation of food production in the
1966
72347
4009
14085
0.055
0.195
wheat belt, enough grains became available
1967
74231
4462
0.060
0.177
13166
to meet the requirements of public distribu
1968
95052
6805
10221
0.072
0.108
tion system from the central pool.
1969
94013
6381
9385
0.068
0.100
1970
99501
An important impact of public distribu
6714
8841
0.067
0.089
1971
108422
8857
tion, which remains unnoticed, is the com
7816
0.082
0.072
1972
105168
7665
11396
0.073
0.108
placency that it has induced among the food
97026
1973
8424
11414
0 087
0.118
deficit states in implementing institutional
1974
104665
5645
0.054
10790
0.103
reforms and agricultural development
99826
1975
9543
11253
0.096
0.113
strategies to raise agricultural output. The
1976
121034
12853
9174
0.106
0.076
fact that the state governments could depend
1977
111167
9874
11729
0.089
0.106
on the central pool to obtain foodgrains to
126407
1978
11098
0.088
10183
0.081
meet their public distribution system seemed
1979
131902
13836
11663
0.105
0.088
to have acted as a kind of insurance against
1980
109701
11178
0.102
14993
0.137
food shortages in their respective states
1981
129588
12975
13014
0.100
0.100
which enabled them from taking hard deci
1982
133295
15419
14768
0.116
0.111
sions and drastic institutional measures.
129519
1983
15571
16206
0.120
0.125
The solution to the elimination of subsidy
1984
152374
18723
13326
0.123
0.087
is not import of grains for feeding the public
145539
1985
20116
15799
0.138
0.109
distribution system. Import of foodgrains
150440
1986
19720
17269
0.131
0.115
1987
would certainly lead to compromising
143418
15667
18700
0.109
0.130
1988
138414
national sovereignty and integrity as we had
14065
18306
0.102
0.132
already found out on earlier occasions. Food
Notes : Production, procurement and public distribution are in thousand tonnes. Procurement
as a powerful weapon to discipline third
and public distribution are for calendar years. Foodgrains production is for agricultural
world countries is part of the international
year.
politics of food-exporting countries. Food
Source: Col 2, India Data Base by H L Chandhok and Policy Group. Vol 2, Cols 3 and 4, Bulletin
security is as important as national security
on food Statistics 1987-89.
and food production should be accorded the
Economic and Political Weekly
November 14, 1992
2489
Mm^ibody ©®m mA®
EnrDonn®^7. ©onlh/ ® f®w
©[r®®^® wssjDUEh
3® ® [rwSgLhsQtb
§pg®fe ffoir '®rs
®QJ®®@S®a
Whether it was the
2490
6,00,000
tonnes
per
setting up of our first ever
annum; or setting up one
paper mill, J.K.Paper Mills
of the most modern audio
at Rayagada in Orissa, or
magnetic tape plants with
going in fora diversification
Japanese knowhow, we
with our Lakshmi Cement
have embarked on all our
Plant, which now produces
projects
with
Economic and Political Weekly
one
November 14, 1992
®JAYKAY
W
philosophy: citherdo it well
by the British as one of the
or not do it at all.
safest industrial units in the
As for the investors -
world. Not to mention the
institutions, public and
JK Magnetics plant, which
ourselves-thisattitude has,
rolls out world-class audio
by 306%. And the El’S from
tapes, right on schedule.
Rs. 5.24 to Rs. 17.47. Our
Straw Products Ltd.
In the past ten years
dividend record has been
As we said, making
alone, our turnover has
consistently growing, and
money could be anybody’s
moved up 294%, gross
today stands at 35%.
business. Creating wealth,
of
course,
brought
rewarding returns.
In capacity utilisation,
our cement plant operates
of course, is an entirely
different business.
at an enviable 110 percent.
In safety standards, The JK
Paper Mills has teen rated
profit by a phenomenal
478%, capital employed
(j) s
LAKSHMI
CEMENT
MAQNETICS
STRAW PRODUCTS LTD.
Registered Office : Jaykaypur-765 017 Rayagada (Orissa)
Administrative Office : Nehru I louse, 4 Bahadur Shah Zafar Marg, New Delhi -110 002.
rrrrrrrrr t c rr cr r rrrnxe rr rrr crcrr rcccc ma
Economic and Political Weekly
November 14, 1992
2491
same priority as defence. Besides, not only
are international prices higher but regular
imports would also require foreign exchange
earnings to pay for such imports. Food im
ports would only lead to further burden to
the balance of payments position.
There is a strong case to procure grains
from all large farmers at below market prices
in all states in India. Agriculture is not taxed
at all and in the circumstances procurement
at below market prices should be looked
upon as an indirect method of taxing
agriculture. The Agricultural Prices and
Costs Commission should lay down each
year targets for procurement of grains for
each state and the allocation of grain from
the central pool should be made on a pro
rata basis.
Finally there is another important reason
why continuous raising of procurement
prices is harmful to the development of
Indian agriculture. There is a close link
between procurement price and open market
price of grains. Whatever be the technical
relations between the two, open market price
is arrived at by adding a mark-up over the
procurement price. Thus the annual in
creases in procurement prices also lead to
annual increases in open market prices of
grains. But the most crucial effect of the
continuous increases in procurement prices
is its adverse impact on agricultural produc
tivity. Price increases which over-compensate
cost increases can discourage measures to
raise agricultural productivity since such
price rises automatically lead to higher pro
fits for the farmer. Once farmers get used
to this softer option to raise profits, they
resort to political action to obtain higher
prices for their products rather than resort
to the adoption of difficult innovations to
raise productivity and thus their profits. The
only guarantee to ensure the continuous
adoption of innovations to raise productivity
is to ensure stability in agricultural prices
rather than continuous increases in agri
cultural prices. While there is a case for the
greater play of price signals for industrial
products, there is a greater need to emphasise
non-pnee factors in agriculture as opposed
to price factors [Raj 1990].
Need for Alternative Food Strategy
The principal finding of our analysis is
foodgrains intake, and hence total food in
take, depends on level of foodgrains produc
tion in that state or region. Procurement and
food subsidy have become essential because
food surpluses are concentrated in two or
three states and are located far away from
consuming centres. The food production
strategy till now have failed to take into
account the distributional implications of
that strategy. The country would not be able
to afford the increases in the distribution
costs to feed the future population of this
country. In the light of our findings, it is sug
gested that we undertake a mapping of the
country into distinct regions within which
grain would move freely. These may be
2492
described as grain marketing regions. Il is
possible that the grain marketing region may
be co-terminus with agro-climatic regions.
This would certainly be an advantage but
not essential for our proposal. After identi
fying such regions, the potential for food
production in each of these regions should
be evaluated under varying assumptions
regarding technology and farm inputs. It is
possible that the potential for increases in
food production is not realised at present
because of institutional constraints rather
than of technological limitations. Wherever
institutional factors are "holding up increases
in agricultural productivity it is necessary to
initiate institutional reforms without delay.
Further, agricultural strategies should be
developed for each of these grain marketing
regions such that each of these becomes selfspfficient in its grain requirements.
The basic strategy that we propose is to
aim for regional self-sufficiency in food
grains as long as it is within the technological
and economic feasibilities of each region.
However, while evaluating economic feasi
bilities it is necessary to include all' the con
siderations mentioned in our analysis. Such
increases in foodgrains output not only
would raise overall consumption levels but
would certainly improve the nutritional
intake of the poor.
Finally, it is also important to mention the
role of the relative price structure within
agriculture for the success of the proposed
strategy. A number of studies exist on the
role of the terms of trade between agriculture
and industry and its impact on agriculture.
However fewer studies exist on crop substitu
tions arising from changes in relative prices.
Land is not like capital. It is a more
malleable factor and is capable of growing
a number of alternative crops where irriga
tion is especially available Therefore it is im
portant to have an appropriate relative price
structure in each region to achieve the
desired goal in food production.
Ill
Wages and Employment
As population grows so too would labour
force. A crucial link between population
growth and poverty is employment. Since
employment is the means to secure entitle
ment a number of questions come to one’s
mind. At what rate would the economy
generate employment? Where would these
jobs be created? How would wage rates res
Tabi.e 9; Percentage Share of States in Total F<x)ix;rains Procurement
Slates
1961-62
1973-74
1983-84
1986-87
19878-8
1988-89
Andhra
Assam
Bihar
Gujarat
Haryana
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
All India
0.00
14 60
nil
nil
nil
12.56
1.70
1.10
0.83
9.42
1.97
0.74
5.26
4.09
2.76
33.32
3.10
5.76
13.52
2.91
100.00
10.81
0.11
0.16
0.04
3.24
0.82
ml
1.84
0.45
0.53
53.88
1.41
2.01
14.14
0.18
100.00
9.45
0.08
nil
nil
17.04
0.72
nil
2.90
0.33
0.79
49.81
0.51
4.88
13.00
0.35
100.00
10.23
0.05
nil
0.02
13.72
0.58
nil
2.32
0.86
0.61
55.71
0 24
4.71
10.17
0.53
100.00
10.28
0.03
0.08
0.16
14.82
0.80
nil
1.76
nil
0.75
51.74
0.71
3.77
14.05
0.53
100.00
nil
nil
2 20
nil
6.06
48.76
nil
nil
28.37
nil
100.00
Tabu 10; Ratio oi Pi bik Djstribi i ion ot Fcxhxirains to Cereai Consumption
Stales
1961
1973-74
1983
1986-87
1987-88
1988-89
Andhra
Assam
Bihar
Gujarat
Haryana
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
0.01
0.08
0.03
0.05
NA
0.02
0.12
0.02
0.10
0 02
0.08
0.01
004
0.04
0.11
0.04
0.09
0.04
0.19
0.07
0.07
0.48
0.04
0.29
0.06
0.10
0.06
0.06
0.04
0.26
0.13
0.16
0.07
0.07
0.09
0.10
0.49
0.05
0.16
0.09
0.12
0.02
0.22
0.05
0.32
0.16
0.23
0.07
0.26
0.05
0.16
0.53
0.08
0.20
0.07
0.12
0.16
0.23
0.04
0.16
0.10
0.18
0.06
0.25
0.07
0.13
0.52
0.06
0.19
0.06
0.04
0.15
0.23
0.05
0.18
0.10
0.15
0.04
0.15
0.03
0.12
0.43
0.05
0.18
0.07
0.02
0.08
0.26
0.04
0.13
Source: Cereal Consumption1—NSS’s Reports on Consumer Expenditure, relevant rounds.
Economic and Political Weekly
November 14, 1992
pond to growth in labour force? Would real
wage rates continue.to rise if increases in
employment lag behind growth in labour
force? What would be the strategy of trade
unions in a labour market with high levels
of unemployment? What would happen to
traditional labour market relations in a con
text of increasing unemploymen*? Would
competition for jobs lead to a brerk down
of traditional labour market relations? All
these issues deserve enquiry for a clearer
understanding of the emerging scenario on
the interrelationships between population
growth, poverty and employment. However,
considering the complexity of the issues
involved, this section is confined only to a
few selected aspects of these relationships.
Trends in Labour Force and
Participation Rates
By the year 2025, the potential labour
force, i e, the population between the ages
15 to 59, based on the United Nations
population projections, would have risen
from 431 million in 1990 to 929 million in
2025 [United Nations, 1991]. But all these
persons may nol actually be seeking or
available for work. A number of them may
still be in schSol or undergoing other train
ing and thus be out of the labour market.
Therefore, we shall follow an alternative con
cept, namely, of work participation adopted
in the 1991 population census. Work par
ticipation rate is defined as the proportion
of total workers in the population, and total
workers are obtained as the sum of main and
marginal workers (Registrar General, 1991).
Thus totally unemployed persons are exclud
ed from this category and to that extent the
size of the labour force thus derived would
be an underestimate. Even on this basis,
India would have to provide additional
employment for nearly 220 millions by the
year 2025. These additional jobs would con
stitute about 70 per cent of the employed
population in 1991 (Table II).
Trends in Unemployment
An interesting aspect of the Indian labour
market is that the growth in labour force did
not seem to have significantly raised the pro
portion of unemployed in the labour force.
The basic data on employment and un
employment collected by the National Sam
ple Survey have been analysed by a number
of scholars [cf Visaria and Minhas, 1991].
I do not propose here to summarise the fin
dings of these studies.
The National Sample Survey employs
three distinct definitions to estimate employ
ment and unemployment, namely, usual
status, weekly activity, and daily activity. On
the basis of these three definitions three
measures of unemployment are provided.
The largest proportion of unemployed are
reported under the daily activity status and
the smallest for the measure based on usual
status. According to the 38th round, the in
cidence of unemployment, based on the
Economic and Political Weekly
usual status definition, was below 4 per cent
for males in all states excluding Kerala where
it was as high as 10.5 per cent [National
Sample Survey, 1987]. The incidence of
unemployment among females was below 1
per cent in many states but was about 12 per
cent and 17 per cent for Punjab and Kerala
respectively. Thbh 12 provides data for a few
states on the distribution of the labour force
according to the number of days seeking or
available for work. These clearly show that
between 85 and 90 per cent are employed full
time and only in three or four states this pro
portion falls below 85 per cent In the rural
sector of Kerala and Thmil Nadu only 64 and
70 per cent respectively are employed full
time compared to 96 per cent in Rajasthan
and in Madhya Pradesh The average
number of days worked, therefore, reflected
these situations in different states. While in
Kerala it was less than five days in a week
in the other states it exceeds six days. These
facts indicate that the increases in rural
labour force either have been certainly ab
sorbed in the economy or none of these
definitions really caught the actual condi
tions of employment within the prevailing
institutional and social structure of the rural
labour market.
In any case, the low level of reported
unemployment has important implications
for the various rural employment program
mes like the Maharashtra Employment
Guarantee Scheme. In fact, the number of
persons available for work under such
employment programmes might reflect the
true incidence of unemployment since the
wage rates offered in such schemes were nor
mally lower than the prevailing market wage
rate. The offer of labour at wage rates below
the prevailing market rate might indicate the
extent of distress faced by the unemployed.
The gender incidence of unemployment
could also form another basis for targeting
of the employment guarantee programmes.
Institutional aspects of Labour
Market Relationships
Indian studies on wages and employment
have concentrated largely on the questions
of measurement of unemployment and on
the trends in money and real wage rates.
Such aggregate studies do not throw light
Table 11: Potential Labour Force and Projected Total Number of Workers
(Figures in 000s)
Year
1971
1981
1991
2000
2010
2020
2025
Population between Ages 15-59
Male
Female
Total
153493
196212
(2.28)
249777
(2.28)
312081
(2.25)
385828
(2.25) ■
452297
(1.62)
477901
(1.01)
143832
182488
(2.28)
231503
(2.28)
289022
(2.25)
358846
(2.25)
425166
(1.63)
451578
(1.25)
297325
378700
(2.28)
481280
(2.28)
601103
(2.25)
744674
(2.25)
877463
(1.62)
929479
(1.12)
Total Number of Workers*
Male
Female
Total
149777
185919
(2.25)
223506
(2.00)
276807
(2.00)
324142
(1.62)
361871
(1.12)
379520
(0.89)
37543
64912
(5.58)
91397
(3.50)
114713
(2.00)
135201
(1.62)
152273
(1.14)
160545
(1 13)
187320
250831
(3.00)
314903
(2.53)
391520
(2.00)
459343
(1.62)
514144
(1.14)
540065
(1.00)
Noles: Figures in brackets show the compound rate of growth.
* Projected assuming the same work participation rale as in 1991 Census.
Table 12: Percentage Distribution of Persons (5 + ) in the Labour Force Classified
ACCORDING TO NUMBER OF DAYS SEEKING/AVAILABLE FOR WORK (CURRENT WEEKLY STATUS)
(Rural)
States
Andhra
Bihar
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Rajasthan
Tamil Nadu
UP
0
3
3, 7
=7
Average
No of Days
Worked
85.29
86.29
86.59
63.53
96.65
87.72
84.50
96.27
70.85
94.42
7.24
6.89
7.77
13.10
1.87
7.04
8.08
1.16
12.27
2.64
3.72
3.03
3.15
8.58
0.37
2.49
3.31
0.69
8.75
1.23
3.75
3.79
2.49
14.18
1.11
2.75
4.13
1.88
8.13
1.71
6.08
6.34
6.15
4.84
6.63
6.20
6.22
6.65
5.41
6.51
Source: NSS Draft Reports on the Third Q uinquennial Survey on Employment and Unemployment.
November 14, 1992
2493
on how the labour market adjusts to the
demographic pressures or how employment
and wages respond to these changes. The in
stitutional framework of the labour markets
vary widely over the Country, from
predominantly feudal or semi-feudal rela
tions in a state like Bihar to the operation
of a relatively free labour market in Kerala.
How employment and wages are determined
under such differing labour market condi
tions is crucial for an understanding of the
adjustment processes in the labour market
for demographic pressures. The very low
rates of unemployment, whatever be the
definition, reported in many states in India
might be partly attributed to the institutional
constraints imposed on the operation of the
labour markets. For instance, the existence
of ‘tied’ or ‘bonded’ labour would not
enable those under such systems to seek
work or even report themselves as available
for work. The point that one would like to
emphasise here is that the perceptions about
unemployment and employment would very
much be coloured by the institutional and
social structure under which the labour
mafket operates. Amartya Sen distinguishes
between three aspects of employment and
one of them is the ‘recognition aspect’ which
is influenced by the institutional and social
structure within which a labourer operates
[Sen, 1975]. Most studies on the measure
ment of the number of unemployed fail to
consider this ‘recognition aspect’. Therefore,
estimates of unemployment or of employ
ment might not truly reflect the extent of the
demographic pressure on the labour market.
On the other hand, levels of household in
come or expenditure and their temporal
changes might provide truer indications of
the consequences of population pressure on
the household and the society. Even an
analysis of trends in real wage rates would
fail to indicate the extent of population
pressure on the economic system, for, it is
possible that real wage rate might go up
while the number of days of total employ
ment might have declined at the same time
(Krishnan, 1991]. In normal situations, the
operation of the labour market has to solve
satisfactorily a number of issues concerning
the determination of the wage rate. First, the
wage rate must be sufficient enough to
enable the continuous reproduction of
labour. Second, it should satisfy the
minimum needs of worker so that it induces
him to put in enough work effort jo main
tain productivity. Third, if the wage rate in
any sector is related to the wage rates in other
sectors then it might also require that the
wage parities are maintained. All these fac
tors are closely related to the institutional
structure under which the labour market
operates. We have very little information and
knowledge as to how wage rates for various
categories of labour in different sectors are
determined. It might be conceivable that
within broadly segmented labour markets
there also exist close interconnections among
certain categories of wage rates. Such a fin
ding emerged from a study of the rural
2494
labour market in Kerala [Krishnan, 1991].
An analysis of the wage structure in the
rural and urban sectors of Kerala brought
out many interesting features of the inter
relationships among wage rates in the
agricultural and construction sectors. We
considered wages of eight categories of
labour in the rural sector, namely, those of
mason, carpenter, unskilled male and un
skilled female labourer for the construction
sector and of paddy field male labour, paddy
field female labour, other agricultural male
labour and other agricultural female labour
in agriculture. During the period 1960 to
1988 it was found that the wage rates mov
ed in such a way as to preserve the wage
relativities. All the wage relativities were
calculated in terms of the wage rate of
mason. When the wage relativities were ar
ranged in a descending order, they fell into
a hierarchical pattern, and varieu wtween
1.0 and 0.41 [Krishnan, 1991]. Male
agricultural worker and unskilled male
worker in construction received about twothirds of the wage of mason whereas those
of female workers were between 40 and 47
per cent. As we moved down the hierarchical
ladder it was found that (he coefficient of
variation of the wage ratios also rose though
the largest figure reported for female
agricultural labour was still only 10 per cent.
The smallest coefficient of variation was in
the wage ratio for carpenters, only 2 per cent
[Krishnan, 1991].
The stability of the wage ratios raised an
important question as to whether the subset
of wages considered in the analysis were
mutually related. If the wage rates were sub
ject to mutual feedback relationships, then
a change in any one of the wage rate,
whatever be its cause, would trigger changes
in other wage rates in order to restore the
wage parities. The existence of the feedback
was verified by testing the relationships for
Granger causality by fitting vector autoregressions and these tests were positive. The
presence of Granger causality implied that
the changes in wage structure could be divid
ed into two separate processes when they
were not attributed to a change in the general
price level. A change in a particular wage
rate brought about by a given factor in that
labour market may be described as arising
from a ‘causal factor’ for the initial wage
change. The next process is the ‘induced’
change in other wage rates in order to restore
parity. This latter wage adjustment we
ascribe to ‘structural’ factors associated with
the institutional structure of the labour
markets for those categories of labour. The
labour markets of those categories of labour
which were structurally related were said to
be interrelated as opposed to segmented.
The most important aspect of the
phenomena of interrelated labour market is
the mutual feedback among wage rates.
Labour markets may be compared to a
group of islands where communication
exists within but not between islands. A
group of interrelated labour markets is like
a single island, but segmented labour
markets arc like separate islands with no
means of communication. Within each
segmented labour market there may exist
interrelationships among wage rates. For
instance the labour markets in agriculture
and construction arc interrelated but the
market for industrial labour may be
segmented from these two markets.
However, within the industrial labour
market, the wage rates for various categories
of labour in a particular industry or even
between different industries may again be
interrelated. In that case the latter constitutes
another interrelated labour market but
unrelated to the first one.
The very existence of interrelated labour
market is anathema to the application of
supply and demand analysis to the problems
of the labour market [Solow 1991 J. So what
explains the prevalence and persistence of
wage relativities? The persistence of wage
relativities in the building trades over along
historical period was reported for England.
We attribute the prevalence of wage relati
vities to the adherence to social norms.
Agriculture and construction were in
existence from time immemorial and
therefore the wage parities developed over
time through custom and usage. Such
parities probably naturally got established
themselves easily in hierarchical societies.
Therefore, it was likely that some form of
wage parities did prevail among longestablished traditional occupations. When
new occupations were created in'the process
of industrialisation, possibly the labour
market got initially segmented but later
established its own wage rules within such
segmented labour markets.
The hierarchical nature of the Indian
society was closely bound by the caste struc
ture of our society. Caste and traditional
occupations were also closely inter-linked.
Any person living in such a society would
be conscious of his rank within that society
and would do everything necessary to pro
tect his position. He would be constantly
comparing his status with those above and
below him. The wage differential might
reflect partly social position and partly a
premia for the embodiment of skill in dif
ferent occupations.
Trends in Nominal and Real Wages
The caveats discussed above are important
in an analysis of nominal and real wages.
Therefore, we have derived indices of
nominal and real wages relative to the wage
rate in the state of Punjab. We have only
considered the wage rate for male agri
cultural labour. Punjab reported the highest
wage rate among all the states in India. Real
wage rates were estimated by deflating the
nominal wage rates by the agricultural
labourer’s cost of living indices of the-respec
tive states. We have also estimated the
unweighted coefficient of variation in these
wage rates. These estimates are shown in
Tables 13 and 14. There are wide divergences
in the wage rates between the states. The
Economic and Political Weekly
November 14, 1992
lowest wage rate, nominal as well as real, is
only ab<?iit one-third of that of Punjab.
Though wc have not made any attempt to
test the relationship, it appears that the more
backward a state is with respect to social and
institutional development the lower the
prevailing wage rate.
The inter-state inequality in wage rate ap
peared to have increased by 1973 compared
to 1960 but seemed to have decreased to
some extent by 1987. During the period 1960
to 1987, the inter-state inequality in the
agricultural wage rates appears to have per
sisted and even shown a mild increase. The
factors underlying these trends deserve
detailed analysis which is not attempted here.
Wage Productivity Nexus
One of the factors underlying the dis
parity in wage rates might be the differences
in agricultural productivity between different
regions or states. The relationship between
productivity and wages is closely tied to the
marginal productivity theory of wages.
While there are many problems in accepting
the marginal productivity theory to explain
the level of wages, it is important to
recognise that levels and changes in produc
tivity in relation to the wage rate do play an
important role in determining the demand
for labour. A V Jose hypothesised that “the
single-most important variable influencing
the movement of real wage rates in any state
has been the level of agricultural output in
the respective state" [Jose 1988.]. In support
ot this Jose nad compared the estimates of
compound rates of growth in real wages and
in net domestic product in agriculture for
each state during the period 1964-84 and
found that they moved together [Jose 1988].
We verified Jose’s hypothesis by regressing
real wage rate on net value of agricultural
product per agricultural worker (cultivators
plus agricultural labourers under the
category of main workers) and on the pro
portion of agricultural labourers in total
rural main workers. We also estimated alter
native regressions where we substituted the
value of agricultural product per worker
with foodgrains output per agricultural
worker. We also Titled another equation for
money wages where cereal price was
included as an additional explanatory
variable In this case we used foodgrains out
put per agricultural worker instead of value
of agricultural product per worker in order
to avoid problems of multi-collinearity. The
results of these estimates are given below:
Variables
Mw
= Money wages of male agricultural
labour
Rw
= Real wages of male agricultural
labourer.
AGLR = Ratio of agricultural labour to
total main workersVAD = Per worker value of agricultural
product.
PFD = Per worker quantity of foodgrains
output.
Economic and Political Weekly
Rw - 2.0881+0.00043(VAD)-L2I03(AGLR) rates but through an adjustment in the
average number of days of work available
(4.677) (3.7534)“* (-0.8203)
to each worker or through otjicr social
R:2 = 0.60
(2) Rw = 2.3762+ 0.00035(PFD)-L4075(AGLR) mechanisms. On the other hand, our findings
confirm Jose’s hypothesis that the level of
(4.6400) (2.5344)’’ (-0.7955)
agricultural productivity was indeed a major
R: = 0.42
(3) Mw = 4.2025+3.9073(CPR) + 0.0029(PFD) factor. States with higher agricultural pro
ductivity were also states with higher per
(0.6214) (1.9640)’ (3.1922)*”
capita foodgrains production and hence
-15.6445 (AGLR)
inter-state variations in per capita foodgrains
(-1.6493)
production also proved to be an equally
R2=0.57
Notes: Figures in brackets show the respective powerful explanatory variable in inter-state
variation in real wage rates. Though one may
t-values.
not attribute a causal relation between per
Significance level: ”•= I per cent,
capita
foodgrains production and the level
•• = 5 per cent, * = 10 per cent.
We find that the proportion of of real wages, it provides another important
agricultural labour in total agricultural link between poverty and wage rate
labour force (including cultivators) was not
The proportion of agricultural labour in
significant in the real wage equation imply total agricultural work-force proved to be
ing that the supply of labour was probably insignificant in the equation for money
not an important factor in determining the wages also. However, cereal priced and per
wage rate. This probably implies that
worker foodgrains output turned out to
the demographic pressure operates in the
be significant. It so happens (as mentioned
labour market not through changes in wage
in the second lecture) that cereal prices
(I)
Table 13: Index of Money Wage Rate for Agricultural Labourers
Male
1960
Female
Male
1970
Female
Male
1980
Female
Male
1987
Female
53.72
Andhra
95.87
Assam
52.89
Bihar
77.27
Gujarat
na
Haryana
68.60
Karnataka
66.53
Kerala
Madhya Pradesh 43.39
62.40
Maharashtra
54.96
Orissa
100 00
Punjab
na
Rajasthan
47.52
Tamil Nadu
66.94
Uttar Pradesh
78.93
West Bengal
0.25
cv
57.14
112.42
68.94
86.96
na
78.26
68.94
51.55
53.40
57.14
100.00
na
52.80
32.36
96.27
0.27
40.03
65.95
41 56
44.17
89.57
44.48
72.55
31.90
46.47
36.04
100.00
51.99
37.88
40.21
49.85
0.38
40.00
65.89
46.11
46.74
84.21
35 58
63.58
31.16
40.63
33.26
100.00
37.47
35.37
51.73
50.53
0.39
47.69
56.35
44.14
57.92
95.96
49.50
79.04
37.29
43.56
44.80
100.00
61.14
48.18
47.60
59.08
0.31
35.92
50.00
36.08
46.88
66.64
37.36
58.32
34.88
29.36
39.36
100.00
46.48
34.24
59.45
48.80
0.36
52.22
71.19
53.23
58.13
92.99
43.73
90.07
45.08
42.25
37.18
100.00
71.82
47.32
52.94
79.13
0.31
50.00
71.46
68.97
62.65
65.57
51.35
85.89
49.41
48.65
40.54
100.00
56.76
33.35
na
64.38
028
States
Source: Agricultural Wages in India, relevant years.
Table 14: Index of Real Wage Rate for Agricultural Labourers
Male
1960
Female
Male
1970
Female
Male
1980
Female
Male
1987
Female
55.60
Andhra
97.41
Assam
50.86
Bihar
75.86
Gujarat
na
Haryana
71.12
Karnataka
67.67
Kerala
Madhya Pradesh 43.53
59.48
Maharashtra
53.88
Orissa
100.00
Punjab
na
Rajasthan
na
Tamil Nadu
Uttar Pradesh
na
West Bengal
71.55
CV
0.25
59.35
113.55
65.81
85.16
na
80.65
69.68
51.61
50.97
56.13
100.00
na
na
na
87.10
0.26
45.90
63.22
37.08
48.63
na
47.72
67.48
30.70
48.63
31.91
100.00
53.19
44.07
32.83
46.81
0.37
46.03
63.18
41.00
51.46
na
38.49
59.41
30.13
42.68
29.71
100.00
38.49
41.00
40.59
47.70
0.37
59.93
57.88
45.21
71.23
89.36
57.88
90.41
36.99
47.95
41.10
100.00
61.64
55.82
50.68
61.30
0.30
45.18
51.50
36.88
57.48
84.10
43.85
66.78
34.55
32.56
36.21
100.00
46.84
39.53
46.84
50.50
0.33
64.45
69.57
53.71
64.19
66.78
46.80
76.98
45.52
42.97
35.55
100.00
75.70
48.59
53.20
78.77
0.28
61.76
69.93
69.61
68.95
108.15
54.90
73.20
50.00
49.35
38.56
100.00
59.80
34.31
na
64.05
0.26
Slates
Source: Agricultural Wages in India, relevant years.
November 14, 1992
2495
tended to be high in low per capita
foodgrains states and vice versa. Thus,
higher money wages probably compensate
the higher cereal prices where per capita
foodgrains production is lower. While we
have not analysed the relationship between
wage and productivity in the industrial sec
tor, one of the important factors in inter
state differential in agricultural wage rate ap
pears to be productivity per agricultural
worker. This is not only confirmed by the
inter-state analysis but also by another
analysis of the behaviour of acreage under
paddy in Kerala [Krishnan 1991]. Kerala had
the highest agricultural wage rate after
Punjab and Haryana and agricultural
labourers were also unionised in the slate The
changes in acreage under paddy fell into two
distinct phases: first phase from 1960-61 to
1974-75 when acreage rose steadily, from 779
thousand hectares to 881 thousand hectares;
a second phase beginning from 1975-76
when paddy acreage began to decline
steadily and was only 577 thousand hectares
in 1988-89 [Krishnan 1991]. Throughout this
period agricultural wages continued to in
crease. Duringjjhe first phase, farm harvest
price rose at a higher rate than that of wage
rate and productivity also increased. These
trends got reversed during the second phase.
During the first phase, wage increase arose
from a combination of two factors, the im
plementation of minimum wages and the
bargaining strength from unionisation. In
the second phase, wage increases appeared
to have been triggered by entirely different
reasons. From 1975, migration to West Asia
began and the emigration of semi-skilled
construction workers led to an increase in
their wage rates. Besides, the remittance
incomes also resulted in a sudden increase
in construction activity within Kerala pro
viding further fillip to wage increases. As
mentioned earlier, the construction and
agricultural labour markets were inter-linked
and hence when construction wages rose, the
agricultural wages also did likewise.
However, in the absence of either produc
tivity increases or rise in product price to
compensate for the wage rise, the effect of
the wage increase was to reduce the acreage
under paddy. An acreage response function
fitted to lagged price of rice and the wage
rate indicated that there was a strong
negative response to wage rate. The response
coefficient to wage rate was highly signifi
cant whereas it was not so in the case of farm
price [Krishnan 1991]. As a result of the
decline in acreage under paddy either the
paddy land was kept fallow or it was shifted
to the growing of crops which required less
labour input compared to that of paddy
cultivation. A male agricultural labourer had
employment for 198 days in a year in 1964-65
but could secure only 147 day’s work in
1983-84 [Krishnan 1991]. This result was due
to the operation of two distinct factors, one,
increase in the number of agricultural labour
households partly arising from population
growth, and another due to the shift in crop
2496
ping pattern brought about by rise in wage
rates and in changes in relative prices of
agricultural products.
The relationship between wages and pro
ductivity has important implications for
employment and alleviation of poverty.
While low wage rate and declining number
of days of employment are contributory fac
tors to the persistence of poverty, the long
term solution do not lie in raising wage rates
when productivity continued to remain stag
nant. This observation may sound trite but
without increases .in productivity attempts
to raise wages would tifrn out to be counter
productive. This factor is especially impor
tant in revising minimum wages. In a situa
tion of low wage-low productivity trap, at
tempts to implement minimum wages ignor
ing this relationship either would not suc
ceed or forced would shift cropping patterns
and labour use.
The construction of minor irrigation
works toextend cultivation and the develop
ment of social forestry in all villages are
closely related to agricultural development.
Reforestation will not only provide relief to
the shortage of domestic fuel but retaining
moisture in the soil would also arrest soil
erosion and decline in soil fertility. The
extension of area under irrigation would
enable double cropping and in crop diver
sification. Apart from raising the volume of
employment, these measures would raise
productivity, food consumption and also
make it feasible the increases in wage rates.
While the importance of foodgrains in the
Indian economy should not be belittled, the
time has come for a major diversification
in agricultural production. Given the
massive increase in the numbers needing
employment in future, we should realise that
industry might not be in a position to ab
sorb this labour force. Therefore, agriculture
and allied activities would have to continue
to expand their employment opportunities.
This expansion of employment is vital to
stem the tide of flows of employment seekers
to the cities. Therefore, diversification of
agricultural production and the development
of agro-processing industries in and around
rural areas would go a long way in tackling
these issues.
Women's SiaTUS. Poverty and
Employment
The importance of women’s status for fer
tility reduction and alleviation of poverty
needs to be explicitly recognised in develop
ment programmes. Women constitute not
only about half the population of the coun
try but, as various studies have shown, they
bear the brunt of poverty. As mentioned in
the first lecture, women’s status and levels
of education in society and family are vital
for the success of family planning program
mes. We would, therefore, like to’ suggest
special measures for generating employment
for women. It is suggested that women’s co
operatives comprising of female agricultural
labourers and women belonging to small and
marginal farmers be constituted. Such co
operatives should be provided with technical
and administrative help to set up agro
processing industries. The advantages of
agro-processing industries are that their
technologies are scale neutral, can be ubi
quitous in location and are ideal as rural
industries. The creation of such enterprises
would enable the rural women to supplement
their income, increase their level of work
participation, promote social intercourse and
provide some degree of autonomy in family
decision-taking. All these factors might
ultimately impact favourably on the accep
tance and practice of family planning and
in reducing population growth. An increase
in female employment and income would
raise food and nutritional intake and reduce
malnutrition and the incidence of Sickness
arising from malnutrition. Thus women’s
employment can be a potent weapon for
breaking the linkages between population
growth and poverty.
Concluding Remarks
Our analysis of the issues of population,
poverty and employment indicates,that the
achievements of Indian planning represent
at best only partial successes. Primarily, the
whole planning effort in India critically
failed in achieving the most important social
and human values any society normally
cherished, namely, to have an educated and
a healthy population. In large regions of the
country, illiteracy, especially among women,
continued to be still rampant, nearly 2 million
children would die before they completed
their first year of life and the bottom 30 per
cent of the population still consumed only as
much grain as they used to consume 30 or
40 years ago. In terms of absolute numbers,
the number of illiterates, the total number of
infant deaths and the total number of
malnourished or partially hungry persons
have all risen in 1992 compared to 1952 when
we started our planning. The failure to
eradicate illiteracy, to improve the status of
women in society and to assure access to
health care also accounted for the failure to
reduce the rate of population growth. The
real shortcomings of Indian planning lie in
its overall neglect of social sector planning
and thus in ns failure to integrate popula
tion issues with development programmes
and policies. Therefore the social objectives
either enjoined in the Constitution of India
or often politically articulated could never
Economic and Political Weekly
Available from
M/s. NEBS.
Dharampeth,
Nagpur - 440 010,
Maharashtra.
Economic and Political Weekly . November 14, 1992
be fulfilled. Indian planning beginning with in the coming years as the challenges posed
the Second Fivc-Year Plan over-cmphasiscd_ by population growth, poverty, and the need
physical and financial planning at the to generate additional employment would
prove to be more daunting during the next
expense of sociaTplanning
An important partial success of Indian 35 years than they were hitherto.
planning was its ability to increase [This is a revised version of the text of the
foodgrains output commensurate with the author’s Kunda Datar Memorial Lectures
increase in population. At the time when the delivered at the Gokhale Institute of Politics
food production strategy was put in place, and Economics, Pune, on April 28, 29 and 30,
1992, incorporating some of the comments
it appeared as a spectacular achievement, but
with hindsight, it was probably a path made by the participants at the lectures. The
guided by a softer option to solve the food author is also grateful to S Mohana Kumar for
crisis. Basically, the food policy was a wheat his diligent research assistance.)
strategy and it widened (he inter-state
disparities in per capita grains production
References
and necessitated the transport of large quan
tities of grains across the country. This
Bhat, Mari and S Irudayarajan {1990]:
strategy also led to the neglect of the
‘Demographic Transition in Kerala Re
development of mino^and unconventional
visited’, Economic and Political Weekly,
sources of irrigation to support the cultivaSeptember 1-8, 1990, pp 1957-80.
tioQ_of inferior cereals on which the poorer
Government of Kerala {1992): Sample Registra
popukmon_o£Jarge tracts of the country
tion, Kerala—Rural, Department of
depended for their grain consumption. Bar
Economics and Statistics, Trivandrum.
ring three or four states, per capitajhains out Gupta, J P and Indira Murali {1989]:
put declined in all states during the period
National Review of Immunisation Pro
1961-1939. Thus, the agricultural production
gramme in India, Appendix XXIV, p 230,
strategy failed to raise productivity m states
and Appendix XIV, p 216, National In
or regions where it was low. Our analysis
stitute of Health and Family Welfare, New
showed that not only per capita consump
Delhi.
tion of grains depended on per capita out Jose, A V {1988]: ‘Agricultural Wages in India’,
Economic and Political Weekly, June 25,
put but agricultural wage rates too appeared
1988.
to depend on the level of per'capita grains
Kabir,
M and T N Krishnan {1992]:‘Social In
output or level of agricultural productivity.
termediation
and Health Transition:
Therefore, a regionally balanced agricultural
Lessons from Kerala’. Paper presented in the
development strategy would have provided
workshop on Health and Development in
a double benefit, namely, it would have con
India, jointly organised by National Coun
tributed directly to reducing poverty by
cil of Applied Economic Research and the
enabling the poor to raise their level of grain
Harvard Center for Population and
consumption and also indirectly by raising
Development Studies, at India International
wage rates which would have further helped
Centre, New Delhi, January 2-4, 1992.
them to purchase larger quantities of grains.
Kakwani, N and K Subbarao {1990]: ‘Rural
Though we did not undertake any detail
Poverty and its Alleviation in India’,
ed examination of the process of labour
Economic and Political Weekly, March 31,
market adjustment to the demographic
1991.
pressure, some tentative hypotheses seem to
Krishnaji, N (1988J: ‘Foodgrain Stocks and
emerge from our analysis. First and
Prices in Amiya Kumar Bagchi (ed),
Economy, Society and Polity, Essays in the
foremost, there is no strong evidence to sug
Political Economy of Indian Planning, Ox
gest that wage rates respond to demographic
ford University Press, Calcutta.
pressure at all. This might be attributed to
the institutional and social structure within
Krishnan, T N {1976]: ‘Demographic Transition
which the labour market operated in a poor
in Kerala: Facts and Factors’, Economic and
Political Weekly, Annual Number, August.
agrarian economy. However, there is some
—{1991]: 'Wages, Employment and Output in
indication that wage rates are sensitive to
Interrelated Labour Markets in an Agrarian
grain prices and to agricultural productivity.
Economy: A Study of Kerala’, Economic
These conclusions are based on the analysis
and Political Weekly, June 29.
of cross-section data and not on a study of
—(1991a): ‘Kerala’s Health Transition—Facts
wage movements over time. Similarly, the
and Factors’. Paper presented at the Health
reported rates of unemployment in rural
Transition Seminar of the Harvard Centre
areas are also probably insensitive to
for Population and Development Studies in
demographic pressures. We tentatively
November 1989 (revised version).
hypothesise that both unemployment and
Mahendra Dev, S and M H Suryanarayana
rural wage rates are closely tied to the
(1991): ‘Is PDS Urban Biased and Pro
institutional and social structure of the
Rich?’ Economic and Political Weekly,
organisation of the labour market and we
October 12.
have only a minimal understanding of these
Mechanic, David (1992]: ‘Research Possibilities
factors. However, the labour force entering
on Facilitating the Health Care Transition’
the labour market during the next 35 years
in L C Chen el al (eds), The Role of Social
would be close to 220 million and this would
Research, Auburn House Publishing Com
pany, Westport, CT.
not be a small figure to reckon with. Indian
Ministry of Agriculture [1990]: Bulletin on
development would be put to a severe test
Economic and Political Weekly
November 14, 1992
Food Statistics, Directorate of Economics
and Statistics, Government of India.
Ministry of Health and Family Welfare (1989]:
Family Welfare Programme in India, Year
Book 1987-88, Government of India, NewDelhi.
National Sample Survey Organisation (NSSO)
[1977]: Tables with Notes on Consumer Ex
penditure, October 1973-June 1974, 28th
Round, No 240, Department of Statistics,
New Delhi.
— [1987]: Report on the Third Quinquennial
Survey on Employment and Unemploy
ment, 38th Round, Jan-Dec 1983, Draft
Report, Survey Results—All India, Table
(25), p 79, Department of Statistics, New
Delhi, June.
—[1991]: Sarvekshana, Vol XIV, No 3, Issue No
46, January-March.
—(1991]: Sarvekshana, Vol XV, No 1, Issue No
48, July-September.
Pathak, K B and Faujdar Ram (1985): ‘Trends
in Reproductive Values of (he Females and
Momentum of Population GrowthAnalysis for the Major Stales of India’
(mimeo), P 30, International Institute of
Population Studies, Bombay.
Planning Commission (1985): Seventh Five-Year
Plan 1985-1989, Government of India, NewDelhi.
—[1992]: Eighth Five-Year Plan 1992-97,
Government of India, New Delhi.
Premi, K Mahendra [1991]: India’s Population:
Heading Towards a Billion—An Analysis of
199! Census Provisional Results, B R
Publishing Corporation, Delhi.
Radhakrishna, R and C Ravi (1992]: ‘Does In
flation Hurt the Poor More?’ Economic and
Political Weekly, January 25.
Raj, K N [1990]: Organisational Issues in Indian
Agriculture, Oxford University Press, Delhi.
Registrar General of India [1991]: Census of
India 1991, Series I, India, Provisional
Population Totals, Paper 1 of 1991, New
Delhi.
—[1991]: Sample Registration Bulletin,
Vol XXV, No 1, Ministry of Home Affairs,
New Delhi, June.
—[1989): Census of India 1981, Child Mortality,
Age at Marriage and Fertility in India, Oc
casional Paper No 2 of 1989.
—(1989a): Census of India 1981, Fertility in
India—An Analysis of 1981 Census Data,
Occasional Paper, No 13 of 1988.
Sen, A (1975): Employment, Technology and
Development, Delhi, Oxford University
Press.
Solow, Robert M (1991): The Labour Market
as a Social Institution, Basil Blackwell,
Oxford.
The Hindu (1992): March 21, Madurai edition.
United Nations (1975): Poverty, Unemployment
and Development Policy: A Case Study of
Selected Issues with Reference to Kerala,
New York.
—[1991]: The Sex and Age Distribution of
Population—The 1990 Revision, Depart
ment of International Economic and Social
Affairs, Population Studies No 122, United
Nations, New York.
Visaria, Pravin and B S Minhas [1991]: ‘Evolv
ing an Employment Policy for the 1990s:
What Do the Data Tell Us?’ Economic and
Political Weekly, April 13, 1991.
2497
Poverty m West IBengal
What Have We Learnt?
Biswajit Chatterjee
This article reviews some of the existing studies on the incidence ofpoverty in West Bengal, with particular
emphasis on the decades of 1970s and 1980s. Discussions of the implications of all-India or inter-state
poverty studies on the poverty scenario in West Bengal are followed by a critical review of some of the
major studies of rural and urban poverty in the state. The aggregative state level scenario as portrayed
by these studies has been contrasted with the grass roots level picture of rural poverty as it has been brought
out by some of the recent case studies based on household surveys /resurveys. The implications of all these
studies for formulation of effective anti-poverty policies in West Bengal are also emphasised in this paper.
I
Introduction
INDIA is a poor country by world stan
dards. There are considerable variations
in poverty across the states of India, and
wide fluctuations in living standards over
time. Naturally, poverty is an important
question which has attracted the attention
of researchers and policy-makers during
the 50 years of our national economic
development through planning in a mixed
economy framework. In the state of West
Bengal, in 1977 a left-front government
came to power with massive popular
mandate. It took a pro-poor stance in its
economic development policy. That gov
ernment has now completed 20 years of
rule in the state. The question is whether
the incidence of poverty in this state has
gone down over the years. In this article,
we shall reflect on this question on the
basis of available studies on the incidence
of poverty and its variations overtime. We
shall also use the insights of some of the
recent case studies on the subject to gauge
the incidence and seventy of poverty and
related deprivations at the micro level on
a selective basis. The relevance and sus
tainability of the existing poverty allevia
tion programmes in this state will be
examined in view of our reappraisal oflhe
available studies.
In Section II. I shall briefly review the
major concepts of poverty norms and the
results of some major studies on the esti
mates of poor in West Bengal as part of
all-India and inter-stale poverty estimates
attempted by various scholars and insti
tutions. The ranking of the state of West
Bengal in terms of indices of poverty
among all oilier states of India, its change
over time, and attempts to identify the
determinants of rural and urban poverty
in the state will he critically reviewed in
this section. This will be followed by
appraisals of the results of some studies
on rural and urban poverty undertaken
Economic and Political Weekly
exclusively for the state of West Bengal.
In Section III, I shall present my own
estimates of the incidence of poverty in
rural and urban West Bengal during the
1970s and 1980s. using state-specific
dietary norms and state-specific retail
paces. I have also chosen some case studies
al the micro level to report some of the
important findings and argue that theextent
of poverty in some areas is much greater
than is revealed through the state level
aggregative estimates. These micro-level
case studies wi 11 be reviewed in Section IV.
In Section V, I shall briefly comment on
the policy of poverty alleviation in the
state of West Bengal, particularly in the
context of the market friendly economic
reforms currently pursued in India. Sec
tion VI contains concluding observations.
Some general observations may be made
at the outset about the relative backward
ness of the state of West Bengal in India.
The level and the rate of growth of per
capita SDP of West Bengal consistently
lags behind the per capita GDP for India,
both measured at 1980-81 prices. The trend
growth rate of SDP of West Bengal during
1970-71 to 1988-89 is 3.24 per cent
compared to 4.17 per cent trend growth
rate of India’s GDPduring the same period.
In fact, the percentage deviations of per
capita SDP of West Bengal from the per
capita GDP for India as a whole, consis
tently widened during the decades of 1970s
and 1980s, indicating clearly relative
deceleration in West Bengal’s standard of
li ving vis-a-vis the all-India average. West
Bengal's reliance on the primary sector for
delivering increments in output has been
much greater compared to the rest of India
and the contribution of the industrial sector
much lower. The relative growth perfor
mance in the industries in the state has
been low compared to India as a whole,
especially since the 1970s, and worsened
since the late 1980s. The only redeeming
feature in the decade of 1980s for the state
of West Bengal has been the spurt in the
November 21, 1998
trend growth rate of agricultural produc
tivity much ahead of the all-India growth
rate. This rise was accompanied by effec
tive distribution of surplus agricultural
land among the landless labourers, thus
raising the share of landholding of mar
ginal and small farmers in the state com
pared to the rest of India. Although West
Bengal’s performance is much better than
the national averages with respect to male
and female literacy rates, its relative rank
ing among the states of India has declined.
The proportions of poor to total popula
tion in rural and urban areas of the state
are higher than the corresponding all-India
averages. The issues of poverty and depriva
tion are critical both for political economy
discourses as well as for policy-making.
II
Aggregative View
Poverty is a complex phenomenon, with
multiple dimensions, which cannot be
captured by a single definition applicable
to all societies and regions at all times.
According to Ravallion (1994), poverty
can be said to exist in a given society when
individuals/households do not attain a level
of material well-being deemed to consti
tute a reasonable minimum by the stan
dards of that society. Thus poverty is a
phenomenon of relative deprivation where
people arc regarded as poor if they do not
have access to diets, comforts and other
entitlements that arc customary in that
society. Therefore, although for some
specific purposes, we may define absolute
poverty corresponding to a specific abso
lute or average norm, the phenomenon in
all its dimensions has relative connota
tions. The inherent pluralism in the con
cept of poverty has given rise to differ
ences in conceptualisation and measure
ment of the deprivations which have not
only income, but also social, cultural.
political and even physiological dimen
sions, relevant for agiven society or region.
Attempts to quantify the incidence of
3003
povcrtv involve a study of the level and
pallet it of indiv iduals' personal consump
tion as well aslheir access tosocial benefits.
For the practical purpose of segregating
the poor front the non-poor. it would be
useful to locus on a single criterion rather
than multiple criteria, although from an
analytical and conceptual point of view,
one should adopt a multi-dimensional
approach. In what follows. I preferto avoid
use of composite indices, because weight
ing arrangements for the construction of
such a composite index generally remain
ad hoc. I shall concentrate on income/
expenditure deprivation aspect of poverty.
The empirical literature on poverty in
India is vast, and has covered many dimen
sions and generated a lot of controversy.
The major data base has been provided
by the household consumer expenditure
surveys conducted by the National Sample
Survey Organisation (NSSO). supple
mented by other sources like the data
generated by the Central Statistical
Organisation. Planning Commission, and
reports of various ministries of the central
and the state governments. The first major
issue has been the choice of the poverty
line to demarcate the poor from the non
poor. The exact estimates of the incidence
of poverty depend on the location of the
poverty line and the methods and norms
underlying it. The idea of a basic mini
mum calorie requirement and the estima
tion of the cost of the diet that satisfy
these requirements were the subjects of
intense debate among researchers during
1960s and 1970s. While Bardhan (1970)
and Minhas (1968) differed on the basic
elements of the consumption basket for
the poor in both rural and urban areas, their
estimates of rural and urban poverty in
India did not differ more than 5 per cent,
and were stable during the decadeof 1960s
Bardhan (1973) also obtained estimates of
inter-state variations in the incidence of
rural poverty for 1960-61 and 1967-68.
and found that between 1960-61 and
1967-68. the rural head count ratio in
West Bengal exhibited a sharp jump from
22 per cent to 74 per cent, while the all
India proportion rose from 38 per cent to
53 percent during the same period. It may
be noted that a sharp jump was noticed
for most of the states by Bardhan (1973),'
and this was mainly due to his use of all
India CP1AL. and all-India consumption
basket as the basis of inter-state variations
in rural poverty. Among other studies
attempting to estimate statewise poverty
indices forthe rural areas, one may mention
Vaidyanathan (1974) and Bhatty (1974).
Dandekar and Rath (1971) had used the
calorie norm of 2,250 per capita per day
3004
for rural and 2,100 per capita per day for
urban population io estimate the poverty
line and the poverty percentages for the
country as a whole The debate following
their work raised the valid point of non
equivalence between an average diet and
a minimum diet, and the need to take into
account bot h i ntcr-personal as well as i ntrapersonal variation in dietary pattern de
pending upon age. sex, climate, food habit
and more importantly the type of work one
is engaged in to estimate the extent of
malnutrition and poverty which are two
related hut distinct phenomena. Their study
look the country as a whole as the unit
and did not contain any specific estimates
for the stales.
Nayyar (1991) attempted loestimate the
incidence of rural poverty in different states
of India for 1961-62. 1970-71 and 197778. She used two alternative calorie norms
of 2,200 and 2.000 per capita per day with
NSS data on consumption of different
food items fordiffcrcntcxpenditurc classes
in rural India. She used the consumer's
price index for agricultural labourers
(CPIAL) for different states, to estimate
the head count ratio (HC), poverty gap
ratio (PG) and Sen's P-mcasurc for rural
areas of different states and India as a
whole. Her method is similar to that of
Dandekar and Rath in the sense that she
used calorie norm as the basis of her
estimates of rural poverty, but she ex
tended the analysis to state-level disaggre
gation. Table I summarises her estimates
of the incidence of rural poverty for West
Bengal and India for the three years
mentioned above. It is clear that her poverty
estimates arc sensitive to alternative
calorie norms chosen to compute the
poverty lines, and West Bengal on the
whole exhibited a higher HC ratio and
greater severity of rural poverty compared
to all-India averages during the period.
In a country as diverse and large as
India, the use of an all-India poverty line
can be misleading. For meaningful esti
mates, one needs to consider the slate
specific diclary patterns and slate-specific
retail prices faced by different fractilc
groups of income distribution io arrive al
state-specific poverty lines. The inter
temporal variations in these poverty lines
provide a chic to success or otherwise of
various poverty alleviation programmes
introduced by the government from time
to time. This was first suggested in a
seminal paper by Monlek Ahluwalia
(1978) and extended in 1986 in response
to Dharam Narain’squery about the impact
of price variations on poverty estimates.
He adopted the per capita consumer expenditurcofRs 15 per month (for 30 days)
at 1960-61 prices adjusted with the help
of CPIAL as poverty line for estimating
the trends in the incidence of rural poverty
(both HC ratio and Sen’s P-measure)during
the period 1956-57 to 1973-74 for India
as a whole and for the different states. He
found that the percentage of population
below poverty line declined through the
1950s. rose to a peak in 1967-68, and then
declined substantially though in an un
even manner, through the 1970s. and a
similar pattern was discernible forthe Sen
index as well. His time-series regression
analysis revealed that (a) rural poverty
was inversely related to agricultural in
come per head in rural areas, and this
relationship was found to be strengthened
when agricultural income was lagged, and
(b) there was a positive relationship be
tween poverty indices and price indices.
but the absolute size of the response
coefficient of price indices was found to
have been much moderated when lagged
agricultural income was included. Regard
ing state-level estimates of rural poverty
(both HC as well as Sen’s measure),
Ahluwalia (1978) derived two types of
estimates, one based on all-India dietary
norm but state-specific retail prices based
on Chatterjee and Bhatlaeharya (1974)
estimates of price relatives, and another
based on state-specific dietary norms
coupled with slate-specific retail prices as
was estimated by state-specific CPIALs.
Tabi e I ■ Estimates of Rural Poverty os the Basis or Alternative Calorie Norms
Calorie Norm
2,200 Per Capita Per Day
2,000 Per Capita Per Day
HC
PG
P-Measure
P-Measure
HC
PG
(Per Cent)
(Per Cent)
1961-62
West Bengal
All-India
1970-71
West Bengal
All-India
1977-78
West Bengal
All-India
70 92
33 98
0 28
0.25
0.25
0.12
48.33
21 14
0.20
0.32
0.14
0.07
79.56
56.64
0 38
0.27
0 39
0.22
66.26
41.84
0.20
027
0.28
0 15
72.77
54.64
0.36
0.32
0.34
0.22
62.88
42.52
0.21
0.23
0.26
0 14
Simrce: Nayyar (1991). Chapter 2.
Economic and Political Weekly
November 21. 1998
His second estimate taking into account
inter-state variations in both the dietary
norms as well as prices is a significant
contribution to our poverty studies, be
cause it incorporates all possible proxi
mate causes of variation in poverty
estimates across states. His regression
analysis at the state level confirmed his
hypotheses, but the significance of the
relations differed across the states.
Gaiha (1989) extended Ahluwalia's
analysis by developing a logistic model,
where
,
I H, 1
t Sen Index I
loeUOO-H,/
----- andlo" ll----------were regressed
Sen Index/
on the index of agricultural productivity
(IAPP) and fluctuations of CPIAL from
its trend values (FCPIAL) for all years for
the states. While Gaiha could not find
uniformity In the response pattern of
those variables across the states, his
findings on West Bengal is worth noting.
The co-efficient of LAPP was found to be
positive but weakly significant, and (hat
of FCPIAL was positive and strongly
significant, but there was a rising residual
time trend which attested to the Bardhan's
(1985) hypothesis that agricultural growth
in West Bengal is immiserising. Gaiha
found similar results with the Sen index
as well, with the difference that the co
efficients of IAPP were also statistically
significant. He concluded that West
Bengal constituted a special case inas
much as agricultural growth reinforced
the immiserisation resulting from price
fluctuations and the factors underlying the
time trend. It may be noted that beyond
1974. NSS has been conducting quin
quennial surveys of consumer expendi
tures on the basis of large samples of
households, while small or thin sample
size is relied upon for consumer expen
diture surveys for periods in between, thus
making proper time-series studies on
poverty difficult, and forcing scholars to
make ad hoc adjustments. We shall return
to this point later.
Planning Commission Estimates
We now move to the estimates of the
incidence of poverty in rural and urban
areas in West Bengal and India made by
the Planning Commission from time to
lime. Here several approaches are discern
ible: (a) use of all-India poverty line for
states, and their updating for inflation
adjustment over the period, (b) use of
slate-specific poverty lines on the basis of
Expert Group methodology and (c) the
consensus approach as is used currently
(also denoted as modified Expert Group
method). Tables 2 and 3 present the
Economic and Political Weekly
percentage and number of poor in rural showing the dramatic success of the
and urban West Bengal according to the government in poverty alleviation, as both
Planning Commission and Expert Group the percentage and number of poor people
methodology, and Table 4 presents the in West Bengal and India as a whole show
estimates according to modified Expert dramatic decline during the period, de
Group methodology, which is being spite a high population growth and dis
adopted by the Planning Commission as parate movement in relative prices of food
official estimates from March 11, 1997. and other articles in the states including
It may be noted that the original Plan West Bengal. This procedure is not only
ning Commission methodology uses all arbitrary and defective because it sup
India poverty linesofRs49.1 and Rs 56.60 presses facts deliberately, but it also
per capita per month respectively for exposes the hollowness of poverty studies
rural and urban areas at 1973-74 prices in our country since the early 1960s.
The Expert Group Methodology, popu
corresponding to 2,400 and 2,100 calorics
norm per capita per day for rural and urban larly known as the Lakdawala Committee
areas, with the help of NSS 28th round estimates, has made a number of refine
(1973-74) consumer expenditure data. It ments in estimating the percentage and
has updated the national poverty lines for number of poor across the stales and India.
subsequent years with the help of whole First of all. the Expert Group has consid
sale prices and GDP deflator for private ered the valuation of the national con
consumption estimated from the National sumption basket corresponding to all-India
Accounts Statistics, and assumes that the calorie norms implicit in the poverty line
price vector of the consumption basket of Rs 49 and Rs 56.6 per capita per month
implicit in the estimated all-India poverty at 1973-74 prices as used by the Planning
lines for 1973-74 and its movement over Commission for rural and urban areas
lime is identical across all slates and for respectively with the help of state-specific
rural and urban areas. Such a restrictive prices to estimate the slate-specific poverty
assumption of using all-India poverty lines lines. These state-specific poverty lines
and all-India deflators for estimating the are plugged into state-specific NSS con
incidence of poverty across the states was sumption expenditure distributions to
perhaps motivated by the urgency of derive the stalewise head count ratios for
Table 2: Percentage or Poor in West Bengal ano India
Planning Commission Methodology
_____ Expert Group Methodology______
West Bengal
India
West Bengal
”India
RU
CRUC
RU
CRU
C
Year
1972-73
1977-78
1983
1987-88
1993-94
64.00
53.16
35 84
24.73
13.34
35.90
30.35
19.40
16 44
8.32
56.80
47.30
31.45
22.49
11.94
54.10
46.13
32.73
28.37
19.24
41 20
33.82
21.68
16.82
10.11
51.50
43.40
30.08
25.49
16.82
73.16
68.34
63.05
48.30
40.80
34.50
38.71
32.21
32.84
22.51
63.39
60.65
54.72
43.99
35 69
56.44
53.07
45.61
39.06
37.27
49.23
47.40
42.15
40.12
33.66
54.93
5181
44.76
39 34
36 31
Source: Planning Commission, 1993; and Malhotra, 1997.
Table 3: Number of Poor in West Bengal
(In lakh)
Year
Planning Commission Methodology
R
U
C
1972-73
1977-78
1983
1987-88
1993-94
220.90
20097
152 68
114.37
68.64
41.60
39.68
30.07
28 24
16.57
Expert Group Methodology
C
R
U
257.96
259.69
266.65
219.09
209.90
262.50
240.65
182.75
142.60
85.21
41.14
51.55
50.45
57.63
44.86
299.10
311.24
317.10
276.72
254.76
Source: Planning Commission, 1993, and Malhotra. 1997
Table 4: Percentage and Number of Poor by Modified Expert Grout Methodology
Year
R
(Per
Cent)
West Bengal
U
C
(Per
(Per
Cent)
Cent)
1972-73
1977-78
1983
1987-88
1993-94
73.14
68.34
63.05
48.30
40.80
34.67
38.20
32.32
35.08
22.41
63.43
60.52
54.85
44.72
35.66
Number of
Poor
(Lakhs)
R
(Per
Cent)
India
U
C
(Per
(Per
Cent)
Cent)
299.30
310.57
318.69
283.61
254.56
56.44
53.07
45.65
39.09
37.27
49.01
45.24
40.79
38.20
32.36
54.88
51.32
44.48
38.86
35.97
Total
Number of
Poor(Lakhs)
3213.36
3288.95
3228.97
3070.49
3203.68
Source: Malhotra 1997.
November 21. 1998
3005
rural and urban areas. Secondly, the group
has added a federal dimension to the
estimation of poverty when the all-India
poverty proportions for rural and urban
areas are derived as a ratio of aggregate
statewise number of poor to the total all
India population for rural and urban areas
respectively, and not as a sequel to using
all-India poverty lines on the all-India
NSS consumption expenditure distribu
tion adjusted for parity with the national
accounts statistics (NAS) estimates of
private consumption expenditure. Thus
the Expert Group has used statewise NSS
consumption expenditure distribution
along with state-specific poverty lines in
order to arrive as poverty percentages for
different states. Thirdly, in order to arrive
at state-specific poverty lines correspond
ing to the baseline consumption basket for
1973-74. they have estimated statewise
price differentials separately for rural and
urban areas with the help of specially
constructed cost of living indices. The
CPIAL with 1960-61 as base year for the
states is used for rural cost of living in
dices. while a simple average of consumer
price index of urban non-manual employ
ees (CPIUE) and consumer price index for
industrial workers (CPIIW) with 1960-61
as base is used to construct the urban cost
of living indices for the slates. As per the
suggestionsofMinhaset al (1988). weights
corresponding to the consumption pattern
of middle 40 per cent of the rural popu
lation. and about middle 42 per cent of
urban population by expenditure strata for
1973-74 have been used to construct state
specific consumer price index for middle
rural population (CPIMR) and Consumer
Price Index for middle urban population
(CPIMU) for 1973-74 with 1960-61 as
base, and using Fisher's index and state
wise price differentials constructed by
Chatterjee and Bhattacharya (1974). the
Expert Group has estimated the price
differentials across states relative to all
India for rural and urban areas respec
tively to arrive at state-specific poverty
lines for 1973-74. and updating them for
subsequent years with the help of CPIMR
and CPIMU for the corresponding years.
The state-specific poverty lines for each
year and the state-specific unadjusted NSS
per capita consumption expenditure dis
tributions enabled the Expert Group to
arrive at state-specific rural and urban
head count ratios. By using the statewise
population figures of Registrar General of
Census for rural and urban areas on the
poverty proportions of states, the total
number of poor in each state for each of
the years have been calculated, before
aggregating them to derive the number
3006
and proportions of poor people in the rural for different states. The Expert Group did
and urban areas for India as a whole. It not differentiate between state-specific
may be noted that the Modified Expert dietary norms, but made detailed inves
Group Methodology that is now adopted tigation into the construction of state
as the official consensus approach of the specific prices. Since differential and
Planning Commission differs from the intertemporal information about state
Expert Group method described above on specific diets arc difficult to obtain, a
two counts: (a) it drops the CPIUE and better way could be to do away with the
uses only CPIIW for updating the urban roundabout ways of estimating the state
poverty lines across states, and thus the specific poverty lines by taking the state
estimates of urban head court ratio are specific mean per capita total (food + non
slightly adjusted in all the states; and (b) food) consumption expenditure (MPCE)
it uses revised population figures so that as the cut-off levels to differentiate the
the number of poor people in rural areas poor from the non-poor for rural and urban
of the states arc slightly different even if areas of each state of India separately for
the rural head count ratios arc identical for different time periods. The variations in
the Expert Group and the Modified Expert MPCE include variations in dietary norms,
Group estimates.
consumption basket, and retail prices for
We are now in a position to reflect on that state on an average over the period,
the state of the poor in West Bengal. Both and if the state-specific MPCE is chosen
the percentage of poor and the number of as the poverty cut-off level for a particular
poor in rural West Bengal has declined state, then we need not bother about base
sharply during the 1980s as per the Expert period adjustments in consumption basket
Group and Modified Expert Group esti as well as the retail prices. Such a proce
mates, and this decline is largely due to dure may be objected to on grounds of lack
a combination of technology-induced of comparability across states from a
•productivity upsurge in agricultural pro common reference point. A common ref
duction, and institutional reforms like erence point is however meaningless when
operation barga, land redistribution and we try to capture variations in deprivation
decentralised planning through elected levels in different states, and the purpose
panchayats, which ensures equitable dis of poverty alleviation measures in a given
tribution of the gains of growth across state is to alleviate those who fail to
population groups. So far as number of command goods and services that an
urban poors is concerned, there seems to
Table 5: MPCE, HC Ratio, Gin, and Sen's
be an inverted U pattern between 1973-74
Measure at 1960-61 Prices
and 1993-94, and this suggests the
relative lack of effectiveness of urban
Rural
Rank
Urban Rank
poverty alleviation schemes, and slacken
in Indiai
in India
ing of industrial growth in the state,
A: MPCE
particularly after the mid-1980s. But the
1961-62
19.83
3
38.00
15
overall picture in poverty reduction in
2
1968-69
14
14.93
31.30
West Bengal appears to be much better
1973-74
17.21
5
37.39
16
compared to all-India scenario, and this
1977-78
4
18.46
31.91
13
1983
19.92
35.77
16
suggests that in West Bengal, despite
5
4
1986-87
22.90
38.43
15
limitations at the micro-level, the poor
have benefited more than the non-poor B- HC Ratio
1961-62
70.92
1
62
97
13
from the developmental programmes
1968-69
17
59.21
61.45
13
undertaken by the Left Front government
1973-74
63.90
12
62.20
8
for the last 21 years.
1977-78
66.50
10
63.90
12
Derivation of a state-specific poverty
2
1983
62 20
14
67.00
9
line requires information about (a) state
1986-87
60.40
69,90
18
specific dietary norm and the composition C: GIN1
10
1961-62 0.2251
14
0.3246
of the basket that satisfies this norm, (b)
1968-69 0.2277
14
0 2810
16
state-specific retail prices of food articles
1973-74 0.2960
5
0.3170
3
to which people belonging to different
1977-78 0.2920
0.3240
12
12
expenditure classes arc exposed, (c) state
1983
0.2840
3
9
0.3350
specific distribution of consumption ex
1986-87 0.2370
4
18
0.3320
penditure. and (cl) state-specific provision D:P-measure
of public goods and free goods which arc
1961-62 0.3244
12
0.4018
10
1968-69 0.2670
16
enjoyed by people in rural and urban areas
15
0.3593
1973-74 0.3573
3
0.3915
6
of the given state. While information about
1977-78 0.3555
12
11
0.3953
(d) is difficult to obtain and evaluate,
1983
0.3418
9
0.4211
5
information about (c) is provided by the
1986-87 0.2826
2
16
0.4597
NSS consumption expenditure survey data
Economic and Political Weekly
November 21, 1998
average person in that state is entitled to
the prevailing price income configurations
of that state. A person is deemed as poor
in West Bengal if he faces entitlement
failure compared to what others in that
state, on an average, are capable of achiev
ing. and not others, pooras well non-poor.
in other parts of the country.
Chatterjee and Bhattacharya (1997) have
estimated HC ratio. Gini-eoefficient and
Sen’s P-measure for rural and urban
areas of different states of India taking the
state-specific MPCE for rural and urban
West Bengal at 1960-61 prices as the cut
off level during the period 1961-87 with
the help of NSS quinquennial consump
tion expenditure distributions for the
state. Their estimates of HC ratio, Ginieoefficient and Sen's P measures for rural
and urban areas of West Bengal arc pre
sented in Table 5. Certain interesting
features about the incidence and severity
of poverty in West Bengal are evident
from Table 5. They are: (a) West Bengal’s
rank among all states of India both in rural
and urban areas in terms of MPCE has
remained more or less unchanged between
1961-62 and 1986-87: (b) rural HC ratio
for West Bengal registered more than 10
percentage points decline during the pe
riod. hut the urban HC ratio went up
significantly. As a result. West Bengal
which was first among Indian states in
terms of rural poverty in 1961-62. ranked
only 18th in 1986-87. whereas its 13th
rank in the incidence of urban poverty
deteriorated considerably and it ranked
second in urban HC ratio in 1986-87;
(c) the extent of inequality in rural con
sumption distribution in West Bengal has
gone up slightly during the period, but the
rank of the rural Gini has declined. A
completely reverse picture emerges with
respect toconsumption inequality rankings
in urban areas of West Bengal (it has gone
up from 10 to 4), although the Gini-coefficient in urban areas has registered only
mild increment over the years; (d) the
severity of poverty as measured by Sen’s
P-measure has appreciably dismissed in
rural West Bengal, but in urban areas of
the state, it has gone up significantly,
pushing its rank to 2 in 1986-87 compared
to 10 in 1961-62. It thus appears that the
urban areas of West Bengal pose serious
and intricate challenges to policy-makers'
aim to diminish I he incidence and severity
of poverty in West Bengal. But the relative
success in poverty alleviation in rural areas
of West Bengal is clearly a lesson for India
as a whole, although the head count ratios
for both rural and urban areas as estimated
on the basis of MPCE as the cut-off level
are much higher than those arrived at by
Economic and Political Weekly
the Expert Group, or the modified con
sensus approach now being accepted by
the Planning Commission.
Determinantes of Poverty Indices
The rationale of the above relationships
has been to test the extent to which the
relative importance of the agricultural and
the non-agricultural sectors in the state
domestic product of West Bengal and the
per capita income generated in the two
sectors respectively have a bearing on the
behaviour of the incidence of poverty in
rural and urban areas of the state over the
period. It is true that structural shifts in
the economy in the form of regime switch
ing make the linear time series estimates
inappropriate. To avoid this problem, the
sectoral shares in the SDP, rather than the
annual growth rates have been used as
explanatory variables to ascertain the nature
of associations, and not causality, bet
ween the variables. The estimated equa
tions for the rural and urban areas of West
Bengal respectively during 1961-87 are:
What could he the proximate determi
nants ol the behaviour of poverty indices
over time? This is an important yet tricky
question whose solution requires time
series information on the relevant eco
nomic variables As a first step, one would
have to construct the time-series of pov
erty indices - HC. Gini or P-measures,
which again would require mixing esti
mates based on large sample quinquennial
surveys of NSSO with its annual expen
diture survey data based on small/thin
samples. This poses the danger of biased
estimates based on too thin sample sizes
being used for prediction, as N S lyenger
(1997) has rightly pointed out. There HCr = 85.761 - 0.50920SHAG
seems to be no way out but to use the
(14.00) (-3.340)
estimates based on two types of samples.
-0.0082 PCYAG
Chatterjee and Bhattacharya (1997) have
(-4.13)
estimated the HC ratio. Gini-cocfficient R2 = 0.454; DW = 0.710; F = 9.58.
and P-measure for rural and urban areas
PR = 0.3951 - 0.0020SHAG
of the states for periods in between the
(4.52) (-0.956)
quinquennial survey years of NSS with
-0.00001 PCYAG
the help of annual expenditure data.
(-0.358)
There is a debate in the literature on
R2 = 0.675; DW = 0.30; F = 13.932.
poverty studies in India regarding the role
ofgrowth and distribution factors. Whether HCu = 53.202 - 0.1336 SHNAG
(14.00) (-2.02)
the 'trickle down’ effect of growth is
-0.0040PCYNAG
sufficient to eradicate poverty or whether
(-9.72)
state-sponsored distribution mechanism
is necessary for the purpose, has been R2 = 0.901; DW = 0.30; F = 104.51.
debated by our policy-makers for long.
It is clear that the share of agricultural
According to Ravallion and Dutt (1996a, sector and per capita income in the agri
1996b), economic growth factors domi cultural sector have significant negative
nate the distribution factors in diminish impact on rural HC ratio for West Bengal.
ing the incidence of rural and urban poverty but the effect is not significant on Sen’s
in India, and further that growth in the P-measure for rural areas of the state during
1961-87. For the urban areas of West
agricultural sector is seen to have exerted
significant diminuilive impact on the Bengal, both the HC ratio and P-measure
incidence of rural poverty in India as a have been influenced significantly in the
whole. To check the relevance of the above inverse direction by the share of the nonmentioned hypothesis of Ravillion and agricultural sector and per capita income
Dutt, Chatterjee and Bhattacharya (1997) from the non-agricultural (secondary plus
havccstimatedthe following relationship.'; tertiary) sector. Thus growth effects ap
HCR=o^+a|SHAG+a2PCYAG+ul ..(I) pear to be significant in poverty reduction
in both rural and urban areas of West
PR = Po + P, SHAG + p, PCYAC + nt ...(2) Bengal, but the effects appear weak on the
where HCR is the estimated HC ratio for severity of poverty in rural areas and
rural areas. SHAG is the share of agricul significant on the severity of poverty in
ture in the state domestic product (SDP), urban areas. Further investigations into
and PCYAG is the per capita income such relationships are required to test the
generated in the agricultural sector. For significance of the success of poverty
the urban areas, non-agricullure sector is alleviation endeavours in the stale. With
used in place of agriculture sector such suitable specifications, one may test the
that the estimable equations for the urban importanceof movementsin inter-sectoral
terms of trade, technological progress.
areas are:
public investment in infrastructures and
I ICu = cto+a, SI 1NAG+ a, PCYNAG +u,...(3) also the role of institutional reforms in
I’ll = 0O + 0, SHNAG + 0, PCYNAG + u, ...(4) rural and urbqn areas as are undertaken
November 21. 1998
3007
by the state government since 1977. One
may also surmise the relationship, if any.
between the decline of the industrial sec
tor in the state during the 1980s. and the
behaviour of urban poverty indices. All
these would, however, require suitable time
series information on the relevant variables,
which at present, are difficult to obtain.
The aggregative view surveyed in this
section suggested a mixed picture of
poverty in West Bengal. The arbitrary
choice of the poverty lines in various all
India and inter-state studies have some
times created false impressions, or erro
neous signals to the policy planners in the
state as well as in India as a whole. On
the whole, however, the decade of 1980s
saw significant reductions in the percent
age of rural proverty in the state, although
no such general clai m cou Id be made about
the incidence of urban poverty in West
Bengal. We need to supplement the
aggregative view with the insights of some
studies exclusively made for the state of
West Bengal, followed by selective mi
cro-studies which would help us focus on
some serious problem areas.
Broadly speaking, available studies can
be categorised into two groups, namely,
those dealing with rural poverty and those
dealing with urban poverty in West Ben
gal. The Planning Advisory' Board (1979)
had attempted to estimate the percentage
of poor in rural West Bengal during
1963-64 and 1973-74, on the lines of
Dandekar and Rath methodology, while
Dasgupta (1989) attempted toestimate the
extent of urban poverty in the state during
1966-67 and 1973-74. Jose (1984) used
alternative indicators like land distribu
tion. pattern of landholding by occupa
tional groups, availability and actual con
sumption of cereals, real wage rates of
agricultural workers, and access to health
and educational facilities in the rural areas
during 1960s and 1970s. Bardhan (1987)
has used the NSS survey on employment
and unemployment in urban areas of West
Bengal during 1977-78 to relate the urban
poverty indices for various occupation
groups to the different socio-economic
characteristics and found that household's
level of living was higher in districts
belonging to main urban agglomeration
and where the importance of manufactur
ing and repair services is larger and in
districts where the average foodgrains yield
is larger compared to other districts.
Ill
The 1970s and the 1980s
We have mentioned the importance of
using state-specific dietary norms along
with state-specific retail prices to arrive
at the poverty lines for rural and urban
areas of a state. Fortunately for West
Bengal, the Calcutta-based All-lndia
Institute of Health and Hygiene has es
timated an average Bengali diet for rural
and urban areas of the state which conform
to intakes of 2,400 calories per capita per
day for rural areas and 2.100 calories per
capita per day for urban areas. These
dietary norms represent the average
requirements of sedentary male workers
in rural and urban areas of the state
repcctively, and not any nutritional mini
mum. Variations across gender and other
types of work arc ignored to simplify
computation.
With the help of these average Bengali
diets for rural and urban areas of West
Bengal, Chatterjee (1991) has estimated
the incidence of rural and urban poverty
in the state during 1970s and 1980s, using
the quinquennial consumption expendi
ture data of the NSSO. The estimates of
HC ratio, Gini-coefficient, poverty gap
index and Sen’s P measure for rural and
urban areas of West Bengal are presented
in Tables 6 and 7, respectively.
The average Bengali diet is biased to
ward cereals, particularly rice, includes
milk, and shows a preference for fish as
a source of protein. For both rural and
urban areas, the cost of the recommended
diet is calculated using the consumer retail
prices made available by the National
Sample Survey Organisation (NSSO) and
the Bureau of Applied Economics and
Statistics (BAES), respectively. Having
calculated the monthly cost of the food
intake, all that is needed to arrive at the
poverty line is to add an appropriate non
food component. One minus the average
Engel ratio for all classes has been as
sumed to be the required proportion of
non-food expenditure to total expendi
ture. The average Engel ratio is observed
to be around 73 per cent for rural areas
and around 61 per cent for urban areas for
most of the years considered in the study.
The formula for the poverty line (per capita
per month) is therefore the monthly per
capita cost of the minimum average diet
divided by 0.73 for rural areas and by 0.61
for urban areas. Rents have been excluded
fromthecalculationofnon-foodconsumption expenditure for urban areas because
of estimation problems. The ultra poverty
line is defined as 80 per cent of the usual
poverty line.
Rural Poverty
Regarding the estimates of rural poverty
presented in Table 6. some observations
may be made. First, the very high rural
HC estimate for 1973-74 (nearly 80 per
cent) is partly due to the prevalence of
drought in that year. By 1974, the CPIAL
food index had risen by 55.5 per cent
compared to 1972. This HC estimate is
therefore clearly abnormal. The year
1977-78 saw heavy rainfall and floods in
many parts of the state. In 1983, the rainfall
was moderate to low, but after 1983 West
Bengal experienced consistently good
rainfall. The HC estimates for 1973-74
and 1977-78 may therefore be interpreted
as poverty during the worst of times. The
figure for 1986-87 similarly represents
poverty during relatively better times.
Second, the poverty lines and their
associated HC estimates for both rural and
urban areas may be compared with the
mean expenditures (at current prices) and
Table 6’ Poverty and Inequality in Rural West Bengal. Selected Years
1973-74
1977-78
1983
1986-87
1988-89
Poverty line per capita per month (Rs)
Head count ratio (per cent)
62.95
79.42
Poverty gap measures
0.572*
Sen’s P-measurc
0.454
72.29
76.85
(-0.81)
0.497
((-3.2)
0.383
(-3.90)
124.51
74.95
(-0.50)
0.474
(0.92)
0.366
(-0.89)
139.14
60.50
(-6.43)
0.343
(-9.21)
0.229
(-12.47)
156.33
53.10
(-6.11)
0.302
(-5.98)
0.196
(-7.21)
Gini-coefficient
No intra-class inequality
0.300
Intra-class inequality
0.305
0.296
(-0.33)
0.299
(-0.49)
64.70
(-0.76)
30 94
50.50
(1.6)
0.289
(-0.47)
0.293
(-0.40)
58.83
(-1 810
31.47
55.40
(1.9)
0.241
(-5.54)
0.247
(-5.23)
38 27
(11.65)
27.07
63.78
(5.04)
0.216
(-5.19)
0.230
(-3 44)
32.60
(-7.41)
25.00
68 98
(3.46)
59.27
64.00
104.59
62.16
139.02
60.41
169.98
61.23
Measures
Head count ratio for ultra poor (per cent)
66.74
Number of poor (millions)
Mean per capita consumption
expenditure at 1973-74 prices (Rs)
Percentage of population below mean
expenditure line
29.40
45.50
47.50
62.40
Note:
Figures in parentheses are the annual average growth rate since the previous observation.
Sourer: National Sample Survey Organisation (various years).
I
3008
Economic and Political Weekly
November 21, 1998
the percentages of people living below
them. For the years 1973-74, 1977-78 and
1983 even the average standard of living
fell below the poverty line in rural areas,
and this only reached the poverty line
level in 1986-87 in rural areas. A look at
the percentage changes in the rural areas
(Table 6) leads to the conclusion that
poverty has fallen significantly during the
two decades under consideration The
pattern, however, has not been uniform
Three sub-periods can be identified ac
cording to the rate of change of the poverty
indices: (a) the period of moderate to low
rates of fall, (b) a somewhat stationary
situation marks the period 1977-78 to
1983, and (c) the rate of fall is significant
and faster after 1983 compared to the
previous periods. The patterns of change
of the HC ratio and of the mean expen
diture (ME) at 1973-74 prices arc similar.
The correlation coefficient between the
annual average growth rate of the ME and
the annual average growth rate of HC is
a high - 0.925 The Indian government
started to implement its direct anti-pov
erty programmes after 1981. Whether the
relatively faster rate of fall in HC and rise
in ME in rural areas of the state after 1983
compared to the previous periods are due
to the direct attacks on poverty or to pure
economic growth is an open question. The
broad pattern of change in poverty indices
is. however, much more pronounced for
the poverty gap (PG) measure and Sen's
measure (P) than for the HC ratio. The
correlation coefficient between the annual
average growth rates of PG and HC is 0.99
and that between the growth rates of P and
HC is 0.993.
Although the rate of fall in the rural HC
ratio is negligible during 1973-78, the
rates of fall in PG and P are much more
significant. This implies that during this
period the average standard of living of
the rural poor went up (PG measure) and
the inequality among them (P measure)
went down. In other words, the intensity
of poverty fell even though the percentage
of poor people might not have changed
much. It is in this sense that we characterise
this period as a period of moderate to low
fall in the incidence of poverty. The period
between 1977-78 and 1983 is marked as
a stable period so far as the incidence of
rural poverty in the state is concerned
because both the percentage of poor people
and the intensity of poverty showed little
change. Poverty fell significantly both in
terms of the proportion of poor people and
intensity after 1983. The absolute number
of poor also fell. The pattern of change
in the HC of the ultra poor is roughly
similar to that of the HC of the poor, with
Economic and Political Weekly
one major difference, namely, the rate of
fall in the HC of the ultra poor seems to
be somewhat larger during 1977-78 to
1983 Once again the rate is especially
significant between 1983 and 1986-87.
On the whole, the situation of the ultra
poor in rural West Bengal seems to have
improved more than that of the poor after
1977-78.
A comparison of the two estimates of
the Gini-coefficient reveals that intra-class
inequality, although a small percentage of
overall inequality, has increased over the
years. Overall inequality has diminished
over the years, though not at a uniform
rate. The fastest rate of fall occurred during
1983 to 1986-87. From 1973-74 to 1983
the rate of fall was negligible. The cor
relation coefficient between the average
growth rates of ME and the Gini-coef
ficient is -0.99. Economic growth in the
rural areas may therefore be said to have
benefited the poor more than the non-poor
during the two decades, particularly after
1983.
The PG index, which may be interpreted
as the proportion of total expenditure that
must be transferred from the non-poor to
the poor, to raise the latter's consumption
upto the poverty line, tells a similar story.
For the years 1973-74, 1977-78 and 1983
the PG estimates were 57 per cent, 50 per
cent and 47 percent, respectively. Expen
diture of the non-poor however consti
tuted only about 39 per cent, 32 per cent
and 44 per cent, respectively, of total ex
penditure. This seems to suggest the view
that reduction in inequality during this
period is related to the implementation of
direct anti-poverty programmes in rural
areas. The increase in rural inequality in
many states between 1983 and 1986-87
may have been due to a shift of emphasis
from pure anti-poverty programmes to
ward growth-promoting policies.Thecontinued importance of the former in West
Bengal may explain why inequality fell
significantly after 1983.
Urban Poverty
The growth rales of the poverty indices
for urban areas of West Bengal (Table 7)
reveal once again three distinct phases of
change: (a) poverty increased during
1973-74 to 1977-78 ; (b) the indices fell
at a significant rate during 1977-78 to
1983; and (c) the fall was much less dis
cernible after 1983. The pattern of change
of urban poverty was thus quite contrary
to the pattern of change of rural poverty.
For the period 1973-74 to 1977-78, in
urban areas of the stale the mean expen
diture and poverty moved in the same
direction, although the rate of growth ME
was negligible. Thus, economic growth
was immiserising in urban areas. ME and
poverty indices moved in opposite direc
tions after 1977-78. The correlation co
efficient between the annual average rates
of growth of ME and HC is around -0.44
in urban areas, which is significantly
smaller than the corresponding figure for
rural areas. This suggests that the trickledown mechanism was more effective in
rural areas than in urban areas. In rural
areas, the presence of direct anti-poverty
programmes may have ensured that the
fruits of economic growth went largely to
the poor. The absence ofsuch programmes
in urban areas may have weakened the
trickle-down effect. Not only did the HC
ratio go up over 1973-74 to 1977-78, but
Table 7: Poverty and Inequality in Urban West Bengal, Selected Years
Measures
Poverty line per capita per month (Rs)
Head count ratio (per cent)
1986-87 1988-89
1973-74
1977-78
1983
69.51
53.00
90.00
58.22
(2.46)
0.315
(0.5&0
0.264
(063)
129.00
46.71
(-3.59)
0.226
(5.65)
0.182
(0.67)
175.51
45.15
(-1 11)
0.209
(-250)
0.183
(029)
188.42
44 19
(-1 06)
0.191
(-4.31)
0.178
(-.1.37)
0.328
(0.63)
0.333
(0.46)
53.61
(12 52)
6.00
81.57
(0.25)
0.339
(0.67)
0.350
(1.02)
30.77
(-8.52)
7.00
90.21
(2.12)
0 336
(-029)
0.372
(2.09)
30.80
(0.03)
7.17
101.20
(4.06)
0 330
(-0.89)
0.381
(1.21)
25.00
(■9.41)
7.24
101.53
(0.16)
97.13
63.70
169.95
64.81
242.62
67.62
268.67
66 rs
Poverty gap measures
0.308
Sen’s P-mcasurc
0.230
Gini coefficient
No intra-class inequality
0.320
Intra-class inequality
0.327
Head count ratio for ultra poor (per cent)
35.72
Number of poor (millions)
Mean per capita consumption expenditure
at 1973-74 prices (Rs)
Percentage of population below mean
expenditure line
3.80
80.76
80.76
64.00
A'vtr: Figures in parentheses are the annual average growth rale since the previous observation.
Source: National Sample Survey Organisation (various years).
November 21. 1998
3009
the average level of living of the poor (1973-74 for 1977-78, 1977-78 for 1983,
‘green revolution’ technology across the
also fell. The inequality among the urban and 1983 for 1986-87) remained un state. The onset of direct anti-poverty
poor, that is. the intensity of poverty, rose changed. whereas the expenditure of in programmes throughout India with mas
significantly. This clearly was the worst dividuals increased by the same percent sive public investment in rural asset/infraperiod for the urban poor. The best period age as the ME forthe total. The difference structurc creation also contributed to this
was between 1977-78 and 1983. when between the actual HC of the previous growth and prosperity in rural West Bengal
all these indices of poverty recorded the round and the simulated HC for the next and the consequent decline in the inci
fastest rates of decline. The period 1983 round is defined as the pure growth effect, dence and intensity of rural poverty,
to 1986-87 was somewhat mixed. Whereas because if distribution had remained particularly among the ultra-poor. The
the HC fell at a moderate rate and the unchanged, but mean expenditure had gone importance of the institutional reforms
average level of living rose, inequality up. then the simulated HC would have introduced by the Left Front government
among the poor worsened somewhat. The been the actual HC for the next period.
in West Bengal since 1981 in the form of
correlation coefficient between the rates The difference between the simulated HC registration of bargadars (share croppers),
of growth of PG and HC is 0.99. higher and the the actual HC for the same period distribution of surplus land among the
than that for rural areas. The correlation gives the distribution cffect.The results of landless, and the decentralisation of
coefficient between P and HC is 0.76, the above simulation exccrcise arc given administration through elected rural
which is less than in rural areas. The pattern in Table 8.
panchayats in the state, cannot be ignored
of change forthe urban ultra poor is roughly
Forthe rural areas for all the three periods, in this context, but the exact statistical
similar to that for the urban poor in gen the growth effects have been larger than decomposition of the effects of techno
eral. The ultra poor seemed to have been the distribution effects. This implies that logical change, institutional change and
more adversely affected during 1973-74 to for the poor in the rural areas growth of changes in public investment are dif
1977-78 and better off between 1977-78 promoting strategics were likely to have ficult to obtain. This may form the re
and 1983 and after 1986-87 than the urban made a greater impact on poverty than search agenda for the future. For furtl^k
poor in general. Unlike in rural areas, the redistributive strategies. During 1977-78 details, one may refer to Chatterjee. Rr^
absolute number of urban poor rose sub to 1983, distribution has partially offset and Bhattacharya (1991).
stantially during the two decades, with the the favourable impact of economic growth
To assess the extent of inter-district
worst change occurring once again be on rural poverty. From Table 7, the Gini- variations in the standard of living in West
tween 1973'74 and 1977-78.
coefficicnt is seen to have gone down over Bengal during the last two decades, the
Analysis of changes in urban inequality the period. The positive distribution effect first important question that we address
reveals some interesting features. First, isthusquiteunderstandablc. In urban areas is the nature and extent of interdistrict
inter-class inequality rose slightly during the distribution effect is larger than the disparities and the ranking of districts by
the two decades, although growth rates growth effect for the first period, which direct and indirect indicators of the stan
were positive up to 1983 and negative is not surprising as the growth rate during dard of living. Due to the absence of
thereafter. The rise is. however, more this period was negligible. During the first consumption expenditure data at the dis
marked for overall inequality, especially and last periods the distribution effects are trict level in rural areas, we have used such
after 1983. Thus, intra-class inequality' quite significant and positive, which means indirect indicators as agricultural produc
has become much more noticeable over that a worsened distribution has largely tivity office per hectare, the money wage
the years in urban areas. Inequality in destroyed the benefits of economic growth rate of agricultural labourers, and the per
general, and intra-class inequality in par for the poor. Inequality thus seems to capita rural district domestic product to
ticular, is higher in urban than in rural contribute more to poverty in urban areas rank the districts. For the urban areas we
areas. The correlation coefficient between than in rural areas. This calls for a rank the districts by the head count ratio
the average growth rates in the overall reorientation of distributive policies pur and the Gini-cocfficient, calculated from
Gini index and ME is 0.99 in urban West sued by the government from rural to the Family Budget Survey data of 1976-77.
Bengal Thus in urban areas, as compared urban areas.
Table 9 presents the ranking for rural and
to rural areas, growth seems to have been
Chatterjee (1991) has also found that urban districts of West Bengal.
accompanied by worsening inequality, there has been an acceleration of agricul
Certain interesting features emerge froi^
although in both cases, the situation of the tural productivity in West Bengal after Table 9. First, since average food produc
non-poor has improved more than that of 1983, which coincides with the period tivity per hectare is closely correlated with
the poor. The direct anti-poverty when the incidence of rural poverty in the poverty in rural West Bengal, the second
programmes seem to have played an state has registered the sharpest decline. column, which ranks the districts in de
important role especially in rural areas.
He argues that the diminution of rural scending order based on rice productivity
poverty since the early 1980s. was mainly per hectare, gives us an idea about the
Decomposition
due to the upsurge in agricultural produc districts which arc relatively poor, with
Chatterjee (1991) has also presented an tivity in the state during the period, which the poorest five being Jalpaiguri.
analytical decomposition of the changes was due to the adoption and spread of Darjeeling, Cooch Behar, West Dinajpur
in poverty over the different periods into
Table 8- Decomposition of Total Chance in Poverty into Growth ano Distribution Effects
changes due to pure growth (growth
Rural
Urban
effect) and those due to changes in distri
Actual Change Growth Distribution Actual Change Growth Distribution
bution (distribution effect) following Jain Period
in HC Ratio
Effect
Effect
in HC Ratio
Effect
Effect
and Tendulkar (1990) and Ravallion and
-0.80
-0 81
Dutt (1990) methodology. For each of the 1973-74 to 1977-78
-0003
2.46
2.82
-0.32
-0.50
-1.52
-1.12
-3.95
-2.27
1.89
years, 1977-78, 1983, and 1986-87, head 1977-78 to 1983
1983 to 1986-87
-4.67
-6.43
-2.04
-LI 1
-6.06
5.12
count ratios were simulated by assuming
that the distribution of the previous period Source: Computed from Tables 6 and 7.
3010
Economic and Political Weekly
November 21, 1998
and Purulia. Rice, which is the state's picture about the poverty scenario in the in 1976 to assess the actual performance
principal food crop, is used as a proxy lor state than can be derived from all-India of minimum needs programmes and the
total food because of the unavailability of or inter-state studies. The real difficulty provision pf health and education among
data on the latter. Second, since agricul in extending the various hypotheses about the poof, classified according to alterna
tural labourers constitute the single most rural and urban poverty in West Bengal tive socio-economic strata, Bhattacharya
important subclass of poor people, the for subsequent periods is the lack of cl al (1987) have undertaken a resurvey
third column ranks the districts by the comparable time-series information on of the three districts of West Bengal agricultural money wag? rate in descend major economic variables, and their dis Bardhaman, Birbhum and Purulia - to
ing order. Considerable variations exist in aggregation across the districts. The of cover the villages and households which
money wage rates across districts. Al ficial machinary on disaggregated data were covered by the NSS survey in its 27th
though no discernible trend in the co collection and their preservation in West round (1972-73) and 28th rounds. Both
efficient of variation has been obtained Bengal is rather weak, and beset with the studies point to the pathetic state of
over the years, the average value of this conceptual difficulties in theirestimations. social infrastructure in the state of West
is no less than 17 per cent. This finding As a result, state-level aggregative data on Bengal with very little improvement over
also complements the evidence provided poverty and deprivations in West Bengal the years, except for Bardhaman district
by the data on the effect of food produc need to be supplemented by micro-level which shows some remarkable improve
tivity on the existence of regional dispari case studies which focus on not only ments. This calls for serious policy inter
ties. The rankings of the districts in second income-based poverty and food depriva ventions as market forces and private
and third columns, however, do not match tions, but also give valuable qualitative initiative seem to have failed to distribute
to any great degree, suggesting the exist information about entitlement failures on the gains of growth in an equitable and
ence of other major influences on the a number of aspects like education, health, perceptible manner.
A recent study on the replicability of
agricultural wage rate apart from food shelter, credit availability, wage-employ
Productivity. The rank correlation co ment solutions, etc. In section IV wediscuss block based special public works schemes
efficient is only 0.51.
the findings of some of the major micro of ILO at Bundwan andManbazarll blocks
Neither of the rankings presented in the studies on poverty, deprivations and stan of the Purulia district in West Bengal by
Bagchi, Chatterjee, Chattopadhyay and
second and third columns ofTable 9 match dard of living in West Bengal.
Moitra (BCCM, 1995) reveals some in
well with those of the fourth column,
IV
teresting lessons about poverty alleviation
which is ranked on the basis of per capita
Micro Views
in West Bengal. Although inequality in
district domestic product in descending
order. The ranking of the fourth column
The aggregative view as dicussed above landholding in rural areas is considered
is. however, closer to the second than to basically gives average picture about the one of the main reasons for inequality in
the third: The rank correlation coefficient extent of poverty in the state, but the income distribution and abject poverty in
between the former is about 0.61, whereas distribution of the facilities or depriva rural areas, and the state government has
that between the latter is only 0.49. As we tions across socio-economic groups or initiated measures for effective redistribu
have argued in earlier sections, poverty in other dimensions of living standards could tion of surplus land among the landless
rural Bengal is primarily linked to low be gauged through micro-studies based on labourers, there was hardly much differ
agricultural productivity, and thus the field surveys/resurveys such that panel ence in the living standards of marginal
second and fourth columns should be given comparisonscould also be made. One may farmers and landless labourers in the
greater weightage in deciding which of the refer to two important studies in this state, both of whom were extremely poor.
districts are relatively backward in rural connection Maitra (1988) reports the But the successful completion of minor
areas. In the final analysis. Burdwan, results of a sample survey of households irrigation schemes of the ILO and effective
Howrah and Hooghly appear to be among m rural areas of West Bengal conducted distribution of water to the fields through
the most prosperous rural districts, while
Table 9: Ranking of Districts by Indicators of Standard of Living and of Poverty
AVestDinajpur, CoochBeharandJalpaiguri
Gini
Agricultural
Agricultural
Per Capita
Head Count
Ire among the poorest. In so far as urban Districts
Co-efficient
Productivity
Money
District Domestic
Ratio
poverty is concerned, the coefficient of
(Rice) Per Hectare
Wage
Product
variation of the HC ratio between the
districts was as high as 31 per cent. The Bankura
1
7
4
5
10
9
5
5
12
2
fifth and sixth columns of Table 9 arc Birbhum
1
13
6
5
I
ranked in ascending order of magnitude, Burdwan
na
7
na
na
1
and the rank correlation coefficient be Calcutta
13
15
Cooch Behar
14
12
13
tween the two columns is quite significant Darjeeling
2
14
9
3
10
at 0.66. Thus, those districts of West Bengal Hooghly
2
3
10
7
4
that suffer from higher levels of urban Howrah
5
6
1
3
9
4
15
8
3
11
poverty were also the ones with higher Jalpaiguri
10
9
7
12
8
levels of inequality. This corroborates the Maida
8
14
9
6
16
earlier result which indicates that in urban Midnapur
9
6
Murshidabad
10
13
4
areas of the state, poverty and inequality Nadia
9
7
8
3
13
go hand-in-hand and arc more important 24-Parganas
11
15
6
16
15
from the point of public policy than rural Purulia
12
14
15
11
11
16
West Dinajpur
16
16
16
15
deprivations.
The important point to note about fc'oies: na - Not applicable as Calcutta has no rural area/activity.
poverty studies specific to West Bengal Source: Family Budget Survey 1976-77 for urban estimates and State Statistical Abstract. 1977-89.
is that they reflect a much better and realistic
for information on rural variables.
Economic and Political Weekly
November 21. 1998
3011
field channels have enhanced the income
levelsofeven marginal fanners in Purulia
district, who, with assured irrigation and
better marketing networks, are now en
gaged in triple cropping and are diversi
fying their production base considerably
so much so that they now willingly con
tribute for the maintenance and extension
of irrigted water to distant fields within
the command areas of the Jorebundhs
created under the ILO schemes. The in
comes of the landless labourers have
however gone up only marginally with
extra off-season labour demand on the
field as a result of such irrigation exten
sions. Thus, public works and irrigation,
even where they were effectively admini
stered, have widened the inequality be
tween the poorest of the poors, with land
owners, however, marginal, gaining much
more than the landless. Such a process of
accentuation of inequality among the poor
through poverty alleviation schemes poses
a real challenge for the long-term
sustainability of these schemes. Secondly.
in order to sustain even a minimum con
sumption standard, the landless agricul
tural labourers are heavily dependent on
off-season and off-farm jobs under the
Jawahar Rojgar Yojana (JRY). Thus even
after so many years, schemes like JRY
Taele 10: Poverty Profiles of Households,
Kashipur Block. Purlua District, (ACRP
Pilot Project Area)
(Per cent)
Name of .Mouzas
Agard ih
Lakhipur
Chaka
Mekhyada
Shalava
Pakhariathal
Pabrahari
Chitra
Seja
Bhatin
Mirgipahari
Kashidih
Saharbera
Isanda
Kaliyada
Ledagora
Ichamara
Mehi
Jiara
Kharikagora
Bhalukgazar
Panja
Telaboni
Lajhna
Kusuingora
Sonathali
Kushjuri
Head Count Measure
36.04
74.07
50.00
69 6!
55.41
50.00
71.61
67.70
76.50
60.45
60.45
65.82
84.42
66.28
43.55
71.43
00 00
78.95
66.66
67.96
63.49
51.28
82.14
59.38
70.52
50.63
60.54
Note: The head count measure is the percentage
of households with an annual income
below Rs. 11,000.00 which is the poverty
line defined by the Planning Commission.
Source'. Estimated from the Survey Data.
3012
continue to be essential in rural poverty (villages) was conducted by Chatterjee
alleviration in West Bengal. All these and Moitra (1995). Table 10 shows that
suggest that the beneficiaries of rural in 7 out of the 27 mouzas under this
development and poverty alleviation survey, the poverty percentage exceeds
programmes in the state must be tapped the Purulia average, and in most of the
to finance the enormous expenditure re mouzas, this percentage exceeds the West
quired to uplift the bottom-most layers of Bengal average arrived at through any
rural .population in the state, and the method. This means that in some pockets
decentralised rural panchayats have to of the backward districts of the state, the
devise effective mechanisms for such benefits of poverty alleviation expendi
resource mobilisation and their deploy ture have so far eluded most of the rural
ment towards alleviating the ultra-poor on households, and West Bengal average
a sustainable basis. The Budwan experi figures are clearly underestimates of the
ment in Purulia is a pointer to this.
extent of deprivation in these villages. A
According to the estimates of the de look at the percentages of households
partment of rural development and indebted either by current deficit status
panchayats, government of West Bengal, (Table 11) or by outstanding debt status
70.70 per cent of the rural households in (Table 12) in the pilot project areas of the
the district of Purulia are below the pov Kashipur block of Purulia district as re
erty line of Rs 11,000 per annum atcurcnt vealed in the ACRP Benchmark Survey
prices during 1992-93. Thus Purulia is by Chatterjee and Moitra (1995), clearly
found to be the second poorest district in indicate that the incidence of indebtedness
West Bengal behind Darjeeling. This is by either status is heaviest among house
in sharp contrast to the West Bengal holds with zero operational holdings
average head count ratio of45.73 per cent, (ZOH) and with marginal operational
according to the same BPL (below the holdings (MRG) in that area. Poverty and
poverty line) survey. If one goes further indebtedness are two interrelated phenom
down up to the village level, the picture ena, and such a high incidence of indebted
becomes worse for some of the villages, ness among rural households clearly
whereas some other villages are found to indicates that something is seriously
be better off in terms of the incidence of wrong in the rural economy of West
poverty compared to the Purulia average. Bengal, despite claims to the contrary in
The ACRP Benchmark Survey for the official statistics about the aggregative
pilot project area of the Kashipur block scenario of poverty in rural areas of this
in the Purulia district spanning 27 mouzas state. Our review of the micro-level case
Table 11: Percentages of Indebted Households by Current Deficit Status, Pilot Project
Mouzas Kashipur Block, Purulia District
Name of Mouzas
__________ ’______
Operational Holdings
Marginal
Small
Small-Medium
Zero
Agardih
Lakhipur
Chaka
Mekhyada
Shalaya
Pakhariathal
Pabrapahari
Chitra
Seja
Bhatin
Mirgipahari
Kashidih
Saharbera
Isanda
Kaliayada
Ledagora
Ichamara
Mehi
Jiara
Kharikagora
Bhalukgazar
Panja
Telaboni
Lajhna
Kusumgora
Sonathali
Kushjuri
76.92
00.00
78.00
33.33
70.00
28.57
40.00
25.00
28.57
00.00
100.00
60.00
22.22
100.00
25.00
28 57
16.67
40.00
100.00
60.00
76 92
24.00
33.33
45.00
25.00
21.74
68.00
41.18
00 00
75.00
42.86
45.45
26.92
48.00
55.26
80.57
48.78
42 86
31.25
44.44
00 00
00 00
61.53
18.18
00.00
54.00
27.77
16.66
36.66
46.15
39.29
63.64
25.00
16.66
20.00
25.00
55.02
26.32
35.00
8.00
41.66
76.92
38.09
85.18
12.12
36.11
46.99
11 11
42.86
3000
55.55
100.00
50.00
71.43
13.33
25.00
19.05
00.00
12.50
00 00
20.00
50.00
33.33
72.72
11.11
33.33
00.00
33.33
00.00
00.00
14.00
12.50
11.11
10 00
14.29
10.00
50.00
50.00
66.66
14.28
Medium
33.33
00.00
00.00
00 00
00.00
00.00
00.00
66.66
00.00
00.00
100.00
00.00
Source: ACRP Benchmark Survey Report. CM (1995).
Economic and Political Weekly
November 21. 1998
studies shows that the extent of depriva
tion in living standards in rural West Bengal
are much more pronounced than can be
comprehended through state level esti
mates. Unfortunately, there is hardly any
serious micro-level studies for the urban
areas of the state, but our presumption is
that we shall get roughly similar pictures
of acute deprivations for the urban areas
of West Bengal as well.
V
Poverty Alleviation and
Globalisation
The poverty alleviation programmes as
are undertaken in the state of West Bengal
are basically the same ones that are put
to use throughout India, with the differ
ence that perhaps these schemes are some
what better managed in West Bengal than
in other states. As a result, leakages and
shortfalls from targets are kept at a mini
mum in West Bengal. But since most of
these schemes are designed centrally, they
often fail to take into account special
features pertaining to local conditions in
West Bengal. Moreover, the share of the
state government in financing these
schemes is low. and in many cases the
state's share is only notionally shown in
the official statistics without correspond
ing actual release of funds to the zilla
parishads and panchayats. While the
overall budget constraint of the slate
government may be the reason for actual
non-disbursemenl of state’s share (the
states share is shown as the value of the
land where the JRY operations are under
taken), the overall poverty of resources
compared to the requirements in the state
is ultimately responsible for the slow
process of poverty alleviation in the state,
particularly in urban areas, where the
economic growth factor is considerably
weak since the middle of 1980s. The only
way out seems to be to make the rural
panchayats much more powerful and
accountable so that they are authorised to
tax the beneficiaries of rural development
and poverty alleviation expenditure in the
state, and raise resources locally to finance
locally designed low-cost and human
resource-intensive projects to provide
additional sources of both on-farm and
off-farm employment for the poor land
less workers. This would require a proper
identification of the beneficiaries and an
assessment of their resource and assets
position, including acquisition of new land.
Such a scheme, if designed, would make
the elected panchayats more aligned to the
poor rural people and would ensure
decentralisation of power and decision
making at the grass roots level in the true
Economic and Political Weekly
sense of the term. Large-scale public
investment schemes expecting to generate
economics of scale in the long run may
be financed with the help of central grants
released under alternative heads of the
poverty alleviation schemes.
The Indian economy has been set on the
path of globalisation since July 1991, and
whatever be its pace and character over
the years, the agricultural sector is likely
to be opened up sooner or later. This
would offer higher price signals to farmers
to increase and diversify their production
base forexports. No doubt, such a process
would increase the prices of foodgrains
and adversely affect the incidence of rural
poverty not only in India, but also in West
Bengal. Subsidisation across all rural
people in the state is not feasible economi
cally, and whatever success has been
achieved should not be allowed to be
washed away through inflation induced
by globalisation. The only alternative
seems to be to orient the rural labourers’
organisations in the state to ensure rise in
money wages faster than the increment in
food prices, and this seems possible be
cause the scale of agricultural production
and hence demand for labour are likely
to increase. There is also likely to be on
expansion of agro-based processing in
dustries in rural areas of the state along
with the expansion of horticulture and
floriculture for which the state possesses
required technical know-how. The state
government, which has so long concen
trated on populist redistribution, has to
reorient its policies and take effective steps
to strengthen the growth impulses in the
rural areas of the state with the help of
accumulated surplus in the rural areas
(which arc now being channelled gradu
ally to trading and service sector activi
ties) and take advantage of the positive
aspects of globalisation towards mitigat
ing the plight of the rural poor and
underpriviledged in this state. Adherence
to all-India norms, standards and strate
gies may not always bring the optimal
results, at the same time, the state gov
ernment cannot and should not ignore the
possibilities thrown open m the national
and international economies. Designing
such an appropriate policy would be the
real task of political economy of poverty
VI
Concluding Observations
This critical review of the studies in
rural and urban poverty in West Bengal
reveals that the incidence of urban poverty
and inequality in the state poses a more
serious challenge to policy-makers than
rural poverty as a whole, which has shown
signs ofsharperdecline in the 1980s thanks
to the upsurge in productivity growth in
West Bengal agriculture. But industrial
growth in the state has been showing signs
of non-revival. and the service sector has
not grown at a fast rate to absorb the
Table 12’ Percentage of Indebted Households by Outstanding Deficit Status, Pilot Project
Mouzas Kashifur Block, Purulia District
Name of Mouzas
Agard ih
Lakhipur
Chaka
Mekhyada
Shalaya
Pakhariathal
Pabrapahari
Chitra
Seja
Bhat in
Mirgipahari
Kashidih
Saharbera
Isanda
Kaliayada
Ledagora
Ichamara
Mehi
Jiara
Kharikagora
Bhalukgazar
Panja
Telaboni
Lajhna
Kusumgora
Sonathali
Kushjun
Operational Holdings
Small-Medium
Small
Zero
Marginal
76.92
50.00
45.00
45.83
47.83
72.00
47.06
00 00
75.00
44 90
47.73
57.61
68.00
71.05
76.92
4 1.46
5000
59.37
66.66
50 00
58.33
54.54
38.46
61.00
81.22
50.00
66.67
20 00
00.00
44.44
33.33
00.00
33.33
37.50
55.55
30.00
42.86
60.00
00.00
50 00
66.66
28 57
55.02
36.84
40.00
52.00
54.17
61.54
38 09
77.77
36 36
58.33
66.27
22.22
100.00
80.00
88.88
87.50
60.00
78.57
53 33
33.33
42.86
33.33
87.50
75.00
100.00
50.00
66.66
81.82
44.44
66.66
76.00
58 33
77.77
28.57
40.00
100.00
57.14
00.00
100.00
80.00
77.77
100.00
25.00
71.43
54 17
60.00
00.000
30.00
61.54
44.44
50.00
•
44.44
00.00
38 46
50.00
54.54
12.50
Medium
66.66
00.00
00.00
00.00
00.00
00.00
100.00
100.00
00.00
00.00
100.00
50.00
Source: ACRP Benchmark Survey Report, CM (1995).
November 21. 1998
3013
Minimum Needs of Poor and Priorities
Attached to Them
V Sitaramam
S A Paranjpe
T Krishna Kumar
A P Gore
J G Sastry
From an examination of the NSS data covering 1951-1991 and taking the cereal consumption deprivation as a measure
ofpoverty the authors present an estimate ofpoverty in India without using the dubious concept of the poverty line. They
argue that there is no need to have a poverty line to measure the degree ofpoverty ofany community or group of vulnerable
households. The method developed here reveals that cereals constitute the commodity group that occupies the top
position in the hierarchy of needs, both in rural and urban areas. Next item ofpriority, both for rural and urban areas,
is fuel and light and not clothing partly because one cannot make a ‘roti’ out of wheat without the cooking fuel.
If the misery of our poor be caused not by
the laws of nature, but by our institutions,
great is our sin
- Charles Darwin, Voyage of the Beagle
I
Introduction
ALLEVIATION of poverty has become one
of the most important items on the policy
agenda of many a government, particularly
in the developing countries. Economic
research so far has concentrated on the issue
of measuring and monitoring the extent of
poverty, rather than on the issues of designing
the appropriate poverty alleviation pro
grammes. Designing such programmes
requires some insights into who the poor are
- for whom such programmes must be
designed, and what their needs are. The view
of a major segment of the economics pro
fession on these two issues has been that all
those who are below the poverty line are the
poor who need poverty alleviation pro
grammes, and that their needs are based on
the common perception of hierarchy-of
needs, such as food,clothing, shelter, health,
education, etc. There are two problems asso
ciated with these economists’ views. First,
the idea of identi fying the poor by the poverty
line is neither acceptable to the policy-makers
nor is it feasible, as the poor do not have
a regular and stable source of income. Also
it is not based on good scientific and objective
reasoning. Second, there is no clear-cut
empirical evidencethat thehicrarchy of needs
corresponds to the oft-repeated slogan ’ foodshelter-clothing’ or ‘roti-kapada-aur makan’.
These priorities may vary from com
munity to community, and from place to
place. The ordering of needs depends on the
circumstances facing the people. For ex
ample, for people living in colder climates
and on forest slopes, clothing and shelter
may be more important than for people who
live on the plains with a more favourable
climate. Similarly, the food habits may vary
Economic and Poli'ical Weekly
from place to place. Hence, what one needs
is a measure of consumption deprivation
that is commodity specific and community
specific. Theeconomists have, inouropinion,
put undue emphasis in defining first who the
poor are and then defining their poverty. It
is our view that it is more meaningful and
useful to define poverty as consumption
deprivation, which is the opposite of welfare,
and then to decide, on a case by case basis,
who ought to be the beneficiaries of any
poverty alleviation scheme.1 The choice of
the beneficiaries should depend on social,
economic, political, and administrative
considerations. The targeting of the poverty
alleviation schemes, in terms of the com
modities for which subsidies are needed
and the people who ought to receive those
■subsidies, should be region-specific. From
this perspective, and given that the notion
of poverty is basically relative, it is even
preferable to call such schemes as welfare
improving schemes rather than poverty
alleviation schemes.
The new United Front government an
nounced its commitment to a ‘Common
Minimum Programme’. What the UDF and
the prime minister seem to imply by this
term is a minimum needs programme, as
there can be a consensus (and hence the word
‘common’) on such minimum needs. This
concept of minimum programme raises
several interesting economic policy issues.
It is suggestive, from the attitude of the new
government, that the new government’s focus
has shifted from poverty alleviation to
providing the minimum needs. This change
in policy focus is quite consistent with the
line of research we have been engaged in
for the past few years on poverty measure
ment. We have been arguing that poverty has
to be measured as commodity-specific con
sumption deprivation of a community,
without any reference to an arbitrarily and
subjectively chosen poverty line. The identi
fication of the poor has to be based, we argue,
Special Number September 1996
not on a difficult to measure income, but on
socially, politically, and administratively,
and unambiguously determined criteria. This
suggestion of ours is also quite consistent
with the actual practice. It may be noted that
the really poor have very irregular employ
ment and income, and hence it is difficult
to measure their incomes to check the eligi
bility for a poverty alleviation programme.2
This is the line of work we have been
doing. In this connection we needed to
identify those commodities, called neces
sities, in terms of which we need to assess
the consumption deprivation. The identi
fication of the most essential commodity,
whose consumption saturates at the lowest
income posed no problem, and it turned out
to be cereals. The budget share of this com
modity at the limiting income is the highest.
Hence, we measured poverty through cereal
consumption deprivation. Butas the economy
develops and the welfare of people in general
improves, people move on to consume the
next item on the hierarchical ladder of com
modities, often by even lowering the con
sumption of cereals. One way of monitoring
the course of economic development is to
see how theconsumption pattern has changed
over the years in terms of bringing into
consumption commodities which were on a
higher rung of the ladder of commodity
hierarchy. Providing to the poor only cereals
at affordable prices is not enough if such
cereals have to be cooked in order to consume,
and if the poor have difficulty in procuring
the cooking fuel. These comments suggest
that there is a need to have a detailed investi
gation into commodity groups other than
cereals that enter into the priority list of
consumers.
In order to get some empirical insights into
the hierarchy of needs among the households,
and how this pattern has.altered over years,
we had examined the consumption pattern
from the National Sample Survey data for
various rounds, starting from the 3rd round
2499
to the 46th round, covering the period 19511991. It is the purpose of this article to share
our findings with the readers and to suggest
some policy implications of our findings in
designing the ‘mini mum needs programmes’
for the vulnerable sections of the household
sector.
The plan ofthe paper is as follows. Section
II presents very briefly the method we use
to examine the consumption expendituredata
of the NSSO employing a new form of Engel
curve. In Section III we present the levels
of cereal-basedconsumption deprivation for .
rural and urban India for various rounds of
NSS. These are our alternate measures of
poverty, based on cereal consumption
deprivation. In Section IV we develop a
method for determining the hierarchy of
consumption needs and apply this method
to the NSSO data. Finally, in Section V we
present the important policy implications
suggested by our method and our findings.
II
Relation between Quantity
Consumed of a Commodity and
Income: Engel Curve
As income is difficult to measure, and as
there are no reliable estimates of household
income levels, we proxy income of a
household by the total expenditure of that
household. In this section we are therefore
concerned about the relationship between
expenditure on a specific commodity or
commodity group and the total expenditure.
Such a relationship is known as the Engel
curve. This relationshi p can assume di fferent
forms. Three very commonly assumed forms
are depicted in Figure. Type I relation shows
that the consumption of that type of
commodity increases with income but at a
decreasing rate. Type II curve shows that the
consumption of this type of commodity
increases with income, but at a constant rate.
Type 111 curve shows that the consumption.
of this type of commodity increases with
income, but at an increasing rate.
Per cent change in the consumption of a
commodity fora 1 percentchangcinincome
is called the income elasticity of demand for
that commodity. ForType I commodities the
income elasticity is less than one. One very
commonly used mathematical form for the
Engel curves of all the three types is:
Log Cj = y + n log y,
'
—(2.1)
where, c, is the mean consumption in
expenditure class i and yj is the mean income
in the same expenditure class. In this form
the income elasticity of demand for the
commodity defined above turns out to be T|.
which is assumed to be constant. It is quite
likely that for necessities such as food, the
income elasticity of demand is not only less
than unity but it may also decrease with
increase in income, i e, the per cent increase
in consumption of food per 1 per cent increase
in income may dccrcaseas income increases.
The above functional form cannot take care
of this possibility.
It is the analogy between the equilibrium
relations in kinetic models of catalysis in
biochemistry and the above Type I Engel
curve of economics that had provided the
major impetus for our research on con
sumption analysis and poverty by an inter
disciplinary team consisting of a biochemist,
an economist, and three applied statisticians.
In fact the Engel curve is an equilibrium
relation between the two flow variables,
expenditure on a specific commodity and the
total expenditure, and hence this analogy is
Table 1: Non-Linear Least Squares Estimates of Engel Curve Parameters (V ano K) for Cereal
Consumption: India 1960-61 to 1990-91
NSS
Round
Period
16th
17th
18th
20th
21st
22nd
24th
27th
28th
32nd
38th
42 nd
43rd
44th
45th
46 th
1960-61
1961-62
1963-64
1965-66
1966-67
1967-68
1969-70
1972-73
1973-74
1977-78
1983
1986-87
1987-88
1988-89
1989-90
1990-91
V
Rural
K
R2
V
Urban
K
R2
19.207
24.267
16.598
23.338
19.863
18.704
16.165
18.053
19.122
12.340
11.830
10.584
10.043
9.701
8.114
9.033
23.030
41.157
21.417
31.157
24.584
27.398
21.397
24.404
25.492
16.217
16.120
15.989
13.898
12.470
9.718
13.216
0.9916
0.9265
0.9878
0.9955
0.8481
0.9953
0.9925
0.9903
0.9854
0.9872
0.9840
0.9757
0.9693
0.9767
0.9432
0.9242
9.645
9.363
8.628
7.6371
8.844
9.225
9.413
8.733
10.794
7.968
8.044
6.767
6.406
6.038
5.968
5.998
9.11
9.87
8.95
8.00
10.87
12.25
12.06
9.46
14.97
11.05
11.98
11.54
9.29
7.48
7.87
9.74
0.9569
0.9519
0.9520
0.8333
0.8992
0.9404
0.9399
0,8910
0.9302
0.9463
0.9668
0.9401
0.9339
0.9524
09356
0.9575
The V and K estimates are in rupees per capita per month (in 1960-61 prices). V and K are first
estimated separately for each NSS Round. The tabulated values above are V and K adjusted for
price changes between rounds.
Source: Estimated using NSSO data.
Note:
2500
quite appropriate. The saturation kinetic
models in bioch.cmistry use a hyperbolic
relation of the following type to represent
the kinetic equilibrium:
Cj = Vy/tK+yJ
... (2.2)
When we fitted equations (2.1) and (2.2)
to the NSSO data we found that (2.2) always
gave a better fit than (2.1 \ When we say
better fit, the criteria we used to compare the
two models are: (i) coefficient of
determination. R2, and (ii) randomness of
errors with’ a Gaussian distribution. It is also
interesting to note that the income elasticity
of demand for specification given by (2.2)
does vary with income and decrease with an
increase in y, adesirable property cited above.
In this study we used model (2.2) for
determining the hierarchical basic minimum
needs. We used thesame model in our studies
on poverty through consumption deprivation.
Some properties of this Engel curve are worth
noting, and these are given below:
(1) Dividing both the numerator and the
denominator of the right hand side of (2.2)
by yi we get:
c, = V/(K/y, + 1)
...(2.3)
From equation (2.3) we note that as y,
tends to infinity tends to V. Thus, V can
be interpreted as the saturation level of
consumption. (V-c)/V is the proportional
shortfall in consumption from the saturation
level, and it lies between 0 and 1. We had
proposed that, for any community, the mean
proportional shortfall of consumption of a
basic necessity such as cereals from its
saturation level be taken as a poverty index
of that community [sec Gore, Kumar,
Paranjpe, Sastry.andSitaramam 1994,1996
and Kumar, Gore, and Sitaramam 1996].
(2) Dividing both the right and the left
hand sides of (2.2) by y, we get:
c/y; = V/{K + yj
...(2.4)
Since income and consumption move
together we can assume that ci/yi tends to
a constant as yi tends to zero. From equation
(2.4) it follows that this limit is V/K. Thus,
V/K can be interpreted as the proportion
spent on the commodity, or the budget share
of the commodity, at limiting (or low levels
of) income.
(3)
From equation (2.3) we get:
V/ci = K/y; + 1 or (V-ci)/ci = K/y. ... (2.5)
From equation (2.5) it follows that yj = K
when c. = V/2.
Thus, the parameter K may be interpreted
as that level of income at which consumption
is at half-saturation level. Hence, parameter K
Economic and Political Weekly
Special Number September 1996
is often called the haif-saturation constant. programmes, would give erroneous conclu
(4) The equilibrium quantity consumed sions as the measured poverty could exclude
depends directly on the forward rate constant some of the actual beneficiaries whose
V (need) and inversely on the backward rate incomes could be above the poverty line.
constant K (cost). The proportion of income Hence what is needed in this connection is
spent on a necessity(commodny) decreases an insight into the commodity-specific
with increasing income.
consumption deprivation among a variety of
The hyperbolic Engel curve was fitted to vulnerable groups of people, identified by
the Indian data on household consumption some policy relevant criteria. If it is desired
published by the NSSO. The model was to choose between alternate groups so as to
fitted using non-linear least squares method exclude the creamy layers from the benefits
of estimation. This method requires an initial of the poverty alleviation programmes one
guess of the unknown parameters. Although can measure the commodity-specific
the programme usually has certain default consumption deprivation for such alternate
values forthe initial guesses, the convergence groups and choose, for implementing the
to the final estimates would be faster, and poverty alleviation programmes, that group
we can also be reasonably sure of a global which has more consumption deprivation
minimisation of the error sum of squares, if than others.
the initial guesses are chosen carefully.
We used the NSS data for the computation
Hence, we provided, as initial guesses of the new poverty index that does not use
estimates derived from the following the poverty line. The consumer expenditure
linearised version of the model:
data for cereal expenditure, and total expen
ditures by various total expenditure classes
I/O = 1/V + (K/V)(l/y.)
...(2.6) for various NSS rounds starting from 16th
round (1960-61) to the 46th round (July
1/C; was regressed on Ity , and the 1990-June 1991) were used. The commodityreciprocal of the intercept estimate is taken specific poverty indices for India for the
as the initial guess of V, while the ratio of period 1960-61 to 1990-91 were computed
the slope estimate to the intercept estimate using the method described in Section II.
is taken as the initial guess of K.
Engel curves of type (2.2) were fitted
separately for each year (round), and
Ill
separately for rural and urban India using
Poverty without Poverty Line: Measure non-linear least squares method of estima
of Poverty Based on Cereal-Based
tion employing RATS computer software.
Consumption Deprivation
The estimates of saturation consumption
(V) for cereals, were deflated with the food
The concept of poverty line has been a component of consumer price index (CPI)
very controversial and subjective concept, while the estimates of K were deflated by
which had placed economic research on the overall consumer price index. For rural
poverty in a very shaky and vulnerable households thcCPI foragricultural labourers
position. We had argued elsewhere [Gore, was used whereas for urban households a
Kumar, Paranjpe, Sastry, and Sitaramam weighted average of CPI for non-manual
1994,1996 and Kumar, Gore and Sitaramam workers and industrial workers was used
1996] that the identification of the poor can along the lines suggested by Minhas ct al
be made on the basis of commodity-specific (1987), by giving them weights of 0.625 and
consumption deprivation among different 0.375, respectively. This de flation wascarried
vulnerable groups of people, those groups out to make the parameters V and K compar
having been identified by a priori criteria able over time.
such as rural landless labourers, unemployed
There is a hierarchy of needs, the cereals
or seasonally employed persons, female being the first and most essential commodity.
headed households with dependent children, The estimates of V and K (adjusted for
etc. Our method did not require a poverty changes in food prices and overall prices),
line level of income for either identifying the for cereal consumption are presented in
poor or for measuring poverty.
Table 1. Trend lines fitted to the estimates
The beneficiaries of poverty alleviation of V and K show that there is a secular
programmes are also normally and actually decline in both. This could imply that over
chosen by criteria other than a poverty line time households started substituting non
level income. If the poor are so identified cereal and non-food items for cereals. This
for the poverty alleviation programmes, by could be due to increased -availability over
criteria other than poverty level income, then time of non-cereal and non-food items. The
it makes no sense to measure their degree estimatesof V iriTable 1 arealmost uniformly
of poverty through a measure that depends larger for the rural data compared to the
on an arbitrarily chosen poverty line. Such urban. The explanation given above, viz, an
a procedure, of applying the traditional increase in availabilit: of non-cceal options
measure of poverty (based on a poverty line), (inurbanareas)possiuiy explains this pattern
when used to monitor the poverty alleviation as well.
.
Economic and Political Weekly
Special Number September 1996
Table 2: Proportion of Total Expenditure
Spent on Cereals at Limiting
Income (V and K), India:
1960-61 to 1990-91
Year
1960-61
1961-62
1963-64
1965-66
1966-67
1967-68
1969-70
1972-73
1973-74
1977-78
1983
1986-87
1987-88
1988-89
1989-90
1990-91
Cereal
Rural
Urban
0.8340
0.5896
0.7750
0.7491
0.6949
0.6827
0.7555
0.7398
0.7501
0.7609
0.7339
0.6620
0.7226
0.7779
0.8350
0.6835
1.0586
0.9490
0.9639
0.9542
0.8135
0.7529
0.7804
0.9233
0.7209
0.7212
0.6715
0.5861
0.6889
0.8075
0.7586
0.6158
V and K arc first adjusted for price
changesbetwecnroundsandthcn V and K
was calculated.
Source: Estimated from Engel Curves using
NSSO data.
' Note:
Table 3: Estimates of Poverty Measured
through Cereal Consumption Deprivation:
With Separate Engel Curves
for Each Round
Year
Rural
Urban
1961-62
1963-64
1965-66
1966-67
1967-68
1969-70
1972-73
1973-74
1977-78
1983
1986-87
1987-88
1988-89
1989-90
1990-91
0.6838
0.5918
0.6459
0.5998
0.6460
0.5519
0.5749
0.5641
0.4757
0.4541
0.4260
0.3768
0.3628
03007
0.4428 '
0.3015
0.2816
0.2734
0.3202
0.3577
0.3188
0.2744
0.3638
0.3118
0.3123
0.2875
0.2533
0.2122
02142
0.3133
Table 4: Estimates of Poverty through Cereal
Consumption Deprivation: With Pooled or
Common Engel Curve for All Rounds
Urban
Year
Rural
1960-61
1961-62
1963-64
1965-66
1966-67
1967-68
1969-70
1972-73
1973-74
1977-78
1983
1986-87
1987-88
1988-J9
1989-90
0.3195
0.5766
0.3132
0.5759
0.3119
0.5981
, 0.3261 .
0.5980
0.3124
■
0.6045
0.3244
0.6269
0.2907
05875
: 0.2938
05814
0.2890
05567 .
0.2954
0.5769 "
0.2801'
0.5577
0.4238 ... 0.2873
0.3768. ■ : 0.2442
0.2122
0.36z8 . '
0.2142
03007 . ■
• ••
■'
2501
As mentioned earlier (refer to equation
(2.4) and the comment below that equation)
the proportion of expenditure on the specific
commodity (cereals) turns out to be V/(K+y)
and this becomes V/K as income tends to
zero. Thus V/K is the limiting proportion of
expenditure on cereals. Table 2 presents
estimates of V/K for cereals.
An interesting aspect to note is that this
V/K ratio for cereals or "proportion spent
on cereals at limiting income" has been higher
in the urban India up to 1970 than in rural
India. But from 1970 onwards (except for
1972-73) this proportion is less in urban
India than in rural India. This seems to be
partly due to the green revolution. This may
also be partly due to the PDS being more
urban-oriented as V/K showed a declining
secular trend in urban’area only. The over
all constancy of V/K for cereal consumption
in rural India also justifies using cereal
consumption deprivation for measuring
poverty, as most of the commonly under
stood poor (agricultural labourers and mar
ginal farmers) live in rural areas and the
proportion of total expenditure that they
spend on cereal consumption expenditure
is very high being 0.75 on an average and
stable.
Table 3 presents the poverty indices based
on the cereal-based consumption depriva
tion.3 The poverty estimates ofTable 3 show
a time trend in this poverty index. The cereal
based poverty index clearly demonstrates
that there is a higher incidence of poverty
in rural India and that the difference between
rural and urban poverty has reduced between
1960-61 and 1990-91. It must be mentioned
that theurban and rural poverty indices given
here are based on different saturation norms
(Vs). Hence we cannot strictly compare the
rural and urban poverty indices. The maxi
mum cereal consumption differs between
rural and urban areas, partly because the
commodity spectra available are different in
rural and urban areas. We can, however, talk
about the rate of decrease in poverty between
urban and rural poverty and note that this
decrease is much more in rural areas than
in urban areas.
The poverty index presented here is based
on deprivation from saturation norm that is
specific to each data set. Since this saturation
norm (i e, estimated V) is different for each
year as well as for urban and rural samples,
comparison of the poverty indices needs a
careful explanation. If our concern is about
consumers’ feelingofconsumption depriva
tions from their own saturation point (this
maybe termed ‘felt-deprivation’derived from
the concept of felt need) then the comparisons
of above indices are alright. Our measure of
poverty is a relative measure relative to the
maximum expenditure on cereals, which
differs between rural and urban areas. While
we may, under certain circumstances, be
Economic and Political Weekly
Table 5A: Sequence in Which Commodity Groups Appear, along with Their Budget Shares at
Limiting Incomes (Adjusted V/K, Presented in Parentheses): Rural
Years
Round No
Cl
C2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
24
25
27
28
38
42
43
44
45
CE(.71)
CE(.68)
CE(.65)
CE(.75)
CE(.99)
CE(.89)
CE(.82)
CE(.58)
CE(.87)
CE(.89)
CE(.78)
CE(,83)
CE(.79)
CE(.83)
CE(.63)
CE(.77)
CE(.75)
CE(.75)
CE(.72)
CE(.76)
CE(.82)
CEG84)
CE(.8)
CE(.83)
CE(.81)
CE(.72)
CE(.78)
CEG82)
CE(,91)
FL(.O567)
CL(.05!2)
FLG0717)
FLC0404)
FL(.0018)
FLG0253)
FLG044)
FL(.O42)
FL(.O234)
FLG0233)
FL(.O346)
FL(.O345)
FL(.O42)
FL(.O442)
FL(.07l)
F1X.0445
FL(.O525)
FL(.O518)
FLGO568)
FL(.O53)
FL( 0356)
FLG0333)
FLG0398)
FL(.0322)
FL(.O435)
FL(.O632)
FU.O459)
FL(.O392)
FLG0175)
Apr 1951-Mar 1954
Apr 1952-Sep 1952
Dec 1952-Mar 1953
May 1953-Sep 1953
Oct 1953-Mar 1954
Jul 1954-Mar 1955
May 1955-Nov 1955
Dec 1955-May 1956
Aug 1956-Feb 1957
Mar 1957-Aug 1957
Sep 1957-May 1958
Jul 1958-Jun 1959
Jul 1959-Jun I960
Jul 1960-Aug 1961
Sep 1961-Jul 1962
Feb 1963-Jan 1964
Jul 1964-Jun 1965
Jul 1965-Jun 1966
Jul 1966-Jun 1967
Jul 1967-Jun 1968
Jul 1969-Jun 1970
Jul 1970-Jun 1971
Oct 1972-Sep 1973
Oct 1973-Jun 1974
Jan 1983-Dcc 1983
Jul 1986-Jun 1987
Jul 1987-Jun 1988
Jul 1988-Jun 1989
C4
C3
CL(.O435)
FLG051)
CLG0595)
CL(.0375)
MEF(.00074)
EOG0085)
EO(.0I08)
MEF(.O34)
EOG009)
MEFGOO73)
MEF(.OI67)
MEFG0U3)
MEF(.O134)
MEFG0128)
MEF(.0I7)
P and P(.0276)
SU(.O3)
S and SG0229)
S and SG0334)
S and S(.O264)
SPIG0194)
VEG(.O16)
VEGG0144)
VEGG0157)
VEGG0232)
VEGG036)
VEGG0286)
VEGG0243)
VEGG0I26)
EOG0197)
EOG0I09)
EOGOI 45)
MEFG0076)
MEF(.O234)
EOG0055)
EOGOI1)
EOG0085)
EOG0077)
EOG0105)
EO(.0148)
P and P(.O245)
VEG(,02)
VEGG0252)
VEG(.O2I6)
VEGG018)
SPK.0158) A
SPIG014) ”
EOG125)
PandP(.0131)
EOG0308)
PandP(.O182)
Pand P(.018)
EOG008)
Table 5B: Sequence in Which Commodity Groups Appear, along with Their Budget Shares at
Limiting Incomes (Adjusted V/K, Presented in Parentheses): Urban
Years
Round No
Cl
C2
C3
C4
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
24
25
27
28
38
43
44
45
CE(.83)
CE(.77)
CE(.68)
CE(.98)
FL(. 1263)
CLGO3O3)
FLGO478)
FL(.0028)
CL(.0I87)
FL(.O26)
CL(.033)
CLG0024)
EOG0151)
EOG0138)
MEF(.O2O4)
REG0009)
CEG9)
CE(.78)
CE(.87)
FL(.O139)
FL(.O3O4)
FU.0174)
EO(.007)
EOG0154)
EOG0089)
MEFGOO63)
SUG0143)
SU(.OO74)
CE(.83)
FLC0273)
EOG0142)
SUG0119) A
CE(.86)
SU(.0252)
PandP(.0I61)
CE(.92)
CE(.87)
CE(.88)
CE(.86)
FL(.0l22)
FLG0204)
FLG0182)
FL(,O2O8)
CE(.84)
CE(.75)
CE(.73)
CE(.92)
CE(.87)
FL(.0224)
FL(.O442)
FLG0389)
FLG0103)
FLX-0199)
FLG0197)
.. ' ■
1:, •’
SandS(.0112)
Sand S(.0161)
SPI(.O156)
SPI (.0168)
' ':
;
PANG0055)
VEGG0228)
EOG0297)
VEGGOO88)
VEG(.0I43)
Apr 1951-Mar 1954
Apr 1952-Sep 1952
Dec 1952-Mar 1953
May 1953-Sep 1953
Oct 1953-Mar 1954
Jul 1954-Mar 1955
May 1955-Nov 1955
Dec 1955-May 1956
Aug 1956-Feb 1957
Mar 1957-Aug 1957
Sep 1957-May 1958
Jul 1958-Jun 1959
Jul 1959-Jun 1960
Jul 1960-Aug 1961
Sep 1961-Jul 1962
Feb 1963-Jan 1964
Jul 1964-Jun 1965
Jul 1965-Jun 1966
Jul 1966-Jun 1967
Jul 1967-Jun 1968
Jul 1969-Jun 1970
Jul 1970-Jun 1971
Oct 1972-Sep 1973
Oct 1973-Jun 1974
Jan 1983-Dcc 1983
Jul 1987-Jun 1988
Jul 1988-Jun 1989
Abbreviations used:CE
: Cereals
FL
: Fuel and Light
EO
: Edible Oil
CL
: Clothing
SU
: Sugar
VEG : Vegetables
PAN : Pan, Tobacco and Int
MEF
S&<
P&P
M&P
F&N
SPI
RE
Meat, Egg and Fish
Salt and Spices
Pulses and Products
Milk and Products
Fruitsand Nuts
Spices ,
Rents
SUG0061)
PANG0065)
EOGOI 09)
VEGGO133)
SALT(.0013)
SPIG0175)
M and P(.O272)
M and P(.OO82)
EOGOI 27)
•' •
*•
Nole: For some rounds when V/K estimate is outside the plausible range they are omitted.
Special Number September 1996
2503
expenditure at the limiting income. The
adjusted V/K are presented in the parentheses
in Tables 5A and 5B. From these adjusted
V/K figures it. becomes quite clear that fuel
and light occupies a high position next to
cereals in terms of budget share at limiting
incomes.
/■-I
; ,,
„■
Policy Implications of Our Results
justified in making temporal comparisons
within rural and urban areas separately, it
is not quite proper to make comparisons
between urban and rural poverty indices.4
To shed more light on this problem we
deflated the expenditure data of each NSS
Round with the appropriate price index and
examined to see if there is a long-run stable
Engel curve that fits the data with a single
V that can be used in computing the poverty
rndex for each round. It was felt that urban
and rural consumption patterns are not
comparable anyway.. Hence two separate
long-run Engel curves were estimated, one
for the rural areas and another for the urban
areas.- — ............. —.............
Table 4 presents the poverty indices based
on the assumption that there is a common
Engel curve for all the rounds of NSS after
adjusting the data for price level changes
from year to year. These estimates seem to
suggest that rural poverty had increased in
India from 1960-61 to 1967-68 and then
recorded a secular decline until 1989-90.
There is again a sharp increase in rural
poverty in 1990-91. An examination of
urban poverty indices of Table 4 suggests
that the urban poverty more or less remained
stable until 1967-68 and then registered a
slight decline and thereafter remained
stable until 1989-90. Likerural poverty urban
poverty also registered a sharp increase in
^>90-91.
I
IV
Prioritisation of Household Needs
When the above Engel curve was fitted
to the various groups of commodities for
which NSSO presents its expenditures we
observed that only for some of them,
particularly cereals, the fit was good. In
some cases Type II and Type HI curves of
the Figure seemed more appropriate. If the
Engel curve for cereals has a Type I shape
it follows that thcexpenditure on a commodity
group consisting of all other commodities
should have an Engel curve of Type 111.
Noting that the expenditure er. cereals forms
a major portion of total expenditure
a
household we felt that the slope of the curve
m some cases could be imperceptible if the
curve is drawn against total expenditure, and
2504
that it may be perceptible if it is plotted
against budget available after deducting the
expenditure on cereals. This in fact turned
out to be the case.
The above observation suggested that there
could be a hierarchy of household needs
among the poor, the intensity of need being
the greatest for cereals. Once the cereals
need is fulfilled the household may spend
a part of the remaining income on a com
modity, expenditure on which saturates
next. Having met the expenditures on the
two most important items the household
may move on to spend on the third item,
from the remaining income, and so on.
We tested this model of needs-hierarchy
using the NSSO data from 1951-1991, viz,
from the third round to the 45th round. After
fitting the Engel curve of model (2.2) for
cereals we asked the question which of the
remaining groups of items takes the second
position in terms of saturating next, with the
best fitting hyperbolic relation plotted against
the remaining part of the total expenditure.
Having thus chosen the second most needed
item, separately foreach round and separately
for rural and urban samples, we asked the
question likewise - what item, out of the
remaining items, would qualify to take the
third position, and so on.
Our findings arc reported in Tables 5A,
5B, and 6. In Tables 5A and 5B we present
the sequence in which commodity groups
appear along with their budget shares at
limiting income (V/K values). In Table 6 we
present the frequency with which a com
modity is selected (out of a total of29 rounds)
as one of the top five priority items. From
these results it appears that next to cereals
comes the category ‘fuel and light’, the major
component of that being possibly thccooking
fuel. After fuel and light comes ‘edible oil’.
After edible oil comes ‘meat, egg, and fish’
in rural areas and ‘sugar’ in urban areas.
The parameters 'W and ’K’ estimated at
each stage refer to the situation where the
independent variable of the non-linear Engel
curve is the ’remaining total expenditure’,
after the expenditures on commodities of
earlier stages ate sub''ac*id from the total
expenditure. The V/K at each stage needs
to be adjusted'to express expenditure on
that commodity as proportion of the total
It is our view that as a part of its common
minimum programme the present UDF
government rhustdesign programmes aimed
at improving the welfare of the vulnerable
sections of the community (households). Such
programmes must not be looked upon as
poverty alleviation programmes of the tradi
tional variety with questionable, dubious and
outdated methods of defining and measur
ing poverty. Instead, we propose that a new
thrust be given to poverty alleviation through
minimum needs programmes. In designing
these programmes the vulnerable groups may
be chosen from a set of alternative groups
through criteria other than the poverty line,
criteria based on social, political, and ad
ministrative considerations aided by our
measure of consumption deprivation. What
we. mean by this is that among a set of
alternate groups chosen a priori according
to sociological, economic, political and ad
ministrative criteria one group may be chosen
as the beneficiary group for the government’s
programme on the basis of the criterion of
having the highest level of commodity
specific consumption deprivation.
From the findings reported in the previous
section on the hierarchy of needs it becomes
quite apparent that after cereals the next most
important commodity group is fuel and light,
which includes cooking fuel. After these two
comes the group edible oils. In view of these
results it can be suggested that in revamping
the PDS the new government should omit
the creamy layer from the PDS beneficiary
list and spend more on providing cooking
Table 6: Frequency with Which a Commodity
Was Selected in First four Priori ty Items
According to Adjusted V/K
hems
Rural
Urban
Cereals
Fuel and Light
Edible Oil
Clothing
Sugar
Vegetables
Pan, Tobacco
Meat, Egg and Fish
Salt and Spices
Pulses and Products
Milk and Products Spices
Rents
Salt
29
29
15
4
I
12
0
10
3
18
18
9
4
5
4
2
2
2
1
o
Economic and Political Weekly
0
3
0
0
1
1
Special Numbci S.-.tei’. i.ei I'.T''
fuel and edible oil to the vulnerable groups
through PDS.
We also feel that more detailed analysis
needs to be done along the lines proposed
here to assess the commodity-specific con
sumption deprivation among different
vulnerable groups in order to design proper
welfare-improving government program
mes based on the minimum needs approach.
The poverty line and the traditional poverty
measures based on the poverty line can be
dispensed with altogether.
While fuel and light have been clubbed
together in this analysis of NSSO data there
are other studies that have examined the role
of cooking fuel (fuel wood, charcoal and
kerosene) in household consumption. In
particular we may refer to the work of Reddy
and Reddy (1985) that examined cooking
fuel consumption by a sample of households
in Bangalore city. A similar survey ofcooking
fuel consumption in rural North India was
undertaken by NCAER (1978). The study
of Reddy and Reddy clearly shows the
importance of consumption of cooking fuel
among low-income households even in highly
urbanised areas such as Bangalore. The long
lines of the urban poor to get kerosene is a
pathetic sight we confront today, even after
the government permitted the import and
sale ofkerosene by private parties. One strong
implication of our study is to highlight the
importance of cooking fuel for the poor
households.
Another major policy implication of our
study is that in the currently prevailing attitude
of giving primacy to local bodies in design
ing and monitoring the poverty alleviation
programmes the targeting of the programmes
in terms of the choice of the beneficiaries
and the choice of commodities must be
specific to each local community. Our
research emphasises this point and also
provides a method of choosing these targets
for each local community.
Although we used the NSS data and
presented our results for the country as a
whole we emphasise that this sort of exercise
has to be done at disaggregated levels,
possibly at the district, taluk, and village or
(urban) block levels. The NSS type of data
which has very few observations at such
levels of disaggregation are not suitable for
this purpose. We hope our study will drive
home the need to generate data bases at grass
roots level to design and monitor social
welfare programmes. We also hope that the
various NGOs whih are actively engaged in
social welfare schemes all over the country
will come together to develop standardised
data bases at the village and block levels.
In short, we hope that this study of ours
will convince both the researchers and the
policy makers that the concept of poverty
line can be dispensed with. We also hope
that our study will form the beginning of
Economic and Political Weekly
worthwhile social science policy research
through research studies on consumption
deprivation at the village and block levels.
of consumption for rural households and rural
saturation levels for urban households, and
then comparing them as transferable public
expenditure equivalents.
Notes
References
[This work is based on an ongoing research by
these authors on measurement of poverty without
using a poverty line. The.authors’ new method
for measuring poverty is based on commodity
specific consumption deprivation, thccommodities
chosen being the most essential basic needs. This
is a radically different method, compared to the
highly discredited traditional methods that employ
a questionable and subjective poverty line. The
interested reader may refer to Gore, Kumar,
Paranjpe, Sastry, and Sitaramam (J 994 and 1996)
and Kumar, Gore and Sitaramam (1996). This
research was initiated by V Sitaramam at the
National Institute of Nutrition several years ago
in collaboration with J G Sastry. It is now being
continued by Sitaramam, during the last two and
a half years, in association with Gore, Krishna
Kumar and Paranjpe. The authors thank S
Subramanian and Vinod Vyasulu for their
comments on an earlier draft of this paper. Krishna
Kumar thanks Sushant Mallick, G Nagaraju, and
N S Manjula for their research assistance.]
Gore, A P, T K Kumar, S A Paranjpe, J G Sastry
and V Sitaramam (1994): ‘Measurement of
Poverty through Consumption Deprivation’,
Paper presented at a Seminar on Quantitative
Analysis of Economic Inequality and Poverty,
March 17-18, Institute for Social and Economic
Change, Bangalore.
- (1996): ‘Measurement of Poverty through
Commodity-Specific Consumption Depriva
tion’, Proceedings ofthe Thirty-Second Annual
Conference ofthe Indian Econometric Society,
March 21-23, Vol II.
Kakwani, N C (1980): ‘On a Class of Poverty
Measures’, Econometrica, Vol 48. pp 437-46.
Kumar, T K (1993): ‘Measurement of Poverty
Suited to Design and Evaluate Poverty
Alleviation Programmes’, Economic and
Political Weekly, December, pp 2893-95^^
Kumar, T K, A P Gore, and V Sitaramam, (19^^
‘Some Conceptual and Statistical Issues in
Measurement of Poverty’, Journal ofStatistical
Planning and Inference, A Special Issue on
Econometric Methodology, Vol 49, No I.
1
It is our view that an attention to semantics and
pp 53-71.
linguistics is quite useful here. The economists’
Minhas. B S, L R Jain, S M Kansal, and M R Saluja
focus so far has been on the ‘focus axiom’ that
(1987): ‘On the Choice of Appropriate Con
requires the noun ‘poverty’ to be associated
sumer Price Indices and Data Sets for
with the substantivised adjective ‘poor’, or the
Estimating the Incidence of Poverty in India’,
associated noun ‘the poor’. The English
Indian Economic Review, Vol 22, No 1,
language, however, gives the nown ‘poverty’
pp 19-50.
a position that does not necessarily depend on
National Council of Applied Economic Research
the identification of ‘the poor’. In other words
(1978): Survey ofRural Energy Consumption
the English language does not say that poverty
in North India.
is only what ‘the poor’ possess. The Webster’s
Reddy, A K N, and B Sudhakar, Reddy (1985):
New Collegiate Dictionary gives three different
‘The Energy Sector of the Metropolis of
meanings to the word poverty. (1) the state of
Bangalore’ in Vinod Vyasulu and A K N
one who lacks a usual or socially acceptable
Reddy (eds). Essays on Bangalore, Vol 3,
amount of money or material possessions;
Karnataka State Council for Science and
(2) scarcity or dearth; (3) debility, due to
Technology, Bangalore, pp 59-137.
malnutrition. The second meaning refers to
Sen, A K (1976): ‘Poverty: An Ordinal Approach
deprivation; and. that is what we wish to
to Measurement’, Econometrica, Vol 44, No 2,
emphasise.
'
'
*
pp 219-31.
2 It is trivial that there should be a direct relation Vyasulu, Vinod, and B P Vani (1996):
between unemployment and poverty. This
Poverty and Unemployment in Selected IndW^
relationship is vividly brought but in terms of
States: Implications’, Development Research
published official statistics for urban areas in
Foundation, TIDE, Bangalore, July.selected Indian States by Vyasulu and Vani
(1996).
INDIA’S POLITICAL AGENDA:
3 Wehaveshowninanotherpapcrthatthepoverty
index we propose satisfies all the major axioms
Perspectives on the Party System
that such an index should satisfy (axioms such
as those proposed by Sen (1976), and Kakwani
(1980)) except the focus axiom. We argued that
there is no need to have the focus axiom if we
Editors
define poverty first without defining who the
Mahendra Prasad Singh
poor are [see Gore, Kumar, Paranjpe, Sastry,
Rekha Saxena
and Sitaramam 1994 and 1996]. The proofs that
the poverty index satisfies all the major axioms
were, based on thc observation that-the index
' 1996
is the mean of deprivation, which ^expressed
as a function of income, the mean being taken
with respect to the income distribution [Kumar
KALINGA PUBLICATIONS
1993].
10A, Pocket I, Phase I,
4
The problein here is quite similar to the problem
Mayur Vihar, Delhi - 110091
of real income comparisons posed by
Samuelson. One can make Scitovsky type of
comparisons between rural and urban poverty
ISBN 81-85163-75-8
Rs. 550.00
measures by substituting urban saturation levels
Special Number September 1996'*
2505
India’s Checkered History in Fight Against Poverty
Are There Lessons for the Future?
Martin Ravallion
Gaurav Datt
Looking back 40 years or so, progress against poverty in India has been highly uneven over time and space. It
took 20 years for the national poverty rate to fall below - and stay below - its value in the early 1950s. And trend
rates ofpoverty reduction have differed appreciably between states. This paper provides an overview of results from
a research project which has been trying to understand what influence economywide and sectoral factors have played
in the evolution of poverty measures for India since the 1950s. There are some clear lessons for the future.
THERE has been much debate about how
best to fight absolute poverty. Total numbers
of poor people in the world - by almost any
accepted standard - are continuing to rise.'
The urgency of resolving the debate, and
taking effective action, is greater than ever.
The extent to which poor people share in
economic growth has been one of the most
contested issues. Some observers have argued
that “distribution must get worse before it
gets better” in developing countries, and that
this puts a severe brake on the prospects for
pro-poor economic growth. There have also
been debates about the effects of growth in
specific sectors. For example, some have
argued that the benefits of the 'green
revolution’ (which resulted in substantial
gains in agricultural yields through new seed
varieties and irrigation) were captured by
relatively well-off farmers, and brought little
or no gain to the rural poor. Others have
pointed to farm-output growth as the key to
poverty reduction, both directly and via its
effects on rural wage rates?
There would be little risk of exaggeration
in saying that the position one takes in such
debates has great bearing on long-standing
issues of development strategy and policy
reform. The link between growth and poverty,
and the interaction with other factors
(including human resource development),
has also taken on new urgency in the wake of
recent macroeconomic difficulties and adjust
ment efforts in many developing countries?
However, these are difficult, issues to
resolve empirically, not least because of the
paucity of representative and reliable data
over, time on the living standards of poor
people. Amongst developing countries, India
has. relatively good data for addressing these
issues. At the time of writing, one could
compile a time series of consumption data
from 34 National Sample Surveys spanning
1951-92. This is one of the longest series
of national household surveys suitable for
tracking living conditions of the poor. Most
of the surveys are large enough to be con
sidered representative at the urban and rural
levels for most states, arid they appear to be
reasonably comparable over time since the
when it moves across sectors unless its
standard of living has changed)? Following
now well-established and defensible practice
for India and elsewhere, the standard of
living was measured by consumption
expenditure (including imputed values for
consumption from own production). Weonly
studied ‘poverty’ in the narrow (though
unquestionably important) sense of
"command over commodities”; we do not
deny that there are aspects of a broader
concept of ‘well-being’ which are not
captured by our poverty measures?
The poverty lines used were those defined
by India’s planning commission? The rural
poverty line is Rs 49 per month and the urban
line is Rs 57 per month at October 1973June 1974 all-India rural and urban prices
respectively? The nominal consumption
distributions for each survey data were then
converted to constant prices using consumer
price indices for urban and rural areas which
were, anchored to the consumption patterns
of low-income workers? 10 .' ■
Three .different poverty measures, were
used: (i) The headcount.index, given by the
percentage of the population who live in
households with a. consumption per capita
less than the poverty line. This measures the
incidence of poverty; (ii) The poverty gap
index, defined by the mean distance below
the poverty line expressed, as.a proportion
of that line (where the mean is formed over
the entire population, counting.the non-poor
as having zero poverty, gap). This reflects
the depth of poverty, as .well,as its incidence;
(iii)The squared poverty gap index, defined
as the mean of the squared proportionate
poverty gaps. Unlike the poverty gap index,
Data on Poor People
this measure reflects theseveriryofpoverty,
To address the questions posed above, we in.that it will be sensitive to inequality
constructed a new set of consistent estimates . amongst the poor,11'12 The estimated poverty
of various poverty measures for India over measures were then collated with a variety
the period 1951 to 1992 from the National. of macro-economic and sectoral variables.1’
Sample Survey (NSS) data. We aimed ,to:.
How Much Progress Has India Made in
measure ‘absolute'poverty’, by which..we.
Fighting Poverty Since the 1950s?
mean that the extent of any household’s:.
poverty depends solely on its own absolute.
Table 1 gives.our estimates of.the three
standard of living (for example, a household' poverty measures for eight periods formed
does not switch from being poor to non-poor by aggregating NSS rounds; Figure 1 gives
basic survey method has changed relatively
little. Other data (on price indices and ex
planatory variables) are also available on a
reasonably consistent basis. Although there
are data problems (some of which we can
make corrections for), they are modest by
the standards of cross-country comparative
studies. The existence of a time series of
consumption distributions spanning 40 years
represents a unique opportunity to study the
link between living conditions of the poor
and the key macro-economic and sectoral
variables which are thought to have important
influences on progress in reducing poverty.
We have used these data to study the past
evolution of living standards in India. We
have asked: How have comparable measures
of poverty in India evolved since the 1950s?
Has the experience been different between
urban and rural areas and between different
states? How have measures of poverty
responded to changes in economy-wide and
sectoral variables? What has been the relative
importance of economic growth versus
changes in distribution? What role has been
played by the sectoral composition of
economic growth? How important have
changing wages and prices been? Why have
some states of India done so much better
than others, in the fight agjinst poverty?
What role have differences in the initial
levels of human developmentplayed, versus
other factors such as physical infrastructure
endowments? This paper provides an
overview of the results .of this research.'.We
avoid details bn data'and methods, Which
are described more fully,in a series of papers
from the project.4^. :
Economic and Political Weekly ; Special Number September 199.6.
2479
the estimates of the headcount index for each
survey round. (The pattern of change was
very similar for the other two poverty
measures.) Several points emerge:
The period from the early 1950s up to the
mid-1970s was characterised by fluctuations
in poverty without a real trend in either
direction.14 The average headcount index
was 53 per cent in 1951-55 (marked in
Figure 1), about the same as its average
value in 1970-74.15 After that there was a
significant decline in poverty incidence (and
the depth and severity of poverty fell too),16
though this was not a continuous decline.
It thus took over 20 years for the poverty
measures to finally fall below - and stay
below - their values in the early 1950s.
Changes in rural poverty closely follow
those at the national level, which is not
surprising given that a large proportion of
India’s population lives in rural areas (about
74 per cent even at the end of the period).
It is more notable that a similar pattern over
time also holds for urban poverty (Figure 1).
Common causative factors appear to be at
work.
The reduction in poverty since the early
1970s has been sizeable; between 1969-70
and 1992, the national headcount index
declined from 56 to 41 per cent. Yet India's
progress against poverty has been modest
when compared to the standards set by some
countries in east Asia. For example,
Indonesia’s headcount index was 58 percent
in 1970—very close to ourestimate forlndia
at that time. But by 1993 (keeping the same
real poverty line over time), we estimate that
2480
the headcount index for Indonesia had fallen
to 8 per cent, about one fifth of India’s
headcount index in 1992.”
a
How Important to India’s Poor was
Economic Growth and Contraction?
We look first at the effect of aggregate
economic growth and contraction on
poverty. Comparing successive survey
rounds, we regressed the percentage change
in each of the three poverty measures on
various measures of the rate of aggregate
economic growth between the same rounds.
Based on the regression coefficients,
Table 2 gives our estimates of the percent
age change in each poverty measure to be
expected from a 10 per cent growth rate
for (i) the mean consumption per person as
estimated from-'the'NSS; (ii) mean
consumption per person estimated from the
national accounts and population census;
and (iii). mean net , domestic product, per
person, also from the national accounts and
census.1’
The national poverty measures responded,
significantly to all three measures of.
economic growth. For example, a 10 per
cent increase in mean consumption resulted
in a 12-13 per cent drop in the proportion
of people who are poor, representing a 1011 per cent drop in the number of poor, at
India’s rate of population growth. The
responses are higher if one uses the NSS
estimate of mean consumption, father than.
the national accounts estimate, though the
difference is small for a given poverty
measure. The responses are lowest'for net
domestic product. This may be due to intertemporal consumption smoothing which may
make poverty (in terms of consumption) less
responsiveintheshort-tefmtoincomegrowth
than to consumption growth.'
Notice too that the responses tend to be
greater if one uses the poverty gap index
rather than the headcount index, and the
response is largest for the squared poverty
gap, which is sensitive to both the depth and
Table 1: Poverty in India 1951-1992
Period
Rural
Urban
National
1951-55
1956-60
1961-65
1966-70
1971-75
1976-83
1984-90
1991-92
54.77
53.96
48.59
60.44
55.27
47.96
37.94
39.44
42.70
47.06
45.46
50.90
46.04
38.08
34.99
33.24
52.66
52.74
48.02
58.60
53.39
45.68
37.20
37.84
1951-55
1956-60
1961-65
1966-70
1971-75
1976-83
1984-90
1991-92
19.69
17.91
14.28
19.80
17.01
13.84
9.26
9.47
14.04
15.36
14.04
16.08
13.46
10.60
9.11
8.58
18.70
17.46
14.23
19.08
16.28
13.09
9.22
9.24
1951-55
1956-60
1961-65
1966-70
1971-75
1976-83
1984-90
1991-92
9.42
7.94
5.73
8.67
7.08
5.45 . .
3.24 .
3.23
6.20
6.69
5.85
6.76
5.28
4.04
3.24
3.11
8.86
7.72
5.76
8.30
6.71
5.13
• 3.24
3.20
NSS Rounds
Headcount index
3-8
9-15
16-19
20-24
25, 27, 28
32, 38
42-45
46-48
Poverty gap index
3-8
9-15
16-19
20-24
25, 27, 28
32,38
42-45
46-48
Squared poverty gap index
3-8
9-15 '
16-19
20-24
25, 27, 28
32, 38
42-45
46-48
Economic and Political Weekly
Special Number September 1996
Figure 2: Cumulative Change in Headcount Index
Did the Pattern of Growth Matter?
(Total of Growth and Redistribution Components)
-X-
Growth component
-O-Redistribution
severity of poverty. This means that the
impacts of growth and contraction in India
were not confined to those near the poverty
line, but reached deeper.
Redistribution played a role in the long
run changes in poverty in India. Any change
in a poverty measure can be decomposed
into agrowth component and a redistribution
component.19 Roughly speaking, the growth
component is the change in the poverty
measure which would have occurred if
inequalities had not changed, while the
redistribution component is the change in
the poverty measure that one would have
found if there had been no change in the
mean. By adding each component over time
we can assess the cumulative total impact
of growth or redistribution. Figure 2 gives
the results.20
It can be seen that the redistribution
component did help over the whole period.
Thus, for India, our results reject the old
view (still held in some quarters) that
distribution must get worse as a low-income
country grows. Nonetheless, the overall
contribution of redistribution to change in
the headcount index has not been large in
the long run. The growth in mean
consumption has been more important,
accounting for about 80 per’ cent of the
cumulative decline by the end of the period.
Redistribution mattered more to the other
two poverty measures. For the poverty gap
index, the redistribution component
accounted for about 40 per cent of the
cumulative decline by the end of the period;
Economic and Political Weekly
— Total change
its contribution was 47 per cent for the
squared poverty gap index. Favourable
redistribution has thus been quite important
for changes in the depth and severity of
poverty.
Most of the pro-poor impact of
redistribution was realised early on, during
the early to mid-1960s, well before the onset
of the sustained decline in the national
poverty measures. Since the mid-1960s, the
redistribution component fluctuated without
making a further addition to its total longrun impact on national poverty. (This holds
for all three poverty measures.) The gains
to the poor since about 1970 have been
almost entirely due to growth.
The latter fl nding might be taken to imply
that public efforts at pro-poor redistribution
in the 1970s and 1980s failed. However, one
should be wary of drawing that conclusion
since we do not know the counter-factual
of what would have happened without those
efforts. Possibly distribution would havegot
worse.
Turning next to the sectoral composition
of growth, we found that the changes in
national poverty have been for the most pan
driven by changes in rural poverty. Figure
3 gives the cumulative (population-share
weighted) contributions of both the urban
and rural sectors to the national headcount
index. The rural sector accounted for more
than three-quarters of the total decline in
national poverty measures over the whole
period.2'Nonetheless-despitethesubstanti al
sectoral shifts in national output that have
occurred over the last 40 years or so poverty in Indiais still overwhelmingly rural.
In 1992, three-quarters of India's poor lived
in rural areas.
We also looked closely at the interlinkage
between the sectoral composition of
economic growth and the urban-rural
composition of poverty, using econometric
methods to disentangle the various effects
within and across sectors.22 The main
conclusion was that the relative effects of
growth within each sector, and its spillover
effects to the other sector, reinforced the
importance of rural economic growth to
national poverty reduction in India. Both the
urban and rural poor gained from growth
within the rural sector. By contrast, while
urban growth reduced urban poverty, it also
had adverse distribution effects within urban
areas which militated against potentially
higher gains to the urban poor. And urban
growth had no discernible impact on rural
poverty. The process of. growth through
rural-to-urban migration contributed very
little to poverty reduction in India.
When the growth in national income was
broken down by output-based sectors, we
found that there were marked sectoral
differences in the poverty impacts. Both
primary (mainly agriculture) and tertiary
(trade, services, transport et al) sectorgrowth
reduced poverty nationally, and they also
did so within both urban and rural areas. By
contrast, secondary (mainly manufacturing)
sector growth brought no discernible gains
to the poor in either sector. In the historical
shift from primary,to secondary and tertiary
sectors it was the latter sector which
delivered the bulk of the gains to India’s
poor..
V...
Table 2: How Responsive Were National Poverty Measures to
Economic Growth in India?
Percentage Change in the Poverty Measure
___________ Attributable to a 10 Per Cent Increase in_______
Mean Consumption ■
Mean Private 1^“'"
Mean Net
from National
Consumption from
Domestic
Sample Surveys
■'‘ National Accounts^
Product
Headcount index
Poverty gap index
Squared poverty gap index
Special Number September 1996
-13.3
-18.8
-22.6
Figure 3: Urban-Rural Composition of Change in the Headcount Index
(Cumulative Changes in the Urban and Rural Components and Population Shift Effects)
inflation. Nominal wages catch upeventually.
But we found that the adverse short-term
impact on the rural poor was sizeable. For
example, we estimate that a once-and-forall 20 per cent increase in the price level
would result in a drop of 13 per cent in the
current year's real wage rate in agriculture,
and an increase of 5 per cent in the rural
headcount index. The impacts on the other
poverty measures would be even higher;
the squared poverty gap would, rise by
9 per cent.
Did Some States Perform Better Than
Others in Reducing Rural Poverty?
-o- Rural component "^-Urban component
--Population shift —Total change
as to average farm yields. And wages also
responded significantly to farm yields,
presumably through effects on labour
demand, such as due to multiple cropping.
Higher yields thus helped reduce absolute
poverty through induced wage effects, as
well as the more direct channels, including
effects on both employment and own-farm
productivity.
Neither the poverty measures nor real
wage rates adjusted instantaneously to
Dm the Rural Poor Benefit from
changes in farm yields. The combined effect
Agricultural Growth?
of this stickiness in both variables was that
Since rural poverty has been so important, the short-run gains to poor people of
we turned our attention to this sector. Here agricultural productivity growth were far
we examined how much India's rural poor lower than the long-run impacts. Also, the
benefited from agricultural growth, what short-run effects on rural poverty operating
role the labour market played, whether the 'via the real wage rate were minor compared
impacts were distributionally biased one way to the direct effects of higher own-farm
or another, and how important macro yields. But in the long run, the wage effects
did matter, accounting for about one-third
economic stability was to the rural poor.
We collated the household survey data of the long run response of absolute poverty
with data on agricultural wages, prices and (for all three measures) to a yield increase.
output, and estimated a dynamic econometric The process through which India’s rural
model jointly determining rural poverty poor participated in the gains from agri
measures and real wages.1* The model, had cultural growth did take time, though about
a triangular structure in which the rural half of the long-run impact occurred within
poverty measure was hypothesised to be a three years of an initial gain in farm yield.
function of both the real agricultural wage
Dm Inflation Matter?
rate and the average farm yield per unit area
We found evidence of an adverse short(as well as other variables), and the real wage
rate was also a function of the farm yield run impact of inflation on real agricultural
and other variables.
wages and (hence) absolute poverty in rural
The resets indicated that all three poverty areas. The effect of inflation was to reduce
measures responded signi ficandy in the short real wages in the short term, because nominal
run to changeshrvagricultUTvL^ag^ wel) agricultural wages responded sluggishly to
The relative lack of an impactof secondary
sector growth on poverty reflects the type
of development strategy India pursued since
the second plan in the late 1950s, which
emphasised capital-intensive industrialisat
ion within a largely closed-economy regime.
It is not surprising that such industrialisation
brought negligible direct gains to the nation’s
poor, who depend heavily on the demand
for relatively unskilled labour.21
2482
The regional disparities in levels of living
in India have been large. For instance, the
proportion of the rural population of the
state of Bihar living in poverty in 1990-91
was about 58 per cent, more than three times
higher than the proportion (18 per cent) in
the(combined) states of Punjab and Haryana.
Some of these differences appear to have
persisted historically, though there were also
differential trends across regions. Looking
back over time, the more striking - though
often ignored - feature of the Indian
experience has been the markedly different
rates of progress between states; indeed the
ranking of states around 1990 looks quite
different to that 30 years earlier, as can be
seen in Figure 4 which compares headcount
indices around 1960 with those around 1990.
(The picture looks very similar for the other
two poverty measures.25) For example, the
southern state of Kerala moved from having
the second highest rural poverty rate around
1960 to having the fifth lowest around 1990.“
Regressing the logofeach poverty measure
against time, there was a trend decrease
(significant at the 5 per cent level or better)
in all three measures in 9 of the 15 states,
viz, Andhra Pradesh, Gujarat, Kerala,
Maharashtra. Orissa, Punjab and Haryana,
Tamil Nadu, Uttar Pradesh and West Bengal.
The trend was hot significantly different
from zero at the 5 per cent level in the other
six states of Assam, Bihar, Jammu and
Kashmir, Karnataka. Madhya Pradesh, and
Rajasthan; therewas not asignificant positive
trend for any state for any poverty measure.
There is a tendency for the absolute size of
the trend to be higher for the poverty gap
than the headcount index, and it was highest
for the squared poverty gap.
In terms of progress in both raising average
household consumption and reducing rural
poverty, the state of Kerala turns out to be
the best performer over this period. The
second, third and fourth highest trend rates
of consumption growth were Andhra
Pradesh, Tamil Nadu, and Maharashtra
respectively. In terms of the rates of poverty
reduction, the second, third and fourth stales
were Andhra Pradesh. Punjab and Haryana,
and Gujarat; the ranking is invariant to the
Economic and Political Weekly
Special Number September 1996
4-
Figure 4: Poverty by State, 1960-90
Averages for first three survey rounds and last three
choiceof poverty measure though differences
in their rates of poverty reduction are not
large. The worst performer was Assam by
all measures. The other poor performers
were Bihar, Jammu and Kashmir, Karnataka,
Madhya Pradesh and Rajasthan; the exact
ranking varies by the measure used.
The states which had the highest trend
rates of growth in mean consumption tended
to have the highest trend rates of poverty
reduction, and the correlation is very strong
(the correlation co-efficient between the trend
rate of reduction in the headcount index and
the trend rate of consumption growth is 0.85;
the correlation is about the same for the other
two poverty measures). Both these variables
may well have been influenced by similar
factors. Next we look at what those factors
might be.
What Accounts for the Differing Rates
of Progress in Reducing Poverty?
Every state has its own story, with a mixture
of both successes and failures at public action
against poverty in different periods.27 There
were di fferences between states i n the i mpacts
of (ostensibly similar) interventions, as well
as differences between states in the package
ofinterventions pursued; and in both respects
experiences in a given stale changed over
Economic and Political Weekly
time. We cannot hope to capture all this
variance in experience - for one thing we
would quickly run out of degrees of freedom.
Here our aim is solely to look for any
empirical regularities that can account for
at least a reasonable share of that variance.
The inter-state differences in progress at
fighting poverty allowed the project to study
the impact on the trend rate of poverty
reduction of a range of variables, including
regional differences in human and physical
resource development. A pooled model was
estimated, giving 31 Oobservations (15 states
over 21 NSS rounds, though with some
missing observations, or inadequate sample
sizes). A model was estimated for each
poverty measure, with both time varying
and static explanatory variables. The key
explanatory variables were current and
lagged real agricultural output per hectare,
current plus lagged real non-agricultural
output percapita, the rateof inflation, lagged
real state development spending per capita,
and thestate’s initial (around 1960) irrigation
rate, infant mortality rale, and female literacy
rale; the latter three variables were allowed
to influence the rate of change in the poverty
measures (thus entering the model interacted
with time).211 The estimated models could
account for about 90 percent of the variance
Special Number September 1996
over time and across states in the poverty
measures.2’
The results indicate that higher growth
rates in agricultural yields and real nonagricultural output per capita, lower rates of
inflation and higher growth in s'; tc
development expenditure all led to higher
rates of progress in both raising average
consumption and reducing all three measures
of absolute poverty.
The results also suggest that inter-state
differences in initial conditions of human
and physical resource development played
an important role; higher initial irrigation
intensity, higher literacy rates and lower
initial infant mortality rates all contributed
to higher rates of consumption growth and
poverty reduction. Initial inequalities in
access to physical and human infrastructure
appear to have been an important factor in
longer-term rates of poverty reduction.’"
Consider Bihar, one of the worst performers
in poverty reduction (Figure 4). The poor
in Bihar suffered from the state’s slow growth
in agricultural yields. But the state's poor
initial conditions were also an important
factor. The incidence of poverty in Bihar
declined at a trend rale of only 0.1 per cent
per year. We estimate that this would have
risen to 1.2 per cent if Bihar had started off
with Kerala’s level of human resource
development in the 1960s.
By and large, the same variables
determining growth in average consumption
mattered to rates of progress in reducing
poverty. Most of the effects on absolute
poverty were transmitted through growth in
average consumption rather than
redistribution, though none of the factors
which reduced absolute poverty had adverse
effects on distribution. Thus, there was no
sign of a trade-off between growth and pro
poor distributional outcomes.
Lessons for the Future
Our investigation suggests that economy
wide variables do matter to India’s poor;
they have generally gained from economic
growth, and lost from contraction; they have
also been hurt by inflation. The net gains
to the poor since the early 1970s can be
attributed in large part to economic growth
- distribution changed little from the point
of view of the poor, though it appears to have
been more important in the 1950s and 1960s,
when there was rather less growth.
The experience of the past 40 years offers
support for the view that a stable macro
policy environment, combined with micro
policy reforms conducive to economic
growth, can help greatly in reducing absolute
poverty in India. However, our results also
reveal important nuances concerning the
pattern of growth, and the importance of
other contingent factors, including human
and physical infrastructure.
2-183
corrects this by replacing the firewood sub
trickle down associated with agricultural
series in the CPIAL by one based on mean
growth” with Saith’s (1981:205) claim that
rural firewood prices (only available from
"there can be little doubt that current growth
1970) and a scries derived by assuming that
processs have served as generators ofpoverty”;
firewood prices increased at the same rate as
both were using data for India over roughly
all other items in the fuel and light category
the same period (1957-73). The debate
(prior to 1970). For details see Datt (1996).
continues; in recent literature on India one
10
These are fixed weight price indices. Thus,
can find claims that “rapid agricultural growth
they ignore substitution in response to shifts
has benefited all classes of the poor" [Singh
in relative prices. To test sensitivity to this.
1990] and “acceleration in agricultural growth
Ravallion and Subramanian (.1996) compare
by itself is unlikely to make a dent in rural
poverty measures for India with and without
poverty” [Gaiha 1995:285].
an allowance forsubstitution effects consistent
3 For an overview of the theory and evidence
with demand behaviour, as modelled by a set
on the effects of adjustment on the poor see
of full rank Gorman Engel curves. Ignoring
Lipton and Ravallion (1995, section 5.3). In
substitution matters far more for some
the Indian setting, see Ravallion and Subbarao
measures and applications than others. It leads
(1992).
to overestimation of inequality, but level
4 See Datt (1996), Ozleret al (1996), Ravallion
effects on poverty measures are generally
and Datt (1995,1996), and Datt and Ravallion
small and turning point errors are rare.
(1996). Later we identify which paper is most
11
These are members of a class of measures
relevant to each topic covered here.
proposed by Foster, Greer and Thorbecke
5 A number of the popular methods of making
(1984). A transfer of income from a poor
poverty comparisons over time and space do
person to a poorer person (for example) will
not satisfy this consistency requirement; see
not alter either the head-count index or the
Ravallion (1994) for further discussion.
poverty gap index, but it will decrease the
6 For further discussion of this point see Sen
squared poverty gap index. Furthermore, the
(1987); in the context of India also see Dreze
squared poverty gap index satisfies the
and Sen (1995).
subgroup consistency’ property, namely that
7 See Planning Commission (1993).
if poverty increases in any subgroup (say the
8 We compared this difference in the poverty
urban sector), and it does not decrease
lines to independent estimates of the urbanelsewhere then aggregate poverty must also
rural cost of living differential. For 1973-74,
increase [Foster and Shorrocks 1991].
Bhattacharya et al (1980) estimated that the
12
The poverty measures are calculated using
cost-of-living for the poor was 16 per cent
parameterised Lorenz curves. We use either
higher in urban areas, exactly the same (to
the beta Lorenz curve of Kakwani (1980) or
the nearest integer) as the differential in
the general quadratic model of the Lorenz
poverty lines. So it can be en argued that the
curve of Villasenor and Arnold (1989),
planning, commission’s urban and rural
depending on which fits the data best (bothpoverty lines represent the same standard of
satisfied the theoretical conditions needed for
living, and (hence) that we are making
a valid Lorenz curve in all survey rounds for
consistent comparisons of absolute poverty
both sectors). Using the formulae derived in
between urban and rural areas. For further
Datt and Ravallion (1992), the poverty
discussion see Ravallion and Datt (1996).
measures are calculated from the estimated
9 For the urban sector after August 1968, the
parameters of the Lorenz curve and the mean
all-India consumer price index for industrial
Notes
percapitaconsumption expenditure. A number
workers (CPIIW) is used as the deflator. For
of checks are made on the results, including
the earlier period, the Labour Bureau’s
[For their comments we are grateful to Jyotsna
consumer price index for the working class
both the theoretical conditions for a valid
Jalan, Peter Lanjouw, Dominique van de Walle,
Lorenz curve, and consistency checks, such
is used, which is an earlier incarnation of the
Quentin Wodon, and the EPW’s referee. This
CPIIW albeit with asmaller coverage of urban
as that the estimated value of the head-count
paper summarises results of a research project,
index must lie within the relevant class interval
centres (27 against 50). The rural cost of
‘Poverty in India, 1951-92’, supported by the
living index series was constructed in three
of the published distribution. The estimation
World Bank’s Research Committee, under RPO
parts. For the period since September 1964,
technique has been set-up in a user-friendly
677-82. However, these are the views of the
the rural cost of living index is the all-India
computer programme ‘POVCAL’ [Chen, Datt
authors, and should not be attributed to the World
consumerprice index for agricultural labourers
and Ravallion 1991] which is available on
Bank, or any affiliated organisation.]
(CPIAL) published by the Labour Bureau.
request, so interested readers can readily check
For the period September 1956 to August
our calculations and their sensitivity to our
1
See Ravallion and Chen (1996) for aggregate
1964 (for which an all-India CPIAL does not
poverty measures for the developing world
assumptions.
i
exist), a monthly series of the all-India CPIAL
13
A complete descriptions of the data set and
over the period 1987-93. They estimate that
was
constructed
as
a
weighted
average
of
the
the percentage of the population consuming
all sources can be found in Ozler, Datt and
_ less than Sl/day at 1985 international prices
state-level CPIALs, using the same state
Ravallion (1996) with an accompanying set
level weights as those used in the all-India
(with'currency conversion^ at purchasing
of diskettes.
power parity) decreased only slightly over
CPIAL published since September 1964. For
14
The first subperiod is marked by three
the initial period August 1951 to August
this period, from 30.7 per cent in 1987 to 29.4
significant peaks in poverty around the years
per cent in 1993 implying that the numbers
1956, forecasts were obtained from a dynamic
1953-55 (rounds 7,8), 1956-58 (rounds 11.
of people living under $ 1/day rose from 1.23
model of the CPIAL as a function of the
12, 13), and 1966-68 (rounds 21,22), the last
billion to 1.32 billion over this period.
CPIIW and the wholesale price index. Our
of these coinciding with the worst drought
(Sl/day is about equal to India’s urban poverty
new CPIAL series also dealt with another
in the post-independence period.
line.) The gains to the poor in east and (less
problem which has to do with the fact that
15
Based on poverty measures averaged over
so) south Asia were roughly counter-balanced
the Labour Bureau has used the same price
NSS rounds, weighted by the duration of the
of firewood in its published series since 1960by the losses in other regions, notably Subsurvey.
Saharan Africa, Latin America and Eastern
61. Firewood is typically a common property
16
The absence of fluctuations in poverty over
Europe and Central Asia.
resource for agricultural labourers, but it is
the period 1975-85 may be somewhat illusory
also a market good, and so the Labour Bureau’s
2 Contrast, forexample, Ahluwalia’s (1978:320)
as we have only two NSS surveys in the
conclusion that “there is evidence of some
practice is questionable. Our CPIAL series
intervening period, viz, those for 1977-78 and
Our results point clearly to the quantitative
importance of the sectoral composition of
economic growth to poverty reduction in
India. Fostering the conditions for growth
in the rural economy - both primary and
tertiary sectors - must be considered central
to an effective strategy for poverty reduction
in India. At the same time, the relative failure
of India’s past industrialisation strategy from
the perspective of the poor points to the
imp'ortance of successful transition to a
strategy capable of absorbing more labour,
particularly from rural areas.
But our results also point to the longerterm importance of investing in human and
physical infrastructure as a complement to
pro-growth reforms in India. Controlling for
growth in farm and non-farm sectors, we
find significant effects on trends in absolute
poverty reduction of the differences between
states in initial conditions related to
infrastructure.
A final lesson concerns the importance of
being able to credibly assess an economy’s
performance in reducing poverty. Though
less than ideal in some respects, the data base
available for poverty analysis in India is
good by international standards. Many other
countries have had far fewer objective socioeconomic surveys on which poverty
monitoring can be based, or their surveys
have been severely wanting in terms of
coverage (lacking, for example, a sound
consumption module) or comparability over
time. The very fact that for India we can
obtain the data needed to address the
questions posed above carries an important
message for other countries today, and India
in the future.
2484
Economic and Political Weekly
Special Number September 1996
19S3. In particular, we do not have poverty
and other papers in the October 14-21, 1995
Foster, James and A F Shorrocks (1991).
estimates for the two drought years 1979 and
issue of the Economic and Political Weekly.
‘Subgroup Consistent Poverty Indices’,
1982.
Progress does not appear to have been even
Econometrica, 59:687-709.
17 The Indonesia estimates are from the national within states either, though our data do not Gaiha, Raghav (1995): ‘Docs Agricultural Growth
socio-economic surveys (SUSENAS) done
allow us to disaggregate further. For evidence
Matter to Poverty Alleviation?’, Development
by Indonesia’s central bureau of statistics.
on changes between 1972-73 and 1987-88 at
and Change, 26(2):285-304.
The 1970 number is from World Bank (1990.
the level of the NSS regions see Dreze and
Kakwani, Nanak (1980): ‘On a Class of Poverty
Table 3.2); the 1993 number is our estimate
Srinivasan (1996).
Measures’. Econometrica, 48:437-46.
from the 1993 SUSENAS using the same real
28 Theestimated model also included a correction Kakwani, Nanak and K Subbarao (1993): ‘Rural
poverty line. The consumer price index was
forserial correlation in the residuals (allowing
Poverty and its Alleviation in India’, in
used to convert to constant prices.
for the uneven spacing of the NSS rounds,
Michael Lipton and Jacques van dcr Gaag
18 The estimates in Table 2 are based on
requiringnon-linearestimation methods); see
(eds). Including the Poor, World Bank.
regression of the percentage change in the
Datt and Ravallion (1996) for details.
Lipton. Michael and Martin Ravallion (1995):
poverty measures against the percentage
29 For full details on the estimated models and
‘Poverty and Policy’ in Jere Bchrman and T
change in consumption or net product per
various specification tests see Datt and
N Srinivasan (eds) Handbook ofDevelopment
person using 33 household surveys spanning
Ravallion (1996).
Economics, Volume 3, North-Holland.
1951-1991 for the surveys-based mean
30 Kerala stands out as an unusual case in Ozler, Berk, Gaurav Datt and Martin Ravallion
consumption, and 23 surveys spanning 1958India, given its high level of human resource
(1996): ‘A Database on Poverty and Growth
91 for consumption or income from the
development, including at the beginning of
in India’, (mimeo). Policy Research
national accounts. All regression coefficients
the period under study. [For further
Department, World Bank.
were statistically significant at the 1 per cent
discussion of Kerala’s achievements in this
Planning Commission (1993): Report of the
level or better, and all regression comfortably
respect see Dreze and Sen 1995.] However,
Expert Group on Estimation of Proportion
passed all the standard specifications tests.
even if we drop Kerala from the regressions,
and Number of Poor, Government of India,
For full details see Ravallion and Datt (1996).
the initial conditions in human resource
New Delhi.
19 Using the methodology outlined in Datt and
development (and indeed all other variables
Ravallion, Martin (1994): Poverty Comparisons,
Ravallion (1992).
in our model) remain significant [Datt and
Harwood Academic Press, Chur, Switzerland,
20 The observed poverty measures are subject
Ravallion 1996]. Their effect is not just due
Fundamentals in Pure and Applied Economics,
to large fluctuations from one NSS round to
to Kerala.
Volume 56.
another, this is particularly true of some of
Ravallion, Martin and Shaohua Chen (1996):
the shorter initial NSS rounds. The cumulative
‘What Can New Survey Data Tell Us About
References
series can thus be somewhat misleading for
Recent Changes in Living Standards in
arbitrary choice of the starting point. To deal
Developing
and Transitional Economies?’,
Ahluwalia, Montek S (1978): ‘Rural Poverty and
with this problem, we selected NSS round 6
Working Paper 1, Research Project on Social
Agricultural Performance in India’, Journal
(May-September 1953) as the reference date
of Development Studies, 14: 298-323.
and Environmental Consequences of Growthfor the first decomposition; the poverty
Bhattacharya, S S, A B Roy Choudhury and P
Oriented Policies, World Bank, Washington,
measures for this round reasonably
DC.
D Joshi (1980): ‘Regional Consumer Price
approximate the average poverty measures
Ravallion, Martin and Gaurav Datt (1995):
Indices Based on NSS Household Expenditure
for 1951-55 (weighted average for rounds 3Data’, Sarvekshana, 3:107-21.
‘Growth, Wages and Poverty: Time Series
8, with weights proportional to the number
Chen, Shaohua, Gaurav Datt and Martin Ravallion
Evidence for Rural India’, (mimeo): Policy
of months in the survey period of each round).
Research
Department, World Bank.
(1991): ‘POVCAL: A User-Friendly
The graphs of the decompositions for NSS
- (1996): ‘How Important to India’s Poor is the
Programme for Poverty Analysis Using
round 9 onwards in Figures 2 and 3 can thus
Sectoral Composition ofGrowth?' World Bank
Grouped Data’, Policy Research Department,
be interpreted as referring to the cumulative
Economic Review, 10:1-26.
World Bank.
change in poverty (and its components) since
Ravallion, Martin and K Subbarao (1992):
Datt, Gaurav (1996): ‘Poverty in India, 1951the mid-1950s.
‘Adjustment and Human Development in
1992: Trends and Decompositions’ (mimeo).
21 See Datt (1996) for details.
India’, Journal of Indian School of Political
Policy Research Department, World Bank.
22 See Ravallion and Datt (1996) for details on Datt, Gauraii and Martin Ravallion (1992):
Economy, 4:55-79.
the methods used.
Ravallion, Martin and S Subramanian (1996):
‘Growth and Redistribution Components of
23 This has long been recognised in discussions
‘Welfare Measurement With and Without
Changes in Poverty Measures: A Decom
of poverty in India, and the Second Plan was
Substitution’, (mimeo), Policy Research
position with Applications to Brazil and India
criticised for this reason at the time (see, for
Department, World Bank.
in the 1980s’, Journal of Development
Saith, Ashwani (1981): ‘Production, Prices and
example. Vakil and Brahmanand’s (1956)
Economics, 38:275-295.
Poverty in Rural India’, Journal of
comments on the Second Plan). For a recent
- (1996): ‘Why Have Some States of India
discussion in the Indian context see Dev et
Development Studies, 19:196-214.
Performed Better than Others in Reducing
Sen. Amartya (1987): Commodities and
al (1992). Recent affirmations of the
Absolute Poverty?’ Policy Research Work
Capabilities, Oxford University Press, Delhi.
importance of labour-demanding growth to
ing Paper 1594, World Bank, Washington,
poverty reduction in developing countries
Singh, Inderjit (1990): The Great Ascent: The
DC.
Rural Poor in South Asia, Johns Hopkins
include World Bank (1990), Eswaran and
Dev, S Mahendra, M H Suryanarayana and
University Press for the World Bank.
Kotwal (1994) and Lipton and Ravallion
Kirit S Parikh (1992): ‘Rural Poverty in
Vakil, C N, and P R Brahmanand (1956): Planning
(1995).
India: Incidence, Issues and Policies’, Asian
for an Expanding Economy, Vora and
24 For details on methodology see Ravallion and
Development Review, 10:35-66.
Datt (1995). On the link between real wages
Dreze. Jean and Amartya Sen (1995): India:
Company, Bombay.
in agriculture and rural poverty measures for
Economic Development and Social van de Walle, Dominique (1985): ‘Population
India also see van de Walle (1985).
Growth
and Poverty: Another Look at the
Opportunity, Oxford University Press, Delhi.
25 See Datt and Ravallion (1996) for more Dreze. Jean and P V Srinivasan (1996): ‘Poverty
Indian Time Series Data’, Journal of
detail.
Development Studies, 21:429-39.
in India: Regional Estimates 1989’,
26 Some of national samples, particularly in the
Villasenor, J and B Arnold (1989): ‘Elliptical
Development Economics Research Pro
1950s, were too small to allow reliable
Lorenz Curves’, Journal of Econometrics,
gramme Working Paper 70, London School
estimation at the state level, so this part of
of Economics.
40:327-38.
the analysis focused on the period from about
Vyas, V S and Pradeep Bhargava (1995): ‘Public
Eswaran, Mukesh and Ashok Kotwal (1994):
1960 onwards.
, Intervention for Poverty Alleviation: An
Why Poverty Persists in India, Oxford
. -27 For discussion of the range of experiences in
Overvie\v’, Economic and Political Weekly,
University Press, New Delhi.
direct interventions for poverty reduction see
30 (October 14-21): 2559-72.
Foster, James, J Greer and Erik Thorbecke (1984):
Kakwani and Subbarao (1993), Dreze and
‘A Class of Decomposable Poverty Measures’, ' World Bank (1990): World Development Report:
Sen (1995), and Vyas and Bhargava (1995)
. Poverty, Oxford University Press, Delhi.
Econometrica, 52:761-65.
Economic and Political Weekly
Special Number September 1996
2485
Economic Reforms, Employment and Poverty
Trends and Options
Abhijit Sen
E.
f
If poverty reduction is to be a serious part of the agenda of economic reforms, the reforms will have to have an
explicitly redistributive content. This will require cuts in subsidies to the rich and also higher taxes to maintain and
F increase the expenditure relevant for the poor. In addition, the old issues of land distribution and provision of universal
primary education and health must be put back on the agenda.
■ ■ But, more than anything else, it must be recognised that a reforms strategy which aims to withdraw the state from
investment, liberalise finance and thus divert finances from the state to the private sector, liberalise agricultural
trade and thus enrich the rich at the direct cost of the poor and seeks to control inflation and balance of payments
problems through deflation and devaluation is at its root a fundamentally inequitous adventure.
THIS paper is concerned with the possible
impact of the economic reforms undertaken
) nAithe government of India in the 1990s on
nature and incidence of poverty in India.
The point of departure is the observation
from NSS data that poverty, which had not
) i showed any time trend at all till the mid■ 1970s, declined significantly between the
j j mid-1970s and the end-1980 but appears to
?■ j have increased again in the 1990s. In other
; '• ■’ words, poverty appears to have declined
• only in the decade and a half beginning the
( : mid-1970s during which there was an
- i explosion in public expenditure leading up
I
lothefiscalcrisiswhich,amongotherthings,
1 precipitated the economic reforms in 1991.
f This suggests that there might be a much
■' stronger link between public expenditure
ad poverty reduction than is usually
5 ; appreciated, and this in turn has the
f
implication that the reforms process may
.' actually impinge adversely on the poor if its
' .? focus continues to be on the reduction of
/
expenditure.
7 For this reason it is important to identify
the direct and indirect effects of public
expenditure, and of other aspects of the
economic reform policies, on poverty
alleviation. This paper is a very preliminary
effort in this direction. In the first section.
1 brief outline of the trends in and structure
of poverty in India is presented, with a view
to identifying the important characteristics
of the poor population. This allows for an
■ estimation of the likely effects of such policies
on the material condition of the poor and
those close to the poverty line. Since poverty
' is found to be closely related to employment
and occupational characteristics, a discussion
on past employment trends and poverty trends
is included in the next section. In the
following section, there is a more detailed
consideration of the recent trends in poverty;
and this is followed by a section which deals
specifically with the statistical determinants
1 of poverty as well as the relationship between
• Sris and economic growth. Finally, the last
)
'T? Economic and Political Weekly
section sets out some brief conclusions in
terms ofdifferent policy options for economic
reform which make poverty reduction an
explicit objective.
I
Long-term Trends and
Profile of Poverty
The Economic Survey 1995-96 has claimed
that “the percentage of India’s population
below the poverty line has declined from
25.94 per cent in 1987-88 to 18.96 per cent
in 1993-94”. This claim is based on estimates
made by the Planning Commission using a
methodology whereby the consumption
distribution obtained from the National
Sample Survey (NSS) are applied to total
estimates of consumption expenditure as
obtained from the Central Statistical
Organisation’s (CSO) compilation of
National Accounts. On this basis, the rural
poverty ratio declined from 28.37 per cent
in 1987-88 to 21.68 per cent in 1993-94
while the urban poverty ratio fell from 16.82
per cent to 11.55 per cent. These figures,
which have been used to claim that there has
been no increase in poverty following the
economic reforms initiated in 1991, have in
turn been challenged by independent
analysts.1 The criticism takes two forms.
First, that even using the Planning
Commission method, poverty in 1993-94
was higher than in 1990-91 just before
reforms began and so the comparison with
1987-88, a drought year, gives a misleading
trend. Second, and much more importantly,
that the Planning Commission method is
itself flawed as was pointed out in 1993 by
the high-level Expert Group on Estimation
of Proportion and Number of Poor. Using
the methodology suggested by this expert
group, not only arc the poverty figures
much higher, these show that there is no
real trend decline in poverty since around
1986, that poverty increased massively
between 1989-90 and 1992, and that although
Special Number September 1996
poverty fell in 1993-94 this was still higher
than in the immediate pre-reform years
1989-90 or 1990-91.
Some of the issues which arise from these
different estimates are discussed in a later
section. Here, we need to outline the long
term trends in poverty, and for this we present
in Table 1. estimates from a third source
altogether - that compiled by the Poverty
and Human Resources Division of the World
Bank,2 also using NSS data. This source
gives a long scries from 1951 onwards, and
the main message which emerges is
important. This is that there was no long
term time trend in poverty from 1950-51 to
1973-74 but that there was thereafter a
sharp decline in poverty till 1986-87. After
1986-87, the decline continued at a slower
pace till 1989-90 when it was reversed, with
a particularly sharp increase in poverty in
1992. Poverty declined again in 1993-94 so
that rural poverty in 1993-94 although higher
than in 1989-90 or 1990-91 just before the
reforms, was at about the same level as in
1986-87. Urban poverty, which had not
increased particularly in 1992, was, however,
lower in 1993-94 than in any pre-reform
year.
These trends are important for a number
of reasons. First, the trend in rural poverty
shows a very close similarity with trends in
agricultural wages. Estimates of real
agricultural wages from a number of sources
also show stagnation till the mid-1970s with
sharp increases thereafter till the end-1980s
when there is a slow-down again. Second,
the period of declining poverty (mid-1970
to end-1980) was relatively short, and one
which was marked by increasing government
expenditure leading to severe fiscal
imbalances by 1990. Third, that this period
of declining poverty was infact one when
rural poverty declined faster than urban
poverty. Fourth, that rural poverty stopped
falling, and indeed increased, as soon as
fiscal stabilisation was attempted after 1991,
and during this latest period the gap between
2459
• : and urban poverty has again tended to
rease. These trends require explanation
J analysis, and this is the main focus of
is paper. In the remaining part of this
,-tion, we provide a brief outline of the
ofile of Indian poverty?
The most comprehensive data on the
tructureof poverty remains the information
tat can be gleaned from the NSS large
unple survey of 1987-88, since details of
iC more recent large sample survey
■nducted in 1993-94 are not yet available.
seems reasonable to assume that in broad
■•ntours the picture that emerges for 1987>8 remains valid for the early 1990s.
Some of the evidence on the structure of
xiverty in India in 1987-88 is provided in
'able 2. The first and most obvious point
o be made relates to the dominantly rural
t’lure of the poor population. The poor in
oral areas constituted around three-fourths
f the total poor population. This has to be
uxtaposed with the fact that subsequently
urban poverty has declined at a faster rale,
to that poverty has become even more rural
n nature. Within the rural areas, there is also
rvidence of greater regional concentration
af poverty, with some backward regions
displaying a very high incidence of poverty
discussed below.
In the rural areas at an all-India level, the
worst off economic group is that of rural
labour, both agricultural and nonagricultural. This is true both in terms of
depth of poverty and its severity in terms
if distance of average incomes from the
..’lual poverty line. Within this broad
r-tegory of rural labour, casual labour on
ion-permanent contracts is the most
susceptible to absolute poverty. There is no
discernible difference in poverty ratios
between agricultural and non-agricultural
casual labourers, which is not surprising,
since the casual labour populations tends to
move between agricultural and nonrtgricultural occupations as they become
available. The self-employed rural
households, whether agricultural or nonagricultural, tend to experience much lower
levels of economic deprivation than other
rural groups.
Female-headed rural households recorded
a higher than average incidence of poverty,
both in terms of prevalence and severity.
Those rural households classified as poor
tended to have higher than average
representation of adult females and lower
than average representation of adult males.
Also, poor households in general tended to
have higher dependency ratios, so that
children dominated in the number of poor
persons, and were ovcr-rcprcscntcd in the
poor population relative to the total
population. Also, there arc definite social
dimensions to material deprivation, with the
category of scheduled castes and scheduled
’160
tribes recording higher extent and severity
of poverty than the general rural population.
In fact, scheduled tribe groups arc even
worse off than scheduled castes on average,
and tend to be the most economically destitute
of all the rural population.
The urban areas present a slightly different
picture. Firstly, the poor are more
economically and socially heterogenous.
Thus, the most important occupational groups
among the poor urban population are those
employed in casual labour, as well as a
section of the self-employed. The selfemployed category is highly heterogenous
in urban areas, comprising both highly paid
professional occupations as well as informal
sector low paying activities. The latter
constitutes among the poorest of the urban
population, along with workers employed in
insecure casual contracts. Clearly, the
irrcgularandinsecurcnatureofsuch incomes,
which are also typically low, is the major
source of poverty in urban households.
Scheduled castes and tribes were less
significant among the poor in urban areas
than in rural ones, and there was no real
■..I. .... ,:f iugioiiai Ji..parities ; t urban
i------- ,.,vbl.i.. •?..;v-’rt.among Icn.aie-iieudcu households was tai
more serious in the urban areas. Despite this,
the dependency ratio among poor urban
households was slightly lower than among
their urban counterparts.
In terms of regional concentration of
poverty, only two states - Bihar and Uttar
Pradesh - together accounted for 34 per cent
of the total poor population in 1987-88. In
Bihar in particular, there was a large over
representation of poor people, and there is
no reason to believe that this has altered
dramatically. Another six states - Andhra
Pradesh, Madhya Pradesh, Maharashtra,
Orissa, Tamil Nadu and West Bengal accounted for a further 43 per cent of the
poor. For rural poverty in particular, there
was over-representation of the poor in
Madhya Pradesh, Maharashtra, Orissa and
Tamil Nadu. In thcstalcsofGujarat,Rajasthan
and Orissa, scheduled castes and tribes
accounted for more than half of the poor well above their share in total population.
Scheduled tribes, especially in these states,
real ■
g.c
impo
°C
Q
. inpo
i. I>W
tib
: P£-
■ The j
Table I: Poverty Estimates 1951-94
NSS
Round
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
27
28
32
38
42
43
44
45
46
47
48
50
H
Rural
PG
SPG
H
Urban
PG
SPG
47.37
43.87
48.21
54 13
61.29
64.24
51.83
48.34
58.86
62.11
55.16
53.26
50.89
45.40
47.20
48.53
53.66
57.60
64.30
63.67
59.00
57.61
54.84
55.36
55.72
50.60
45.31
38.81
39.60
39.06
34.30
36.43
37.42
43.47
38.74
16.05
14.64
16.29
19.03
21.95
25,04
18.44
15.65
19.45
21.69
19.01
17.74
15.29
13.60
13.60
13.88
16.08
17.97
22.01
21.80
18.96
18.24
16.55
17.35
17.18
15.03
12.65
10.01
9.70
9.50
7.80 '
8.64
8.29
10.88
9.41
7.53
6.71
7.56
9.12
10.26
12.50
8.80
6.71
8.50
10.01
8.78
7.88
6.13
5.53
5.31
5.49
6.60
7.60
10.01
9.85
8.17
7.73
6.80
7.33
7.13
6.06
4.84
3.70
3.40
3.29
2.58
2.93
2.68
3.81
3.27
35.46
36.71
40.14
42.77
49.92
46.19
43.92
43.15
51.45
48.88
47.75
44.76
49.17
44.65
43.55
44.83
48.78
52.90
52.24
52.91
49.29
47.16
44.98
45.67
47.96
40.50
35.65
34.29
35.65
36.60
33.40
32.76
33.23
33.73
30.03
11.14
10.91
13.25
13.83
17.24
15.76
14.65
13.34
18.16
16.31
15.95
13.75
15.83
13.84
13.79
13.29
15.24
16.82
16.81
16.93
15.54
14.32
13.35
13.46
13.60
11.69
9.52
9.10
9.31
9.54
8.51
8.51
8.24
8.82
7.62
4.82
4.41
5.96
6.29
7.74
7.02
6.40
5.41
8.51
7.25
7.00
5.87
6.75
5.83
6.05
5.17
6.38
6.98
7.19
7.22
6.54
5.86
5.35
5.26
5.22
4.53
3.56
3.40
3.25
3.29
3.04
3.12
2.90
3.19
2.76
Period
Aug 51-Nov 51
Apr 52-Sep 52
Dec 52-Mar 53
May 53-Sep 53
Oct 53-Mar 54
Jul 54-Mar 55
May 55-Nov 55
Dec 55-May 56
Aug 56-Feb 57
Mar 57-Aug 57
Sep 57-May 58
Jul 58-Jun 59
Jul 59-Jun 60
Jul60-Aug6!
Sep 61 -Jul 62
Feb 63-Jan 64
Jul 64-Jun 65
Jul 65-Jun 66
Jul 66-Jun 67
Jul 67-Jun 68
Jul 68-Jun 69
Jul 69-Jun 70
Jul 70-Jun 71
Oct 72-Sep 73
Oct 73-Jun 74
Jul 77-Jun 78
Jan 83-Dcc 83
Jul 86-Jun 87
Jul 87-Jun 88
Jul 88-Jun 89
Jul 89-Jun90
Jul90-Jun9l
Jul 91-Dec 91
Jan 92-Dcc 92
Jul 93-Jun 94
I
i drop
Nines: H: head count ratio of poverty; PG: poverty gap ratio; SPG: squared poverty gap.
Source: B Ozler, G Dutt and M Ravallion, 'A Database on Poverty and Growth in India', The World
Bank, January 1996, for estimates up to the 48lh round; For 50th round, NSS data has been used
to calculate the estimates using exactly the same methodology ns in the rest of the series.
Economic and Political Weekly
Special Number September 1996
: toSk
l Ucnd
thr
’
econ
:
; sugp
< orga
•
[ buttl
I. ‘C
f grow
K
6 gro«
pc
tpopc
be
2® were found to be among the most absolutely
it deprived and destitute of all Indians.
'® There is a close relationship between the
' extent of poverty and patterns of employment
mdrcal wages. In the rural areas in particular
-as argued in the section below-two factors
> ire critical (in addition to food prices) in
explaining the incidence of poverty both
* overtime and across states and regions: the
behaviour of employment including the
V degree of diversification away from purely
igricultural employment, and movements in
> real wages. For this reason, trends in the
jmwth and pattern of employment arc very
important indicators of theextent and severity
of poverty.
II
Trends after Mid-1970s
e essential point which emerges from
cvious section is that a sustained decline
h. m poverty is observable only after 1973-74,
• md that this process was over by 1990-91.
Io fact, the decline was almost certainly a
phenomenon which began after 1975-76.
The years 1972-75 were difficult years with
bigh inflation was low growth, and the
top in poverty in I977-7S as compared
» 1973-74 is to a large extent attributable
lothccomparisonofan excellent agricultural
tear with a year when output was below
s rend. For this reason it is more reasonable
• Iodate the beginning of the poverty decline
to around 1977-78.
As is well known by now, during the
period 1977-91 (and particularly during the
1980s) the Indian economy underwent a
cnsumption-lcd boom, spurred on by increa
sing revenue deficits of the government, and
financed in large part by high deficits on the
ext^Bl current account. This is the boom
which went bust in 1991, laying the basis
- fcr‘reforms’. But since this boom and bust
cycle is paralleled fairly closely by what
' ■
happened to rural poverty, it is worth
recounting some of its more important
features. First, the boom was possible at all
_. because, with increased access to external
debt and with agricultural growth higher
• dun the long-term average, the Indian
, economy was much better placed on the
' supply side, with both of her two traditional
■ ^supply constraints greatly cased.
j ;. Second, during this boom it was the
■ '. cejanised sectors of the economy which
. pr# fastest in terms of incomes and output,
V \ fathis growth did not lead to much increase
* feoeganised sector employment. The rate of
■ 'ipwrth of organised sector employment
_ .^federated significantly, and the 1980s
■ r powth of such employment was, at 1.5 per
I' 'Scat per annum, much less than the rate of
m tfpcpulation growth. Within this, employment
; -C nthe private organised sector was the most
#:rgish, averaging a growth rate of only
•
' ' T.xnomic and Political Wi
0.2 per cent per annum, and there was slow
growth also of employment provided by the
Central government and its industrial
undertakings. In fact, whatever employment
growth occurred in the organised sector was
provided mainly by state governments and
certain quasi-government organisations, for
example the nationalised banks. Moreover,
during this period there was also a sharp
drop in labour absorption by the agricultural
sector, and agricultural employment also
grew at a rate substantially below the rate
of population growth, and below rates of
growth achieved in the past at times of lower
output growth. Thus, the rapid growth of
output in agriculture and in the organised
private sector failed to translate itself into
highcrdirect employment in these important
sectors.
Nonetheless, and this is the third important
point, this decade was characterised by rising
real wages and a fairly sharp drop in both
the incidence and the severity of poverty,
particularly in rural India According to
calculations made for this paper, using data
from the National Sample Survey and
following the methodology recommended
by the Expert Group on Estimation of
Propotion and Number of Poor, there was
a steady decline in the head count measure
of poverty for the rural population from 56.4
in 1973-74 to 53.1 per cent in 1977-78 to
45.6 in 1983, 38.3 in 1986-87 and to 37.9
per cent in 1989-90.4 The urban poverty
ratio similarly fell steadily from 49.2 per
cent to 32.4 percent during the same period.
This meant that the incidence of poverty
which had fluctuated, positively with
inflation and negatively with agricultural
output, with it anything a positive underlying
trend up to the mid-1970s began to decline
thereafter. The most important reason for
this was the fact that real wages of unskilled
labour increased significantly in both urban
and rural areas. Several alternative sources
of data are available .for agricultural wages,
and all of these suggest that real agricultural
wages increased by around 50 per cent during
the decade - an increase almost double the
increase in labour productivity in agriculture
during this period.
These observations indicate that there were
important changes in the nature of inter
sectoral and other linkages in the economy.
One important point is that, although
agriculture continues to be the largest
employer of the workforce and productivity
increases here arc of major weight in the
economy, the rest of the Indian economy
appears to have become progressively less
dependent on the behaviour of agricultural
output during the 1980s. This is evident
from the fact that the period of relative
stagnation in agricultural output 1983-87
was nevertheless marked by high growth
rates in non-agriculture, and, more generally,
econometric evidence suggests that the earlier
dependence of aggregate economic growth
on the behaviour of the monsoon seems to
have diminished. There arc three major
reasons for this. First, the sharp decline in
the employment elasticity of output of the
organised sectors of the economy meant that
increased output in industry and services
today involves a much lower concomitant
increase in the demand for wage goods.
Table 2: Profile of Poverty in India 1987-88
Groups
Self-emp agriculture
Self-employed nonagriculture
All self-employed
Agricultural labour
Other labour
Others
Scheduled castes
Scheduled tribes
Female-headed HH
All households
Population
Share
Rural
Per Cent
Poor
44.3
38.3
37.9
12.5
56.7
27.1
8.1
7.9
18.4
10.5
39 0
38.5
62 7
48.7
26.4
56.1
62.7
47.0
44.9
10.8
48.7
41.8
9.2
4.5
22.9
14.7
38.8
12.1
43.7
5.5
11.7
3.8
100.0
100.0
100.0
Per Cent of Population
Total Poor
Share
Urban
Per Cent
Poor
41.5
68.1
25.9
32.6
53.3
48.3
43.4
36.5
Per Cent of
Total Poor
43.0
25.9
27.3
47
17.0
5.0
100.0
Table 3: Composition of Rural Employment (NSS Usual Status Data)
1977-78 (July-June)
1983 (Jan-Dec)
1987-88 (July-June)
1989-90 (July-June)
1990-91 (July-June)
1991 (July-Dee)
1992 (Jan-Dec)
Primary
Males
Secondary
Tertiary
Primary
Secondary
80.6
77.5
74.5
71.7
71.0
74.9
75 7
8.8
10.0
12.1
12.1
12.1
1 1.2
10.4
10.5
12.2
13.4
16 2
16.9
13.9
13.9
88.1
87.5
84.7
81.4
84.9
86.3
86.2
6.7
7.4
10.0
12.4
8.1
7.9
7.8
'-.-r September 1996
Females___________
Tertiary
5.1
4.8
5.3
6.1
7.0
5.8
6.0
2461
Secondly, the share of the traditional agro
based industries fell sharply so that
agricultural raw materials played a less
significant role as industrial inputs than
earlier. The boom sectors of the 1980s chemicals, consumer durables and hightech services - had very little linkage to
agriculture. Thirdly, the combination of an
easier import situation and an enhancement
of government operations meant that
government policy instruments were more
effective in insulating the non-agricultural
sector from the effects of monsoon
fluctuations.
This last, i e. government policy, operated
on both the supply and demand sides. Given
its higher foodgrain stocks and easier access
to foreign debt, the government could better
ensure agricultural supplies to non
agriculture during periods of low agricultural
output by running down its stocks and by
resorting to higher imports of other
agricultural commodities. And. the demand
consequences on non-agriculture of lower
agricultural incomes during such periods
were also belter mitigated because, at such
limes, the government stepped up its revenue
expenditure in rural areas, by expanding
employment programmes and by generating
more self-employment opportunities either
directly through its own rural development
schemes and/or by instructing hanks to extend
more credit. Thus, although there were
features in the nature of organised sector
growth which tended to weaken agricullurenonagriculture linkages, the extent of this
weakening depended considerably on a
particular type of government involvement.
Because of this, the continuing importance
of agriculture cannot be wished away easily,
since a fall in agricultural output can still
have severe negative implications for the
economy, both in terms of output and
inflation.
Indeed, what is striking about the
experience of the 1980s is that despite the
declining dependence of non-agricultural
sectors on the performance of agriculture,
the prices of agricultural goods rose faster
than the general price level. This meant a
reversal of the earlier terms of trade
movement against agriculture, and this also
went against the international trend of a
worldwide movement of terms of trade
against agriculture. That this increase in
agricultural prices did not have unbearable
inflationary implication was partly because
of the tendency described above of weakened
inter-sectoral linkages, and partly because
of the way the food procurement and public
distribulionsysiems functioned. Government
procurement of foodgrains was more than
adequate and government stocks generally
sufficient, the procurement prices were
typically close to the market prices and
domestic food prices were also not too far
2462
from world prices so th..:
ie were fewer
speculative pressures, and the PDS, along
with the government's free market
operations, worked to some extent to keep
the prices of essential foodgrains under
control.
In fact, from the point of view of poverty,
an important trend during this period was
that while agricultural prices as a whole
increased faster than the general price level,
cereals prices increased slower, so that it was
possible for real wages to rise without
increasing product wages correspondingly.
This was an important contributory factor
behind the decline in poverty which occurred
during the period. As will bediscussed later,
these equations appear to have changed in
the post-reform period.
Moreover, there was another important
development concerning linkages in the
economy. This was the rapid growth of nonagricultural employment in rural areas. After
a long period during which agriculture’s
share in the labour force had remained
constant, there seems to have been a change
somewhere in the mid-1970s when this share
began to decline. Since the urban population
has grown faster than total population, this
is of course related to some extent with
urbanisation. But it is important to note that
during the 1980s, the pace of urbanisation
was in fact less than in any decade since
independence. For this reason it may be said
that the really important development was
the growth of the rural non-agricultural sector.
According to NSS surveys, the share of
agricultural workers among all male rural
workers declined steadily from 80.6 per
cent in 1977-78 to 77.5 per cent in 1983
to 74.5 per cent in 1987-88 to 71.7 per
cent in 1989-90. For rural females this share
dropped from 88.1 percent in 1977-78 to
87.5 per cent in 1983 to 84.7 per cent in
1987-88 to 81.4 per cent in 1989-90. The
true 'ignificance of this shift is probably
better understood in incremental terms: these
figures imply that non-agriculture absorbed
about 70 per cent of the total increase in
the rural work-force between 1977-78 and
1989-90. And this rapid growth of rural nonagricultural employment provides the main
explanation for what would otherwise be a
puzzle: how did agricultural wages rise and
i,:::.l poverty fall during a period
employment in botn agriculture and
organised sector was growing slower t To- ihx,
population? That this development, r
iWiti
than the somewhat faster growth
agricultural output during the 1980s, was
to al
major driving force behind rising wages
-JdaC
declining poverty becomes clearer when
. to b
is noted that while agricultural growth
regionally diverse (with agricultural out
van
per capita decreasing in many states),
rapid growth of rural -non-agricult
muc
employment was a phenomenon whi
occurred in almost every state in thecou
I Of I!
and almost every stale recorded rising
real wages and falling rural poverty belw
;, higr
1977-78 and 1989-90. In other words,
the*
is need to modify the conventional vic
among Indian economists that the mi c' OVel
factors determining rural poverty r I ““P
agricultural productivity and the rate < >: gn?
inflation. Although both these continues
be very important, the growth of rural net
agriculture has emerged as an addition
crucial link from the mid-1970s onward.
What then explained this growth of
non-agricultural employment? The I
that
literature on the subject has been domi
by two debates. First, whether the groat
wor
of rural non-agricultural employment is
positive development at all, or is it si
a reflection of the fact that agricult
employment has been sluggish and ce
ta a
non-agricultural activities have emerged
‘residual sectors’. Second, to the extent
Ti
the growth of rural non-agricultu
employment is not a 'residual', is it driit
in t
by developments in agriculture or ared
growth impulses external ? The idea tk
out
non-agrieullural employment is ‘rcsidual’i
now somewhat discredited because notod
and
are average wages seen to be higher in sue
employment than in agriculture, but, ma
importantly, becauscagricultural wages ha
increased as non-agricultural employma
has grown suggesting that what is in volva
is a pull factor which tightens the agriculon
labour market. Nonetheless, NSS data six* »■". ire
that the actual picture is more complicaa
and suggests that ‘distress' movement in f
non-agricullurchascominucdtobeimpottn jk *■for a significant section of rural worken.1
well as that the main dynamic sourceofnn f
nn<
Table 4: Composition or Urban Employment (NSS Usual Status Data)
______
SelfEmployment
1977-78 (July-June)
1983 (Jan-Dec)
1987-88 (July-June)
1989-90 (July-June)
1990-91 (July-June)
1991 (July-Dee)
1992 (Jan-Dee)
40.4
40.9
41.7
42.3
40.7
42.9
41.2
Males___________
Regular
Casual
46.4
43.7
43.7
41.3
44.2
39.9
39.4
13.2
15.4
14.6
16.4
15.1
17.2
19.4
Economic and Political Weekly
___________ Females_________
SelfRegular
Casual
Employment
49.5
45.8
47.1
48.6
49.0
47.0
42.5
24.9
25.8
27.5
29,2
25.9
28.0
28.8
25.6 .
28.4
25.4
222
25.1
25.0
28.7
Special Number September 1
Is
burr
r;'
1 CCKU
arc
rura
P. employment generation over this period has
been the external agency of the slate rather
than forces internal to the rural economy.’
j
There arc several planks to this argument.
i Within agriculture, all the available evidence
■ points tothcdecrcasing ability of agriculture
\ to absorb more labour, as the overall crude
elasticities ofemploymcnt to output arc seen
to be low in other major states and on an
all-India basis. However, there arc substantial
variations across states, with the
agriculturally less advanced regions showing
much higher elasticities than the developed
states like Punjab and Haryana. Since some
of the less advanced states (such as West
Bengal and Bihar) actually showed the
highest rates of output growth over the period.
there was less of a dampening effect on the
overall elasticity as well as a pointer to the
importance of regional spread of agricultural
growth for employment generation.
.Moreover, an interesting observation relates
0b the Bow of person-day employment in
agriculture, which, after 1977-78, was seen
to be increasing more than stock measures
of usual or weekly status workers. In a very
rough and approximate sense, this suggests
that the supply of agricultural labour (as
measured by the stock of agricultural
workers) was actually increasing slower than
the demand for agricultural labour measured
inperson-days. Simultaneously there appear
to have been contractual changes underway
in agriculture, with a greater emphasis on
casual contracts.
The natural question consequent upon such
a finding is what caused the slow growth
in the stock measures of workers in
agriculture. Here it was found that pull factors
out of agriculture were significant. The
relationship between agricultural prosperity
and the growth of non-agricultural
rtunities was found to be weak and nonr. being significant only in states such
as Punjab and Haryana where not only have
agricultural incomes crossed a threshold but
where further increases in agricultural output
are accompanied by labour displacement
rather than absorption. Outside this limited
region, thepull is provided mainly by external
stimuli. In certain regions, forexample along
the Bombay-New Delhi and the BombayBangalore highways, there is clear evidence
that industrial development, and the growth
of services linked to this,have made deep
inroads into rural society creating
opportunities not only in the tertiary sector
but also in small-scale industry. In addition.
in the hinterland of industrially or
commercially developed regions, there is
growing incidence of workers who live in
rural areas but commute to urban areas - a
tendency which has been enhanced by the
fact that the organised sector has tended to
prefer casual workers to regular employees,
and because rising urban rents and falling
B
Economic and Political Weekly
transport costs have influenced workers’
development' schemes with an explicit
choice of residence. However, given the
redistributive concern/' This included not
limited geographical spread of such direct
only the various rural employment and
links to modern industry and commerce, in
IRDP programmes but also a plethora of
most areas the pivotal role in the expansion
special schemes for a variety of identifiable
of rural non-agricultural employment appears
’target’ groups. Motivated by the realisation
to have been played by the expansion of that income growth by itself would not
government expenditure.
'trickle down’ in adequate amounts, these
As noted earlier, the 1980s were a period
programmes were however less than entirely
when, along with a rapid increase in all sorts
successful. They spawned a large
of subsidies and transfers to households
bureaucracy and they became a focal point
from government, there was a very large
for the politics of ‘distributive coalitions’.
increase in revenue (as opposed to capital)
Yet, though the intended beneficiaries often
expenditure on agriculture by state and
got short-changed because of such leakages,
central governments, and this wits also a
these programmes represented a fairly
period when the expenditure on Rural
massive net transfer to rural areas. The second
Development expanded manifold. More
avenue by which resources flowed from
generally, throughout the period political
government to rural areas was through the
developments tended to give rural interests
greater accessibility of the rural elites to the
greater power and they were able to command
government's normal gravy train. In part
an improvement in the historically low share
this was a result of greater mobility due to
of government expenditure benefiting rural
better infrastructure, but to a large extent it
areas. Although this improvement in share
was also the outcome of the fact that with
should not be exaggerated, an indication
governments
changing
frequently
may be had from the fact that nearly 60 per (particularly at the state level) more new
cent of all new government jobs created
favours, not just jobs, but also various types
accrued to rural areas during the decade.
of agencies and contracts, had to be
Moreover, NSS data suggest that, despite a
distributed more often and the rural areas
low average contribution of only around 5
got a greater share in such electorally
per cent of total rural employment, the
motivated largesse than they get at other
government’s contribution was around a fifth
times. The resulting (low of resources and
when it comes to either total rural nonthe resulting generation of rural demand led
agricultural employment in 1987-88 or the
to growing opportunities for diversification
increments in total rural employment between
of the self-employed from agriculture to
1977-78 and 1987-88. Moreover, in 1987-88,
non-agriculture.
about 60 per cent of the regular nonTo a very large extent, the direct access
to government permanent employment and
agricultural employees in rural areas were
also to many of the other resources was
employed by the government which created
confined to the better-off and more powerful
almost 80 percent of the increments in such
groups in rural society, to whom such
regular jobs during the decade covered.
incomes were more lucrative than agriculture
Thus, given the magnitude of what is now
Associated with this was a large and
commonly accepted to have been a profligate
significant increase in the proportion of the
growth of government expenditure, the total
15 to 29 age cohort which continued in
quantum of increased flowof public resources
education rather than join the work force.
into rural areas must have been significant.
In part this must have been a result of the
Besides the large growth in agricultural
expansion of educational facilities as part
subsidies already mentioned, this flow ol
of the general expansion of government in
resources took two predominant forms. There
rural and semi-urban areas, but to a subwas, first, a fairly large expansion of ‘rural
Table 5 Changes in EMPLOYMENT 1989-90 ro 1992 (NSS Usual Status Unadjusted)
(Million persons)
Self-cmp agriculture
Regular-emp agriculture
Casual-emp agriculture
Self-cmp secondary
Regular emp secondary
Casual-emp secondary
Self-cmp tertiary
Regular-emp tertiary
Casual-emp tertiary
Unemployed
Total workforce
Total population
Special Number September 1996
Rural
1989-90
1992
Urban
1992
1989-90
122.2
5.6
70.5
17 7
3.3
11.2
18.9
106
4.1
2.8
266.9
602.7
4.6
0.3
2.8
6.8
7.3
4.6
14.5
15 4
3.1
2.3.
61.7
176.3
132.3
38
74.9
11.3
4.1
9.9
16 9
9 1
3.6
2.9
268.8
608.9
5.3
0.3
3.3
7.0
8.8
6.9
15 6
16.0
4.2
33
70.7
200.9
Total
1989-90
1992
126.8
5.9
73.3
24.5
10.6
15.8
33.4
26 0
7.2
5.1
328.6
779.0
137.6
4.1
78.2
18.3
12.9
16.8
32.5
25.1
7.8
6.2
339.5
809.8
2463
Iff
; off
i c
i ma
c
lhai
c
met
g
1
G
ago
G
Thc
acc
G
alth
G
poo
G
GT
(ent
G"
siln
c;
agri
setc
c7id
SI COM'S ROOTS GO V E R V D EE P INTO TH E SO IL
OF MAHARASHTRA -and reach out to the remotest
DEPENDS
ON
corners of the State. Stimulating industrial growth in
the less des eloped regions of the State for nearly
three decades. Evolving and growing continuously
throughout that period. Today, it offers Project
Merchant Banking, Underwriting, Project Appraisal,
p&Oi
days
Scheme of Incentives and Escort Services for all
.Vo
Of ll
clearances and infrastructural amenities. S1COM also
HOW
DEEP
YOUR
Finance, Corporate Loans, Short Term Loans, Direct
Equity Participation, Loan Syndication, Leasing,
diyc
Bill Discounting, Mergers & Amalgamations, Package
acts as a nodal agency in Maharashtra both for NRls and
Foreign Direct Investment. There arc no upper limits on
the size of the projects assisted or on the quantum of
largi
loan. Today, SI COM is also the first to change from a
ROOTS
ARE
must
government enterprise Io a company with public
participation
- to
the
pres
full extent of 51%.
and i
o
G
Ihcd
G
'
grow
SICOM
I. I M I T F. I)
'NIRMAI.'. NARIMAN I’OIN I
. MUMBAI - 400 021.
: 202 3013. 202 1X55 FAX : 202 300X Tl-I.l X • Oil
83066
extet
Mon
only
c
64
Economic and Political Weekly
Special Number September 1996
" stantial extent this must represent also a
' motivational change (to acquire necessary
■ qualifications for a regular non-agricullural
job) among the youth in the relatively welloff sections of rural society. There was thus
a movement out of agricultural work at the
margin by workers and potential workers
from such better-off rural groups, which meant
that sections of the relatively rich vacated
agriculture either to obtain regular employ
ment, mainly in the service sector, or to take
up non-agricultural self-employment.
This increased the ability of members of
the less well-off rural households to find
agricultural work, and also created a demand
for certain types of rural servicesand industry.
The relative tightening in the agricultural
labour market which resulted, helps to
account for the increase in real wages
observed from the late 1970s. However,
although such increases in employment and
wages did improve the condition of the
forest rural workers, their employment
^versification into non-agriculture
continued to have many characteristics of
a ‘distress’ process, given the overall
tendency of labour use in agriculture.
Dictated by the need to ensure economic
survival, they increasingly entered into casual
work not only in agriculture but also in non
agriculture. The main sectors providing this
type of non-agricultural employment were
secondary sectors like construction, mining,
and small-scale manufacturing, and there is
evidence that over time the incidence of
poverty among those employed in some of
these sectors became larger than in
agriculture. Moreover, the agency of the
state was important in terms of the
diversification of opportunities for the rural
poor.Thus,22.3 pcrccntofall casual labour
days spent on non-agricultural activity in
-88 were on public works programmes
c government, this percentage having
increased from 17.7 in 1977-78 and 14.9
in 1983. And, although there is little evidence
of any increase in non-agricultural self
employment among the bottom 40 per cent
of the rural population (such increase was
largely among relatively richer households),
income generation scheme such as the IRDP,
must also have had some effect.
This is of course an extremely schematic
presentation of what is a much more
multifarious and regionally divcrsesccnario,
and there were variations in the pattern across
states and over time. However, the fact that
the developments described above occurred
in every state, irrespective of the rate of
growth in agriculture or organised industry,
does imply the increased importance of
external stimuli to rural employment and,
in particular, the crucial role of the state.
More importantly, these trends mean that the
rural labour demand is no longerdetermined
only by what is happening within the
■
Economic and Political Weekly
agricultural sector, but is determined crucially
also by macro-economic processes and
policies which do not at first appear to have
any direct link with rural well-being.
Moreover, because much of the govern
ment spending involved is project funded,
because most of the private enterprises
involved are small and lack staying pow’er.
and because most of the wage employment
thus created are casual, the vulnerability of
the rural non-agricultural sector to overall
public expenditure cuts and to restrictive
monetary policy is almost certainly greater
than for its urban counterpart. This has very
important ramifications in the current macroeconomic context, in which the reform
measures have particular implications for
patterns of government expenditure as well
as on internal and external trade.
It is important to note in this context that
the pattern of structural adjustment and
government economic strategy since 1991
has been one which has involved a continued
stagnation in employment generation in the
organised sector, both public and private.
Moreover, this strategy involved:
(1) actual declines in Central government
revenue expenditure on rural develop
ment (including agricultural programmes
and rural employment and anti-poverty
schemes), as well as on the fertiliser
subsidy, in the budgets of 1991 -92 and
1992-93. Some of these cuts were how
ever reversed subsequently in 1993-94.
(2) declines in public infrastructural and
energy investments which affect the rural
areas.
(3) reduced transfers to state governments
which have been facing a major financial
crunch and have therefore been forced
to cut back their own spending, parti
cularly on social expenditure such as on
education and on health and sanitation.
(4) reduced spread and rising prices of the
public distribution system for food.
(5) financial liberalisation measures which
have effectively reduced the availability
of credit, especially to small borrowers
particularly agriculturists.
Thus, in the early 1990s, there was a
reversal of several of the public policies
which contributed to more employment and
less poverty in the rural areas in the earlier
decade. It should, therefore, not be entirely
surprising that rural non-agricultural
employment appears to have declined fairly
sharply as soon as the stabilisation and
structural adjustment policies were put into
place in 1991. According to NSS survey
data, the rion-agricultural proportion among
rural male workers was 28.3 per cent in
1989-90 and 29 per cent in 1990-91, before
the reforms, and this fell to 25.1 per cent
in July-December 1991 and 24.3 per cent
in 1992. For rural female workers, the
corresponding figures were 18.6, 15.1, 13.7
and 13.8 per cent. This represents a decline
of somewhere between 9 and 11 million in
the number of workers in rural non
agriculture, or a drop of 13-15 percent in
the first 18 months of the initiation of the
reform process.’
This fall occurred almost all over India,
with only Karnataka and Madhya Pradesh
being significant exceptions. In terms of
sectors, this decline in employment was
divided roughly equally between manu
facturing, construction and community and
other services, along with a smaller drop in
transport; while mining, electricity, trade
and financial services were immune among
the self-employed and casual workers that
the decline was greatest, with regular
employment being largely maintained, except
for some drop among regular malccmployces
in the tertiary sector. Thus, the pattern of
the decline in rural non-agricultural work
suggests that it occurred not because of any
large-scale retrenchment of regular
employees by the government or the
organised private sector, but because of a
cut back in activity in the unorganised sector
and. possibly, some retrenchment of casual
workers by the organised sector. In this
context, it is significant that, according io
the NSS, this drop in rural non-agricultural
employment was not accompanied by a
corresponding drop of such employment
Table 6: Tendulkar-Jain Estimates of Poverty
H
1970-71
1972-73
1973-74
1977-78
1983
1986-87’
1987-88
1988-89’
1989-90*
1990-91*
1991 (July-Dec)’
1992*
45.89
47.00
49 20
42.98
38 33
35.39
36.52
36.98
32 41
32 43
32.02
33.87
_________ Urban___________ ______________Rural___________
PG
SPG
H
PG
SPG
17.57
73?
5.32
57.33
13.39
17.93
7.54
5.32
57 21
13.57
16.75
6.72
5.31
56.17
13.88
16.59
6.88
4.81
54.47
12.16
5.45
13.86
3.66
49.02
9.95
4.60
12.21
3.54
45.21
9.48
4.04
1 1.26
3.38
44.88
9.34
3.54
10.20
9.61
2.95
8.80
8.03
3.03
8.81
8.03
10.02
3.39
7.90
4.58
12.59
8.43
Notes: ’ Denotes small sample; Poverty njeasurcs arc same as in Table 1
Special Number September 1996
2465
among urban workers. Even among rural
workers this did not lead to any large increase
in open unemployment or to any large fall
in the work participation rate. Rather, the
self-employed and the casual workers dis
placed from non-agriculture appear to have
reverted back to agriculture, leading to
disguised (rather than open) unemployment.
However, as a result, real agricultural valueadded per agricultural worker dropped
significantly, by over 8 per cent, even if
comparison is restricted to the years 1989-90
and 1992-93 when monsoon conditions were
very similar. Unfortunately, later data
(particularly from the 1993-94 NSS large
sample) is not yet available to verify whether
this reversal of trend has continued, but
clearly the early post-reform impact was
adverse.
In the urban areas, regular employment
has continued to stagnate, especially in the
organised sector. During 1991 -95. the growth
rate of employment in the organised sector
halved from its already low growth rale
during the 1980s to only 0 8 per cent per
annum, mainly because of a massive slowing
down of employment growth in the public
sector. The increases in employment that are
discernible arc essentially in casual
employment, and this is evident for the
secondary and tertiary sectors according to
both usual and weekly status definitions.
However, these increases in employment are
still below the estimated increases in urban
population over this period. The continued
process of casualisation of work in urban
areas has to be seen in relation to two other
recent tendencies which arc highlighted by
several micro-level studies. First, there is the
growth of subcontracting in manufacturing,
which increasingly integrates formal and
nformal sector productive activities, and
allows for a substantial part of the production
to be undertaken by very small informal and
unorganised units at the bottom of the
production chain. These i mply that a growi ng
part of manufacturing production is
undertaken by units in which there is no
formal protection of any sort to labour.
Second, and related to the first tendency,
there is evidence ol some Teminisalion' of
employment, that is the growing share of
female employment to the total, particularly
in export-oriented activities, and with wages
and working conditions that are typically
inferior to thoseof male counterpart workers.
It is evident that these processes will have
direct and indirect links to the spread of
poverty, throughout India but especially in
rural areas. These links, and the more general
relation between economic growth and
poverty, arc considered below.
HI
rost-reionn Trend in Poverty
In earlier sections it has been observed
that there was a declining trend in poverty
after the mid-1970s but that this trend was
reversed in the 1990s. However, while the
earlier declining trend is officially accepted.
the reversal during the 1990s is not. As
discussed earlier, the main difference between
the official view and those of independent
observers arises because till now (he official
estimate is based not on the NSS data
directly but on adjusted figures obtained
by blowing up the NSS consumption
estimates forevery dcci Ic group by a common
adjustment factor equal to the ratio between
the CSO estimate of private consumption
and the corresponding NSS estimate. Because
this adjustment factor has increased sharply
in recent years, the official estimate has
diverged increasingly from any estimate
based directly on NSS data. The Expert
Group which recently went into this matter
concluded quite categorically that the practice
of ‘adjusting’ NSS data was arbitrary and
was likely to give wrong results because as
against the implicit assumption in the official
method that any underestimation of
consumption is distributed uniformly over
the entire population, it is better to assume
., :iimd 're.'”1 is
for those
hi fuel, several alternative series which
use the unadjusted NSS figures are available.
In addition to the World Bank scries given
earlier, a scries calculated by Tendulkar and
Jain is available for the period 1970-92.*
This uses the same reference poverty lines
at 1973-74 prices as the official and World
Bank scries, but using a different deflator
they obtain an even larger increase in poverty
between 1990-91 and 1992 (Table 6). In
addition, it has been possible to obtain
measures of rural poverty for All India and
the major states, based on the Expert Group
Method using NSS data covering the years
from 1972-73 to 1993-94'* (Table 7). Unlike
the other senes this is not obtained from the
national-level NSS data but is obtained by
applying state-specific poverty lines to stale
level NSS data. Il must be noted that the
figures given here for 1993-94 are
preliminary, being based only on partial data
(not yet officially released) from the 50th
round of the NSS. Moreover, the data for
1986-87 and for 1989-90 to 1992 arc based
on the so-called ‘thin’ surveys by the NSS
involving a much lower sample size than the
other survey points. With only three survey
points available for the post-reform period,
and given the above qualifications for
whatever data is available, any conclusion
about post-reform trends must necessarily
be rather tentative.
Nonetheless, using the mutually compar
able thin samples alone, it is evident that
poverty increased sharply during the first 18
months of the reform period (i c, the second
half of 1991 and 1992), particularly in the
rural areas. The partial data relating to
1993-94 suggests, however, that this upward
trend was reversed thereafter. Taken together,
these data suggest that there was a very large
increase in rural poverty in the first 18
months of reform but that this trend has been
moderated thereafter. Rural poverty in
1993-94 continued to be higher than in
Taiii e 7: Estimates or Rural Headcount Poverty iiv tiu: Expert Group Method
Andhra Pradesh
Assam
Bihar
Gujarat
Haryana
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
' 'unjab
.<n.jasthan
Pamil Nadu
Alar Pradesh
v/est Bengal
Ail India
2466
1973-74
1977-78
1983
1986-87
1987-88
1989-90
1990-91
1992
1993-94
48.4
52.7
63.0
46 4
34.2
55.1
59.2
62.7
57.7
67.3
28.2
44 8
57.4
56.5
73,2
56.4
3S.I
59.8
63.3
41.8
27.7
48.2
51.5
62.5
64.0
72.4
16.4
35.9
57 7
47.6
68.3
53.1
26.5
42.6
64.4
29 8
20.6
36.3
39.0
48.9
45.2
67.5
13.2
33.5
54.0
46.5
63.1
45.6
14 6
39.7
50.1
30.3
19.5
36.6
33.5
47.8
44.6
55.2
13.0
29.2
41.2
36.6
47 3
38.3
20.9
39 4
52.6
28.7
16.2
32.8
29.1
41.9
40.8
57.6
12.6
33.2
45.8
41.1
48.3
39.1
19.5
35.2
52.4
14.8
1.3.3
45.4
34.4
39.5
34.8
52.9
3.2
26.1
38.4
.30.5
37.2
34.4
22.1
33.7
46.3
21.6
19.5
34.9
30.3
42.4
35.9
36.5
9.3
25.9
37.5
34.8
49.5
35.0
27.4
51.7
61.1
33.7
17.7
45.5
26.0
47.9
53.6
49.0
10.2
31.7
44.3
47.9
44.0
44.0
16.0
45.0
58.0
22,2
28.7
28.2
25.9
40.8
38.6
49.9
12.5
27.5
32.6
42.6
40.3
37,5
Economic and Political Weekly
Special Number September 1996
■ 1989-90 and 1990-91 but was less than in
11987-88. Urban poverty, on the other hand,
S' tppears not to have increased much during
, tic first 18 months of the reform period and.
indeed, appears to have declined significantly
io 1993-94. Nonetheless, there were about
JO million more people in poverty in the
fatter year than before the reforms began.ln
Thus, the post-reform trends in poverty
do not suggest either an unambiguous
improvement or an unambiguous worsening.
They do suggest, however, that the initial
impact of the stabilisation/structural
adjustment package was adverse, that this
impinged particularly on the rural sector,
with less impacton the urban sector, and that
there was somegeneral reversal of the adverse
trend subsequently. Nonetheless, it is
important to note that the state-wise figures
show that, as far as rural poverty is concerned,
in most states the poverty ratios in 1993-94
were significantly larger than in the
fcmediate pre-reform period. This is
particularly true of the two largest Indian
states, Uttar Pradesh and Bihar, and also of
the hitherto successful ‘green-revolution’
states of Haryana and Punjab. The exceptions
ire the Southern states of Andhra Pradesh.
Karnataka, Kerala and Tamil Nadu in all of
which the poverty ratio in 1993-94 was
lower than in the immediate pre-reform
period. However, it must be noted that in
these states, and in Maharashtra and Gujarat,
the year 1993-94 was exceptional in that the
food prices actually fell in absolute terms
is against rapid increases in both preceding
and following years. For this reason, the
calculated poverty ratios for 1993-94 arc
likely to be somewhat lower than the
underlying trend.
More importantly, these trends in poverty
need to be viewed in the general context,
divpissed earlier, that me stabilisation and
S^^iral adjustment policies carried out so
farin India involved a fairly sharp contraction
m fiscal and monetary policy in 1991-92 and
1992-93, followed by a return to high fiscal
deficits from 1993-94 onwards. The revival
ofgrowth after an initial period of stagnation
also followed a pattern broadly coincident
with that of the government’s fiscal stance,
so that both the initial worsening of the
poverty situation and the subsequent
improvement seem to be broadly in line with
the overall growth performance of the
economy. Yet, there arc a few surprises, the
most important of which is that although the
reform measures did not directly involve
much changes in agriculture, it was rural
poverty which appears to have been more
sensitively affected by the post-reform
developments.
This has led some analysts, for example
Tendulkar and Jain (along with many other
economists who arc generally in support of
the reform process), to argue that, although
Economic and Political Weekly
poverty did increase during the first 18
months of reform, the reforms were not the
principal cause for this increase.This
argument rests on the observation that "the
rural sector in general and agriculture in
particular were not the major focus of
structural adjustment and were only indirectly
affected by fiscal compression" and on the
assumption that the reforms “would have
adversely affected primarily the urban
organised sector with second-order impact
on the urban informal sector and weaker
effect on the rural sector”. Because, in fact,
it was rural poverty which increased sharply,
they attribute this increase not to reformsper
se but to weather and to the higher post
reforms inflation for which, moreover, they
hold the reforms only indirectly responsible.
However, although the reform process has
indeed neglected agriculture, there arc two
major difficulties with the argument that,
therefore, it could not have increased rural
poverty.
First, although there was a 2 percent drop
in agricultural production in 1991-92
compared to 1990-91, and although the
inflation rate did increase sharply
(particularly for foodgrains), these adverse
factors were simply not large enough to
explain the very large increase in the
incidence of rural poverty. As discussed in
the next section any econometric model fitted
to the data prior to reforms linking the
incidence of rural poverty only to some
measure of agricultural production or
productivity and to the inflation effect, breaks
down as soon as the post-reform data is
included. And. indeed, all such models are
outperformed by models incorporating
relative food price, rural non-agricultural
employment and some measure of
commercialisation, in addition to agricultural
production. The latter not only fit the prereform data better, but when fitted to prereform data accurately track the post-reform
increase in poverty, unlike models not
including government expenditure and/or
rural non-agricultural employment which
predict much lower poverty increase than
that which look place actually.
The important point to note is that this
phenomenon of rural non-agricultural
employment, which Tendulkar and Jain
ignore, was almost certainly the major factor
which drove rural real wages up, and caused
poverty to decline, during the 1980s. And,
in turn, this was based largely on increasing
government expenditure and on the
availability of cheap credit to the small-scale
sectors. As has been poi nted out in a previous
section, the onset of stabilisation and
structural adjustment appears to have led to
a rather quick and large decline in such rural
non-agricultural employment, pushing
millions of self-employed and casual rural
non-agricultural workers back to agriculture,
thus reducing per-worker incomes in
agriculture. It appears, therefore, that just as
the expansion of the 1980s involved a rapid
increase in rural non-agricultural employ
ment without any concommitant increase in
organised sector employment, the stagnation
in the first 18 months of reform saw a cutback
in rural non-agricultural employment with
not much effect on cither organised sector
or urban employment. This suggests that,
contrary to popularopinion. thccmploymcnt
multipliers associated with thegovernment's
fiscal stance are larger for rural non
agriculture than for the urban or organised
sectors. If this is accepted, the trend in the
magnitude of employment decline and
poverty increase arc not surprising,
particularly because, as noted earlier, the
contractionary tendencies generally
impinged much more adversely on smaller
enterprises than on the corporate sector"
In addition, it is obvious that the effect
on poverty of the rather small decline in
agricultural output in 1991-92 could have
been mitigated if rural employment policies
had been used effectively, as they were
during 1987-88 a year of much larger decline
in agricultural and foodgrains output' than
Table 8: Basic Regression Results: All-India Data
(Dependent Variable Is Log of Headcount Poverty Ratio)
Constant
Rural
1 1.34
9.56
8.83
1.37
4.68
Urban
11.16
8.57
8.24
4.61
5.23
Commerce
Per Capita
Agriculture
Income
Per Capita
Non-Agri
Income
-0.72 (1.2)
-0.68 (1.1)
-0.45 (1.9)
-0.49 (2.2)
-0.45 (2.4)
-0.36 (2.1)
-0.78 (1.2)
-2,38 (4.0)
-1.62(3.3)
-2.15 (4.3)
III (1.5)
3.20 (4.8)
1.73 (3.1)
2 65 (4.6)
-0.75 (2.0)
-0.69 (2.0)
-0 58 (2.4)
-0.59 (2.7)
-0.58 (2.8)
-0 32 (3.0)
1.35 (1.8)
-0.07 (0.1)
0.19 (0.4)
-1.63(2.3)
0.31 (0.4)
-0.26 (0.5)
Public
Dev
Expenditure
R Bar
Squared
Relative
Price of
Cereals
1.45 (9.8)
0.80 (3.0)
0.73
0.77
0.96
0.96
0.97
0.70(4.8)
0.60 (2.7)
0.86
0.88
0.94
0.95
0.95
-0.76 (9.4)
-0.39 (2.7)
-0.34 (4.0)
-0.08 (1.5)
Note: T values in parenthesis.
Special Number September 1996
[ ‘
2467
1991 -92. Indeed, a state-wise analysis shows
that between 1989-90 and 1992 rural poverty
increased in every state except Kerala, i c,
it increased even in those states (Haryana,
Karnataka, Madhya Pradesh, Orissa, Punjab.
Tamil Nadu, Uttar Pradesh and West Bengal)
where the 1991-92 foodgrains output was
higher than in 1989-90, thus indicating a
rather weak link between the fall in output
and the increase in poverty. Moreover, it is
interesting to note that the largest post-reform
increases in poverty were registered in the
two states, Gujarat and Maharashtra, which
were most enthusiastic about the reforms
process. Here, there was a fall in foodgrains
output in 1991-92 as in 1987-88, but. unlike
in 1987-88, the state governments neglected
rural employment and drought relief schemes
in 1991-92.
Secondly, although inflation is clearly
important, Tendulkar and Jain arc not entirely
correct when they claim that the large rise
in inflation (to over 25 per cent for the
Consumer Price Index for Agricultural
Labourers) during the first year of reforms
was only indirectly related to the reform
process. Their argument is that inflation
occurred because crop output fell in a year
when foodgrain stocks were low and the
balance of payments position did not permit
large imports; and because the government
succumbed to rich farmer demands to
increase procurement prices following the
(necessary steps of) devaluation and cut in
fertiliser subsidy.
In fact, because of a record harvest in
1990-91, public foodgrain stocks were high,
over21 million tonnes, when the government
embarked on its reforms at the end of June
1991. And inflation accelerated principally
because of the expectations set-off by
devaluation and the impression given that
all discrimination against exports of
agricultural goods would be removed. l!This
led to an immediate speculative increase in
private slocks, forcing the government to
run down its own slocks faster, and also
contemplate food imports. However, the
decision on such imports was postponed til!
after the next harvest. And when this turned
out to be somewhat less than expected, the
government was faced not only with low
stocks but also with low procurement because
farmers withheld sales in the expectation
that it would be politically and economically
difficult forthe government to justify imports
at prices well above domestic free market
prices which in turn were higher than the
procurement prices. In the event, the
government was forced to increase both its
procurement prices massively (linking this
with withdrawal of fertiliser subsidies) and
also import wheat at prices higher than the
increased procurement prices.
The entire problem was thus clearly driven
by the fact that devaluation was linked
2468
explicitly with the idea of removing trade
discrimination against agricultural goods an.:.
therefore, with a central tenet of the
liberalisation argument. If this had been
followed through fully, food prices would
have risen much more than they actually did
and farmers would have received more than
what they allegedly obtained as a result of
their political clout. In fact, prices began
stabilising only when the government made
it clear that not only would some export
restrictions continue on foodgrains but also
that it would continue with its earlier policy
of importing foodgrain to stabilise domestic
prices, even if this meant making a
commercial loss. For this reason, it can be
surmised that the inflationary problem could
have been avoided to a large extent had
devaluation been accompanied at the outset
by an explicit policy of increasing the wedge
between world and domestic prices through
higher export duties and a definite announce
ment that canalised imports would continue.
That it was not, and led to an inflationary
spurt which was contained only when the
government backtracked, is one among a
number of instances of how fidelity to the
liberalisation world-view was extremely
costly in the short-run without being
sustainable in the longer run.
Thus, the massive increase in rural poverty,
by over 60 million people, in the first 18
months of reform was to a very large extent
a direct result of the stabilisalion-cumstructural adjustment policies. The latcrdata,
for 1993-94. which shows a moderation in
poverty docs not neccsarily contradict this
conclusion because, after all, public
expenditure cuts were to a large extent
restored (and so rural non-agricultural
employment might have risen somewhat)
and stability was returned to foodgrains
markets by removing the expectation that
Indian agricultural prices were to soon reach
international levels. Thus, although nothing
firm can be said about employment trends
till the full data from the 1993-94 survey
is released, the decline in poverty appears
only to confirm that changes in public
expenditure levels and announcements
regarding liberalisation of international trade
,
riculturul products have a large
i.;act on .uial puvc.ly.
IV
Determinants of Poverty
This brief review of trends in poverty E
bring us to the central issue of this paper •
that is how liberalisation and structural ;
adjustment may be expected to affect the ;
incidence of poverty. Given the limited data •
and the somewhat conflicting empirical >
trends reviewed above, the remaining •
discussion will focus on past discussions of
the determinants of poverty in India, and
how the conventional logic needs to be
modified in the light of subsequent develop
ments. Since the Indian literature is mainly
on rural poverty, this discussion will also
focus largely on (he rural sector.
Past literature has tended to focus on two
types of variables: some measure of :
agricultural output or productivity and some .
price variable. And past writings have
debated both the relative significance of these variables, and, more importantly, their ■
proper specification.1-' Thus, although the
level and growth of agricultural production
per capita of rural population is obviously 1
an important variable determining levels of !
welfare in a predominantly agricultural rural ■
community, it is also obvious that such a
relationship would be affected by whether .
agricultural growth is accompanied by ■
increasing inequality and whether there are •
other sources of rural incomes. The link
between poverty and prices is even more '
complex. For example, when Dharm Narain ■
presented regression results showing that ,
poverty was related positively with higher '■
food prices, his specification was challenged
because bis use of the nominal food price ■
as an explanatory variable ran counter to the
prior, common to most economists, that what
really matters are relative prices, and. that ■
if absolute prices need to be incorporate^*
this should be done by considering the rat"
of inflation rather than the price level.
i
As far as the importance of agricultural
output as a determinant of rural poverty is
concerned, it is obvious that, unless the
Table 9: Au. India Rural Poverty Equation
Variables
Ravallion-Dutt Model
I960 io 1989
1960 io 1992
Our Model
I960 to 1992
I960 to 1989
Constant
Ag Productivity
Real Wages
Relative price cereals
Non-ag employment
Commercialisation
Time Trend
Lagged Dependent
R Bar Squared
DW
4.6(5.01)
-0.3(2.69)
-0.5(3.26)
3.3(2.79)
-0.1(0.56)
-0.7(3.53)
-0.6(0.67)
-0.4(4.21)
-0.3(3.29)
0.9(7.69)
-0.5(3.97)
1.0(6.45)
-0.1(0.10)
-0.4(3.36)
-0.4(2.79)
0.9(6.25)
-0.4(3.65)
0.9(4.92)
0.0(3.85)
0.5(4.74)
0.94
1.63
0.0(1.86)
0.4(2.75)
0.89
1.28
0.98
2.27
0.97
2.37
Economic and Political Weekly
Special Number September 1996
manner in which higher agricultural output
is brought about is sharply incqualising, any
increase in agricultural output per capita
would lend to benefit most rural people. It
is precisely the fear of the possible
incqualising impact ofthc 'green revolution'
which had triggered off early work on this
area. But. by now, it may safely be conceded
that although relative inequalities may have
increased, the ‘green revolution' certainly
reduced the incidence of absolute rural
poverty tn the regions where it was successful.
But, although proponents of the trickle down
hypothesis may have proved more correct
than the detractors in this matter, the really
sinking feature of the post green revolution
period is that, nonetheless, there is a rather
weak link across states between the rate of
agricultural growth pcrcapitaand reductions
in rural poverty.
The simple fact is that, with the green
revolution limited in geographical coverage,
most states in India did not record any
significant increase in agricultural valueadded per head of rural population during
the 1970s and 1980s, although almost all of
them recorded significant declines in poverty.
With poverty reduced even where
agricultural output did not increase, there
has thus been a reversal of the earlier
apprehension that agricultural growth could
occur without reducing poverty. But this
very disassociation between poverty
reduction and agricultural growth is a feature
which merits more attention than has been
given so far. For example, it is significant
that, while early work on the subject
invariably chose some measure of
agricultural output per capita, some recent
research finds agricultural output pcrhcclare
to be the measure of agricultural performance
better correlated with poverty decline.'4 This
masure has the advantage, for those
^sistent in viewing agricultural growth as
the main engine for the reduction of rural
poverty, that on this basis almost every region
in India recorded some agricultural growth
during the period when poverty declined.
and, in most cases, this growth was also
larger than during the earlier period when
rural poverty did not decline.
Yet, given the stagnation of agricultural
output per head in most parts of rural India,
this shift in the measure used surely serves
to obfuscate matters rather than to clarify
them, especially because in actual fact the
underlying shift in Indian agriculture from
expansion of cropped area to yield increases
has been accompanied by a sharp decline
in the output elasticity of agriculture's
demand for labour. As a result, poverty has
declined in most regions of rural India in
icontext not only of stagnant agricultural
*■; output per head of rural population but also
i one where agricultural employment has
r, grown much slower than the growth of the
and
n'
ns
Economic and Political Weekly
‘
r
Sp-ci.-l
rural labour force. This latter feature would
normally be expected to depress agricultural
wages and thus affect adversely the poorest
among India’s rural residents. But, in fact,
real agricultural wages increased sharply in
most parts of India between the mid-1970s
and late 1980s, and this was, in fact, one
of the main reasons why poverty declined.
Some researchers do note the rise in wages
but chose to ‘explain’ this rise also by
reference to the increase in output per acre.
This, however, stretches credulity because
by almost every measure the rise in real
wages was at least twice as much as the
increase in real output per worker, and cannot.
therefore, be ascribed mainly to technical
progress in agriculture. As has been argued
already, the real explanation for the rise in
agricultural wages lies in the rapid growth
of rural non-agrieultural employment and
the dynamics behind this. The Indian
literature on this has in the past toyed with
two ideas: that rural non-agriculturc is itself
driven by agricultural growth through the
operation of Engel’s law; and the opposite
idea that the process of commercialisation
of agriculture leads to displacement of
agricultural labour which finds distress
employment in certain 'residual' sectors of
non-agriculturc. However, both these ideas
are now somewhat discredited. The link
between agricultural grow’th and that of rural
non-agriculturc has been found to be non
linear, and, with the exception ofthc limited
prime 'green revolution’ area there is little
evidence that agricultural growth has
provided the impetus to rural non-agriculturc.
Also, despite considerable evidence that
commercialisation is incqualising and leads
to casualisation ol the rural labour force, the
‘residual’ sector hypothesis stands
discredited because this docs not square
with rising real wages. The evidence
overwhelmingly supports the thesis that the
main impetus for the growth of rural nonagriculturc has come from outside the rural
areas, in considerable part from the expansion
of government expenditure. This observation
suggests that rural Incomes are no longer
derived only from agricultural production.
and that the process by which rural areas
have got integrated into the wider economy
are important. One aspect of this has been
commercialisation with its incqualising
effects hut the other is that external stimuli!
have provided employment opportunities
and incomes which arc related not so much
with agriculture but with developments in
the wider macro-economy.
However, many economists continue to
view agricultural growth as the main solution
to poverty. One reason for this is precisely
because such an association fits neatly with
the view that in countries like India, which
protected industry in the past and therefore
require ‘structural adjustment’ today, got it
wrong not only on efficiency but also on
equity. Indeed, during the 1970s, when the
World Bank pushed its growth with
redistribution slogan, the argument was that
the pro-industry policies followed by
countries such as India hampered agricultural
growth and thus meant higher poverty than
was necessary. Even today, some World
Bank analysis tries to show that rural poverty
is unaffected (oreven worsened) by industrial
growth while agricultural growth reduces
not only rural but even urban poverty.15 This
view continues to be a central component
of the.structural adjustment policies, wherein
il is argued that greater liberalisation of trade
and industry would shift resources towards
agriculture and this would not only be more
in line with India's comparative advantage
but would also reduce poverty much more
than earlier policies. In other words, although
no one will dispute that higher agricultural
output is very likely to reduce rural poverty,
it is no accident that in circles where
‘structural adjustment is seen as a good thing
there is also an almost single-minded
obsession with this causation, to the point
of excluding from consideration other
possible determinants of the incidence of
poverty.
However, given Dharm Narain’s critique
of earlier work which concentrated only on
agricultural output, these more recent analysts
of rural poverty in India do not entirely
forget the price dimension as a possible
determinant of poverty. But, interestingly,
the focus in such recent analysis is almost
entirely on how inflation is bad for the poor.
Thus, cither an inflation term is added to
agricultural output in statistical models
explaining poverty or real wages arc added
as an explanatory variable to the poverty
equation and an inflation term is added to
the equation explaining real wages. The
argument is the entirely plausible one that
Table 10. Pooled Time Series and Cross Section Across States
(Dependent Variable: Head Count Poverty Ratio)
Constant
Ag output per rural person
Per capita state domestic product
State development expenditure
Relative food price
Inflation rate
R Bar Squared
DW
amber September 1996
5.3(9.61)
-0.2(2.87)
0.1(0.89)
-0.2(5.42)
0.6(5.16)
0.1(1.42)
0.87
2.21
5.8(16.59)
-0.2(3.67)
-0.2(7.61)
0.6(5.45)
0.88
2.22
2469
rural money wages (and possibly certain
other components of rural income, such as
the proceeds of the previous harvest) are not
index-linked and therefore are not
immediately protected against inflation.
although these are likely to adjust in the
longer-run .
But, although very plausible as an
explanatory variable of short-run variations
in poverty, the choice of inflation as the
preferred price variable is again not entirely
accidental. For reformers, this leads to the
happy coincidence that while structural
adjustment is goodfor poverty because that
is likely to shift resources towards
agriculture, stabilisation is also good for
poverty because this will reduce inflation.
Most importantly, this way of incorporating
prices avoids facing the essential trade-off
that Dharm Narain was pointing to: that
between the possible beneficial effects of
higher agricultural output on poverty and
the possible losses involved if the preferred
strategy for increasing agricultural output
consists of higher agricultural prices.
This is a trade-off particularly relevant for
the structural adjustment policy as it relates
to agriculture. As has been pointed out by
many, this policy seems to have given no
serious consideration to agriculture in terms
of new programmes or investment, although
it purports to be a virtual overhaul of the
entire economy. But this is no accident cither.
In the liberal! sing worldview, most economic
problems can be resolved by a greater
recourse to markets and allowing the price
mechanism ’free’ play. A similar position
governs the attitude to agricultural growth.
in that it is supposed that relative price
movements and profitability ratios will be
sufficient to ensure that supply responsive
ness in agriculture will lead to higher rates
of growth. And the critical variable here is
the ratio of agricultural to other prices in the
economy, which is sought to be increased
by reducing trade protection to industry,_
through devaluation which makes non-traded
goods cheaper relative to traded goods, and
through removal of restriction on
international trade in agricultural goods
which would have the effect of increasing
the domestic relative price of most
agricultural products, including foodgrains.
In other words, the very mechanism
through which agricultural output is expected
to increase under structural adjustment
involves increasing the price of agricultural
goods, notably food, relative to all other
prices in the economy. This essential rise in
'he relative price of agricultural goods is
thus not seen as a transient phenomenon like
inflation, and it would leave the rural poor
unaffected adversely only if the prices of
goods and services they sell rises in line with
the rise in price of food which makes up most
■f theirconsumption basket. The effect would
2470
certainly be adverse if money wages are
sticky. But. even with flexible wages and
full employment, wage workers would
invariably lose if they got a substantial part
of their income from non-agricultural
activities in relation to which agricultural
prices would require to rise. Unlike inflation
whose control may be benign (if not
accompanied by deflation), this essential
relative price implication of structural
adjustment is permanent by design and so
also is its likely adverse effect on poverty.
It is, therefore, noteworthy, and hardly
accidental, that in the new set of poverty
models, such a relative price variable is
hardly ever included.
Moreover, deflationary policies designed
to control overall inflation during a transition
when domestic price relatives adjust to
international levels may have a
disproportionate adverse effect on the rural
poor if the latter obtain sizeable parts of their
income from non-agricultural activity. In
Indian discussions on the subject, it is
sometimes almost assumed that income and
employment multipliers associated with
fiscal and monetary policy neverspill across
municipal boundaries into rural areas.
Nothing could actually be further from the
truth. With agricultural output determined
from the supply side, and agricultural prices
made inflexible downwards by government
support operations, it may indeed be the case
that the income multipliers of a deflationary
package arc borne entirely by non
agriculture. But given that most organised
sector workers still have secure employment
at pre-determined wage rates, the entire
burden of the employment multiplier falls
on the unorganised non-agricultural sector.
This sector does not respect rural-urban
boundaries and cut-backs in employment
demand here are likely to have knock-on
effects on the incomes of the rural poor:
through a combination of lower nonagricultural employment, falling real wages,
and an increase in the extent of disguised
unemployment in the agricultural sector.
That this latter effect might be important
has already been indicated by the data so
far presented on rural poverty and nonagricultural employment. Its plausibility is
enhanced because much of the government
, .
,.i. :.
irtant..
....I.
1 ;..>j^c, .a ptugiainine Itlndcu and tin...
mo. c..'u:w'cpliblc to cxpendilu.e cuts, because
most rural non-agricultural enterprises are
small and lack staying power, and because
most of the wage employment thus created
is casual. As a result, it would be no
exaggeration to claim that the vulnerability
of the rural non-agricultural sector to overall
public expenditure cuts and to restrictive
monetary policy is almost certainly greater
than for its urban counterpart. This has very
important ramifications in thccurrcnl macro
economic context.
Thus, there arc two possible stories which
can be told about the impact of structural
adjustment and stabilisation on rural poverty.
The first is the benign one: that by increasing
agricultural output and controlling inflation,
these act to reduce poverty. Alternatively,
there is the less optimistic but no less possible
outcome that structural adjustment acts
adversely on the poor because ‘getting prices
right’ leads invariably to a rise in the relative
price of food, because greater reliance on
market forces spurs inequalities inherent in
the commercialisation process, and because
these adverse effects are compounded by
contracting non-agricultural employment and
falling wages in the unorganised sector if
the government wishes to contain, through
contractionary stabilisation policies, the
inflationary fall-out of adjustment.
Which of these actually transpires is an
empirical matter, and one would expect
economists to have tested for which of these
effects are more likely. But oddly, the benign
agricultural output/ inflation story of rural
poverty seems to hold the fort without being
tested seriously against the alternative
hypothesis involving relative price changes,
commercialisation, rural non-agricultural
employment and the government’s fiscal
and monetary stance.
On the other hand, the discussion of actual
developments earlier in this paper suggests
that the simple agricultural output/ inflation
story about the determinants of rural poverty
can be quite misleading. Thus, any
explanation of falling rural poverty during
the mid-1970s and 19S0s would appear to
be incomplete if it did not incorporate the
fact of increasing rural non-agricultural
, Table 11: Duit-Ravaluon Cross-Secuon ano Time Series Pooled
(Dependent Variable: Head Count Poverty Ratio)
Time Varying Variables
Mean consumption
Ag productivity
Rate of inflation
State dev expenditure
Initial Conditions
Irrigation
Female literacy
Infant mortality
R Bar Squared
Economic and Political Weekly
-0.1(2.58)
0.6(6.57)
-0.3(5.12)
-1.1(15.62)
-0.0(0.79)
0.2(2.35)
-0.1(3.33)
-0.6(3.08)
-0.4(2.59)
0.9(4.69)
0.86
-0.4(3.32)
-0.1(1.09)
0.4(3.18)
0.94
Special Number September 1996
.
*5
3
0
□
5
3
3
3
□
3
'$££•27
INTERACT
Economic and Political Weekly
Special Number September 1996
2471
employment and the role of government
policy behind this. Similarly, our critique of
the Tendulkar-Jain explanation of the
increase in poverty post-reforms is also
essentially that they fail to go beyond the
agricultural output/ inflation story. In the
remaining part of this section we attempt a
statistical investigation.
To begin with, we regress head count
measures of poverty for the period 1960-61
to 1993-94 at the all-India level in both
urban and rural areas against per capita
agricultural and non-agricultural incomes
(with agricultural incomes defined per head
of rural population and non-agricultural
incomes defined per capita of total
population). As the accompanying table 8
shows, both income variables are
significantly negative in both rural and urban
areas, and in both cases the agricultural
income variable appears more important
Next, the per capita non-agricultural income
variable is split into per capita income
from trade and transport (an indicator of
commercialisation) and other non-agri
cultural incomes. The rationale for this is
that many commentators (e g, Vaidyanathan
1986) have argued commercialisation this
tends to increase rural inequalities. In these
respecified equations, agricultural incomes
continue to be negatively related to poverty,
but now there is a difference between the
urban'and rural equations. In urban areas,
the commercialisation variable appears to
reduce poverty while the remaining nonagricultural incomes have a positive effect.
But exactly the opposite pattern appears in
rural areas. Next, we include a public ex
penditure variable (development expenditure
per capita, i e, government expenditure less
interest payments and expenditure on defence
and administration). This variable is strongly
significant, reducing poverty in both rural
and urban areas, with, interestingly, its
coefficient almost double in rural as
compared to urban areas. Moreover, on
inclusion of this variable, the non-agricultural
income variables become insignificant in
urban areas, while in rural areas the earlier
pattern is maintained.
These observations suggest a much more
complex relationship between growth of nonagricultural incomes and poverty than usually
appreciated. As expected, commercialisation
does indeed seem to be associated with
increased rural inequalities, but it seems also
to be associated with some increased
opportunities for the urban poor. On the
other hand, the expansion of other nonagricultural incomes appears to have reduced
rural poverty while doing nothing to reduce
urban poverty. This suggests that to the
extent that the benefits of such income growth
do percolate down to the poor, this spills
over disproportionately to (he rural sector,
cither because of rural-urban migration or
2472
•efnuit period, is almost the same as that
because, as suggested earlier, employment
■ . - ' i" lheii miili:i;d paper ..ml appears
multipliers arc higher for rural nonagricultural employment. This latter la tie a la.ily good lit. uOtVCa.. u.C sCCi..,..
fit, that of the same model fitted to data
suggestion finds some confirmation from
the coefficients on the government extended to the post-reform period shows
that the model breaks down almost
expenditure variable, whose significance
completely since the most important variable,
suggests also that such expenditure has a
agricultural output per acre, turns totally
much largest mpacl on poverty than a general
insignificant. Furthermore, the breakdown
increase in non-agricultural output.
of the model occurs essentially because the
Surprisingly, on including an inflation
magnitude of the actual increase in poverty
term in these equations, this is found to have
is well beyond anything that this model can
an insignificant effect on poverty in both
predict. In fact, when the model estimated
rural and urban areas. Replacing the inflation
term with a relative price variable (the relative
with data up to 1987-88 is extrapolated, it
is able to explain only a small part of the
price of cereals to all commodities in the
largcaclual increase in 1992, and also predicts
wholesale price index) docs however make
an increase in poverty in 1993-94 as against
a difference. This variable turns out to be
highly significant in both areas, and also an actual decline.'1
This table also gives the details of an
serves to reduce the significance of the public
alternative model fitted to the same data
expenditure variable, which however
incorporating our observations in the
continues to be significant.
preceding discussion. Here in addition to the
These results need to be interpreted with
agricultural output and real wage variables
caution because the mutual correlation
we included the relative price of cereals, the
between these explanatory variables is often
proportion of non-agricultural workers in
high. But three points emerge quite strongly
rural population and the commercialisation
First, that agricultural incomes are important
variable (the per capita GDP from trade and
not only for rural but also urban poverty.
transport). An intci vsting observation is that
Second, that non-agricultural impulses,
on inclusion of these variables, the time
particularly public expenditure, are not only
trend and the lagged dependent variable
important but that they arc especially so in
used by Ravallion-Duttt turn insignificant,
the determination of rural poverty. Third,
suggesting that in fact the adjustment of
that, as far as the price variable is concerned,
poverty to real factors is much faster than
the relative price ef fect is if anything much
that suggested by the Ravallion-Dutt model.
more important than the effect of inflation
This alternative model not only fits past
per se.
Taken together, these results suggest that
data much better than the Ravallion-Dutt
we need to modify the view that the principal
version, but also, unlike theirs, remains
determinants of poverty are agricultural
robust when extended to the post-reform
output and inflation, and that, therefore,
period. In parli-cular, the massive increase
both structural adjustment and stabilisation
in rural poverty in 1992 is predicted with
arc good. To consider this matter further, we complete accuracy by the model fitted up
replicated the version of this story as it
to 1987-88.19
emerges from the World Bank publication
These results not only emphasise the
Growth and Poverty in India.'1 This is more
importance of the relative price variableand
sophisticated than most other versions of the of non-agricultural factors, they cast strong
story in that it is a two-equation model
doubts on the simple agricultural output/
whereby the incidence of rural poverty is
inflation paradigm. This paradigm is further
'explained' by the agricultural real wage and
compromised because it is seen that the
the lagged and current agricultural output
wage equation in the Ravallion-Dutt model
per net sown acre. In addition the model
also collapses in the sense that inclusion of
includes the lagged dependent variable and
alternative variables, c g, real government
a time trend. In turn, the level of the
expenditure per capita, renders the
agricultural real wage is ‘explained’ in
agricultural output variable insignificant.'
another equation by the inflation rate and
Indeed, the significance of the government
the earlier agricultural output variable, in
expenditure variable here confirms lhepos-,
addition to a lagged wage term. In this
sible importance of government expenditure
model, therefore, higher agricultural output
for non-agricultural employment, the rural
reduces the incidenceof poverty both directly
real wage and, therefore, rural poverty. And,
and through its.positive effect on the wage
indeed, including a government expenditure
rate; and infiation works on poverty only
variable along with lagging the employment
indirectly through the wage rate.
variable yields our preferred equation.
Table 9 gives the coefficients of the
Interestingly, this equation fitted to data up
Ravallion-Dutt poverty equation obtained
to 1987-88 accurately tracks the subsequent
when fitted it to the periods 1960/61-1989/90
movement in rural poverty, including both
and 1960/61-1992 with all-India data. The
(he sharp upward movement in 1991 and
first fit. which docs not include the posl1992 and the reversal in 1993-94?"
Economic and Political Weekly
Special Number September 1996
is
tv'ju
ol ,"i
°fr 1
: n'.STi
cxya-
if. Si
IS
£
iV
1 ar
b<
re
'• a
£
-■ a
P
Nonetheless, these observations on the
he attributed to the fact that Kerala began
basis of All-lndia data, though indicative,
with higher female literacy (I per cent) and
cannot be conclusive given that the degrees
lower infant mortality (0.6 per cent). Our
of freedom arc few and because a number own preliminary experiments with such
of possible explanatory variables arc initial conditions confirm the long-run
mutually correlated. For this reason, the
importance for poverty reduction of health
excercisc was repeated using state level data,
and education status, though much less so
in the form of a pooled time-seris and cross
of irrigation, and suggest furthermore the
section analysis with data up to 1992. In this
importance of initial land distribution.
exercise, poverty was regressed against
Clearly, this analysis suggests that both
agricultural output per rural person, state per the benign agricultural outpu (/inflation model
capita SDP, a relative food price index
and the relative price/statecxpenditurc/rural
calculated by dividing the index of the food
non-agricultural employment models
price in the CPIAL by the SDP deflator, an
mentioned earlier arc relevant for the
inflation index based on the SDP deflator,
determination of rural poverty. But the really
and per capita real state development
important conclusion is that, of the two, the
expenditure. All variables except the per second is by far more important: agricultural
capita SDP were significant, but. in addition,
output and inflation do matter, but as
the inflation term was small and just crossed
determinants of the incidence of poverty, the
the significance level. The relative food price relative price of food and the level of
iable was easily the most statistically
government expenditure arc even more
nificant variable and it also was the most
important. In addition, the analysis points
important in terms of its impact. The next
to an important and hitherto ignored longimportant variable was state development
run synergy between efforts at improving
expenditure, followed by agricultural output.
the health and education status of a society
These results with pooled cross-section
and its ability to bring down poverty over
and time series data at the state level are in
time.
many ways similar to results obtained
recently by Dutt and Ravallion (1995) with
V
more or less the same data set, but restricted
Policy Conclusions
to the period up to I989-90.3' In their model
they regress poverty measures against
The results above do not lead to any very
agricultural output per hectare, state optimistic prognosis about the effect of
development expenditure and an inflation
structural adjust mentor of further 'marketist
term, along with certain indicators of initial
reform’ on poverty. It is tnic that ceteris
conditions (c g, irrigation, infant mortality paribus an increase in agricultural output
and female literacy). All the variables had
would reduce poverty and that, therefore,
the expected sign, and, interestingly, they
there is a ease for diverting more resources
find that state development expenditure was
to this sector. It is also true that any expansion
the most significant variablcand that, unlike of employment in the unorganised sectors,
agricultural output which reduced poverty
say through the rapid growth of labouron^^ increasing mean consumption, state
intensi vc exports, would also reduce poverty.
expenditure reduced poverty by increasing
And it remains extremely plausible that any
both mean income and improving income
policy which can moderate inflation without
distribution. They did not include any relative
leading to a cut-back in employment
price variable but our experiments with the opportunities would in general benefit the
same data suggest that this would have
poor. Nonetheless, there arc trade-offs
swamped the inflation term had they done
involved in achieving each of this goals in
so. Hence, the importanccofstate expenditure
the structural adjustment package, and it is
and of the relative food price appears to be
precisely these trade-offs which arc cause
fairly robust as factors explaining poverty
for pessimism.
both across time and space.
The basic thrust of the structural adjustment
However, perhaps the most important
strategy is to allow greater play to market
result of this Dutt-Ravallion exercise is that
forces and to ensure that domestic relative
it shows that, quite apart from the
prices reflect the opportunities available in
contemporaneous effect of prices, output
international trade. In theory, domestic
and government spending, theextent to which
liberalisation would cause a greater degree
a particular state could reduce poverty over of commercialisation, and liberalisation of
time depended also on the initial conditions
international trade would cause shifts in
with respect to physical and human
relative prices in favour of agriculture and
infrastructure, in terms of irrigation, female exportables. Taken together, these arc
literacy and infant mortality, with which that
expected to bring about the desirable shift
of resources towards agriculture and labourstate began. Thus, of the difference of 1.8
per cent per annum between the rates of intensive exports. However, the problem
with this is that not only do these very
poverty reduction in Kerala and Madhya
mechanisms, commercialisation and a rise
Pradesh, fully 1.6 percent per annum could
V
Economic and Political Weekly
Special Number September 1996
in the relative price of agricultural products.
act directly to increase poverty, but also that
the magnitudes of the elasticity of poverty
to these make it extremely unlikely that the
direct loss on this account can be made up
by the indirect benefits accruing from the
better resource allocation that is expected to
result thereby. For example, in almost every
poverty equation reported above, the positive
coefficient on the relative price variable is
twice the absolute size of the negative
cocl Iicient on the agricultural output variable
- implying that the elasticity of food
production to the relative price of food would
have to be greater than two if such a change
in relative price is to reduce poverty through
higher food production.
This cautions against any sudden opening
up of the foodgrains sector to international
trade: and. indeed, the caution here should
be greater than simply one of managing a
careful transition to world prices. The fact
that the relative price specification is more
important than the inflation specification
suggests that the underlying problem is
caused by more than a stickiness of money
wages in the face of price increases. If such
stickiness existed without any long-run
impact of relative prices on poverty, the
problem would have only been a transitional
one which could be managed by keeping
inflationary pressures in check either by a
graduated movement to world prices or
through a morceffectivestabilisation policy.
However, since poverty is extremely sensitive
to relative prices there is more than a
transitory problem with opening up
agriculture to international trade. Also, with
government expenditure important for
poverty, there is the further important trade
off between this direct effect and the indirect
effect, through inflationary pressures, of the
fiscal policy stance. Given the relative
importance of the government expenditure
and inflation variables in the poverty
equations, attempts to use contractionary
expenditure policies to deal with inflation
pressures, say as a result of a greater opening
up to international trade, could prove to be
a ease of the medicine being worse than the
desease.
In any case, there appears to be
considerable evidence that the increased
government spending during 1976-90 was
among the principal reasons why India could
record rather impressive declines in poverty
during this period. However, it is also true
that the sustainability of such expenditure
increases in the future is more doubtful than
ever before. During 1976-90. real per capita
government development expenditure
increased at an annual rate of 6 per cent per
annum as against only a 3 per cent growth
in per capita real GDP. Real government
expenditure percapita fell 15 percent during
1990-93. but increased again by 6 per cent
SJtfustrious
Glorious
PAST
PRESENT
unr.
rad
0!
FUTURE
per
Sig
On July 20,1908, a lamp was lit in
the city of Vadodara. It was the
,
A network of 2476 branches 38'of them in twelve countries
spanning six continents.
lamp of Bank of Baroda, which
A 20 million plus customer-base
A dedicated work force of over
soon lit thousands of lamps in the
45,000 women and men.
Business-base of over
Rs. 44,000 crores. Deposits of
over Rs. 28,000 crores and
advances of over
Rs. 16,000 crores.
Business-per-employee of almost
Rs. 1 crore. Gross profit of
Rs. 877 crores (highest among
nationalised banks) and net
profit of Rs. 204 crores. Capital
adequacy ratio of 11.2%
banking arena. Both across the
National and International
frontiers. We have come a long
way...from a capital base of
Rs. 2.8 lakhs, deposits of
Rs. 1.9 lakhs, profits of
Rs. 2000 and a manpower
strength of 20 to what we are
today. Dedication to
customer service and innovations ’
were moving forces then, are
still so now.
(as against 8% stipulated).
Better financials in terms of
Return on Equity, Return on
Assets and Gross Profits to
Working Funds. And a will to
excel on the National and
International frontiers.
Our sights are set high.
Vision 2000, for which strategic
plans have been drawn up,
envisages an increase in the
customer-base from the present
20 million to 50 million
by 2000 AD and
100 million by 2005 AD.
Our plan is to record a business.
level of Rs. 86,000 crores and a
profit level of Rs. 1000 crores by
the turn of the century. We are
technologically upgrading .. . .
ourselves to transform the Bank
into a highly competitive,
market-oriented,
technology-driven bank dedicated .
<
to better customer satisfaction
and socio-economic development
in the country. And achieve
excellence in International
banking with commitment and
innovations. Our goal : Make
every branch of the Bank a
meeting place of banking pleasure
for every customer.
Bank of Baroda
(A Government of India Undertaking)
the Mumbai Main Office in 1919.
INDIA'S INTERNATIONAL BANK
The proposed Corporate
Headquarters at Bandra-Kurla
Complex, Mumbai.
Kasaane
BSaKKatt
-in i sss-
vfhOI
Ptf-
1908 - 1996
-^IMACEADSI
Economic and Political Weekly
Special Number September 1996
c
nc
P‘
£
i G
i.
•W
i
G
?G
in 1993-94. The earlier expansion of
government expenditure had led to large
fiscal imbalances despite the fact that taxGDP ratios had then grown quite
significantly. On the other hand, both GDP
growth rates have been lower and tax-GDP
ratios have been falling in the post-reform
period. It is therefore unlikely that the pace
of growth of government expenditure can
be sustained unless GDP grows at least 8
per cent per annum or there is a definite
policy of increasing the tax-GDP ratio
significantly.
Given this fiscal reality, and the fact that
non-agricultural GDP does not appear to
have much impacton poverty except through
its effect on the sustainability of government
expenditure, it is obvious that there will be
problems with maintaining the pace of
poverty reduction. Even if GDP growth
increased, the current fiscal priorities make
it unlikely that this would be reflected fully
|n public expenditure. One possibility
discussed in this context is to alter the
composition of government expenditure so
that it is more directly focused on poverty
alleviation. But, although this is possible
and desirable, a note of caution must be
sounded on this at the outset. In our regres
sion exccrciscs, we played around with
different components of government
expenditure, and the results suggested,
somewhat surprisingly, that it was the broader
measures of such expenditure which had a
greater poverty alleviation effect, at least
when poverty is measured by the head count
ratio, than narrower and more focused
measures such as that on agriculture and
mral development?1
There arc two possible reasons for this.
First, it may well be the ease that the existing
poverty alleviation programmes arc not
'^bplarly effective and that their impact
on^verty is no greater than other govern
ment expenditure. If so, there is room for
improvement in the design of expenditure
rocused towards poverty alleviation. And,
indeed, a case can also be made out that it
may be possible to transfer funds from such
programmes to rural capital formation
ithout endangering the poverty alleviation
impact. But, secondly, it also appears that
what is really at issue arc much broader
Multiplier effects of overall government
expenditure. Clearly much more work is
required in this area to identify ways in
which the impact of a given amount of
expenditure can lead to more poverty
alleviation, but, although there are certainly
fltcly to be ways of achieving this, it should
, t.ot be expected that it will be possible to
cut-back overall government expenditure
without any adverse effect on poverty. The
' .ml significance of government expenditure
appears to be that it is this which imparts
a 'trickle-down' characteristic to the
’.'u:iomic and Political Weekly
growth process, something which appears
quite weak ifonlyG DPgrowth is considered.
Nonetheless, since this effect is likely to
be greater if government expenditure is
properly targeted, it is necessary to attempt
a brief evaluation of the contribution of
government anti-poverty schemes in the
reduction of poverty. There have been
numerous evaluations of these made by the
government and by independent researchers,
and no attempt will be made here to review
this literature which attempts to measure the
effectiveness with which particular schemes
have been able to target the poor. Suffice
it is to note that such evaluations have by
and large found that asset-creation schemes,
such as the Integrated Rural Development
Scheme, have had less success in alleviating
immediate poverty than rural employment
programmes, although even the latter have
leakages and are often criticised for being
a pall iative whose effectiveness at permanent
poverty reduction are rather low. However,
comparing the official figures on employment
schemes with independent data from the
NSS, four points arc worth making. First,
with the NSS reporting a quantum of
employment in public works which matches
official data well, fears about large leakages
may be rather exaggerated. Secondly, the
schemes appear to be reasonably welltargeted in that they arc availed of most by
casual labour households which have both
the highest poverty and the highest personday unemployment, but the regional
distribution of employment through such
programmes appears to he concentrated in
a few western Indian states, and also public
works appear to have been m’uch more
effective in 1987-88. a drought year, than
in more norm-’! years. Thirdly, it seems
unlikely that the effective transfer through
such schemes was much lower than the wage
cost as a result of incomes foregone by the
workers to take up such employment.2'.
Fourthly, possibly because they are-well
targeted, public works appear to have been
more effective in moderating the severity of
poverty ratherthan its head count incidence.2J
Conceptually, if viewed primarily as an
anti-poverty measure, a well targeted public
works programmes should not provide
incentives for the non-poor to participate
and nor should there be impediments to
participation by anyone who is poor. At least
till 1987-88, Indian schemes seem broadly
to have passed the first test but, except
possibly in Maharashtra, failed the second
both because of a paucity of funds and a lack
of official commitment, except at limes of
natural disasters. Since then, confusion about
the intention behind such schemes seem to
have increased. First, wages offered have
been increased to the statutory minimum
wage rate which is often higher than locally
prevailing wages, thus making participation
Special Number September 1996
more attractive. Second, despite this, funds
available lor such schemes have been cut in
real terms, causing job availability to be
even more rationed. Finally, as a result of
a misplaced importance given to the head
count poverty incidence measure there is a
feeling that these schemes have failed to
reduce poverty, and this, combined with a
general presumption that investment rather
than doles arc what is really necessary, have
led many to argue for an increase in the
materials and expertise content of these
schemes, at the cost of their unskilled labour
content, so as to make them more viable
instruments of rural investment.
These reactions arc confused because the
primary goal of an anti-poverty measure is
not the creation of assets and nor is its
purpose a general redistribution of income,
say by increasing the general wage level.
This is not to argue that these are not laudable
objectives but simply to point out that
attempts tochase too many objectives without
substantially increasing the budget available
risks diluting the primary goal of poverty
alleviation. Both higher wage rates and a
lower component for unskilled labour in
these schemes reduce their transfer content.
And. these objectives, by attracting richer
workers and/or by directing employment to
regions where viable investment projects,
rather than the poor, exist arc also likely to
make for much less effective targeting.
Possibly, the correct approach would be to
make employment guarantee the primary
concern of such programmes, setting the
wages paid to levels where the demand for
such employment would broadly match the
funds available. Clearly, if more funds can
be directed into such programmes the wage
rale paid can he increased, and with sufficient
expenditure even the general wage rate
influenced within thccmploymentguarantcc
framework. The employment guarantee
aspect should, however, be the primary
concern and higher wages the secondary
concern because only this priority would
keep out the relatively rich while allowing
the poor unimpeded access.
Secondly, the best way to dovetail
productive investment into such a programme
would possibly be to give a wage subsidy
equal to the employment guarantee wage
rate for each unskilled worker working on
a class of well-defined approved investment
projects, delinking project choice from the
agency implementing the guarantee scheme
and treating the rest of the project cost and
benefit on par with any other. With the
employment guarantee scheme in place, this
need not cost the exchequer any more and
yet the linkage between poverty alleviation
and productive investment through labourintensive schemes could be decentralised.
Needless lo say. this means that other project
costs would have to be met from'otitsrdcthc
employment guarantee budget. But this is
the proper way of proceeding because while
there is a case for subsidising employment
if there is paid underemployment at the
normatively chosen employment guarantee
wage rate, there is no case to subsidise any
particular investment more simply because
it is selected by the agency implementing
the employment guarantee scheme.
However, the main lesson from earlier
sections is that the basic thrust towards
permanent reduction of poverty must take
the form cither of increasing employment
7n agriculture, mainly through better
irrigation and multiple cropping, or of
increasing the stock of viable self
employment opportunities ui regular jobs in
non-agriculturc. It is towards these objectives
that rural investment should be encouraged
while employment guarantee provides the
framework within which this can be done
without sacrificing the need to combat
poverty immediately. Yet. because the
reforms themselves have aspects which tend
to increase poverty, and because fiscal
considerations mean that it might be difficult
to increase both agricultural investment and
the expenditure on anti-poverty schemes.
there will be difficulties in the future.
Under these circumstances, it is clear that
if poverty reduction is to be a serious part
of the agenda in the reform period, the
reforms themselves should have an explicitly
redistributive content. This would require
cuts in subsidies to the rich and also higher
taxes to maintain and increase theexpenditure
relevant for the poor. In addition, the old
issues of land distribution and the provision
of universal primary health and education
must again be put back on the agenda. But,
more than any thing else, it must be recognised
that a 'reform' strategy which aims to
withdraw the state from investment, liberalise
finance and thus divert finances from the
state to the private sector, liberalise
agricultural trade and thus enrich the rich
at the direct cost of the poor, and seeks to
control inllalion and BOP problems through
deflation and devaluation is at its root a
fundamentally inequitous adventure.
Notes
[This is a slightly revised version of a paper
delivered at the Workshop on Economic Reforms
and Poverty Reduction organised jointly by the
Institute of Development Studies. Sussex and the
Lal Bahadur Shastri National Academy of
Administration. Mussooric, and held in Mussoorie
in February 1996 ]
■ ■ . ? L'htindrasckh:..
Abhijil Sc...
.........„,u:y Declined will, Reforms?'.
Macruscan, Businessline, January 23, 1996.
Sec also Jaya Mehta. 'Poverty Figures and
the People of India' in Ahernaiive Ecinuiuiic
Survey 1995-96, published by the Delhi
Science Forum for the Alternative Survey
Group.
2 Ozler,B,GDuttandM Ravallion, ‘A Database
on Poverty and Growth in India’, Poverty and
Human Resources Division. Policy Research
Development,The World Bank, January 1996.
This data base which is available in diskette
form contains data up to the 48th Round
(January-December 1992) of the NSS. It also
has compilations of the poverty line for the
50th Round (July 1993-June 1994), but docs
not contain poverty estimates for this round
because the NSS data was not available. The
poverty estimates for 1993-94 are ours, using
the World Bank poverty lines (which use the
Planning Commission benchmarks of Rs 49
and Rs 57 at 1973-74 prices for rural and
urban India but use somewhat different
deflators) and the World Bank computer
programme POVCAL for poverty estimate
calculations with the 50th Round NSS
estimates.
3 This relies heavily on Tendulkar. S, D K
Sundaram and L R Jain, ‘Poverty in India
1970-71 to 1988-89', ILO-ARTEP, 1993.
4 Sec the next section, and particularly
Table 7.
;
We’ve given you a series of firsts.
Because you come first and foremost
Asian Paints.
A series of firsts
add up to make us No. 1
When you're going to be first, you try a little harder. That's why we reach out to you with the widest
distribution. The most competitive prices. And, of course, the largest range.
It takes foresight to come out first.
First with a comprehensive retail network. Our products are sold in over 25,000 outlets spanning the length and breadth of the country.
□
First Indian paint company to set up a subsidiary abroad.
□
First with Super Decoplast - India's No. 1 affordable Plastic Emulsion Paint.
□
First with corrosion-resistant protective paints for Urea Prill Towers.
□
First with computers for integrated management systems. Also the first Colour Control System lor 100% shade fidelity.
First with Krilo - India's first pure acrylic car finish
28 years of leadership
2476
^asian paints
Economic and Political Weekly
Special Number September 1996
O’ !
o
India’s Checkered History in Fight Against Poverty
o
Are There Lessons for the Future?
o
Martin Ravallion
Gaurav Daft
0
o
o
D
□
0
o
o
o
J
J
J
□
□
□
□
□
Looking back 40 years or so, progress against poverty in India has been highly uneven over time and space. It
took 20 years for the national poverty rate to fall below - and stay below - its value in the early 1950s. And trend
rates of poverty reduction have differed appreciably between states. This paper provides an overview of results from
a research project which has been trying to understand what influence economywide and sectoral factors have played
in the evolution of poverty measures for India since the 1950s. There are some clear lessons for the future.
THERE has been much debate about how
best to fight absolute poverty. Total numbers
of poor people in the world - by almost any
accepted standard - arc continuing to rise.1
The urgency of resolving the debate, and
takat effective action, is greater than ever.
'^^extent to which poor people share in
economic growth has been one of the most
contested issues. Some observers have argued
that "distribution must get worse before it
gets better” in developing countries, and that
this puts a severe brake on the prospects for
pro-poor economic growth. There have also
been debates about the effects of growth in
specific sectors. For example, some have
argued that the benefits of the ‘green
revolution’ (which resulted in substantial
gains in agricultural yields through newseed
varieties and irrigation) were captured by
relatively well-off farmers, and brought little
or no gain to the rural poor. Others have
pointed to farm-output growth as the key to
poverty reduction, both directly and via its
effects on rural wage rates.2
There would be little risk of exaggeration
in saying that the position one takes in such
deb^b has great bearing on long-standing
issu^rof development strategy and policy
reform. The link between growth and poverty.
and the interaction with other factors
(including human resource development),
his also taken on new urgency in the wake of
recent macrocconomicdifliculties and adjust
ment efforts in many developing countries.1
However, these are difficult issues to
resolve empirically, not least because of (he
paucity of representative and reliable data
over time on the living standards of poor
people. Amongst developing countries. India
has relatively good data for addressing these
issues. At the time of writing, one could
compile a time scries of consumption data
from 34 National Sample Surveys spanning
1951-92. This is one of the longest series
of national household surveys suitable for
Backing living conditions of the poor. Most
of the surveys arc large enough to be con
sidered representative at the urban and rural
levels for most states, and they appear to be
reasonably comparable over time since the
and Political Weekly
basic survey method has changed relatively
little. Other data (on price indices and ex
planatory variables) are also available on a
reasonably consistent basis. Although there
arc data problems (some of which we can
make corrections for), they arc modest by
the standards of cross-country comparative
studies. The existence of a time series of
consumption distributions spanning 40years
represents a unique opportunity to study the
link between living conditions of the poor
and the key macro-economic and sectoral
variables which arc thought to have important
influences on progress in reducing poverty.
We have used these data to study the past
evolution of living standards in India. We
have asked: How have comparable measures
of poverty in India evolved since the 1950s?
Has the experience been different between
urban and rural areas and between different
states? How have measures of poverty
responded to changes in economy-wide and
sectoral variables? What has been the relative
importance of economic growth versus
changes in distribution? What role has been
played by the sectoral composition of
economic growth? How important have
changing wages and prices been? Why have
some states of India done so much better
than others in the fight against poverty?
What role have differences in the initial
levels of human development played, versus
other factors such as physical infrastructure
endowments? This paper provides an
overview of the results of this research. We
avoid details on data and methods, which
are described more fully in a series of papers
from the project.4
Data on Poor People
To address the questions posed above, we
constructed a new set of consistent estimates
of various poverty measures for India over
the period 1951 to 1992 from the National
Sample Survey (NSS) data. We aimed to
measure ‘absolute poverty', by which we
mean that the extent of any household’s
poverty depends solely on its own absolute
standard of living (for example, a household
docs not switch from being poor to non-poor
Special Number September 1996
when it moves across sectors unless its
standard of living has changed).’ Following
now well-established and defensible practice
for India and elsewhere, the standard of
living was measured by consumption
expenditure (including imputed values for
consumption from own product ion). We only
studied 'poverty' in the narrow (though
unquestionably important) sense of
"command over commodities"; we do not
deny that there arc aspects of a broader
concept of 'well-being' which are not
captured by our poverty measures.'’
The poverty lines used were thosedefined
by India's planning commission.7 The rural
poverty line is Rs 49 per month and the urban
line is Rs 57 per month at October 1973June 1974 all-India rural and urban prices
respectively." The nominal consumption
distributions for each survey data were then
converted to constant prices using consumer
price indices for urban and rural areas which
were anchored to the consumption patterns
of low-income workers.'1
Three different poverty measures were
used: (i) The headcount index, given by the
percentage of the population who live in
households with a consumption per capita
less than the poverty line. This measures the
incidence of poverty; (ii) The poverty gap
index, defined by the mean distance belowthe poverty line expressed as a proportion
of that line (where the mean is formed over
the entire population, counting the non-poor
as having zero poverty gap). This reflects
the depth of poverty, as well as its incidence;
(iii) The squared poverty gap index, defined
as the mean of the squared proportionate
poverty gaps. Unlike the poverty gap index,
this measure reflects the severity of poverty.
in that it will be sensitive to inequality
amongst the poor."-11 The estimated poverty
measures were then collated with a variety
of macro-economic and sectoral variables.11
How Much Progress Has India Made in
Fighting Poverty Since the 1950s?
Table 1 gives our estimates of the three
poverty measures for eight periods formed
by aggregating NSS rounds; Figure 1 gives
2479
1 ,/UndS, V/C
J lllC
w ..
Lt.
-u.- >:
i.i cjjh of me iiu v-:
■ .ncasurc^uc^
various measures of the rate of aggrcgaltfe
economic growth between the same rounds.
Based on the regression coefficients,
Table 2 gives our estimates of the percent
age change in each poverty measure to be
C
expected from a 10 per cent growth rate
for (i) the mean consumption per person as
estimated from the NSS; (ii) mean
consumption per person estimated from the
national accounts and population census;
and (iii) mean net domestic product pa
person, also from the national accounts and
census.'8
The national poverty measures responded
significantly to all three measures of
economic growth. For example, a 10 pa
■101
cent increase in mean consumption resulted
in a 12-13 per cent drop in the proportion
of people who are poor, representing a 1011 per cent drop in the number of poor, at
India’s rate of population growth. The
responses are higher if one uses the NSS
-20
estimate of mean consumption, rather than
the national accounts estimate, though the
difference is small for a given poverty
measure. The responses are lowest for net fi £
domestic product. This may be due to inter
temporal consumption smoothing which may
make poverty (i n terms of consumption) less
responsi ve i n the short- term to income growth
than to consumption growth.
Notice too that the responses tend to be
greater if one uses the poverty gap index r
rather than the headcount index, and the
response is largest for the squared poverty
Io
gap, which is sensitive to both the depth and
e-
C1
9,
National
the estimates of the headcount index for each
survey round. (The pattern of change was
very similar for the other two poverty
measures.) Several points emerge:
The period from the early 1950s up to the
mid-1970s was characterised by fluctuations
in poverty without a real trend in either
direction.1' The average headcount index
was 53 per cent in 1951-55 (marked in
Figure 1), about the same as its average
value in 1970-74.13 After that there was a
significant decline in poverty incidence (and
the depth and severity of poverty fell too),1'’
though this was not a continuous decline.
it thus took over 20 years for the poverty
measures to finally fall below - and stay
below - their values in the early 1950s.
Changes in rural poverty closely follow
those-at the national level, which is not
surprising given that a large proportion of
India’s population lives in rural areas (about
74 per cent even at the end of the period).
It is more notable that a similar pattern over
time also holds for urban poverty (Figure I).
Common causative factors appear to be at
work.
The reduction in poverty since the early
1970s has been sizeable; between 1969-70
and-1992, the national headcount index
declined from 56 to 41 per cent. Yet India’s
progress against poverty has been modest
when compared to the standards set by some
countries in east Asia. For example,
Indonesia’s headcount index was 58 percent
in 1970—very close to ourcstimate for India
al that lime. But by 1993 (keeping the same
real poverty line over time), we estimate that
2480
Rural
the headcount index for Indonesia had fallen
to 8 per cent, about one fifth of India’s
headcount index in 1992.1’
a
How Important to India’s Poor was
Economic Growth and Contraction?
We look first at the effect of aggregate
economic growth and contraction on
poverty. Comparing successive survey
Table 1: Poverty in India 1951-1992
NSS Rounds
Period
Headcount index
3-8
1951-55
9-15
1956-60
16-19
1961-65
20-24
1966-70
25, 27, 28
1971-75
32.38
1976-83
42-45
1984-90
46-48
1991-92
Poverty gap index
3-8
1951-55
9-15
1956-60
16-19
1961-65
20-24
1966-70
25. 27. 28
1971-75
32. 38
1976-83
42-45
1984-90
46-48
1991-92
Squared poverty gap index
3-8
1951-55
9-15
1956-60
16-19
1961-65
20-24
1966-70
25. 27, 28
1971-75
32. 38
1976-83
42-45
1984-90
46-48
1991-92
■■
Rural
Urban
National
54.77
53.96
48.59
60.44
55.27
47.96
37.94
39.44
42.70
47.06
45.46
50.90
46.04
38.08
34.99
33.24
52.66
52.74
48.02
58.60
53.39 •
45.68
37.20
37.84
19.69
17.91
14.28
19.80
17.01
13.84
9.26
9.47
14.04
15.36
14.04
16.08
13.46
10.60
9.11
8.58
18.70 .,
17.46 .
14.23 '
19.08 ■
16.28
13.09 l
9.22
9.24 ■
9.42
7.94
5.73
8.67
7.08
5.45
3.24
3.23
6.20
6.69
5.85
6.76
5.28
4.04
3.24
3.11
8.86 .
7.72 ,
5.76 'i
8.30 1
6.71
5.13 d
3.24
3.20 ,1
Economic and Political Weekly
Special Number September 1991
Itr
ine<
the
me
of[
I
Th<
C
dis:
cor
ine
£
nfc
Did the Pattern or Growth Matter?
Figure 2: Cumulative Change in Headcount Index
(Total of Growth and Redistribution Components)
-X-
Growth component
Redistribution
severity of poverty. This means that the
impacts of growth and contraction in India
were not confined to those near the poverty
line, but reached deeper.
Redistribution played a role in the long
run changes in poverty in India. Any change
in a poverty measure can be decomposed
into a growth component and a redistribution
component.19 Roughly speaking, the growth
component is the change in the poverty
measure which would have occurred if
^equalities had not changed, while the
ribution component is the change in
jverty measure that one would have
found if there had been no change in the
mean. By adding each component over time
we can assess the cumulative total impact
of growth or redistribution. Figure 2 gives
the results.20
It can be seen that the redistribution
component did help over the whole period.
Thus, for India, our results reject the old
view (still held in some quarters) that
distribution must get worse as a low-income
country grows. Nonetheless, the overall
contribution of redistribution to change in
the headcount index has not been large in
the long run. The growth in mean
consumption has been more important,
accounting for about 80 per cent of the
cumulative decline by the end of the period.
Redistribution mattered more to the other
two poverty measures. For the poverty gap
index, the redistribution component
accounted for about 40 per cent of the
cumulative decline by the end of the period;
»
Economic and Political Weekly
— Total change
its contribution was 47 per cent for the
squared poverty gap index. Favourable
redistribution has thus been quite important
for changes in the depth and severity of
poverty.
Most of the pro-poor impact of
redistribution was realised early on, during
the early to mid-1960s, well before the onset
of the sustained decline in the national
poverty measures. Since the mid-1960s, the
redistribution component fluctuated without
making a further addition to its total longrun impact on national poverty. (This holds
for all three poverty measures.) The gains
to the poor since about 1970 have been
almost entirely due to growth.
The latter finding might be taken to imply
that public efforts at pro-poor redistribution
in the 1970s and 1980s failed. However, one
should be wary of drawing that conclusion
since we do not know the counter-factual
of what would have happened without those
efforts. Possibly distribution would have got
worse.
Turning next to the sectoral composition
of growth, we found that the changes in
national poverty have been for the most part
driven by changes in rural poverty. Figure
3 gives the cumulative (population-share
weighted) contributions of both the urban
and rural sectors to the national headcount
index. The rural sector accounted for more
than three-quarters of the total decline in
national poverty measures over the whole
period.21 Nonetheless-despite the substantial
sectoral shifts in national output that have
occurred over the last 40 years or so povertyin India is still overwhelmingly rural.
In 1992. three-quarters of India's poor lived
in rural areas.
We also looked closely at the intcrlinkagc
between the sectoral composition of
economic growth and the urban-rural
composition of poverty, using econometric
methods to disentangle the various effects
within and across sectors.22 The-main
conclusion was that the relative effects of
growth within each sector, and its spillover
effects to the other sector, reinforced the
importance of rural economic growth to
national poverty reduction in India. Both the
urban and rural poor gained from growth
within the rural sector. By contrast, while
urban growth reduced urban poverty, it also
had adverse distribution effects within urban
areas which militated against potentially
higher gains to the urban poor. And urban
growth had no discernible impact on rural
poverty. The process of growth through
rural-to-urban migration contributed very
little to poverty reduction in India.
When the growth in national income was
broken down by output-based sectors, we
found that there were marked sectoral
differences in the poverty impacts. Both
primary (mainly agriculture) arrd tertiary
(trade, services, transport et al) sector growth
reduced poverty nationally, and th>ey also
did so within both urban and rural areas. By
contrast, secondary (mainly manufacturing)
sector growth brought no discernible gains
to the poor in cither sector. In the historical
shift from primary to secondary and tertiary
sectors it was the latter sector which
delivered the bulk of the gains to India’s
poor.
Table 2: How Responsive Were National Poverty Measures to
Economic Growth in India?
Percentage Change in the Poverty Measure
Attributable to a 10 Per Cent Increase in_________
Mean Net
Mean Private
Mean Consumption
Domestic
Consumption from
from National
Product
National Accounts
Sample Surveys
Headcount index
Poverty gap index
Squared poverty gap index
Sp-cml Number September 1996
-13.3
-18.8
-22.6
-12.1
-17.9
-21.8
-9.9
-14.9
-18.5
2481
Figure 3: Urban-Rural Composition or Chance in the Headcount Index
(Cumulative Changes in the Urban and Rural Components and Population Shift
Cumulative change since 1951-52 (percentage points)
.lion, Nou.v.‘,:gescalchi;peveiiui;iil). j
.vc iuur.J
iijvc.'sc siioM-t*.;
iiiipacl on liic
,. - ;
. izcuoic.
t
example, we estimate that a oncc-and-for- J
all 20 per cent increase in the price level
would result in a drop of 13 per cent in the
current year’s real wage rate in agriculture,
and an increase of 5 per cent in the rural
headcount index. The impacts on the other
poverty measures would be even higher;
the squared poverty gap would rise by
9 per cent.
Did Some States Perform Better Than
Others in Reducing Rural Poverty?
-O- Rural component "A-Urban component
The relative lack of an impact of secondary
sector growth on poverty reflects the type
of development strategy India pursued since
the second plan in the late 1950s, which
emphasised capital-intensive industrialisat
ion within a largely closed-economy regime.
It is not surprising that such industrialisation
brought negligible direct gains to the nation’s
poor, who depend heavily on the demand
for relatively unskilled labour?’
Did the Rural Poor Benefit from
Agricultural Growth?
Since rural poverty has been so important,
we turned our attention to this sector. Here
we examined how much India's rural poor
benefited from agricultural growth, what
role the labour market played, whether the
impacts wercdistributionally biased oneway
or another, and how important macro
economic stability was to the rural poor.
We collated the household survey data
with data on agricultural wages, prices and
output, and estimated a dynamic econometric
model jointly determining rural poverty
measures and real wages?* The model had
a triangular structure in which the rural
poverty measure was hypothesised to be a
function of both the real agricultural wage
ate and the average farm yield per unit area
as well as other variables), and the real wage
ate was also a function of the farm yield
nd other variables.
The results indicated that all three poverty
measures responded significantly in the short
run to changes in agricultural wages as well
241)2
--Population shift —Total change
as to average farm yields. And wages also
responded significantly to farm yields,
presumably through effects on labour
demand, such as due to multiple cropping.
Higher yields thus helped reduce absolute
poverty through induced wage effects, as
well as the more direct channels, including
effects on both employment and own-farm
productivity.
Neither the poverty measures nor real
wage rates adjusted instantaneously to
changes in farm yields. The combined effect
of this stickiness in both variables was that
the short-run gains to poor people of
agricultural productivity growth were far
lower than the long-run impacts. Also, the
short-run effects on rural poverty operating
via the real wage rate were minor compared
to the direct effects of higher own-farm
yields. But in the long run, the wage effects
did matter, accounting for about one-third
of the long run response of absolute poverty
(for all three measures) to a yield increase.
The process through which India's rural
poor participated in the gains from agri
cultural growth did take time, though about
half of the long-run impact occurred within
three years of an initial gain in farm yield.
Did Inflation Matter?
We found evidence of an adverse shortrun impact of inflation on real agricultural
wages and (hence) absolute poverty in rural
areas. The effect of inflation was to reduce
real wages in the short term, because nominal
agricultural wages responded sluggishly to
The regional disparities in levels of living I
in India have been large. For instance, the |
proportion of the rural population of the I
slate of Bihar living in poverty in 1990-91
was about 58 percent, more than three times
higher than the proportion (18 per cent) in
the (combined) states of Punjab and Haryana.
Some of these differences appear to have '
persisted historically, though there were also ■
differential trends across regions. Looking •
back over lime, the more striking - though
often ignored - feature of the Indian
experience has been the markedly different
rates of progress between stales; indeed the
ranking of states around 1990 looks quite
different to that 30 years earlier, as can be
seen in Figure 4 which compares headcount
indices around 1960 with those around 1990.
(The picture looks very similar for the other
two poverty measures?5) For example, the
southern state of Kerala moved from having
the second highest rural poverty rate around
1960 to having the fifth lowest around 1990?
Regressing the log ofeach poverty measure
against time, there was a trend decrease
(significant at the 5 per cent level or better)
in all three measures in 9 of the 15 states,
viz, Andhra Pradesh, Gujarat, Kerala,
Maharashtra, Orissa, Punjab and Haryana,
Tamil Nadu, Uttar Pradesh and West Bengal.
The trend was not significantly different
from zero at the 5 percent level in the other
six states of Assam, Bihar, Jammu and
Kashmir, Karnataka. Madhya Pradesh, and
Rajasthan; there was not asigni ficant positive
trend for any stale for any poverty measure.
There is a tendency for the absolute size of
the trend to be higher for the poverty gap
than the headcount index, and it was highest
for the squared poverty gap.
In terms of progress in both raising average
household consumption and reducing rural
poverty, the state of Kerala turns out to be
the best performer over this period. The
second, third and fourth highest trend rates
of consumption growth were Andhra
Pradesh, Tamil Nadu, and Maharashtra
respectively. In terms of the rales of poverty
reduction, the second, third and fourth stales
were Andhra Pradesh, Punjab and Haryana,
and Gujarat; the ranking is invariant to the
Economic and Political Weekly
Special Number September 1996
c
c
c
£•
•c
c
Figure 4: Poverty by State. 1960-90
Averages for first three survey rounds and last three
choiccof poverty measure though differences
in their rates of poverty reduction arc not
large. The worst performer was Assam by
all measures. The other poor performers
wcreBihar, Jammu and Kashmir, Karnataka,
N^Jhya Pradesh and Rajasthan; the exact
r®lng varies by the measure used.
The states which had the highest trend
rates of growth in mean consumption tended
to have the highest trend rates of poverty
reduction, and the correlation is very strong
(thecorrelation co-cfficicnt between the trend
rate of reduction in the headcount index and
the trend rate of consumption growth is 0.85;
the correlation is about the same for the other
two poverty measures). Both these variables
may well have been influenced by similar
factors. Next we look at what those factors
might be.
What Accounts for the Differing Rates
of Progress in Reducing Poverty?
Every state has its own story with a mixture
of both successes and failures at public action
against poverty in different periods.” There
wercdiffercnccs between states in thcimpacts
of (ostensibly similar) interventions, as well
as differences between states in the package
of interventions pursued; and in both respects
experiences in a given state changed over
Economic anil Political Weekly
time. We cannot hope to capture all this
variance in experience - for one thing we
would quickly run out of degrees of freedom.
Here our aim is solely to look for any
empirical regularities that can account for
at least a reasonable share of that variance.
The inter-state differences in progress at
fighting poverty allowed the project to study
the impact on the trend rate of poverty
reduction of a range of variables, including
regional differences in human and physical
resource development. A pooled model was
estimated, giving 310 observations (15 states
over 21 NSS rounds, though with some
missing observations, or inadequate sample
sizes). A model was estimated for each
poverty measure, with both time varying
and static explanatory variables. The key
explanatory variables were current and
lagged real agricultural output per hectare,
current plus lagged real non-agricultural
output per capita, the rate of inflation, lagged
real state development spending per capita,
and the stale’s initial (around 1960) irrigation
rate, infant mortality rate, and female literacy
rate; the latter three variables were allowed
to influence the rale of change in the poverty
measures (thus entering the model interacted
with time).2" The estimated models could
account for about 90 percent of the variance
Special Number September 1996
over time and across slates in the poverty
measures.2’
The results indicate that higher growth
rates in agricultural yields and real nonagricultural output per capita, lower rates of
inflation and higher growth in state
development expenditure all led to higher
rates of progress in both raising average
consumption and reducing all three measures
of absolute poverty.
The results also suggest that inter-state
differences in initial conditions of human
and physical resource development played
an important role: higher initial irrigation
intensity, higher literacy rates and lower
initial infant mortality rates all contributed
to higher rates of consumption growth and
poverty reduction. Initial inequalities in
access to physical and human infrastructure
appear to have been an important factor in
longer-term rates of poverty reduction.1"
Consider Bihar, one of the worst performers
in poverty reduction (Figure 4). The poor
in Biharsuffered fromthestate’sslowgrowth
in agricultural yields. But the state’s poor
initial conditions were also an important
factor. The incidence of poverty in Bihar
declined at a trend rate of only 0.1 per cent
per year. We estimate that this would have
risen to 1.2 per cent if Bihar had started off
with Kerala's level of human resource
development in the 1960s.
By and large, the same variables
determininggrowth in averageconsumption
mattered to rates of progress in reducing
poverty. Most of the effects on absolute
poverty were transmitted through growth in
average consumption rather than
redistribution, though none of the factors
which reduced absolute poverty had adverse
effects on distribution. Thus, there was no
sign of a trade-off between growth and pro
poor distributional outcomes.
Lessons for the Future
Our investigation suggests that economy
wide variables do matter to India’s poor;
they have generally gained from economic
growth, and lost from contraction; they have
also been hurt by inflation. The net gains
to the poor since the early 1970s can be
attributed in large part to economic growth
- distribution changed little from the point
of view of the poor, though it appears to have
been more important in the 1950s and 1960s.
when there was rather less growth.
The experience of the past 40 years offers
support for the view that a stable macro
policy environment, combined with micro
policy reforms conducive to economic
growth.can help greatly in reducing absolute
poverty in India. However, our results also
reveal important nuances concerning the
pattern of growth, and the importance of
other contingent factors, including human
and physical infrastructure.
Our results point clearly to the quantitative
importance of the sectoral composition of
economic growth to poverty reduction in
India. Fostering the conditions for growth
in the rural economy - both primary and
tertiary sectors - must be considered central
to an effective strategy for poverty reduction
in India. At the same time, the relative failure
of India's past industrialisation strategy from
the perspective of the poor points to the
importance of successful transition to a
strategy capable of absorbing more labour,
particularly from rural areas.
But our results also point to the longerterm importance of investing in human and
physical infrastructure as a complement to
pro-growth reforms in India. Controlling for
growth in farm and non-farm sectors, we
find significant effects on trends in absolute
poverty reduction of the differences between
slates in initial conditions related to
infrastructure.
A final lesson concerns the importance of
being able to credibly assess an economy’s
performance in reducing poverty. Though
less than ideal in some respects, the data base
available for poverty analysis in India is
good by international standards. Many other
countries have had far fewer objective socio
economic surveys on which poverty
monitoring can be based, or their surveys
have been severely wanting in terms of
coverage (lacking, for example, a sound
consump?ion module) or comparability over
time. The very fact that for India we can
obtain the data needed to address the
questions posed above carries an important
message for other countries today, and India
in the future.
Notes
[For their comments we are grateful io Jyoisna
Jalan, Peter Lanjouw. Dominique van de Wallc,
Quentin Wodon. and the f/’IVs referee. This
paper summarises results of a research project.
'Poverty in India. 1951-92', supported by die
World Bank’s Research Committee, under RPO
677-82. However, these arc the views of the
authors, and should not be attributed to the World
Bank, or any affiliated organisation.]
See Ravallion and Chen (1996) for aggregate
poverty measures for the developing world
over the period 1987-93. They estimate that
the percentage of the population consuming
less than $ 1/day at 1985 international prices
(with currency conversions at purchasing
power parity) decreased only slightly over
this period, from 30.7 per cent in 1987 to 29.4
per cent in 1993 implying that the numbers
of people living under S 1/day rose from 1.23
billion to 1.32 billion over this period.
(S 1/day is about equal to India's urban poverty
line.) The gains to the poor in east and (less
so) south Asia were roughly counter-balanced
by the losses in other regions, notably SubSaharan Africa, Latin America and Eastern
Europe and Central Asia.
2 Contrast, for example, Ahluwalia’s (1978:320)
conclusion that “there is evidence of some
I
2484
corrects u..s by replacing the firewood $ubtrickle down associated with uf.iioilii-i.-..
.serics in the CPI Al. by one based on n.eo;
growth" with Sailh’s (1981:205) claim il.’.t
rural firewood price:.
.-..’.ukible froa
“there can be little doubt that cimcm
1970) and a series derived by assuming tlm
processshave served as generators ol poveity :
firewood prices increased at the same rate as
both were using data for India over roughly
all other items in the fuel and light category
the same period (1957-73). I he debate
(prior to 1970). For details see Datt (1996).
continues; in recent literature on India one
10 These are fixed weight price indices. Thus,
can find claims that “rapid agricultural g.i owth
they ignore substitution in response to shifts
has benefited all classes of the poor" | Singh
in relative prices. To lest sensitivity to this,
1990) and “acceleration in agricultural growth
Ravallion and Subramunian (1996) compart
by itself is unlikely to make a dent in rural
poverty measures for India with and without
poverty" [Gaiha 1995:285].
an allowance for substitution effects consistent
3 For an overview of the theory and evidence
with demand behaviour, as modelled by a set
on the effects of adjustment on the poor see
of full rank Gorman Engel curves. Ignonng
Lipton and Ravallion (1995. section 5.3) In
substitution matters far more for some
the Indian setting, see Ravallion andSubbarao
measures and applications than others. It leads
(1992).
to overestimation of inequality, but level
4 See Dall (1996), Ozlcr et al (1996). Ravallion
effects on poverty measures are generally
and Datt (1995,1996), and Datt and Ravallion
small and turning point errors are rare
(1996). Later we identify which paper is most
11 These are members of a class of measures
relevant to each topic covered hcic
proposed by Foster, Greer and Thorbeckc
5 A number of the popular methods ol making
(1984). A transfer of income from a poor
poverty comparisons over time and space do
person to a poorer person (for example) will
not satisfy this consistency requirement; see
not alter either the head-count index or the
Ravallion (1994) for further discussion.
poverty gap index, but it will decrease the
6 For further discussion of this point sec Sen
squared poverty gap index. Furthermore, the
(1987); in the context of India also see Drcz.e
squared poverty gap index satisfies the
and Sen (1995).
subgroup consistency' properly, namely that
7 See Planning Commission (1993).
if poverty increases in any subgroup (say the
8 We compared this difference in the poverty
urban sector), and it does not decrease
lines to independent estimates of the urbanelsewhere then aggregate poverty must also
rural cost of living differential. For 1973-74,
increase [Foster and Shorrocks 1991].
Bhattacharya et al (1980) estimated that the
12 The poverty measures are calculated using
cost-of-living for the poor was 16 per cent
parameterised Lorenz curves. We use either [
higher in urban areas, exactly the same (to
the beta Lorenz curve of Kakwani (1980) or
the nearest integer) as the differential in
the general quadratic model of the Lorenz r
poverty lines. So it can be on argued that the
curve of Villasenor and Arnold (1989),
planning commission’s urban and rural
depending on which fits the data best (both
poverty lines represent the same standard of
satisfied the theoretical conditions needed for s
living, and (hence) that we are making
a valid Lorenz curve in ail survey rounds for
consistent comparisons of absolute poverty
both sectors). Using the formulae derived in
between urban and rural areas. For Imiher
Dan and Ravallion (1992), the poverty
discussion sec Ravallion and Datt (1996).
measures are calculated from the estimated
9 For the urban sector after August 1968. the
parameters of the Lorenz curve and the mean f
all-India consumer price index for industrial
pcrcapita consumption expenditure. A number
workers (CPIIW) is used as the deflator. For
of checks are made on the results, including
the earlier period, the Labour Bmcau's
both the theoretical conditions for a valid
consumer price index for the working, class
Lorenz curve, and consistency checks, such
is used, which is an earlier incarnation ol die
as that the estimated value of the head-count
CPIIWalbeit withasinallercovcragcol uiban
index must lie within the relevant class interval
centres (27 against 50). The rural co-.i of
of the published distribution. The estimation
living index series was constructed m three
technique has been set-up in a user-friendly
parts. For the period since Scptcmbci 1964.
computer programme ‘POVCAL’ [Chen, Datt
the rural cost of living index is the all India
and Ravallion 1991] which is available on
consumerprice index for agricultural labouiers
request, so interested readerscan readily check
(CPIAL) published by the Labour Bmeau.
our calculations and their sensitivity to our
For the period September 1956 to August
assumptions.
,
1964 (for which an all-lndia CPIA1. does not
13 A complete descriptions of the data set and
exist), a monthly series of theall-lndia (Tl AL
all sources can be found in Ozler, Datt and
was constructed as a weighted average ol the
Ravallion (1996) with an accompanying set
state-level CPIALs, using (he same ’.tale
of diskettes.
level weights as (hose used in the all India
14 The first subperiod is marked by three
CPIAL published since September 1961. For
significant peaks in poverty around the yean
(he initial period August 1951 to August
1956, forecasts were obtained from a dynamic
1953-55 (rounds 7,8), 1956-58 (rounds II,
12, 13), and 1966-68 (rounds 21,22), the last
model of the CPIAL as a function ol the
of these coinciding with the worst drought
CPIIW and the wholesale price index. Our
new CPIAL series also dealt with another
in the post-independence period.
problem which has to do with (he fact that
Based on poverty measures averaged over
the Labour Bureau has used the same price
NSS rounds, weighted by the duration of the
survey.
of firewood in its published scries since 196016 The absence of fluctuations in poverty over
61. Firewood is typically a common piopcrty
the period 1975-85 may be somewhat illusory
resource for agricultural labourers, but it is
also a market good, and so the Labour Bmeau' s
as we have only two NSS surveys in the
practice is questionable. Our CPIAL. series
intervening period, viz, those for 1977-78 and
£
£
*
f £
ic
c
ic
: C
r
c
c
c
£
c
I- £
Economic and Political Weekly
Special Number September 1996
is-tlo.
DISCUSSION________________________ _
Rural Poverty and Its Alleviation
in India
N Kakwani
K Subbarao
WE are happy to see a critical scrutiny of
our paper [Kakwani and Subbarao 1990] by
S Tendulkar and L R Jain (TJ for short
hereafter) [Tendulkar and Jain 1990].
The measurement of poverty involves a
number of conceptual and practical diffi
culties. Many a time we need to settle for
second best methods because of non-availa
bility of appropriate information. For in
stance, like most other researchers on Indian
poverty, we have used per capita household
expenditure (PCHE) as a measure of house
hold economic welfare. A better measure of
household welfare will, of course, be the per
adult equivalent consumption which corrects
for the differing needs, of adults and
children. But this measure cannot be
employed in Indian studies because the NSS
data are available to researchers only in
grouped form (the groups formed on the
basis of per capita household expenditure),
although the NSS organisation collects the
expenditure data for each household. The
grouping involves considerable loss of infor
mation which may lead to biased estimates
of poverty. To estimate poverty from such
data, one needs to employ some interpola
tion device Most Indian studies have
employed a two-parameter lognormal
.distribution [Minhas, Jain, Kansal and Saluja 1987], with the exception of Ahluwaua
(1978) who employs Kakwani-Podder’s
[1976] Lorenz function. Since the NSS does
not regularly correct the income ranges in
order to take account of inflation, inap
propriate interpolation devices may induce
large estimation errors. These errors will be
particularly serious when one uses a single
density function such as lognormal to the
entire consumption range. In our study wc
used a general interpolation device [Kakwani
1980] which uses a different density func
tion within each consumption range.
Although this procedure is an improvement
over those employed by previous researchers,
it is still the second best solution.
TJ do not seem to recognise these and
many other problems associated with pover
ty research in India. Had they appreciated
these, they would have been more construc
tive in their evaluation of our paper; instead
they adopted the negative approach of
attacking—wrongly and unfairly in most in
stances as we shall soon demonstrate—
everything in the paper. In what follows, we
respond to their criticisms not in the order
chosen by them, but in order of the impor
tance of the issues raised by them. These fall
into five groups: (a) our choice of price
deflators; (b) problems with the decomposi
1482
tion methodology; (c) growth elasticities;
(d) regression results; and (e) other miscel
laneous issues including validity of our
conclusions.
Price Deflators
The central issues raised by TJ are (a) em
pirical inconsistency in using current prices
for calculating Gini and Theil measures;
(b) price deflators used by us for measuring
poverty are inappropriate; (c) adoption of
“conceptually more appropriate” deflators
would vitiate our conclusions; and (d) data
problems and alleged errors in our estimates
of per capita household expenditure (PCHE)
growth fates.
We first turn'to the alleged empirical in
consistency in using current prices for
calculating Gini and Theil’s measures of in
equality. We used the current prices because
both these measures are relative measures of
inequality and, therefore, will not be affected
if per capita consumption of all households
is multiplied by the same price deflator.
These indices computed at the current and
constant prices will differ only if we assume
that households with different per capita
consumption have different price indices.
The households in the current period can
have different price indices if the consump
tion patterns of households in the base
period are different. If these differences are
significant, w must compute the price index
for each household. The price index of a
household with PCHE x in the base period
will be
m
I(x) = 2
Pli
— w, (x)
(1)
where w/x) is the expenditure share of the
ith commodity (i = 1, 2, .... m), which will
depend on x and P0, and PL are the prices
of the ith commodity in the base and cur
rent period respectively. Then the real in
come of that household in the current
period, which we denote by xR will be
xR = .JSL
(2)
■ I(x) .
where x* is the current per capita consump
tion of that household. In our study we
assumed that wf(x) is independent of x,
. which would imply I(x) will be exactly the
same as that.of x*.
Following TJ let us assume that house
holds with different income have different
consumption patterns or in other words
Wj(x) is not independent of x. Then poverty
and inequality measures (and also per capita
average consumption) must be derived from
the distribution of xR. To compute these
measures, we must rank the households in
the ascending order of their real income xR
but the available NSS data are ranked ac
cording to the current per capita consump- a
tion x*. If these rankings are significantly
different from each other, one cannot do any
meaningful analysis of poverty and inequali
ty. Even if the data were available for indi
vidual households, we could not construct;
fractile groups on the basis of the real
household income because the households
in the base period are not the same as those
in the current period. Since we have no ex
cess to Jain and Tendulkar’s 1989 paper
published in the Journal of Indian School
of Political Economy, we can only wonder
how they succeeded in measuring real levels
of living for different fractiles. We conjec
ture, however, that they have constructed the
fractile groups on the basis of the assump
tion that the ranking of households by xR
is exactly the same as by x*. If so, it would
be methodologically a wrong procedure.
Next, we turn to the issue of inappro
priateness of our choice of price deflators. .
This is not a new issue; it was earlier debated
by P Bardhan and B S Minhas in the early
1970s. TJ point out that we have used the
well established consumer price index for
agricultural labourers (CPIAL) for adjusting
both the poverty line.and the mean PCHE.
Recently Minhas, Jain, Kansal and Saluja
[1990] have worked out state-specific con
sumer price indices separately for the total
rural population (CPITR) and Minhas and
Jain (forthcoming) for the middle range of
the rural population (CPIMR). They suggest
that the conceptually appropriate deflator
for the mean PCHE to be CPITR, and
CPIMR for the poverty line.' They argue that
since we have not used these “conceptually
more appropriate” deflators, our findings
and hence conclusions are wrong.
At the outset we must point out that when
we wrote our paper, the alternative deflators
were not available. The Minhas, Jain, Kan
sal and Saluja paper was published in June
1990 whereas our paper appeared in March
1990. The Minhas and Jain (1990) is still for
thcoming. Therefore, TJ criticism for dur
not using these alternative deflators is un
warranted and unfair. Be that as it may, let
us assume that this set of deflators were
available to us. How much do we gain (or '
lose) by using these deflators? Should one
choose a set of deflators merely because they
are the latest to arrive on the shelf?
While advocating use of two.deflators, TJ
have not comprehended the empirical and
conceptual pitfalls. Their suggestion of
course implies that there exist two
homogeneous groups of households, which
have different consumption patterns. Em
pirically, how should such groups be form
ed? One possible suggestion is to use
CPIMR for the households which are below.
Economic and Political Weekly
June. 15, 1991'
the poverty line and CPITR for those above
the poverty line. Minhas, Jain, Kansal and
Saluja. [1987] have correctly argued that
“methodologically speaking this is not a
sound suggestion. The proportion of the
people below the poverty line is a variable •
entity which itself is the very object of pover
ty measurement”
There are conceptual problems as well.
The correct procedure to compute the mean
of real PCHE is to find the mean of xR in
(2). TJ suggest that the mean of real PCHE
should be computed by deflating the mean
of current PCHE (which we denote by"x) by
CPITR, which is given by
It should also be obvious from the above
ranges and over time should make little dif
^discussion that while Minhas et al have made
ference to the calculation of the consumer
a valiant effort to improve the quality of
price indices. One would, therefore, expect
price deflators, the new indices have their
the difference between CPIMR and CPIAL
own limitations so that our reliance on
to be small because bojh these indices are
CPIAL is not entirely unjustified.
based on the same retail prices. The values
This does not mean that CPIAL, the
of these indices at the all-India level are as
follows:
deflator used by us, is the best. A major
objection raised against this is that the
agricultural labour households constitute
CPIMR
CPIAL
only about 30 per cent of the total rural
1972-73
.121.9
122.9
population and the remaining 70 per cent
1973-74
148.7
151.6
(which include a large number of poor small
1977-78
174.6
168.6
farmers) may have quite a different con
1983
282.2
, 267.0
sumption pattern. And, therefore, for pover
ty analysis Minhas et al [1990] have ad
It is puzzling to note that the differences bet
vocated use of CPIMR which is derived on
the basis of consumption pattern of the mid ween the two indices are quite large for the
where
recent years. The calculations performed at
dle income range in the base period. The
the State level showed even larger differences.
m Pl,
argument that the' CPIAL is inappropriate
How do we explain these differences? Since
I(x) = 2 ----- w.,
because it is based on the consumption pat
tern of only 30 per cent of the rural house this is an important issue, it is worthwhile
to
understand in more detail how Minhas et
is the CPITR, wbeing equal to the average hold population is not that appealing. As
al [1990] have computed their new indices.
expenditure share of the ith commodity for a matter of fact, the CPIMR covers only
The CPIAL series is constructed on the
the entire rural population. It can be seen those households which belong to the middle
basis of 62 consumer items, of which 37
that mean of xR in (2) will be equal to x^, income range which comprise only 20 per
belong to the food group, four to fuel and
only if we assume that w^x) is equal to vy cent of the total rural household population.
light group, 11 to clothing, bedding and foot
for all i. Thus, TJ’s suggestion to compute As such a large proportion of the poor is
ware
group and 10 to miscellaneous goods
real PCHE by deflating by CPITR implies not covered by this index.
and services group. In the construction of
that the consumption patterns of all
Gaiha [1990] argues that CPIAL is the
CPITR and CPIMR, 62 consumer items
households must be the same, in which case most appropriate deflator for measuring
were aggregated into 49 items (37 food items,
one deflator is good enough direct con poverty. He makes two points in its favour.
10 miscellaneous and two item groups, viz,
tradiction to their suggestion to use two First, “agricultural labour households
fuel and light and clothing, bedding and
deflators.
(ALH) are the largest occupational group
footware).
As we do not know why and how
We have so far examined whether or not among the rural poor; not only are the bulk
this aggregation was done, it will be difficult
it is appropriate to. use the recently developed of them poor but they also account for a
to evaluate its effect on the price indices.
two price deflators. We now address , the large share of the rural poor”. Second, since
question whether it is at all necessary to use “ALH are typically net buyers of food, the . Since the consumption patterns of the 49
items were not available for the 1970-71 year,
more than one price deflator. It depends on CPIAL can be expected to provide a close
all these items were further aggregated into
whether or not the consumption patterns. approximation to the prices confronting the
differ significantly across the PCHE ranges. net buyers of food among the rural poor,' 13 major groups (nine relate , to food and
four to non-food). The prices for each of
If they do not, the use of one price deflator which would be much larger than the share
these major groups were computed using the
is justified. It is desirable to test this of ALH among the rural.poor”.
hypothesis but it is a major undertaking
The most attractive feature of CPITR and . consumption patterns observed in the 1960^61
year. The State-specific aggregate consumer
beyond the scope of the present study. For CPIMR is that they are based on the con
price indices were then constructed from the
tunately, /Minhas and Jain (forthcoming) sumption patterns observed in more recent
13 major group indices using the consump
have presented the price indices (CPIMR years (1970-71) than the CPIAL (1960-61).
tion patterns of households in 1970-71.
and CPITR) for the middle range and en The principal question therefore is: have the
tire rural population, respectively, which consumption patterns changed so drastically
The claim that the proposed indices
throw some light on the issue. The two in that CPIAL has become unusable? To
(CPITR and CPIMR) are the best ones
dices computed by them for the years answer this question, we again refer to the
because they are based on the latest represen
1972-73, 1973-74, 1977-78 and 1983 do not Minhas, Jain, Kansal and Saluja [1990]
tative consumption patterns as the weighting
seem to differ much as can be seen from paper which presents CPITR based on both
diagram is also not entirely valid because in
their following numerical results at the all the weighting diagrams, viz, 1960-61 and . fact the consumption patterns of households
India level:
1970-71. We present their results at the allof both years, 1960-61 and 1970-71, have
India level but patterns at the State level are
been used. Assuming that the within-group
CPIMR CPITR Per Cent quite similar:
consumption patterns are the same in the
Difference
1960-61 and 1970-71 periods, Minhas et al
1960-61 1970-71 Per Cent
[1990] make allowance for changes in the
121.8
-0.08
121.9
1971-73
Weighting Weighting Difference consumption patterns between the groups.
148.7
148.1
-0.40
1973-74
Diagram Diagram
TJ overlook the fact that this is a highly
176.5
-1.09
1977-78
174.6
283.5
-0.46
restrictive assumption. One would normally
282.2
1983
1972-73
121.5
121.1
-0.3
expect that the ^consumption patterns 'of
1973-74
147.6 . 146.7
-0.6
Although the differences are somewhat 1977-78
households within the groups would change
173.9
172.2
-1.0
larger at the State level, they are still not large 1983
more readily than between the groups.
274.8
270.1
-1.7
enough to alter the direction of our results
Because of changes in incomes, the substitu
or the broad conclusions about poverty and These results indicate that thfc price indices
tion of one food item for another food item
inequality. •
may be more prevalent than the substitution
do not vary significantly when the weighting
It would thus appear neither appropriate diagram for the more recent year is used.
of food for non-food. Since the withinnor necessary to use two deflators to con
group consumption pattern is assumed to be
This suggests that the differences in con
vert the nominal PCHE into the real PCHE. sumption patterns across different PCHE
the same in the two periods and the between-
Economic and Political -Weekly ■’ June 15, 1991
- 1483
group consumption pattern is expected to be
stable, it is not surprising to find that the
Minhas et al indices do not vary much when
calculated using the weighting diagrams-of
the two periods.
Because the Minhas et al indices are
computed on the basis of two weighting
diagrams, 1960-61 and 1970-71, their com
putation of CP1MR is also problematic.
Since the income ranges are constructed on
the basis of current prices, the households
in the middle range in the 1960-61 year may
not have the same level of welfare and con
sumption patterns as those in the middle
range of the 1970-71 year. This creates a
problem as to which households should be
selected in 1970-71 so that they are compati
ble with those selected in 1960-61. Moreover,
CPIAL is the only index available for the
most recent years 1986-87. for which the
poverty estimates are presented in our most
recent version of the paper [Kakwani and
Subbarao 1990a].
Before we respond to the next set of issues,
we recapitulate the discussion so far. Our use
of the CPIAL is perfectly justified; adoption
of alternative price deflators as suggested by
TJ, even disregarding the numerous concep
tual and empirical problems stated by us,
would have introduced fresh, unknown (in
deed, unknowable) biases into the poverty
and inequality estimates; the inferences
drawn by TJ (their Table 3) are consequently
untenable; there are no errors in our com
putations and the data problems, we believe,
are not serious enough to vitiate our con
clusions....
The Decomposition
This technique enables one to separate the
impact on poverty of changes in average
consumption and in its inequality. The pure
growth effect is measured as the per cent
growth in poverty if the mean PCHE were
to change but the Lorenz curve remained un
changed. Similarly, the inequality effect is
measured as the per cent change in poverty
if the Lorenz curve were to change but the.
mean PCHE remained constant. Since the
poverty measures are non-linear, the total
percentage change in poverty will not be
equal to the sum of growth and inequality
effects. In the EPW version, thej-esidual
(usually called the interaction term) was
combined with the inequality effect. In the
recent revised version of our paper [Kakwani
and Subbarao, 1990a], we have separated the
interaction term from the inequality effect.1
Again, the direction of the results remained
unchanged.
TJ are at pains to drive home the point
that this methodology does not permit one
to establish causal relationships} It is quite
obvious that decomposition is an exact
mathematical relationship and, therefore,
cannot be used for establishing causality. We
have ourselves pointed out this in our paper.
Nonetheless, the methodology enables one
to draw interesting inferences. For example,
the inequality effect tells us whether or not
growth in the economy is accompanied by
1484
a redistribution of income in favour of the
poor. It should not, however, be interpreted
to mean that growth is causing a redistribu
tion of income. Even if we do not know what
is causing the redistribution, it is nonetheless
important to know whether or not growth
is accompanied by a redistribution. For
example, if the relationship between growth
rate and inequality effect is found to be
significantly positive, we may conclude that
the growth process tends to benefit the rich
proportionately more than the poor, even if
the causality is not known. Since the rela
tionship between growth and inequality is
expected to be non-linear (Kuznets 1955), the
correlation coefficient which measures the
deviations from the linearity may invariably
show that the variables are either not related
or weakly related. In such situations, the best
procedure is to use the rank transformations
which have been found to be robust and
powerful. In this section, we present results
on Spearman’s rank correlation coefficient
to test whether there exists an association
between the two variables. The test statistic
q being the rank correlation is distributed
approximately as student’s t-distribution
with (n-2) degrees of freedom? This'approxi
mation suggested by Pitman {1937] has been
shown to perform better than the usual nor
mal approximation (Iman' and Conover
1978). Table 1 presents the correlation co
efficients between growth rate and inequality
effect and also between growth rate and total
poverty effect. The student t value is calcu
lated using 14 degrees of freedom.
The correlations in Table 1 suggest that
the association between the growth rate and
the inequality effect is positive and signifi
cant only in the first period (1973-74 to
1977-78). The relationship became insignifi
cant during the subsequent periods. Thus,
the adverse effect of.jncome redistribution
on poverty is supported only in the first
period but not in the-two subsequent
periods. Although the factors that may have
contributed to the virtual absence of this
relationship in the subsequent periods are
difficult to establish empirically, it is at the
same time difficult to dismiss entirely the
role of anti-poverty interventions introduc
ed in this period especially because the scale
and delivery of this effort stood in contrast
to the piece-meal efforts of the past.
The correlation between the growth rate
and the total percentage change in poverty
was found to be negative and significant
during all the three periods (with one Excep
tion being the poverty-gap ratio for the ultra
poor during the 1973-74 to 1977-78 period
when the correlation was not significant at
5 per cent level). It implies that a positive
growth generally tended to reduce the pover
ty. Although these observations tend support *
to the ‘trickle-down’ hypothesis, anti-poverty
interventions may have played an important
role in suppressing the adverse effects of
•income redistribution which might have
resulted because of the growth. The impact
of the growth without anti-poverty interven
tions would have been smaller as we Observed during the 1973-74 to 1977-78 period.
TJ went to such a great length just to
point Out that the growth effect is more
dominant than thp inequality effect. This is .
quite obvious from our tables. An impor
tant observation made in our paper was that
th^inequality effect was generally positive
in the 1973-74 to 1977-78 period but it
became generally negative in the subsequent
period. Thus, in the first period the
redistribution of income increased poverty
whereas, in the second period it reduced it.
. Consequently, the proportional reduction in
•jxiverty was greater in the second period
than in the .first period despite the lower
growth rate in the second period. It is'not
Growth Rate arid Tbtal Poverty
Effect______ _
Correlation
t-value
Growth Rate and Inequality
_________Effect
Correlation
t-value
038
-0.02
0.04
. Head-count ratio poor ...
3.2*
-0.68
■ -0.75
1.4
-0.62
0.1
Head-cout ratio ultra poor
-0.59
2.9’
. -0.83
-0.1
. -0.56
02
2.9*
—6.1*
-2.7’
1973-74 to 1977-78
1977-78 to 1983
1983 to 1986-87
0.66
0.13
-0.01
Poverty gap ratio poor
33«
-0.57
-0.79
; 0.5
—0.57
-0.1
-2.8*
-5.1*
— 2.8*
1973-74 to 1977-78
1977-78 to 1983
1983 to 1986-87
0.66
0.05
-0.06
Poverty gap ratio ultra poor
-0.39
3.5*
-0.68
0.2
-0.53
-0.3
-1.7
-3.7*
— 2.5’
1973-74 to 1977-78
1977-78 to 1983
1983 to 1986-87
0.62
0.33
-0.03
1973-74 to 1977-78
1977-78 to 1983
1983 to 1986-87
—3.7*
-4.5*
—3.1*
Note: Asterisk indicates that the correlation is significant at 5 per cent level.
. Economic and Political Weekly
... . . •
.
June 15, .1991
‘
*
>
;
j
•
j
J
j
I
J
j
Table 1: Rank Correlations between Growth Rates in Per Capita Household Expenditure
and Inequality and Total Poverty Effects .
Period
j
-J
clear from TJ’s writing why these conclu
sions do not follow even descriptively from
a careful examination of our tables. They
repeatedly make criticisms without providing
explanation.
The decomposition proposed in the paper
has important implications for the ‘trickledown’ mechanism which is widely talked
about by economists. Unfortunately, TJ
dismiss this approach by calling it descrip
tive or non-causal. They further assert that
there are interpretational problems with the
decomposition of which we are not aware.
The authors make such criticisms without
providing any explanation. The only point
on decomposition methodology raised by TJ
is that depending upon in which year (base
or terminal) Lorenz curve is kept unchanged
the results can have very different interpreta
tional implications. This is a valid point. The
choice of the year is abritrary. We chose the
base year Lorenz curve because we felt that
this was a natural choice. If TJ provided a
rationale for selecting the terminal year
Lorenz curve, we would, of course, adopt it
but we believe that there exists no such ra
tionale. It might interest EPW readers to
note that the same authors have written a
paper (June 1990 after our paper was pub
lished) using the same decomposition.2
Growth and Inequality Elasticities
TJ argue that the growth and inequality
elasticities estimated by us are of no value
in the absence of an explicitly formulated
economic model incorporating the mecha
nisms and processes connecting poverty,
mean PCHE and relative inequality. We
ijever implied that these elasticities would
provide any clues to causality. Our aim in
computing these elasticities was more
modest, viz, to examine whether or not there
is a temporal tendency for the poverty ratio
to exhibit greater (or lesser) responsiveness
to changes in growth and inequality. The
question ‘how’ growth was impinging on
poverty was never asked. Of course, it would
be nice to formulate an economic model
which incorporates all the processes concer
ning poverty but such a model has to be a
fully blown general equilibrium model en
compassing all the sectors of the Indian
economy and their linkages to the foreign
sector. We invite TJ to attempt such a model
and throw light on causality.
In this context, TJ also point out that the
elasticities are point elasticities and,
therefore, are conditional upon the points
of evaluation which should be kept constant
in order to detect true inter-temporal
changes. It seems that they have completely
missed the meaning of these elasticities. It
is true that the Engel elasticities are com
puted at a given point because they are
generally not invariant to the points at which
they are evaluated. TJ are extending the same
idea to the growth and inequality elasticities.
These elasticities are computed from given
income or expenditure distributions. They
are in fact fixed for a given distribution. Any
provided the methodology which can en
compass alternative assumptions. It is not
clear whether TJ are criticising our assump
tion or the methodology itself. All economic
models are constructed on the basis of some
assumptions and, therefore, the conclusions
emerging from them are never unambiguous.
Thus their criticisms relating to our elasticity
calculations are unnecessary.
The final point on elasticities made by TJ
is that growth and inequality elasticity can
not be related to growth and inequality ef
fects. (This is correct but we have not made
any attempt to relate the two. Elasticities
measure the responsiveness whereas effects
measure the actual change due to growth
and income redistribution. Elasticities are
computed on the basis of expenditure distri
bution for one year only whereas to com
pute growth and inequality effects we require
expenditure distributions in two periods.
Both these concepts convey quite different
information about the characteristics of
poverty. The elasticities are useful for
simulating the effect of alternative policies.
But it is equally useful to explain the actual
changes in poverty. It is puzzling to know
what point TJ are making by alleging that
we are relating the two concepts. On the
basis of the magnitudes of inequality elasti
cities, we do make two statements: (1) if the
inequality deteriorates during the course of
economic growth poverty may increase ever.
with a faster economic growth because of
temporal change in them reflects the changes
taking place in the income distributions.
It makes little sense to fix a point of
evaluation.
As pointed out in our paper, the computa
tion of poverty elasticity with respect to
inequality is difficult because keeping per
capita consumption constant, inequality in
distribution can change in infinite ways. To
compute this elasticity we need to make an
assumption as to how inequality is changing;
for instance, whether inequality is increasing
by decreasing the share of the poor or in
creasing the share of the rich. In our paper,
we have clearly stated that inequality
elasticities are based on a proportional shift
in the Lorenz curve and, therefore, are not
unambiguous. This assumption may not be
acceptable but one can compute these elasti
cities with alternative assumptions. We have
—Published §
A GIAN - PT1 PUBLICATION
Forty Years of World’s Largest Democracy
A Survey of Indian Elections
G.N.S. Raghavan & G. Balachandran
Based on factual data on electors, voters and fortunes of all
India and regional parties in the elections held during the last
forty years for Parliament and state Assemblies.’Complete
with party manifestos, main issues, tables, charts and maps.
Also photographs.
130pp
Ulus
23cm
Rs. 85
GIAN PUBLISHING HOUSE
4C, Ansari Road. Oaryaganj. New Delhi-110002 Phone: 3261060
Economic and Political Weekly
June 15. 1991
1485
increasing inequality elasticity; and (2) both
growth and inequality elasticities are con
siderably higher for the ultra-poor than for
the poor implying that increasing inequality
will hurt ultra-poor more than the poor.
Then TJ go on to say, “both the statements
imply general reduction in inequality and
not a particular characterisation used in B”.
This statement is based on the mistaken im
pression that we related inequality effects
and elasticities. We made no effort to esta
blish causality between growth, inequality
and poverty. It seems that TJ have complete
ly misinterpreted what has been said in our
paper.
Data-Related Problems
TJ raised two other data-related points.
They argue that 1977-78 NSS poses problems
as it reports unusually high proportion of
expenditures on durables at the top openended class interval. We have no access to
their all-India analysis of this problem; we
did, however, look at this problem at the
State-level and found it relevant only for two
States, Maharashtra and Rajasthan, and
made appropriate adjustments. It is worth
stressing that a judgment on how serious is
this problem is essentially a subjective judg
ment. We believe the problem is more serious
in the two States mentioned. The expen
diture figures on durables in the other States
did not lead us to suspect their accuracy. TJ,
however, are right in pointing out that the
adjustments made for Maharashtra and Ra
jasthan in real mean PCHE have not been
carried over to our decomposition exercises
for period I. This was done in our revised,
recent version of the paper [Kakwani and
Subbarao, 1990a], and we found no change
in the direction of results.
TJ also point out that we have made some
calculation errors in computing PCHE. To
clarify our position, we only need to repeat
the procedure we adopted in computing all
India PCHE. We have adjusted the expen
ditures in current prices for each State for
State-wise price differences (both over time
and across States) and then aggregated to
derive all-India real PCHE estimates. Our
estimates, therefore, need not correspond
with .those all-India PCHE estimates derived
on the assumption that the price levels in dif
ferent States are the same each year. More
over, our all-India estimates of the real
PCHE are based on the average for 15 States
whereas the published all-India tables are
averages for more than 15 States. Thus, the
observed differences in the numerical results
are to be expected owing to. differences in
methodology and coverage; they do not
reflect lack of care on our part but a lack
of understanding on the part of TJ about
the procedure we used in estimating all-India
PCHE.
the hypothesis whether or not there exists
a significant association between the varia
bles. If the association is found to be
statistically insignificant, it would most like
ly imply a non-existence of causal relation
ship. But if the association is statistically
significant, it would only mean monotonici
ty in the relationship between the variables.
One would then require further investigation
to establish causality. This task could have
been accomplished by using correlation
coefficients. But we used regressions because
they immediately provided us with the
t-values.
Other Issues
Finally, TJ object to our use of two-point
comparisons. It is well known that NSS is
now available only quinquennially and one
can only compare five-year periods so that,
contrary to TJ’s hope, there is no way one
can get a complete time series for the 1970s
and the 1980s. TJs references to the problems
created by the dance of the monsoons is a
real one; but we disagree with their view that
the years chosen by us are “exceptional
years”, especially when analysis is conducted
at a disaggregated State-level; we also do not
see much logic in ignoring observations on
the basis of such factors as the rate of in
flation. It is worth stressing no two years can
ever be “identical years” from the viewpoint
of sectoral, macro, fiscal1-and monetary
angles even if data were available on a time
series basis; every year has some year-specific
factors associated with it. . Variation is the
.reality and it is the analysts’ job to explain
this variation, of course with due regard to
exceptional extraneous occurrences (such as
the drought of 1987). We believe there was
no such exceptional occurrence in the chosen
years; we hasten to add that it is our per
sonal judgment. Moreover, purists might
argue—and TJ may be in sympathy with
them—that we need at least 20 observations,
and given five yearly surveys, we need to wait
for a hundred years to do any meaningful
analysis of poverty “trends” in India. In that
sense, we have no hesitation in saying that
we do not belong to the category of purists.
The above discussion shows that there was
nothing wrong with our methodology and
the deflator chosen; causality was never im
plied (let alone established) by us; the
qualitative conclusions of our paper never
theless hold good; that there were no errors
in our estimation, and that TJ’s suggestion
of haste in our publication is without any
basis.
"Notes
1 We have greatly benefited from discussions
with Martin Ravallion who suggested to us
to separate the interaction term from the in
equality effect. For an excellent discussion
of the decomposition see Datt and Ravallion
REGRESSIONS
[1990],
It is quite obvious from even a casual 2 The decomposition proposed by TJ makes
reading of our paper that regression relation
little intuitive sense. It is beyond the scope
of the present note to provide a detailed criti
ships were not estimated to establish causali
que of it.
ty. Our purpose was again limited to testing 4
1486
References
Bardhan, P K (1970), ‘On the Minimum Level
of Living and the Rural Poor’, Indian
Economic Review, Vol 5, pp 129-36.
— (1973), ‘On the Incidence of Poverty in Rural
India in the Sixties’, Economic and Political
Weekly, Annual Number.
Datt, G and M Ravallion (1990), ‘Growth and
Redistribution of Changes in Poverty
Measures: A Decomposition with Applica
tions to Brazil and India in the 1980s’,
mimeographed paper, The World Bank,
Washington, DC.
Gaiha, R (1990), ‘Poverty Alleviation Program
mes in Rural India: An Assessment’,
mimeographed paper, Faculty of Manage• ment Studies, University of Delhi.
Iman, R L and W J Conover (1978), ‘Approxi
mation of the Critical Region for Spear
man’s Rho With and Without Ties Present’,
Communications in Statistics, series B, 7,
pp 269-82.
. Jain, L R, K Sundaram and S D Tendulkar
,
(1988)
‘Dimensions of Rural Poverty: An
Inter Regional Profile’, EPW, Special
Number (November), pp 2395-408.
Jain, L R and Suresh D Tendulkar (1989),
‘Inter-Temporal and Inter-Fractile Group
Movements in Real Levels of Living for
Rural and Urban Population of India:
1970-71 to 1983’, Journal of the Indian
School of Political Economy, Vol 1, No 2
(July-December), pp 313-34.
Jain, L R an^S D Tendulkar (1990), ‘Role of
Growth and Distribution in the Observed
Change in Head-Count Ratio Measure of
Poverty: A Decomposition Exercise for
India’, Technical Report No 9004 (June),
Indian Statistical Institute, Delhi Centre.
Kakwani, N and N Podder (1976), ‘Efficient
Estimation of the Lorenz Curve and
Associated Inequality Measures from
Grouped Observations’, Econometrica,
Vol 44, pp 137-48.
Kakwani, N (1980), Income Inequality and
Poverty: Methods ofEstimation and Policy
Applications, New York: Oxford University
Press.
Kakwani, N ^nd K Subbarao (1990), ‘Rural
Poverty and Its Alleviation in India’, EPW,
March 31, 1990, pp A2-A16.
— (1990a), ‘Rural Poverty in India 1973-86’,
PRE Working Paper No 526, The Workd
Bank, Washington, DC.
Minhas, B S (1970), ‘Rural Poverty, Land
Distribution and Development’, Indian
Economic Review.
Minhas, B S, L R Jain, S M Kansal and M R
Saluja (1987), ‘On the Choice of Appro
priate Consumer Price Indices and Data
Sets for Estimating the Incidence of Poverty.
in India’, Indian Economic Review, Vol 12,
pp 19-49.
— (1990), ‘Cost of Living in Rural India:
1970-71 to 1983, Statewise and All India’,
Indian Economic Review, Vol XXV, No 1,
January-June 4. _
Pitman, E J G (1937), ‘Significance Tests which
May Be Applied to Samples from Any
Population: The Correlation Coefficient
Test’, Journal of the Royal Statistical Socie
ty, Supplement 4, pp 225-32.
Tendulkar, S and L R Jain (1990), ‘Rural Pover
ty and Its Alleviation in India: A Critical
Scrutiny’, EPW, September 22, pp 2165-68.
Economic and Political Wwk.Jj
June 15, 1991
ea that society
ite various inid secular, is a
ccular instiiueligions withso on an equal
perhaps also
vish to secure
istitulions will
on. A univer;r institution if
o matter how
nt of religious
late will cease
to carry every
its shoulders.
for reading th'e
publication My
for drawing my
Act of 1922.]
ism and Equa. Vol 27. No 2.
d Equality. Re
Vol 28, No 5.
n the Compara■Wirica! Weekly,
for Sociology’.
<Vol27.No35.
w’ch.v. Galli mard.
nard. Paris.
•ary Forms ofthe
*'in, London.
zones of Primi-
s. Oxford.
ami Fall of the
London.
ory ofReligion,
inford.
.tiation of Sociks. New York.
n its Place’, The
/ol 46. No 4.
sm’. The Tinies
:ience and Re-
auvenir Press,
of India, Asia
ztu’eandFunc-
hen and West,
•nge in Modem
i lYess, Berke-
■>hy’. The Tunes
the Decline of
pjnds worth.
y of Religion,
SCs and STs in Eastern India
Inequality and Poverty Estimates
Mridul Saggar
Indranil Pan
This paper employs inequality and poverty measures to consumption expenditure data for SCs, STs, and other
households in four eastern states published by NSSO. Inequality and poverty differences among SCs, STs and other
households, rural-urban disparities and inter-state variations in these respects are considered. The eastern region as
a whole suffers from economic backwardness and large incidence ofpoverty and this is specially true for SCs and STs.
Rural-urban disparities in consumption also exist with the urban sector better-off than (he rural sector. Also, the SCs
were found to be comparatively better-off than the STs. However, the generally held view that large inequalities exist
among SCs and STs is ill-founded. State level comparison shows Assam to be performing better than the others, with
lower poverty levels and more egalitarian distribution of consumption.
I
Introduction
THIS paper employs inequality and poverty
measures to consumption expenditure data
for SCs, STs and oilier households in four
eastern slates, viz, Assam, Bihar, Orissa and
West Bengal. Consumption expenditure
pattern is analysed to examine: (i) inequal
ity and poverty differences among SCs, STs
and other households, (ii) rural-urban dis
parities. and (iii) inter-state variations. Data
generated by the 38th Round of National
Sample Survey (NSS) is used for this pur
pose.
While studies abound on the consumption
inequalities among rural and urban house
holds for various expenditure classes, little
effort has been made to compare the con
sumption levels of SCs and STs with those
of other households in both urban and rural
sectors. To our knowledge, the only study
on the levels of living of SCs and STs is by
Vijay Nayak and Sailaja Prasad (1984).
Using ungrouped NSS data for the 28th and
32nd Rounds, they compare the levels of
living, inequalities in the levels of living and
disparities in the levels of education and
occupational structure of SCs/STs vis-a-vis
the non-SCs/STs (i e, other households) in
Karnataka during 1973-74 and 1977-78.
They conclude that the SCs/STs have a
lower standard of living than the non-SCs/
STs. Though there has been a fall in the
standards of living in real terms for both
SCs/SIs and non-SCs/STs over the two
Periods, the SCs/STs seem to have suffered
H’a’re. It is also found that inequality is less
u>lhin SCs/STs as compared to non-SCs/
Inequality within all groups has also
^creased, with the SCs/STs registering a
'harper rise than (he non-SCs/STs. Inequal^'iihinall groups has also increased, with
t'<c S(.s/STs registering a sharper rise than
!e n°n-SCs/STs. .Such analysis can provide
seiul input for framing development
^"grammes in the country. Unfortunately,
^■On°mic and Political Weekly
larch 5, 1994
ungrouped data arc not published by the
National Sample Survey Organisation
(NSSO), noris the age or class composition
of its sample available. Only a select few
have access to original NSS tapes and per
haps this has dissuaded researchers to work
in this area. However, information on con
sumption of SCs. STs and all households
generated by the 38th Round was published
by the NSSO in Sarvekshana (1989) in the
form of grouped data.
The study is presented under the follow
ing sections. Section II gives information on
the data set and the choice of methodology.
Section III gives some preliminary observa
tions and Section IV presents the results on
disparities in the consumption levels. Sec
tion V covers findings on poverty indices.
Section VI concludes the study by
summarising the findings, makes brief com
ments on policy aspects and implores for
more research in the area.
Il
Data Source and Methodology
West Bengal. However, data for other house
holds for all the slates considered were not
directly obtained from the publication. These
were derived as a residual from the data for
all households after taking out the effects for
SC and ST population through the use of
proportion of SCs and STs in total popula
tion as published by the NSSO in
Sarvekshana (1989). Also, data for urban
STs was available only for Orissa. The
survey captured expenditure on domestic
account, including consumption out of non
marketed self products, gifts and transfers
but excluded transfer payments, expendi
ture on residential housing and expenditure
on household enterprises. The sampling
design for both rural and urban sector was a
two-stage stratified sampling1. While the
NSS sampling design is based on scientific
principles despite not being random, it nev
ertheless suffers from some limitations.
These limitations are, however, not dis
cussed here as they have been already widely
documented in the literature2.
Methodology
Data Source
Inequality Measures
Household budget surveys have tradition
ally provided enormous insight into the
problem of inequalities, particularly in re
spect of income and consumption. Contem
porary literature on the subject has increas
ingly relied on longitudinal studies. In the
absence of institutional support in India for
panel data, longitudinal studies are scanty
and no generalisations are possible to sup
port macro policy changes. As such,’this
study also lakes recourse to the NSS data.
The paper uses estimates of the 38th Round
of the NSS which relate to the period De
cember 1982 to December 1983. The refer
ence period was 30 days prior to the enquiry
conducted during calendar year 1983. We
analysed the data for both rural and urban
SCs, Si's, and other households for the four
eastern states—Assam. Bihar, Orissa and
March 5. 1994
Consumption inequalities in the four east
ern states is captured in the study by em
ploying the following alternative measures:
coefficient of variation, Gini coefficient,
Atkinson’s index and Theil’s entropy mea
sure. The choice of these measures was
influenced by the properties of these indi
ces. All the above mentioned indices satisfy
the three basic properties that an inequality
index should ideally possess, viz, (1) mean
or scale independence: the index should be
in vari ant to scaling up or down of everyone’s
income by a constant proportion; (2) popu
lation size independence: index should be
invariant to scaling up or down of popula
tion at each income class by a constant
proportion; and (3) Pigou-Dalton criteria:
index value should decline on any transfer
567
from richer to poorer person that does not
disturb the relative ranks of these persons3.
The geometric definition of thcGini coef
ficient is used in this paper. It is given by
G= 1 -[I(FrJ-F) (Q...+QJ
where F = In /nQ is the cumulative popula
tion share in each class. n‘ being the cumu
lative population up to the ith class and n0
being the total population. Qj = Inx/noc,
where, x is the mean consumption expendi
ture of the ith class and e is the mean
consumption expenditure over the entire
distribution. The data is of course arranged
in ascending order of per capita expenditure
for the whole distribution for computing F
and Q. Alternatively, other methods devel
oped by Kendall and Stuart (1963), Rao
(1969). Sen (1973) and Fci and Ranis (1974)
may be used to compute the Gini ratio’.
However, all these measures can be shown
to be equivalent to the geometric definition
a^s demonstrated in Sudhir Anand (1980,
A1-316).
1 heil’s entropy index T is defined as
T = |l[It]T|+[E^- log (Y./Y) (n/n) |
where Y. denotes the mean expenditure in
the ith class and Y is the mean expenditure
of the entire population. The first term cap
tures the within group inequality and the
second term captures the between group
inequality. Since the grouped data from
NSS docs ndt permit us to capture within
group inequalities, our computations of
Theil’s index was reduced to computing
only the second term in the above expres
sion.
Atkinson’s index (1970) is of particular
interest to economists as it is normative in
nature and based on social welfare evalua
tion of income distribution y = (yr y,....... .yn)
among n individuals. The Atkinson's Index,
I is defined as
where Y^E is equally distributed expendi
ture obtained from a given social welfare
function. It gives the level of per capita
expenditure which, if equally distributed
would give same social welfare as at present.
p denotes the average for the whole popula
tion. The formula can be translated for com
putational convenience as
where E represents the inequality aversion
■parameter. Higher the value of E. more
568
averse the society is to consumption in
equalities. Since I is a normative index,
the choice of E depends on value judge
ment. Atkinson (1970) uses 1.0 < E < 2.5
in the empirical sections of his paper.
Stern (1977) presented a number of argu
ments in support of £ values between 1.5
and 2 5.. Wc computed the index for E
values of .5, 1.0, 1.5, 2.0, 2.5 so as to
provide a thorough picture.
Poverty
By the very nature of the word ’pov
erty’, normative clement creeps into its
measurement. The concept of poverty de
pends upon some absolute or relative no
tion of who is poor. Traditionally, the
absolute poverty concept has been adopted
in India in which absolute minimum stan
dards of living arc fixed in terms of income/consumption expenditure translated
from nutritional requirements measured
in calories5. The Dandckar-Sukhatmecontroversy in this regard is well entrenched
in Indian economy literature. Here, wc
only wish to attract the readers’ attention
to the vexed issues relating to the choice of
appropriate consumer price indices (CPI)
for estimating the incidence of poverty,
and recommend Minhas et al (1986) as a
ready reference. The problem was com
pounded in our case as for analysing pov
erty in select stales, the all-India price
indices or poverty lines were not represen
tative of these states. Therefore, we found
it more appropriate to use statewise pov
erty lines for rural and urban areas for the
year 1983, computed by Jain andTcnduIkar
(1990). The authors had updated the pov
erty linedefined by Planning Commission
for 1962 by using a middle-range CPI at
all-India level and then adjusted for the
differential in prices in a given state rela
tive to all-India. The poverty lines in terms
of mean per capita expenditure (MPCE)
for the country and the four eastern states
in 1983 arc shown in Table 1.
Using the above poverty lines, two al
ternative measures of poverty were com
puted for the four eastern states—the head
count ratio (H) and a variant of the FostcrGrcer-Thorbcck (henceforth FGT) index.
The former was chosen for its simplicity
and the latter for its properties. The head
count ratio is defined as H=n./N where n is
the number of poor among the total popu
lation, N. II measures the incidence of
poverty but docs not quantify the aggre
gate income needed to lift all the poor to
the level of poverty line.
In view of the limitations of the head
count ratio, Amartya Sen (1976) proposed
that poverty measures should satisfy the
monotonicity axiom, the transfer axiom and
the transfer sensitivity axiom. Wcopt to use
the FGT index in our poverty analysis for the
eastern states, as this is one index whid)
satisfies Sen’s three basic axioms as well as
meets the decomposability property. The
index is defined as
where■'Z is the poverty line, Y. is the
income of the ith person and a is the
poverty aversion parameter. A larger a
gives greater emphasis to the poorest poor.
The computation of the FGT is not a
straightforward affair in the Indian con
text, as the published NSS data gives the
expenditure distribution in the form of
grouped data. Instead, we chose to use the
FGT variant developed by Suryanarayana
and Geeta (1992), which derives an ex-*
pression for Pa assuming income distribu
tion to be lognormal with two parameters,
mean, 0 and the standard deviation, yof
log variable. The Pa measure so defined is
applicable to grouped data and takes the
form
Table 1: Statewise Poverty Lines in Terms
of MPCE at 1983 Prices
(Rs per month)
Rural
Urban
Assam
Bihar
Orissa
West
Bengal
103.50
96.23
105.33
116.81
103.53
129.94
109.69
100.12
Table 2A: Mean, Median and Mode of
Monthly Per Capita Expenditures by
Household Groups for the States
in Different Sectors
Assam
Rural SC
Rural ST
Rural others
Urban SC
Urban ST
Urban others
Bihar
Rural SC
Rural ST
Rural others
Urban SC
Urban ST
Urban others
Orissa
Rural SC
Rural ST
Rural others
Urban SC
Urban ST
Urban others
West Bengal
Rural SC
Rural ST
Rural others
Urban SC
Urban ST
Urban others
Mean
Median
Mode
111.08
108.78
114.02
113.73
—
167.08
103.8
100.14
105.64
105.56
—
144.29
89.25
82.87
88.87
89.22
—
98.70
76.38
80.77
100.84
121.21
—
142.08
68.47
70.23
88.66
95.19
—
119.27
52.62
49.15
64.30
43.15
—
73.66
87.31
72.71
111.60
115.75
111.84
164.60
79.27
65.64
97.74
106.63
95.50
136.92
63.20
51.50
70.02
88.38
62.83
81.57
94.99
85.26
111.42
133.07
—
175.96
82.33
76.93
95.76
109.47
—
139.01
57.01
60.27
64.44
62.27
—
65.12
Economic and Political Weekly
March 5, 1994
-(f)
G(——y—-^2 )
+ a j 2“! e:,: '2’. G (lnZ ~ (8 + 2f) j +
- '
Y
+ (_!)'(“ ) Z-'/’‘'’’<'.G d!^2l£ +!X>)
r
V
+ ...]
where r is any rational number, reflecting
the choice of a greater than or equal to 0.
We computed the Pa for a = 0, 1, 2, 3. For
a > 1, the monotonicity and transfer axioms
are satisfied, while for a > 2, transfer sensi
tivity axiom is also satisfied. Il is also
interesting to note that as a increases. Pa
decreases and for increasingly large values
of a, Pc tends to zero as it gives an overriding
weightage to the transfer sensitivity axiom.
Thus, it was considered appropriate to
compute the measure only for some re
stricted values of cc terminating at 3 at the
upper end.
Ill
Preliminary Observations
Before analysing the results obtained from
a wide array of measures employed from
among the ones cited in the preceding sec
tion on methodology, a few preliminary
observations about the consumption levels
by comparing the MPCE would be in order.
The averages for household groups in rural
and urban areas for die four states is pre
sented in Table 2A. Table 213 gives the index
numbers of MPCE in the four states, taking
the all-India MPCE as 100. Table 2C gives
a comparative picture of MPCE by house
hold groups in these slates, by taking the
MICE for all households in each state to be
100. Thus Table 2B gives us an across-slate
comparison of average MPCE and Table 2C
gives us an across-houschold comparison of
average MICE in each of the four states.
1 able 2D presents the MPCEs in rural areas
as a percentage of MPCEs in the urban
areas, to facilitate finding out the ruralurban divide in each state.
llic following observations emanate from
Tables 2A, 213, 2C and 2D :
II) Hie MICE of SCs and STs is consider
ably less than that of other households for all
c states in both rural and urban areas
I ables 2A and 2C). Though this appears to
a trivial finding that conforms to general
cxpcctatians it nevertheless is central to the
issue as it establishes that SCs and
f<>
CC(>nomically weaker and, there•°re’ there is a prima facie case for policy
nitrvcinion to improve their consumption
5landards.
cc\^°ra^ Assam constitutes a striking expl,“n to the first observation as the MPC ’!•
Eco
"°mic and Political Weekly
of SCs and STs is only marginally below
is concentrated in the Santhal Parganas and
thabof the other households in this region. north and south Chotanagpur district. Of the
In terms'aT the Index numbers of MPCE 193 blocks in this region, sub-plans for
with respect to MICE for all households tribal area development were operational, in
of Assam, the average consumption of early 80s. in 112 blocks. These were the
SCs is only 1.7 per cenl below the state blocks in which ST population exceeded 50
average and that of STs is only 3.8 percent per cent of the total population. As part of
lower, while that of non-SC/ST household the sub-plans, the blocks were grouped un
is merely 0.9 per cent above the stale der 14 Integrated Tribal Development
average (Table 2C). This clearly testifies Programmes and provisions were made for
to near absence of inter-class disparities in agriculture, animal husbandry, milk, educa
rural Assam. Il may be pointed out here tion. health, etc, and steps were undertaken
that assistance of term lending institutions to s t re ng th c n ad m i n is tra t i ve struclu re. S i nee
and banks did not reach the desired levels
1976-77, separate provision was made in
due to low absorption capacity of the budget of the slate government for these
region; the credit-deposit ratio for Assam sub-plans with a share in state expenditure
was as low as 46.8 percent in March 1982.
ranging from 13 to 19 per cenl of the total
As for the public expenditures, though state plan. Supplementary policy actions
tribal sub-plans and plans for SCs were in were also undertaken in the form of special
vogue, no major thrust was made in the allocations for STs under education, hous
direction, Therefore, we found no ready ing, etc, and launching of the Mada
explanations for low inter-class dispari programme for socio-economic develop
ties. To an extent the answers would lie in ment of STs
the realm of sociological structure but we (5)
1
he rural-urban divide in consumption
would not wish to speculate We would standards exist (Table 2D). The dispari
rather welcome supportive research effort ties arc higher in case of West Bengal than
in this area from those well-versed with other eastern slates, both for SCs and
micro-level realities of the slate. Unfortu non-SCs, except in case of Bihar SCs. This
nately the 1981 census could not be held in is somewhat surprising as the political
Assam due to disturbed social conditions.
base for the ruling parly in West Bengal
As a result not enough social indicators comes largely from the rural areas. In
are available for the slate.
ca.se of Assam, the absolute consumption
(3) The MICE is marginally higher forSCs levels of SCs are same in rural and urban
compared to STs except in case of rural areas.
Bihar. This makes out a case for special (6) The MPCE of non-SC/ST households in
development programmes for regions in all four eastern states are below the all-India
habited by STs. The broad division of re
Table 2D; MPCE in Rural Areas as a
source allocation forpoverly alleviation and
Percentage of MPCE in Urban Areas
rural development should reflect some bias
for SC and ST in Different States
in favour of STs vis-a-vis SCs.
(4) Die average consumption level of STs in
Orissa West
Assam
Bihar
Bihar compares favourably to that of SCs
Bengal
( fables 2A, 2C). Illis inter alia reflects the
97.67
75.42
71.39
SC
63.02
effective targeting of special development
_
—
—
65.02
ST
programmes for STs. A large segment Others
67.80 63.32
68.24
70.98
(93 per cent) of the ST population in the state
Table 2B: Index Nos of MPCE by Household Groups for States in the Two Sectors wmi
Respect to All-India MPCE
Rural
ST
SC
Assam
Bihar
Orissa
West Bengal
All-India
111.77
80.99
92.57
100.72
100.00
124.81
92.68
83.43
97.83
100.00
SC
Urban
ST
Others
88 20
94.00
89.76
103.20
100.00
—
—
84.02
—
100.00
97.08
82.55
95.64
102.24
100 00
Others
94.69
83.74
92 68
92.53
100.00
Table 2C Index Nos of MPCE by Household Groups for Sta hls in the Two Sectors with
Respect to Staif MPCE for Ai t. Ilousiaiot ns
SC'
ST
Others
All
SC
Urban
Others
ST
All
98.27
81.47
89.56
90.81
96.24
86.14
74.59
81.51
100.88
107.55
1 14 49
106.52
100
70.87
86 84
76 48
78.30
—
—
73.89
—
100
100
100
100
Rural
Assam
Bihar
Orissa
West Bengal
March 5. 1994
100
100
100
104.12
101 79
108.75
103.51
569
Public Issue of 1,12,50,000 Equity Shares of Rs.10/each for cash at par aggregating to Rs, 1125.00 Lacs.
HIGHLIGHTS : n Company promoted by Akar Laminators Limited, profit making and
dividend paying Company u Ready market - 50% captive consumption by group
companies a Term loan assistance by Syndicate Bank- Company s Banker. ■ ideal plant
location, eligible for Income Tax & Sales Tax exemption, backward area benefits.
proximity to input sources & cosumer centres ■ Tax concession Sec. 80 L, 80 M and
Wealth tax benefits a Easy liquidity—listing at Ahmedabad, Bombay. Calcutta. Deihi and
Madras Stock Exchanges ■ Technical knowhow and machinery to be supplied by M/s.
Gahleo Vacuum Tec . Italy, who are noted lor high vacuum coatings and space
technology
Investors may note that in case of oversubscription, a SEBI nominated public
representative shall be associated in the process of finalisation of basls,of allotment.
RISK FACTORS AS PERCEIVED BY THE MANAGEMENT: ■ Competition from existing
and new units. However Company proposes to manufacture highly specialised grade of
metallised polyester and due to the promoters experience in this line as well as captive
consumption by group companies The Company is confident of facing the competition.
b Exchange rate fluctuations can adversely affect the project as a part of the machineries
are impodea However adequate provisions of contingency is made to take care of
exchange rate fluctuation. ■ The process does not generate any effluents However
application has been made for N O.C to Pollution Control Board Their clearance is
awaited
Investors are advised to refer to the contents of para on "Promises Vs Performance"
before making an investment in this Issue
Note : NRI Forms are available at the Registered Office of the Company, at the Lead
Managers Office and at Advisor (NRI) Indian Investment Centre, New Delhi.
LEAD MANAGERS TO THE ISSUE
Economic and Political Weekly
|l ft
’1$!
Canan Bink
(Mtrchanl Banking C»H)
Na.ayan Chambers.
A:h<am Read.
Ahm«;abad-3600C6
REGISTRAR TO The ISSUE
SrndkMl« Bank
(Marchanl BanUng Buraau)
Zonal OH.ce,
Neplune To»«rj. Ajhram Road
Anmedabad-360009
@
5 Santojh" Plot No M.
Su»amacur< Jeiaipcr Road
Barola-390 005
ADVISORS TO THE ISSUE
rpCTl Anagram Finance Llmilad
|7~)| (AMcmoerollhelalbhi. Group)
SOI. -SAKARNeniu B'.dge, Aahram Road,
Ahmeoaoad-MO 003
jtyf Ktynoti Cocpcxik
SarncwLki
Mia 403. Embassy Centre.
/T-V
Nauman Po<nL
Bombay-400 021
March 5, 1994
Regd. Office : Isl-Floor, Trade Centre. Opp. Stadium, Navrangpura. Ahmedabad-380 009
Admn. Office: 517, Maker Chambers V, Nariman Point, Bombay-400 021
ISSUE OPENS ON 7TH MARCH, 1994.
MINIMUM SUBSCRIPTION
IF ’HE COMPANY DOES NOT RECEIVE MINIMUM SUBSCP'PTION OF 30^ OF THE ISSUE AMOUNT !NClUD*G OtVOLVEMfNT OF
UNDERWRITERS WITHIN 120 DAYS FROM THE DATE OF OPENING OF THf <SSuE THE COMPANY Shah. REFUND ENTIRE SUBSCRIPTION
AMOUNTW1THIN128OAYS WiTHiNTERESTFOfiDEi*’0E’ONO7«OA’SfhOMTn£OATEOF ClOSURE OF Th£ ISSUE AS PER SECTION’3OF
THE COMPANIES ACT
I
Econoi
levels, except in case of urban West Bengal
(Table 2B). '1’he consumption standards are
strikingly low in case of Bihar with the
indices of 83.74 and 82.55 for rural and
urban areas, respectively, b'or SCs and
S Is. the consumption standards arc re
markably above the all-India levels for
rural Assam. However, it is worrisome to
note that MPCE in both rural and urban
areas of Bihar and Orissa compares poorly
with all-India levels for all categories.
1'he relative backwardness of these two
slates should L>e considered in shaping the
policies and perhaps the Planning Coin
mission and the Finance Commission need
to impart a still sharper edge to their
redistributive policies in an effort to strike
a regional balance in growth and develop
ment process.
IV
Inequality Estimates
We had earlier brought to fore the striking
result of MPCE of rural SCs and STs in
Assam being only marginally below that of
non-SC/ST rural population of Assam
(Table 2C). ’Hie inequality measures com
puted here (Tables 3A and 3B) bring to fore
an additional feature in case of SC/ST popu
lation of rural Assam. Not only are the
MPCE of SCs and STs of rural Assam in
tune with general rural population of the
state, but also their consumption expendi
ture is fairly equally distributed across the
SC/ST population. Tins is reflected by a
relatively low coefficient of variation of
0.33 in case of rural SCs and 0.37 in case of
rural STs. These coefficients arc lower than
the coefficient for all the categories for all
the states. This observation is valid even for
other measures of inequality. The Gini ra
tios, excluding the slate of Assam range
from a low of 0 23 in case of Bihar rural SCs
to a high of 0.33 in case of non-SC/ST urban
population of West Bengal. In comparison,
the Gini ratios for SCs and STs of rural
Assam were as low as 0.17. Theil and
Atkinson index also revealed similar results
in terms of inter-state and inter-class com
parisons. In contrast, there are strikingly
high inequalities in consumption expendi
tures in case of urban SCs of Bihar and nonSC/ST population of urban West Bengal.
Consistently, the inequality jnensures were
higher for these two categories than for any
other category of any other state. While the
high inequalities are a disconcerting feature
by nature, one can draw some comfort from
the 1 act that the MPCE for these two catego
ries is not very' low compared to the all-India
levels and this is reflected in the MPCE
indices in Tabic 2B.
•Some other striking observations cun also
be made in light of Tables 3A and 3B. First.
lheeconomic inequalities remain high in the
state of West Bengal despite the conscious
Economic and Political Week I \
endeavour for a designed state intervention
in favour of redistributive measures. In ab
sence of inter-temporal data on consump
tion pattern of SCs and STs a definitive
conclusion is not possible on whether the
inequalities are accentuating or going down
over time. However, from the available
information from the 38th Round it is axi
omatic that urban inequalities are the high
est from among the four eastern slates and
even the rural inequalities are high enough.
For most of the cases, the Gini, Theil and
Atkinson indices for SCs. STs and other
households in West Bengal are higher than
iu two other eastern states, viz, Bihar and
Orissa, which are known fortheir iniquitous
structure and class conflicts. A second strik
ing resul t is that while the rural-urban divide
exists in terms of higher MPCE for urban
areas, the divide works the other way round
when inequality measures are considered
and the urban areas suffer from distinctly
higher inequality. Both the exponents as
well as detractors of the inverted-U hypoth
esis would agree that in initi.d phasesol high
growth inequalities would accentuate. Ur
ban areas are now set on a higher growth
path than the rural areas and it is only natural
that they record a higher level of inequali
ties. We find that except for the case of
Orissa, the inequality measures are higher
for urban population of all the other states in
all categories compared to the rural seg
ments. Incidentally, the result also provides
a rationale for the prevailing urban bias in
raising tax revenue, but should not be inter
preted to negate the possibilities of taxing
agricultural sector altogether. On the other
hand, the higher inequalities in urban areas
should awaken policy-makers to the need
for implementing urban poverty alleviation
and employment generation programmes
with as much vigour as they do in case of
rural areas. Hie third and the Imai striking
result brought out by this section is that, in
general, the consumption inequalities arc
lower in case of SC and ST population as
compared to other households. Exceptions
to this rule are rural Orissa, where inequality
indices are more or less same across all the
three categories and urban Bihar where there
is disconcertingly high inequality among
SC population. The lower inequality among
SC/ST population in general is indeed an
important result that lends to provide a
rationale for slate intervention through
programmes designed specially for SC and
ST population. Such direc t targeting for SCs
and STs would have lost moral force in case
inequalities were high among SCs and STs,
in which case economic criteria and not the
caste criteria would be preferable for target
ing of development programmes. Our find
ings need not be interpreted to altogether
discount the possibility of a conjunction of
caste-cum-economiccriteria. But, certainly
the lower inequalities among SCs and STs
March 5, 1994
provide a rationale for targeting SC and ST
population as a group.
'1'he analysis till now was made at nn
aggregative level using the total per capita
ex pe nd i t u re across al 1 com mod i t i cs. 11 wou 1 d
be interesting to see how the consumption
profile shifts across classes and whether
expenditure pattern is more unequal for
some commodity groups compared to oth
ers. While far greater insight on this issue
can be drawn from estimates on Engel elas
ticities (which is not taken up in this paper),
we nevertheless feel that interesting obser
vations can be made by comparing the con
centration ratios for select commodity
groups. 'Iliese ratios are presented in sum
mary form in fable 4 A to Table 4D for four
major commodities—total cereals, milk and
milk products, fuel and light and clothing;
and for total expenditure on food and total
expenditure on non-food items.
It is notable from the tables that concen
tration ratios for expenditure on non-food
items is much higher than on food expen
diture and this observation is true for all
groups and all stales covered in the analy
sis and in several cases the former is more
than twice the latter. This result is hardly
surprising as it is in complete conformity
with consumer theory. Food items are a
necessity and should, therefore, have a
lower income and price elasticity. There
fore, expenditure on food items can be
expected to be more equal across expendi-
Table 3A: Measures of Inequality: Coefficient
of Variation, Gini Ratio and TheiCs
Entropy Index
Assam
Rural SC
Rural ST
Rural others
Urban SC
Urban ST
Urban others
Bihar
Rural SC
Rural SI’
Rural others
Urban SC
Urban ST
Urban others
Orissa
Rural SC
Rural ST
Rural others
Urban SC
Urban ST
Urban others
West Bengal
Rural SC
Rural ST
Rural others
Urban SC
I than ST
Urban others
Coeff of
Var
Gini
Theil
.33
.37
.41
.40
—
.64
.171
.174
.197
.218
—
.275
.022
.025
.030
.033
—
.066
.49
.54
.53
.83
—
.60
.225
.270
.246
.325
—
.296
.041
.053
.048
.096
—
.065
.49
.50
.52
.4 7
.53
.58
.242
.245
.244
.236
.239
.294
.044
.046
.047
.042
.047
064
.64
.47
.57
.60
—
.66
.276
.237
.288
.293
—
.327
.063
.041
.061
.064
—
.079
571
Lure groups than one can expect in case of
non-food items.
Within the food group, the concentration
ratio is considerably low for cereals than
for milk and milk products. This again
appears logical as cereals intake is neces
sary for survival and enters the consump
tion basket of poor and rich in accordance
with the body needs. The higher total
expendi turc of rich is mostly on account of
items other than cereals, such as clothing
or milk and milk products. /Xmong the
non-food items, the concentration ratio
for fuel and light was considerably lower
than clothings. This result may surprise
some but those conversant with rural re
alities would realise that fuel and light is
as much a necessity for rural households
as for urban and as much a necessity for
poor as for rich, the only difference may
be in the mix of energy sources for fuel
and light and not necessarily in the expen
diture thereupon.
Comparing the concentration ratios
across rural and urban households, it is
n^BM'est that while the urban Gini ratios
fSRtal expenditure may be somewhat
higher (Table 3A). the concentration ratio
for cereals is mostly higher in case of rural
households, perhaps because part of the
rural population may not have access to
minimum cereal requirements. The cloth
ing concentration ratio is somewhat higher
in case of urban households except for
some anomalies. In any case this ratio is
quite high, both for urban as well as rural
areas. This reflects the broad social milieu
in which quality of clothing and, therefore
the expenditure thereupon, reflect the so
cial status. This prompts the higher in
come, brackets to spend progressively and
disproportionately higher amounts on
clothing.
The comparison of concentration ratio
for SC and ST population with those for
other households shows that in most cases
inequality is higher in non-SC/ST popula
tion This trend is mainly on account of
hiMfer variations in non-food expenditure
oi^wn-SC/ST population as that on food
items is found to be much more egalitar
ian. The concentration ratio for cereals, in
general, was found to be somewhat higher
for SCs and STs as compared to non-SC/
ST households. Considerably higher in
equalities existed in case of milk and milk
product consumption of SC/ST households.
with the concentration ratio being quite
high ( 533 to .784) in case of Orissa, both
in urban and rural areas. Perhaps the milk
marketing network needs to be strength
ened in eastern India. The Operation Flood
gains in case of western and northern India
are well known. The programme needs to
shift its attention to hitherto neglected
eastern region and the NDDB should be
asked to spread its network in the four
572
eastern states in close co-operation with
the state governments. As for the high
concentration ratio in case of clothing,
nothing much can be done except that low
cost clothes can be made free of excise
duty while most of the revenue may be
raised from high cost clothes. To some
degree such a fiscal policy design already
exists but it may not have contributed a
great deal to the lowering of concentration
ratio as the producers of high cost clothes
are in belter position to shift the incidence
of lax back to the consumers, because the
high income bracket has the ability to pay.
V
Poverty Estimates
Using the poverty lines in 2.2.2. for com
puting the head count ratio and the FGT
variant for grouped data, poverty estimates
were obtained for SCs and STs and other
households for the four eastern slates. The
results obtained arc presented in 'Fables 5
and 6.
From the head count measure it is obvious
that the incidence of poverty is alarmingly
high among SC and ST population, both in
rural and urban areas of Bihar and Orissa
Table 3B. Measures of Inequality—Atkinson Index-
Assam
Rural SC
Rural ST
Rural others
Urban SC
Urban ST
Urban others
Bihar
Rural SC
Rural ST
Rural others
Urban SC
Urban ST
Urban others
Orissa
Rural SC
Rural ST
Rural others
Urban SC
Urban ST
Urban others
West Bengal
Rural SC
Rural ST
Rural others
Urban SC
Urban S I'
Urban others
0.5
1.0
Values of Epsilon
1.5
2.0
2.5
.024
.027
.033
.037
—
.068
.047
.051
.064
.074
—
122
.068
.072
.092
.108
—
.167
.088
.091
118
.141
—
206
.108
.109
.144
.17
—
.24
044
.059
.052
.094
—
.07
.083
.112
.098
.163
—
.133
.118
.161
.140
.217
—
.186
.150
206
.179
.260
—
.233
.18
.247
.215
.296
—
.274
.049
.05
.051
.046
.05
.07
.093
.096
.097
.087
.092
.132
.135
.14
.14
.124
.128
.187
.175
.181
.180
.158
.160
.232
.213
.219
.219
.188
.189
.282
.066
.046
.069
.069
—
.086
.121
.088
.133
.131
—
.161
.17
.128
.196
.184
—
.226
.214
.165
.257
.232
—
.282
.256
.200
.315
.274
—
.332
Table 4 A: Concentration Ratios for Select Items for Assam
Total cereals
Milk and milk prods
Food total
Fuel and light
Clothing
Non-food total
SC
Rural
ST
Others
SC
Urban
ST
Others
.063
.331
.118
.118
.589
317
.093
.335
.137
.137
.259
.282
118
.290
.151
.150
.530
.323
113
.396
183
.1 12
.274
.311
—
—
—
—.
—
—
.074
.304
.173
.116
.546
.449
Table 4 B: Concentration Ra tios for Sfi ecir Items for Bihar
Total cereals
Milk and milk prods
Food total
Fuel and light
Clothing
Non-food total
SC
Rural
ST
Others
SC
Urban
ST
Others
.164
.520
.190
.119
536
.331
199
.630
.232
179
.680
.383
.143
.483
204
.121
.569
356
.098
.4 15
.192
.189
.730
.558
_
—
—
—
—
—
.082
.480
.220
.197
.596
.445
Economic and Political Weekly
March 5, 1994
and in the rural West Bengal. Three-quar
ters or more of SC and ST rural population
is poor in Bihar, Orissa and West Bengal. In
West Bengal and Bihar even the non SC/ST
rural population has a high poverty inci
dence. In general, poverty is higher in ease
of SCs and STs, when compared to non SC/
ST population, though the difference is not
large in ease of rural Assam. In ease of urban
areas, the poverty incidence is distinctly
lower than in rural areas, though in case of
Orissa, nearly three quarters of the SC/ST
population was poor even in urban areas.
Tlie incidence of poverty among non SC/ST
urban population was below 50 per cent in
case of Bihar and Orissa and was quite low
in case of West Bengal (26 per cent) and
Assam (17.6 per cent). For Assam, the pov
erty incidence is lower compared to all other
states, reflecting again the low inequalities
in consumption expenditure amidst high
MPCE, as shown in the preceding two sec
tions. Furthermore, poverty incidence among
rural households is only marginally higher
for SC and ST households compared to
non SC/ST households. In urban areas,
however, poverty is distinctly higher in
case of SCs compared with non-SC/ST
population.
The results obtained by the FGT variant
are in broad conformity with the head
count ratio. It may however be mentioned
that our FGT measure Pa for a = 0 is not
exactly equal to the head count ratio, though
according to theory it should have been.
This is due to the fact that for computing
head count ratio, the frequency in the class
where the poverty line lies was distributed
equally across that class width and the
percentage of people within that class who
lie below the poverty line was added with
the cumulated frequencies of the classes
having income less than the poverty line.
In contrast, for P measure we had im
posed a log-normal distribution and the
expenditure class with poverty line formed
part of this distribution. Thus, the fre
quency of the expenditure class with
poverty line was divided among poor and
non-poor according to the parameters of
the log-normal distribution. It may also be
noted that with a higher a, the value of
Pu decreased. In fact at a =3. Pa ranges
between a low of .005 in case of urban
Assam non SC/ST households and a high
of only .094 in case of rural Orissa ST
households.
We have seen that the poverty analysis
undertaken has brought to fore some inter
esting Findings. Poverty incidence in east
ern India is indeed high and is on a much
higher side than the official all-India pov
erty ratio figures. While the inequality mea
sures showed the poor performance of the
region on equity front, the alarmingly high
poverty incidence shows that the area must
be lagging on the growth front as well.
buonomic and Political Weekly
Increasing the pace of development in east
ern India must, therefore, be put high on the
country’s economic agenda 'Hiosc conver
sant with thcsocio-economicprofilcof east
ern India would acknowledge that economic
inequities are rooted in the rigidities in
social structure prevalent in the region. Even
as late as the early 80s the economic and
social relations were al best feudalistic in
structure. For instance, till December 1992,
of the 7,888 bonded labourers identified in
the country 7,325 were in Bihar and the bulk
of the remaining in Orissa. Furthermore, the
agricultural growth has been low in the
region, except during 80s when particular
endeavour was made to step up rice produc
tion through special rice production
programme (SRPP) and special foodgrains
production programme (SFPP). 'Die indus
trial growth has been lopsided on the side of
heavy and infrastructural industries, mainly
in the public sector, and the industries in the
region arc plagued by low labour-intensity
and poor productivity. The tertiary sector
growth has also been rathcr poor. As a result
unemployment backlog has become a mat
ter of concern, making it imperative that
employment generation capacity of the new
investments in the region is kept in mind.
Also, considering the high poverty inci
dence the crucial role of direct poverty
alleviation and employment generation
programmes cannot be neglected. Consid
ering that poverty is indeed a widespread
malaise in rural regions, increased outlay
for rural development needs to be sustained
over a fairly long period.
VI
Conclusions
The study has thrown up a number of
interesting results, which have been dis
cussed in various sections of this paper.
Table 6: Poverty Measures—FGT Index
0.0
Assam
Rural SC
.470
Rural ST
.503
Rural others .465
Urban SC
.409
—
Urban ST
Urban others .186
Bihar
Rural SC
.852
Rural ST
.788
Rural others .630
.604
Urban SC
—
Urban ST
Urban others .458
Orissa
Rural SC
.731
Rural ST
.847
Rural others .525
Urban SC
.691
Urban ST
.724
Urban others .428
West Bengal
Rural SC
.714
Rural ST
.792
Rural others .593
Urban SC
.389
—
Urban ST
Urban others .250
Values of a
1.0
2.0
3.0
.093
.103
.103
.093
—
.039
.027
.009
.011
.030
.033
.013
.012
.031
__ —
.013
.005
.322
.310
.196
.210
—
136
.150
.153
.081
.097
—
.056
.078
.084
.039
.051
—
.027
.251
.343
.147
.218
.237
.124
.108
.171
.057
.091
.101
.064
.045
.094
.026
.043
.049
.063
.259
.287
.206
.108
—
.066
.121
.132
.095
.042
—
.026
.064
.068
.050
.020
—
.012
Table 4C: Concentration Ratios for Select Items for Orissa
Total cereals
Milk and milk prods
Food total
Fuel and light
Clothing
Non-food total
sc.
Rural
ST
Others
SC
Urban
ST
Others
.159
.533
.205
.128
.505
.354
.189
.655
.223
145
.479
.320
.127
.470
.186
.177
.495
.400
.099
.695
.185
.096
.613
.367
.097
.784
.197
.179
.483
.347
.053
.473
.213
.172
.559
.436
Table 4D: Concentration Ratios for Select Items for West Bengal
Total cereals
Milk and milk prods
Food total
Fuel and light
Clothing
Non-food total
SC
Rural
ST
Others
SC
.173
.484
.227
.177
.630
.419
.172
.474
.204
.129
.549
.337
.174
.489
.246
.174
.523
.406
.108
.440
.240
.206
.622
.399
Urban
ST
—
—
—
—
Others
—
.063
.399
.243
.230
.605
.454
Urban
ST
Others
Table 5: Poverty Measures—Head Count Ratios
Assam
Bihar
Orissa
West Bengal
March 5. 1994
SC
Rural
ST
Others
SC
49.57
84.59
74.49
74.27
53.70
79.20
84.84
78.54
47.58
65 77
55.38
60.57
43.03
67.06
73.23
44.29
—
—77.10
—-
17.60
48.31
44.45
26.06
573
Many of these have important policy impli
cations. We synthesise some of the key
findings to facilitate thinking on important
policy issues.
First, the eastern region as a whole suffers
from economic backwardness and this is
reflected in low MPCE which is below the
all-India levels. This is particularly true for
the SCs and STs. The incidence of poverty
is also considerably higher than the all-India
figure. Secondly, the rural-urban divide ex
ists in eastern India, with a lower MPCE and
higher poverty incidence in rural areas as
compared to urban. The consumption in
equality measures are however, higher for
urban areas. This is further confirmed by
looking at the concentration ratios for a
select few commodities. Thirdly, the study
shows that the SC and ST households have
lower consumption standards than non SC/
ST households and also suffer from high
poverty incidence. More importantly, con
sumption inequalities are low among SCs/
STs compared to other households. The
analysis of the concentration ratios reingBjees this view. Also among SCs and STs,
are somewhat better-off compared to
STs in terms of marginally higher MPCE
and lower poverty.
The first two results are incidental to the
study but are nevertheless important as it
confirms the need to expedite development
in eastern India through direct measures.
particularly those for employment genera
tion and poverty eradication. Tie third re
sult is central to the issues that this paper set
out to raise. SC and ST households arc
unequivocally worse off than the general
population and, therefore, merit special pref
erences in government plans to uplift the
consumption standards of the population.
Furthermore, the view prevalent in certain
quarters that there arc large inequalities
within the SC and ST households is shown
to be ill-founded. A general line of attack on
specific social benefit programmes for SCs
and STs is that in presence of large inequali
ties within these groups, economic benefits
cM^rnercd by the higher rung of the SC/ST
closes and this lot graduates themselves to
still higher expenditure classes, leaving the
downtrodden unaffected or worse-off. If
this were true, one is likely to see high
inequalities in consumption expenditure for
SCs and STs. Our results are quite contrary
to this. While inequalities are low, poverty
incidence is high. This makes a strong case
for targeted intervention for SCs and STs to
raise their consumption standards. As STs
are. in general, relatively worse-off than
SCs, special area programmes exclusively
targeted for STs may supplement the
programmes targeted for SCs and STs.
Apart from the general findings stated
above, the study also yielded some striking
state-specific results. First. Assam appears
to be an outlier among the eastern states
574
suffering from low consumption standards
and high poverty and inequalities. In Assam,
lheconsumption levels are high, the poverty
incidence appear to be lower than all-India
trends and consumption expenditures are
well distributed across SC, ST and Other
Households, as also within these house
holds. Hiereforc, it is difficult toexplain the
social unrest in the state merely in terms of
economic reasons. Second, inequalities re
main high in West Bengal despite the redis
tributive policies. 'Hie incidence of rural
poverty is also at a high level. Among the
four eastern states poverty ratio is the sec
ond highest in West Bengal, next only to
Bihar. Thirdly, in Bihar STs arc marginally
better-off than SCs. perhaps due to spe
cially targeted programmes for STs. Fourth,
poverty incidence is alarmingly high in both
the rural and urban areas of Bihar and Orissa,
more so in the rural areas.
Lastly, we hope the study will stimulate
research towards objective analysis of stan
dards of living for SCs and STs in various
parts of the country. Such information will
go a long way in forming necessary policy
response to socio-economic issues and also
help in evolving a socio-political consensus
on the policy measures. We also hope that
more up-to-date data will be generated and
published by the NSSO on consumption
expenditure of SCs and STs to facilitate
research in this direction.
References
Atkinson. B (1970): ‘On the Measurement of
Inequality’../w/nr// ofEconomic Theory 2
Vol 44. No 2. pn 244-63
Anand, Sudhir (1980): Inequality and Poverty in
Malaysia: Measurement and Decomposition,
Oxford University Press.
Bardhan. P K (197.3): ‘On the Incidence of Pov
erty in Rural India’, Economic and Political
Weekly Annual Number February.
Bardhan, P K and T N Srinivasan (1974): Poverty
and Income Distribution in India, Statistical
l*iib|ishing Society, Calcutta.
Dandekar V M (1981): ‘On the Measurement of
Poverty’, Economic and Political Weekly,
Vol XVI, No 30. July 25. pp 124 1-49.
Dandekar. V M and N Rath (1971): Poverty in
India, Indian School of Political Economy, Pune.
Dutla. Bhaskar (1978): ‘On the Measurement of
Poverty in Rural India’, Indian Economic
Review, Vol 13, No 1, pp 23-32.
Foster, Greer and Thorbcck (1984): ‘A Class of
Decomposable
Poverty
Measures’,
Econometrica, Vol 52, No 3, pp 751-65.’
Government of India (1986): A Technical Note on
Seventh Plan, Policy Perspective Division,
Planning Commission. New Delhi.
Jain and S Tendulkar (1990): Role of Growth and
Distribution in the Observed Change in Head
Count Ratio Measure of Poverty : A Decom
position Exercise for India, Indian Economic
Review, Vol XXV No 2. pp 165-205.
Kakwani. C Nanak (1980): Income Inequality and
Poverty—Methods of Estimation and Policy
Applications, Oxford University Press.
Mrnhas. B S, L R Jain.S M Kansal and M RSaluja
(1986): ’On the Choice of Appropriate Con
Notes
sumer Price Indicesand DataSets for Estimat
ing the Incidence of Poverty in India’, Techni
[The authors thank M II Kirit Parikh,
cal Report No. 8615, August 1986 (revised
Suryanarayana and R Nagaraj for their kind com
January 1986). Indian Statistical Institute.
ments on a previous version of the study. Any
Nayak. Vijay and Prasad, Sailaja (1984): ‘On
remaining errors and omissions are the responsi
Levels of Living of Scheduled Castes and
bility of the authors.)
•Scheduled Tribes’, Economic and Political
Weekly, July 28. Vol XIX. No 30. pp 1205-13.
1
For details on the sampling structure refer to Paul, Satya (1989): Inequality, Poverty and Con
Sarvckshana (1989).
sumption. Commonwealth Publishers.
2 The interested readers may refer to Satya Paul Rao, VKRV (1981): ‘Measurement of Poverty’,
( 1989) for a comprehensive discussion on these
Economic and Political Weekly, Vol XVI,
limitations.
No 35. August 29, pp 1433-36.
3
The commonly used statistical measures of
Sarvckshana (I9<S9)‘. Journal of the National
range, variance and variance of log income
Sample Survey Organisation, Department
were not used on the ground that they do not
of Statistics, Ministry’ of Planning, Govern
meet all the desirable properties stated above
ment of India. Vol NIL No 3, Issue No 38,
simultaneously. Theil’s entropy index besides
January-March 1989
meeting all the three properties has added ad
Sen. Amartya (1973); On Economic Inequality,
vantages of being defined for distributions with
Oxford. Clarendon Press.
zero income recipients and is also additively
(1976): ‘Poverty An Ordinal Approach to
decomposable.
Measurement’. Econometrica, Vol 44. No 2,
4 For detailed references ofthese measures, please
pp 219-31
see Sudhir Anand (1980).
Stem. Nicholas H (1977): ‘Welfare Weights and
5 Fora quick recapitulation of the poverty debate
the Elasticity of the Marginal Valuation of
in Indian context, it may he recalled that the
Income’ in M Artis and R Nobay (eds),
first effort in (he direction was made in 1962 by
Current Economic Problems, Basil
a high level working group consisting of emi
Blackwell. Oxford.
nent economists. D R Gadgil, VKRV Rao.
Sukhalmc. P V (1981): ‘On Measurement of
Pitamber Pant and some technocrats and nutri
Poverty’. Economic and Political Weekly,
tional experts. The group used norms set by the
Vol XVI, No 32. August 8. pp 1318-24.
Nutrition Advisory Committee of the K’MR in
Suryanarayana. M 11 and S Gcetha (1992): ‘Es
1958 to arrive at the conclusion that the na
timation of P-cx Poverty Measure for
tional minimum consumption expenditure per
Grouped Data using Two Parametric
household (of five persons) should not be less
LogNonnal Distribution’, Discussion Paper
than a hundred rupees per month al 1960-61
No 78. Indira Gandhi Institute of Develop
prices.
ment Research.
Economic and Political Weekly
March 5, 1994
SOCIAL SECTORS
2. The government has relied mainly on three
approaches for reduction of poverty and
unemployment: the first entails pursuit of higher
economic growth which will improve the levels
of living of all groups of people in the society
including the poor: the second involves direct
anti-poverty and employment programmes; and
the third has stressed high priority to government
expenditure on social sectors. The reforms
underway since the economic crisis of 1991 have
sought to strengthen these approaches.
Growth wi*n social justice and alleviation of
poverty have been primary objectives of Indian
planning since its inception in ‘951 Several anti
poverty measures aiso have been in operation
for decades focussing on the poor as the target
groups. These include programmes for the
welfare of weaker sections, women and children,
and a number o' special employment
programmes for seif and wage employment in
rural and urban areas.
TABLE 10.1
Central Government Expenditure (Plan and Non-Plan) on Social Services
Item
(Rs. crore)
1990-91 1991-92 1992-93
1993-94
1994-95
1995-96
1996-97
1997-96
(RE)
1996-99
(BE)
5380
5892
6397
8150
9223
11631
1 3659
16520
21159
■< VCS
'■332
1 878
'.722
237 8
2146
2799
2413
3630
2542
3988
2751
5*34
3369
6535
4366
934
7 86
0/1
1262
392
- 0 c,
4 1 7
479
1 756
596
2957
593
3449
896
4495
923
L1C
488
564
744
800
833
728
1101
f
1686
1273
-iee-"- arc za~v:v Welfare
’Veate' Suaoiy. Sanitation.
Housmc anc Urcan Development 826
m’c'maccr ' ana Brc.accastma
436
Warfare o; SC
ST ana other
348
BaCKwa’d classes
Labcf Employment and
g
Laocu' V.'e'tare
Socia Welfare & Nutrition
289
520
357
628
347
803
526
880
440
997
507
1800
587
1950
651
2293
S31
2908
2578
2283
3211
4680
5803
6609
5081
5321
5890
2466
2673
3760
1. Social Services
a
Ec-cat-o'' Art 8 Culture ana
2
c
c
e
2. Rural Development
3. Basic Minimum Services (BMS)*
including Slum development
4. Social Services, Rural
Development and Basic
Minimum Services (1+2+3)
8058
8175
9608
12830
1 5026
18240
21206
24714
30809
5. Total Central Government
Expenditure as % of GDP
at current market prices
18.1
16.6
16.0
16.2
15.5
14.6
14.3
15.0
15J_
7.7
7.3
7.8
9.0
9.3
10.2
10.5
10.5
11.5
1.4
1.2
1.3
1.5
1.4
1.5
1.5
1.6
6. Social Services. Rural
Development and Basic
of Total Expenditure
7. Social Services. Rural
Development and Basic
Minimum Services as % of
GDP at current market pricesS
Note*
; a
■
V
: Figures for the years 1990-91 to 1995-97 are actuals.
’ : Came into operation from T996-97.
fitted
: The ratios to GDP are basec or. new series of National Account Statistics with 1993-94 as base year released by the C antral
Organisation (CSO) on February 3. 1999. For the purpose of comparability. GDP at current market prices for t’.e years
19S1-92 <=nd 1992-93 have been interpolated by using an average link factor of 1.0878 obtained from the over app’mg yea‘..i|^
1993-94.1994-95.1995-96 and 1996-97 tor which both the old series and new series for the GDP at current market prices are ava
Source • Budget Papers.
S
143
•
. 2.
_______________________________
TABLE 10.2
Central Plan Outlay for Major Schemes of Social Sectors and Rural Development
(Rs crore)
t-jj-
IKinistry/Department/Scheme
1990-91
1997-98
1996-97
1995-96
1998-99
(BE)
(RE)
(BE)
(RE)
(BE)
(RE)
(BE)
(RE)
(BE)
865
education
ol which
’« a) Elementary Education
265
96
b) Adult Education
1
2 HEALTH INCLUDING I. S. M. 4 H.
275
3 family WELFARE
675
h 4 WOMEN AND CHILD DEVELOPMENT 330
ot which
s
Integrated Child Development
Services
268
364
j: / 5 WELFARE
p RURAL DEVELOPMENT AND
r RURAL EMPLOYMENT &
3130
POVERTY ALLEVIATION
•A- ' ''
A < ' '..
of which
a) Jawahar Rozgar Yoiana (JPY)
2100
drv*
b) Employment Assurance
837
1825
2504
3388
2574
4095
3350
4245
224
1443
170
649
1506
821
2264
225
815
1535
847
1567
112
818
1547
847
2542
127
2265
81
918
1829
1026
2779
94
255
785
313
651
234
670
1581
730
255
366
588
890
669
890
682
890
682
890
1389
600
804
603
1539
2975
7700
8248
8632
7775
9001
8290
9811
2001
3862
2955
1865
1655
2078
1953
2095
1570
1816
1970
1840
1970
1905
1990
550
932
550
700
490
700
,• £ ’
■1
1
X
1 ‘1 •
1 ‘
1
I
-1 * Z
Scheme (EAS)’
c) National Social Assistance
Programme-;d) Integrated Rural Develocment
Programme including
Rural Artisans
e) Rural Water Supply and
Sanitation
f) Indira Awas‘Yojana +*
'-1 \ ?
g) Million Wells Scheme ++
7 OTHER PROGRAMMES
a) Nehru Rozgar Yojana (NRY).
1
b) Scheme for Self Employment for
!
Educated Unemployed
Youth (SEEDY) S
c) Prime Minister's Rozgar
2
Yojana (PMRY).
d) Swama Jayanti Shahari
Rojgar Yoiana (SJSRY)'@@
1 •
-f /St
1
V
'-Z- ,
955
1829
900
734
1195
2489
1226
377
356
656@
656
656
646
61 1
552
800
443
421
1170
1170
492
1170
1194
1155
1194
1402
1402
1727
1144
1600
21 1
448
388
1190
448
373
450
71
68
71
50
80
31
145
145
145
115
145
95
110
103
189
120
1 10
57
53
(A) Total Central Plan outlay on Major
Schemes on Social Sectors
5816
(1 to 7)
(8) Total Plan Expenditure
30466
(C) A as percentage of
Total Plan Expenditure
19.1
(D) A as Percentage of GDP
at current market pricesSS
1.0
5694
13612
14831
20804
48684
54894
18394
62852
16446
48500
16323
54685
14616
29956
60630
72002
19.0
28.1
30.5
29.8
26.6
29.3
27.1
28.9
1.0
1.1
1.2
1.2
1.0
1.2
1.1
1.2
’
Came into operation on October 2. 1393.
~
©
The scheme was announced on 1Sth August 1995, Rs. 550 Crcre was provided at RE stage.
BE (Rs 640 crore) was revised upward within total BE for the Deptt. of Rural Development.
+-!■
-1 :
131
The Indira Awas Yojana (IAY) and the Million Wells Scheme (MWS) were earlier the sub schemes of JRY From 1 1 1996 thev
have become separate schemes.
S Integrated with PMRY.
@@ Is a rationalised version of the erstwhile schemes'of Urban Basic Services, NRY and PM's Integrated Urban Pov* rty
Eradication Programme.
SS : The ratios to GDP are based on new series of National Account Statistics with 1S93-94 as base year released by the Central
Statistical Organisation (CoO) on February 3, 1999 . For the purpose of comparability. GDP at current market prices for 199091 has been interpolated by using an average link factor of 1.0878 obtained from the overlapping years viz 1993-94, 1994-95,
1995-96 and 1996-97 tor which both the old series and new series for the GDP at current market prices are available.
Source : Budget Papers.
144
3. The Central government expenditure on
social sectors (comprising education, health &
family welfare, water supply, sanitation, housing,
social welfare, nutrition, rural employment and
minimum basic services) as a ratio to total
expenditure increased from 7.7 per cent in 199091 to 10.5 per cent in 1997-98 (RE) and further
to 11.5 per cent in 1998-99 (BE). As a ratio to
the GDP at current market prices, the central
government expenditure on social services
increased from 1.4 per cent in 1990-91 to 1.6
per cent in 1997-98 (RE) and further to 1.7 per
cent in 1998-99 (BE) (Table 10.1).
TABLE 10.3
Basic Indicators of Human Development
Year Life expec
tancy at
birth
(years)
4. The Central plan outlay on major schemes
on Social Sectors as a percentage to the GDP
at current market prices increased from 1.0 per
cent in 1990-91 to 1.2 per cent in 1998-99(BE)
(Table 10.2). The central outlay for Welfare of
Weaker Sections increased by 91.4 per cent in
1998-99(BE) over 1997-98(RE), Family Welfare
by 36.1 per cent. Health by 30.2 per cent and
Education by 26.7 per cent.
’
Birth
rate
Literacy
rate
(per
cent)
Death
infant
rate mortality
____________ rate
(per thousand)
1951
32.1
18.3
39 9
27.4
146
1961
41.3
28.3
41.7
22.8
146
1971
45 6
34.5
36.9
14.9
129
1981.
50.4
43.6
33.9
12.5
110
1991
59.4
52.2
29.5
9.8
80
1996
62 4
NA
27.5
90
72
1997
NA
62’
27.2
8.9
71
Source : Registrar General and Census Commissioner
of India.
’ As per NSSO.
economic growth has brought about an
improvement in living standards of people in
general.
5. Increased availability of health care and
family welfare services have resulted in reduction
of all India death rate, birth rate and infant
mortality rate. The crude death rate declined
from 14.9 per thousand in 1971 to 9.8 in 1991
and further to 8.9 in 1997. Similarly, the infant
mortality rate per thousand declined from 129
in 1971 to 80 in 1991 and further to 71 in 1997.
The birth rate per thousand also declined from
36.9 in 1971 to 29.5 in 1991 and further to 27.2
in 1997 (Table 10.3). These tentative trends
are consistent with the view that rapid
6 However, there are wide inter-state variations
in indicators of human devefopment. For
instance, in Kerala the life expectancy at birth at
72 years and overall literacy at 90 are
significantly higher than those in states like Bihar,
Madhya Pradesh, Orissa, Rajasthan and Uttar
Pradesh where concerted efforts are required
to improve overall quality of life. In fact the
indicators of human development in Kerala are
comparable with several Asian developing
countries like China, Malaysia, Indonesia,
Thailand and Sri Lanka which have made
significant progress in human development over
the years (Table 10.4).
TABLE 10.4
Indicators of Human Development for Some Asian Countries
Country
Adult
Literacy rate
(Per cent) .
Life
Expectancy
at birth
(Years)
1995
Infant Mortality
rate (Per
thousand
births)
1996
India
62.4
72
52,:..
Kerala Srate (India)
72.0
13
China
69.2
38
90 :
82 .
Inaonesia
64.0
47
Korea. Republic
71.7
6
Malaysia
71.4
Philippines
Thailand
67.4
69.5
1995
84
98-T
84
32
31
Source: UNDP - Human Development Report. 1998.
For India, estimates are trom Registrar General & Census Commissioner of India & relate to tne year 1996 (P).
95
■
94 . r.
145
j - Average real wages for unskilled agricultural
1 wDO^r. which reflect the economic conditions of
’A'ricultural labourers, declined by 6.2 per cent
crisis year of 1991-92 (agriculture year
I
;jiy 10 June) for the country as a whole, but
| %-reased in subsequent years except in 1994-
:■ -"§5 (Table 10.5). However, there were no uniform
Trends across the states implying that local
' conditions exert significant influence on
'agriculture wages.
Poverty
8. Together with the overall economic growth.
the anti-poverty and employment generation
programmes have helped in reducing the
incidence of poverty over the long run. The
poverty ratio declined from 56.4 per cent in
1973-74 to 37.3 in 1993-94 in rural areas and
from 49.0 per cent in 1973-74 to 32.4 per cent
in 1993-94 in urban areas. For the country as a
whole, the poverty ratio declined from 54.9 per
TABLE 10.5
Annual Percentage Change in Real Wages for Unskilled Agricultural Labour for Selected States
Percentage Change for
agricultural year (July to June) over previous year
1991-92
1992-93
1993-94
■; Andhra Pradesh
(-) 11.40
(+)
1.57
(+) 3 60
(+)
2.71
(-)
Assam
(-)
8.73
(+) 0.58
(-)
6 58
(■)
167
Bihar
(-)
4.39
(-)
5.00
(-) 5.98
(*)
• Gujarat
(-)
4.31
(♦)
7.92
(<•)
2.86
(+)
Karr.ataKa
(-) 13.25
Keraia
(+)
Machya Pradesh
(-)
State
(-) 14.39
(*) 41,31
4.07
(+)
(■)
3.89
Maharashtra
Orissa
9.74
1995-96
1994-95
1996-97(P)
1997-98(P)
I-73
(+)
1-51
(+)
1.40
(+)
2 68
(+)
2.95
(+)
0.93
1.69
(-)
2.30
(-)
2.79
(-)
0 79
1 27
(+)
2 92
(+)
5.08
(*)
14.52
(-) 15.60
(-)
8 61
(+)
21.48
(+)
15.22
2.34
(-)
5 24
(+)
13.20
(+)
1 4 60
(+)
15.34
(+) 12.57
H 3.53
(+)
4.93
(+)
1 24
(+)
1.31
(+)
0.96
(-) 14.79
(+) 0.66
(+) 25.58
(-)
0.68
(-)
7 89
(■)
7.98
(+)
31.73
(■)
3.89
(+) 11 03
(-)
0.14
(-)
3.52
(+)
0 55
(■)
0 27
(+)
2.56
Punjab
(+)
3.65
(+) 4.25
(*)
1.51
(■)
1-17
(-)
6 50
(-)
0.42
(-)
3 35
• Rajasthan
(-) 6.12
(-) 3.56
(-)
7.66
(♦)
1 05
(+)
10.33
(+)
1781
(+)
"3.78
Tamil Nadu
(-)
4.85
(+) 13.29
(+) 11.60
(+)
1.03
(+)
3.63
(+)
28 05
(■)
16.67
Uttar Pradesh
(+)
1.02
(+) 7.56
(-) 6.77
(-)
2.31
(+)
14.78
(-)
6.37
(+)
18.29
West Bengal
(■)
625
(-) 24.39
(-) 5.50
(-)
5.29
(-)
0.28
(+)
10.21
(-)
5.16
' (-)
6.19
(+)
(+) 5.61
(-)
0.39
(+)
0.72
(+)
4.67
(+)
4.88
All India
5.21
(P) : Provisional.
. Notes : (i)
Data on state average wage rates for unskilled agricultural labour in current prices are taken
from Ministry of Agriculture. The same have been converted into real wages by deflating with the State level
Consumer Price Index Numbers for Agricultural Labourers (CPIAL) with 1960-61 as base. (CPIAL has been
sourced from Labour Bureau, Shimla). Having estimated real wages for agricultural year percentage change
over previous year has been worked out.
(ii)
New series of CPIAL with base 1986-87 = 100 were released w.e.t Nov., 1995. To maintain continuity of old
series of CPIAL, the new series have been converted by using the linking factor of each State and then, the
average for each State has been worked out on the basis of converted series.
(i'i)
The real wages for unskilled agricultural labour for each State have been weighted by total agricultural labourers
of the State for working out alk India average. The weighted average real wages for all India are based on 14
States as reported above. Having estimated weighted average real wages for all India, percentage change over
previous year has been worked out.
Source: Ministries of Agriculture and Labour.
146
'Cent in 1973-74 to 36 per cent in 1993-94 (Table
10.6). Large Sample surveys on common
' expenditure on the basis of which poverty ratios
are estimated are not available in subsequent
years.
9. Although reduction of the overall poverty
ratio in India from 55 per cent to 36 per cent
during a period of two decades is significant,
India's performance in poverty reduction has
been weak as compared with some of the East
Asian countries. It may be observed from Table
10.7 that the success of some of the East Asian
countries (like China and Indonesia) lies in
faster economic growth. In general, the faster
the rate of overall growth, the faster is the rate
of poverty reduction. It is, therefore, reasonable
to expect that a sustained and long lasting'
solution to the problem of poverty depends on
creation of opportunities for broad based
economic development and higher growth.
Employment
10. It may be observed from Table 10.8 that the
average annual growth rate of overall
employment (in both organised and unorganised '■
sectors) declined continuously from 2.75 per cent
in the period 1972-1978 to 1.77 percent in 19831988, but increased to 2.37 per cent in 1987-
TABLE 10.6
Number and Percentage of Population Below Poverty Line
(Number in million and poverty ratio in percentage)
Rural sector
Year
Urban sector
Combined All India
Number
(million)
Poverty
ratio
Number
(million)
Poverty
ratio
Number
(million)
Poverty
ratio
1973-74
261
56.4
60
49.0
321
54.9
1977-76
264
53 1
65
45.2
329
51.3
1983
252
45.7
71
40.8
323 '
44.5
1987-88
232
39.1
75
38.2
307
38.9
I 1993-94
244
37.3
76
32.4
320
36.0
I Source: Planning Commission.
1
TAB _E 10.7
1
Poverty incidence and growth rates in India and selected Asian countries
(in per cent)
Country
Poverty ratio
1975
Poverty ratio
1995
Annual Reduction
In 1975-95
Percentage point
Average GDP
growth 1970-1980
india
54.9
36.0
0.9
3.2
5.6
China
59.5
22.2
1.9
5.0
11.1
Indonesia
64.3
114
2.6
7.8
6.6
Korea
23.0
5.0
0.9
9.0
8.7
Malaysia
17.4
4.3
0.7
7.8
64
Philippines
35.7
25.5
0.5
6.2
1.4
Thaiiana
8.1
7.2
7.9
0.9
0.4
Average GDP
Growth 1980-1995
Source :
For India, Planning Commission; tor others World Bank Report on Social Consequences ot the East Aslan
Financial Crisis, September, 1998.
Note :
For tndia, poverty ratios refer to the years 1973 and 1993 respectively and GDP growth rates are based
on old senes with base 1980-81.
147
■B'gg4. However, the growth rate of organised
Kgectcr employment maintained its declining trend
;even during the period from 1987-88 to 1993■ g4 There was a significant improvement in the
iJSqrowth rate of organised employment in the
- private sector from 0.43 per cent in 1983-1988
‘■■■to 1.18 per cent in 1987-1994. For the first time
■ jhe growth rate of employment in the organised
- private sector exceeded the employment growth
.
in the pubic sector.
TABLE 1C.8
Grown of Employment (per cent)
Growth rate of
overall
Period
. employment
Growth rate of employment in
the organised sector
Public
Private
TABLE 10.9
Growth rates of Employment in organised
sector (per cent)
Year
Public
sector
Private
sector
Total
organised
1991
1.52
1.24
1.44
1992
0.80
2.21
1.21
1993
0.60
0.06
0.44
1994
0.62
1 01
0.73
1995
0.11
1.63
0.55
1996
(->0 19
5.62
1.51
1997
0.67
2.04
1.09
Total
Source
#72-73
1577-73
2 75
1977-78 to
1983
2.36
2.99
1.41
2.48
1983 to
1987-38
1 77
2 17
0.43
1.38
1937-38 to
1993-94
2 37
1.C0
1 13
1 05
2.45
Sou.-ce: Planning Commission.
■-11. Table 10.9 presents the annual growth rates
.of employment in the organised public ano
' private sector during 1991 to 1996. It can be
■ observed from the table that the private sector
contributed predominantly to the increase in the
. ...Organised sector employment in the reform
; period since 1991 except in the year 1993. The
■ i ‘ government has decided to set up tne Second
National Commission on Labour with a view to
provide protection to millions of workers. The
i main focus of the Commission would be to
< suggest rationalisation of the existing labour laws
i in the organised sector and also to suggest an
7 umbrella legislation for ensuring a minimum level
’ . of protection to the workers in the unorganised
"H. sector.
Poverty Alleviation and Employment
Generation Programmes
| .'...12. India’s anti-poverty strategy comprises of
j -: a wide range of poverty alleviation and
■| - |ieniploymen.t generation programmes, many of
. .Which have been in operation for severa'l years
.jg'^-'Snd have been strengthened to generate more
Wjjllrenpioyment, create productive assets, impart
i
Planning Commission.
-technical and entrepreneurial skills and raise the
income level of the poor. Under these schemes,
both wage employment and self-employment are
provided to the people below the poverty line. In
1998-99, government proposed to unify the
various poverty alleviation and employment
generation programmes under two broad
categories of Self Employment Schemes and
Wage Employment Schemes. Funding and
organisational patterns will also be rationaiisea
to achieve maximum beneficial impact of these
programmes. The budgetary (plan) support on
Rural Development and Rural Employment &
Poverty Alleviation has been enhanced to
Rs.9811 crore in 1993-99(BE) from Rs.8290
crore in 1997-98(RE).
13. The salient features of some of the major
employment and anti-poverty programmes are
given below:
(a) Integrated Rural Development Programme
(IRDP) and its allied programmes of Training
Rural Youth for Self-Employment (TRYSEM) and
Development of Women and Children in Rural
Areas (DWCRA) are major self-employment
programmes for poverty alleviation. The basic
objective of IRDP is to enable identified rural
poor families to augment their incomes and cross
the poverty line through acquisition of credit
based productive assets. Assistance is given in
the form of subsidy by the government and term
credit by the financial institutions for income
generating activities.
This is a centrally sponsored scheme funded
on 50:50 basis by the Centre and the states. It
is stipulated that at least 50 per cent of the
148
.assisted families should belong to Scheduled
Caste and Scheduled Tribe categories. It is also
•required that at least 40 per cent of those
assisted should be women under this
programme. About 535 lakh families have been
covered up to November 1998 since 1980-81
under the programme out of which coverage of
SC/ST families had been 45 per cent. The level
of per family investment is currently more than
Rs.17441 compared to Rs 1642 during 198081. A sum of Rs.800 crore (including Rs. 60
crore for Rural Artisans) has been provided in
1998-99 (BE), an increase of about 45 per cent
over 1997-98 (RE).
(b) The Training of Rural Youth for SelfEmployment (TRYSEM) is to train rurai youth
from the target group of families in skills so as
to enable them to take up self/wage
employment. It has been laid down that the
^^iverage of youth from SC and ST communities
should be at least 50 per cent of the rural youth
trained. Out of the total beneficiaries, at least 40
per cent should be women.
(c) The Programme of Development of Women
and Children in Rural Areas (DWCRA) aims to
improve the socio-economic status of the poor
women in the rural areas through creation of
group of women for income generating activities
on a self-sustaining basis. Up to November, 1998,
1.97 lakh women were benefited during 199899. A sum of Rs. 100 crore has been provided in
1998-99 (BE).
(d) Jawahar Rozgar Yojana (JRY) is a wage
employment programme with its main objective
^f generation of employment in the lean
agriculture season to the unemployed and under
employed rural people both men and women
living below the poverty line. The significant
aspect of the scheme is that it is implemented
by the Panchayats at the village, block and district
levels in the ratio of 70:15:15 respectively. An
amount of Rs.2095 crore has been allocated
during 1998-99 (BE) for JRY. Against a target of
396.66 million man-days during 1998-99. a total
of 190.28 million man-days were generated up
to November 1998 with an expenditure of
Rs. 1244 crore.
(e) The Employment Assurance Scheme (EAS)
has been universalised so as to make it
applicable to all the rural blocks of the country.
It aims at providing 100 days of unskilled manual
work up to two members of a family in the age
group of 18 to 60 years normally residing in
villages in the lean agriculture season, orj|t
demand, within the blocks covered under EAS '
A sum of Rs.1990 crore has been provided
during 1998-99 (BE). During 1998-99, a total of237 61 million man-days have been generated:
under the scheme with an expenditure of
Rs.1572 crore up to November 1998.
(f) The Million Wells Scheme (MWS) which was .
earlier a sub-scheme of JRY, is funded by the
Centre and states in the ratio of 80:20. The.
objective of the MWS is to provide open irrigation :
wells free of cost to poor, small and marginal
farmers belonging to SCs/STs and freed bonded f
labour. A sum of Rs.450 crore has been provided :
in 1998-99 (BE). Up to November 1998, a sum \
of Rs. 225.90 crore has been incurred during ?
1998-99 and 49821 wells were constructed. <
(g) The National Social Assistance Programme' .
(NSAP) recognises the responsibility of the1--Central and state governments for providing '
social assistance to poor house-holds in case of j;
maternity, old age and death of bread earner.
NSAP is a centrally sponsored programme with ?
100 per cent central funding to the States/UTs C
that provides benefits under its three components ■
viz., (i) National Old Age Pension Scheme
(NOAPS): (ii) National Family Benefit Scheme .
(NFBS); and (iii) National Maternity Benefit
Scheme (NMBS). On the basis of suggestions
made by the Central Advisory Committee onNSAP. the Government has since approved
changes relating to enhancement in the rate ofbenefits for NFBS and NMBS A sum of Rs.700
crore has been provided for the above three'
components of NSAP in 1998-99 (BE).
Y
(h) The Swarna Jayanti Shahan Rozgar Yojana
(SJSRY) which came into operation from
1.12.1997, sub-summing the earlier urban....
poverty alleviation programmes viz., Nehru
Rozgar Yojana (NRY), Urban Basic Services ;
Programme (UBSP) and Prime Minister’s
Integrated
Urban
Poverty
Eradication.
Programme (PMIUPEP). The scheme aims to,.
provide gainful employment to the urban
unemployed or underemployed poor by .:
encouraging the setting up of s’elf-employment
ventures or provision of wage employment. Itis .
being funded on a 75:25 basis between Centre
and the states. It comprises two special schemes
i.e. The Urban Self-Employment Programme
(USEP) and the Urban Wage Employment.'
Programme (UWEP).The scheme gives a special
impetus to empowering and uplifting the p°°r I.
women and launches a special programme, ..
•iCnely, Development of Women and Children
s^urban areas under which groups of urban
women setting up self-employment ventures
^eligible for subsidy up to 50% of the project
'rtst During the year 199/-98, a sum of Rs.98.63
-^re was released to States and UTs under
TgjSRY- A sum of Rs-189 crore has been
-provided in 1998-99 (BE) out of which Rs.64.59
has been released to twelve states till
30.11-1998.
0 Prime Minister’s Rozgar Yojana (PMRY) for
providing self-employment to educated
unemployed youth had been designed to provide
employment to more than a million persons by
F
setting up of s even lakh micro enterprises in
Eighth Plan. During the Eighth Plan, loan in 7.70
lakh cases were sanctioned and 5.76 lakh cases
disbursed. The scheme is beina continued in
the Ninth Plan. Since inception of the scheme
up to the programme year 1997-98, over 7.52
lakh cases have been disbursed. During 199899, 57527 cases have been sanctioned loans
and 27533 cases disbursed bv the end of
October, 1998. A sum of Rs.110 crore has been
provided in 1998-99 (BE).
The achievements of the special anti-poverty
programmes are indicated in Table 10.10.
TABLE 10.10
Performance of Special Employment and Poverty Alleviation Programmes
(In lakh)
Programmes
A. Programmes in Rural Areas
1. JRY- Mandays of employment generated
2. EAS ■ Mandays of employment generated
3. IRDP- Families assisted
■ 4. TRYSEM- Youths Trained
5. DWCRA- (a) Groups formed
(b) Membership
6. lAY-House Constructed
7. MWS- Wells Constructed
8. ARWSP- Habitation/viilages
9. CRSP-Sanitory latnne
W. NSAP —
(a) NOAPS- Beneficiaries
(b) NFBSBeneficiaries
(c) NMBSBeneficiaries
1996-97
1997-98(P)
1998-99(P)
(upto Nov. 98)
Target Achievement
Target
Achievement
Target
Achievement
4141.4
3864.9
1.0
18.8
3883.7
4717.7
17.1
2.5
0.4
4.6
7.7
1.0
1.2
11.6
3966.6
1.0
8.3
4006.3
4030.0
19.2
3.6
0.4
5,3
8.0
1.1
1.0
12.2
1.1
16.0
1902.8
2376.1
7.7
0.8
0.2
2.0
3.6
0.5
0.5
5.6
53.7
4.6
46.0
46.6
1.6
14.0
48.7
2.7
25.7
40.3
1.8
15.4
48.8
2.1
17.8
40.2
1.4
6.6
1.2
135.8
1.2
0.6
44.6
0.4
2.2
4.4
2.2
3.2
2.2
44
1.8
2.6
2.2
4.4
0.3#
0.4#
B. Programmes in Urban Areas
1. NRY-(a) Families assisted
(b) Mandays of employment generated
(c) Persons trained
C. Other Programmes
1. PMRY -(a)Micro-enterprises @
(b)Employment generated S
SJSRY SS
of which
• (I) USEP —
(a) Beneficiaries
?'
(b) Persons trained
(ii) UWEP- Mandays of employment generated
1-------
2.9
0.3
11.2
3.0
0.3
7.2
P Provisional.
Targets are not fixed.
5 Estimated @ 1.5 per case disbursed for the concerned programme years.
Merged with SJSRY.
Cases disbursed.
Came into operation from Dec., 97.
1 Up to Oct.98
• JAs per report ending Dec., 98
Source: Ministry of Rural Areas & Employment and Dther concerned Departments.
2.9
0.6
9.9
0.2##
0.2##
12.9##
150
Literacy and Education
14. Building on educational priorities set out in
the National Policy on Education, 1986 as
modified in 1992 and its Programme of Action,
the National Agenda for Governance (NAG) has
education amongst its highest priorities. The
following educational agenda has been
specifically identified:—
c Education for All - Free and compulsory
primary' education up to 5,r standard and total
eradication of illiteracy.
e Education of prioritised groups - girls, SCs/
STs and Backward classes and educationally
backward minorities.
e Access and quality - equal access and
opportunity for all up to school stage and
improvement of quality at all levels.
e Financing of education - increase in
government and non-government spending on
education, and bringing this up to 6 per cent
GDP level.
In addition, the item on harnessing of youth
power makes a specific .mention of its
involvement in the total eradication of illiteracy.
The items dealing with Constitutional and Legal
Reforms, and Information Technology have
important implications for educational planning
and management in particular.
15. In pursuance of the emphasis embodied in
the National Policy on Education and reiterated
in the NAG, several schemes have been
launched by way of central intervention, primarily
for meeting the needs of the educationally
disadvantaged and for strengthening the social
infrastructure in the sector. The important
schemes by way of illustration are Operation
Black Board (OB), Non-Formal Education (NFE).
Teacher Education (TE). National Programme
of Nutritional Support to Primary Education
(NPNSPE) (Mid-day Meal Scheme), District
Primary Education Programme (DPEP), Total
Literacy Campaign (TLC), Community
Polytechnics (CP), Shiksha Karmi Project (SKP),
Area Intensive Programme for Educationally
Backward Minorities (AIPEEM) and Integrated
Education for Disabled Children, etc. Several
resource institutions have either been
strengthened/established to achieve the
objectives of the NPE.
16. As per the report given by the National
Sample Survey Organisation (NSSO), the overall
National figure for literacy has gone up frorn
1
52.2 per cent in 1991 to 62 per cent in 1997
The male literacy has gone up from 64.1 per,
cent to 73 per cent and the female literacy f;Crn. j
39.3 per cent to 50 per cent during the same /'
period. The literacy percentage has also goneup substantially in some of the educationally
backward states. Since independence, India has'
tripled its literacy (female literacy increased by.
five times).
.
17. The Gross Enrolment Ratio (GER) in theprimary stage (classes l-V) increased from 42.6?
per cent in 1950-51 to 89.7 per cent in 1997-98?
and in the upper primary stage (classes VI-VIII)
from 12.7 per cent to 58.5 per cent over the
same period. The percentage of giris' enrolment
to total enrolment has increased from 28.1 in
1950-51 to 43.6 in 1997-98 in the primary stage
and increased from 16.1 to 40.1 over the same m
period in the upper primary stage.
18. The dropout rate of girls is much higher than
that of boys at both the stages. The enrolment1
of SCs and STs has increased considerably at
the primary stage. The share of enrolment of
SCs has increased from 17.1 per cent in 1986 '
to 19.6 in 1993 at primary stage and from 14.7
to 15.6 over the same period at upper primary
stage. Similarly, the share of enrolment of STs
has increased -rom 7.8 per cent in 1986 to 9.1
per cent in 1993 at primary stage and from 5.1 .
per cent to 5.9 per cent over the same period
at upper primary stage. Moreover, substantial
increase in the share of girls' enrolment oelonging
to these communities has also taken place.
19. In order io improve the interna! efficiency
and minimum level of learning, at the school
level, the Central government has ta^en three
important ini iatives since 1993, namely. Area
Intensive Prorramme for Educationally Backward
Minorities (AiPEBM), District Primary Education
Programme iOPEP) and National Programme
of Nutritional Support to Primary Education
(NPNSPE) (Mid-Day Meal Scheme).
20. The government proposes to formulate and...
implement plans to gradually increase the_
governmental and non-governmental spending on educatior up to 6 per cent of GDP. Planning .
Commission has set up an Expert Commiitee to /
assess the current status of expenditure on '
education, both in public and private sector.Them
central plan > allocation on education has been /
enhanced from Rs.3350 crore in 1997-98
to Rs.424 5 ci ore in 1998-99 (BE). In order to...-
151
Initiate plan to implement the provisions of
Rational Agenda for Governance, the current
■', year's budget (plan and non-plan) provides for
■ nearly 50 per cent increase I.e.from Rs.4716
■i Cfore in 1997-98(RE) to Rs.7047 crore in 1998-
99(BE)-
' population and Family Welfare
21.
An enhanced outlay of Rs.2489 crore has
been provided for various programmes of Family
’ .Welfare in 1998-99 (BE) as compared to Rs. 1829
crore in 1997-98 (RE) registering an increase cf
, 36 per cent. Through various on-going family
welfare programmes, government has been able
to achieve a decline in fertility rate from 4.5 in
; 1981 to 3.5 in 1995. The crude birth rate (CBR)
declined from 33.9 per thousand population in
to 27.2 in 1997. The crude death rate
]BdR) also declined from 12.5 to 8.9 per
thousand population ever the same period. As a
result, the natural rate of growth of population
further declined to 1.83 per cent in 1997
compared to 1.35 in 1996. The Eighth.Plan target
of CBR of 26 has been achieved by major states,
except the States of Assam, Bihar, Haryana,
Madhya Pradesh. Orissa, Rajasthan and Uttar
Pradesh.
22. The Reproductive and Child Health (RCH)
. .programme has been introduced during the Ninth
Plan. Under the RCH Programme, several new
schemes for improving quality and coverage of
services are under implementation. The
replacement of the system of setting
contraceptive targets from above by
decentralised planning based on community
needs assessment has been well received in all
parts of the country.
23. The Pulse Polio Immunisation has been
continued for the fourth year during 1998-99
with the objective of eradicating polio. Two
supplemental doses of oral polio vaccine were
administered on 6.12.1998 and on 17.1.1999.
Medical and Health Care
24. Central sector emphasis continues to be on
control and eradication of communicable and
non-communicable diseases like Malaria,
Tuberculosis, Leprosy,.AIDS, Blindness, Cancer,
etc. and various programmes to this effect are
being implemented with the World Bank
assistance. Measures for upgradation of central
and state levels organisations for drug quality
control and food safety and strengthening of
medical store organisations are also being
undertaken. It may be observed from Table 10.11
that there had been marked expansion in
infrastructure for health services since 1951.
25. An enhanced outlay of Rs.1195 crore
including Rs. 50 crore for Indian System of
TABLE 10.11
Expansion of Health Services
(in numbers)
1S51
Item
1. Medical Colleges "
2- Hospitals "
3- Dispensaries "
4 Community Health Centres ©
5- Primary Health Centres ©
.®- Sub-centres ©
J- Hospital bedsfall types)"
8. Doctors "
'8- Dentists ”
'0. Nurses ”
Na
©
7 ' »
1 P
28’
2694
6515
0
725
-
117178
61840
3290
16550
1961
1971
1981
1992
1996
1997
60
98
3362
111
6804
146
13692
165
15097#
12180
16751
217
5740
51405
569495
268712
8648
154280
27403
165
NA
NA
3094
9406
0
2565
230000
83756
3582
35584
Not available
Pertains to 1350
As on December 31
As on March 31
As or. 31.12.95
Provisional
: Ministry of Health and Family Welfare.
0
5112
28489
348655
151129
5512
80620
2186
20701
131370
834650
395851
11300
385410
28225#
2572
21917
134931
870161#
375291(P)
23953#
565696
2628
22446
136379
NA
484401(P)
NA
NA
'•■192
••
•
. ^Medicine and Homeopathy has been provided
in 1998-99 (BE) as compared to Rs.918 crore in
• 1997-98 (RE). The Department of Indian System
of Medicine and Homeopathy (ISM&H) has
identified thrust , :eas namely improving the
standards of education and strengthening the
educational institutions, standardisation of drugs
and quality control, enhancing the availability of
raw material, information, education and
communication, etc. Specialty clinics of Unani,
Ayurveda and Homeopathy have been set up in
major allopathic hospitals like the Ram Manohar
Lohia Hospital and Safdarjung Hospital
respectively.
Development of Women and Children
26. The strategy of women’s development
comprises social and economic empowerment
^of women through attitudinal change towards
"girl 'child, and education, training, ’employment,
support services and emphasis on women' rights
and law. Similarly, for the child, various schemes
to raise nutritional and early childhood care and
pre-school educational status have been
implemented. A provision of Rs.1226 crore has
been provided in 1998-99 (BE) as against
Rs.1026 crore in 1997-98 (RE), for
implementation of various welfare schemes for
women and children.
27. Under the scheme of Balika Samridhi
Yojana launched on October 2, 1997 with a
specific objective to encourage the enrolment
and retention of girl child in the schools,'the
mother of a girl child bom on or after August 15,
B 1997 in a family, below the poverty line in rural
and urban areas is given a grant of Rs.500
besides a scholarship for education of the girl
child when she attends school. 12 lakh girl
children were benefited during 1997-98. A sum
of Rs.60 crore has been provided in 1998-99
(BE) under the scheme.
28. The scheme for school drop-out adolescent
in the age group of 11-18 year is being
implemented in 507 blocks of ICDS projects. So
far four lakh adolescent girls have been benefited.
Six projects under Support to Training and
Employment
Programme
(STEP) were
sanctioned to benefit 53325 women during 199798. A sum of Rs.16 crore has been provided
during 1998-99 (BE).
29. Under Indira Manila Yojana (IMY) for
empowerment of women, 28000 small
homogeneous women's groups were formed up' 'Y
to 1997-98. A sum of Rs.10 crore has been ' Y
provided in 1998-99 (BE). The scheme of Mahila '
Samridhi Yojana (MSY) to inculcate habit of ' '■
saving among rural women is being revised and ■’
merged into IMY to have an integrated package ■ ■
of 5 components including formation of viable ''
women's group.
-jYT
30. Under the scheme of Employment and'
Income Generating Training-cum-Production
centres, partially funded by Norwegian Agency
for International Development and Cooperation A '
(NORAD), 178 projects were sanctioned to '
benefit 36095 women in 1997-98. A budgetary!;.
provision of Rs.18 crore has been provided in ’.
1998-99 (BE) under this programme
31. The Rashtriya Mahila Kosh (RMK) set up for'meeting the credit needs of the poor women YT
has sanctioned loan to 250312 women during 1997-98. Under the scheme of Short Stay Home ;-~~
for Women and Girls, there are 361 short homes .
running in the various parts of the country tobenefit 10830 women. Since inception, 811'9:working women hostels have been sanctioned '■
to benefit 56974 women for single, divorced, :
married and widowed women who migrate to
towns and cities in need of employment.
32. For the welfare and development of the
children, the Integrated Child Development
Services (ICDS) scheme aims to provide an-.
integral package of services of health check- j.
up, immunisation, supplementary nutrition, \
referral services, pre-school education, nutrition
and health education to children, pregnant ;
women and nursing mothers. The scheme which V,
started in 1975-76 on an experimental basis in
33 projects has been extended to 5614 centrally. isponsored ICDS projects of which 4200 projects .
are now' operational. The programme benefited ,)
3.81 million expectant and nursing mothers and j
21.0 million children under six years of age. A..
sum of Rs.603 crore has been provided in 1998-r^
99 (BE). Apart from ICDS the other programme -:
for child development are day-care centers for y
children below five years belonging to weaker.;
sections of the society, Balwadi Nutrition ;
Programme, Early Childhood Education and
National Institute of Public Cooperation and Child;-|.
Development.
..<9
y-T
Welfare of Weaker Sections
33. The National Scheme for Liberalization and.-;
Rehabilitation of Scavengers aims to provide-
153
s ’ alternate dignified and viable occupation to each
■ S scavenger and his/her dependents. The scheme
' T which was modified w.e.f. 1.4.1996, inter-alia.
T- includes TRYSEM norms for training, release of
'■'-'centra! assistance direct to Scheduled Caste
Development Corporation and adoption of cluster
-.■ approach in the training and rehabilitation
■programme. The National Safai Karamchari
Finance and Development Corporation set up in
January 1997 provides
loans for higher
. education to students from safai karamchari
community besides providing assistance in self-"
employment ventures and technical and
- entrepreneurial skills.
34. The National Commission for Minorities,
reconstituted w.e.f. 26.11.1996 to focus on
• Effective implementation of 15 point programme
•. for the welfare of minorities, has constituted a
High Powered Study Committee for socio
economic conditions of minorities in India. The
Central Government has raised equity share
towards National Minorities Development and
Finance Corporation from 25 per cent to 60 per
cent.
35. The welfare of STs and SCs is being closely
monitored by the state governments through the
Special Component Plan (SCP) and Tribal Sub
Plan (TSP) with the support of special central
assistance provided by the Central government.
The consolidated achievements during the Eighth
Plan (1992-97) had been 51.53 lakh ST families
.^against the target of 49.78 lakh families. The
target for 1997-98 was fixed as 10.97 lakh ST
families out of which 9.86 lakh ST families were
assisted. A target of 11.01 lakh families has
been fixed for 1998-99 out of which 4.13 lakh
ST families have been assisted up to 30.11.1998.
On going schemes like pre-matric and postmatric scholarship and providing hostel facilities
■ .. to SC boys and girls continued to be operative
. - in 1998-99.
’
36. In addition to various schemes for the
development and growth of welfare of disabled
persons, a national centre for drug abuse
J
prevention has been established. A sum of
4'
Rs.1539 crore has been provided in central
.4 . V sector plan during 1998-99 (BE) for various
‘.' J
schemes of welfare of weaker sections, of the
-.7
society including minorities, persons with
'7-4 . disabilities and others covered under ..Social
^.defence.
Housing
37. The Central government provide policy
guidelines for housing programmes for various
target groups and supplements the efforts of
the state governments by implementing certain
central sector/centrally sponsored schemes
mainly for the benefit of urban poor.
38. A new National Housing and Habitat Policy
1998 has been formulated which was approved
and laid before the Parliament on 29.7.1998.
The objectives of the policy are to facilitate
construction of 20 lakh dwelling units each year
with emphasis on the poor. Out of 20 lakh
additional houses, 7 lakh houses would be
constructed in urban areas and 13 lakh in rural
areas. HUDCO is expected to meet more than
55 per cent of the target i.e. 4 lakh units and the
balance 3 lakh units per year will be met by
other housing financial institutions recognised
by the National Housing Bank, corporate sector
and cooperatives.
39. A scheme of Night Shelter and Sanitation
Facilities for the urban footpath dwellers is being
implemented as a centrally sponsored scheme
in the metropolitan and other major urban
centres. 61 schemes benefiting more than 40000
footpath dwellers have been sanctioned by
HUDCO as on 31.10.1998 in various parts of
the country since April 1991. A sum of Rs.1600
crore has been provided as central outlay in
1998-99 (BE) for rural housing with a target to
construct 923908 houses under the Indira Awas
Yojana (IAY) in the current year. The existing
housing programme of the IAY for construction
of new houses free of cost for the target group
below the poverty line comprising SCs/STs, freed
bonded labourers and also non-SC/ST families
will continue. In addition, a new component for
upgradation of kutcha and unserviceable houses
will be introduced.
40. The Government has promulgated an
ordinance to repeal the Urban Land (Ceiling and
Regulation), Act 1976 (ULCRA). It will be
implemented in all Union territories and two
states of Haryana and Punjab. This was aimed
at speedy growth in sectors like housing and
transport, land assembly, development and
..
disposal of excess land would be facilitated. The
(f
Delhi Rent Act, 1995 was enacted . for ■ .s
overcoming the shortcomings in the existing Rent
Act but after taking into consideration various
representations,, the government decided to ■;
amend'the Act to'make it more' acceptable and
Y,
u .
I!
■ '■...X
C-O'Vl H •
UNDP’s Human Development Index
A Computation for Indian States
A K Shiva Kumar
An attempt to construct the Human Development Index (HDI)
for 17 Indian states and to rank these states with the countries for
which the HDI has been computed in the UNDP’s Human
■Development Report 1990.
THE HUMAN DE VELDPMENT REPORT
1990 of the UNDP argues for viewing
development not merely as an expansion of
Income and wealth, but as a process of
enlarging people’s capabilities.1 The Report
emphasises three elements of living stan
dards: longevity, literacy, and, what may be
. called, a measure of necessary income.
Longevity as an indicator of human develop
ment captures several aspects of welfare
because of its close correlation with nutri
tion, health, and other important biological
and social achievements. The relevance of
literacy to human development, is of course,
self-evident. The third element of human
development.discussed in the Report is the
command over resources needed for a de
cent living, covering those aspects’of living
which are not well represented by life expec
tancy or literacy. In the absence of more
Table I: Classification of 130 Countries According to HDI, 1987
HDI
^cow (HDI below 0.500)
Lowest
44 countries including India 0.116 Niger
Medium (HDI 0.500 to 0.799) 40 countries including
China, Sri Lanka
High (HDI above 0.800)
■ 46 countries including
. Cuba, Costa Rica
0.501 Egypt
Highest
0.489. Morocco
0.790 Albania
0.800 Malaysia 0.996 Japan
_
I Source-. From data in UNDP (1990).
7-C
specialised indicators of this command over
resources, a practically useful indicator is per
capita income.2 But incomes are only
means of good living and must not be con
fused with it. The Report takes the logarithm
of per capita income to reflect the conver
sion of income into good living.3 The
Report uses data on life expectancy, literacy,
and “income for a decent living standard’’
to construct a composite Human Develop
ment Index (HDI) for each country.4
This note corfstructs the HDI for 17
Indian states for which data are available
and ranks the states along with the countries
for which the HDI for 1987 has been com
puted and presented in the Report.
Construction of HDI
In order to construct the HDI, the first
step is to specify a minimum value (the max
imum deprivation set to one) and a desirable
or adequate value (no deprivation equal to
zero) for life expectancy (X, ), literacy (X2j
and the logarithm of real GDP per capita
(X3). A life expectancy at birth of 78 years
(which is the figure for Japan) has been
taken as the maximum value, and of 42 years
(the figure for Afghanistan, Ethiopia and
Sierra Leone) as the minimum value. The
lowest value for adult literacy’is 12 per cent.
(the rate in Somalia), and the maximum is
100 per cent. The minimum value of pur
chasing power adjusted GDP per capita is
Zaire’s S220 (log value 2.34). The average of-
Table 2: Basic Daia Relating to 17 Indian States.
Andhra Pradesh
Assam*
Bihar
Gujarat
Haryana
Himachal Pradesh
Jammu and Kashmir
Karnataka
Madhva Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
India
Females
1,000 Males
1981b
Adult Literacy Ratcd (Percentage)
Per Capita Income0 •'
Life Expectancy 1981-86°
Per Capita Net State Real SDP
1971
1981 •
Males.
Females Combined
1987
(Projected) Domestic Product at Per Capita
1970-71 Prices (Rupees) ’ . 1987
1980-81
1986-87 . (PPP S) .
975
901
946
942
870
973
892
963
1 032
’ 941
937
981
879
919
977
885
911
933
56.1
52.7
55.2
55.3
61.4
60.0
■ 52.0
52.9
58.3
59.6
60.2
65 2
53.2
59.9
54.1
64.3
54.8
58.3
51.1
57.0
55.6
61.1
69 9
51.5
60.7
51.9
64.3
55.4
57.9
46.9
56.3
56.4
28.3
58.0
36.4
52.4
54.1 “ •23.5
56.8
42.0
29.7
. 60.6
56.6
32.3
' 21.1
568
35.9
60.6 .
69.1
67 6
' 52.4
■ •'26.6
44.9
60.2
<31.0
.53.0
35.2
64.3
220
55.1
, 42.9
58.1
24.5
42.1
5^6
.40.5
56.0
34.0
32.5
43.6
29.4
48.3
39.2
■ 43.6
29.9
43.0
78.1
35.6
.51.8
■ 38.7
42.6
28.2
50.4
30.8
48.1
40.8
35.3
48.5
33.6
52.5 •
46.2
52.2
36.9
48.0
84.1
-42.4 -<
56.5
44.2 .
47.7
32.7
55.4
35.3
. . 53.3
45.5
• ;
■
647
558
441
904
1,060
711
642
087
621
516
. 957
477
1,354
535
584
519
797
698
758
605
• 482
860
1,233
855
684
799
639
583
1,039
535
1,652
666
828
607
860
812
983
785
.' 625 .
1,116
1,600
1,109
887
1,037
829
•
756 /
1,348
694
2,143 864
1,074
787
1,116
1.053
Notes and Source-, a Assam’s population figure for 1981 is an official projection obtained from Government of India (1987). The national growth
rate in adult literacy between 1971 and’1981 has been used to project Assam’s adult literacy rates for 1981 and 1987.
b Figures on sex ratios are from the Census of Indi^J98I.
: / .
c Projections of life expectancy for males and females for all slates except Himachal Pradesh and Jammu and Kashmir are based on the ‘Report
of the Expert Committee on Population Projections’ and quoted in Government of India (1989). Figures of life expectancy in Himachal Pradesh
and Jammu and Kashmir are based on the gains in life'expectancy Vecocded between 1970-75 and 1976-80 given in Office of the Registrar
General (1985).
d Rates of adult literacy have been computed from the Census of India, 1971 and 1981.
■ z -■
c Figures of real per capita SDP for the states are from Directorates of Economics and Statistics, and are quoted in Government of India (1989).
F^onnmic and Politic?!
October 12, igoj
2343
ficial poverty line in nine industrial coun
tries, $4,861 (log value 3.68) has been taken
a desirable upper value.
The deprivation indicator, 1 for the j'h
country with respect to the invariable is
then defined as:
(max X,j - X,,)
i
(max X ■ - min X )
j
j
A simple average of three indicators is then
taken:
1, = s’,
1/3
The final step is to measure the as one minus
the average deprivation, index.
(HDI^ = (1-1.)
The Report has calculated the HDI in 1987
for 130 countries and grouped them as
shown in Table 1.
The HDI for 17 Indian states has been
constructed for 1987 using the same
methodology as in the Report. It would have
been useful to calculate the HDI for the
(smaller states and union territories in India.
However the absence of disaggregated data
on health and life expectancy for the union
territories and for the states of north-east
India prevents the computation of the HDI
for these regions. The selected states are
those for which data on the three indicators
are available as shown in Table 2. The
following procedures have been adopted for
estimating the values of life expectancy, adult
literacy and per capita inconies-for 1987:
Life expectancy at birth: The report on the
expert committee on population projections
gives projections of the life expectancy for
males and females in different states of India
for 1981-86. These life expectancy figures for
1981-86 have been weighted by the 1981 sex capita net state domestic product measured'
ratios to arrive al an estimate of the com at constant prices to the national per capita
bined life expectancy figure for each state. income at constant prices in 1987 has been
Adult literacy rate: The figures on adult * computed.
literacy for 1971 and 1981 are available from
(2) India’s real per capita GDP.(in PPP $)
the Census of India. The annual growth estimated at $1,053 for 1987 has beep multi
rates in adult literacy rates between 1971 and plied by K for each state to arrive at an
1981 have been used as the basis to project estimate of the level of the state’s real per
adult literacy rates for 1987 for each state. capita GDP.
For Assam, where the Census was not con
The basic data used for.arriving at the
ducted in 1981, the average national growth estimates of the thrce.indicators for 1987 are
rate in literacy between 1971 and 1981 has presented in Table 2. Table 3 shows the com
been used for making the projections.
putation of HDIs for the 17 Indian states.
Per capita GDP. According to the Human The HDI index for India as a whole is 0.439'
Development Report, India’s real GDP per placing it in the category of low HDI coun
capita (in PPP $) for 1987 was $1,053. The tries. The diversity within India is con
value of each state’s real GDP per capita has spicuous. The HDI ranges from a low of
been estimated as follows:
0.292 in Uttar Pradesh to a high of 0.651 in
(1) For each state, Kn, the ratio of'ihc per
Kerala. Only four Indian states out of the
Table 4: Ranking of 17 Indian States by HDI, 1987
Low HDI
1 Uttar Pradesh
2 Bihar
3 Madhya Pradesh
4 Rajasthan
5 Orissa
6 Assam
7 Jammu and Kashmir.
8 Andhra Pradesh
9 Himachal Pradesh
10 Gujarat
11 West Bengal
12 Karnataka
13 Tamil Nadu
Medium.HDI
14 Haryana
15 Maharashtra
16 Punjab
17 Kerala
“
HDI
Per Capita Net State Domestic Product
• ' at 1970-71 Prices '
• :;
Rs
Rank
0.292
0.306
0.344
0.347
0.348
0.372
0.381
0.397
0.462
0.465
0.467
v0.475
0.483 .
'
0.514
0.532
0.586
0.651
607
482
583
666
535
605
• ' 684
758
.855
860
860
. 799
828 '
. 1.,.
1,233
1,039
1,652
639
.
......... 5
1
. 3 ’ '
• •
T
2 ■' • ■
4
8
9
, 12 ' .
' 13
13
10
•
11
16
15*
17 ’
6
Table 3: Human Development Index ior 17 Indian States. 1987
Life
Adult Real SDP Per Capita
Expect . Literacy
(PPP S)
Life
ancy
Rate ~ Actual
Expect
Log
at Birth (Per Cent)
ancy
1 Uttar Pradesh
2 Bihar
3 Madhya Pradesh
4 Rajasthan
5 Orissa
6 Assam3
7 Jammuu and
Kashmir
8 Andhra Pradesh
9 Himachal Pradesh
10 Gujarat.
11 West Bengal.
12 Karnataka
13 Tamil Nadu ■
14 Haryana
15 Maharashtra
16 Punjab
17 Kerala
7344
Deprivation
Adult
North
Average
Literacy' Minimum of the
Purchasing Three'
Power
Human
SDP HDI Rank
Develop Per Capita Minus
ment
Rank SDP Rank
1987
Index
49.1
54.1
52.4
55.1
53.0
52.4
35.3
33.6 ■
42.4
32.7
44.2
48.5 ■
787
625
756
864
694
785
• 2.90
2.80 .
2.88
2.94
2.84
2.89
0.80
0.66
0.71
0.64
0.69
0.71
: 0.74
0.76
0.66
0.77
0.64
0.59
. 0.59
0.66
0.60
0.56
0.63
0.59
0.708
0.694
0.656
0.653
0.652
0.628
0.292
0.306
0.344
0.347
0.348
0.372
56.8
58.0
56.6
56.8
56.6
60.6
58.1
60.6
60.2 64.3
67.6 '
36.9
35.3
*52.2
52.5
53.3
48.0
55.4
46.2
56.5
47.7
84.1
887
983
1,109
1,116
1,116
1,037
1,074 •
1,600
1,348
2,143
829
2.95
2.99
3.05
.3.05
3.05
3.02
3.03
3.20
3.13
3.33
2.92
0.59
0.56
0.60
0.59
0.59
0.49
0.56
0.49
0.50
0.38
0.30
0.72
0.74
0.55
0.54
0.53
0.59
0.51
0.61
0.50
0.6(5
0.18
0.55 .
0.51
• 0.47
0.47,
0.47
0.50
0.49
0.36
0.41
0.26
0.57
0.619
0.603
0.538
0.535. .
0.533 ‘
0.525
. 0.517
0.486 :
0.468 '
0.414
0.349
6.381
0.397
0.462
0.465
0.467
0.475
0.483
0.514
0.532
0.586
0.651
.
'
<5"-'- ’ ‘ ■ 4 ’
1 —1 • •
0
3
7
3
2
3 ■
4
-2
8
9
• 12
V.13
13
10
'. 11
' ' 16
15
,17
6
Fcoribrh’c and Po!:‘;ca1 Weekly
1
T
3
3'
2
-2 '
•' -2
2
0
1
-11
Octcbe*- 12, 199!
Table 5: HDI of 17 Indian States and of Countries in
•Low’ and ‘Medium’ Categories
Country
HDI
Low HDI
1 Niger
Mali
3 Burkina Faso
4 Sierra Leone
5 Chad
6 Guinea
7 Somalia
8 Mauritania
9 Afghanistan
10 Benin
11 Burundi
12 Bhutan
13 Mozambique
14 Malawi
15 Sudan
16 Central- African Republic
17 Nepal
18 ’Senegal
19 Ethiopia
Uttar Pradesh
20 Zaire
21 Rwanda
22 Angola
Bihar.
23 Bangladesh
24 Nigeria
25 Yemen Arab Republic
26 Liberia
27 Togo
. Madhya Pradesh
Rajasthan
Orissa
28 Uganda
29 Haiti
30 Ghana
31 Yemen, PDR
Assam
Jammu and Kashmir
32 Cote d’Ivoire
33 Congo
Andhra Pradesh
34 Namibia
35 Tanzania
36 Pakistan
37 India
38 Madagascar
Himachal Pradesh
Gujarat
West Bengal
39 Papua New Guinea
0.116
0.143
0.150
0.150
0.157
0.162
0.200
0.208
0.212
0.224
0.235
0.236
0.239
0.250
0.255
0.258
0.273
0.274
0.282
0.292
0.294
0.304
0.304
0.306
0.318
0.322
0.328
0.333
0.337
0.344
0.347
0.348
0.354
0.356
0.360
0.369
0.372
0.381
0.393
0.395
0.397
0.404
0.413
0.423
0.439
0.440
0.462
0.465
0.467
0.471
Country
HDI
40 Kampuchea, Dem
0.471
41 Cameroon
0.474
Karnataka
0.475
42 Kenya
0.481
43 Zambia
0.481
Tamil Nadu
0.483
44 Morocco
0.489
Medium HDI
45 Egypt
0.501
46 Lao PDR
0.506
Haryana
0.514
47 Gabon
0.525
0.532
Maharashtra
48 Oman
0.535
49 Bolivia
0.548
50 Myanmar
0.561
51 Honduras
0.563
52 Zimbabwe
0.576
. .
53 Lesotho
0.580
Punjab
0.586
54 Indonesia
0.591
55 Guatemala
0.592
56 Viet Nam
0.608
57 Algeria
. 0.609
58 Bostswana
0.646
59 El.Salvador
0.651
Kerala
. 0.651
0.657
. 60 Tunisia
61 Iran, Islamic Republic
.
. .0.660
62 Syrian Arab Republic.'
0.691
63. Dominican Republic
0.699
64 Saudi Arabia •*. •
0.702
65 Philippines
......
. 0.7J4
66 China
0.716
67 Libyan Arab Jamahiriya
0.719
68 South Africa
0.731
69 Lebanon
0.735
70 Mongolia
0.737
71 Nicaragua
0.743 •
72 Turkey
0.751
73 Jordan
0.752
74 Peru
0.753
75 Ecuador
0.758
76 Iraq
0.759
77 United Arab Emirates .
0.782
78 Thailand
0.783
1
79 Paraguay
0.784
80 Brazil
0.784
81 Mauritius
. >
0.788
82 Korea, Dem Republic / . - .0.789
83 Sri Lanka
. 0.789
.84 Albania
'■
0.790
Zaire, ranked 19th and 20th respectively. The
HDIs for Bihar, Rajasthan, Madhya Pradesh
and Orissa are in the same region as
Bangladesh, Nigeria, Uganda, Haiti, and
(jhana. Kerala, the state with the highest
HDI in India, comes between Botswana and
Tunisia, ranked 58th and 60th in the world
respectively. Kerala’s achievements are excep
tional given that the level of human develop
ment has been achieved despite low per
capita incomes. While 101 countries had per
capita incomes that were higher than
Kerala’s, there were only 51 countries in the
world with a higher life expectancy and 53
countries with a higher adult literacy than
the levels achieved in Kerala.
Notes
[I am grateful to Lincoln Chen, T N Krishnan*
V K Ramachandran, and Amartya Sen for
comments.]
'
* 1 UNDP [1990]. On the concept of capabilities,
see Sen [1985] and Dreze and Sen [1989] ’
2 “The • third key component of human
development—command over resources
needed for a decent living—is perhaps the
most difficult lb measure simply. It requires
data bn access to land, credit, income and
other resources..But given the scarce data on
many of these variables, we must for the time
being make the best use of an income in
dicator’; [UNDP, 1990,'p 12].
3. “A further consideration is that the indicator
should reflect the diminishing returns to
transforming income into human capa-.
bilities. In other words, people do not need
excessive financial resources to ensure a de
cent living. This aspect was taken into ac- . ”
.. count by using the logarithm of real GNP
per capita for the Income indicator” [UNDP,
1990,p 12].
4 The HDI is not conceptually or statistically
' equivalent to the Physical Quality of Life In-.
dex (PQL1). The PQLI treats development....
as achieved well-being. The focus of the
Report is on socioeconomic development, .
with development viewed hot as an expan
sion of commodities andwealth, but as the
widening of human choices; For a further
discussion on the differences, see UNDP
[1990], Technical Notes No 1.
References
Dreze, Jean and Sen, Amartya (1989) ‘Hunger
Source: Table 1, Human Development Index, UNDP Report’, and from Table 4 of this note.
and Public Action’, Clarendon Press,
Oxford.
17, Haryana,. Maharashtra. Punjab and
better than Somalia, Ethiopia, Nigeria and Government of India (1987), Health Informa
tion of India, 1987 (Ministry of Health and
Kerala, had an HDI in the medium category.
Uganda, but not as well as.Botswana, Zim
Family'Welfare, New Delhi).
Ranking the states of India along with the
babwe, Zambia and Egypt. If we compare —(1989), Family Welfare Programme in India:
countries ranked in the Report makes for an
India’s HDI with, those of developing coun
Year Book 1987, (Ministry of Health and
interesting comparison.
tries in Latin Aperic-a, we find that with the
Family Welfare, New Delhi).
.
exception of Haiti, all other countries in . Office of Registrar General (1985), SRSAbridgIndia as a whole ranks 37th in terms of
Latin America have a higher HDI than
ed
Ltfe-Tables
1976-80
(Occasional
Papers
HDI. Among the Asian countries! it has a
India.
No 1 of 1985, Census of India, New Delhi).
higher HDI than Afghanistan, Bhutan,
Sen, Amartya (1985), . Commodities and
Nepal and Bangladesh, but its HDI is much
There were only 19 countries with a lower
Capabilities, North-Holland, Amsterdam. .
lower than Sri Lanka, Thailand, China and
HDI than Uttar Pradesh, the state with the United : Nations . Development Programme
Philippines. If India’s HDI is compared with
lowest HDI in India. Uttar Pradesh with an
. (1990), Human Development Report 1990,
those of African countries, India has done
HDI of 0.292 lies between Ethiopia and
Oxford University'Press, New York.
Economic and Political Weekly
October 12, 1991 .
2345,
Poverty Estimates and indicators
Importance of Data Base
M H Suryanarayana
This paper emphasises the importance of the data base in any discussion of poverty and identifies the major data
gaps for policy studies. Beginning with the identification of the poor based on a measure of standard of living and
a minimum norm and going up to thefinal stage ofpolicy prescription, an awareness ofthe data base and the constraints
it imposes on interpretations is crucial. Conventional approaches to poverty identification and measurement presuppose
a stationary economy. But in an economy subject to changes in institutional parameters involving increasing
commercialisation of product markets and growing casualisation of labour markets, as in India in recent years, the
conventional approach can yield misleading results and policy prescriptions.
I
Introduction
IDENTIFICATION of the poor and
estimating the magnitude of poverty has
received considerable attention during the
reform programme for stabilisation and
structural adjustment in different countries
all over the world. Identification of the poor
is importantforeffective targeting of various
poverty alleviation programmes and safety
nets so as to ensure their budgetary cost
effectiveness in minimising the social cost
of the reform programme. The magnitude
of poverty has to be estimated for assessing
the budgetary implications of the various
safety net programmes and for evaluations
of these programmes. As part of its efforts
to minimise the social costs of the reform
programme and the associated budgetary
costs, the Indian government is also seized
of the problem of identifying the poor, the
magnitudeof poverty and theircost-effective
solutions. Ever since the reform programme
began in July 1991, there has been
considerable debate on efficient targeting of
the various poverty alleviation programmes
and safety nets, the public distribution system
in particular.
As for any empirical policy study, effective
solutions for the issues raised above
presuppose an efficient information set. The
information requirement varies with the
questions raised and the proposed objectives
of policies. If the questions are who are the
poor, what is the magnitude of poverty and
if the objective is to examine the welfare
levels and consequences of policy changes,
then a measure of economic (material)
welfare or standard of living would be the
relevant variable. If the question is what
makes the people poor and the purpose is
to formulate efficient policies for poverty
alleviation then causes and correlates of
poverty assume relevance. In either case, identi
fying the poor assumes a lot of importance.
Poverty is generally defined as the inability
to secure a minimal level of economic
Economic and Political Weekly
welfare. This raises the questions: How to
measure economic welfare? What constitutes
a minimal norm or poverty criterion? The
answers for these questions depend, in tum,
upon the scope for policy action as dictated
by government’s resource constraint, social
consciousness and academic perception of
the problem. This is because economic
welfaremaybemeasuredintermsofdifferent
variables: per capita income, per capita
consumer expenditure, per capita food
consumption, Engel ratio, calorie intake,
anthropometric measures or basic needs.
Poverty groups identified, estimates of their
magnitude and policy measures designed
will vary depending upon the economic
welfare measure chosen for defining
poverty.1 All through the post-independence
period the focus of academic research as
well as government policy reports in India
has been on absolute poverty or the
proportion of the poor population. Absolute
poverty is estimated on the basis of private
consumer expenditure with reference to a
poverty line defined on a normative caloric
intake basis. Even when we address such a
narrow set of questions, the scope for
misinterpretation of statistical measures by
disregarding data considerations is
considerable and hence, the question of data
base is quite important.2 The most widely
used data base for poverty studies in India
is the household consumer expenditure data
collected by the National Sample Survey
Organisation (NSSO). Conceptually,
consumer expenditure as a measure of
standard of living has the advantage that it
is amenable to welfare interpretations subject,
of course, to the limitation that it docs not
consider the welfare derived from leisure
and pure public goods. In a predominantly
monsoon dependent agricultural economy
like India where agriculture constitutes the
source of livelihood of about two-thirds of
the population, consumer expenditure
measure has an additional advantage that it
may not reflect the periodic variations in
income. But, when it comes to empirical
Special Number September 1996
verifications it bristles with many problems.
Empirically, conclusions from comparison
of real consumer expenditure levels would
vary depending upon how different items of
consumption are valued. In a country where
income distribution is not optimal, market
price is not a proper welfare measure. In
addition, as regards the NSS data base, there
areanumberofstatisticalquestionsregarding
its reliability. There are other irsues like
sample data presentation, tabulation and
institutional parameters which have been
ignored in the poverty studies resulting in
misleading inferences.
Even though the government has been
carrying out poverty alleviation programmes,
much remains to be done on the poverty
identification and measurement fronts for
policy operational purposes. This is so in
spite of the policy emphasis on cost
effectiveness of the poverty alleviation
programmes.’ Cost-effectiveness can be
achieved either by maximising benefits to
the poor for a given budgetary cost or by
minimising the cost subject to a poverty
alleviation target. One approach to achieve
cost-effectiveness is by ‘better targeting’ of
(Programmes, i e, by maximising the coverage
of the poor and minimising the leakages of
benefits to the non-poor. How do we pursue
such goals and formulate different policy
actions? Different country experiences
including that of India have shown that
identifying the poor for purposes of targeting
by ‘means-test’ is administratively costly.
This is because people have an incentive to
under-report incomes so as to secure more
benefits than they are eligible for/ If so,
targeting is desirable so long as the
administrative cost of identifying the poor
does not exceed the saving involved in
excluding the non-poor. This problem is
sought , to be overcome by ‘indicator
targeting’, i e, by making eligibility for
benefits conditional on correlates ofpoverty,
such as land holding, caste, or place of
residence.5 In that case, we need to ask, what
are the appropriate indicators or correlates
2487
of poverty and their implications? What are interviews of households. The households
the data requirements in such a context?
are selected following simple random
This paper addresses some of these sampling. The sample design is stratified
questions pertaining to the two themes of and two-stage in both rural and urban sectors.
poverty estimates and targeting of The stratification of rural areas is with respect
programmes. The paper is organised as to homogeneity of population density,
follows. Section 2 deals with the first set of cropping pattern, etc, while that of urban
issues on data base for estimating the areas is with respect to population sizes of
magnitude of poverty in India, re, consumer towns and cities. The first stage units are
expenditure data collected by the NSSO, its villages in the rural sector and urban blocks
limitations, scope for wrong inferences and in the urban sector. The households constitute
the need for an integrated approach. Section the second stage units in both the sectors.7
3 deals with the shift in emphasis towards The NSS conceptof consumer expenditure
targeted welfare programmes for the poor includes all the non-productive expenditure
and their implications for methodology and incurred by the households. It includes
data base. The final section sums up the paper. consumption out of home-grown produce,
gifts, loans, etc. Data on perquisites like
n
food in the employer’s house are not included
Data Base for Poverty Estimates
in the NSS estimates of consumption of the
employee households. This must have
Poverty analysis in India is generally resulted in underestimation of foodgrain
carried out using private household consumer consumption by the employee households
expenditure as a measure of standard of who in the rural sector are generally poor
living." Income/consumer expenditures are landless households. In-kind wage payments,
sufficient measures so long as they include including prepared food at the employer’s
own production. Even then they capture home, used to be quite common during the
only certain economic aspects of welfare.
1950s and 1960s. The consequent
They do not take into account other underestimation of foodgrain consumption
dimensions of welfare like health, life by the labour households must have resulted
expectancy, literacy, access to safe drinking in overestimation of poverty, particularly
water, public goods or common property distributionally sensitive measures like the
resources. In fact, these other dimensions are Sen index [Sen 1976] and the Pa measure
important since they do not remain the same [Foster-Greer-Thorbecke 1984], According
over time in the course of economic to the second agricultural labour enquiry of
development. Poverty, defined and estimated 1956-57, agricultural labourers received
using a narrow concept of standard of living, wage payments in kind for about 50 per cent
has its own limitations in the context of a of the mandays worked." As studies [see, for
developing economy.
instance, Vaidyanathan 1986b] have shown,
Much has been written about the there has been increasing landlessness and
adequacies and inadequacies of the NSS casualisation of labour since the mid-1970s,
methodology and their implications for which must have resulted in increasing
estimates of consumption aggregates and market dependence of the poor. Such
distributions [Dandekar and Rath 1971; progressive market dependence of the
Dandekarand Venkataramaiah 1975; Iyengar landless poor must have involved increasing
and Bhattacharya 1975; Kadekodi 1992; monetisation of the labour marketand hence,
Minhas 1988; Mukhetjee 1986; Srinivasan also that of their consumption. As a result.
and Bardhan 1974; Suryanarayana and the extent of underestimation of foodgrain
Iyengar 1986; Tyagi 1982; Vaidyanathan consumption by the poorer households and
1986a, 1988]. Most ofthese studies examine hence, overestimation of poverty must have
thequestionof reliability of theNSS estimates declined. This is one factor that has to be
pf the underlying population parameters by taken into account in interpreting the
different statistical criteria. This section of observed declining trends in poverty
the paper examines the implications of some estimates since the mid-1970s.
of the features of the NSS estimates of
The NSS does not take into account the
consumer expenditure for poverty estimates imputed rental value of owner occupied
and their interpretation in the course of houses. This would affect estimates of
development.
consumer expenditure inequality measures
The NSS is a socio-economic enquiry and hence any estimation of poverty based
carried out in the form of successive rounds. on the inequality parameter." The consumer
The period of enquiry has been varying expenditure data are collected by the NSS
across rounds and has varied from a few using a moving reference period by which
weeks to months. During these rounds, the the interviewing of sample households is
NSS collects information on various socio spread over the year. This introduces a
economic aspects of households, household seasonality bias into the data. Its implications
consumptionbeingoneofthemosti mportant. have been well documented in the literature.
The data are collected on the basis of What is little recognised is that such a
----- '
2488
." ’
a:.*
•
procedure also superimposes price variation
on actual variation in consumption during
a period of rising prices. This will affect
estimates of poverty and inequality based on
consumercxpendituredistribution.Thismust
have been a serious problem for estimates
of poverty and inequality for the 1960s and
1970s when there was considerable inter
year price dispersion caused by restrictions
on inter-state foodgrain movements [Ray
1970], This is one aspect which has not been
taken into account in the studies on trickle
down of agricultural growth in rural India
[Suryanarayana 1996b],
The NSS surveys are generally based on
the consumer expenditure schedules. The
NSS surveys were integrated household
surveys during the 19th to 25th rounds
inclusive. That the NSS used integrated
schedule only for some select rounds has
affected adversely the inter-temporal
comparability of its estimates of consumer
expenditure and hence poverty estimates.
Available evidence indicates that people
under-reported theirconsumption duringthe
rounds covering the integrated household
surveys [Mukherjee and Saha 1981;.
Suryanarayana 1996b]. Accordingly, the
NSS estimates of consumer expenditure for
these years arc substantially lower than the
National Accounts Statistics (NAS)
estimates. This could be one reason why
NSS based poverty estimates show
substantially higher levels of poverty during
the 1960s. Thus, such questions on inter
temporal comparability of data base affect
inferences on trends in poverty.
The data base also gets affected by the
method of valuation. Poverty estimates based
on such data using conventional statistical
deflators may be valid in a given stationary
context but loses its meaning and relevance
when the context itself undergoes a change.
For instance, the NSS distinguishes between
consumption from homegrown stock and
that from market purchases, and values the
former at farm harvest prices and the latter
at market retail prices. In a semi-monetised
economy where a substantial segment of the
rural population consists of subsistence
farmers, this approach may be valid. For
instance, consumption out of home grown
stock accounted for 50.9 per cent of total
food grains consumption in the rural areas
and 7.15 per cent in the urban areas during
1964-65 [GOI 1972:4], In the development
context, as observed in recent years in India,
institutions, labour markets, and production
conditions change. As already noted, with
progressive market dependence of the poor
there must have been increasing monetisation
of consumption. By the NSS method of
valuation, the same amount of physical
consumption would get exaggerated in
nominal terms due to changing price weights.
whereby an increasingly larger part of
Economic and Political Weekly
Special Number September 1996
consumption is valued at (generally) higher
retail prices. On the other hand, with the
green revolution, there have been changes
in the crop composition of foodgrain output
from the cheaper inferior cereals to superior
but costlier superior cereals like rice and
wheat involving a decline in per capita
availability of coarse cereals necessitating
a shift in consumption patterns in favour of
the latter. In such a context, inferences on
trends in measures of poverty based on
estimates of consumer expenditure obtained
by dual valuation can be misleading [see,
for details, Suryanarayana 1995].
Such questions are fundamental for any
empirical study on poverty during the
development process. Few studies have
addressed these questions or cared to relate
them to their main query. Both are needed
because a poverty measure, like any other
descriptive statistical measure, is a
conditional measure, i e. conditional on the
prevailing structural features like market,
production and other institutional parameters.
Unless one considers these conditional
parameters, one is likely to end up making
wrong inferences. The following is an
example. The very concept of poverty line
and its updating by simple price ratios does
not make sense since with changes in
structural parameters the base year poverty
line is not valid any longer. Similarly, various
poverty measures which are supposed to
aggregate and summarise characteristics of
the poor will not convey the dynamics of
underlying changes. For instance, Ahluwalia
(1978) does not examine these institutional
parameters; instead Ahluwalia regresses
poverty estimates on agricultural production
and confirms trickling down of growth to
the poor. But actually due to structural
changes in production conditions and labour
markets, cereal consumption of the general
population and some poorer sections has
actually declined contrary to what one would
expect during a period of poverty reduction
[Suryanarayana 1995). A similar limitation
applies to the recent debate on economic
reform and increase in poverty. Studies,
using poverty measures, have not examined
how far these measures summarise the
dynamics underlying the explanation put
forward by them. For instance, Tendulkar
and Jain (1995) make estimates of poverty
for 1990-91 and 1992. They find that between
1990-91 and 1992, there was a substantial
increase in rural but only a marginal increase
in urban poverty. They contrast these rural
and urban changes and attribute the increase
in rural poverty to bad harvest. This is because
economic reform measures in the initial years
involved fiscal contraction and import
compression which must have adversely
affected the urban sector first. Hence, urban
poverty should have increased more sharply
than the rural. That this did not happen was,
Economic and Political Weekly
according to them, due to the decline in
agricultural production particularly coarse
cereals along with speculative hoarding by
traders and farmers which resulted in an
increase in cere ' pr:'-»r -•-’.iicing “’ereby
the economic entitlement of only the rural
poor. The urban poor did not suffer to the
same extent because they were relatively
better protected by the urban-biased public
distribution system (PDS). Thus, Tendulkar
and Jain conclude that economic reforms
contributed indirectly and was not a major
cause of increase in rural poverty. They
arriveat these conclusions without examining
how far the data or the summary measure
reflects the postulated economic processes.
They build up poverty estimates and their
explanations from diverse non-comparable
data sources. As shown in Suryanarayana
(1996a), the price indices used in obtaining
rural-urban poverty estimates are outdated
and non-comparable. Further, the NSS data
itself shows that percentage reduction in
cereal consumption was more in the urban
than in the rural areas in majority of the
states and particularly in the most urbanised
states of Maharashtra, Gujarat and West
Bengal. Thus, the statistical details under
lying the summary statistics do not provide
any support for the explanation put forward
by Tendulkar and Jain. Instead, they support
the contrary, by their own reasoning. These
two illustrations show the importance of
data base and its verification in any discussion
on poverty analysis, explanation and their
policy implications.
Thereis another limitation stemming from
the fact that poverty estimates based on
consumption data are conditional measures,
namely, these consumption estimates do not
take into account the access to common
property resources'” and their importance
for the poor. For instance, as Jodha (1990)
shows, in dry regions of India in 1982, about
84 to 100 per cent of the poor population
depended on common property resources
for fuel, fodder and food. Over time, there
has been a decline in area, productivity and
maintenance of these common property
resources due to large-scale privatisation,
inappropriate policies and programmes for
productivity improvements, increased
commercialisation resulting in overexploitation and resource degradation, etc.
This has resulted in reduced reliance of the
poor on common property resources and
hence, increased cost of achieving the basic
minimum level of living. Therefore, the
current approach of updating the poverty
line by adjusting for only price changes may
not be valid; instead the very poverty line
may have to be redefined before obtaining
the conditional poverty measures.
Accordingly solutions for poverty
eradication may differ and get varying
emphasis.
Special Number September 1996
Further, the NSS estimates of consumer
expenditure would serve the purpose so long
as the questions to be addressed are limited
and confined to a poor economy subject to
serious budget constraint which restricts the
policy options and size. In such a context,
calorie based ‘minimal’ poverty line reflects
an approach to poverty alleviation which is
concerned with eradication of mass hunger
and starvation. Accordingly, the studies in
India have al! along considered a calorie
based poverty norm and confined themsel ves
to measuring absolute poverty. This explains
why factors other than hunger got low priority
in policy formulations, and plan achieve
ments with respect to primary education,
primary health, etc, have remained quite
modest. Therefore, it is time that we defined
a poverty line with adequate provision for
education, nutrition, health, housing, etc.
This would call for data on various socio
economic aspects of rural households. •
But, even access to public services and
common property resources vary across
households. No single survey data gives
information on all these aspects. This would
raise questions regarding comparability and
integrability of the data sets on various
aspects. For instance, the NSS estimates of
household size based on the survey of social
consumption is quite different from that
based on the consumer expenditure survey
for 1986-87. The average household size is
5.26 in rural all-India and 4.79 in urban all
India as per the NSS annual consumer
expenditure survey but 5.20 and 4.90
respectively as per the NSS social
consumption survey [GOI 1989 and 1990],
Therefore, even though there are various
NSS surveys of different socio-economic
aspects of households, there are questions
regarding the extent to which such data can
be combined to throw up a complete and
integrated picture of poverty.
Pol icies for a sustai ned solutionforpoverty
can be formulated only with an adequate
understanding of the nature and causes of
poverty. This calls for a distinction between
chronic and transient aspects of poverty.
Such aspects can be examined only with
panel data. Of course, it will raise questions
as to how far the results based on such panel
data can be generalised for the population.
Their value is essentially from their policy
implications and suggestions. One approach
to overcome such constraints could be to
examine qualitative aspects of economic life.
Jodha (1988) shows that there is a need for
supplementing macro investigations with
micro studies. Of course, often they end up
with contradicting conclusions. Jodha
explains such contradictions in terms of
methodological deficiencies of the micro
approaches. Such deficiencies are due to (i)
restrictive concepts and categories used for
identifying rural realities; (ii) restrictive
2489
norms and yardsticks used for assessment
of rural realities; and (iii) communication
gaps between the researcher and the
respondent. Jodha (1988) illustrates the need
for supplementing researcher’s approach by
the respondent’s approach, citing results on
incidence of poverty in two villages of
Rajasthan during 1963-66 and 1982-84.
Poverty identification was done by the
conventional approach using the income
criterion., and by qualitative indicators of
economic well-being. Households that had
become poorer by the income criterion were
actually betteroffby thequalitative measures
of economic well-being. The villagers
measured the changes in their economic
status in terms of the following criteria: (i)
decline in their reliance on the traditional
patrons, landlords, and resourceful people
for sustenance, employment, and income;
(ii)decline in dependence on low pay-off
jobs; (iii) improved mobility and liquidity
position; (iv) shifts in consumption patterns
and investment in durables. The changes
reflected by these qualitative indicators are
essentially the result of gradual changes
over a long period of time unlike those
reflected by the changes in income which
capture only the transitory components.
In fact, the very purpose of the de velopment
process and policies is to achieve such
improvements in economic status and
opportunities and any analysis of changes
in poverty has to consider such
improvements. This underlines the need for
supplementing information on consumption
by other qualitative indicators so as to arrive
at correct inferences. One effective solution
seems to be a decentralised approach with
the participation of the local people in
identifying the poor, determining solutions
for the poor and targeting them for the poor.
Ill
Poverty Alleviation Programmes
and Indicator'Targeting
How do we identify the poor by some
simple indicators so as to ensure targeting
of benefits only to the poor? The Government
of India set up a working group to evolve
an acceptable methodology for identifying
the poor through criteria alternative to per
capita income/calorie requirement. The
working group considered the household as
the unit of enquiry for identification [GO1
1985b). The group sought to identify poor
households by characteristics “which are
simple, easily verifiableand easily amenable
to collection and recording of information
about them without observational errors”.
The list of household characteristics
considered for field survey for identification
of the poor is given in appendix I. However,
the actual study was based on the results of
the survey conducted by the directorate of
economics and statistics, Maharashtra of the
standard of living of the rural households
during 1982-83. This was a re-survey of the
same setof households canvassed duringthe
32nd round of the NSS during 1977-78.
Essentially the group attempted a
dichotomousclassificationof the households
into poor and non-poor on the basis of not
only consumer expenditure but also other
household characteristics. The efficiency of
classifying households as poor or otherwise
by other characteristics was assessed by the
extent of mis-classification vis-a-vis that by
consumer expenditure.
The analysis was conducted for the sample
data from the two districts of Aurangabad
and Solapur. For example, one exercise was
based on a sample of 171 households from
Aurangabad. The study found that six
characteristics, namely, (i) land possession,
(ii) irrigated land, (iii) type of wall,
(iv) possession of sofa-cot, (v) floor area per
capita, and (vi) possession of account in
bank/post office, were highly correlated with
per capita consumption expenditure. A
comparison of the classification of
households into poor and non-poor by these
characteristics with that by consumption
expenditure by the discriminant analysis
showed that the overall mis-classification
was as high as 34.5 per cent. Among the six
characteristics considered, land possession
and irrigated land had low correlation with
per capita expenditure. Hence, the exercise
was repeated by dropping these two
characteristics but the extent of mis
classification reduced to only 26.9 per cent.
A similar exercise was carried out on the
basis of a sample of 300 households from
Solapur. The exercise considered 41
characteristics consisting of 24 qualitative
characteristics (converted into quantitative
form by standardisation assuming normal
distribution), five quantitative characteristics
and 12 new characteristics constructed
from the sample data. From these, 14
characteristics having high correlation with'.
per capita expenditure but low correlation
among themselves were selected for
discriminant analysis." The study found the
extent of mis-classification was about 35 per
cent under two alternative threshold values
and the results did not change substantially
even for analyses based on (five-year) lagged
and current per capita consumption. The
group concluded that there is no alternative
to the identification of the poor except by
annual income or expenditure. Of course,
when the estimates of percapitaconsumption
itself has so many limitations, one may raise
questions regarding the classification by
expenditure criterion as the reference
classification. It is also difficult to justify the
assumption of normality white transforming
the attribute values of qualitative
characteristics into standard normal variate.
This seems to be a major limitation of the
exercise. For, normality of variables is a
basic assumption of thediscriminant analysis.
which is very sensitive to the fulfillment or
otherwise of the basic assumptions.
Attempts have been made in India for
targeting some of the poverty alleviation
programmes on the basis of the means-test.
For example, the ministry of rural
development identifies household by income
criterion for targeting the benefits of the
Integrated Rural Development Programme
(1RDP). The target group is the rural poor
consisting of scheduled castes, scheduled
tribes, agricultural labourers, marginal
Table 1 : Proportion of Households Benefiting from Specified Poverty Amelioration Programme by Socio-Economic Status: All-India Rural”
(a): Across Monthly Per Capita Expenditure Classes
Monthly
Per Capita
Expenditure
(Rs) Class
Less
Than
65
Number per 1000 of
households receiving
IRDP assistance by
MPCE (Rs) class
60
Number per 1000
of households
participating in
public works by
(MPCE) (Rs) class 81
65-80
80-95
95110
110-125
(Poverty
Line
122.63)
60
64
66
63
67
63
63
60
67
57
60
75
76
72
67
66
67
56
60
51
*
49
42
125-140 140-160 160-180 180-215 215-280 280-385 385 and . All
above
63
1 J-
!•
’ 64 .
Source: GO1 (1993).
2490
Economic and Political Weekly
Special Number September 1996
farmers with annual income less than a
specified threshold level (Rs 6.400 per annum
per family in 1987) and bonded labour
families. The beneficiaries are identified in
two stages. To begin with, a family income
survey/census is carried out to ascertain
family income from all sources. In the second
stage, a list of all potential beneficiaries is
prepared. The list is further screened by the
gram sabha or in the general meeting of the
village residents. But, this is not a fool-proof
criteria. Studies by the National Bank for
Agriculture and Rural Development
(NABARD) and programme evaluation
division of the planning commission have
shown that there are leakages of IRDP
benefits to ineligible households.12
Recent estimates of distribution of number
of beneficiary households per thousand
households of different poverty alleviation
programmes across economic classes,
occupational groups and social classes also
show considerable leakages to the nonketed groups. The NSSO collected
ormation on receipt of IRDP assistance,
participation in public works, etc, during the
43rd round survey on consumer expenditure.
The enquiry, among other things, was about
households which received IRDP assistance
sometime during the last five years, those
which participated in public works during
the last 365 days and their consumption
during the month preceding the date of
survey. Tables 1 to 4 provide the proportion
of beneficiary households across different
socio-economic status. The results on
distribution of proportion of IRDP
beneficiary households across monthly per
capita expenditure (MPCE) classes show
that they are uniformly spread over all
expenditure classes (Table 1). Since IRDP
benefit refers to assistance received some
time during the last five years while MPCE
relates to the month preceding the survey,
the results cannot be treated as unambiguous
^Mence of misallocation of IRDP since the
Wrerved pattern could also be due to the
very success of IRDP in poverty alleviation.
Therefore, the question on IRDP targeting
should be verified further and one option
would be to examine the spread of beneficiary
households across six different ‘land
possessed’ classes of households (Table 2).
The data bring out that IRDP benefits have
been conferred on all classes of households,
and not on only the poorergroups. The state
wise data show that in Andhra Pradesh,
Assam and Bihar, the proportion of
beneficiary households increases with size
class of land possessed; but in Kerala, Punjab
and Himachal Pradesh the proportions of
beneficiary households are more in the lower
classes of land possessed [GOI 1993:101].
The findings an IPDP ’eakages to noneligible households get only confirmed by
the data on distribution of beneficiaries
S
2492
inferior food items. This has happened partly
due to availability constraints, partly due to
increased incomes and hence entitlement,
and due to greater market exposure [see
Suryanarayana 1995].
Similar suggestions for indicator targeting
have been made with respect to poverty
alleviation programmes. Experts have
recommended employment-oriented strategy
towards poverty alleviation on the ground
that it permits self-selection of the poor and
hence leakages to the non-poor will be
minimum. Thisis based on the understanding
that there is considerable scope for targeting
by setting the wages at the reservation level
in programmes like public works. The
reservationwage.forunskilledworkersbeing
across other socio-economic categories (see
Tables 3 and 4).
During the ongoing economic reform
programme, there is renewed stress on the
need for cost-effective safety nets to protect
the poor without exacerbating the;
government budget deficit. This has called
for targeting of various welfare programmes
like the PDS. Any targeting exercise
presupposes identification of the poor by
means test or some other criteria. In the
context of the PDS reform,. experts have'
argued for commodity based targeting also
by re-orienting the PDS in favour of coarse
cereals, little realising that the consumption
patterns of even the poor have undergone
a change against coarse cereals and other
Table 2 : Proportion or Households Beneftting erom Specified Poverty Amelioration
Programme by Socio-Economic Status : All-India Rural
(b): Across ‘Land Possessed’ Classes
Less than 0.01-0.4 0..41-1.00 1.01-2.00
0.01
Size Class of Land
Possessed (Hectares)
:
2.01-4.00
Above
4.00
All
Number per 1000 of
households receiving
IRDP assistance by
land possessed class
49
64
72
71
60
52 .
63
Number per 1000
of households
participating in public
works by land
possessed class
59
57
69
71
70
75
64
Source: GDI (1993).
Table 3 : Proportion of Households Benefthno from Specified Poverty Amelioration
Programme by Socio-Economic Status: All-India Rural
(c): By Type of Household
Self-Emp Agricultural
loyed in
labour
Agriculture
Household Type
Number per 1000 of
households receiving
IRDP assistance
by household type
Number per 1000 of
households participating
in public works by
household type
Other
Labour
Self-Emp
loyed in
Agriculture
Others
All
. 78
7
66
62
27
63
59
66
168
40
30
64
Source: GOI (1993).
Table 4 : Proportion of Households Benefiting from Specified Poverty Amelioration
Programme by Socio-Economic Status : All-India Rural
(d): By Social Group
Scheduled
Tribe
Scheduled
Caste
Neo-Buddhist
Others
All
Number per 1000 of
households receiving
IRDP assistance by
household group
82
94
156
51
63
Number per 1000 of
households participating
in public works by
household group
122
75
129
52
64
Household Group
Source: GOI (1993)
Economic and Political Weekly
Special Number September 1996
inversely related to poverty, only the poor
are expected to participate in such
programmes, whereas the opportunity cost
of participation is higher for the non-poor.
However, Indian experience even with
indicator targeting of poverty alleviation
programmes has not been that successful.
One important reason seems to be the faulty
design of such programmes like guaranteed
minimum wages and not reservation wages.
The NSS results about the beneficiaries of
public works programmes from the 43rd
round survey are as follows. As regards
public works beneficiaries, the NSS considers
a household as a participant in public works
during the reference period (365) days if at
least one member of the household has
worked for 60 days or more during the
reference period. From Table 1, it can be
seen that the proportion of household
beneficiaries per 1000 households declined
from 81 for the poorest expenditure class to
42 for the richest expenditure group for all
India. ForOrissa, Bihar, Assam,Tamil Nadu,
Andhra Pradesh and Uttar Pradesh, no such
declining trend can be observed. The
proportion of beneficiaries is higher in lower
expenditure classes only in Gujarat,
Rajasthan and Maharashtra (GOI1993:104).
The data on household classification by
classes of land possessed for all-India show
that the proportion ofhousehold participation
is higher for groups possessing more land
than those with less land (Table 2). When
the public works participation is examined
by occupation type of households, the allL.dia cellmates show the participation to be
highest among other labour households
followed by agricultural labour households
(Table 3). But in Tamil Nadu, West Bengal
and Assam, it is the ‘self-employed in non
agriculture’ who seem to have benefited
most than other categories. In all otherstates,
‘other labour’ households’ participation is
proportionately more; in Punjab and
Maharashtra ‘agricultural labour’ and ‘other
labour’ households participated more or less
equi-proportionately [GOI 1993:104]. The
data by social group, shows the SC/ST
households to have benefited much more
than 'other households’ (Table 4). In sum,
the evidence presented above brings out two
salient features: (i) They indicate
considerable leakages of benefits of poverty
alleviation programmes like the IRDP and
public works and hence call for their reform
on the identification front for effective
targeting. This would improve cost
effectiveness of these programmes and
enhance the poor’s economic access to food;
and (ii) the pattern of leakages across different
socio-economic categories vary across
states suggesting that there is no single
correlate or factor by which issues regarding
targeting can be considered across the entire
country. -
Economic and Political Weekly
There is also a need for studies to evaluate
strategies and policies for poverty alleviation
and their micro and macro implications in
view of the current emphasis on efficient
utilisation of resources. Even though there
have been such studies based on computable
general equilibrium (CGE) models, they have
their own limitations. One major merit of
such models is supposed to be their ability
to endogenise the price vector and take into
account the substitution effect of relative
price changes. These CGE models ” arc based
on traditional demand models for the rural
sector in spite of the fact that a substantial
portion of the rural resident households are
subsistence and surplus farmers and their
market participatiori rates for food
consumption are very low. As per the
available estimates, market participation rates
for the all-India total rural population for
rice and wheat were 57.19 and 30.2 per cent
respectively in 1986-87 (GOI 1990]. For
these households, an increase in the price
of a staplecommodity has not only a negative
substitution and income effect but also a
positive profit effect. Hence, there is a
Appendix I
List of Household Characteristics Considered by the Expert Group for Identifying the Poor
1 Principal occupation of the household
2 Secondary occupation of the household.
3 Household size by age and sex.
4 Number of earning members of the household.
5 Number of members working as attached labourers.
4
6 Number of workers engaged in non-agricultural occupation.
7 Monthly household expenditure.
8 Average monthly income of the household.
9 Total land possessed (irrigated + dry + others).
10 Total land owned by the household.
11 (a) Educational status
(b) Skill and training
12 Number of pairs of clothes in use.
13 General use of footwear, tea/coffee/soft drinks, hair oil, toilet soap/washing soap.
14 Housing :
Area under roof
Type of walls of thc house
Type of roof of the house
Height of the roof from the floor
Height of the floor
Type of fleer
15 Lighting, electricity, lantern, etc.
16 Fuel - electricity, gas, kerosene, etc.
17 Livestock possession
18 Possessions: (a) Bullock cart
(b) Bicycle
. (c) Motor cycle
- (d) Car
(e) Time piece, torch
■
(f) Electric fan
(g) Sewing machine
j
(h) Electric iron
. -' t*
1
(i) Refrigerator
(j) Chairs and tables
(k) Utensils
(I) Cots
\
(m) Almirahs
(n) Carpels, mats, rugs
(o) Horse-drawn car (tonga) . (p) Rickshaw
(q) Hand-pulled cart
.<
(r) Agricultural implements
19 Possession of an account in bank or post office.
20 Household reads newspaper.
■ '
21 Household members belong to co-operative societies.
22 Household indebtedness (with reason/purpose of loan)
23 Expenditure on food
24 Prinking water
25 Sanitation ’
"
•
26 Expenditure on ceremonies and functions*
'
27 Pattern of entertainment availed by the family. '
. .
28 Nutrition and Health parameters:a) Height and weight of children aged below five..
b) Measurement of mid-arm circumference of children. • .
c) Number of children bom in the household during preceding five years and how many of these
were found to be surviving on date of survey.
Special Number September 1996
2493
specification error involved in estimating a consumer responses and hence, the price to conventional approach based on narrow data
base and concepts can yield misleading
traditional demand model, which cannot take be paid in using such models is ‘too high’
into account the profit effect, for the rural [Deaton 1974],15 An attempt may be made results and policy prescriptions. Therefore,
sector. The specification error must have to overcome these limitations by estimating rhe-e is a need for an i'tegr”—1 ■’nproach
been quite serious for India as most of these demand systems that permit non-linear and for a comprehensive analysis.
Available evidence on the distribution of
traditional demand models are estimated on non-separable preferences like the Almost
the basis of the NSS data collected using a Ideal Demand System (Deaton and benefits across socio-economic classes of
moving reference period spread generally MucIIbauer 1980].16 Such models can also administratively targeted poverty alleviation
over an agricultural year. Actually, consumer be used to answer welfare related questions programmes like the IRDP shows
preferences in the rural sector have to be in a partial equilibrium framework at least considerable leakages to the ineligible. Seme
specified and estimated in terms of an for theurban sector. But the methodological holds good for public works programmes,
agricultural household model that takes into problem is that the statistical properties of where scope for self-targeting by fixing
account the interdependence of production such models based on the NSS data are wages at the reservation level is considerable.
and consumption decisions. Studies [Singh suspect. This is because the NSS estimates Much needs to be done in the area of
et al 1986] have shown that such specification of total consumer expenditure at constant indicator targeting. The studies conducted
errors really matter and the results differ prices do not show any sustained growth by the expert group on identifying
depending upon whether the profit effect is or even substantial variation during the alternative indicators of poverty have their
considered or not. Estimates of own price entire post-independence period [See, own methodological limitations and their
elasticities of demand for agricultural Suryanarayana 1995] and hence, the scope conclusion that there is no alternative to
commodity obtained using the two alternative for obtaining statistically efficient estimates identifying the poor except by means-test
approaches, that is, traditional demand and of parameters of non-linear consumer cannot be generalised for all time across the
agricultural household models, differed preferences is limited. In addition to this country. Same holds good for suggestions
significantly with respect to size for Japan, limitation, some of the CGE models do not for indicator-based PDS reform by changing
Thailand and Sierra Leone and with respect have an explicit labour market and cannot its commodity basket. In sum, the only
to both size and sign for Taiwan, Malaysia, account for the dynamics of labour market effective solution seems to be decentralised
Korea and Northern Nigeria. While the own changes under the various poverty all approaches based on perceptions at the grass
price elasticity estimates obtained by the eviation programmes.'7 This is not to say roots level.
Questions on macro strategies for poverty
traditional demand models show negative that these models and inferences based on
consumer responses to agricultural them are invalid but to stress the need for alleviation can be answered with appropriate
commodity price changes for all the countries appropriate modifications in their computable general equilibrium models.
considered, those by agricultural household specification. Towards overcoming some Much needs to be done both in the
models show much smaller negative of these limitations, it would be worthwhile specification and the estimation of such
responses for Japan, Thailand and Sierra to collect data on both consumption and models for India. On the specification level,
Leone but positive consumer responses for productive activities of the households there is a need for agricultural household
the remaining four countries [Singh et al through integrated household surveys. models for a predominantly rural and semi
1986:27]. Equally striking are thedifferences Integrated surveys are important for monetised agricultural economy like India.
in estimates of elasticities of demand for evaluating the welfare implications of Such models cannot be estimated without
non-agricultural goods with respect to the questions like increases in foodgrain prices. appropriate integrated household surveys
price of agricultural goods. Cross price For answering such questions, it is not and without relevant data on all aspects of
elasticities based on traditional demand enough to have information on household household economic activities, which are
models are small and negative because of food consumption but data on household presently not available in India. Such studies
are also important for understanding the
negative income effects; thoseobtained from food production are also needed.
dynamics of changes in poverty and their
theagricultural household models are positive
IV
welfare implications.
andlargebecauseofthe positive profit effect.
Summing Up
Unless the data required for such a type of
Notes/.
specification and estimation are available, it
is not possible to make accurate analysis of
The preceding discussion emphasises the
welfare consequences and policy implications importance of data base in any discussion [This is a revised version of the paper presented
of various market intervention and poverty on poverty and identifies the major gaps for at the workshop on ‘Data Base for Rural Poverty
alleviation programmes. But, even while policy studies in India. Beginning with the Indicators’ held atthcNational Institute for Rural
using the traditional demand models for the primary question of identification of the Development, Hyderabad, during April 17-19,
...
general equilibrium analysis, some of the poor based on a measure of standard of 1996.]
available CGE models are based on living and a minimum norm till the final
1 See, forinstance, the study for Cote d’Ivoire
characterisation of consumer preferences by stage of policy prescription, an awareness
by Glewwe and Van Def-gaag .(1990).
the Stone-Geary utility function and hence ofthedatabase and the constraints it imposes . 2 The question of aggregation, i e, how one
obtains an index of poverty also matters
the Linear Expenditure System.1' The on interpretations, etc, is quite important.
from the point of view of policy design
underlying preference structure of the The paper throws light on some such issues
and implementation. As the planning com
Stone-Geary utility function is additive, ignored in studies for India. It is noted that
mission admits, one of the'reasons for
separable and linear. Demand rriodels based " the -conventional approaches to poverty
ineffective targeting of Integrated Rural
on additive separability of preferences' ■ identification and measurement presuppose
Development Programme (IRDP)'is the
cannot account for substitution effects of a stationary economy. In a developing
tendency for the administrators to’focus on
relative price changes and price elasticities economy subject to changes in institutional
those who arc in the neighbourhood of the
based on such demand rriodels will be ’ parameters
involving
increasing
poverty line so as to claim higher success,
proportional to income elasticities [Deaton commercialisation of product markets and
■ measured in terms of head-count ratios, in
1975]. Thus, the use of such demand increasing casualisation of labour markets
poverty alleviation [Government of India
models involves distorted measurement of as experienced by India in recent years, a
. 1983:49]. .
.......
Economic and Political Weekly
Special Number Scp'embcr 1996
2495
The Seventh Five-Year Plan states: “Cost 6 The working group appointed' by the
Subbarao (1985).
effectiveness of the programmes and
Government of India in 1962 distinguished
13
See, for instance, Janvry and Subbarao
minimisation of leakages should be the two
between private expenditure and public
(1986), Mitra and Tendulkar (1986),
guiding principles in the implementation of
expenditure, the latter supposed to be incurred
Narayana et al (1991) and Quizon and
poverty alleviation programmes. ^Economic.
by the state on. health, education, etc. In
Binswanger (1984).
viability should be understood primarily in
keeping with this distinction, the expert
14
See, for instance, Janvry and Subbarao (1986),
terms of cost effectiveness, i e, maximum
committee defined poverty in terms of private
Mitra and Tendulkar (1986) and Narayana
income generation per unit oftotal expenditure
household consumer expenditure [Perspective
ct al (1991).
incurred. This is to be distinguished from
Planning Division 1962].
15
The ongoing research at the Indira Gandhi
economic viability defined as level of'
7 For details regarding how the sample size,
Institute of Development Research seeks to
investment sufficient to enable a family to
number of strata, etc, has changed over time
overcome some of these limitations.
cross the poverty line. The ability of a poorer
and their implications, see Vaidyanathan
16
For studies on India in this direction, see
household to cross the poverty line depends
(1986a)
Coondoo and Majumdcr (1987), Majumder
on its overall income, i e. income from the
8 Cited in Jose (1978)
(1986), Ray (1991), Radhakrishna and
poverty alleviation programmes and other . 9 For instance, Ahluwalia (1978) estimates
Murty (1995), Suryanarayana (1996c) and
wage and non-wage incomes accruing to
poverty measures in terms of parameters of
Suryanarayana, Roy and Parikh (1993).
them” [GOI 1985a;51).
Lorenz curve, proposed by Kakwani and
17 These issues are also quite important. It really
4 Glewwe, and Kanaan (1989) show how
Podder (1976), of consumer expenditure
matters whether the response functions are
household incomes can be predicted on the
distribution.
estimated taking into account the.inter
basis of household survey data on observable
10 The term common property resources refers
dependence, of different decisions as shown
characteristics like area of residence and
to that of the community wealth or resources
by Singh et al (1986).
characteristics of household dwelling. These
over which every member of the community
predictions in tum can be used to decide on
has equal rights for use.
References
budgetary allocations and income transfers to
11 These were (i) Principal occupation; (ii) Use
the poor so as to reduce poverty subject to
of toilet soap; (iii) Possession of torch; (iv)
Baker, Judy L and Margaret E Grosh (1994):
the resource constraint.
Reading newspaper; (v) Lighting; (vi) Having
‘Poverty Reduction through Geographic
5 see, for instance, Baker and Grosh (1994),sofa-chair, (vii) Having cot-diwan; (viii)
Targeting: How Well Does it Work?’, World
Ravallion (1989) and Rayallion and Sen
Possession of cup-board; (ix) Possession of
Development, Vol 22, No 7. pp 983-95.
(1994). Similarly the question of alternative
fan; (x) Height of door; (xi) Irrigated land per
Coondoo, Dipankor and Amita Majumdcr (1987);
indicators of food and nutrition security for
capita; (xii) Land possessed per capita; (xiii)
‘A System of Demand Equations Based on
use in food and nutrition monitoring, and
Floor area per capita; and (xiv) Per capita
Price Independent Generalised Linearity’,
evaluation systems has received considerable
income.
International Economic Review, Vol 28, No 1,
attention [Haddad et al 1994],
12 These studies and their findings are cited in
pp 214-28.
3
This vivid and lively account of a
LITERACY AND EMPOWERMENT
celebrated mass literacy campaign
VENKATESH B ATHREYA and
in the backward district of
SHEELA RANI CHUNKATH
Pudukkottai in Tamil Nadu is set
in the context of India’s policies and programmes to advance adult education
over the past 50 years. The book provides a detailed process account of the
campaign, showing how the combined energies of literacy activists and a
supportive administration managed to convert one of the most backward
districts in the country into a fully literate one. Of particular note is the role of
women—both as teachers and learners.
1996 ■ 209 pages ■ Rs 295 (cloth) ■ Rs 150 (paper)
Students of economics have for long been
uneasy about the claims of the discipline to
universality and institutional neutrality.
Transcending the dualities of substance
CTKURIEN
and form, and theory and practice, this
bookidentifies the essentials ofthe economy
SAGE Publications India Pvt Ltd
Post Box 4215, Now Dllhi 110 048
and provides insights into the changes requ
(til: 6485884, 6444958;
ired in economic theory to make it more
Solis Offici—6463794, 6463820
Fox: 91-11-6472426)
responsive to real-life economic problems.
RETHINKING ECONOMICS
Reflections Based on a
Study of the Indian Economy
1996 ■ 272 pages ■ Rs 335 (cloth)
■ Rs 195 (paper)
2496
IE-55, Soli Loki, Colona 700 064
(Til: 3377062) ■ 27, Molony Hood, I Hagar,
Madras 600 017 (Til: 4345822)
Economic and Political Weekly
Special Number September 1996
Dandckar, V M and N Rath (1971): Poverty in
India, Indian School of Political Economy,
Lonavala.
Dandckar. V M and P Vcnkataramaiah (eds)
(1975): Data Base of Indian Economy, II:
Role ofSample Surveys, Statistical Publishing
Society. Calcutta, and The Indian Econometric
Society. Hyderabad.
Deaton. Angus (1974): * A Reconsideration of the
Empirical Implications of Additive
Preferences’, Economic Journal, Vol 84,
pp 338-48
-(1975): ‘The Measurement of Income and Price
Elasticities’, European Economic Review,
Vol 6, pp 261-73.
Deaton, Deaton and John Muellbauer (1980):
‘An Almost Ideal Demand System’, American
Economic Review, Vol 70, pp 312-26.
Foster, James, Joel Greer and Eirk Thorbecke
(1984): ‘A Class of Decomposable Poverty
Measures’, Econometrica, Vol 52. No 3.
pp 761-66.
Glewwe, Paul and Oussama Kanaan (1989):
Contributions and Crisis’, Economic and
Political Weekly, Review of Agriculture, Vol
XXV, No 26, pp A65-A78.
Jose, A V (1978): ‘Real Wages, Employment,
and Income of Agricultural Labourers’,
Economic and Political Weekly, Review of
Agriculture, Vol XIII, No 12, pp A16-A20.
Kadekodi, G K (1992): Poverty in India: Data
Base Issues, Vikas Publishing House, New
Delhi
Kakwani, N C and N Podder (1976): ‘Efficient
Estimation of the Lorenz Curve and Associated
Inequality Measures from Grouped
Observations’, Econometrica, Vol 44, No 1,
pp 137-48
Majumder, Amita(1986): ‘ConsumerExpenditure
Pattern in India: A Comparison of the Almost
Ideal Demand System and the Linear
Expenditure System’, Sankhya, Vol 48. Series
B, Pt I, pp 115-43.
Minhas, B S (1988): ‘Validation of Large-Scale
Sample Survey Data: Case of NSS Estimates
of Household Consumption Expenditure’.
Targeting Assistance to the Poor Using
Household Survey Data, WPS 225, Population
Sankhya: The Indian Journal of Statistics,
and Human Resources Department, The World
Bank, Washington, DC.
Glewwe, Paul and Jacques Van Der Gaag
(1990): ‘Identifying the Poor in Developing
Countries: Do Different Definitions
Matter?’, World Development, Vol 18, No
6, pp 803-14.
Govemmentoflndia(1972):77ie National Sample
Survey. Nineteenth Round: July 1964 - June
1965, Number 192, Tables with Noles on
Consumer Expenditure, Delhi.
- (1983): Sixth Five-Year Plan: A Mid-Term
Appraisal, Planning Commission, New Delhi.
-(1985a): The Seventh Five-Year Plan 1985-90,
Volume II, Planning Commission, New Delhi.
- (1985b): Report of the Working Group for
Evolving an Acceptable Methodology for
Identification of the Poor, Central Statistical
Organisation, Ministry of Planning, New
Delhi.
- (1989): Sarvekshana, Vol XII, No 4, Journal
of the National Sample Survey Organisation,
New Delhi.
- (1990): Sarvekshana. Vol XIII, No 4, Journal
of the National Sample Survey Organisation,
New Delhi.
-(1993). ‘Proportions of Households Benefiting
from Specified Poverty Alleviation
Programmes by Socio-Economic Status’,
Sarvekshana, Vol XVII, No 2, pp 100-12.
Haddad, Lawrence. Eileen Kennedy and Joan
Sullivan (1994): ‘Choice of Indicators for
Food Security and Nutrition Monitoring’,
Food Policy, Vol 19, No 3, pp 329-43.
Iyengar, N S and N Bhattacharya (eds) t1 978):
A Survey of Research in Economics: VII,
Econometrics. Allied Publishers, New Delhi.
Janvry, Alain De and K Subbarao (1986):
Agricultural Price Policy and Income
Distribution in India, Oxford University Press,
Delhi.
Jodha, N S (1988): ‘Poverty Debate in India: A
Minority View’, Economic and Political
Weekly, Special Number. Vol XXIII, Nos 45,
46 and 47, pp 2421-28.
- (1990): ‘Rural Common Property Resources:
Economic and Political Weekly
Vol 50, Series B, Pt 3, pp 279-326/
Mitra, Pradecp K and Surcsh D Tendulkar (1986):
Coping with Internal and External Exogenous
Shocks: India, 1973-74 to 1983-84, CPD
Discussion Paper No 1986-21, The World
Bank, Washington, DC.
Mukherjee, M( 1986): ‘Statistical Information on
Final Consumption in India and the National
Sample Survej'. Economic and Political
Weekly, Vol XXI, No 5, pp 206-09.
Mukherjee, M and Somesh Saha (1981):
‘Reliability of National Income and Allied
Est i mates '.The Journal ofIncome and Wealth,
Vol 5, No 2, pp 131-36.
Murthy, M N (1977): ‘Use of Empirical Studies
in Evaluating Sample Design for Estimating
Frequency Distribution’. Bulletin of
International Statistical Institute, Vol XLVII,
No 3. ’
Narayana, N S S, Kiri t S Parikh and T N Srinivasan
(1991). Agriculture, Growth and Redistr
Ravallion, Martin (1989): ‘Land Contingent
Poverty Alleviation Schemes’. World
Development, Vol 17, No 8, pp 1223-33
Ravallion, Martin and Binayak Sen (1994);
‘Impacts on Rural Poverty of Land-Based
Targeting: Further Results for Bangladesh’,
World Development, Vol 22, No 6, pp 823-38.
Rudra, Ashok (1972): ‘Savings, Investment and
Consumption’ in C R Rao (cd), Data Base
of Indian Economy. I: Review and
Reappraisal, Statistical Publishing Society,
Calcutta and the Indian Econometric Society.
Hyderabad, pp 148-62.
Sen, Amartya (1976): ‘Poverty: An Ordinal
Approach to Measurement’, Econometrica,
Vol 44, No 2, pp 219-31.
Singh, Indegit (1986): Agricultural Household
Models: Extensions. Applicationsand Policy,
The Johns Hopkins University Press,
Baltimore.
Srinivasan. T N and P K Bardhan (1974): Poverty
and Income Distribution in India. Statistical
Publishing Society, Calcutta.
Subbarao, K (1985): ‘Regional Variations in
Impact of Anti-Poverty Programmes: A
Review of Evidence’ .Economic and Political
Weekly, Vol XX, No 43, pp 1829-34.
Suryanarayana, M H (1995): ‘Growth, Poverty
and Levels of Living: Hypotheses. Methods
and Policies’. Journal of Indian School of
Political Economy, Vol VII, No 2, pp 203-55.
- (1996a): ‘Economic Reforms, Nature and
Poverty’, Economic and Political Weekly,
Vol XXXI, No 10. pp 617-24.
-(1996b): ‘IndianTrickle Down Debate: Missing
Methodological Links’ (mimeo), Indira
Gandhi Institute of Development Research,
Mumbai.
- (1996c): ‘A Hierarchic Demand System based
on nested AIDS-Translog-LES Cost
Function’. Proceedings of the 32nd Annual
Indian Econometric Society Conference,
Volume 3. Economic Analysis Unit, Indian
Statistical Institute, Bangalore.
Suryanarayana. M H and N S Iyengar (1986): ‘On
the Reliability of NSS Data’, Economic and
ibution of Income: Policy Analysis with a
Political Weekly. Vol XXI, No 6, pp 261-64.
General Equilibrium Modelfor India, NorthSuryanarayana, M H, Tirthankar Roy and Kirit
Holland and Allied Publishers, New Delhi.
S Parikh (1993): ‘Growth, Distribution and
Perspective Planning Division (1962):
Demand for Textiles.in India’, Journal of
‘Perspective of Development: 1961-76’,
Quantitative Economics. Vol 9. No 2.
abridged version reprinted in P K Bardhan
pp 263-80. -■■■■
and T N Srinivasan (eds) (1974): Poverty and
Tendulkar.SurcshDandLRJainf 1995): ‘Economic
Income Distribution in India, Statistical
Reforms and Poverty’; Economic and Political
■ Publishing Society, Calcutta.
Weekly, Vol XXX, No 23. pp 1373-76.
Quizon, Jaime B and Hans Binswanger (1984):
Tyagi,D S (1982): .‘How valid arc the Estimates
The Impact of Agricultural Growth and of
of Trends in Rural Poverty?’, Economic and
Selected Government Policies on the
Political Weekly, Review of Agriculture, Vol
Distribution of Income in India, Report No:
XVII, No 26, pp A54-A62.
ARU 21, World Bank, Washington, DC.
Vaidyanathan, A (1986a): ‘On the Validity of
Radhakrishna, R and K N Murty (1995):
NSS Consumption Data’, Economic and
Econometrics of Complete Demand Systems
Political Weekly,’Vol XXI, No 3, pp 129-37.
with Illustrations for India, Centre for
- (1986b): ‘Labour Use in Rural India: A Study
Economic and Social Studies, Hyderabad.
Ray, Ranjan (1991): ‘Complete Demand Systems . . of Spatial andTemporal Variations’, Economic
and Political Weekly. Review of Agriculture,
and Recent. Welfare Applications in India: A .
.'. Vol XXI, No 52, pp AI30-A146.
Review’, Journal < •! Quantitative Economics,
- (1988): ‘Agricultural Development and Rural
Vol 7, No 1, pp 1-22.
Poverty’, in Robert E B Lucas and Gustav
Ray, S K (1970): ‘Imbalances, Instability and
F Papanek (eds). The Indian Economy: Recent
Government Operations in Foodgrains’,
Developments and Future Prospects, Oxford
Economic and Political Weekly, Review of
University Press, New Delhi, pp 77-90.
Agriculture, Vol V, No 39, pp Al 15-A214.
Special Number September 1996
2497
RIGHT TO WORK AS A FUNdWeNTAL RIGHT
Right to Work as a Fundamental Right:
Solution to the Problem of Poverty in
India
T.V. Lucy
According to Article 23 of the International Bill of Human Rights,
“Everyone has the right to work, to free choice of employment, just and
favourable conditions of work and to protection against unemployment.”
This Bill further affirms “the right to equal pay for equal work and the right
to just and favourable remuneration to ensure for himself and his family an
existence worthy of human dignity.” The right to work and a decent living
are the central themes of the above Bill. A decent living implies alleviation
of poverty and eradication of the causes of the same. The success of poverty
eradication measures to a great extent, depends on the ability of the govern
ment to provide employment facilities to the umemployed and underem
ployed. Here comes the importance of the right to work as a fundamental
right of the citizens of India. Incidentally, it may be noted here that India is
a signatory to the U.N. Charter and, therefore, has the bounden duty to
abide by and to implement the provisions of this Bill to the maximum
benefit of its citizens.
Five Year Plans, Poverty Alleviation and Employment Generation
No doubt, at independence, India inherited the problem of mass rural
poverty (Govt, of India: 1), and a shattered economy. Therefore, the govern
ment of India was faced with the difficult task of a war against abject
poverty and hunger. Different programmes were framed and implemented
in successive Five Year Plans to improve the economy (Banja 1988: 1).
The First Five Year Plan gave the highest priority to the increase of agricul
tural production and irrigation. During the Second Five Year Plan priority
was accorded to rapid industrialisation, expansion of employment opportuni
ties and reduction of inequalities in income and wealth. Self-sustaining
growth by securing 25 per cent increase in the national income, self-suffi
ciency in foodgrains, reduction of disparities in income and wealth etc.,
T.V. Lucy is editor of a Telugu monthly "Mana Marpukosam” and is involved in the
activities of “Weaker Community’s Action for Development and Liberation," Hyderabad.
359
were the main objectives of the third plan. Growth with stability, selfreliance etc. were to be achieved by the fourth five year plan. The fifth
aimed at eradication of poverty, while the removal of poverty was the fore
most objective of the sixth plan. Rapid generation of productive employ
ment, alleviation of poverty, attainment of self-sufficiency in food and a
higher degree of self-reliance were proclaimed to be the objectives of the
seventh plan. The eighth plan proposes to emphasise growth and
modernisation. It also proposes to reduce poverty levels to 18 to 20 per cent
(Anon. 1992: 13-23).
Gains and Setbacks at a Glance
As a result of these measures, from a food deficit State India has
become a surplus producer. It has attained the sixteenth position among the
industrially developed nations in the world. In terms of its technical and
scientific humanpowcr, India is the third largest in the world, next only to
the USA and the former USSR (Bhanja and Vcnkatadri 1988: 1). Yet the
other side of economic development is equally disheartening.
The plight of the majority of the rural population has not improved.
Chowdhari (quoted in ibid: 1), reported evidence of steady increase in pov
erty and remarked that the economic development and economic planning
have largely bypassed the people who were supposed to be its beneficiaries.
It is little wonder, then, that despite the high investment, the number of the
poor continues to swell year after year and plan after plan. The Fifth Plan
stated, “at present over 220 million are estimated to be living below this
level.” lite sixth plan (1878-83) has stated that “according to a recent esti
mate using norms of caloric consumption, the percentage of population be
low the poverty line in 1977-78 may be projected at 48 per cent in the rural
areas and 41 per cent in the urban areas. The total number of the poor so
defined would be about 290 million” (Anon. 1992: 13-23). According to
official figures recently released, there are nearly 254 million people in our
country still living below the poverty line. This figure is not, however,
acceptable to several non-official agencies. They say that it is more than
that. Whatever may be the extent of poverty and the manner of people
living below the poverty line, the fact remains that poverty in our country is
substantially a rural problem (Govt, of India 1992a: 15).
Q
3
The Problem and the Causes Identified
lite slogan Garibi Ilatao raised during the parliamentary elections in
1971 is a clear indication of the identification of the problem of poverty.
Therefore, the fifth plan approach paper stated that unemployment, under
employment and the low resource base of a majority of producers, particu
o
r
360
SOCIAL ACTIO~’OL.I3 JULY-SEPT. 1993
larly in agriculture, are the principal causes of poverty. Little wonder then
that the fifth plan resolved to launch a direct attack on the problems of
unemployment and underemployment. According to the sixth five year plan
poverty was a reflection of the problem of unemployment and underemploy
ment (Anon. 1992: 34-35).
Additional Employment Generation Programmes
In order that the benefits of development might reach the poorest of the
poor, that is, the working class in the rural areas, the government of India
undertook to implement several special programmes. Some of these arc
Small Farmers’ Development Agency (SFDA), Marginal Farmers’ and Agri
cultural Labourers’ Development Agency (MFALDA), Drought Prone Area
Programme (DPAP), Integrated Rural Development Programme (TRDP),
National Rural Employment Programme (NREP), Rural Landless Employ
ment Guarantee Programme (RLEGP), Development of Women and Chil
dren in Rural Areas (DWCRA), Training of Rural Youth for Self-employ
ment (TRYSEM) and legislation like land reforms (Mathur 1985: 73-80).
The Minimum Wages Act of 1948 for agricultural labourers, the Bonded
Labour Abolition Act of 1976, besides the 20 point programme of the prime
minister were some of the other measures (Mahajan 1984: 333-338). Jawahar
Rojgar Yojna (JRY) seems to be the last in the series of programmes under
taken by the government of India to alleviate rural poverty by means of
employment generation.
The End Result of the Programmes
Each of these programmes, though conceived in good spirit, failed to
achieve the desired results mainly due to the loopholes in the planning and
setbacks at the implementation level and due to the lack of political will
(Anon 1992a). Each successive programme was meant to rectify the loop
holes in the earlier one, without any structural change or innovative meth
ods. A critical examination of the national development efforts revealed
that, the marginal farmers, small farmers, agricultural labourers and the rural
artisans, who composed the development scenario of rural India, had not
benefited from them (Bhanja and Venkatadri 1988: 1-2). Out of 850 million
Indians, over 600 millions live in the rural areas. Around half of them are
stated to be living in absolute poverty. They suffer from chronic malnutri
tion and lack of access to health care, education and proper housing. Women
and girls are often the worst off. In 1981 about 70 million people were
officially registered as workers whose main occupation was agricultural labour.
However, this number docs not include women, because they arc not regis
tered as such or are only looked upon as house-keepers. One can safely
RIGHT TO WORK
FUNDAMENTAL RIGHT
361
conclude that there arc now more than 100 million agricultural labourers.
Altogether it is reported that there arc about 250 million men, women and
children whose livelihood depends on agricultural labourers. Altogether it is
reported that there arc about 250 million men, women and children, whose
livelihood depends on agricultural labour (Oonk. 1992: 10). According to
R.K. Paramahamsa, (1984: 5-6) 254.7 million people in rural India were
V.
living below the poverty line in 1981.
Sad Plight of the Agricultural Labourers
For agricultural labourers, work is available for not more than 120 days
in a year. There is high competition in the agricultural labour market, due to
the non-availabilily or the seasonal nature of work and surplus labourers.
There is the Minimum Wages Act for the agricultural labourers, but the
provisions of this Act arc seldom known to the employers and the employees
alike. With the result, the agricultural labourers get much less than the
stipulated wages. Disparity in wages between men and women continues to
be the order of the day in spite of the law on equal wages for equal work.
From backward and drought affected districts like that of Mahabubnagar in
Andhra Pradesh, agricultural labourers are forced to migrate to far away
places in search of work and livelihood. In this migration the workers,
especially women and children, are put to untold difficulties, harassment and
exploitation.
But little attention is paid to the problems faced by them. According to
M.L. Mehta (1985: 191) since independence only two significant initiatives
have come from the States in terms of launching innovative programmes of
rural development. One came from Rajasthan in the shape of Antyodaya and
the other from Maharashtra in the mid-SOs in the shape of the Employment
Guarantee Scheme.
On 2nd October 1977, the Rajasthan government launched the Antyodaya
programme. The scheme was formulated after interviewing many poor
persons and conducting a survey of several families drawn from 25 villages
situated in five agro-climatic zones of Rajasthan. The Scheme, development
of the last person (antyodaya) covered all the villages of the State. The
beneficiary group included artisans, educated unemployed and non-agricultural labour along with the old and infirm. The programme focused on a
small segment of the poor called “the poorest of the poor” (ibid: 190).
States like Himachal Pradesh, Gujarat, Orissa and Bihar later launched this
programme on the Rajasthan pattern (ibid: 192).
362
SOCL-xwIc'nON VOL 43 JULY-SEPT. 1993
The Maharashtra Employment Guarantee Scheme
The idea of an employment guarantee scheme in Maharashtra arose in
the 1960s. During this period the income gap between the urban and the
rural areas increased, which resulted in large scale migration to the cities,
especially to the city of Bombay. From 1969 onwards some experiments
with Employment Guarantee had been started in a number of villages. The
scheme itself was launched at the State level in 1974, and in 1979 the
Employment Guarantee Act made employment a legal right.
According to the Employment Guarantee Scheme of Maharashtra, ev
eryone who registers himself or herself as unemployed has to be offered
work on a project within 15 days. One or more villages together have to
guarantee the participation of at least fifty labourers. Two persons out of
every family have to supply free labour for one day per month. Normally
work is offered within a range of 5 to 8 k.m. from the village. Work is
offered anywhere in the district, if it is not available within this range.
Labourers are paid the official minimum wage, which is equal for men and
women. The wages are partly paid in money and partly in coupons with
which they can get subsidised food at the fair price shop. Creche facilities
are also arranged at the work site, which helps the older children to attend
school instead of minding their younger siblings, when both the parents are
working. If the government is not able to offer work, an unemployment
allowance of two rupees per day is paid.
The scheme is financed 50 per cent by the government and 50 per cent
by the special taxes paid mainly by the city population. A major part of the
work undertaken under this scheme is on the lands of small farmers. The
kind of work undertaken is digging of wells, land reclamation, afforestation,
horticulture etc. Decisions about the kind of work to be undertaken, are
taken by a committee consisting of technical experts, politicians and at times
representatives of the workers. Most decisions are taken at the district and
sub-district level. Two thirds of the village council has to endorse the
project to be undertaken in a village. As of March, 1991, more than 1,95,000
such projects were completed and about 13,000 were in progress.
Achievements and Advantages of the Scheme
The scheme has provided six months’ employment to about one mil
lion people per year. About 45 per cent of the participants were landless
labourers, the rest being small farmers. The scheme is more attractive for
women who form two thirds of the labourers, particularly since wages paid
for them in the scheme in comparison with the prevailing agricultural wages
are relatively high. Often more than one family member is working on the
R1GI rr
AS A FUNDAMENTAL RIG1TF
363
same project. Therefore, women feel more protected against sexual harass
ment.
The city population has less reason to be afraid of rural migration since
rural workers get more work. Farmers benefited from the improved infra
structure such as irrigation, land reclamation etc. which resulted in opportu
nities for multiple cropping, which in turn could absorb more labourers.
Due to more employment opportunities in the completed project, demand for
work under the Employment Guarantee Scheme has decreased. The agricul
tural production in the completed works has gone up at least by 25 per cent
per year. Labour unions have been formed in about half of the districts in
the State. Because employment is made a right to work and not a favour,
the bargaining power of the rural workers has improved. The caste rigidity
is mitigated as members of different castes have to work together on a
project. Due to alternative employment facilities the bonded labour system
has got reduced. Since 1989 “Rural Development Through Labour” has
become an additional objective of the scheme which combines activities
such as water and soil conservation, land reclamation, afforestation, cattle
breeding, agriculture and horticulture, avoiding overgrazing etc. Thus the
ecological balance is maintained.
The Employment Guarantee Scheme of Maharashtra is not altogether
free from loopholes. But it is an innovative plan to assist the unemployed
landless agricultural labourers. The positive benefits outnumber its very few
negative effects. If the State of Maharashtra can succeed why not Andhra
Pradesh and the remaining States in India? Most States are at present facing
the same situation which Maharashtra faced in 1960s. This sort of an
Employment Guarantee Scheme should serve as a first step towards the
realisation of the right to work as a fundamental right.
International Concern for Employment Guarantee in India
Some NGOs in the West are of the opinion that the Western countries
should support programmes like the Employment Guarantee Scheme for the
poor. The India Committee of Netherlands for example, is of the opinion
that the conversion of India’s foreign debt to western countries should be in
the form of financial support to the Employment Guarantee Programmes.
Until now the landless agricultural labourers are not a target group of west
ern development cooperation with India, though they constitute the majority
of the rural poor. Many western governments are stating that their aid is
primarily meant to tackle poverty, but in fact they are mainly supporting
economic growth. It is only now acknowledged by some important interna
tional donors that the rural poor have been neglected as far as their develop
ment is concerned.
364
SOCIAL ACTION VOL. 43 JULY-SEPT. 1993
In a letter written to C. Francis of WCADL, Hyderabad, E. Fossati,
Director of the Commission of the European Community expresses his feel
ings in the following way: “poverty in India is, however, also a problem of
numbers and limited national resources. With a population of about 860
million people and an annual increase of about 2.1 per cent or 18 million to
be fed from an already overexploited environment, short term measures by
way of simple employment schemes to feed and activate the poor will only
aggravate tomorrow’s problems.” Thus the concern of the human rights
groups in Europe is not accepted by official organs.
Human Rights and the Right to Work
N. Vaidyanathan (1993: 159) says that, the ILO’s work in the field of
human rights aims at safeguarding the freedom of association, abolition of
forced labour, elimination of discrimination in employment, promotion of
equality of opportunity, prevention of economic exploitation, minimum wages,
social security and adequate conditions of work and life.
During 1993, the year of the Indigenous People, the U.N. is holding a
conference on Human Rights. This concern of the U.N. about the rights
relates to every aspect of life, the violation of which is almost universal. It
is impossible to imagine the creation of a sustainable world without the
protection of rights. Right to work is one of the human rights. It is,
therefore, worth taking the unemployed millions of India to the 21st century
with the right to work as a fundamental right. This would be the best tribute
India can pay to the U.N. cause of human rights.
Indian Constitution and the Right to Work
The Constitution of India is quite conscious of the right of its citizens
to work. Article 41 of the Constitution says: “the Slate shall within the
limits of its economic capacity and development make effective provision
for securing the right to work. . . and to public assistance in case of unem
ployment.” Article 43 authorises the States to make suitable legislation to
the effect that all the workers get a living wage ensuring a decent standard of
life. Unfortunately, this part of the Indian Constitution is not of the funda
mental rights, but the Directive Principles of Slate Policy and, therefore, not
enforceable by any court of law. It is left to the discretion of the States to
apply these principles in making laws. Here it may be noted that in the 43
years of the Republic of India, only the government of Maharashtra has
guaranteed the right to work to every adult person in the rural areas (Govt.
of India 1992: 115). The Constitution of India accepts the right to property
as a fundamental right but it docs not accept the right to work as fundamen
tal. It accepts as basic, the principle that a proprietor is entitled to compen
RiGtrr to work a^Tfundamental RtGtrr
365
sation when his property is acquired for public purpose. But it docs not
accept the principle of compensating those whose labour power is not utilised
(Paramahamsa 1984: 6).
No doubt the Government of India at various levels planned and imple
mented various employment generating programmes as has already been
mentioned above. But no efforts has been made to make the right to work a
fundamental right except by the National Front government under the former
prime minister V.P. Singh (1989-90) which, as a first step, wanted to
implement a national employment guarantee scheme for the rural poor. The
government fell before it could implement its plans. The present govern
ment through its eighth five year plan promises full employment by 2000
A.D. But it does not seem to have any plan to make right to work a
fundamental right or to implement any employment guarantee scheme seri
ously.
Role of Political Parties in Relation to Employment Guarantee
The ultimate goal of any political party is to capture power, both at the
centre and in States, and to rule the country. But the paradoxical situation
which prevails in a democracy is that power is said to be vested with the
people. In order to get into power the party needs a majority vote to be cast
by a majority of the citizens. To get this majority, political parties make use
of the powerful elements of the rural elite. This sort of vote catching is a
heavy financial burden on the political parties. In such a situation, these
parties would do well to approach the rural poor directly and organise them
on their right to work as a fundamental right. This sort of a vote catching
mechanism would contribute a great deal towards the development, unity
and integrity of the nation and would be a healthier practice than any other
for the stabilisation of a secular India, and would be a real boon to the
teeming millions of unemployed and unorganised agricultural labourers of
this country.
Role of the Voluntary Sector in Relation to the Right to Work
The voluntary sector in India, in spite of its separate identity, aims and
objectives, method of approach, and kinds .of work etc., could work jointly
towards the realisation of the right to work as a fundamental right. Already
there is a movement in this country in this direction. Since the last five
years, there has been a growing movement in India promoting the right to
work. Some of the participating organisations emphasise the inclusion of
the right to work as a fundamental right in the Constitution. This movement
to organise the unemployed and underemployed leading to the right to work
as a fundamental right, could be further strengthened by many more volun
RIGtrr TO WORK AS FUNDAMENTAL RIGHT
366
367
SOCIAL ACTIO~-'OL 43 JULY-SEPT. 1993
tary agencies joining hands, and through more coordinated efforts than hith
erto.
Bhanja, S.K. and S. Venkatadri. 1988. Milch Cattle in Integrated Rural
Development. Hyderabad: National Institute of Rural Development, p.l.
Govt, of India. 1992. Occasional Papers: The Rural Poor and Institutional
Credit. New Delhi: Ministry of Rural Development, p.l.
Conclusion
Most of the development programmes of the government of India,
contribute to the development of those who already have assets like land.
The landless agricultural labourers benefited the least from these programmes.
The physical labour is the only asset which they have. Their development,
therefore, depends on the planning and implementation of programmes suited
to make use of their physical labour for their benefit as stated by the Interna
tional Bill of Human Rights for “an existence worthy of human dignity” and
as stated by the Constitution of India “for ensuring a decent standard of
living.” Improving the standard of living of the workers amounts to improv
ing the standard of life of the entire nation.
—. 1992a. News Item. The Indian Express, December 16.
Taking into account the benefits of the employment guarantee scheme
of Maharashtra, particularly for women, and its contribution towards the
curtailing of population, it would be worthwhile for the government of India
to make the right to work a fundamental right. Tire present government
through the eighth five year plan and the AICC session on 27th March 1993,
has promised to provide employment for all by 2000 A.D. (Anon 1992a).
From the past experience one can presume that politicians who arc never
weary of making such promises, will not implement it. Moreover, by 2000
A.D. the government has to fulfil many more such promises, e.g. health for
all, education for all etc. Hence, the unemployed agricultural labourers of
this country need not be carried away by such promises. The unfortunate
aspect of Indian development is that the influential and the vociferous sec
tions of society have benefited the most out of the programmes. Hence it is
up to the hitherto silent majority, the landless labourers, the Scheduled Castes,
the Scheduled Tribes etc. to work unitedly for the realisation of their right
to work as a fundamental right. And it is up to the political parlies, the
voluntary sector and anyone who cares for the unemployed landless millions
of this country, to be in solidarity with them and support their movement.
Oonk, Gerard. 1992. Work Against Poverty: Development Aid and Debt
Conversion for Employment Guarantee in India. Amsterdam: India
Work Group.
REFERENCES
Anon. 1992. Political Sociology and Economy. M.A. Final, Political
Science. Instruction Material. Distance Education. Hyderabad: Osmania
University.
---- .
1992a. Literacy Mission. Vol XV, No. S, August. New Delhi:
Directorate of Adult Education, Ministry of Human Resource Develop
ment.
Mahajan, R.K. 1985. Revised 20 Point Programme: An Analysis ofAchieve
ments and Projections for the Future. Hyderabad: Rural Transforma
tion: National Institute of Rural Development.
Mathur, Y.B. 1985. Rural Development in India. Hyderabad: National
Institute of Rural Development.
Mehta, M.L. 1985. Perspectives in Integrated Rural Development.
Adminstrative Arrangements for Rural Development. Hyderabad: Na
tional Institute of Rural Development, p. 191.
Paramahansa, V.R.K. 1984. Development with Social Justice: Concern of
the Weaker Sections, A Constitutional View Point. Hyderabad: Na
tional Institute of Rural Development, p. 6.
Export
Not viewed