7576.pdf
Media
- extracted text
-
4
fin the.
'Dc.filining ^ex. TZ-atifi
Social DSgra^ of Excess Female Mortality in India
■;®.New Directions •
•
c
•'?i'' Alice W Clark
Discussions of sex differentials in mortality in South Asia have suffered from a lack oftheoreticaldepth On
what basis do we decide that a certain mortality difference must be socially rather than purely bioloStcally derived.
The answer is usually to compare the observed mortality difference to the expected dtfference based on western
experience, but this glosses over the need to examine South Asian environments and epidemiology for their own
expected effects on sex differentials. In addition, what are the expected biological sex-differential outcomes of
levels of mortality that are, by international standards, extremely high? Are age-specific patterns of mortality
for both sexes differerit'frpm those based on western models, and why? Is the relationship between these sex
specific patterns different than it is in western experience?
In trying to answer some of these questions, the author advances two presuppositions: (’) that excess female
mortality, in its socially conditioned aspect, forms a part of an overall reproductive strategy, which in most cases
refects strategic calculations of social units considerably larger than the household: (2) that class and gender
relations may have primacy over many other independent variables as determinants of demographic rates. In other
words, variables which operate at a more proximate level may be decisively influenced by changes in class and
gender relations, and in their interrelations.
Net Reproductivity within A Mode o"
THE term ‘social demography* usually con of the issue.
Production Framework
’ In reading the literature on sex differen
notes a primarily descriptive, and highly
statistical, approach toward examining the tials in mortality in the South Asian culture
In the first place, a theoretically rigorous
empirical relations between sodo-cconomic area, a lack of theoretical depth emerges. conception of class ought to illuminate Doth
and demographic phenomena, particularly Tb/s sense is strongly felt from a purely gender inequality and sex differentials in
those empirical relations that can be demographic perspective, and gains even mortality more statisfactor'y than the mixeu
examined given the kinds of data available further momentum given a social historical concept of socio-economic status, which is
(based on the kinds of questions asked) in perspective.
sometimes used. What is needed is research
Some demographic doubts can be sket on the phenomenon of differential mortality
surveys and censuses. Given .a theoretical
framework which is often rather narrow (and ched briefly. On what basis do we decide as it operates within the tctal demographic
is, in part, often derived from an accumula that a certain mortality difference must be and social framework of which it is a part.
tion of earlier empirical examinations of socially rather than purely biologically
/X theoretical basis for this kind of think
similar data sets), this approach can be derived? The answer is usually to compare ing may be found, in part, in a literature
revealing of very broad patterns along im- the observed mortality difference to the which attributes differences in demographic ■
portant lines. Based on such enquiries, we expected difference based on western ex- behaviour to differences in modes of pro
surely start with the notion, for example, that pcrience, but tHii glosses over the need to duction within which households arc embed
female education has a powerful relationship examine Sout^
Sout^Asian
Asian environments and ded (Caldwell, 1983, Mcillassoux, 1981).
epidemiology fpr their own expected effects This literature has not been adequately criti
with child mortality.
Such an approach, however, leaves us in ^on sex differentials. In addition, what are qued and mined for hypotheses specifying
no position to address adequately more com- the expected biological sex-differential out- demographic behavioural differences,
plex and fine-grained patterns of socio- comes of levels of mortality that arc, by although Caldwell’s ideas arc more widely
demographic interrelationships. This is not international standards, extremely high? Are paraphrased than those of any other demo
really the fault of statistical methodology, age-specific patterns of mortality for both graphic theoretician today.
which is able to simplify very complex tables sexes different from those based on western
of cross-classified data using such techniques models, and why? Is the relationship betOne problem lies with Caldwell’s some
as log-linear analysis (of which hazard
ween these sex-specific patterns different
what rigid conception of modes of produc
models arc a variant); but the most powerful than ibis in western experience? I will shortly tion. which apparently allows for very little
methods available fail to illuminate problems discuss some results of analyses that I have of the dimension of social stratification and
that are not adequately addressed thco- . performed in trying to answer some of these dassformationiniisapplicationtodcvclopretically. In terms
of the
°
° example given regar demographic questions.
ing countries. Caldwell secs the contrasting
ding the relationship between female educa
The more sociological doubts are mutli- modes .there as dichotomous: “The real
tion and child mortality, we are for the most laycred and cannot be sketched quite so reproductive divide”, he states, “lies between
part left groping with guesses and genera- briefly.
. My own thinking about these from modes of production based largely on net
lilies about what education.actually means an historical perspective has resulted in two works of relatives and those in which the
and does to people under different presuppositions: (J) that excess ffemale individual may sell their labour to complete
circumstances.
mortality, in its socially conditioned aspect, strangers” (1983, p 568). My own work on
Finer data based on better questions are forms a part of an overall reproductive historical and contemporary South Asia,
surely needed: but without theory, we do not strategy, which in most cases reflects howtver (Clark. 1979, 1983. 1984), convinces
know the right questions to ask. What we strategic calculations of social units con me that there is a reproductive divide bet
would wish for is the chance to test ade siderably larger than the household; and ween classes within given modes of produc
quately specified hypotheses based on an (2) that class and gender relations may have tion, and that mortality diffcrcniials arc
adequate theory, using adequate data. I will primacy over many other independent involved in differential rcproductivity. (See
deal with these requirements as they relate variables as determinants of demographic also De Vos. Clark and Murty, 1987.) Thus
to the topic of excess female mortality in rates. In other words, variables which there can be expected to be interrelations bet
India, for the most part all too briefly, but operate at a more proximate level may be ween demographic rates and such socio
with the hope of suggesting a more syste decisively influenced by chances in class and economic phenomena as class position
matic consideration of them to researchers gender relations, and in their interrelations. within a given mode of production.
\VS-i2
Economic and Po!iti*ii Urrkh Vol X.\!l No |7
Rene* of W«-r.
A-ril 25. WR
Claude Mcillassoux, thus far not in uide
use as a theoretical reference point among
demographers, identifies an interdependence
between the domestic and capitalist modes
in developing countries, in what might well
be called the “interpenetrative mode” of pro
duction. According to this theoretical
framework, within a situation of uneven
development the market sphere partially
depends upon various formerly autonomous
domestic spheres (which originate in tribal
social organisation) for the production of
labour-power in the form ot children
(Mcillassoux. 1981). Thus decisions about
marriage and reproduction do not only con
cern the members of households. While the
household participates in its own micro-level
political economy, encompassing gender
relations (of which marriage networks form
a partic’darly important element), it faces
constraints and incentives from within the
larger political economy which affect its
reproductive options. Within this context,
class relations between and among house
holds, which mirror interrelations among
modes of production, play an important role
as determinants of reproductivity. A similar
theoietical framework has engendered im
portant studies in European historical
demography (e g, Tilly, 1978) in which class
has been found to be a significant context
for differential demographic behaviour.
We may start, then, from a theoreti
cal framework which presuppose* inter
relationships between class and gender rela
tions. and a set of pathways from these
relations through family decision making
processes to reproductivity (which in
corporates child mortality). Figure 1 pro
vides a sketch of the full theoretical model,
which conceptualises a mode of production
in both its productive and reproductive
aspects—the reproductive aspect being more
closely specified in terms of demographic
measures here than in most such theoretical
formulations.
The final outcome, net reproductivity,
, corresponds to the familiar demographic
measure, the net reproduction rate. It can
easily be computed for sufficiently large
samples if one has either age-specific fertility
and mortality rates, or estimates of the total
fertility rate and female survival to the mean
age at child-bearing. These measures can be
straightforwardly computed for populations
and groups, or estimated using logit models
for much smaller units such as households,
given that this is done within a large data set.
The dominant technology (cell A) serves
to classify the mode itself. The health
technology (cell C) is closely linked to the
productive technology, and forms the nexus
of proximate determinants (as called for in
Mosley and Chen. 1984); but its actual uses,
in the context of family decision processes,
arc intermediated by the social relations of
production (class relations) and reproduc
tion (gender and generational relations), as
found in cells B and D. An important
example of (his kind of intermediation will
be found in the following exposition of some
of the implications of the theory.
In agrarian and tribal societies under
patrilineal kinship organisation, women
have long been objects of exchange along
with other wealth in the forms of dowry and
brideprice, originally as an outcome of the
very process of social articulation (LeviStrauss, 1969). The economic use of women
as exchange objects clearly predates one in
which an expected wage from their labour
is calculated. Rosenzweig and Schultz (1983)
Figure 1: Stricture ano Reproductive Outcomes of a Mode of Production
Panel 1
Diagram of a Mode of Production
Social Relations’
Forces
Material
Production
Human
Reproduction
(A)
Technology
Levels of skill and human capital
.
(B)
Surplus appropriation system
Location of units in class system
Rules governing labour supply
(C)
Available medical and contraceptive
technology
Levels of nutrition and health
Family authority systems, marriage
networks and gender roles
Rules of distribution and inheritance
Panel 2
.; .
Reproductive Outcomes-.
>
D)
(A
(C)
Family decision processes
Child + young adult mortality
Fertility
Net reproductivity
FlGcaE x-'At^JOALnY Sex Ratios, 25 WFS Countries
(• . Coaparcd
UdiTi (1 «x>atb-3 yean)
’•*
------ n—------- -----------------------------
t.J
,
/•.z
... •:
i
Indio* •
---/
1.2 -
X
o
1.1
o
5
0
1
0
0
x 0.9
&
I
I
OS
(r
0.7
r-------- r0
02
04
o
0.6
0.8
1
Proportion with Rotio < X
Countries Ranked
WS-13
ECONOMIC AND POLITICAL WEEKLY
Review of Women Studies April 1987
4
see a value difference between the'sexes,
caused by lower wages for females than
males; I would prefer to view the laUer as
but one manifestation of a more general and
complex phenomenon. The.value of females
as objects of exchange significantly out" J for a time, the
■ wages they
•
—
weighs,
can
command when market relations become a
factor. The strong positive association bet
ween low labour force participation, which
reflects in part differential wages, and
dowry-giving, which reflects the value of
women within an exchange context, is
demonstrated for India in Miller (1981). My
own work has shown how the high value of
women in the exchange context can be
enhanced by the limitation of their numbers
(Clark, 1983).
: '•-.v
The use of women as exchange goods can
continue in a market setting for sometime,
especially in a situation where class forma
tion overlaps with existing castc/tribe/ethnic
segmenution (Clark, 1979,1983). The
emergence of women as subjects of their
own history is a turning point whose occur
rence cannot be entirely predicted. *
During the initial process of class forma4ion, the roles that women are expected to
play, however, become differentiated by
class. Class itself is at the outset articulated
corporately, along pre-existing group lines,
and not simply according to a family’s indi
vidual rank on the scale of wealth and
property-holding. Due to various political
historical factors, formerly autonomous
tribal and descent groups emerge at the
upper and lower rungs of a new social and
economic hierarchy within the new reality
of property. During this phase (which can
be long-lived), class solidarity remains to
some degree linked to kinship and its wider
analog, descent. In the South Asian context,
class is linked to caste, with its endogamous
marital limits. Different rules emerge for the
proper conduct of upper and lower, caste
women which reflect an emergent class dif
ference, which then in turn becomes encoded
as a cultural one.
What are some of the demographic effects
to be expected from the process of class
formation and female role differentiation
that has been sketched here? The examples
suggested below are based on the analysis
of social demographic phenomena in central
Gujarat during the last century (as found in
Clark, 1979).
One example is the following, applicable
to classes with some property. Where female
children, due to the structure of marriage
networks (cell D) which continue to exist
within given classes (cell B), are given in
marriage accompanied by large dowries,
some of them, because they are so expensive
to invest in. may be neglected via with
holding of nutrition and health care, raising
their mortality. Above some benchmark
level, excess female child mortality has been
shown to reflect a relatively greater neglect
of the nutritional and health care needs of
girls than boys (D’Souza and Cberb 1981).
•
Constraints on female labour force pyucipation within such classes (a_produd of
influences from both cells B and EQ can
f^'r this process. Tlie net rcproduttivuy
of mothers would be lowered m such a case
jf there were no compensating nse in the
level of fertility. Given overall levels of
mortality that make it relatively unpredict
able how many children will survive, we
would expect high fertility and high child
mortality to go together; and under con
ditions where sons are disproportionately
desired and several sons are sought—that is,
often, where dowries arc
are high and male
labour is predominant—we would expect
both high fertility, and child mortality which
is elevated by socially conditioned excess
female mortality. Some female education
might not appear to improve these factors,
because its chief effect might simply be
added to a girl’s marriageability.
A second example pertains to classes with
no property and no tenure rights over pro
ductive resources. We would expect family
constraints on proper female behaviour here
to be fewer, and dowery to be small enough
to be disregarded (cell D). Women should
be participating in the labour market at rates
(although not necessarily wages) nearly
equal to those of men (cell B). In this case,
it would be desirable to have several children
FIGURE 3(1): SMOOTHED CENSUS SEX RATIOS, MADRAS
(After Migration Adjustment)
I.Z.
i.i -
5o .
r
i -
0.9 -
ao
1
zo
o
o
et-9i
60
40
Age-Q^oup
01-11
x
a
FIGURE 3(b): SMOOTHED CENSUS SEX RATIOS. BOMBAY
(After Mitralion Adjusimem)
u
1.Z -
o
1.1 -
X
J
0.9 -
o.8 -r
0
*0
70
60
Age-group
WS-14
81—91
91-0’
11-21
x
21-31
Review of Women Studies April 1987
ECONOMIC AND POLITICAL WEEKLY
The second major variable from the
to help support the family, with little or no ficantly to human capital, io the improve
sex preference; in such a case, child mortality ment of child-bearing conditions, or to the model which is expected to influence morta
would be high, due to poverty, but would not equalisation of gender relations and the lity outcomes, in terms of both levels and
be expected to be elevated by excess female reduction of sex preferential practices? At sex differential^, is gender inequality. The
child mortality. Education would be expec what point does it help transform a woman's gap between the age of the mother and her
ted to have a straightforward linear effect own relationship to her assigned roles? spouse, and the educational differential
between them, may represent part of this
Below this, as when its expected effects are
on________
mortality.
The most important determinant variable overwhelmed by those of class, does educa factor. Gender inequality is expected to
in this model is then the class position of tion still have an important meaning? operate interactively with class—and per
the family. (It is probably within class that Trussell and Preston (1982) have stated that haps indepenoently as well.
Adequate data to test hypotheses genera
gender is constructed, for the most part.) there is a need to understand the paths
The concept of class has been recast, in through which education operates; it may ted by the theoretical framework given here
much of the demographic work on develop well be equally important to understand would be microdata, like those of the World
ing countries (for example, that cn Sri those through which it fails to operate as Fertility Survey, but ideally with many more
socioeconomic and ethnographic questions.
Lanka), into a statistical construct called might be expected.
‘socio-economic status’ which is often
FIGURE 4(a); MADRAS, 1901-11, SEX RATIOS
measured by (and therefore confounded
(of l(x)s of Alieruaxivc Life Tables)
with) parents' educational level (Trussell and
Preston, 1982, Meegama, 1980, Trussell and
Hammerslough, 1983, Hobcraft et al, 1984,
i
Lime and Perrera, 1981).
x.z
Education is clearly crucial, but there is
a need to separate more clearly its effects
from those of class—that is. the level of
control over productive resources cnjo>cd by
one’s group, and whether individuals within
•. ---------- 1.
--i--------—----------------------------------- 1--------- J
it work for others
or are self-employed
(control their own labour-power).
What are our expectations about educa
tion’s effects, given a mode of production
framework for generating hypotheses? At
the individual level, first, education is itself
a potentially productive resource, in its role
as human capital; and second, increased
education can affect both child mortality
and fertility through its impact on the con
ditions of child-bearing, such as the mother’s
age, the length of birth intervals, and the
. practice of breast-feeding. Third, at the level
of the village or region, the average number
of years of schooling completed forms a
global, ‘ecological’ variable (in the social
sense). There can be threshold effects on
demographic rates at both the household
and the ecological levels (Tienda, 1984).
Further, within the theoretical framework
presented, the global educational level serves
as one indicator of the dominant technology,
and thus plays a part in the classification of
the mode of production which is applicable
Class in some cases can be seen as /runculing the expected effects of additional
education. A study of mine on child morta
lity in rural Sri Lanka (Clark, 1984) found
(hat class—in which ownership or control
of productive resources is the defining
element—plays an important role in exacer
bating or alleviating sex differentials in
mortality in early childhood, once mother’s
education level is controlled. In addition,
within classes which possess means of pro
duction, but at very low levels, a small
amount of mother’s education reduces
female child mortality less than it does male.
It is strongly suggested that class interest
operates in some cases in the opposite direc
tion from that of educaion. Above what
threshold, then, does education add signi-
J
1.1 -
0.9 -
I
o.a -r
0
2G
O HypochctkaJ
40
vS MoJe/SA FenxiJ*
.
•
60*
•
o
> *•. ■?
80
.
MS M0le/M5 Feme*
■. • •-
FIGURE 4(b) MADRAS FEMAJ-ES, 1901-11
(Life *Ublc Comparison)
L
0.6
■
•
I
i
0.5 -
I
i
A* ■
0.4 -
o
O.J -
(•- •
O.2 -j
%
XI -
o --
(to J
0
/7
a
vs MO*
40
Age
VS female
60
o
80
SR fer
• WS-15
Review of Women Studies April 1987
•
‘
J'
'
•
■
,
<
•
1
■
ECONOMIC /XND POLITICAL WEEKLY
’
In particular,
like
to have ques- propensity
male infants
die at
al higher
. ■ one would
’
r--------------------FtvpcMiuy of
ui mate
imams to
to me
higher excess female child mortality above it. It is
tions on
- marnao^j
n^tu/nrlfc
anrf
rt
___ .t‘__
efemale
___ • <>"«. a .large pan of. which
.
F^T
r,a8e'?'
tTk’’?
n<1
d0Wry
'
ha"
---- —•vimuicUUCJ,
oi wmen important to set a conservative standard
cvcls. Family reconstitutions from parish is captured in the first month of life.
here, for it is too early in the study of this
records might also serve to provide some of
I he measure of an excess sex effect on
the needed data, though in the case of India, child mortality is not whether male and topic in general to say what the ‘true* nor
where few such records exist, this point is female lates differ significantly from each mal level is.
In calculating the cumulative distribution
a little academic. The data need to be on a other in statistical terms, but rather whether
large scale: to estimate believable age-specific their ratio differ significantly from the of sex ratios of mortality from one month
mortality rates requires cell sizes no smaller expected ’normal’ ratio. A normal value to five years of age for 25 World Fertility
than several hundred—though smaller could be expected to be a ratio of slightly Survey (WFS) countries, based on data
numbers can work when mortality rates are less than LOO, female oxer male, due to con given in a recent WFS report (Rutstcin, ‘
very high. In order to use the.data existing tinued biological excess male mortality dur 1984), wc find that India’s ratio (based on j
in the real world, ingenious creation of pro ing infancy above one month o' age. Adop Padmanabha, 1982) lies above those of 95
per cent of the countries. (Sec figure 2. India 1
xies is required..
ting a ratio of LOO would seem to set a con- was not, of course, one of the WFS coun- '
’???•'V -V
. servativc standard for the identification of tries.) If these ratios can be assumed to 1
Age-Specific Sex-Differential
Mortality’^ •
FIGURE 5(a): BOMBAY, 1901-11, SEX RATIOS
(of l(x)t of Alternative Life Tables)
Wc return now to the more purely demo
graphic questions that need better theoretical ?'
“i
‘ and empirical grounding. ILwas necessary for me to work at this broader demograhic
level in doing research on Indian sex dif
ferentials in the past, specifically .1881 to ..
1931, because there are no microdata to use.
‘ '■ 1 ^ed ordinary census data along with some ■ .. J
.new methodologies, and then experimented
widely with alternative explanations for the o
^A rcsull^^planafipns^^^ould incor- »
:'; porate varying patterns of differential mor-$- > ’ •
tality such as appeared to fitlhe particular Z*J? ’
i
r
X case^at hand. Before presenting some of c i. *- i these results, 1 return to a demographic ques- '
lion raised earlier: when is a sex diffcrenual
C.93 J
in mortality likely to have socially condi- *
tidned, rather than purely endogenous,
0.9 -J
biological causes? .
J:'
t
An important part of the above question
o.es -r
is contained in the following one. What
' c
ought-the ratios of mortality rates to be
in infancy, if there is no excess female mor- Q Hypocb«kxi
tality? This very normative question cannot
be answered without any qualifications. In.
western models set at the same UevcT, accor- :•
ding to the level-setting methods of Coale JV. .»
and Demeny, the ratio differ both according \
to model and according to the level within '
each model. For example,’Kfodd South,
X6
Level 18 (Coale and Demeny, I$66), has a
, ratio of mortality from zero to age one of
0.893, female over male. But those of the
other model families at the same level are
0.5
lower, implying that in the South family of
life tables, female over male mortality was
somewhat elevated relative to what it might
have been. Cultural differences between
southern and western Europe are incor
porated in these ‘levels’ as they_differ
___ among - O.J -r
the models. This gives us 1little help in J
answering the question, since models are not
culture-free.
A more conservative approach To the
question is to compare India’s mortality sex
ratios with those of other third world coun
0.1
tries. What really counts is the ratio of mor
tality between one month and five years of
age. This accounts for as much as half the
o
mortality between birth and five, and is far
more likely to be free from the biological
WS-18
2G
40
-I
80
60
us Mole/SA Fe^noie
►
* • MS Male/MS Fwrxye•
FIGURE 5(b) BOMBAY FEMALES, 1901-11
(Life Tabte Comparison I)
20
60
SR Femole
80
ECONOMIC AND POLITICAL WEEKLY
follow approximately a normal distribution
curve, then their mean and standard error
could be used for significance tests of
specific sex ratios of mortality in subsets of
the Indian population. The mean of these
25 countries’ ratios is in fact 1.004; the 95
per cent confidence limit for the maximum
is 1.057. Female over male mortality ratios
higher than 1.057 for the interval between
one month and five years of age can
therefore be said to be significantly above
the normal level for less developed countries.
If we use western models, the mean ratio and
variance are considerably lower. Thus the
WFS ratio of 1.004 can be called an ap
propriately conservative standard.
This presentation clearly shows the far
greater than normal post-neonatal and child
female over male mortality in India, and all
the qualitative evidence that has been gather
ing over the last several years points in the
direction of a greater preponderance of
social rather than endogenously biological
causes. This is only a first approach to the
quantitative issue, however, and further work
needs to be done using clinical field data.
One would like to know more clearly what
; is the benchmark, ‘normal’ mortality sex
ratio for each age, including 0-29 days, under
various Indian conditions, from which signi
ficant deviations could be calculated.
Using data from Padmanabha (1982) these
arc the ratios of age-specific female to male
mortality rates (nmx) in India:
Review of Women Studies April 19S7
and the Status of Women in India, 1881-1931’,
which will be published in a book being
edited by Tim Dyson.) It is possible, however,
to explore ditferent ways of configuring
these adult levels, using different assump
tions about possible life table patterns and
about the ways male and female life (able
interrelate, based on an understanding of
some of possible patterns of sex differentials
in early childhood.
There are problems to address in under
taking such an attempt. The first of these
is that we have no historical life tables for
India that we can trust. The second is that
there' is no way of creating life tables for
"————————
0.6
1
fi
I
f•
0.3
0.4 -
I
:*•
«
■
0.3 -
■■
C.2 -
<29 days
0.863 0-1 year
0.977
29 days-I year 1.202
1-4 years
1.303 29 days-4 years 1.262
It is abundantly clear from these ratios that
female mortality in India does not
take place only in toddlerhood, but also in
the first year of life. This is obscurcd by the
ratio for ages 0-1, because this ratio is so
heavily dominated by mortality in the first
month of life. In that interval, male mor
tality dominates in all demographic regimes
except those of swift and outright female
infanticide. When we sec mortality sex ratios
for ages 0-1 which are as high as 0.977,
then, it seems clear that infant excess female
mortality is involved.
I have dwelt in some detail on ratios of
mortality between 0 and 5 because a very
large share of mortality, particularly in high
mortality regimes, takes place within that
interval (which is, in turn, a combination of
several important intervals). My research on
historical Indian mortality requires some
such baseline understanding of what one can
expect of mortality differentials befor the
age of five.
My data for analysing sex differentials in
mortality in India are only the census age
distributions, plus the adult mortality level
calculations that arc based on them with the
help of the new methodology devised by
Preston and Bennett (1983). (This research
is reported in detail in ’Mortality, Fertilitv
extremely low levels of life expectancy which
is not artificial and mechanical. Using he
Preston-Bennett method we can ascertain
the adult level, but not the shape at the
youngest and oldest ages of the total sur
vival curve. Wc can produce an approxima
tion to the total curve, however, if wc have
some model of survival rates, some modi
fication of which we believe may be close
to the mortality experience of this region.
Recent demographic research on India has
preferred either the Coale-DcmcnV (1966)
Model South, or the new United Nations •
(1982) South Asia pattern; therefore, in the
study mentioned, I produced alternate life
FIGURE 5(c): BOMBAY FEMALES, 1901-11
, * (Life Tkble Comparison 2)
1 ■'
5s I
'•
•
0.1
o
O
SAsia Fe-rva'c
70
Aje
MS Female
(FIGURE 5(3) BOMBAY FEMALES. 1901-11.
(Ltfc Table Comparison 3)
0.5
o
2
2
O.J -i
0.2
0.1
o
60
ECONOMIC AND POLITICAL WEEKLY
■
■■
'
V 'X
tables based on modifications of those
extra female deaths took place in infancy
tables were created for the decade 1901-11 for
models.
and early childhood. Bombay contained
each presidency. This was done by inflating
For the following discussion, however, 1 ■ groups and areas that had been known to
or deflating the male table's survival column
(adjusted for the sex ratio at birth) by a still
went beyond this, starting by fitting the male practice female infanticide. The custom had
been outlawed in the Infanticide Act of 1872,
further smoothed series of sex ratios leading
adult mortality levels which result from the
Preston-Bennett method of life tables accor- but its incidence had also been recorded in
to and then descending from such a peak.
considerable detail in the reports on the
ding to the Model South pattern, and then
Results are presented in figures 4 and 5.
computing female life tables directly from
Figures 4 (a) and 5 (a), in addition, contrast
Act (sec Clark 1983). We believe that the
the male. This has been done in such a way
my hypothetical, smoothed sex ratio curves
decrease in infanticide was followed by an
that the resulting female tables, first, reflect , increase in eccess female mortality at early with those produced by two different pairs
the age pattern of census sex ratios after this . ages/due to neglect of female children of life tables based on adjustments of
pattern has been highly smoothed, and
(Clark 1979 and 1983, and Miller 1981).available models; the model pairs cause
second, have a good fit with the female adult
Pursuing die examination of what difrather wildly anomalous patterns.
dif
mortality levels calculated via the Prestonfcrent
The result of the sex-ratio estimation
ferent sex ratio patter is might imply about
Bennet method. This procedure was per- ■ the different patterns of mortality between technique for Madras is a female life tabic
formed for Madras and Bombay Presiden
regions, the age-specific sex ratio curves for which crosses the male at two points. See
cies, for the decade 1901-11. ./<••
each presidency were smoothed out by the curve labelled SR Female in figure 4
The issue is, how can wc know whether
amalgamating several age groups and pin (b): there it is contrasted with the male table
we have the right life tables for females in
on which it is based, and with the appro
pointing their joint ratios on central ages.
relation to males? We can start by considerpriate female table from the Model South
This got rid of the insistent peaks and
ing the population sex ratios, as computed • valleys characterising these data,"which are pattern, if the sex-ratio based table is more
from census totals. The curve of actual agenotorious for their high degree of age accurate than that from Model South, it
specific sex ratios at a point in time
can be
.
i__
misreporting and underenumeration. The implies that excess female mortality, at ages
seen theoretically as a cumulative measure
where it is not found in matched pairs of
resulting patterns, on inspection, suggest
of mortality at all earlier ages. In a popula
where each real curve may have had its single western models of the same family, did occur
tion with stable rates over time, the ratios
1high point.
in Madras. The
.................
Model South patterns for
of male to female age-specific survival pro
Graphs of the smoothed age-specific both sexes which are appropriate here,
babilities, inflated by the sex ratio at birth,
iratios for each decade are shown in figure however, create the appearance of too much
should .be very, close to the age-specific sex
3. It appears from this exercise that for mast excess (emale mortality, because they are at
ratios of the, population. This is simply ucxaiucs
<
decades,
—
, Bombay
Dornoay’s peak
peax may have
nave occurred different levels.
because age-specific sex ratios are, clearly,
at around thirty or thiny-rnr, and Madras’s
Based on the sex-ratio based table, Madras
ratios of survivors to each age.> •• /.
. at around age forty-five or fifty. This sugfemales had life expectancies close to or
There are both data problems and assure- gests the different age patterns of excess surpassing those of Madras males. But they
j ption problems in proceeding w.kh this line female mortality by region, including the still had higher age-specific mortality than
of thought; in particular, rates were not in strong likelihood that Bombay had some of Madras males, concentrated within the ages
fact stable over time. However, it is clear that
it in childhood, (rhe'curves for 1921-31 may ‘between early childhood and old gge. They
• •J
there are very different patterns both of L-.
have been affectedjn pan* V/
by UIV
the JU4WUUllg
smoothing clearly
had •far belter life chances than
overall mortality and of excess female mor
of the age data (Of me 1931 census which was
females fromi western India, but so did
I tality by age between the two regions. Tne done by the census office, and in part by Madras
-------males. Even in Madras, it appears
latter is clearly suggested by an examination
actual changes in the age-specific patterns tF^at there must have beeni some excess
of the census sex ratios (even admitting that
of mortality. Since we do not know how
rfemale child mortality, for though the infant
these data include underenumeration and
much effect may have been due to each
mortality ratio (female over male) in the
| age misreporting by sex). For Bombay the possible cause, this decides pattern is not
newly created life table is very low, that for
overall sex ratios arc very much higher than
included in my rough estimate of the average' children one through four is 1.058. But there
those for Madras. (This is after adjusting for
peak age.)
was relatively more excess female mortality
migration.)
Having established in this way some in middle adulthood.
We have considerable reason, in any case,
estimate of where the single peak of the age
The Madras female life table constructed
to believe that in the west of India, many
specific sex ratios occurred, new female life in this way turns out to have a fairly dose
fit to one from Model West: thus, if male
Table : Alternative Life Table Pairs for 1901-11
another, due to unusual differences (relative
Male Base
•
Paired with Alternatives
to western experience) in mortality between
Male
Female : • Sex
Female
Sex
Female
Sex
the sexes.
Model
Model
Ratios
South
Ratios
Sex Ratio
Ratios
The new life table for Bombay, in contrast,
South
South
• •
Asia
Table
is quite different from any model-based
ones. For this life table, I have presented three
Bombay
e(0)
different sets of comparisons. Figure 5
18.91
17.63
1.13
18.99
1.05
17.39
1.15
e(5)
38.75
3721
(b) contrasts the new table with the male
43.16
3826
1(5) ■
■ 0.418
0.401
table it is based on (the Model South table
1.10
0.380
1.16
0.3S5
1.14
1(30)
0.287
0.261
that fit the male adult level found via the
0279
. ... .1-16
1.08
0253
120
GRR
3.343
3.014
Preston-Bennett method), and with the
3.342
Madras
female table from Model South which does
e(0)
27.50
25.69
1.13
26.69
1.09
28.11
1.03
likewise for females. Here we see that the
c(5)
44.48
43.03
46.25
42.35
Model South female table does not do
0.546
K5)
0.523
1.10
0.510
1.13
0.581
0.99
justice to the full amount of excess female
1(30)
0.417
0.383
1.15
0397
0.411
1.11
1.07
mortality at early ages, which the sex ratios
GRR
3.057
2.889
2.878
imply.
Figure 5 (c) shows that a South Asia pat
Note: For details of estimation procedures, see Clark. 1987.
tern table runs close to the sex-ratio based
WS-20
Review of Women Studies April 19S7
ECONOMIC AND POLITICAL WEEKLY
table at early ages, but then improves too
much at later ones. The newly created life
table meets the low of the South Asia tablat young ages and the low of the Model
South table at older ones. In other words,
this procedure suggests that excess female
mortality was high at most ages of life in
Bombay.
Figure 5 (d) compares the sex-ratio based
female table both with the South Asia table
and with one from Model West which has
the same life expectancy at birth; clearly the
new one is very different from each.
What the sex-ratio based table postulates,
going back to figure 5 (b), is that female
survival waj well below male until old age
in Bombay; and that early excess female
mortality was particularly high. The infant
mortality ratio, female over male, based on
this table is i.014, and that for children one
through four is 1.128. It is worth noting that
the second ratio is not as high as contem
porary all-India ones. Yet significantly
excessive female child mortality on the
international scale presented earlier is clearly
implied.
,
Of course, these new life tables (sum
marised in the table) have been hypothetical
creations but they make a point. The Indian
data that we have suggest that we continually
go wrong for major parts of India by fit
ting female mortality schedules to available
models. The models themselves have to be
questioned.
Certain additional questions have also to
be asked, and asked again. Do we assume
that mortality regimes are primarily a
‘natural’ working out of certain environ
mental conditions? Is the social environment
continually going to be naturalised in our
thinking, as if female education, for exam
ple, were an ecological variable not just in
the sociolog’ist’s sense but also in the
biologist’s? There appears to be powerful
i factors strengthening the dowry system even
among the educated classes in India today
and dowry is one of the conditions of son
preference
While we do not yet know what constitutes a valid expectation of the endogenous,
biologically-based sex differential at each age
for the various regions of South Asia, the
date that wc do have for Indian regions sug
gests that socially-derived excess female mor
tality has created enduring demographic pat
terns over at least the past century. This
evidence must take its place as an exemplar
of the social forces shaping demographic '
history in a larger sense, and lead us to re
examine the modelling of mortality in
general.
(Prepared for presentation to the Workshop
on Different’-*.! Morality and Health Care
for Females in South Asia, held in Dhaka,
Bangladesh, January 4-8, 1987, under the spon
sorship of the Social Science Research Council
(New York) and the Bangladesh Association for
Maternal and Neo-Natal Health (Dhaka)]
* 7 ■. national Statistical Institute.
Mcegama S A, 1980. “Socio-Economic Deter
Alam, Iqbal and John Cleland, 1981. “Illustra
minants of Infant and Child Mortality
tive Analysis: Recent Fertility Trends in Sri
in Sri Lanka: An Analysis of Post-War
. Exoeriencel’ World Fertility Survey, Scien
Lanka", World Fertility Survey. Scientific
tific Reports, No 8, Voorburg, Netherlands:
Reports, No 25. Voorburg, Netherlands:
International Statistical Institute.
»
International Statistical Institute.
p -- Mdllassoux, Claude, 1981. “Maidens, Meal and
Caldwell, John C, 1982. “Theory of Fertility
Money: Capitalism and the Domestic
Decline". New York: Academic Press.
Economy”, Cambridge University Press.
Chen, Lincoln, Emdadul Haq and Stan
Miller, Barbara D, 1981. “The Endangered Sex:
D’Souza, 1981. ‘’Sex Bias in the Family . . Neglect of Female Children in Rural North
Allocation of Food and Health Care in
India", Ithaca: Cornell Univcrsi y Press.
Rural Bangladesh’, Population and ■ Mosley, Henry and Lincoln Chen, ed, 1984.
Development Review, 7:1.
J'
“Child Survival7, ‘Introduction’, New York:
Clark, Alice W, 1983. ‘Limitations of Female :Population
__ Council.
Life Chances in Rural Central Gujarat’,
Padmanabha P, 1982. ‘Mortality in India: A
Indian Economic and Social History
Note on Then ds and Implications! Economic
Review, 20:1, Special Number on Sur
and Political Weekly, August 7.
vival, Politics and Work: Indian Women,
Preston S H and N G Bennett, 1983. ‘A Census
1880-1980.
based Method for Estimating Adult Mor
—, 1984. ‘Effects of Class and Sex on Child •*;' tality! Population Studies, 37, 91-104.
Mortality in Rural Sri Lanka’, University of Rosenzweig, Mark R and T Paul Schultz, 1982.
Wisconsin. Center for Demography and
. •Market Opportunities, Genetic EndowEcology Working Paper No 84-17. (Revised
ments, and Intrafamily Resource Distribuversion available from author.)
tion: Child,^Survival in Rural India’, .
—, 19S7. ‘Mortality, Fertility and the Status of}?, ■ American Economic Review, 72:4.
\
Women in India, 1881.1931’, forthcoming in J. Rutstcin, Shea Oscar/1984. “Infant and Child
Tun Dyson, ed, “Famine, Disease and.^; Mortality: Levels, TYends, and Demographic
Development in India: Studies in the
Differentials", ; World Fertility Survey,'
Subcontinent's Historical Demography",
Comparative Studies, No 43, Voorburg, ‘
Curzon Press.
'
'Netherlands:!.'International Statistical
Coale, Ansley J and Paul Demeny, 1966,Y!*: ' Institute.<-.y“Regional Model Life Tibies and StableH-Tienda, Martayi984:i‘Community CharacteriPopulations", Princeton: Princeton Univer-/> A - sties. Women's Education, and Fertility in
sity Press.
»
Peru’, Studies in Family Planning, 15:4.
De Vos, Susan, Alice Clark and K R Miirty.?'- Tilly. Charles, ed, 1978. “Historical Studies of ’
1987. ‘Family and Fertility in Context- Gom-< ,
Changing Fertility’’, ‘Conclusion’, Princeton
ments on Some Caldwellian Themes’,) ? . * University-Press. ■
Journal of Comparative Family Studies' Trussell, James and Charles Hammerslough,
(forthcoming).
• .
. 1983. ‘A.Hazards Model Analysis of the
Levi-Strauss. Claude, 1969. “The Elementary
Covariates of Infant and Child Mortality •
Structures of Kinship". Boston.
. ■■■:
' in Sri Lanka’, Demography, 20:1..
LitUe,.Rodcrick J A and Soma Perrera, 1981. Thisscll, James and Samuel Preston; 1983.
Illustrauve Analysis: Socio-Economic Dif- ’ ,
‘Estimating the. Covariates of Childhood
ferentials in cumulative Fertility in Sri
Mortality from Retrospective Reports of
I^anka—A Marriage Cohort -Approach", ■ Mothers’, Health Policy and Education, 3.
World Fertility Survey, Scientific Reports, United Nations, 1982. “Model Life Tables and
No 12, Voorburg, Netherlands: InterDeveloping Countries”, New York.
References
.J-;/.-...
, .’^APPOINTMENTS
International development support group working at grassroots level
is looking for a chief executive for their programme in India to be
based in Pune Incumbent must be over 35 years with grassroot
voluntary agency experience and proficient in administration. Must
be a team leader of proven jmerit. Emoluments shall depend
on qualifications and experience Shall be in the range of Rs 8000/plus perks.
: ■ ' ■
Apply in strict confidence to
Box No. 102, Economic and Political Weekly/284, Skylark, Shahid
Bhagatsingh Road, Bombay 400 038
A-
SPECIAL ARTICLES
T
-----------------
• <
.
-------------------------------------------------------------------------------- ----------------------------- -------------------------------------------------------------------------
■-
Poverty and Sex Ratio
. 'W- ■
-/Some Data and Speculations
N Krishnaji
;
T consumption pffamili^ and he sex ratio.
males Semales and''th'’35
“"“/J ■
responsible for the£ deficit^$ f3^ rn,ain.ly
'
°ther WOrds' are the differences wider in
'wlhin ,nd‘a which has remained fairly
su-iis.ts:
P°Ver,y> defined by per caPil^
••4
in a relatively weak manner in their case
because there is not much to be transferred,
from one poor family to another. If the sugJ
gested arguments are valid, these factors can^
be expected to promote higher female pro-?
portions among the poor.10
,
The question about the relationship bet^
-rring in the north-west and distinct.y
■ ‘raw in Xral h? i|r
wh,'e death ■ ‘h' speculation that the ratio is governed by ? lhc rec'nt famincs in Bangladesh have?
ranidTj
thJoughout (he and hence related to women's role in agrb- fOr cxanlple’thown ,hal severe food shorf4!iof d«nne ha< a27lnde^cncc. the rate, cultural work, more prominent in the poorer
,cad 10 a 8rea‘er ‘chess’, i e. famine^females, at least in
,ha?
.regions and mediated to some extent through "J?.'!,'®4,1dca.tlls among females, especially
^S&The “often
” ^^.the cropping patterns.6 While neat correla- children. Likewise, the data-for the 1943
^iSmin^o^^r^*0^*5ubd?;>‘ions acro“ regions between agriXSl ^al famhe exhibit a higher excess female
'4i\heea o •
etldren"d ; ' Pn»P«ity and crop-mix on the one hand mo.rtallly «“ong children. “ The implifaaors" nderMnJ“ ?U,‘
“d the sot ra,i° on "he o'her are lacking.
this note-could be that when
ff£to which Vklril^t
X“ nevcrtheless true that the agricultural^; food '? 'n.’.hort ’“PP1*-as itnormally is in'
&c£nsti uterhe <?Hk
bretf reference rich wheat-based states of Punjab and roor r«n>>hcs. the mal.s get a dispropor^‘research work Thk wnr?"°f
' Haiyana hav' '«>• low female ratios and low
ly b,gg.cr $har5-To di£r“s a little for;
-cero/d
Jhls.w?rKls■maj‘!Iy con-’:'rates of participation of women in work
reason'“nlayPointed°ul‘hatthe
: '• neglect of f
. 'a.cr?minat10? againsl and whereas the poorest states in India, viz
British-Indian Census compilers looking at
■ ’ determining Of
rCSr CCt Of '*'? CrUClal ■ orissa- Bihar and ‘Madhra Pradesh arc more f3m,nc morlali>y have observed higher male
-H hXTrTn
m0r,3bt«iJlulriti0n “nd- balanced with respect to numbers oE mOr,ali,ies ,il1 about ‘h' f.rst decade of this’
health (or, more precisely, access to food and and males, and the paddy-based south has cen,ur>" ,hey say that earlier, under faminelldn^S o^®-,(1S86)P"^‘U aS-UC V-*he hi£h“'501 ratio/The deficit of femmes cfonditio^
had to go out in search of';
Mshovs th’t the availlbleV^rf1^
in ,he north-w-est has been repeatedly noted f??d ?“d,were ,“biect t0 hl8her mortality^
^ gestint; dhlrS ion ^n“-the early British-Indian Censuses, and ™ks but lat®r' W1,h famine administratiJ
'inadequate in coveraee/nT!!
“ In<kcd $o wcrc RajpuC vi,,a8CJ without bcc°m!ng cffecuve, could stay at home fotf
J“’/“I In- fcmalc
!,drcn or uilh
-’ The
,c[-.” The insights provided^
conclusive The <diffinL
diffirfemale ch
children
with on,
onlyy a fcw
few?
"J"™1 of rc
r«licr
provided!
net a; aH eas£’
crs Wlthin families is to what is now mildly described as discri- Ween,lhe poor and lhe rich
for the ■
r.-cnrr/ri
Jr ....l:he aIlcrnallve» usually ruination mainly to landowning families of Slmplc rcason that famine is a rare occur-^’
resorted to, of looking at anthropometric certain castes and their strong preference for rcnce Carfccting also sections of the non'6
measurements such as body weights and male progeny ascribed to both economic and poor) *n contrast to the chronic deprivation -j
heights m relation to certain standards or cultural factors/ Moreover, it has been amon8 lhc P00’’norms, leads in turn to numerous problems
Against this hazy background, two rela/
of interpreting data unambiguously. These suggested that for easily understandable
reasons,
apart
from
the
role
of
women
in
tionships
arc analysed here. The first refers^
difficulties are discussed in an influential
work, property and asset transfers follow- to sck ratio variations across families with >1
study of some data from Bangladesh (Chen
et al, 1981). On the other hand, it seems ing marriage also play an important part in landholdings of different size and an cnhe determmauon of sex preferences. Miller bedded contrast between agricultural labour i
a pnon that observation of health care prac
(1981)
argues, for example, that there is and other rural households in India with'*
tices and underlying sex biases may be easier
. but so far there have not been many studies greater reciprocity between families of part- respect to the balance between the sexes. The 1
ners in marriage in such transfers in the second is concerned with poverty defined
based on such observation. ■
This note is not, however, concerned with south than m the north-west where it tends as it usually is, by per capita consumption^
ln fav°ur of the of farniIies and its relationship with the sex*
the factors behind the level and the declin hrirtrern0105’ f001!?1
bridegrooms family. All this has obvious ratio. Needless to say, this is a speculative I
ing trend in the fcmale-to-male ratio (FMR)
in India but with a related question: How implications for the question raised here for exercise, for. in a field wherein direct obser-1
fa7,I,CS
Vati°n Can prOvidc no morc than useful '
arc the mortality differentials and the im have'to wnrLfrrSl’ ,n
Drooenv i™<f SUr'*'3,.and SC€Ond’ lhc . insights, secondary information cannot be
balance between the sexes related to powny?
propertj transfer mechanism can act only more illuminating.
892
Economic and Political Weekly
Vol XXII. No 23. June 6, 19S7
ECONOMIC AND POLITICAL WEEKLY
• June 6. 1987
Labour and Landholdings
Demographic data, especially on morta-
* •
household size and the FMRj'lhe'highKt
lily rates by sex, relevant to the discussion
ere, arc not available separately for agricultural labour and other poor households,
so that for comparing the latter with the
nn inHOOr.rUr? househo,ds one has to rely
on
indirect information
variables aUM
such* as
household
size and sex on
ratio.
hotKchoM c,^
--------- •
femaIc Proportion is observed among the
IandI«s labour households who have also
lhc
average size (table 1). They have
an avcra8e size of slightly over 4 with sex
^tios close to 990 in contrast to all rural
ratio of with
950. a- size of
V. about
uoou. 5.5
no and a sex
raljo of 95Q
a sex
(see; tabic I). For agricultural labour households covered by the rural labour enquiries
■ XT ,7“n’“i ss ‘“.s
These correlations hold in a more general
sense with respect to the variations accordin
?landholding
size
3' Firct
the
omg
to landholding
size (ubb
(i-ble 3)
First,
the
a
^ge
householdin^s
s^dHy
average L----- *
-
*•
171 in households with less than half an acre
to 8.73 among families with over 50 acresand second, in contrast, the female-male
ratio among the adults declines as the land
holding increases, with females outnumber
ing jpales in households with small land
’^d^^
with
sm^i
h°1dings and b,8 Emilies go
?ClWCCn lhc scx«’
•»
5
All Labours. All Rural
better balance between the sexes than other
Average
A',CTa8c household
hou«hQ|d
size
.
..
------------------- --------
5.52
$NnriaPS |d° ‘° even ,he lowe“ '"d of (hi
gjaml scale, among the labour families.
T rati0J''aries f«>m one age group
'•d°!^ , and ?ence ,he overa" "'io
, dejxnds also on the age distribution of the
■"«!es; the sggiegate ratio is a weighted
' .aherag- ?f 'u'.Va’Ues within thca8egroups.
■ he weights being the proportion of males.
Aurora: “Rural Labour Enquiry-, I%3-65 and l97417S'<>„d^I
-•: 'IWwi r
T*« Sex Ratio
'■
963
949
952
t-
,, .. ------■■ ■
various
issues.
Category of
Household
LSeCn fr°m lable 2 tha[ [hc f«"a‘e
>000 ^fU254;^ 41.9
473
Agricultural labour [1964-65
f-1 no in the prime-adult age group (15-49)
10.8
J974-75
All rural
hnlXTh Can"y. h'sher an’°ng labour house■: ,45.4 '■ ' 10.7
1971
n ™ ha" ,n-‘he rural Population as a whole
- 44.3<
12.7
.1,000 as against 971); the difference between Source: Same as for Table I.
,rlhe two categories of households is even
Table 3: Sex Ratio
more marked in respect of the 50+ age
age
--------- -------- group But the age distributions of males are
Household Operational
Average
not radically different, despite a higher pro
Holding (Acres)
'-Proportion of
Household Size
portion of males in the 50+ group in [he
_
J______ Children (Per Cent)
0.00.5
.population as a whole (arising possibly from
2.71
0.5 - t.0
402
ower levels of life expectation
1061
among - l.o - 2.5
41.6
abourers). Thus thc differences in thc FMR
2.5- 5.0
' 41.3
mamly arise from higher sex ratios among
' 527 .
;
969 ■
5.0- 7.5
40.6
adults in the labour families. L.J^U, „ ulc
' 5.85
923 '
Indeed, if the
7.5 - 10.0
412
6.13
age distribution of males for the whole rural
.;> •?? -94i
10.0 - 12.5
41.0
6.54
population in 1971 is taken as 'standard- the
. 4’:
910
12.5 - 15.0
41.6
6.70
sandardtsed FMR for the agricultural
922
15.0 - 20.0
41.5
6.91
’ ox?LhOUSC‘,olds Iurns OUI t0 be 988 in
. 947 •
20.0 - 25.0
41.4
. 7.40
^65 and 973 in 1974-75, as against the
25.0 - 30.0
- 42.4
■"'
724
■ actual values of 986 and 976 respectively,
30.0 - 50.0
42.0
7.84
Above 50.0
<s owing that age distributions do not con42.7
■^• 959^
-8.73
inbute significantly to the differences being
__________________ _39.9
/discussed here.
raiio (female to male) refers to the value amonj ’
--------- e
*
.
t'
— X-- • ! «a»«fca me pCIUCIl; -Sex ratios in the aggregate—and among
are derived from .he ■National Sarnple(0-14)
S^-fReX
°U1’Th
da'a
age population in the ‘total.
The'data
v adults in particular—are influenced by sex? Snrv^v*’
xt. « «
-selectivc outmigration, apart from differed
• tial mortalities. Leaving aside migration for
household for .his excluded Xo “ V4 52
h0UScholds- The
average size
of
The average
size of
_ the moment, one can discern another cor-
SV
■
S ■ 40
. Os?
: ;rclatl0n from the data presented so far: an
893
June 6. 1987
ECONOMIC AND POLITICAL WEEKLY^
landholding.
However, the observations arc based on
the sex ratios in the adult populations.
Fortunately, data covering persons of all ages
classified by land size arc available for
the different states from the 1961 Census
(table 4). Once again the FMR can be seen
to be ov’er 1,000 in the smallest land-size class
of less than an acre in the different states
of India, with only a few exceptions. It is
noteworthy that even in Punjab (including
Haryana) and Uttar Pradesh, where the
female deficits arc among the largest in the
country, females outnumber males in the
smallest landholding class. Equally note
worthy arc the sex ratios below' 900 cor
responding to huge female deficits observed
only in the middle and big landholding
groups. Even in the south, where the FMR
is generally high, it is low at the upper end
of the land scale.
Migration: What do these mutual associa
tions between no land or small land
holdings—associated with poverty of an
undeniable sort—small families and a
balanced sex ratio convey to our understan
ding of mortality differentials and the
underlying discrimination against females?
As hinted earlier, these variations in the sex
raiio are influenced io some extent by the
rural to urban migration, demonstrably
more prominent among males seeking work.
Such a sex-selective migration tilts the
balance in favour of females in the rural
areas, especially among population groups
- with high migration propensities.
Some arithmetic will show that while this
kind of migration accounts for a part of the
high female ratio observed among agri
cultural labour and land-poor households
can by no means be the whole explanation
for it. The calculation can be done in a crude
* but quick fashion on the basis of the 1971
Census migration data (Mehrotra. 1974).
Remember that for 1971. the FMR was
roughly 980 and 940 for the agricultural
(about and non-labour populations respec
But, more conclusive at least in a quali
tively. The 1971 Census counted, among the
tative sense are the fragmentary data thrown
438.8 million (mn) rural people, about 23.7
up by small-scale studies of migration. Thus, \
mn birth-place rural-to-urban migrants,
for example, a study of four villages in east
i e, persons enumerated in urban areas but UP (Saxena, 1977) notes that out of 300 •
born in rural areas, with a FMR of 919. villagers migrating to Gorakhpur, only 129 v
Assume that all these migrants were from (i c, 43 per cent) came from households with '
less than 5 acres of land back at home. It j
labour households (a wholjy unrealistic
is said that even well-to-do farmers with over
thing to do as will be presently seen and least
favourable to the hypothesis being developed
50 acres migrate to towns aspiring for higher
standards of living. In Saxena’s sample only ?
here). Migrants (23.7) then constitute about
5 per cent (23.7 mn out of (438.8 + 23.7) mn)
39 migrants out of 300 had an agricultural '
labour background. Similarly, a comprehenof the redefined rural population, but 20 per
cent of the labour population (labour house sive study of the Ludhiana district (Oberai }
holds constituting roughly a quarter of ill
and Singh, 1983) shows that despite the\
households). If all these migrants are thus
landless constituting the bulk of the out- '
added to the observed labour population in
migrants the ‘propensity’ to migrate is no
rural areas, the sex ratio for labour will come
less high among those who have large
down from 980 to 965. but the last figure
landholdings.
is still far higher than the FMR for the non
In this study of the agriculturally most •
labour rural population (940).
prosperous district in India, viz, Ludhiana,
A similar arithmetic can be done with
the authors find that in the surveyed villages
respect to the landholding data. All we need
only about a half of the migrants to urban
to remember is that more than a half of the
areas were landless or land-poor; about 9.5
rural households in India have holdings
per cent of the households with outmigrants
had over 15 acres of land in the village
below 2.5 acres, so that even under the implausible assumption that all rural-urban
(table 5). While the average migration pro
migration arises from these households, the
pensity for all households (defined as the
differences in the sex ratio between them and
ratio of households with outmigrants to the
the big land-owners will remain substantial. , total in the village in each landholding class)
Table 5: Outmigrants from Ludhiana Villages by Landhoi ding Size
Landholding
(Ac.cs)
Households with '
Outmigrants
Per Cent N( = 504)
All Household^
Per Cent N(=2I24)
Relative Migration
Propensity
Landless
Below I
1 - 2.5
2.5 - 5
5 - 7.5
7.5 - 10
10 - 15
Above 15
45.8
3.0
2.0
10.3
8.9
13.1
7.3
9.5
47.6
2.4
3.3
13.7
7.4
9.5
7.9
96
121
65
74
120
136
93
114
8.2
Nofez N refers to the total number.
Source: A S Oberai and H K Manmohan Singh (1983).
Table 4: Sex Ratio (Female: Male) among Cultivating Households by Landholdings. 1961
*
(per 1000)
Sute
Andhra Pradesh
Assam
Bihar
Gujarat
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
Punjab
Rajasthan
Tamil Nadu
Uttar Pradesh
West Bengal
1
1-2.5
2.5-5
______ Landholding (Size in Acres)_______
5-7.5
7.5-10
10-12.5
12.5-15
1025
921
1085
1005
1074
1026
1013
1162
1052
1005
967
1023
1003
978
991
942
1041
969
1027
989
1005
1070
1033
960
942
1000
972
975
974
927
1004
955
9871 .
962 '
984
1026
1015
936
927
977
931
. 95?
960
912
987
951
959
934
973
999
1005
892
919
965
904
938
957
901
977
953
953
934
968
988
994
875
912
961
893
923
949
897
982
947
939
941
966
987
995
863
908
961
890
916
944
878
982
951
931
946
966
972
986
847
913
959
889
902
Source: Census of India”, 1961 (Table Cl, Pan II C—‘Social and Cultural Tables”, different issues).
894
15-30
30-50
50 +
942
873
955
949
930
933
964
970
999
843
906
955
896
897
952
862
912
935
951
897
978
966
1024
847
899
955
905
896
974
~“_867
880
934
915
951
1003
957
1027
859
887
937
899
888
ECONOMIC AND POLITICAL WEEKLY
; •?
June 6, 1987
■'
i-
^??:?
2.r
crccn.t* households
households with
with over
over 15
15
r.I acomL^?
was
/z
per
1?cent,
05’ ,andho!ding ca,
cgory) had
states
of
India
at the
r per •
capita
expenditure
while
veryJo^rario/of
states of
India at
bottom
end of
propensity nf
of 27.5
per cent.
j/ a corresponding Dronensirv
77 < r^r
—.
: ?_ 50,116 of the rural poor migrate, but so, it
^.appears, do some of the non-poor; the
I..
---------- 4 may not be very
fu different.
jC‘ The reasons
reasons for
for migration
migration differ,
differ, of
of
.
c er> low ratios of
1
i
/ acre arc £rouPcd, with
2 .haIf and one
per capita expenditures of over Rs 21 in the
same category as about 24 per cent of
**llh.O.ve.r J00 acrc! This shows that
for example, of a woman and a man, both
working and, earning incomes,
.. —"•••©'
v«miiig incomes, can rub
.. w°hU85
as bct
between
WCCn the Endless
landless and
and those
those
HIt«
?Ou!dc7 7™
istempting
temptingto
toconclude
concludethat
thatthese
thesedata
data ’^houldere
with big
big landowning
landowning families
families of
of
can, remaining in villages, earn high
v?who
no can.
ar
e consistent with the observation of * ,, 8 $.,ZC ,n
capita expenditure
are
,n3lhc
hc:Nss
NSS
h howler
incomes ifrom
agriculture.
point
srOrhata
?hriCU,-UrC- The P
°in'tios among
: •, .
balanceds«
sexraratios
amongthe
theagriculturalagricultural y c,a^>ncations.; :?-,
£ nowever.
however, is
is that
that there
there is
is a
a marked
marked sex— >.. labour and land-poor households. TheconThe con2:;;>
’-<„':_
rhat
uaiv >,’5^cquany
^ualIy. pov
obvious but, cannot.
r"’ . lv,.ly ln outm’8ration, with
............
V
males out- elusion may be
----valid but there is the nuisance' • agai
...n. be^substantiated
-a.,-.
J on the basis of
^numbering females, and this can—and p^oof two-way determination: the direction of ’ Pub,lshcd .Information of the sort being
S bably dow—influence the balance between
betweet
causation runs both ways betwcen low per '
here, is the implication of the dif< the sexes in not only the poor but also th<
capita incomes and high female proportions.
fcrcnl,aI between the earnings of man and
non-poor rural families.
Tn.hkn , 1- .
c
’ ■ •
^^bonr as well as cultivator
v z p16 ^Planation for the observed, more t Toc,a?oratcaI1U,c:,hcpurposehcreis . fam,hes with >mall landholdings women
• ;balanced, sex ratios among the labour and
o see whether and how economic status generally work in contrast to those with larce
.land-poor families does not. therefore, lie 1
proporlions- However, given holdings; ;'and these earning differentials
..’wholly in male emigration but possibly in
‘andhoId|ng and other resources, the size/- ; Table TJPe^age Distribution of
’v
klat.lVC]y
morality Qinerentiais
differentials
VL7
' narr0WC.r •nviiaiiiy
mposnionof families (with respect to Households in Each Landholding Class by
between males and females. No doubt, the agc and
of members) in turn influence:' fb<PER Cahta Expenditure.
^magnitude
,ar?i*>y’ In
income
two-way-V .>
^magnitude of the influence of migration on thc
Chc ,ani,I
<»nre and give rise to a two-way
I^URAVIndia 1955-56
fisex
ff.sex ratios in rural
^ausation
?ation of <he
the
per
nd •
r
r~ '
lul<u areas
areas has
nas to
to be
be worked
worked out
out cau
—
me relationships
relationships between
between pcr'?r^
per J' Land
L™
™?rc. satisfactory
M!i?factory manner
manncr than has been ^f.
,t.aJincome
t?coTc and demographic
^mogrzphic variables.'
variables: Po^r^d
in
a more
^Plta
Monlh <Rs)
y-possible in this note to make this more con- generally,
^cnc‘^Hy. family income increases• uith;
but
not\
prononiyatMy. wthaf dAcrCT) ^^9-12^:13-20
214-.,< t..
gdusive:
^elusive: studies of migrant backgrounds are family size "
-------^‘needed
for states
such----as ■'•'••■w®
Orissa ®*«U
and ____
Bihar
021,113 ,ncom« tends (p be immdy to
.^313 * S.
.
...
------------------------------.. .
s.w
.23.34 ;■.> - 43.33
related to fami|y
I
s> . having high
rates of
labour migration And
rain • <45'
'
holding class as it docs even among agri- X. 03 - 1.0,j^426'Wv.|g.55
-JT
i^lo conclude this section, it must be emc
7.19
phasised that mortality differentials between
cultural labour households. As a result, in 5 IX) - I.5^4.85>?"
1.5 ^^4.85^ -io.91
-20.91
4.24
fjhe sexes among th'-’ poor-as among
sample surveys large families get classified 13 ’67.62-a ? 27
74
7 2.5
2 5 ^§
f^67.62.-A
27.74
4.64
bthers-need not exclusively depend on
into the lower per capita expenditure groups.
33 ?$63.40?>
^63.40
31.46
“23.5
3146
. 5.14
For
iHustrative
purposes
some
out-dated
3
5
"
50
£357.76
34.84
, mscnmination’ (more on this later).
3.5 - 5.0 .^57.76
. 7.39 ’
NSS data (1955-56) are given in table 7,^60J2 •
30.32
9.45
r
_
which
shew
that
about
15.5
percent
of
big
7
*
5
“
>0
'->56.93
33.91
7.5
10
'X>56.93
Per Capita Arithmetic
8.66
[5
-:>54.32
landowners with over 100 acres appear with 10•10 ~15
4*32 ' >f 34 23
15 W
>>54.32
& Poverty calculations
11.43
per —
capita
of |less
than xr
R<12
- ”20
20 A-.>54.74
^54.74 ':;' 31.47
.--------- are usually based on
*--•
»-•— expenditures vii
C55 man
iz 15 I?
13.79
••a classification of households by levels of along with 33 per cent of landless house- - 20 - ::
30 :;?48.67. . . 35.90
15.44
<n
45
59
30
’
50
4'38.39y
/
.
.Pereapn3 (income or) expenditure It has
hoIds» and 77 per cent of households
16.02
.been argued elsewhere that while the latter
Possessing less than half an acre; correspon- 50
IAa - 100 <->31.58,
«.. •--- /V-:. • 33.41
35.01
- 100 4- - * /60.12. .<. 24.27
.^■15.61^
hhev^ZTn/ W-hCihyr Or not thcy arc poor’
..economic variables. The main reason is that
^such relationships are expected to depend on
^Ong-run, 1 e durable, characteristics of
j families whereas per capita incomes are sub
ject to wide fluctuations even in the short
period, caused by vital events such as births
,4,deaths. and migration of members within
; fam-hes (Krishnaji. 1984). Another reason
<is that per capita classifications create in--homogeneous groups in which agricultural
^labour families of small size may appear
^alongside landowning households o£ large
<size The resulting pictures can and do cause
7(much confusion.
< ''.But, for what they are worth, sex ratio
^variations according to levels of per capita
•expenditure are presented in table 6. These
':.a8^n refer to female ratios among adults
>
Sh.°W ,hat’ With a fcw exceptions,
• he.FMRtendsto be the highest in the lowest
s per capita expenditure group and vice versa.
Jhesex ratio is close to or over 1,000 in many
,
caplla consumpllon-
abOUt 7
■ 8.23
0001 Of thosc wi,h holdings bet- • 7
~
‘
Source: NationalSampleSurvey, Report No 140.
1973'74
-________
State
-
India
Per Capita Expenditure Per Month (Rs)
0-34
34-55 55-100
Above 100
Andhra Pradesh
1031
943
939
802
Assam
994
952
861
555
Bihar
1072
1126
1041
844
Gujarat
936
955
883
813
Haryana
834
939
942
847
Jammu and Kashmir
865
902
■
845
812
Kerala
1204
1048
. If'1030
1036
Madhya Pradesh
996
981
..-ft: 931.
841
Maharashtra
1045
1026
985
850
Karnataka
1021
964
;'918.
920
Orissa
1080
1010
.>.958
800
Punjab
1022
979
>•927
913
Rajasthan
986
992
\\ 953
988’
Tamil Nadu
1080
1021
•;.1002
854
Uttar Pradesh
997
956
?.-942
748 '
West Bengal
962
975
‘
828
788
^ote. Thr,Qr
.
f
-------------------—
Note:' The
.flarger
SCr- tnumhcr of c-xpendnure groups available in the NSS have UV11 v
ive been collapsed into
the four given above to av
Source: NationL^mp^/^-^-—^
by SmaI153011,16 sizcs in
---------1 some groups.
, le Survey, Report No 240.
895
ECONOMIC AND POLITICAL WEEKLY^
June 6, 1987
imply a distinct disadvantage to families
with more women than men, so that, other
things remaining the same, families with
higher proportions of females earn lower
incomes per capita. The same reasoning
applies to children. Children in labour
and small peasant families work and add
marginally to the household income but
their contribution tends to be far less in per
capita terms than that of adults. For this
reason, a large proportion of children, just
as a high female ratio, is an economic dis
advantage to the family. It can be clearly
seen then, that families with high female
male ratios and high proportions of children
get classified in surveys into the relatively
lower per capita expenditure groups within
i each landholding class. This is one reason
why high female proportions and high child
adult ratios and large families are systemati
cally observed at the lower end of the per
capita consumption scale.14
It has been shown by Visaria and Visaria
the property, depriving women not only of a
headship in a formal sense but of much elsc^
(1983) that among cultivators with small
landholdings the incidence of females as
This is speculative but Important noncthless,
for it highlights the role of the mechanisms 1
heads of household tends to be higher than
in other rural families (table 8). Moreover,
of property transfer and control in the deter-jj
minaiion of the relationships between sc$
they note that households with female heads
ratio and landholding characteristics.
>'4
tend to be smaller in size and to have higher
female proportions. There is some evidence '
A
At any rate, the smallness in family sizej
to show that female headship and high sex
ratios constitute a disadvantage to families among the land-poor arises in part from aJ
in terms of per capita consumption. In a greater incidence of widows heading the]
speculative vein the Visarias say. “Also
families constituted by them and their child)
females form a higher proportion of house dependents. The sex composition of such?
holds with small .t landholdings perhaps
households has a priori to be in favour of j
females because of the absence of the male?
because they are gradually forced to sell a
part of their landholdings or get a smaller spouse. They are likely to be concentrated;
share relative to the brothers of the deceas- among the poor—howsoever defined—as a',
cd spouse. Partly as a result, the female- consequence of the female disadvantage in'
headed households might include a high pro earning incomes and acquiring property.'^
portion of rural labour households..
Concluding Remarks
It is possible to indulge in further specula
Factors such as male outmigration and the<
tion. The Visarias find that female heads are
bunching of families headed by widows coo-.i
Thus the data based on classifications ac
mainly widows (or divorced or separated
tribute to the distinctly better-balanced see
cording to the per capita expenditure of
ratios among the agricultural labour and!
persons). Now (here is no reason why the
families are less informative about the deterincidence of widowhood should be very
small cultivator households, but they cannot:
minants of the sex ratio than those based
significaintly higher among those with small
decisively explain why these ratios often tendj
f - on landholding data. However, they reflect
plo^s of land; it can be so however, if male
to be over 1,000 in contrast to an average of
a different aspect of ‘discrimination’, that of
mortalities in the relevant age groups are 950 for the whole rural population, and even
unequal earnings, too well known and
higher than female ones by margins rela- ii lower values among households with big
. . documented to be pursued any further here.
lively wider than in (he bigger land size
landholdings uniformly all over India. The.
classes. This dearly points to the lack of conclusion must be that although mortality,
Small Families and Female heads
one-to-one correspondence between widow rates in general can be expected to high—'
~ According to the landhold: ag data as well
hood and female headship. To the extent that
higher than in the rest of (he population^
. as those from the rural labour enquiries.
______ r____________________
o._
female headship
results from male outmigrain such poor households, the differential bet;
labour and land-poor families tend to be
lion, this is obvious. What is not obvious is
ween the sexes may be relatively narrower in
small in size. The determinants of family size • that, whatever the spirit of law and custom,
some age groups. The consistency with
in relation to agrarian structure have been' ’ women fail in general to obtain rights to which pronounced deficits of females are
examined in detail in other studies.15 The
ownership of (or control over) land. Thus
observed in households with large land
association between small families and high
when landholdings are substantial, even if
holdings—for example in the south, when
female proportions deserves further attenwomen acquire legal titles, male relations
generally the sex ratios are more evenly
.tion, however.
may step in quietly or otherwise, to ‘manage*
balanced—lends weight to this conclusion
Table 8: Percentage of Households woth Female Heads among Cultivators according to Size of Landholding. 1961
Area
Below 1
1.0-2.4
23 -4.9
India
14.5
Andhra Pradesh
143
‘ 8.0
Assam
■'ihar
16.1
Gujarat
14.9
Himachal Pradesh
15.7
Jammu and Kashmir 10.9
Kerala
15.1
Madhya Pradesh
12J
Maharashtra
27.1
Manipur
202
Meghalaya
292
Karnataka
20.1
Orissa
11.5
Punjab
20.6
Rajasthan
10.9
Sikkim
4.9
Tamil Nadu
14.7
Uttar Pradesh
143
West Bengal
8.5
9.8
10.8
5.0
10.0
83
10.6
73
10.0
92
16.4
10.1
24.8 .
14.4
7.6
11.4
7.1
6.4
7.9
2.7
6.0
5.8
79
6.0
7.6
6.2
10.9
5.6
16.1
11.0
4.7
6.3
__ J
Size of Landholding (in Acres)
5.0-7.4
73-9.9
10.0-14.9
15.0-49.90
50+
5.1
7.1
2.0
4.4
6.5
3.7
5.9
3.4
4.3
4.3
3.5
52
7.4
7.8
4.9
8.4
32
18.0
9.1
3.7
1.6
3.8
4.4
69
4.2
6.3
1.5
3.4
3.4
1.3
3.1
2.6
5.7
5.5
4.5
2.0
5.5
3.2
4.3
6.4
7.5
7.6
4.0
6.2
6.1
12.0
3.9
6.9
6.1
3.7
4.5 "
5.0
22.1
3.9
3.3
3.2
17.8
8.9
3.2
4.3
17.1
7.8
3-2
1.7
3.2
4.8
6.2
3.0
2.2
6.1
7.3
32.8
4.8
3.5
1.5
2.1
6.7
6.5
All
5
7.2 V
'8.8 <
3-5
8.6
4.9
9.5 ;
73 ■
12.0 .
5.5 .
9.5 A
7.9 .
20.4
10.0
.
3.2
5.8 <
1.5
3.9
4.0
2.8
4.1
8.6
6.5
4.7
5.3
3.7
6.1
11.7
8.9
7.5
6.5
5.4
10.1
93
5.6
4.0
3.4
2.9
3.3
6.9 4
5.8
3.1
2.4
2.1
4.6'2.8
5.0
Sourer. “Census of India". 1961. Part IIC (i). “Social and Cultural Tibies", table C-l. published in state volumes; cited in Visaria and Visaria (198
896
3.2
2.1
3.4
ECONOMIC AND POLITICAL WEEKLY
June 6. I9S7
This indirect inference drawn from the data
on sex ratio balances discussed in this note
no doubt requires more direct empirical sup
port; also, there is scope for further analysis
at the regional level of even this kind of
secondary information.
However, it will be harty to conclude that
there is less discrimination against females
among the poor, for mortality differentials
cannot be wholly attributed to 'discrimina
tion'. For instance the higher male death
rates observed in some regions among some
age groups do not imply discrimination
against males. But the discrimination against
females in India is real and its different
dimensions are well known, what even
statistic reveal or do not. It is possible that
-&crii»inatorv practices —especially in rela
tion to nutrition and health care—are more
effective among the landowning classes than
in poor families, given the very low standard
; of living of the latter; the bias may be univer.sal but it seems to be stronger among some
caste groups and in some regions of the
country. Theorising about this bias is not
easy, however.
U No doubt, the economic value of a
•woman, calculated (by economists who do
not distinguish human beings from com
modities) on a long-run basis, may turn out
to be higher for a labouring or small culti
vator family than for a big farmer, given th$
cropping patterns and the possibility of
women earning incomes. But as an explana; tion for the observed sex ratio variation with
better balance among the labouring poor
> this calculation is suspect because it is dif^ficult to believe that it is actually made in
some fashion or ingrained in culture through
economic consciousness and underlies sex
. bias or the lack of it. For the poor, work is
directly related to survival from day-to-day
-and long-run calculations may not be rele
vant. It is necessary therefore, to look
•beyond a simple economic calculus for map
ping and understanding sex ratio variations.
\
Noles
>1 The number of females for every 1.000
' males in India as a whole has declined,
although not steadily, from 972 in 1901 to
-. 935 in 1981 (Padmanabha, 1981).
• 2 Visaria considers other factors such as
spatial mobility, under-enumeration of .
.. females and sex ratios at birth, before arriving at the importance of mortality dif£: fcrentials for explaining the deficit of
females. During the pre-independence
* • period, the census authorities thought that
the deficits of females arose mainly from
the undercounting (resulting from under
reporting) of females in certain age groups.
•s Even while recognising female infanticide,
p the ‘neglect of female life’, and ‘bad treat• ment’ of women, they refer to the ‘conceal• ment’ of women belonging to certain age
S . groups among some agricultural and other
high caste communities as an important fac. tor underlying low female proportions
(Natarajan, 1972).
3 Some recent data show that upto the age
of 35 years mortality tends to be higher
among females (Padmanabha. 1982).
Female mortalities are lower after the age
of 35 years but life expectation at birth is
lower for females than for males.
4 Visaria (1961), p 66. He suggests in conclu
sion that the discrimination “denies to many
women the benefits of the normal biological
superiority of their sex...“ The superiority
is presumably inferred from higher male
mortalities generally observed in the western
countries.
5 Apart from the difficulty of observing and
assc sing amounts consumed by individual
members at a single meal, it has been sug
gested repeatedly that there is also the
possibility of a bias introduced by the
presence of an observer.
6 For a comprehensive discussion see
Agarwal, op cit.
7 The 1911 Census says: In the Benares divi
sion Moore personally made most minute
investigations into the facts in three hundred and eight villages; in sixty-two of these
villages he found that there were no female
children • ader the age of six years. In
another part of the division. Moore found
a community of Hara Rajputs regarding
whom he said. “Not only are there no girls
to be found in their houses now, but there
never have been any. nor has such an event
as the marriage of a daughter taken place
for more than two hundred years...
“.. .the extraordinarily low female ratio of
the Shckawant branch of the Kachwaha dan
of Rajputs in Jaipur sure. 530 females per
1,000 males, is indubitably suggestive cf
deliberate interference with the natural ratio,
when considered with the Rajpur tradition”
(quoted in Natarajan, op dt. p 4). These
may be extreme cases but they illustrate the
point.
8 Extracts from the different early Censuses
in this respect are given in Natarajan, op dL
9 For a discussion of Miller’s work and some
extrapolations, see Bardhan (1982). The
author of this note has not read Miller’s
book.
10 Agarwal (1986) discusses this expectation in
some detail and refers to some micro
evidence showing the contrary, higher levels
of discrimination among the landless poor
households. However, such higher levels
may coexist with better female proportions
(aggregated over all age groups) for reasons
to be clarified later in t^iis note
11 During baseline years fe/nale mortality consistently exceeded mal^ fnortality in all age
group* except infant deaths. The age
specific sex differentials were more pro
nounced in children 1-4 and 5-9 years and
in the childbearing years. Disaster tended
to accentuate these sex differentials among
children. In 1971-72 [a year of food crisis]
mortality of female children 1-4 'rars was
57 per cent higher than mortality of males
in com parison to a differential of 40 per
cent in the preceding five baseline years”
(Chowdhury and Chen, 1977. p 53)
12 Sec Greenhough (1982), p 311.
13 Referring to fewer deaths among men than
women in the 1908 famine, the 1911 United
Provinces Census Report says: “This is at
tributable chiefly to the absence of wander
ing. This absence of wandering was... due
to the fact that the people by 1908 had
learnt by experience that government was
anxious and willing to assist them. In
1897... they had not yet obtained such con
fidence in government and took to...
wandering in search of work... It is these
wanderers who feel the worst effects of
famine, it is chiefly they who starve. And
it is amongst them that man would most
severely feel his disadvantages and women
would reap the fullest benefit of her advan
tage” (quoted in Naurajan, 1972).
14 For further discussion of such correlations
sec Krishpaji (1984 and 1986).
15 Ibid.
References
Agarwal, Bina, 1986, 'Women Poverty and
Agricultural Growth in India’, Journal of
Peasant Studies, 13. pp 165-220.
Bardhan, Pranab, 1932. ‘Little Giris and Death
in India’. EPW, pp 448-50.
Chen. Lincoln C, Emadul Huq and Stan
D’Souze, 1981. ‘Sex Bias in the Family Allo
cation of Food and Health Care in Rural
Bangladesh'. Population and Development
Review, 7, pp 55-70.
Chowdhury, Allauddin A K M and Lincoln C
Chen, 1977. ‘The Interaction of Nutrition,
Infection and Mortality during the Recent
Food Crisis in Bengladesh’. Food Research
Institute Studies, XVI, pp 47-61.
'
Greenhough,
Paul R, 1982. “Prosperity and
Misery in Modem Bengal—The Famine of
1943-44”, Oxford. New York.
Krishnaji, N. 1984, ‘Family Size, Levels of Liv
ing and Differential Morulity in Rural
India—Some Paradoxes', EPW, pp 248-258.
—, 1986, “Size and Structure of Agricultural •
Labour Households in India’, Working
Paper. IIM, Calcutta.
Mehrotra. G K. 1974. “Birth Place Migration
in India". Census of India 1971, Special
Monograph No 1.
Natarajan. D, 1972, “Changes in Sex Ratio”.
Census of India, 1971, Census Centenary
Monograph No 6.
1 Oberai. ASandHK Manmohan Singh. 1983,
“Census and Consequences of Internal
Migration”, Oxford, Delhi.
Padmanabha, P, 1981, “Provisional Population
Totals, Series-I, India, Paper I”, Census of
India. 1981.
—, 1982, Trends in Mortality', EPW, pp
1285-90.,
Saxena, D P, 1977, “Rururban Migration in
India”, Popular Prakshan, Bombay.
Sen. Amartya, 1981, “Family and Food, SexBias irr Poverty* (mimeo).
Visaria, Pravin M, 1961, “The Sex Ratio of the
Population of India", Census of India 1961,
Monograph No 10.
—, and Leela Visaria, 1983, “Indian House
holds with Female Heads—Incidence
Characteristics and Level of Living", paper
presented to the workshop on Women and
Poverty, Calcutta. Centre for Studies in
Social Sciences.
*
897
SPECIAL ARTICLES
Differentials in Mortality by Sex
Malini Karkal
of males and females at birth m India has shown a steady decline since
/PW Xp
THE United Nations, "Estimates and Pro
jections of World Population (1982)’’,
reports, among other Figures, life expectan
cies for populations of the world from
•1950-1955. A review of these Figures shows
that there are only 7 countries that had life
expectancies for females that were low than
that for the males. These countries were
.Bangladesh, Bhutan, India, Nepal, Pakistan,
Sri Lanka and Papua New Guinea. For all
Hthe countries, excepting for Sri Lanka, the
^pattern of low expectancy for females con
tinued throughout the 20th century. Sri
kLanka, hower. showed higher life expec
tancy for the females, compared to that for
i^nales, during 1960-65 and thereafter. It is
^interesting to note that the death rates for
s^hese countries differ and no'relationship'is
Bndicated between the level of life expectancy
>d differentials in mortality by £^Xh
4“^ h“
® ’,g,uf‘ca"t dedine “ t*.■ that improved hralth is not only a social goal V
mmXme^T'inJb“l « « a means, ahd'indeed an indispenshe
I V 1
°f-b a-"8 °f ab ' comP°nent and a prerequisite of social .
erad>P
d^Se an<1 anomic development.
'
•
M the above mentioned countries are among
nhe higher mortality group.1
■ *
?: .The acceleration hypothesis postulated
;that the later a count^ enure
Semo-
support 7o thVmJdiil ?'hr bCe1’suggested that differentials in
th«t
"xhnology hypo- ; mortality of the two sexes reHect the difh“ no^ ^"le apparem; ferences in’their biological make up In
en^ "a"” -cX^r
’ndl". l^her modify for
graphic transition, the period between thtraditional pre-industrial high wasmge of
(population replacement (high mortality and
mentis nrX^X r
T
-'7
- 7" ”,S a renection of the role and status
mXlS«
’ subs“ ^d . of fernales. Jpth'within the family and in '
of^nm.
T*0" ‘n mortahty. This hake society at large, as much as they represent ■
;nanh fer,iliry)/nd 'hC m°dcrn ‘"nnomicaP
jPatlerns or demograpbic behaviour (ijw
advakc^enTbut ^a/XTcul u^'T" ' *hC.hea]tH
yz-y../ table
^'“.hmulh'Th fcr,il-'y)‘ 'b' raS,<^t - P°Pulali°^the discussion hereafter will \
hypothesis implies an automatfcXfer of • jgjSl
,cchno'°R'^'P0'hesLSpostu-. understanding of levels of mortality not
not ( ’
broVe^hrr1?"^
,cchnol°W h“? only as averages, but as experienced ^dif-:
development and
Rates by Sex 1nd[a
—
-_______ ,ndia
1970-75
fe(lj.^.:r. (2)
Female
■ (3)
fe®?
-1^ .13505
I-1
-07567
.10301
gO"""
.02613
.01139
^5■'•-01401.
Jo2266
-01546
.02266
.02593
.02852
.03260
.04618
.06918
-02050
.•■•°3854 ■
psax ..
BRw25?8i>>;-22r6
and Political Wccklv
Difference
(2-3) = (4)
-.00507
■ -.02734
-.00347
-.00134
-.00587
-.00865 .
-.00717
-.00543
.00034
.00594
.01218
.01500
.02823
.02295 .
.02495
0
^dition?afto“ge<he >
CORRESTOSD|SC
-- ------------Pattern India-
' •
•.
■ South Asiaji Pattern
1970-75 V •
Male
(5)
1976-80_________
Female Difference
(6)
(5-6) = (7)
Male
(8>
.12100
.06774
.01815
.00946
.01055
.01322
.01361
.01844
.02471
.03830
.05239
.08165'
.12051
.18317 .
-24215. '
1
.12720
.08992
.02266
.00976
.01460
.01972
.02129
.02060
.02354
.02871
.03744
.05865
.08975
.14687
20472
I
.13704
.09063
,.02047
.00871
.01030
.01215
.01480
.01834
.02480
.03532
.05120
.07714
.11017
.16409
.23169
1
-.00620
-.02218
-.00451
-00030
-.00405
-.00650
-.00768
-.00216
.00117
.00959
.01495
.02300
.03076
.03630
.03743
0
UN
of social, economic
, 1976-80
Female Difference . Male
Female Difference
(9)
(8-9) = (1Q)
(II) f. (12) (11-12) = (13)
.13901
-00197
.12656
.12584
.00072.10845
-.O17n3 y: .07977
.09083 ‘ -.01106
.02472
-.00425 <’■ .01803
.02035 -00232
.01058' • .-.00187 3i;OO775
.00873 -.00098
-01516 -.00486 ;
00922 •- .01242
-.00320 ■ ■
.01839
-.00624 i, • : .01088
.01504
-.00416
.01989 -.00509 ': .01331
-.01645
-.00314
.02339
-.00505 . ‘.01650
.01957 -.00307
..02713
-.00233
.02247
.02316 -.00069
.03295 ( .00237 ; ,.03234
.02879
.00355
.04291 .00829 *.• ?. .04743 . .03822
.00921
.06481.;
_ ' ______
__
<.t07238
-05855
.01383
?•
' -09893
.. .01124 C*So456
.01408
' -15139M .01270 ^15720
^K-790' • .09048
■1.15I39M;
- .14014
' .01706
. 21781 .01388 ; * 22337
22337- '.20453
.01884
i
o '
1
0 ■
August 8, 1987
■1343
for females it was 3.146 years. The gain in TI.
significance. One, women constitute almost arc
arc at
at greater
greater advantage
advantage in
in comparison
comparison to
to
I half of the population and to that extent the males from the now developed c uniries
the life expectancy for females was therefore i
larger.
*
their condition contributes to the overall when they passed through mortality transiColumns 4 and 5 show that for males as S
hcalth situation and second, women’s lion. However, this advantage has shown
health determines the health of the future some decline over time.
well as for females the gain was larger for |
higher ages than for lower ages and for
population.
Tabic 1 gives a comparison between mor1 he gap between life expectancies at birth tality rates for India and the corresponding
females the percentage gain was higher for ’
of males and females in India showed a figures for the South Asian model pattern.
ages 55 and above. Significantly, largest y
decline from 1.7 to 1.5 and to 0.4 years for Table 2 gives the percentage difference in
share of the gain was taken by ages 70 and x
the periods 1961-70/1971-75 and 1976-80 mortality rates of females compared to those • above. Here again the share was 25.10 per >
respectively. Along with tliis decline has been, of males by age-groups for India and cor- cent of the total for males and 33.67 for
.
the change in the sex ratio of the popula- • responding South Asian pattern.
females.
lion from 941 (1961) to 930 (1971) and to 933
The figures in Thble 3 therefore show that X
The figures in Table 2 show that females
(1981). The reversal of the trend in the sex in both groups of populations arc at a dis- whatever gain in life expectancy that has^<‘
ratio from 930 to 933 has been welcomed by advantage when compared to the males in been made by females in the period interven- X
several experts. Chhabra, while reporting on their population in the younger ages. For
ing between 197C-75 and 1976-80. the larger y
the provisional results of 1981 said “the only India this age-group is upto age-group 30-34 share has gone to older women in contrast.^ silver lining to be noted from the census • and for -the South Asian Model it is upto to the agewise distribution of gains for{
results is the marginal improvement in the age-group 35-39. Thereafter females arc at males.
T
s sex ratio’’.2 The Registrar General, com an advantage.
Table 3: Gain in Life Expectancy in IndiaT
menting on the census results, opined that
It is worth noting that over lime in the
Between 1970-75 AND 1976-80 by Age f
“the tendency of the sex ratio to deteriorate SouthjAs^an pattern the difference between .
has been halted and that in fact there has the mortality rates of the two sexes has
Age Group
Absolute
Percentage
been a slight improvement”. He further narrowed at ages when women were at a
Male Female Male Fcmale
added that “one’of the conclusions that one disadvantage, whereas in ages above 35 the
(1)
(2)
(3)
(4)
could, at this initial stage, come to is that male disadvantage has increased probably
Total
1.966
3.146
100.00 100.00 I
. probably maternal .and child care pro due to greater gain in mortality of females
0
0.014
0.006
0.73
0.18 f
grammes are yielding results”?
in .higher ages.
1-4
0.034
0.052
1.70
1.65
/ • In the light of overall belief that in the
In contrast to the above pattern the
3.12 |
5-9
0.086
0.098
4.35
< recent years there has been an improvement. females in India not only have larger dif
10-14
3.37 %
0.095
0.106
4.84
in the health of women—the sample of such • ferences, compared to the South Asian
15-19
0.095
0.114
4.83
3.63 «■
v belief is provided by the above mentioned ; pattern, for both the time periods, but for
3.821
20-24
0.096
0.120
4.90
comments on the trend in the sex ratio of first year of life for the age-group 5 to 9 and
25-29
. 0.099
0.126
5.04
4.02 g
30-34
the Indian population—it is necessary to
0.106
for 25 to 29 the disadvantage has increased.
0.136
' 5.42
4.32 Jj
5.84’
35-39
0.115
0.150
• make a careful analysis of the changes in the
The disadvantage in ag^ 40 and above for
40-44
0.120
0.161
6.11
• mortality patterns over time and inithe males, for both the time periods, is not only
45-49
6.37
0.125
0.176
relative changes in the two sexes.
larger in India compared to that for the
50-54
0.132
0.198
6.69
6.29 R
The narrowing of the gap between'the 5?u\h
Pat^rn, but over time the
55-59
0.130
0.211
6.70$
6.61
disadvantage
for
males
has
increased.
expectation of life of the two specs does
6.75 J
60-64
0.120
0.212
6.09
In the light of the above discussion it is
indicate that there is a convergence in the
6.99 Vi
65-69
0.106
0.220
5.38
figures over time but this needs to be analy worthwhile looking into the change in
70 +
0.493
1.059
25.10
33.67?
mortality rates of the two sexes o^rr time.
sed to understand the changes in different
is essential
Tablc 3
lhc 8ain in lifc expectancies
• age-groups. *This
___ ..
________ because
_______
Source: Computed from SRS life tables.
betwren 1970-75 and 1976-80 it. India.
Sample* R'egistritTon’sJitem”i97O-7j’averages do tend to hide large differentials
______
r______
From column 2 in Table 3 it is seen that
and Sample Registration Systcti;
in some.............................
groups when these are compensated
studied the
1979-80.
|
1 on
on othe^Lopez
omcrs-jcopcz siuoiea
me relationship
reiauonsnip .lhc ^ain for males was 1.966 years, whereas
between m\le and female life expectancies Table 2: Per Cent Difference in Mortality Rates of Females Compared to Those of Males;
at birth based.on trends in population of 139
by Ages for India and Corresponding UN South Asian Pattern
;
developed countries. He concluded that the
South Asian Pattern________
India
relationship is highly linear (q = 0.98901. Age
Corresponding :
Corresponding
1976-SO
1970-75
He, therefore, presented an equation for the
to 1976-80
to
1970-75
linear form.4 The equation is
(5)
(4)
(2)
(I)
(3)
e(m) = 0.8788 eo (f) + 3-541
Applying the above equation to the data
0.57,
- 1.44
- 5.12
- 3.90
0
-13.87
-19.67
-32.74
available from the Indian Sample Registra
-36.17
1
-12.86
-20.76
' -24.85
-«5 31
tion System (SRS) following results were
5
-12.65
-21.46
- 3.17
10
-13.33
obtained:
/x
-34.71
-47.18
46.6 15
-38.39
1970-75 for eo(f) = 49.0, e(m)
-52.45
-38.24
-49.17
51.36
-61.74
20
e^)-e0(m) = -3-9
-23.59
-31.39
-56.42
-46.29
25
1976-80 for eo(f) = 52.1, e(^) = 493 30
-18.61
-27.54
-11.71
-26.49
- 3.07
- 9.40
4.73
1.18
35
eo(m)-eo(m) = -3.2
10.98
6.73
25.04
15.41
It is observed that during 1970-75 the life 40
19.41
16.19
28.53
20.87
-xpectancy at birth of the Indian males was 45
19.11
15.98
28.17
17.82
’ 50
higher by 3.9 years than the one expected by
13.47
10.20
25.52
22.41
55
the observed life expectancy of females of
10.85
7.74
19.82
12.45
60
49 years. Over the years this gap was reduced
8.43
5.99
15.48
9.95
65
to 3.2 years by 1976-80 when the life expec
70 >
tancy of females increased to 52.1 years.
Sourer. Computed from data in Table 1.
Thus Indian males, relative to the females
— I-
4
s-60*-
____________ f
Economic and Political Weekly
August 8,
Table 4 presents the differences in life ex
pectancies of males and females by ages and
lor 1970-75 and for 1976-80. The difference
in life exoectancies of the two sexes in the
earlier years was 1.544 years whereas in the
later it was 0.365 years. It is interesting to
note that in earlier years upto age 50-54 the
percentage difference was much smaller
compared to the later years and in first year
. of life it was the females who were at an
^ advantage. For age group 55-59 in earlier
/ years males had an advantage, howev _t by
■ 1976-80 the females were at an advantage.
The difference is in favour of older females
■ for ages 60 and above in 1970-75 and for
.‘ages 55 and above in 1976-80.
i It is therefore, observed that Indian
females are at a disadvantage as compared
to Indian males and over time whatever
gains that are made jn life expectancy of
^.females a disproportionately larger share has
gone to females in older ages. This observaJ^°n belies the argument (hat there is
S:?C yD,un8er
B
relative neglect of the female child is erident
kom the fact of greater prevalence of growth
retardation even in infancy among girls than
l ,tSU/h nulrJt,onal ,insu,t’ commencing right
infancy and continued
1 from
"
through all stages of development that even
tually results in maternal health/nutritional
status which harms not just the women but
the succeeding generation as «.di.”7
International standards for satisfactory
performance at delivery of women indicate
that women with weights of 38 kg or less
during Riegnai,^ and 42 tg or les. duri.g
me last month of pregnancy, and those with
heights less than 145 cm are to be considered
as being at risk during pregnar .-y. Such
women experience complications during
pregnancy or at delivery and have babies of
low birth wnght whose growth and development arc unsatisfactory. The data from differcnt states in India clearly indicate a
distressingly high proportion of women with
. ‘pregnancy risks’.
In keeping with the above observations..
and especially those in th<
gHEACTH Status of Younger Women ‘
gfeg-t’*’
■
g^jn the context of the above discussion it
£will be worthwhile looking at the data that
Jgives a clue to thc health status of women
i^y^CTagwnndthatofthechildrenbom
‘Birth weight is strongly conditioned by the
^health and the status of the mother, her early
£dict.and maternal malnutrition. Ill-health
Sand other deprivations are the most com®mon causes of low birth weight and retarded
fe^fetal .growth.3 Countries with lower life
Ktxpectanaes for females, mentioned earlier,
gsreport Ingher incidence of low birth weight
Thcsc countries also report higher
Mincidenee °f perinatal mortality and a larger
^share of infant mortality during first four
Sleeks of life. Indian data show that the
fijwk8111 morta,ity is around 60 per 1,000
P^^ 5ndurOU.nd 50 10 60 pcr ccnt °f the
ilftr-f1 ^hS takC placc durin8 f*rst 4 weeks
pattern of deaths, and with the'
agTeportcd higher incidence of low birth weight
fcbabies^are . dear Judications of poor
Kvmaternal health.' .
C^owdhury point out that
angladesh and Pakistan and in parts of
Kn^diSad^n’ag' Of Sirls begins
Pcriod-6 Gopalan
and says ,hat ‘■,hc ■
The effect of malnourishment as an
Table 5: Life Expectancy at Different Ages
by Sex-India
f, c .
Age
thc yean thc health of younger women
> :ha: improved was also provided by (he SRS
J^Iife expectancies at each age (Thble 5). These
^data show that in 1970-75 as well as in
LJ976-80 from age 5 females have higher life
’•.expectandes compared to males for thc same
^periods. What needs to be realised is that
^hclifc©cpectancies calculated for younger
gages!are influenced by the changes in
{chances of survival at older age. Hence the
JMrger gains observed for older women and
f presented in discussion so far were respon
sible for this misleading picture
born of the malnourished and infested
mothers in the developing countries arc poor
in health. The data, therefore, show (hat late
neonatal and post-neonatal deaths are now
uncommon in developed countries. In many
of the developing countries however, they
account for about two-thirds of all infant
mortality. The inter-American investigation
of mortality in childhood points out that
malnutrition is an underlying cause for 57
per cent of the infant deaths in some of the
countries.8
----------unlls wln
‘--Special
care’ or ‘premature’ units
with
well-equipped set-ups for these low births
weight babies do show survival rates com
parable to those found in some of the
developed countries. However, a follow-up
cf
,v per
of these babies showed that 70
per cent
cent 01
of
the infants ^discharged from the premature
.....
nursery were dead
within three months.9
22 million low birth weight babits are born
annually in the world and 21 million of them
are bom in the developing countries. The im
portance of this condition can be well
understood when it is pointed out that there
arc strong indications that these babies con
tribute to a large
of deaths and
_ proponion
. . _________________
child morbidity, the risk of mortality being
upto 20 times higher for these babies than
for other
ocher babies, both in neonatal period and
later. Further it needs to be mentioned that
-- - two-thirds of all babies of low birth weight
bom in developed countries are estimated
to be pre-term (i e, less than 37 weeks of
gestation). In developing countries on the
other hand, three-fourths of all babies of low
birth weight are full term, but significantly,
- ----------undcmourished and small for gestational
“ is
‘ a clear indication that the babies
age. This
0
1
5
10
15
20
25
30
35
45
50
55
60
65
70+
J 1970-75
Male Female
1976-80
Male Female
^;5O3J ' 49.0
52.5
52.1 v 57.0 V/ 55.6 . 58.6
™!^.58.6 .
-^573
57.7< 58.8
" ’ ^ 60^.;'' .
v/53.8
542 - 54.8 ^ 56.6 •
• <49.3‘••'^49.8
50.3
52.1.
44.8 J; 45.6 ; . .45.8 • .,,.0
47.8
^■40.4^ 41.6'; . 41.4;j743.7
,^36.0 ■>‘•.37.5
------36.9 '139.6
.
■<-31.7
33.4
32.6 ■ 35.4
27.6
29.3
28.3
31.2
23.5
252
24.3 ■ 27.0
19.8
21.3.
20.5
23.0
‘ '16.4
17.7
17.1
193
; J’13.4
14.3
14.1. 1?-15.9
;7
j1o.9 :.:K6.
;370.9
11.7 ;;:;-132\
.^8,6 jj92;
. 9:6^10^
Source: Sample Registration System, 1970-75
and 1976-78. , ?
J.-J?
Table 4: Difterexces in Life Expectancy of nTales and Females by Age ; y •.
'
Age Group
(I)
Total
0
1
5
10
15
20
25
30
35
40
45 •
50
55
60
65
70 +
____________
Difference in Life Expectancy •■ • • •
:
________ Absolute
Percentage ________
—1970-75_________ 1976-80
. 1970-75
1976-80
<7)(3)
(5)
■
1-544
-0.004
0.080
0.146
0.151
0.168
0-190
. 0216
0-235
0238 .
0-2210-lfl '
0126
(1'050
-0.030
-0.078
-0-346
'
. .
’
0.365
0.005
0.061
0.134
0.140
0.149
.
0.166
0.189
0206
0203
0.180
0.130
0.060
- 0.030
-0.123
-0.192
—0.911
■100.00
.-027
• 5.18
9.48 ,
. 9.78
10.85 '
1227
.2 j 13.37;. .
■2 ''.:1523”'•
15.43
— '"1428
V
H.7I .
’ / 8.17
3.26
'
-1.95
— -5.03
'-22.37
. 100.00
123
16.83
36.69
. 38.46
40.73
-.45.47
■.51.71 • '
'■ 56.45
<• 55.72
49.18
35.66
16;39
-8.30
■ -33.68
-52.61
-249.95
Source: Computed froi
—Life
“ Tables.
from S R S
.
Sample Reginrrnion s^em 1970-75 and Sample Registration Syst.
:em 1979-80.
rtjgai^Wcek 1 y . August 8, 1987
• .1345 .
underlying cause /becomes, much more libiljty to diseases of childhood, such as
and children.
:
significant in the case of female babies, since measles, pertussis and pneumonia. These in
A medical check-up of the individuals W*
as pointed out earlier in Table 6, there .is ■ combination with continued malnourishshowed thqt women receive lesser share of
cvidence that in countries such as India, ment hast higher fatality rate. For the babies
food compared to the men in their families. «' ’
female babies get less nourishment and show
ho manage to survive through these con- In populations
r_,
such as those in Maha-3 *
greater malnourishmcnt. Further it is now ditions there may be serious chronic damage, rashtra, that seem to have made the transi- J:
conclusively proved that the lite-long under Some of this is already apparent in child- tion to relatively lower mortality for females,
nutrition of the mother, extended into her hood such as blindness and paralysis, while females when compared with males, show |
•pregnancy, is.more serious for the baby than other scquelea become manifest later in life
incidence of morbidity for several 4
an. acute nutritional disturbance during such as lower resistance to infections, chronic
diseases however, the incidence of anaemia
* pregnancy in a previously well-nourished heart disease, and mental retardation.,Thus is higher among
o them.12 thus supporting ?.
' •/ mother. The long-term effect on the child the neglect and under-nourishment of girls the
«t____ t______________________________ . .t
belief that it is not the poverty alone that •?*..
/• are more severe and may be devastating when
vicious cycle of deteriorating health, is responsible for the poor health of females, t
intrauterine malnutrition is followed by of population, stagnation of death rates and
but their social neglect.
’ •
.
malnutrition in the firs’ months of life.10 continuing high mortality among infants
Dyson argues that excess male mortality^
There is now evidence th it undernourished
Table 9: Comparison cf Survivals by 1Birth Weight Between Birmingham (20)
mothers when breast-feeding, do not meet
and
Imesi. Nigeria
the nutritional requirements of their babies,
•Tj ; It is also found that the infant bom to the Birth
---------- «S’illbirths/1.000
Neonatal Deaths/1,000
Mortality
1-12
months/
mother aged 30 and over is likely to receive Wright
_ Total Births___
* 1,000 Total Births
Total Births
too little milk of low fat content, and (8)
Bir n>*ngham
Imesi
Birmingham
Imesi
Birmingham
Imesi
y, .-'. therefore obtains too few calories and later
develops marasmus.1’ Continuation of -1.000
448
429
931
750
118
1,000
1,000fertility among women at higher ages can
436
612
200
64
444
1.500192
.«4.7/4 therefore affect the health of their children
95
224
211
37
286
2,00087
?- in their early ages as well'as adults.
47
77
49
18
103
225046
14
36
13.7
in developing countries takes
19
111
2200+
10.6
22.9
7 f-;: place by the second year of the baby. Poor
6.8
15.2
8
38
22.1
272
15.8
24.9
Posl intra-uterine conditions influence the All weight
9
54
birth weight babies and such babies Source: David Morley (1973). -Paediatric Priorities in the Developing World”. Post-graduate?^
when undernourished shiw I.igher suscepPsedU.rics Series. Buuerwonh and Co Ud.London, pp' 79-°80, p93.‘
** — “ “
®
““
*-*
■*
*
*
•
•
..
-iff• •
_
’F
Centre
r-. < Excellent Normal
I—
. Bombay
•’} Calcutta
Madras
I
Table 6: Malnutmtiox on the Comets Scale in Male and Female Infants
•. '■ =<w.
• : ' a<^5.4- ’
1C.2
'
21.9
14.2
19.9
_______ Male______
Retardation Grade
II
III
NR
N
42J
42.7
41.7
2.0
4.5
2.1
958
667
846
183
25.0
23.0
2.8
82
3.1
Excellent Normal
8.2
5.5
5.6
Female
Retardation Grade
1
II
111
NR
25.4
27.5
28.3
• 4.7
1.8
17.4
432
14.8
36.9
43.8
14.8
4.5
10.6
5.7
1.3
Notes'. N = Total Numbers of Sample. NR = No response.
Source: From C Gopalan’s article, ref 5.
■ '
20-24
’ 25-29
30-34
35-39
• 40-44
;
Kerala
Tamil
Nadu
Karnataka
Andhra
Pradesh
Maharashtra
Gujarat
20
' 21
23
20
22
23
21
22
26
28
22
24 .
27
25
28
24
25
15
20
21
24
25
27 ■
•...■34
23
22
24
30
29
32
Madhya
Pradesh
Orissa
17
15
16
18
19
16
24
22
28
29
883 f
£
. ’. J
Table 7: Percentage of Female with Weight Less Than 38 Kg Calculated- from NNMB Data 1974-79
Age Group : - .■• i/j.
862
710
West
29
32
35 .
42
43
Pradesh
f
17
20
24
25
26
*1
1
Note: • Calculated on the baris of values given for means and standard deviations assuming normal distribution.
distribution
Source: Report for the year 1979 NNMB, NIN, Hjxlerabad, 1980. quoted from Gopalan’s article, ref 5.
Table 8* Percentage of Women wsth Hejc.ht Less than 145 cm Calculated* from NNMB Data 1974-79
Age Group
20-24
25-29
30-34
35-39
40-44
Kerala
• . (1781)
'
20 •
'20
22
:■< 24
30
Tamil
Tamil
Nadu
Nadu
(1827)
Karnataka
- (2573) ,
14
14
14
14
16
12
12
14
18
Andhra
Pradesh
(2131)
16
15
17
16
18
Maha
Gujarat
rashtra
(2376)
(1995) '
15
17
21
24
24
Mad Hi's
Pradesh
(H28)
Orissa
(608)
West
Bengal
(1641)
Uttar
Pradesh
(1577)
12
13
16
17
13
16
14
17
16
18
23
25
22
27
22
21
22
25
29
29
22
15
22
25
26
Note
The percentages have been calculated on the basis of mean and
and standard
standard deviation
deviation values
values using
using normal
normal probability
probability tables.
tables.
+ The value given for standard deviation is not reliable and the percentage figure is not calculated.
Source. NNMB Report for the year 1979, NIN, ICMR. Hyderabad. Figures in bracket indicate total sample size. Quoted from Gcpalans
----------- - _.r r
article,
ref 5.
I '
>
1346
■*
Economic and Political Weekly
August 8, 1987
;
■
‘
Table 10: Sex Ratios (F/M x: icxj) of De.aths-Urban Maharashtra. 1982
Ca^jsc of Death
Group
1
1-4
0-1
5-9
10 14
___________ Age Group
_________________
15-19 2024 25-29 30-34 35-39 40-44 45-49
1
2
86
94
90
75
3
6
96
73
112
97
lf*l
85
82
65
100
83
91
80
66
63
89
56
93
71
67
96
71
72
54
127
81
83
77
75
91
97
78
71
70
103
65
88
75
69
61
79
94
76
88
86
63
83
72
98
54
72
119
90
93
87
96
95
91
82
78
7
8
9
15
16
17
All
Population 1981
Maharashtra
53
55
124
45
50
81
88
83
61
51
127.79 ‘
83*
91
71
57 ■
55
81
47
30
72
77
68
66
64
49
64
90
53
56
78
78
87
88
S3
78
91
’37
90
86
73
51
60
25
32
69
72
• 57
49
53
22
49
53
35
44
■ 45
48
50-54
55 +
29
41
62
71
55
56
73
33
63
66
54
48
49
26
48
33
43
99
54
61
To’.al
56
64
84
70
76
39
75
88
60
66
73
79
76
92
85
Causes of deaths: Group 1: Infectious and parasitic disease; Group 2: Neopl^imi; Group 3: Endocrine, nutritional and metabolic disease; Group 6:
Diseases of nervous sys.ens; Group 7: Diseases of drcuWi^^
3(OrV SVStCTTK CfTrttm R*
nf rr-crUrattvetOu:___ _:___
system; Group 15: Perinatal causes; Group 16: III /defined conditions; Group 17: Injury and poisoning,
I nnton
rvn
C
_____ ____ -1_ _ _ _ _
« «« •
•
* x-w
«
observed in
in Indian
populations
may
in the. Jlattitude
of the ■-• • National
National
University,
Canberra, p
p 89.
89.
- be .due because of any
- reversal
--------------------------------University,
Canberra,
to higher incidence of tuberculosis?3 society to the position of women and treat- ... 5 Nearly three quarters of the world’s LBW
Discussing the incidence of tuberculosis ment meted out to them, but due to the
infants arc born in Asia, two-thirds of them
among children Morley points out that . benefits of some of the public health pro
in middle south Asia, one of the most
association between protein energy mal grammes and more significantly due to
populous area of the world. India, Pakistan
nutrition and vitamin A deficiency in child effects of family planning programmes that ■
and Bangladesh together account for some
hood tuberculosis is not simple. Tuberculosis have influenced maternal mortality as well
10 million LBW infants. ‘The Incidence of
Low Birth Weight, A Critical Review of
may present as kwashiorkor or marasmus as strains on the health of women, resulting
Available Information’, World Health
following a slow deterioration of the child’s in gains in older ages rather than younger.
Statistics Quarterly, Vol 33, 1980. p 202.
: nutrition over many months. Morley says
Data from urban Maharashtra for
.World Health Organisation, Division of
. that adult physicians and those concerned medically certified causes of deaths show
Family.Heallh.
with community health place childhood that even when (he overall death rates for
6 L T Ruzicka and A K M A Choudhury
tuberculosis low in thdr priority, because it females are better, the sex ratio of deaths for
(1978), “Dcmogrphic Surveillance System
j ,is non-infectious and goes largely un- ages 15 to 19 and 20 to 24 are higher than
in Matlab, Vital Events and Migration
s recognised. In developing countries there is those for the population in those age groups.
1974** (Sec C), Cholera Research Labo
? a tendency for infection at an early age, the Also the sex ratios for deaths were higher
ratory Science, Report No 11, Dhaka. A
, nutrition of the child is poor and there is than tho.se for the population for almost all
review of detailed investigations by medical
7 a high probability of an intcrcurrent infec age groups from 0 to 4 to 40 to 44. excepspecialities enquiring into perinatal dea’*•<
tion, particularly measles and whooping
. _ ting age group 5 to 9 (and which showed a
show that in all the cases of perinatal deaths
; cough. Low tuberculin sensitivity is frequent marginal difference) for causes endocrine,
the mothers were anaemic, received inand caseation (degeneration of tissue typical nutritional and metabolic diseases. The
* adequate antenatal care and quite often had
' of tubercular lesions) cither in the primary ratios were also higher for diseases of
several health problems.
lesion or in the associated lymph glands is respiratory system for age groups from 15
See Malini Karkal, ‘Health of Mother and
more common. The majority of primary in- to 39 (Table 16). These data may be indi
Child Survival*.in “Dynamics of Popula
. fections in children over 5 years of age, cative of the effects of maternity on the
tion and Family Welfare” (1985) ed.
undergo a benign course with no detectable already weak females. Since much of the K Srinivasan and S Mukhcrji, Himalaya
Publishing House, pp 358-374.
'dinicial illness.14 Severely of TB .among female mortality takes place in younger ages,
girls may be accentuated by their under at higher ages males show relatively poor
7 C Gopalan (1985), ‘The Mother and Child
nourishment and through pregnancies, con mortality situation.
in India’, Economic bnd Political Weekly,
tributing to higher female mortality in
Vol 20, No 4, January 26, p 162.
younger ages, whereas for males it takes toll
8 World Health Organisation (1980). “Sixth
-x
Notes
in later adult life.
Report ort the World Health Situation”, part
The etiology of malnutrition, which is a
one, ‘Global Analysis’, pp 130-131.
(Paper prepared for the Workshop on Differen-.
• major health problem for the majority of
9 David Morley (1974), “Paediatric Priorities
tial Female Health Care and Mortality, jointly
• the world’s population, is not only inter
in the Developing World”, Post-graduate
sponsored by the Bangladesh Association for
woven with many facets of poverty and Maternal and Neonatal Health, and the Social
Paediatrics Series, Butterworth and Co Ltd,
underdevelopment but in many of the Science Research Council, New York, at Dhaka
London, pp 79-80.
developing countries it is accentuated by the in January 1987.)
10
Victor
C Vaughan III, R James Mckay and
discriminations against women. It is, thereRichard E Berhraman (1979), “Nelson Text
■fore, imperative that the nutritional pro
1 United Nations (1982), “Estimates and Pro
book of Paediatrics", Eleventh Edition,
jections of World Population".
blems in their solutions have basically to
W B Sanders C Philandclphla, p 14.
2 Rami Chhabra (1981). ‘India’s Sobering
depend not only on economic development
11 David Morley (1973), pp cit, p 10°.
8. No 3. p 28.
Census’. People, Vol 8,
: .bet have to assuie equitable distribution to
3 P Padmanabha (1981). The Decisive 12 S M'Sawarkar (1985), “Health Survey in
. both sexes, as much as to all the sections of
Tribal and NonTribal Village in Maharashtra
Decade: A Note on the Provisional Results
the population.
—Year 1982”. Seminar paper presented as
of
the
1981
Census
of
India
’
,Yojana,
. It must*be mentioned here that in parts
Vol XXV, No 9. p 6.
a part of training for Certificate in Popula
.-. of the developing world available data show
4 Alan D Lopez (1983), The Sex Mortality
tion Sciences, International Institute fo.
that mortality transition is underway and
Differential in Developed Countries’ in '
Population Sciences, Bombay.
Tomales have higher life expectancies as com“Mortality Ibends, Determinants and Con 13 Tim Dysor. (1981), ‘Excess Male Mortality
.■ pared to the males in these population.
sequences", ed Alan D Lopes and Lido T
in India’, Economic and Political Weekly,
However, there is reason to believe that this
Ruzicka, Miscellaneous Serie: No'S,
Vol 19, No 10. p 422:
•..change from the earlier pattern is not
Department of Demography. Australian .14 David Morlcy’(1973), op cit. pp 260-262.
•Economic and Political Weekly
August 8, 1987
1347
I
Female Infanticide in Rural South India
Sabu George 'Z
w . I<aicratnar.'. Alxri
• BI) Miller
Infanticide has been practised in al!continent but little dependable primary data exists on thissubject. Presented
here are find,ngs on female tnfanticide for a rural south Indian population. These data were collected as part
'■
-
•'
■
di^wyt°^
on child growth and survival in a 13.000 population and have been confirmed
d.rectly w.th the famdies concerned. Female infanticide is practised in only 6 of the 12 study villages affr-cting
' wdh
Cent Of!,eW:b°Jn
ported here are the demographic consequences and socialfactors associated
inra7ti<:id5 ,is a •
a‘
One
“ Of
one “
case
of infanticide of a are definite and confirmed on.the ba<k of
r ’ cent knew of SI°r 'hC
■ study of infanticide ”mong ^mans-its at’X^ne^Vwhicha^ckThfaThadf
?
at least one case in which a sickly infant had
been killed. AH the women questioned knew .
of at least one such circumstance. Due to
taries of early travellers to contemporary limiratinn^ ti« h
l° of rccall and are thus not as reliable in their
fieldwork-based studies. Long-term scholarprovide little indoht hXk*
^U' dctails 33 lhc others. His detailed research
■X interest in the subject tf"^ - provide little insight beyond this basic infor-,’.
mafirvn r\n
__ ______• C
;
f . .. . .
------------- J-.^. term residence in the area (every
stances of infanticide, the methods used, and
.sx's
^^oAjrshx^qur,i0nedDkncw ne,d-^t^^indud^onX'^i:
! information on inTantic^ espSZS?
..in^nticide h is difficult to obtain firsthand, ?>
carefully confirmed data on infanticide rav*$
toon.,
^I‘7a;'Fpa
',cni-0^wi,chcraft bdiefs
ion of south-western Bolivia,-^
pattenrof;witchcraft
beliefs “
as ideological
ore
’cZZ'TwSZ^
quantitative approach.3^
r JUS,in
justificationTor
“tiOn-for the
«« tafant
infant deaths, and soHaj
■ SSKSSSSsksss:
necessarily mean that such practices are in
*
The larger study, of which the infantidde
fact non-existent. The researchers simply
<;rlnd.,OrOn!y.!.hO$fC?Ses in whic*”he‘°'
infanticide
victim
is the
known?*
They
constitute one
conducted of
in
^*Lh-a^.r.loO,?d ,he.m- in'entionaUy or ?. of
arethe
concerned
mainly
with
reasons
forU:. data
a rural.areaTn
’th'e part,
S^thwas
Fn^mu
'■ t infanticide. Ann their nnalveie
»«f^-
.. *tv — h
_ <««.•
’•
•
.
. _
- »"w);,(;..amu(i,auu.-u ut rcscarcn was earned out
general patterns: mfantiddcs due to the con-^. in 12 villages of K V Kuppam block. North
.■ .. due
, to uncertain
7 social ~ JIZ!. ,■, Arcot Ambed k ar district, Thmil Nadu state,
• Zuah . h • VX Or.,ndlrcc‘ infanticide ticides
liades due to uncertain social or physical > South India,- for-four years beginning in
J on a reser-?:; S€ptember4986;,”
’
vation). Analysis of the marital histories of
The 12 study villages are noncontiguous,
. instance), the number of infanticides that
the mothers involved reveals the overarching scattered ,4g ' the ^peripheral, areas -of K V
3 might occur is small. For example, Sargent.
importance of marital instability.
Kuppam block; Most villagers arc Hindus, .
who has written an insightful study of witch
Larger state-level populations can’ be • and a small proportion are Christian concraft and infanticide in a west African
studied through
asf parish
verts.
- archival data,r such
-------—•
• j— ”• While
• • ••••*. villages
viuo^vq in
nt the
mi study
at uvijr area
<11 td differ
uii ici
population, learned of five cases of ir.fanyeeords or early censuses.7 But this strategy . slightly from each other in their caste cpmrecords
comticide during her field trip of 1976-77? In
is constrained by the necessity of having to ' position, the average distributidri is 56 per
addition to the problem of small numbers
infer inf»nf.r.de
infantidde r^
from the a
data
and
Infer
.........
a .the
u. lack
goundcrS( 3f per cent harijans, 11 per
of cases, the subject of infanticide bears a
of firsthand observational insights on related
cent other backward castes, and 2 per cent
> certain amount of stigma for both the
aspects of the sodety under consideration.’ • forward castes.12 Sixty per?cent.of the
population concerned and the anthroA contemporary analysis of offidal
mothers in the study are illiterate.
• pologist who studies it.
statistics on infantidde cases .in Canada, .
Socio-culturally, the studfarca is Dravi-’.^Therefore, mpst anthropological studies
m direct
®ltc™t,vc,yd‘,rca and confirm- V dian. a term’which implfes, in addition to
>. of deliberate and
direct infanticide
infanticide, in
in n«r
LrX '•
language^distinct; features of marriage,
’ ticular, rely on inferentil?Xide^cr
i '
sorts
from
in
r,wrn.„..
_'_
C
<
t
,C
h,
Sl
?
n<
:
a
’
/ccnsus-baseri
studies,
(his
intrahousehbld dynamics, female status, and ,
secondhand reports from Informants who
one is limited in Its ability< to
— provide
- ------- . other practices, .In; comparison to non
?
°rr? -hc antbropologist of infanticides .
understanding of social context and motiva-V Dravidian‘ north;' India.”. Most notably
X2aV! ,hCard about-’
such local
(ions. The author
authors?
^nf^h
removed from the; characterisations of Dravidian sodoculturai:
, tudicf of direct infanticide, numbers of
people who commuted the infanticide; they. dynamics emphasise consanguineous mar;
arC Stil1 quitc sma11.
t b«™«ng theory testing and analysis. For in-a cannot.interview them and must instead ■, riages. sometimes between uncle and niece,
,Vay°ns/on ‘be basis of; the’ first cousins or more extended kin relations
■ M
,nStudy of thc Tarahumara of
Z r11^ d ■ ava,,ab,c- such as age • within the same;village or micro-region.'4
Mexico, Mull and Mull interviewed 20
wQmcn about their knowledge
of of
infanper cent
the : sMpw The^nf'h' in"an’
daughS in^tc^s
.
■ ticide. They found that 95
J^^-rmant^edstutty'f
‘
women knew of at least g.«
ticide when thc mother had
X«ntaknw;' fa f
? Nu‘,n,.oti7"nes ?50 0350 of in^./same household,Literacy rates of women are
: had ‘too many children. 55
.. ‘ fanticide in an.area of rural Mcx’co which
higher in the south than the north,' and cur-
=• x
—v
■*<
Economic and Political Weekly
__
al
J
••
Mav 30. !<Mn
'•
l.!>
rcntly all new. teachers ip government
pic infanticide dara on which this paper
beginning of the study period (February
primary r*schools
in Tamil ,Nadu. must
be mj? na.Vfl fhfrrfnrr 4Uv
arc UUUoUxlliy
nnntiiallv cjcpc.
dCpCD1987)’ ...i____ <hcmmher bsltotmsba^
. •«»
—• . wr
women, f
.r cases
- ----- . is aand kUlcd (hc
^JOW?’.in‘hc
dablc- ----------------Tbc reported--number of
south than the north.13 The Tamil Nadu
conservative estimate. At least three other
after which the mother remarried. In the*;
state government has instituted, special
female infant deaths during the period arc
case
of the unwed mother, she tried to abort
monetary incentives for weddings of girls
likely to have been infanticides, and uncon
the pregnancy, which was unsuccessful and
above the age of 18 years,who have com
firmed female infanticide may account for
committed female infanticide when it was
pleted the 8th standard. Sex ratios (both for
thc disproportionate number of females (13
born. Maternal motivations for infanticide"
the juvenile and total populations) in recent
ouj of 21) reported as stillbirths (infanticide
may be said, therefore, to vary on the basis
decades have been near, equality at the
at birth may be misreported as ‘sevappu’—
of marital stat us (the mother’s motivations,
district level.16
blue baby syndrome—or as a stillbirth).
in turn, are likely to be influenced by her
None of this, however, should be' taken to
Tncsc deaths arc probable, but not certain
natal family and their concerns for loss of
imply that gender equality prevails in South
infanticides, and thus they are not included
status).
It is likely that, no matter what the*
India. Instead, one should icalisc that gender
in this study as infanticides. Also not indudinlant’s gender,. an unmarried or a newly
inequality exists, but is less extreme than in
cd arc female infanticides that occurred
mamed mother who becomes widowed ma>
India
’s north-west. Furthermore, it
be
remembrrrH
Vshould
—T . - bcfcrc th«
reference, period or
or be
be impelled
impelled to
to commit
commit infanticide.
infanticide. Unwed
Unwed
- ---- 1-------- j
. .
•- - tha\th‘S- typ!n“Uon ,s. subsequently.”
‘ •
• motherhood
motherhood as
as a
a motivation
motivation for
for infanticide
infanticide
drawn in very'general comparison to the
Other information gathered concerns the
has been documented for historic periods in
more patriarchal north-west and should not
village in which the infanticide occurred;
Europe19 and contemporary Canada.20
be assumed to apply to all contexts in the
caste of the household in which the.infanAnother
Another case
case of
of male
male infanticide
infanticide occurred
occurred
south where considerable variation from the
ticide occurred; age, sex and birth order of
after the study period, where the child had
general
pattern
can
be
found
within
a
region.
vJitoa.
c
.------- ----- ------- ‘ thc rcfcrcncc infant; twinship status of the
a severe congenital anomaly. Despite utilisa
ge, or even family. •
- • reference infant; and marital status of thc
tion of the necessary w«
corrective surgery and'
oath
312 °n 7^ *n^ant*c'^c y^cfe
mother. The following discussion reports on
P
051 operative
pest
operative care
care (fre
(free of cost) for over
• - • ST?T? ? a- PSU1
ProsPcclivc study Wr
the analysis of these variables in relation to
two months,
. infanticide
- .
T was committed (he?QRQdUA,ng Apn ’ I987 to September
the cases of female infanticide.< '
. day the child was taken
.
.
...
.
.
was
taken
home.
A?.
' 30, 1989. All pregnancies . in the'
the'.13.000
(13,000 .
The prevalence
..- .7.. ..
vTi/*
.The
prevalence of
of fem
female infanticide in the'
population during this period were follow-.
’
^Patterns of Infanticide ^’'- ■ study villages corresponds with a report of.
ed. The study includes a total of 773 birth : r
7 "v’™’<
gender-specific infanticide in a nearby*
outcomes involving 772 married women and •>■>' ’In
study population qf 13,000 there
population, the Kallars of Madurai-*
one unmarried woman. There were <166were
record
£7'7 a
”total
’7’: of- 773 birth outcomes
vm.il**v* .vvv.udistrict.21 The Kallars discussed are
singletons and seven sets of twins. Total live
‘--.-I,S <?Cwluch 3?8 ' . smallholding farmers and landless
™n™!T.mL7i9,orer^
births were 759 of which 378 were male and
were - male and 381 female. ‘Among the
agricultural labourers who sometimes resort *
381 female. The observed sex ratio at birth
cohort of live born infants, 56,died in the
to poisoning second-born, and subsequent,
is not significantly different (at p .- 0.05)‘ period of two and a half years (from April 1.
daughters. According to local hospital suff.
from the standard sex ratio at birth of 105
1987 10 Sepfsnbcr 30, 1989), and of these
. estimates reported in the article, a very high 7
males to 100 females, observed in large
there were'23 males and 33 females. Thus
percentage of female infants are victims of""
populations worldwide. Of the 21 stillbirths,
the male to female mortality ratio was about
infanticide:
:
eight were male and 13 female.
a very low ratio compared to worldwide
The
statistics
are
shocking.
Nearly
600
female
'
Each village had an assigned village level
statistics for - societies where gender bias
births in the Kallar group are recorded in the ’•
worker whose primary functionwas to protoward infants is not significant which inUsilampatti government hospital every year,
vidc education about child care to village • dieate slightly higher male mortality in in- .
and out of these an estimated 570 babies ’
mothers. The worker in all cases was a local
fancy ■ and • early childhood due to the
vanish with their mothers... Hospital
resident of the village and had been trained
biological higher'vulnerability of boys,
sources estimate that nearly 80 per cent of<
at RUHSA (Rural Unit for’ Health • and :
Of these deaths, 19 were confirmed infanthese vanishing babies—more than..
Social Affairs, headquartered ’in' Kavanur ‘ tiddes. In ether words, of thc total 56 deaths
450—be~ome victims of infanticide.22 • : 7'
village). Thc village worker’s normal duties ’ m°re than one-third were confirmed infanrhlc
.. .
.... ..
This assertion implies that, W,thlr
within'±
the v
^eluded keeping track of reproductive erents ' >«des.Of the 23 male deaths, there was no
..T
among the entire village population.'a task
infanticide. Among thc 33 female deaths,
*r s?0
{1!’out™p'T cen.t (4s0,>
which was accomplished through visiting ‘ thcrc
»infanticides. Thus more than '
8‘r
‘Qfa^' f
each home every 10 to 12 days. Every housT half the female.deaths in the 12 study ■
Such,.a h!gh
“enu ; ;
*n which an infant is born is visited within
villages were due to direct infanticide; in the
7
r7f !h
7"
'
two days of the birth. Such regular home
viUages in which all the infanticides ocX*s
°u‘ °f ,hc
of posstbthty
given historic
dataduring
on similarly
high rates in '•-.i
visiting generates high quality household
"t™1- infanticides constitute 72 per cent of
north-t^st
IndTa
lolh^
demographic data since no pregnancy and " f'tnaJe deaths (oduding the only case of the
and the nanem n-vsn Jt f
■CCnrUryU-J.
its b.rth outcome can be overlooked by the
female
L m reported for a region of confemale infanticide
infanticide to
to the
the unwed
unwed mother).
mother).
health worke.s.
,
'
Thus, the 'natural' death rate in this area for ' '^^T^Jarthan in wh.chvery few gnls
.Thus, the ‘natural’ death rate in this area for
The fact that an infanticide has been comfemale , infants is substantially increased
anj
a‘vc~
j • discussed
••
In terms of the possible histone roots of ‘
milled is widely
among thc village
through the practice of direct infanticide, lt_
female infanticide (direct or indirect) in
women. To the outsider, however, the cause
Using the infant deaths of thc two onc-ycar.
oj
is^mkre^
cohort (obtained by foiled iLg elch^ris- S?uth Ind,a-wc 0211 on,y ^culate concer-^
p C • U-*
3SSKS
E“j.
...
..
._
of the possibility of infanticide.
This is con-. . mortality
rate (IMR.
deaths y pc/l.000 live
firmed with thc mother and immediate
births) of 69 was obtained for thc whole ■
relatives. After about five months following
study population; if we subtract out the .
thc establishment of excellent rapport with
------------.... .....x
deaths
due to-------------------------female infanticide,, the
IMR
the study families, the■' field team had
drops to 46. Put another way, in the six
knowledge of the intent of infanticide even
villages where female infanticide is practised,
before the birth occurred in many cases. The
female infanticides constitute 9.7 per cent of
father or other family members would tell
all female birthi v
the village worker that if the current
Only female infanticide occurred during
pregnancy resulted in the birth of a female,
the study period. However one case of male
it would be killed.
infanticide had occurred just before the
"
iin v cnnu/c
nr V
V
sent study
shows 1for
K V Vunnor.i
Kuppaia block,
though the PR’s field conversations revealcd a pattern of at least several generations.
Adding some confirmation to the possibili
ty that the practice is of longstanding arc
some references to female infanticide in'
South India in the 1800s among the Kallars, ;Khonds, and Todas.26 It is possible,
therefore, that the contemporary situation
has antecedents far back in time, but at
tempts at more detailed historicalreconstruction have yet to be made.
1154
Economic and Political Weekly
May 30. 19^2
One of the most interesting and perplex
of direct infanticide, but the senior author
unlikely, however, that ‘development’ in itself ‘
ing results of this analysis is the clearly
is aware that a female twin may be more sub
would be sufficient in the short run, since
demarcated village clustering of female in
ject to neglect than a male twin, and a
it has been found that, with increased
fanticide cases. All 19 cases of female infan- ' . female infant bom after a set of twins is very
resources, people who disfavour daughters
ticide occurred in six of the study villages . likely-to be killed.
and favour sons follow a pattern of divert; which are in the same part of the block; in ; •
The age at which death occurred is • ing cven more resources. to sons than
the other six villages there were no cases.
predominantly, very young. Seventeen of the
daughters. Second, a simple biomedical ap
. Overall (all ages combined) sex ratios con- •
female infanticides occurred within seven
proach to improved infant mortality rates in
‘ firm the pervasiveness of this pattern and ' days of birth, one on the ninth day after
«vuu have
nave omy
the a.vo
area would
only a
a sman
small enecx
effect m
in
its effects on village demographics. In the' birth, and the remaining one on the 16th day. \ , changing the ‘unwanted’ status of certain
- 12-village study population, the overall sex ’£ In the entire study population, there were : daughters.
daughters. A
Aholistic
holistic approach
approach isis required
required
7 ratio (females per males) at the time of the jV a total of 18 female infant deaths during thefor changing such a complex system of,
‘. study
was
977
J.
In
the
villages
where
female
first
seven
days
after
birth,
and
17
of
these
values
about
girls
and
study was 977.5. In the villages where female first seven days after birth, and 17 of these . values about girls and women, and exteninfanticide was practised, the sex ratio was
were confirmedJ infanticides (the single
/.„6;vnon..vu- sive
' study of the underlying social dynamics
!n !hc.
/viHaSC* it was ■.. infanticide death was due to prematurity of in this micro-region (such as marriage
1018.6. Sex ratios in the under-five age group . the infant). Thus, the first week of life is ex-j payments,-marriage links among villages,
reflect this different distribution as well, with;/ tremely
J, risky 7b.
'
for .b
female infants,
but not .women’s economic opportunities, etc) would
a surplus of boys in the former villages and
because of‘natural’’’causes.
be helpful in constructing needed policies to
■ a surplus of girls iq the latter at all timesNotably, only one female infanticide (by
reduce female infanticide.
.
during four years. Also, the PR observed in-’£*. a married mother) 'involved a first-born
In
terms
of
public
health
involvement,
the
stances of deliberate female neglect more fre-/^ daughter. All the other victims had birth
;quently in. the former than the latter. ...i>2 orders greater than one, and each of these /
tl?n's{stcm. h?s
•.The villages in which female infanticide'^" families had at least one surviving female ■
sa*^rc in India,
occurs tend to be even more remote and have-y. child at the time, and usually they had two. :■ .
m
Wltu
main5^n5d
less educated populations than the villages V? This pattern corresponds to the well known v :
^h?^hroI-d rccord.s ®n
■with no ca« of femak infanticide They are P parity-apcdfic practicc\or female ■ child / ‘
linwa"t«1.dau8hter5
■ located in a hilly and more isolated part of f neglect in north-west Indilwhich seems
T"8 *?
- non-infantiddal villages, all but one have/ '- with birth erfer has also been docume^
■;bus service . ■
■ . > ,
in Tokugawa Japan."
9/ The caste composition of the villages ^thf> Alth^htheg/bdersuwolvedintai’eyf
eOPT-°f
’
the female infanticide cases differs from the
infanticide live in remote villages, they are
jurinitnavmanincantlvmitieethe
Be
• other Villages in that they m predominant- ;.: the upper social stratum of thS vill^Inf;/Pf""®^^^7n"y,TdU“Lh5 nUmber
(if s
‘bk^rTTrihed^^d8^? ttfunthatts (cob-
they do not fed econom.c pressures when -'programmes.'reservations (reserved slots) in
and tint and second cousins. In the six '
-
¥pr?!??
\
-
»'!r.aT5 '“I
. ,n
.reference to unde-niece mam ages, 6 per cent
■
female
fcmaIc infant mortality level. Indeed, if one
were
were seeking
seeking to
to explain
explain high
high rates
rates of
of infant
infant
, mortality in this region, ignoring the role of
direct female, infanticide .would entail.
: cause for- the majority of
Motherhood comments, “Such measures
onjy u- successful if better data—
; it involves a very different kitGhip dyn'a^ite
rauit's™f'This study
<the scales were balanced?54
• .,
:.vS
jn . s HOrth-west where female in-'; challenge to concepts about the benefits of ...- anticide is ; clearly - associated with .v rural socio-economic,, development
-and
development • .and ^7; Z; ■
,c=.l,hcare
rar,prog^-2^^Vxi5
‘Sir i'.i ;^^^^e£ t.. ■ .1...
®nd.cxtc>nsiyc exogamy. ' •> ?
biomedically^riented health
-^Another distinguishing feature of the mes. First, this research suggests that the^f fThesmdySl^
^. villages where female infanticide occurred 'fl; villages in which female infanticides occur->*< which this was an incidental part was funded
;:-is that they also are the villages in which all
-red are less •developed’-in terms of urban"
UNICEF^and .Thrasher. Wc would like to
the twins(were bom; over the entire study
linkages, services, ind wlucation than th^ 5
D v Mavalankar, P Visariaand L Visaria
^odoflv^nd.htdfyeanno^
ther^ug^uons.] ■,
;
" born in lhc
the other villages. 28 As in many
rr.z;
* not assume simply that bringing/develop^^VJ See'’ reviews^-a . Mildred < Dickemann,
other cultural contexts, the chances of a ' ment’ to the more remote .and less/’DemographicConsequences ’of Infanticide
. female twin dying through direct or indirect s; *dcve!oped’ villages would necessarily brine* 'j Man’, Annual Review of Ecology and
..infanticide, in either • a . male-female of--\aboutm. Immediate reduction in fcrntdein-i.^v-^^^Y^S1”5,
> female-female set; are very high.” In lhisl.-\ranticidA.this is a possibility that should
t study area none of the twins died as a result •* « held Open for ‘further investigatibn/lt jsYW? (Ithaca^NY, Cornell University Press. 1981
^Economic and .Political Weekly ■ May. 30 j 1992’ -•
’Ss&'sisra^-
Johansson. ‘Delayed Infanticide’ in Glenn
'7- w
Biology and
Hausfater and Sarah Blaffer Hrdy (eds) ’
• Fkdvf^)
• Human Social Behaviour: An AnthroInfanticide: Comparative and Evolutionary
^lu^ary^^i^ m/Mra!,Ve
Dulbury . Press.'
Approaches, (Aldine Publishing Company
10 HmrnC
i a
'
.North Scituate, Massachusetts, J979,
Hawthorne, NY, 1984, pp 463^85)
r ’
Contern.. pp 321-367; Miller (cited in note 1, ch 3)
2 Carolyn Sargent, ‘Born to Die. Witcncraft . : .
flVin,c?dcJn thc < ? *nd Alicc Clark. .’Limitations on Female
and Infanticide^ in BaribalCuhure’ ‘
’'' •Vl5.Chan«s in
Central Gujarat’. Thc
Ethnology, Vol 27, 1988,’pp .79-9^^'*:^ ; s ^Ethical ^Sr^inlI^d,an Economic and Social History
3 Thc literature on indirect infanticide (which TechnologicalAspects :■<, • .Review, Vol 20.. 1983, pp 1-25..
.
results mainly from lack of food?mcdS
• •
Tc>day' ‘Rajasthan: A -Murderous
car5 and other kinds of life-suppiSS.
H
™
1988, pp 22-24.
. /
tention to an infant) is relativclyabundant
infantSvl
lb?
?£-fcmaJC •■?z;S<€ ladia Today <ci-lcd >n note 22). ‘
and based more on firsthand i^idcncc of
of
°f
U<Jy -v26- ?dgY Thur$lon« &hnog aphic Notes in
mtrahousehold discriminat ionJ For exam-'
ehildgrowth and survival patSouthern India, Ddhi, Coj no Publications,*'
pie, on north India: BD Miller (dtxxfSe
' •*
f‘m,** “onths m the
.1975 11907].
I), on Bangladesh: Sran/D’^ouza^and” ‘ T f
S<7CT^ cases of female in-77 The mother involved in this case is known
Lincoln C Chen, 'Scx^Diff^tiaLlin ■' • 'J
h’".d°nc «W *i<h her rubsequenc.
;
PP 257-70; Lincoln C Ch^C Em^di Huq iV‘ 'J1
and Stan D’Souja, ‘Sex Bias in the Fami?v
•' r
inh
Allocation of. Food and •Health .Care in.’ ■: f ,Mdc
Rural Bangladesh’/-Poputatioh^and
%1
Development Reivew. Vol 7^'1981'pp53-7(y' , 7^-
‘nrant b?In aflcr s—'-Xj
J;2? P' vi“*S« *heee thc twins wen: born are C
remembered that, while
known throughout the area for having a
h “ud^Jan- 'nJ hi«h rale of "winship.For related resetuShri
necessary to do $0 because V.-\ : 'see A H Bittles. A Radha Rima FVvn
"
?°l^^ablVO
APP«j> Rao, ‘Consanguinity TWinnm/’
and Secondar^Scx
7
1o.. Ihi!t
Am™rrE A
uetde among .-the TSuahutaaia^ithe:
?l’'£$5V'0!k
a.™
■Irawati
30
pologtcal Perspectives on theTreatment and'L>avi<1 E Sophet,/Tbe
yuiii'isss. pp
in note Epp. KK-05)- Bettv '■
KssvKsssi
Nutrition Education’ in D SMcLarcn (edk •
gspg
■
Sn^d
DqMi11?-
.oh’ J’'<?
J’ Bj&’tt’i •'“?-»?"! -J
First Result?,.
childh^^zi^^^^ 'TT?
urrc ^
known
nextSi^’34 The
ResuIt?,. Histonif-or
HistoryChildhood
nownfto have had their next
t nc Lancet, ‘Girls Matter Too,PP1991 £•<
Quarterly, Vol 1, 1973, pp’ 98-116:’Josiah
bom after.September3Q,‘1989J?‘. < pp813.
813.
_____ -
• • 7*
Economic and Political Weekly
Regina Schulte, ‘Infanticide /in' Rural' <^’8ther l5™aJc lnfanI
may have been >
Bavaria in the Nineteenth Century tn Hans^.'^CJ^ ?nJa ,1.,C1?CS <ncx confirmed) or in-T
Available from
Mcdick and David. Warren Sabean (eds) -; C ■S:<hrca.,nfan(,adcs brought about through V.
r Ramakrishna News Agencies
Interest and Emotion: Essayson
' "•■nutr‘[ion-'31 or medical neglect.
Of Family ahd Kinship, Cambridge Uni^r-’- . I? Man3 W Picr^ Infanticide (W W Norton/j). <■ 5—4—670 Neo Mysore Cafe Lane
sity 1-rcss. New York; 1988, pp 77-102- and
^n<? ComPany. New York. 1978).
. . S?. • Nampally Starion Road
Linda Gail A mgo. ‘Female Infanticide and ''‘ ?? P^y
W’!^n (cilcd in no(c 5). ''
/
. H YDERABAD - 500 <j01.
*
SoaaJ Stratification in Republican China- ''' 2x Ind,°Toda>\ Bo«n to Die*, June 15, 1986. 1
Nov Perspectives frora thc Buck Survey of
PiPh 1
'
. Lise Mc’dia Agency
■" >
Farm Families’, naoer
\
22 Ibid, p 13.
•• • • . ;
36. A/10 First Floor
V.csteni Conference on the Association for
23 t?"1' ,.‘akra4’’ ‘Fffeci of Infanticide on ScxHalls Road. Egmorc,
Asian Studies, California State University
Ratio in an Indian IVjpulaiion’, Zeitschrft •
Madras • MX) OOH, Tkmil NadG.
Long Beach, 1985. . : >■ ■
Y’
' fur Morphologic and Anthropologic
■■
ri • ’
Vo1 6;L ,97°. PP 214-30; K B Pakrasi and-r
1156
Economic and Political Weekly
May 30, 1992
Sex Ratio, Son Preference and Violence
in India
A Research Note
Philip Oldenburg
Explanations for the different comparative values of sons and daughters have focused on economic and cultural
factors includmg the type of agriculture, kinship systems, customs concerning the linkages between parents and
offspring after marriage and socio-economic activity. Are differences in these factors sufficient to explain the
Bermuda Triangle for girls' of west-central UP and the surrounding districts as revealed by the sex ratio map
of the 1991 Census?
- This article examines the hypothesis that families in west-central UP want (or need) more sons than families
because additional sons enhance their capacity to literally defend themselves or to exercise their power
by investigating the correlation of sex ratio with violence in the state.
■
•
,
i
¥
'
■
I
. . of standard family planning methods once < Das Gupta (1987: 95] has evidence that ex,THE map of sex ratios published in the’; J"
family (skewed in favour of sons) cess mortality. among later birth order
report of lho provisional figures of the 1991
, n ^ched. With this ideal family,' “ daughters increases with the education of the
Census (Nanda, 1991:53] vividly shows that ?
P»<nning strategy that says “stop motherland .through a better ability to
“barring Jaisaimer in Rajasthan and Jind f u^V’ng children after two sons have been,v manipulate, both their fertility and their
in Haryana, all the other districts of 5?ni|?n<! *ncglect unt0 dcath’ fourth-bom children's;mortality, educated women are
J Haryana. Uttar Pradesh, Madhya Pradesh
^ould mean that a normal sex better'equipped than;others to achieve the
and Rajasthan with a low sex ratio below 850
per cent of the family size and sex composition that they
. form a continuous belt” [Nanda. 1991:58] f*m,‘lcs nave had five children,'with three . desired Paradoxically, worsening sex ratios
Twenty-two of the 30 districts in this belt are Of thcm dau*ht«s Oible 1).
. may be in part the consequence ot a suc^.• in UP. These include Delhi and Kanpur
This •strategy’ and this model obviously cwsfuljamily planning programme which
?. (urban). Aside from the other major citydo not accurately represent the actual situa- . encourages parents to have ’two or three1 ;
\ districts there are II other below 850 sex <*on: on the one hand, many families plan children but which cannot control the techniratio districts, largely scattered. Most of the thcir families to stop at two or three, no matby which they achieve that result and
districts in the next lowest category (850-899 tCT what the sex of the children; on the other ' satisfyThar-preference for sons,
females per 1,000 males) form a ring around hand, there are families who apply Meath
>ls6 been a great deal of study
;.r these very low sex-ratio districts, extending by neglect’ before the fourth-bom daughter ‘ on the.question of why Indian parents want
:: into Bihar, with an ’island’ of higher sex (Das Gupta, 1990]. . •
. . . ‘ 7 more sons than daughters, and, overall, there
ratio districts in eastern UP. And the further
r
}
>s now some agreement on the major exsouth the eye travels the more favourabley d°“
r?’u,fc ,ldol>- V planatory factors, which, though intertwin,‘ to-females districts one finds, with Kerala ,,<’n of ,nf,n,,ad« or
Plan- •: - can be
• put under
• ■ the
- two broad headings
• having the best picture by far The data ag8 m^ures; all rt needs is the conscious of economic and cultural (see Miller, 1989:
gregated to the state level is consistent with °r
'reiitrnent of
.
....
’this picture. It is perhaps mot too fanciful • L nran!V*ho
on average, given a.; ... T
able
!: V’ariation
in Live axavx
Births
per
.......
—. —
— "■ "
■ no «
FamiLV? by Factor of Need for Sons ‘For
to see this map as showing a pit with slop lesser share of food and less medical care
Physical Force to Dominate in
ing sides, a pit whose lower depths represent than their brothers. The data are overwhelm
i
j V' Family and Village’
a very large number of ‘missing’ women, ing and the techniques are well-studied, even
sometimes
\(Figure 1). With the exception of Kerala, • if the parental responsibility
t •, •' is
--------------where the sex ratio continues to rise, (he * ?
5)bv,ous,y-11IS not necessary
'
• Sons are
Sons are
for every family to practise this strategy to
.Important Unimportant
4.4
Andhra.
3.7
. Pradesh
. .. .
7a
;
“
■
4.2
3.4
2.5
The pattern of the sex ratio map is vir- • bickward’
'^"oraot: people: Monica
sex ratio map is vir- •
tually the same as (hat of the map of “excess
Table I: PaatEcftONS of Family Size ANb Sex Ratio
female mortality over male mortality by age
(1st child, 2nd child. 3rd child. 4th child. .5th child (cumulative))
; (UNICEF, • 1991: 25]:. the largest dif- ..
Females
ferences, of 20 per cent or more, are to be ■parcn!s Son ' - Dn
Parents
,
Son
found in western. UP and in north Bihar ’ Parcn,s 50,1
Daughter ■ Son,'.
Son ..... ;
f f
Per 9 Males
Raju and Premi [1992: 911], citing the argu- ‘
Daughter
Son j";
.ment of Coale (1991 J, seem to agree that “it
----------------------is the sex differences in mortality rates which
Son
. Daughter . Daughter
females per
affect the balance of the sexes”, but do not Pawn’s L Daughter
Son
Daughter
18 males
provide an estimate of how much effect that Pawn’s Daughter
Daughter
Son
Sori
= 778
has. There is no doubt that the sex ratio pic
ture reflects deliberate choices by parents, Parents Son
Daughter
Daughter
Daughter
Son
30 females
ranging from the rare murder of female in- Parents Daughter
Son
Daughter
Daughter
Sori
per 30
Parents
Daughter
Daughter
Son
.thc **fatal nc8,cci of. female
Daughter
Son
males
chddren iM.ller, 1989: 193) Io the adoption Parents Daughter
Daughter
Daughter
Son
Son
■
=, 1000
Economic and Political Weekly
December 5-12, 1992
2657
''
Figure I: Sex Ratio, 1991
• • ' <? ' .rr-
outmigration of males clearly increases thc& '
sex ratio in the Konkan ^districts of M
' ‘:
f r Maharashlra and in the hill districts of UP^
•
'
*
and in-migration of males may well ’lift* the“^
Punjab sex ratio, for example, Kundu and C
- .v
■ . - Sahu (1991: 2342] argue that ”at thc.statc or<?
district level, mirgation is the single most im-?portant factor explaining the temporal and
cross-sectional variation in the sex ratio” ?*
The slowing down of male outmigration) X-N’ ‘/X/1 ' ’fr°m Bihar and castcrn UP,possibly due to
\Vi <hc dangcr•'of working in"terrorist-tom’
X x’>->’ Punjab, would explain at least some of the s
X^X ?'i:' ?•~<,ec,inin«
ratio, and the.marginal im-j^
\wS\2yi’•♦.-‘iPfoyement.in sex ratio in the central-west<
pr
*9 ‘
i
nets might be due to increased male out-’f*
•; migration? However; it seems unlikely that ;jX/\ ■■'’"‘even if all east UP migrants were to return *
\
■^<’V--5hori1^-thal sex ratio would decline'to the' :
■ rfo c vyp.
; of yw-antral UP.* .
*.
\ i?- while one might be prepared to agree with >
: .z. Kundu and Sahu [1991: 2341] that “increas^
_
. Idiscrimination against women, resulting a
5 •-•> ^tl-^T^vv^inithetrhigher moftalit/alsofstands^dis/S
.■
•<' i
r^an explanation ofrlndiaY^ex^
,*- _ ,
J\
^^Ime.'discrimination^gainst^girl?!
»<» M^g^ jclearly h the best explanation foAthe ab-^
'Rggggdfe] <
■'* ’ ‘•*
level of sex ratio, and therefore should^
N.A?
••
*' v
■
.*
'Va>
• > >. .^cirrJr-
>'■;«;'Almon
<hewhfch
whichTainilieftl^
a all disnets^c^^?-7?^,i^do rt’c“««y in>"the
’fainiliell
■.
y-•
. x {- 7a
<; /-/. ,/.,«> j .‘4
•
■ I MU no1
the area “with'lhe very highestg
.. ............ '. u----------------- 11 "
““dpauern . ^^^^urenHe (male to fanale] sex ratiorreponed^
c, .
Source: Nanda IGO.,
1991: 53
■ 192-93 for a ‘
fxWixS iMaj°ri^’’y
.sld5ls .H; :>
'*Wc mu
sl return <hen
y of
of di
disuictt
:y r e-Wc
must
then to
to social?cconqmi(Aa<
socialcconotnid^
I’v <• •,
r'UtdV/ S*lUraI’ *??• J ^yld ®dd, political faciorsM.
'A
A.. Z- .-!* -i ;:€«r-A“t ' in the 1872 Census" iMiller. 1981:611. Is there*
"^i^n^ ^something special about theoiste compost 7
....
"
. • -; .t J!;-;-'
r
women in agricultural work; the expectations
economic capacity Millt^ 119811-a^s leave it to othws to explore—preferably witlvj
of financial support.not just iridoid age; and - strongly that a ‘nonhem mewir k
' ■ $U?^y
and complementary microthe burden of^providing/dJmaXl
cultural, and economic^
•
boys to find brides even in a.low* sex’ratio'
locale). There is also at least one maior’oersonar factor: the literal phZSS
seems scccptical of that exolanation' d
h^rc ,n
1
l,kc lo propose -i.
Are differrnr^ .*n,h ^p,anatl°"- ? • as one factor among many, but one that has' "
to^n^^StS" "JT" °f,Cn
^‘’on of J
tional support that comes from co-residing
of west central UP and the furroundina
fa^l^V S°nS tO “Ph°d’ W1.lhk*,oknce-
with children. These do not by an/ineans’.- ‘downward sloping- districts
exhaust the factors that come into play? nor ' sex ratio map of the 199i Ccnqi^ Varmnc '
r
• .
ua,ng jmsfolk).
‘
%
ranking in importance,'
factors have been proposed and also rejected:
2“27 b: cap . un^r5"umcrati?n of females in the census, Miller [1989: 192-93] remarks that
lured by Monica Das Gupta’s assertion that ' Jinrl
and fl
differences
irf
lf
/-**•<- erf
»-** • • — —• UT —f* >•••
•
*
1
ratio at birth [Miller,
important roles in local power struggles over ’ '
for India as-a whole, the evidence suggests
1981: t>8-69; but see ^Basu, 1991: 16-17]. ' land boundaries and'rights'to irrigation
that son preference is primarily culturally
Kundu and Sahu [1991] discount the former' -water” Mahadevan and* Jayashree 1l989:
determined, and scarcity of resources may
factor, but argue for the latter as explaining
nsf. surveying 6,500 respondent? in UP?at most accentuate the effects of sex bias"
some of the most recent decline, adding “the
Andhra
Pradesh, and Kerala, report’that’.''
[1987; as quoted with approval, in Vlassoff
relative decline in sex ra’io at birth in recent
parents who thought sons were important5 /1990: 19].
years could be due to amniocentesis” [ibid:
"for physical force to dominate in the family'
Explanations for the different 'com
2342].* It is hard to beliesc that the west
and village” had more children born than
parative values of sons and daughters along
central districts of UP differ from others in
those who thought it unimportant. Only 41.2 .
these dimensions include the type of
access to amniocentesis facilities.
per cent of the respondents in UP (compared
agriculture, kinship systems, customs conMigration patterns arc a genuine factor:
to 48.4 in Andhra and 68.6 in Kerala) said
•
--- .--------------
—
ail
I
2658
Economic and Political Weekly
•/
December 5-12, 1992
' L? ‘
\
*
Figure 2: Trends in Sex Ratio. 1901-91
(India/states having population 25 millions and above)
This is not to deny the importance—
probably the far greater importance in India
overall—of other ’currencies’ of power that
1100exist: economic leverage (landlord-tenant
r•
lender-borrower, etc); socio-cultural control
1080(exclusion from religious or community af
fairs. e g); influence with the political
establishment; and the ability of some to get
their way by persuasion or by calling into
\f‘- 1040play the respect others have for them.
Still, in some places, the strong arms of
!'i; i 1020adult sons and other male relatives, in par
^'LNadj
ticular. are needed for the exercise of day§ 1000- Andhra Pradesh
to-day power. To be sure, they are needed for
_Bjhar_^
the protection of the family as much as for
forcing compliance on others. Patrons also
980;
would value as many males as possible in
•?S•*
••
•
their dependent clients. This perhaps gives
an image of occasions of pitched battles
V
940'(which do occur), but violence as the cur
rency of power exchange is also a feature 6f
day-to-day intimidation, in which the show •
,<5 ‘920••
(wilh thc display of large or vicious
Rajasth;
followers), the reputation for the use of
violence, and the well-timed slap or hard
; shove i>lay an important role.,
. *•••*'*
- . . ’ ’.
-There is no systematic data on this, as far
X
as I am aware?4.! therefore have chosen to
use the murder case rate as a fairly crude
proxy variable for the incidence of violent
j^.MOL
behaviour that would tend to enhance the
---------1
.
1
1901
1911
1921
1931
1941
1951
1971
^ue of dependent males, especially sons.
1961
198! 7: 1991
Census Year
The assumption is, obviously, that the more
•.
•
v
. .. :•»<*£.'*
----- Uio
son O1 this sort of violence
severe and
widespread
• ’•11 is
- ,o «d in
,.<h« «* were Rvalue- for this purpose
important etoption. In onochfflinjte.^
“iJr-."0,e. •••
likel)
i murder.
she describes the endemic vioienS^S
'nber'°r murt?crsvai
■side^iyf
rom
y^r
t0
f I|h|ther Percentages of
environment in the early 1970s:
from year to year ut a give distria’
tespondents.
These results* were not
a uffniGrant r«™
ru
’
501 have calculated the mean mu
, /broken down by district or region within /
component ofthepoGtaal pro- * rate in 1980.-1981 and 1982 in the
Aites, so there is no way of knowing how
y the physical force factor varied within UP*$
of inferiors by superion was an everyday
2,000 respondents.
k
eurrence Father beat their roru. husbands :
S ,md “ S.Umman?^ in
•
c 1
,060‘
K- S.
•
tt960'
■
<
■’•
T*
*.
r-
te
«?■'" • r ./
• •
'
j
* P?n'a^ purely indicative test of my
'4 hypothesis that families in west central UP
£ V"1 .(°r r^d) morc 1001 lhan families
^elsewhere’2 because additional sons
.. enhance , their capacity literally to defend
■ memje/vcs or to exercise their power. I have
r- begun to investigate the correlation of sex
ratio with murder case rate in UP. Murder
•
rate (the statistic used is murder cases
registered per million population * as
t calculated from data in the Crime in India
y t’hl’? hM ^Cn .Ch0SCn as an ind‘cator of
■ the degree of violence in a district.” Ceteris
,SLmorc I,kc,y that people will be
murdered when and where political power
< r d J"(cr*Pcrsoni| disputes are literally
1 ^.out with fists, knives, and guns rathcJ
•; ™n w«h la* suits, shouting matches, or
n er, and other less violent means.
We have tended to overlook. I think, the
degree of variation in chronic violent
/
/,0JLr;. ^ar8ucritc Robinson’s book
^,ltcs: The Law of the Fishes
—
r. (Robinson. 1986). a study of a i
village in'
i^Medak district of Andhra Pradesh,
is a
a'very
i i<
_
< Ecdnomic and Political Weekly
their wives. mothervin-Uw their daughters- yi
impressive number of
in-law. teachers their pupils; vilhge d^
!n
high sex ratio/low
beat any offending youth of the same or'd,StnC1S and thc ,ow scx •
lower caste. Usually a wooden rod was used. ‘ C!8? :,F\urdcr casc rate districts, but
although any nearby object could serve the
,mPressive is how the districts
purpose. Beatings tended to be severe, but
t.^5?,carViP ,cach Quadrant largely arc
usually there were understood limits. For 're8iqh. .all but one .
.ecample, the watchman of (a village] was . .S?a?.,?,5?)^bf 7hc east and centre-east' "
viIU<c rcvcnuc officer imali 'hc u^Per left quadrant and all
u Wl^n on,o°kers determined that the
but three (Unnao; Bulandshahar. Bijnor) of
watchman s wnst was broken, the beating
<he west and centre-west districts in the lower
was stopped.; ? With the exception of the
nght quadrant Three of the districts in the
. Z
'Thisiis clearly a result that cries out for
more detailed testing. Since crime data are
collected byfthana’ and since sex ratio could
easily'be. calculated for the same unit, the
One wonders how much of this rort of
number of cases could be easily expanded
Hon v»rles from vitl^n’o’vHl^diMria io
’nd t^l7'8hlhdi|S<:°VCr sub-districI Paf'enZ
beaten by him (usually privately, sometimes
Pyhlidy). .. Only the patels were considered
to have the righ( (o administer bearings to
other-adults (Robinson. 1988: 40-42);
district, and state to state With??he«st dif
ferenccs in history, caste structure, and struc
-------------anu
tures of economic
and pondcai
political contml
control, ir
it
would be strange if ’naked’ physical force
-----------------nakcd
*orce
was not used to greatly differing degrees.’
---
•»*- V
Ji I ivt
----------------- variables
' "
and test for the influence
of other
of likely relevance, such as caste composiChangc ovcr ‘‘rne—the relationship, if
.a.ny« ^ween the deteriorating law and orddr
- situatIpn.Vrefiectcd in an increase in the
rd« case rate in the data up to 1989. and
December 5-12. 1992
2659
■
,ra,iornc'd .10 »' ■ i"vestigated. wc need to study whether in these
areas of violence daughters are sccrl as an
even greater burden than elsewhere because
of the greater risk of rape and abduction.17
A careful micro-study of the hypothesised
linkage between son preference and violence
as the currency of power has to be made
before we come to any firm conclusions on
this topic.
More important, wc will need to explore
why there is this variation in UP.’.These
districts are indeed notorious for the use of
physical force, because, in one explanation,
they are areas, unlike the ex-zamindari areas
of east UP, where the economic leverage ‘big
men’ exercise is comparatively wcak,kand
resort to force
‘traditional
’.18. Another
suggesdon
is .ha.is.he
Ganges-Jamuna
d'oab
M-on. »nd V«na Til.a, Oldenburg (who is
daughter";
also responsible in large measure for sensitis
3 Eg. UNICEF 1991: 64-70, which presents
ing me to issues of gender and from whose ■ copious data, but carefully uses the passive
descriptions of her own work I may well have
tense and thus helps hide the fact of paren
been pointed in the direction of this note).
tal responsibility. Monica Das Gupta makes
Some of the data and ideas herein are con
the agency very clear, but even she softens
nected to research I am presently engaged in
under a grant from the American Institute of
Table 4: Sex Ratio and Murder Case Rate.
Indian Studies.)
Larger States of India. 1981
I The trend line of a combined Punjab and
Haryana moves steadily upward from a very
Below Median .Above Median
• low level to meet, roughly, the line of UlMurder Case
Murder Case
- tar Pradesh, descending, in 1991. Kerala’s
'
Rate
increased rftale outmigration to the Middle
-------East may account for the recent increase in
Above
Kerala
Bihar its sex ratio.
median Orissa
Gujarat
2 The ’ideal family’ in Thane < istrict consists
sex ratio Andhra Pradesh
of ■ two sons and 1.3 daughters (with
J.;..-;- - Karnataka • A
.
women’s and men’s views virtually the same)
Tamil Nadu
was an area where pastoralists had to adopt
and Rao
Ran (1989:147)
Iiosohti report: "Irrespective of
r
and
Below ’* • West
West Bengal
Bengal Madhvn
settled agriculture where the ecology was uneconomy class a minimum of two sons still median Rajasthan^
Maharashtr.
median Rajasthan
. favourable, thus making the struggle for land
Bihar" (thar cm phasisj witha4! wo sons one .
< Nonh-wX^ ‘
more intense than ini areas where more jntensive agriculture was possible.” ..
daughter* family, the modal preference.
-.
sratA*
It „
is.also
just possible—I note with trepidation—that
Even in «>u.h Indi, (in .hi, cay: in ■
.
Norths
castes or communities that predominate in
Karnataka): “Most families want
_„l a
------------- ----------------- ----------------------certain places have culiural traditiorwjthai...
. n^un^of^^
No/er.a Nonh-wesi states - (weighted)com--'
valorise violence (i e» a ‘macho’ or ‘martial*- • . the danger of losing one, but also because
. bination of Delhi. Haryana.
peoples explanation). All this requires' ■; two are believed to be the minimum size of
Himachal Pradesh. Jammu and .
a male team within the famiIy._Tbere are
further research...
; • Kashmir, and Puryab.^ : *«....
also strong emotions about having c*
■ b Nortlwaa nua - (weighted) comI cannot resist, however, speculating on* ■
’ .daughter”, (Caldwell, Reddy and CaldvrelL
the relevance of the UP data, such as lhev
•./.... bi nation of Arunachal Pradesh,'
are, for .he larger picture of the ya^S
£££
.S
Ve
^STt
•
1988:77). According to Srinivasan and ,
Manipur, Meghalaya, .
7
' 'Tripura?*** NaiaUnd*
and
and.J
4’
We can see rrainnal nati * k V<
‘ conducted in 1980-81 indicates that for a
1981 Murder case rate * mean of
whirh mi
n
P
hC^.aS ”
;
P^cencage of couples the best com-,
1980-82 murder case rates, ct t
which might well require the addition of. . binadon of children was two sons and one
Source: As in Tkble 2. .
s
j
endemic violence* to other north-south dif-' " i"
ference explanations, and other'Tsocial/ /
r"
* Table 3: Sex Ratio and Murdea Case Rate in 1981. UP Districts
economic, and cultural variables. The decline; ~
——---------------------- -——
•i
in sex ratio over lime might also be ilt * /■. '
Bdow the Median Murder Case
Above the Median Murder Case ‘f 5
luminated by this factor (compare Figure2).? . •
■'
Rate (Range 3-46)
Rate (Range 47-109)
J 2?'.
Although the murder case rate in the ’12 •
---------------------largest cities is in fact exactly that of the'
•
"7 HiD Darias
East
country as a whole (33 per million), it may" ■ A. 7^7 Vttarkashi
Azamgarh
be that as the country has become,morel-,
Jaunpur
,; Bulandshahar < * ’ ’ ‘
«'
' .*• :
Ballia
urbanised, and as old systems of social and? "
G^.bwal
, Unmn <
Ghazipur
economic control have broken down£ir£^.
Srfl*al u • •
. creasing violence might<help explain;tS^l.
Deoria
Gorakhpur
‘ Allahabad
fiecline in sex ratio. Might-the precipitous! ’ ?
la ./™KXa
Basti
Han^irpur ’
^c.,Inc ‘.n
Bihar
rat*0, for example,-^* r
Central
Varanasi
.be in an indicator of a deteriorating law.andri ? i
Faizabad
Mirzapur
order situation in that state? But this limb'
- •
of speculation will not, I fear, bear any more
’’
Gonda
weight. Low and declining sex ratios in India ‘ \ 4 ??
SuItanpwZ
■z
are the results of the preference for a fami-7
‘
Pratapgarh
Bijnor
ly with more sons than daughters in it, and ' — ‘
'
)
they will not improve until that preference,'
. ; ’f'... A .. Naini Tai
Bahraich
West
Centre-west •
is altered. Although a less violent and conBdow tbc’; • Dehra Dun
Meerut
Rampur -*
frontational social system may contribute to7 Median Jex’&7 Saharanpur
Moradauad
Bareilly ’
that change, I suspect that the impact would' ratio
7 Mazaffarnagar
Jhansi
Budaun
Pilibhit
be minor, and more fundamental economic i (801-858),744’X Aligarh
Shahjahanpur
’ ^d social changes that enhance the-4
' '
* •' Etah
Farukhabad
autonomy and power of women are ■
Mathura
Kheri
necessary.
a
Agra
Silapur
Mainpuri
Hardoi
. Notes
. Etawah
Lucknow
Kanpur
Barabanki
(I wish to express my gratitude to those who :
Jalaun
read this note in draft, and to absolve them
•A.«
from blame for the-errors that-remain: Rana
Behal, Mecra Chattcriec, Govind Kelkar, Ritu
2660
Noir. 1
lion; sex ratio is females per 1000 males.
Economic and Political Weekly
r>— bcr ' 12. 1992
the language in her concluding statement:
tnan women migrating out from particular
of the persons arrested for murder were
“All these considerations result in strong
districts of UP. See Premi (1989) for a
women (Crime in India, 1981). Computing
and mutually reinforcing incentives for
discussion of th- problem.
an index of ‘violent crime* requires access
parents to successfully rear sons rather than
9 Das Gupta (1987) and Miller (1989: 196-204)
to a level of data that is unpublished: e g.
daughters” (1987:96). Cassen (1979:114)
outline (he way it is done in rural Punjab,
brawls in which less than five persons arc
notes that excess female mortality is “due
but (here r.Jght be other methods elsewhere;
involved arc not registered as a ’riot* but
fundamentally...to worse malnutrition
they both have excellent discussions of the
under another section of the Indian Penal
among young females than males and the
various arguments for the nazsorts for doing
Code, which is not separately listed in the
risks of maternity" without assigning blame
so. to which we now turn.
Crime In India data.
for the former cause. For the clearest
10 See Hasan (1989) fora very thorough sum
14 There may be some indicative evidence in
presentation of the early research and a
» mary of the historical (including post
works of literature, or in the work of
powerful argument on the topic, see Miller,
independence polijics) and land-relation
anthropologists other than Robinson. In my
1981; in her later study (1989:193) she says V 'factors that distinguish western from eastern .?.>
own fieldwork in Meerut. district, as in a
“(There is today] the practice of indirect'
,UP, and regions within that broad division.
number; of incidents I myself have
female infanticide through the fatal neglect
1 !\Mahadevan and Jayashree( 1989:126); other’ experienced, this aspect of on-the-ground
of female children.’’ In Punjab, “being
‘dimensions* scored as follows: ^economicpolitics has certainly been evident.
female significantly increases the probability
support during old age* (98.5 per cent of - 15 While the data is from 1981, the 54 districts
of dying at all stages of childhood after the '
respondents), ‘salvation to the parents by ;
are those of 1972, which was chosen because
.; first month of life, i c, at the ages when
. doing ritual formalities* (923 p^r cent),
I tried to discover whether the rate of inbehavioural and environmental factors play
. ‘provide tradtioGal links (Encage)* (943 per
crease of murder case rate displayed the
\. a larger role in child survival" (Das Gupta,
cent), •meeting family obligations’ 191.1 per'. <;,<• same pattern that the single-point coni
1990:499]. Most of these studies focus on
cent), ‘inherit family property’ (903 per
parison showed. The difference is that I
questions within this broad picture:
cent), ‘becoming adult member and getting'^? - treat Meerut and Ghaziabad districts as one
e g, what effect birth-order has, or level of
sutus* (100 per cent), and •physical support 7 .
(Meenit), and Kanpur Rural and Kanpur
y education, or socio-economic status, or
; - and staying with parents* (993 per cent).
. Cuy as ooe (Kanpur). I have also disregard■
mothers* economic and social autonomy, .
Thcre were other factors that were exprcss-,,. b^^ed the. murder cases registered with the
i.«c.
cd by less than 50 per cent of respondents,
4 See, e g, Mahadevan and Jayashree, 1989:...'>
, . . ,
, .. ..
.
’• :±thOy/nd
^rnr^W>fr- -
including
•receiving
dowry* given by only jj 16 I... would
--------------- --------------o
wvumi like
«■«; to
iu thank
nMiiK Nandita
ix&noiut Aras
zvras for
lor
134
cem)
for me. and
13.4
cent “
in up
HP <♦»*
MAX
—
<1 ta&rScdatattrthe statistics
---------
.
way police in UP.
applies mainly to the propertied groups of
UP from the ocher tub-states was ‘inherit K. 17 I am grateful to Meera Chatterjee for makthe north just as the ‘southern* model fits
family property* (UP: 903 per cent; U.: ing this point
most accurately the propertied groups in the
Andhra: 42.7 per cent; Kerala: 234 per 4 18 Interview with a senior UP police officer,
south.. Unpropertied groups in both the
cent).
.
.
. December 1991. New Delhi. •
north and south have a characteristic sex
12 Coupled with the oft-noted disincentives to
19 Interview with Ashok Thapar, December
ratio pattern more like one another’s than
:r have more than ooe daughter (with they^>399l?»d5!it^"like the propertied groups of their respecresult that “a sizeable proportion of young
tn..'
2*^live regions (Miller, 1981:27).
...♦ < women [In the Punjab area gtidird] dkl not
r» r
6 Miller (1981: 168-69] anticipates that
want to have even one daughter", and ‘
^e^erence8
»
|
i
j
|
possibility. However. Alaka Malwade Basu
(1991:17] argues: “If one assumed
gmenxuJy) that . quarter of the ;
almost none wanted a second dauriner*
(Da* Gupta, I987.-94) this wouldg^^e
• company low J'ratkn. Ajain, I nuie
.
AlakaMalwade (1991) ‘Why the Female
OP"'i°n
[abortions performed in I98445| followed
no guesses as to the contribution of w'^a.
8.’’(M yJ
I
fe sex-determination tests, they would account ’
beJriicXsus •daughterdiabiUtv* ' hiW*; ' Bose'.^d'.«s«sted by Vinod Kumar Singh
kj.. ^cx-oetenninauon tests, they would account
0.75 million of the estimated 31
13 This is true of UP. according to
^d.M!thl‘ ^Jhlkar> lI991l- Demographic
for about 0-75
j
million female shortfall in the country^
UP police official (interview New rMM ‘
India: 1991 Census State and
|| .7
December 1991), though probably ^f
r
B R
.7 Kundu
Kundu and
and Sahu
Sahu [19913324]
(199112324] nou that there
J
seems
every Indian stateTMiXra^^more
PubLshing Corporation, Delhi.
seems to
to be
be a
a connection
connection of
of decline in
... ta
f l r<tio to <back*anincss’—that “sex ratio has .
reliably reported than other crimes (inter- P H Rcddy andTat
J;-' declined in districts that arc poor in terms
view with a senior police official who has
Causes of Demographic
|
of their urban, industrial, agricultural ■■
dealt with the statistics. New Delhi; vl/
ExP^r}me'!taI Ke^arch in South
development**, but what the causal connec
December 1991; sec also Saksena, 1986:151).
"w™\Thc• University of Wisconsin Press,
J tion might be remains unsaid and is not ob•Murder cases’ do not equal ‘murders': if
Madison.
vious. Rajan, Mishra, and Navaneetham
more than one person is murdered in a
Cassen, RH [1979], India: Population
[ 1991 ] sharply criticise Kundu and Sahu and
.
U-! uut-m,grant males am of,™
”?s,e
only one case is registered; ? < Economy. Society. Macmillan. Delhi.
;
propose that out-migrant males are often
I
double<ounted. Raju and Premi 119921 in
lh<^eforeused theawkward but mote r Co*^ * J IWl). 'Excess Female Mortality and
' •'— -----------------[1992], in
accurate term ‘murder case rale*. According
‘
the Balance of the Sexes in the Population:
'•
turn, support Kundu and Sahu (1992) and
to NS Saksena, one of the most know- • • * An Ertimatezof the Number of ‘Missing
C
provide a critique of Rajan, Mishra, and
ledgeable analysts of police matten, this
Females’,. Population and Development
• ’ Navaneetham.
factor plus under reporting of murders,
Revlcvf 17, 3: 517-523.
. ,
’ 8 Najma Khan (1981:492) finds that in the 12
of more than three
Crime in India. .1970-1989. Bureau of Police
.] • villages of six cast UP districts the rate of ,
l&ksX
mkUrdCTJ*50 registered
Re^arch and Development.' Ministry of
■
male out-migration is only 7.1 per cent (with .*•
loaksena, 1992]; i e. the murder rate is three
Home Affairs, Government of India New
a high of 16.7 per cent in two villages of
limes the murder case rate It is probable, '
Delhi..
; . Azamgarh district). Calculafng the
bowevrr.that thisratio docsnot vary much ' Das Gupta;’ Monica [1987], 'Selective
| ; magnitude of (temporary and permanent)
lS?m^h^
T,nOrh“i'Ch^ed
Female Children in
• male outmigration from census data is difpolice off ^T''m'r^"'cnACW w',h sCTuor .-'.
Rural ’ Punjab. India', Population_ and
Bcult, and I have found no studies that pro■ MuX? 7’
’ Fcbnllry IW2)- D^opmem/^iew 13. 1 (March); ri- 100.
’' vide a simple answer to the question of the
males^Xn^S^^^^
---------- — — —Iiu l£IJ * VllIUC Ol 1 F —
, ‘impact on sex ratio of more men
.' 1.. . males against males: in 19S1 only 2 per cent'
- ■
,• ., won and the Determinants of Child Mortality
*■
Economic and Political Weekly
December 5-12, 1992
2661
075 7 6
in Rural Punjab, India’, Imputation Studies. ‘ Singh,' S N and M K Prcmi. P S Bhatia, and
• l^uiriiion in India, Nutrition Foundation of
44:489-505.
:<>Ashish Ikwc (eds). (I989|.Imn• India, New Delhi. ,■
Hz'an, 7ova
llQRQl
*
Pnarr
ar>H
__
*
>•
.
.
Zoy^ [1989] ’Power and Mobilisation:
sition in India.^ (2 volumes). B R Publishing ’ UNICEF [1.991 ], Children and Women in Indi'cc
Patterns
Change in tUttar
erns of Resilience and Ch.nnor
Itiar . -i. Corporation^Delhi, s ’ <•■
................................... ..
~*
A Situation Analysis, UNICEF India Of-'Pradesh Politics’ in Francine Frankel and
Srinivasan. K and Tara Kanitkar (1989).
Gee, New Delhi.
MSA Rao (cds), Dominance and Slate
’Demographic Consequences of Low Sums
Vlassoff, Carol (1990]. ‘The Value of Sons in
Power in Modern India: Decline of a Social
of Women in Indian Society’ in C Copa»an
an Indian Village: How Widows'See It*.
Order, Volume 1, Oxford University Press,
and Suninder Kaur (cds), ki'omen and
Population Studies, 44: 5-20.
Delhi.
Jeejcebhoy, Shircen J; and Sumati Kulkarni
[1989], ’Demand for Children , and
Reproductive Motivation: Empirical Obser
vations from Rural Maharashtra’-in S N
Singh ct al. volume 2, 107-12l./if»-;V‘i
Khan, M E and Sandhya. Rao .[1989], ‘Do
Welfare Services Reach Couples Below (he
Poverty Line?, A Case Stud] of,Family
•.i&i • ; ■
Welfare Programmes in Bihar’ .nSN Singh '
ct al, volume 2, 143-152. . • Vl t .
Khan, Najma [1981], ’Pattern.of Male Out. Migration from Eastern Uttar Pradesh: A
Study of Twelve Villages’ in R B Mandal
(cd). Frontiers in Migration Analysis, Con
cept, New Delhi.
Kundu, Amitabh and Mahesh K Sahu [1991 J.
‘Variation in Sex Ratio: Development Im
plications’, Economic and Poiitical Weekly,
October 12: 2341-42.
■. It ...
’7
Mahadevan, K and R Jayashree [1989], 'Value
of Children and Differential?Fertility
Behaviour in Kerala, Andhra Pradesh and
. Uttar Pradesh’ in S N Singh ctjal, 1989,
volume 2. 123-31.
'
associale of Prinls In<^-New Delhi Multiple Export Awards Winner). A';
Miller, Barbara D [1981], The Endangered Sex:
Neglect of Female Children in Rural North ■
"Prints House". 11 Darya Ganj. New Delhi-110002 »*
India, Cornell University Press,-Ithaca. ;
Phone: 3273347 Fax : 91-11-3275542
- 1 —[1989], *Son Preference, the Household and
a Public Health Programme in India’ in Maithreyi Krishnaraj and Kanina Chanana__
(cds), Gender and the Household Domain: . ——
.. Social uad Cuiluras Shnmzonsi Sage; New
Delhi...
•
•• ■'
Nanda, Amulya Ratna (Registrar General and ‘
Census Commissioner) [1991], Cennu of
India 1991, Series 1 (India), Piapa^ l of 1991,^
.ProvisionalPopulation Totals,'QwmuncnLi^.
!
1______
of India, New Delhi.
Prcmi, M K [1989],' ‘Pattern of Internal Migra- H?
lion in‘India: Some New Dimensions’ in
S N Singh ct al. 277-92. '
‘ '
Rajan. S Irudaya, U S Mishra?, and K
Navan cel ham [1991], ‘Decline in Sex Ratio: . ..Cultural Domination, Hegemony and Schooling in India
An Alternative Explanation?' Economic
and Politir'fl Weekly,, December 21: ..
TIMOTHY J SCRASE
2963-64.
•• ' T;:.
ATTENTION AUTHORS !
Leading Exporters &• Publishers :
. invite Manuscripts in all branches of
■^Humanities & Social Sciences : Arts. History,^.Culture, Philosophy. Religion. Tourism.. Travel. ■
^Himalayan Studies. Environment. Current Affairs .
^’and Women Studies:
<
g^-D. PUBLICATIONS PVT. LTD.
~•-
’—::---------------------------------------------------------------------------------------------------------------------- ,.<
■
Neio
. -4
A■ <*.|-I Publishers of the1
I Professional ■
I Social Sciences
«
from SAGE
ilMAGE, IDEOLOGY AND INEQUALITY
■
Raju, Saraswati and Mahendra K Prtmi [1992],
‘Decline in Sex Ratio: Alternative Explana
tion Re-Examined’, Economic and Political
Weekly. April 25: 911-12.
‘ ■
Robinson, Marguerite S [1988], Local Politics:
The Law of the Fishes, Development
Through Political Change: in Medak
District, Andhra Pradesh (South India), Ox
ford University Press, Delhi.*'
Saksena, N S [1986], Law and Order in India.
Abhinav. New Delhi.
f ’ ■
—[1992], ’Hold Police Chiefs Accountable for Excesses’, Indian Express (April 6). p 9.
Singh. K P [1989], ’Child Survival. Health and
Nutrition: Impact of Green Revolution’ in
S N Singh el al, volume 2, 191-99.
2662
•
•
••
Dr Sc rase explores social and cultural reproduction theories in the context-of
schooling in India. He addresses the question of educational inequality from the
perspective of critical cultural analysis and draws attention to the complex
interplay between education, society and polity.
.
- The central focus of this study is an examination of the content of a selection of
• English language primers and readers prescribed for use in West Bengal schools.
• /r .. The author analyses these books mainly in terms of their portrayal of social
relations and cultural formation and argyes that these textbooks reproduce
. established class-defined patterns of hierarchy, culture and social order.
180 pages • 220 x 140 mm • Rs 200 (cloth) • 1993
SAGE
PUBLICATIONS INDIA PRIVATE LIMITED
Post Box 4215, New Delhi 110 048
Economic and Political Weekly
J
December 5-12, 1992
DISCUSSION
i
Sex Ratio and Violence
marriage harassment of the daughter-inlaw and her parents). In the context of the
•
rapid spread of consumerism and rising
inflation of the 80s, the problem of dowry
. has aggravated further, and its subsequent
Amp Mitra
impact on parent’s sex bias cannot be ig
■' ft". "
nored. In the face*of rising cost of living
THE grand debate on gender ratio in the
urban areas the concept
of social securithe average family size tends to decline but
--- -----------------------:r?' . Indian context gains momentum further
ty associated with ‘son preference’ would
in that case the small family norm, given
as Philip Oldenburg (1992) in his recent
vary substantially from such a connota- the sex bias of the parents, tends to get
4’
piece attaches a new dimension to the
tion given to the term. It would, therefore,
implemented at the cost of female foetus
issue. To explain the ‘Bermuda Triangle
be quite misleading to suggest that “as the :?.'.[Mitra^j991]r Given the level of social
for Girls’ of west-central UP and the surcountry has become more urbanised, and
backwardness the problem' of dowry is
YV;
rounding ‘downward sloping’ districts he
as old systems of social and economic
more severe among those belonging to the
'.
proposes to include the factor, namely, the control have broken down, t increasing ^ middle and higher income brackets than
SS*. **?crccPl,on of a necd for sons lo uphold, violence might help explain the decline m ‘t\ their counterparts ’in thexbottom ^ize
Spurious Results
■
• A * *.**’•
with violence, a family’s power vis-a-vis sex ratio”. Moreover, it would be a wild .classes^Similarly it would be erronenuc
neighbours. In favour of his basic ' guess to treat all the murders
---------- --- across
-------------- t°‘gno.rclhe positive association between
r ^7 - hypothesis (i e, “families in west-central
regions as being of uniform type. Organis- affluency and crimes (like murders) If the
; ®; UP want (or need) more sons than ed violence and the murders committed average level of living in the central-west
A&tk families elsewhere because additional sons
by the mafia groups functioning in the UP can be found to be higher than that
..
enhance their capacity literally to defend
cities are phenomenal. In that case, can-i? in other1 parts;:bf.the state, given the levH
”),”),the
“W themselves
^.emselves or to exercise their power
power
the one associate the ‘son preference* , with
°f social backwardness, the discrimination
•■Ci??'1
__ -j_____ ? V
Vi -^vempinca 1 evidence he ales is the negative' crimes? If Maharashtra falls into4 the
against ©Wcanjdndeed, be iclated to the
u^rnrrriannn
correlation (-0.72)
t.(\
between
— the district
.h.
category of low sex-ratio and high murder • problems^bf dowry
/
■ specific sex
.^spcCTiic
sex ratio
ratio and
ana the
tne rate
rate of
ot murders.
murders,
case rate (Table 4, Oldenburg) it would be ' &
\ X
But such a correlation between the
^variables mentioned above can also
WScmcrgc
l^c d’stricts with low murder
rotes, and thus less disputes and violence
,*5
35 l^c author would have us believe, report
. ^’outmigration ofmales larger than that in
!«..«! In U SU» .nd ns <he 2CgS “SS’""'!,';?''
himself has noted, the murder rates> in
. iXS
"T*
j
these large cities are considerably high:•!iIn
slaSof
I,fact, if regions
.
- with low sex r.uo ha4high
, S?
’ Par'
if regions wh low sex r^io have high
crime
rates
and
vice
versa
it
would
not
,
T
35lhe districts with high-murder rales. The i j.-g- .
------------------- —- —
difficult to
to perceive
perceive both
both crimes
crimes and
and sexjQ
sex^ft
?
2
outmigration of males in the districts with / difficult
A ^J'Jcss violence would turn the sex ratio in - rati
ral’°0 “
35 lhc outcome
outcome ofsome
of^some endemic:
endemic.
/
sodoecoijOmlc conditions.Kundu,^Mnitabh
conditions. Kundu,-Ainitabh and
favour of the women. But such a pdssibiliProblem of the soao-ecoijiOm5c
andM
M K Sahu (1991 j: ‘Varia:*My
’ty has been ruled out in Oldenburg’s .
*na^itiontomoren)ateoutmigration^
; Hidn?^^jRatibrDevelopmentTmplica-kty
•
*
1
!—
^ discussion as he writes “it seems unlikely
in easl^ UP and mory female outmigraEconomic and Poliiical ’Weekly.
i
7
lion in western UP for reasons such as■
cvcn
casl UP migrants were to .
marriage, other factors which may be sug-®
1111 in Gender Ratio?’
home, the sex ratio would decline
?! ^>;return
gested in explaining the inter-district varia^; oidenbu^P^
d •
Sk'to the level of west-central UP”.. But „
if
for
reasons
such
as
marriage
have
!
,ons
,n
I
’
a
‘
,
°
in
UP
are
to
be
perceived
•
j
.
Preference*;
and
.
Violence
in
India- °A
‘women f
“"■J to
• a greater extent 'from the
m tCrmS of thc Pract*^.ofz.doYn; and pro- /-f Research;'Note/Economic and Political
outmignfted
blems associated with it (including postWeekly,' December 5-12.
c^mml-west UP as compared to elsewhere
,
in the
that vail
can aou
also IA.
be aa ica^un
reason of
of
..........
* *'state,
‘*"*v’ *»•«*
...
%®»Attention ;■■■■
.
low sex-ratio in central-west UP.
FRONTIER t
61, Mott Lane, Calcutta-13 ‘1
Oldenburg also discounts the view of
j^WScholars/Readers '>:>■■■■ •<
SUBSCRIPTION RATES* S®
l^lyKundu and Sahu (1991) which relates the
fcv'Vourje’quirernent of Rare, Out-of
'^’decline in sex ratio at birth to rising am-’
' Print ana New Books and government
^isu^niocentesis. In his words, “It is hard o
;■
AsocMter>.
pubiicdtions oh India, write to or visit:M* ' 5
(Annual) .(Annual)!^
rgbelieve that the west-central districts of
Ordinary
.
■,.1,.;
tOS^feiervic^!- ■
!?UP differ from others in access to amnioIndia A Nepal
' Rs 100
■ . ? R$ 250
, ?»lg;.Sadar Bazar.. ‘
'
'* Half yearly - Rs 50
•
yVi • 'ccntcs’s facilities? He would rather explain
Bangladesh
Rs 150
. ./
. . .
Gurgaon,-.
Haryana
122
001
•
-•4
imer-regional differences in sex ratio
Overseas air mail
'
'*
x'
./St^^'neiia
USA. Europe. .
- .?>- ,n terms of “discrimination against
Australia.
- -xrTel: oh STD 01272-20588
’r,
g’ris’—the discrimination that stems from
Canada. Japan,
,
From Delhi 83-20588
i
l-----^c want
for .Iivit
more avili
sons III
in inc
the
-------- - (°r nccd) ivri
and Hong Kong ‘US <50
. US S 70
Asia
and
Africa
US
$
45
,
t
US
S
55
•
facc of violence. But protection from
We are also interested in purchasing '■
Overseas surface mail
.4' ', »•
single ? books/individual collections/
v . violence such as disputes with neighbours
All countnes
US 5 25
US $
’whole libraries of dntique/rare cooks.
(leading to murders) is just one single
Payment should be made to frontier by
.MahatmdGandhi,
Money Order/Bank Draft/Cheque (overseas ^7
‘;’$b/.component of the huge spectrum of social
I'-' Jawaharlal Nehru,.Indira Gandfii, Rajiv •
security the parents expect to derive from ;
having more sons. Particularly in the
towards bank collection charges. -
5;
,
| B.
____._____
I
'•1
"I
I
# ** • *
I*
i
I fcS;
■
Economic and Political Weekly
January 2-9. I<M"
a ■■
*
"4.
’
•
:~g
■
67
, J.i .
On the Demography of the 1991 Census
Tim Dyson
coupltd w IIh a higher rale of dll-India gnnvih during the 1961-71 decade.
Faced with the choice between i'S
<
one’s mind and proving that there is
'synoonneed
th^
to do so. almost everyone gets busy on the
-
proof.
■
- John Kenneth Galbraith
PUBLICATION of the provisional resuhsdf
the 1991 Census of India (Nanda 19911 has
occasionedtheusualveryconsiderabledegree
ol interest and debate. Two key indications
have received particular attention-first, the
fact that the rate of population growth seems
U. have declined, and"second.^he indZZ
mcrcaseinpopulationmasculinitycompared
tothcCensusof 1981. which has been widely
: -Ibken to refieet a worsening of^ relat^e
7 survival chances of female!
XT'
decade [Mazumdar 19911
‘
■
Inlhiscontcxuhcpresentpaperapproaches
r .:
'
^i^’nuowardsthewholccensusexer.ise whether (hero has been anv decline
in the
?Ur n"c ?r P°P"'ation growth.
assessment in this respect for 1991 [see also
This.issue is best explored at the state
Bose 1991:42). In our estimation, therefore, level. But before we do this, some preliminary
the true population as of March 1991 could remarks arc appropriate regarding some of
• well have been liigherthan X75 million. And
c^^sTT
1. Still, some 20
its current nn/i
size (allowing
underenumeration
k
F for L^
n^Ctr2 ycarsJatcr» lhe question of the level of
enumeration
subsequent growth) could
well be veryand
close
C<f d enumcral,on in lhc 1971 Census remains
well be very
close «» 90() million.
For
examnT
rm/
"
F°r cx,rcmc'y3 germane. It was Visaria [1971]
example,
if
the
enumeration
was
just
4
who
cent deficient 'andThl'T Ua| J“s<4per
Pcr •*
ho fin/suggested
P"?1 suggested, from a comparison of
cent
deficient
and mthe
has
*
subscauentlv
crown
7 npopulation
PU a"On £
aS h(HI!ie,lstin
gara)ccnsusP<,Pulationcstimatcs.
itsT^Z Vgrown
mZat|£A
' ‘hCU971
“ bCCn 3 re,a,iVCl*
subsequently
2.11 ZtM
per cent, then
its size(896
as of
would
be «96 P™’r.enumeration in tenns of its coverage. .
million
0 =March
8439 1992
7"! 04
x 1 Ml
million
x 1.04 x 1.021).
.. .. If
If 'this
was Ihc
(he case
case. and there is certainly
T-X (896.0
, . = 843.9 1
02
h's .was
t Tabic 2 summarises ‘the
T assumptions,- ijndirctt^vidcncc
indinecVievidcncc in support-.
- then
ulen clearly
c.cany
S
hemsches drawn TZ
up wnh close reference '• "''^"-'^^tcrccnsal
the nlljridia; intercensal growth rate for .-/'
to SRS estimates
behind the most
recent
1971-81 thTablc
I may
have
to amended
be amended. .
~ W,-.,nd,hc
reccn
' ■ '^^HnTable
I may
have
to be
XJ^CoZ
3"on'^,on$™*fchy'hc i"
all-lndiapopulanonpmjectionsmadebythc
in sucha
such iway
wy that it modifies the indicated
' ty?""
.«-iS!SEX5
SZSZ'SS'""''”""'
STS SSt Xi'i.mi!’
“'■’■Tawi-i.
■ ■:.'.DM.t f. L
s .. nr.w.™
.k-
u__;_ . ------------. .
nasic views wntch
conditioned our interpretation
of wearcue
the I9KI
19811
• provisional results (Dyson l?_*”.
.
both that it is unlikely that India’s rate of
^. population growth has materially declined
i* >;and thal lhcrc has probably not been any
* | deterioration in C
experienced by females.
total ,n 1991 of 843.9 million ts extremely. 'coveragehbmpletcncss is also very relevant
‘r'?T.TTP7J^<'?“rC'Vh,'C’ClC,r1y- ,o'h£^ucs,ionof changes in the population
•h'scouldpaitlybefonuitoos.itnevcnhcless sex ratio.Thus it has been arguedThat when
cT',S C“ns,"u'C ’*ronF support for the . census coverage completeness deteriorates .'
continuabonoftcntlityandmortaliiydcclinc . in India, it does so particularly for females
hXZ ’'.Z
,h°"San<I
^’^’odonThe level of 'enulhcratinn ’
Z
7VzT£
z”'
and 197! - produced '
u
hfe expectation surpnsmgly masculine population sex ratios. *
is around 60 years and the level of total Also. in the subsequent censusesbf 1951 and
fertility is a shade above four live births per 1981 it may be significant that the population
/'
All-India Ri-sults
woman. In other words, tn general terms the sex’ ratio fell:.indeed, these arc (he only
. Table I summarises the main results from
coincidence of the enumerated and projected •’censuses this century when
vruvu mu
va iratio has
the »sex
. the last five censuses, including the provi- population figures almost certainly docs declined compared/lo\hTprevious
~
‘
’
» census
f
sional figures for 1991. As can be seen, the
-vuiugc
annual rate
rate of
ot population
population growth
growth ’
an?1 '
'r h h
\
average annual
fertility decline during the 1980s.
followed by those olhe main sXsTT’
inthef1naXctiXweWe“Xi^uT
principal conclusions.
conclusions. ..
principal
/
loabmxll per cent durine
are“"comenlious. For bing.': Fof if.thd relative level of census
The population sex ratiofm/n in 1991 has
example, today there is fairly general . coveragc'diddelerioratefurtherin 1991 then.
10
i’S^ °f/o'^se/a
, .. decline
.-------- in.n- the
theTn.ercensa,
intercensal
to its highest ever recorded figure of 1.077.
occurring ano mat it is probably happening populateion growth rate becomes still less ■
Apropos the provisionaT 1981 Census
in virtually all the main states (see. for plausible?\Vith this as backgroundJ we now
«count. we have argued previously that the example. Srinivasan 19881. However, it is
the state-level results.1/-,.’
. - ;r
. Indian census probably underenumcratcs the as well to remember the recency of this • examine
■' •
v&W-• ?
•;
> ’ • ’ -; ’
real population size by several (c g. 4 or 5) realisation.Thus(heprovisional f981 Census
rrrtain|la^C POl.n,S |Dy5On 1^1] And • results were scrutinised primarily apropos
Table 4 presents (he 1991 state-level
■
arTe m’nTZwsuTZTh iSmUC.h ■^JSSUC1 ofI.whc"''r
"’ere was a 'popuht’onTotals-and related intercensal
enmnemrion Jheck surveysview <5The '
UndCrWay’hC
can be seen, seven states
increased mobility size and residential • T hl*' I T lmpon-,n' .,ssuc ans,n8 frnm oow have populations over 50 million: that
complexity of thecountry's nonulation - Tor, p^h n0', T’'"
nxluc,ions in • «f Uttar Pradesh is around 140 million .Also
Plus its probable heightened deuichment and from h’-^w7nrV Y T"0 W''“:h T**
'Krsislcncc orespecially high
t c IZH provisional census results ts rates of population growth’in the large poor
'
Economic and Political Weekly
December 17-24. IW4
/i
*
/■ • j,-.
.
WH - ItO
0 1j /6
-C*
!
I
M/Fe ‘
1,0X0.
■
r ••
I’oHa.ATioN.Six Ratio. 1921-91
",
■ ^1? n
1.070-
LOGO-
;;
■. 1 h?,5a
;•
.• ../'-Vi;
i
, LtMO-
reverse el feet. Againlhercisa him (no more)
that there mav have been some further slight
dctcrioralior in census coverage in 1991.
7 bird, and rclatcdly. comparing Tables 4
■■
- na mU!;h W7kcr rcl;"ionsl,'P
(r - 0.45) between the indicated change in
intcrcensal growth rales according to the
:•
&
■j
<
e
'
o
census and lhe SRS. Indecu. for all states,
die census growth rate either indicates a
greater percentage decline in growth rate
. between lhe l970sand 19X0scompared with
5 ' . £ .y ^K’*SRS-<,r,CJ‘s<>l’:iperucniage rise. In other
•
~ *'
' i.74Census jear
•' ■?■•••'• I
<• • .
l,‘un
'O'd-dccadal CRNI’s indicated by
^'R-''- lhe differences ate particulatly
great in Bihar.Guiarat.
Bihar. Gujarat. Harvann
Haryana. KiirnuivL-i
Karnataka.
.;Kcialaand Ihinjab. Ikrliapstlie most puzzling
discrcpaiK’ics’arc lor Gujarat. Haryana and
Karnataka, stales which tend to be net
receivers of migrants. However, the main
point we arc making is that whereas in the
1970s slate intcrcensal growth rates were
‘ generally higher than lhe rates of natural
increase indicated by lhe SRS. lhe 1991
Census results have produced precisely the
■ WV;
l*,e SRS is generally much less
sup|x»nive of a declining rate of population
■ -!g*>w<h <mx ;.!«> Table 3).
6 summarises the position at the
of ^Ute-level. Both Hk: SRS and census arc in
run1
1
^agreement that die rate of populatiu
an growth ■
''n-uuml in KS '’rri,lCS Ofdcdin,,,<ss in Kcra,a- Punjab-Tamil Nadu
...neural increase ^nj^^^verages for &ind ta shade less certainly) Gujarat Note
V
'inAr’™5"1?ri'r'
Mh/r
eraerge
each -:;llu:aco>rdingloTahlc4.wulnliccxccption
>■ Punjab. .he« four Male, alwcxpcJaced
P°inL'
1961-71 and 1971-KI. We can Unis be fairly
sure (hut (heir growth rates, ate indeed
declining - in some, quite sharply. Con
versely. the states of Andhra Pradesh.
Madhya Pradesh. Maharashtra and West
Bengal each experienced tt slightly higher
ratenl intcrcensal population growth during
19X1-91 than during 1971-XI. and a higher
ratenl natural incivase in I he 19X1K compared
to the 1970s according to (he SRS (Table 6).
Thus the available evidence cCtiamlv docs
not support any decline in the rate ol growth
of these lour stales - although it is certainly
possible that improvements in SRS coverage ;
account lor some ol the suggested increases
5'^^’ (perhaps especially apropos.
Maharashtra) andmigration '
1has also
'
. during
.
occurred
the l9K0s to Ixuh ■
Maharashtra and West Bengal.
.
'
7 his leaves us w ith six major states which
according to the 1991 provisional census ’
results have each experienced a decline in
the rate of intcrcensal population growth, yet
according to the SRS have each experienced *
a rise in the rate of natural increase. In brief -!
the issue is: how are we to apportion Bihar, fHaryana. Karnataka. Orissa. Rajasthan and ‘
Uttar Pradesh between the cells in Table 67 ■
Clearly, (here arc several possible. •
resolutions for the indicated differences in ?
the ease ol lhc.sc six states. They encompass
changes in migration, census coverage level 7
and the performance of the SRS. over they
penod 1971-91. However, threeexplanations
seem particularly relevant. First, improvc-
gr,"“h
First, the consistency and quality of statelevel demographic data available lor the
\
’__________ -__________ »: Au-Iswa CtiNsux Statistics. 1951-91
I 9X()s seems to be.belter than (hat available
lor the 1970s. In. this context-Figure 2
Ccasus Year
■
compares state-level Crude Rates of Natural
IncrmsA tCPMO
?•<-«».• .
Increase (CRNI) fronvihc;SRS with .1951 -
corresponding inicrccnsal rates ofpopulation
1961
........... . for both.lhe1.;i.97DXi-and .........
..growthI9XI-9I • „J971
decades. lt is clearthat thcjrclakionship
9KI ;
thcjrclakicmship has 7?
' 19x1
ttw«
•••!« 9 «. .I.l'.- • at_
bcct.mcnmchiighier-ihcsimplc'correlatKm
;-IWI ’•
coelhcients lor.
for. the-'J97(K^and
the: l97.0sWi9X0s
i.)
coefficients,
T9K0- arc
respectively
0.5-\Smd.
bJ9Q^M
regarding dK’ 19xf
9*1"" e^'?fe
Morcover.
!,reovcr- • ,’r
regarding the 19X1
7X1-91
-91 comparison
compariMm it is clear ; G '
that some
of llC
the d
di‘ ".^"^
ffcrcnccsm
'Figure 2 are .V .
cxnh’hf °
“l^gurc^are
plitablcmierm^^
^plicaWeiniernisofnii^aii^
;, > ’ '
Maharashtra and West Bengal lend to be net
n<«onto whilc^Kcrala’jiss a
. • receivers
receivers ’of people.
exporter. 77iis improved Stale-level correspondencc bctwcen.jhc SRS^itd census
measures ol population growth probably
P;,nly rellects lhe better operation of the SRS
Total Population
’
Average Annual
Awrage Annual
Rate
of Population
Ratenl
Population
Gruulh (per cent)
Males
((XXM
36I.OXX
439.235
54KJM)
6X3 329
X43.93I
L25
1.96
2.20
2.22
1X5.547
326.324
2X3.9X7
353.257
437J9K
?
" 406.333
I (157
1.063 \
1.075
1.070 C-j
1.077 <
been adjusted to allow for the fact (hat (be reference date for the 1971 census was April |
rather than March I-The first growth rate given relates to the decade I94I-5|P
’■
r: Nanda (1991).
•
* *
I.
.
7 TAKU 2: Sl'MMA*v,r AssuMrTMKS ami Statistics ikom Rlvishi (19X9) All-Isdia Population ’
.
PR<llf(TK»SX <■ EXTFXT CoMMinih ON PtrfUt.ATlON PtCOJI (TIONS
/ '
; Indices
Date/Penod
19X1
19X6-91
mT
5K.I
59.1
10.8
30.9
140.9
20.1
54.7
. Crodebirth raiv<CBRi
General knility talc (GFKj
Crmle rare-nt n>uof irurvasc (CKNIi
Population ItXXki
A^raZ'tddlillle m<’Tl^-iyxipopulaiioo
w h ! ,
uv '< ,C "Ucrccn5:11 irr;HC-lowth lor Ubl-91 arc marginally lower
175.541
212.911
264.173
330.072
Population
Sex
Ratto (m/f)
The 19X1 census inial given above is based on the recently revised estimate for Assa.n.
cnumeraieu tn the 1981 census,
which was not enumerated tn the 19X1 census.
; '
. (ii) The average annual cues of population growth given above for 1961-71 and 1971 Kt hnvr'
also
-r’ • • - I '
< Lilccxpcclation-iiuk:
n v the 19X
K 1 -9
, ml
inicrccnsal. en>wd>
growth rales . .. lyeaS-k-naKmay have been less ulleeied ,by changes . . Cnak-death ra.dCDRn
in census coverage level .than those lor
1971-XI (which, of course, may be biased
by the probable poor census count of 1971).
Second, for all stales except Maharashtra.
2.11
• Females
((XXk)
|991
?
6X5.1X5
Z“
1991-96
60.6
617
9.4 1
27.5
122.7
18.1
X43.596
figure given above is that taken to be the final figure lor the 19XI census.
Tbebreespv^auonspvvnlor I9X!
are actually tbeSRS rates for 19X0.
I'rm, t/^l S^tutc: Nanda (I w t».
(konomic and Political Weekly
December 17-24. 1994
FiGiiRi: 3: Ciiam.i-s in Sex Ratio I9XI-9I
Pi.orn i> Against Deviaiion i Rom Projecti-.i>
1991 Poh'i.ation Toiai.s. Maiok Stah s
A in/l 19X1-91
Figcre 2 Comparison oi Intirct.nsai Growth Ratis andSRS Crmm- Ratisk Natvrai
|s<m asi . M mor Staii s
.
i\
cd
5
IVXI-VI
I97I-KI
-4(1
• Bihar
§
I
cd
>J.
1
tl
ly
1
1
>es
on
nb
in
ve
co
ie*‘
.ar.
i nd
6
de
ass
vv
the
m
vu
IO
ivn
7
X
S
I
7
■n.
r
’ I
4
5
I
MIS
ir.«
2 5
• Rj|.'ih.>n
(.iuprji •
• K.irnjiuku
liiur Prail.-xli •
2 I1.91.7-
Maiiafu'hlrj •
Bllui •
• Mudnyj Prjik>b
M.hjJavhttJ •
• I’uiiub
W II. njul •
/
2 I
U IknjJjl • * Amllir.i O-Mk-'h
Arvlhru 1‘i.aJcKh • >
luniil
/
KinutaLa**
17
• Kcr.ilj
13
■
GupGit •
• < >i 1\XJ
-4
.
I - Hal l*»a»k->li •
l uniil Nadu *
llatyuiu ■
Karitaluka ■
-3
-I
________ * Mahatj-htra
.1_ 2
l*n«viM'itul/pn>Kvi<.*il «k-vi;iii<>n
.‘-R'A
(per cent*
3
W Rrneal •
• Krrala
• IjJlllI S'jklu
ri
. Onwa • •.
• (Tvjirul
I'unph
1.5-1
ri
-2o
• Rih^i
Sjdu y
13-
-30
!
• UllM Pi*Jr4i
i<>|
1.9-
•<
. |() = • AiKlhra I’l'ack-sh
>
ri
CRNkPcr Ceni)
CRN It Per Cent)
ch
6
7
I l.ll V.111.1
1 -20
is most persuasive for Haryana which also Obviously, these are rough and ieady
menls in SRS coverage over the period may
assumptions and calculations. But they
experienced an iniercensal growth rale
have produced a spurious rise in the CRN!
suggest that the decade 19X1-91 mux not
decline between 1961-71 and 1971-XI
between the 1970s and 19XOs. Looking at
have witnessed any reduction in the rate of
(Table 4).and I'orwhich thcindicaicd increase
the SRS CBRs in Table 5 this explanation
population growth. This conclusion will be
inCRNI between the !97(Kand IVXOsfrom
seems partly applicable in the cases of Bihar
strengthened if (as is to be expected) the final ’
(especially) and perhaps Karnataka. Orissa the SRS is comparatively modest (Table 5).
1991 Census total is a lew million higher
Also, levels ol contraceptive use in Haryana
and Rajasthan. Second, a particularly poor
than the provisional total and if any evidence
are comparatively high | Govern inent of I nd ia
level of census enumeration in 1971 may
arises of a further deterioration in census'
I9X7|. Noteltxi from Table 4 that Orissa also
have biased in an upwards direction the
coverage level in 1991.
1971-XI interccnsal growth rate and hence experienced a decline in its intcrcensal growth
The question of a possible deterioration
rate between 1961-71. 1971-XI and 19X1-91.
produced a spurious growth rate decline
in census coverage level in 1991 redirects
And we have already remarked that im
compared with I9XI-9I. Third, the relative
provements in SRS coverage may help to our attention to the indicated increase in the
level ol census coverage in 1991 may have
population’s masculinity. In this context
explain some of the indicated nsc in CRNI
deteriorated to produce the same el feet.
in Orissa. According ly.it seems very probable Table7 showslhallheincrea.se in masculinity
Obviously, these explanations are not
(hat Orissa’s rate of growth is alsodcdining.
in 1991.compared with 19X1. was widespread
mutually cxclusi ve: they could all be relevant
To sum up. then, it seems fairly certain
(as indeed it was in 1971 compared to 1961).
in the case of some states.
that rates of population growth have declined Hlic sex ratio (m/f) rose in Gujarat. Haryana,
This said, it seems unlikely that the rale
Karnataka. Madhya Pradesh. Maharashtra.
in Kerala.Tamil Nadu. Punjab and Gujarat.
ofintercensal population growth has actually
And the same conclusion seems likely for Orissa. Rajasthan. ’Jamil Nadu and Uttar
declined in Uttar Pradesh. Rajasthan and
Haryana. Orissa and (perhaps a little less Pradesh.The increase in 1991 wasespecially
Bihar. /Analysis of both the 1971 and 19X1
pronounced in Bihar. Indeed.Table 7 shows
Census results strongly suggested that the clearly) Karnataka. On (he other hand, the
generally more populous slates of Andhra that if we simply exclude this one stale then
1971 Census enumeration level was
tbeall-lndia rise in masculinity between 19X1
Pradesh. Madhya Pradesh. Maharashtra and
particularly deficient in all three stales
and 1991 is reduced from seven to three
West Bengal have probably experienced
|Visaria 1971. Dyson 19X11. Note too from
points. Further, i I we exclude Bihar from the
some minor increase in growth rate. And to
Table 4 that Uttar Pradesh. Rajasthan and
last four censuses l hen t he al 1-1 ndia sex rat ios
these states - almost certainly in our view
Bihar together experienced the greatest
- must be added Bihar. Rajasthan and UP. 'arc notably more consistent and any upward
proportional rises in rates of intcrcensal
trend is much less marked. It may also be
Some simple calculations can illustrate
growth between 1961-71 and 1971-XI.
highly significant that the presumed
the implications of such an analysis for
Moreover the indicated intercensai growth
conclusions regarding trends in the all-lndia deterioration in census coverage level in
rate percentage declines between 1971-XI
1971 was also accompanied by an especially
interccnsal rale of population growth. First,
and 19XI -91 lor UP and Bihar are extremely
great increase in population masculinity in
for the six states in Tabic 6 lor which the
small and it would not need much to overturn
Bihar (Table 7). Coupled with the previously
census indicates a decrease in growth rate
them (Table 4). In the case of Bihar in
while the SRS indicates an increase, assume
particular, it may well be that improvements
that the true growth rate during 1971-XI was
in SRS coverage explain part of the indicated
Tabu-. 3: E<timati:I> Crcdi-: Biriii. Di-atii and
the average of the CRNI from the SRS (loo
rise in CRNI (Table 5). But. on the other
. Naiurai. Iscri asi- Raii?» i kom Samii.i
low) and the indicated interccnsal growth
hand, the 1991 Census results lor Bihar
Registration System tSRSi.
rate (t<M» high). Second, back-project the
Ai i.-India. I97I-K9
produce an intcrcensal growth rate that is
19X1 populations for these six states to 1971
particularly low compared with the CRNI
Penod/Ycar
on the basis of these average growth rates.
from the SRS: again, this could also be
I9K9
I9X1-X9
19KK
1971-KO
thought of as indicative of a relatively poor This exercise alone suggests that the 1971 •
30.5
32.7K
34.59
31.5
CBR
Census total should be raised by just over
census enumeration level in Bihar in 1991.
10.2
1 1.54
I 1.0
14.72
C DR
five million. The implied 1961-71 all-lndia
Returning to Table 6. in the cases of
21.24
20.3
19.K7
20 5
CRNI
intcrcensal growth rate is raised to 2.29 per
I laryana. Karnataka and Orissa it seems more
likely-though not certain-that the indicated cent per annum and the rate I or 1971-XI falls
Noirx: (i) Figures gixeu for penods arc incai
to 2.13 percent. The latter figure is virtually
growth rate declines from the 1991
values i»1 corresponding annual rales
the same as the implied growth rale of 2.1 I
(ii) Rales lor I9K‘j arc provisional
provisional census results arc real (indirection
PriiiciptilStHiiTr' Rcgistrai General. InUiai 1990
|kt cent per annum lor I9XI-9I (Table I).
if not in magnitude). Perhaps the evidence
Economic and Political Weekly
323
December 17-24. 1994
,1
noted ..lionlall in Bihar's l9XI>9| interccnsal
lliis
v....‘l.y decline in India during the IVX(K
Illis said,
snid. our
<»ur principal
principal views
views and
and con*
con- i..
inortality
population growth rate compared with the elusions arc as follows. India's true popu*
(life expectation is
- - » now about (»() years an«’
/ I?
. I I I «.» C I) C
I . 1 ...... a
.. t
•
•
|
CRNII <ol
lhe SRS. •this
casts 1*further
doubt
kitionsi/e
is very much larger Ilian is iiulicaicd
total iv*
fertility
around 4 live niiuiNja
births), ii
it is
is niucn
much * si
J
i
........ .''
’
-----------...............................................
vw'Mi
1
uiwuiiuHvu
on Iinc level of census enumeration in. 1991
hy the. census - and tlie post-cnumcrat ion
less certain that the last decade has witnessed
.stale.
.-I...
..It
. ..i>...............
i. ..true
......
......
.
•in the
. rale ol population gmwth.
in this stale.
check
rysiillsare
a poor indicator of lhe
any_reduction
Another way to gel tentative insights into
level of undcrcimmcration. While there has - •»'
•••
We believe the evidence supports fairly .
the possible influence of census coverage ’ probably been considerable icrtilily and
si7.eablc growth talc reductions in Kerala.
changes in 1991 may be to examine state
level dillcrences between the provisional
1 Aim 4. |»r»| Cissux Ri.Mt.ts ASH Inii kcijwsai Gmom hi Raii.s. Makir SiaIix
;:ad piujccicd population totals. Figure 3
Total Population 1991
Average Annual Rale of
Per cent Change m
plots these dil let cnccs against changes in the
((XXh)
Population Growth (Per Cent)
Growih Rate
sex ratio between 19X1 and 1991. As can be
1961-71
I97I-XI
I9XI-9I
61-71 m 71-XI io
seen. Kerala’s enumerated population both
71-XI
Xl-MI
declined in masculinity and fell well short
•Slate
ol its projected total (both changes probably
Andhtu Pradesh
66305
I 911
2 10
2.14
4-11
• *2
icflcctme outmigration). But excluding
22.295
Assam.
2.9K
2.12
2 12
Kerala, all slates which fell short of their
Bihar
X6.339
1.93
2.17
2.11
+ 12
Gujarat
piojected population totals also registered a
41.174
2.58
‘2.46
I K9
Haryana
I6.3IK
2.79
255
rise in masculinity. Figure? provides a very .
233
-9
Karnataka
44.X 17
2.17
2.39
I .XX
tentative indication that at the slate level a
* tio
Kerala
29.1)11
2.33
1.77
131
\ . -24
-2b
relatively poor census coverage’may have
Madhya Pradesh
66.136
252 .
2.27
237
-IO
been associated with increased masculinity.
Maharashtra . 7X.7O7
2.43 ... 2.21
2.26
-9
. +2
Orissa
Moreover. Bihar is clearly an outlier, 'lite
•31313
223
1.85
1.7K
—1 Punjab
20.191
.u I
1.96
2.16
suggestion is. that some additional factors
I.X5
,+10
-14
Rajasthan
43.XXI
2.46
2.X7
2.47
+ 17
-14
may have been operating there.
\
Tamil Nadu
55.63X
2.01
1.63
139
-19
The increase in masculinity in Bihar has
Uttar Pradesh
I3X.76O
I.KO
2.29
2.24
+27
therefore been examined al thedistrict-lcvel.
West Bengal
67.9X3
23K
• 2.10
220
*5
All India
where necessary recombining 1991 Census
X43.93I
220
2.22
2.11
-5
data lor districts tavailable in’Prcmi 1991)
PrinciiHil
Registrar General. India (I99(>): Dyson (19X1).
so that they can be compared with the 31
Tawx 5: EcitMAtw Cmmi: Bi«m. Diaui asu Nah'kai. iNmiutsii Raii« hum Samhjdistricts of Bihar which existed at the time - r - ?
Rrc.«T«Aii<r<Svsuj4 (SRS). Mmou Si ails. 1975-77 am» I9X5-X7
of the 19X1 census. Three points emerge
* +i
Irom this exercise. Firsts virtually every
------------ I97V77----------- -------- __ I9X5-X7_______
per Cent Cliange
district tn Bihar experience! an increase in ■■ Slate
CBR
Cl)R
COR
CRNI
CBR . CDR
CRNI CBR
,CDR . CRNI
masculinity bciwecn 1981 and 199;!: the sole .:
Andhra
Pradesh
33.6
IX.9
14.7
30.6
10.0
20.6
-9 • -32
+9
exception was Dhanbad -jhc state’s most
Assam
14.9
33.1
1X2
34.4 .
124
22 0
+4 :
. +21
.masculine’district al both enumerations M .9 X
Bihar
297
123.
17.4
13.9
23.0
+24
+32
which experienced a slight decline in sex -. Gujarat
36.X
152
21.6
32.0
10.4
21.6
0 ‘
-13
-32
I laryana
ratio from J.229 to 1.200 in 1991. Second,
363
110
.233 . 352
X.9
263
-3
-32
+ 13
Karnataka
27.X
113
165
292
X.7
the greatest increases in population
. 20.5
+5
+24
Kerala
7.9
27 2
225
193
6.2
163
-17
-22
-16
masculinity appear to have occurred in
Madhya Pradesh
395
I? 6
21.9
37.7
13.7
24.0
-22
+ 10
districts located in north-central Bihar (c g.
2X4
Maharashtra
ILK
16 6
293
K.4
209
+3
-29
+26
Nawada. Sitiiinathi, MuzalTarpur. Vaishali.
Orissa
Ml
153
14.K . 31.4
13 4
+4
IK(I
-22
+22
Punjab
10.9
31.5
20.6
and Darbhanga). Third, at the district-level,
2X6
K4
-9 •
20.2
-23
-2
Raps!han
34.7
15.1
19.6 . - " 37.1
12.2
24.9
+7
there is a fairly strong and highly significam
-19 .- *27
Tamil Nadu
30.4
14.4
16.0
24.1
...
9.6
14.5
-21
-33
-V
positive relationship between the increase in
Uttar Pradesh
41 2
20.7
205 <37.7
-15.0
22.7
-X
-2X
+11
the sex ratio between 1981 and 1991 and the
122 '
West Bengal
30.9
IX.7
?'•
“
29.9
9.1
20.X
-25
+11
. intcrccnsal rate of male [jopulation growth
All-india
342
152
19.0
32.6 .
113
21.3
-26
+12
between those years; (r = 0.40. p < 2.5 per
rntici[Hil Stnircr: Rcgisirar General. India 11990].
cent (excluding Darbhanga. r = 0.53. p < 1
percent)).! his last lindingcanbe interpreted
Tabll 6; SiAtE+Evii.Changes in Potv-LAiiaN Growth between 1970s and 1980s
as suggesting that a lower rate of male
Census
outmigration from Bihar, perhaps coupled
tkxrva.se in Inlcrwnsal
. Incivase in Intervensal .
with an element of return male migialion
_____Growth Rate_____
_______ Growth Rate
during the l9X0s. has contributed to the
(Cl
(SRS)
sudden increase in masculinity in lhe stale.
’
(X-crva.sc
tn
CRNI
However, one has lo ask why any such
Gujarat
(-23)
(0)
migration effect has been so pronounced in
Kerala
(-26)
(-16)
Bihar compared with UP. ’•
Punjab
(-14)
(-2)
Tamil N;kJu
(-15)
(-9)
Discussion and Conclusions
Increase in CRNI
(C)
(SRS)
Bihar
(+32)
Andhra Pradesh
(+9)
(+ 2)
Il is dilltculi io he other than tentative in
Haryana
(-91
<+ 13)
Madhya Pradesh
(+4)
(+ 10)
interpreting the provisional’ 1991 Census
Karnataka
Maharashtra
(+ 261
results. Here we have approached them in
Orissa
(+22i
West Bengal
(+ II)
(+ 5)
a similar vein as we previously approached
. Rajasthan
<-H)
(+27)
'Uttar
Pradesh
( + 11)
the 19X1 results. And il might be objected
that this essentially constitutes an example
A7«/r: (C) dviuHcs |»criy.igc tkxlmc in intervensal growth rale.
ol getting •busy on the proof’.
(SRS) dciuxcs jyrvcntagv decline in SRS CRNI bciwecn the 1970s and 1980s.
3
323S
' c.
Economic and Political Weekly
December 17-24. 1994
perhaps l«scf dcc!'n„ In Ihr^na.Ori™
Mm^gjl'j
T"“-fc.“ ■' "9^'heSC
■» meroasei
^=2- ““xiiS'Es “sj“" «*•.~Jsaasc^s
mmnic ^wrr
^h1’ Y
farCr pr?'-' and ,980s- PcrbaPs Maharashtra is the mos:
CrinTiimi T’7 ..•difficuk slalc 10 ^count for from this per
^oTam
rf rnar.y factors other than ■ spcctive. since 53.1 per cent of its elitjibk
ndi^itedTracerS ' COUplCS WCrC cs,!malcd >° be protcuec
Pradesh. Maharashtra. Wc.st Bengal Bihar
Rajasthan and Uttar Pradesh. Overall we
consider that a picture of a constant rate of
population growth between 1971-81 and
aS?",te-.. .
as
'
decline during the 1980s has been age- •
structural change; specifically, increases in compared to 1961 7D in its
f* t
a- l|hC- ,9?°S haVC sccn 3 s!owcr rate o!
the proportion of the population in the
i .
7210 of’nlc[‘C;: Population growth than did the 1970s
principal reproouctivc years
' relationshippySO.nI98P*^^\-?.^
last decade has certain!}
Despite the fact that (he quality of India’s
r P
^CT-Iaftcr ^^w.lnessed much progress in terms of bed
demographicdatabascsccmstob^improvingrzprotect^r^7986\ST^nratVnndiaW-^
(as in Figure 2) the 1991 Census*
reductions in the
•certainly can be interpreted in such 7^“ w
of growth and indeed the SR5
issssxjsh
* $UrV.,Va1 ^P^’ation growfcFor^n^
Vr chances Rathl/theSRS*
.(forthcoming).]
more plausible (Table2^ kpu^kmgrowth
...and Dyson 1988). In this context it is pcrhaps\tand Uttar Pradesh were inflated ^nd thes^%5^^I^,):r"z’","//"n' 'lndia' /99^Censu.
ki<AWortk rc.mar^,nS't*iat the state-level pattern states also performed Dooriv'in'terms
Met^^f>Ry, BR Publishing
.
I-.- be conlributine to the similaT
’ P?pU,anon 8rowth nites rafter, the. 1991^^iflerential’ in K Srinivasan and S Mukeri
relationshin ktrv ™ u ’lanty land the.. Census. Thus the all-Indiajcvel of family;MWf(cds^-Dynamics of_ Population and Fami'l
* < R;h ;c^.Yri 2jnUA^ W^crfr ^°-28]:?^weJfare protection in 1986 wk34.9 percent^®Wc//<?"« Himalaya-Publishing House
Cf^cd)-'A!so2ti5Jik^lha^>dmostbfthesut«
5
’i'tinno05
-C
,n ®n^‘a-s ^PH^-^-pCObablyexpcricnabgaieducfionirithe'rateSi^^^jS'^-;^ C1971): Economics. Peace an
®r^ contnb?t,.ng 'sI,ghtIy to.the in-:;Xof growth did comparedvdy welli'o'n this^^^^" ” Cbntemparary Guide: a i
'
.mascul|,n!t* '<f°r example,' the ^measure: Kerala(41^tan^\^\41J)^J^!,1^s <«D. Penguin Books.
^asse.?ssaat
ZSlZ™ xE“'“ ?s*^■ •*tt"****hj^^
.. ....■ .----------- —
^Ahdhra Pradesh
L0I9
‘
r--------------------------- <■> ^Nand^R. (l991):-T^Monil/ Popula.!,,
1.023XL 1.038 • ^H.O29
99,):^iafsPripulation: Headir.
004*^^0991
7 J.069vM;l.0%'^h'i^'^fe^ s.^9cnc^rd"di=<l'972):Pr.cI<-Ie<mI-i
. |;144
I J52nfel.,w :jMl.I4O
Statixtic^ Registrar General ar
OQf&^l
tetl£!sB8«^teS£~:
i.VolXXP
Mxil-lndiaV’•? ‘2!
‘IS1 097 • ^.l-090?;
<0.986\^; .0.994Issues’^Ecotu,mic and Politic
• 32:
• •••'»**»<•<* 1«.. •
.
.
•
w
•'*
*'
’‘
*
*• -
•* *
'**-■.*
‘ '
** *'. a .'*** A ■•’•■'
*’ •'.T •«■
P* ••*«.■* * f
Sex Ratios: What They Hide and What
They Reveal
K Srinivasan
The widespread lar^e decliii.e in the sex ratios in the country in the last decade in the context of an increasing trend
in female life expectance and
It other facto, ouxcx the question of whether there was large scale under-enumeration
of females m the /99/ Censu.x.
INDIA shares a distinctive feature of the
south Asi an and Chinese populations with
regard to the sex ratio - the centuries old
deficit of females to males - the opposite
of non-Asiun countries. In India the deficit
is largely atlribulcd to woman's lower status
in soeicty which has contributed to their
higher mortality in all ages up to 45. Of more
senousconcern to the country and vocalised
by women's groups in recent years is that
(he sex rat io. de lined as the number of lemales
per I .(MW) males (which is the opposite ol
the intcrnalionally used dehnition as males
per I .(XX) females), has been declining almost
consistently over the decades, except for a
small improvement in 19X1. The sex ratio
for the country as a whole, computed to be
972 in the 1901 Census, declined steadily
to 930 by 1971. rose marginally to 934 by
19X1, but declined subsequently to 927 in
the 1991 Census. The table gives the sex
ratio figures for the years 1901.51. 71. XI
and 1991 for the states and union territories
ol the country.
The secular trends and the inter-state
differentials revealed by the figures in
Table I arc perplexing and hard to explain.
Over the decades w hen 11 ic sc x rat ios dec 111 led
from 972 in 1901 to 927 in 1991. the
expectation of lilcof females increased from
23.3 years during 1901-1910 to 40.6 years
during 1951-60. 44.7 years during I97I-K0
and lo 59.1 years during the quinquennium
19X6-91. For males, the expectation of lite
during the corresponding years was 22.6.
419. 46.4 and 5X. I years rcs|>cciivcl\ The
increase in the expectation ol hie during the
live decades. 1905-55. lor lemales was 17.3
years compared to 19.3 years for males,
lower by two years, partially justifying the
decline in the sex ratios during this period.
In the next 33 years. 1955 to I9XX. female
expectancy increased by IX. 5 years compared
lo 17 X years for the males, higher by 0.7
years. Butevcnduring this period, according
(V the censuses of 1951 and 1991. the sex
ratios have declined, a phenomenon not
consistent with the relatively larger increase
in the life expectancy of females compared
to males From demographic theory we can
easily show that in a period when the life
expectancy of females increases by a greater
length than for the males, it is impossible
for the sex ratio to decline except in the case
of large scale surplus emigration ol lemales
over males (in millions), or a very sudden
Economic and Political Weekly
drop in the percentage ol female births to
male births (increase in the sex latio at birth,
delined in this case as male births to l(X)
female births) again tn millions, both
implying events of such colossal magnitude
that they could not have gone unnoticed in
a democratic society such as India. This
phenomenon, observed m India as a whole,
is also noticed in most of the stales, but (he
thorny question of inteistate migration,
especially ihedilfcieniial migration of males ■
and lemales. compounds the issue. All the
estimates ol interstate migration lout ami in
migrants) during any mieicens.il period are
based on the data from the censuses and
using them to prove the point of inconsistency
between the increase in hie expectancy of
females t relative to males) and the population
sex ratio al the state level, adjusting tor the
migration figures, would amount to begging
the question for an answer. At the national
level the arguments arc not affected by the
magnitude of interstate migration, nor by the
modality estimates after 1966 that are based
on the system of sample registration, which
is indc|K'ndent of the census.
In many states, especially the large Hindispeaking areas, the sex ratios have been
quite low over time and still there have been
further declines between 19X1 .:nd 1991 For
example, in Bihar, it tell li.nn ‘>46 to ‘>11,
and in Uttar Pradesh front an aliv.ids quite
low Figure of XX5 in 19X1 to X79 in the next
10 years. Surprisingly, even Maharashtra,
considered to be one of the most progressive
stales tn the country with a belter status for
Tahii • Tk1) \(>s is' Si x Kavkis in mi Staii s am» Imua
Sialcx
/• rxii
Andhra Pradesh
Aruiuchul Pradesh
Assam
Bihar
Goa
Gu|.irai
Harsaaa
Himachal Pradesh
Jainmti ami Kashmir*
Karnataka
Kerala
Madh>a Pradesh
Maharashtra
Manipm
Me tlialasa
Mi/ot am
Nagaland
()rissa
Punjab
Rajasthan
Sikkim
Tamil Nadu
Uttar Pradesh
West Bengal
IJiinm T< rrihirie\
Andaman and Nicobar Islands
Chandigarh
Datlru and Sugar Ikncli
Daman and I )iu
Delhi
Lakshadueep
Pondicherrs
All India
NA
919
|liS4
|O9|
954
xr»7
XX4
XX 2
9X3
1004
990
97X
1037
I (136
1113
97 3
1(137
utis
*>lb
KM4
»>37
94 S
3IX *
771
9X0
99 5
X62
1063
NA
972
Sex Kano (No of Ivmalcs jx-r I OOP males t
I95|
1971
19X1
1991
9X6
NA
X6X
^99(1
II2X
952
X7I
912
X73
9(»6
I02X
967
941
1036
949
1041
999
1022
XI I
921
907
1007
910
X65
977
K6I
X96
954
9X1
934
K67
95X
X7X
957
1016
941
9 30
9X0
942
94(»
X7I
9XX
X65
91 I
X63
97 X
X79
X9I
97?
X62
91(1
946
975
942
X7(»
973
X92
96 3
I03X
94 1
9.3?
•>7 |
954
919
X63
9X1
X79
919
X35
977
XX5
91 I
9 72
X59
923
91 I
967
9.34
X65
976
923
960
1036
93|
‘>34
9.5X
9S5
921
XX6
971
XK2
910
X7X
974
X79
917
625
7SI
946
6’4
749
I (X)7
10*79
XOI
97 X
9X9
930
760
769
974
1062
KOX
975
9X5
934
XIX
790
952
969
X27
943
979
927
1125
76K
104 X
1030
946
* Hie 1991 Census uj\ noi conducted in Janiinu and Kashmir. Hcrce. the population
pri)|ccieil by ’he' Slan-tnv r'.Hiiinim'c has been used.
Stunt <• Regisir.ir General 11‘>‘>2 W>-K7. IO2-OS|
(Vo/c
December 17-24. 1994
3233
•; 1991 (Irudaya Rajan cl al 1991
ai
1992)' $ ’
nas the population favoured *frinnU.-4- i
wu'-»»v igcs. However, since
throughout this century with 1036?imSeC^ll^|fV^?<X’cffo^ have been
_ --------- •■•stations
such as China, the Middle East and south
Asia, including India. The differential
both in 1981 and 1991. the sex ratio w w in n^ h^C’S1”0**1 a s,a,us
«> men
treatment meted out to the girl child might ’
lower in lhe later census. (In 1991 thccensus nosri nnh^’
u'- “ ’’ rcnain lha‘ lhcir
be due to their understanding of the
was not conducted in Jammu and Kashmir. » lo ?:‘f?Orated °VCr l,nlC such
differential survival capabilities of the two
and in 1981 it was not conducted in Assam ) fcm-i'c<rPy 3 ,ghcr 'nonali,y ,cvcl for sexes. From centuries of observation of the
.
In all these 15 states the Icmale IhT lc(.nl‘1,|c:’to‘,’Parcd '= males. The possibility
r—•
•
____ _
expcctancyhasincrcascdmorethanform'lcs females'i^he!^" Undcr-Cnumcralion of mortality of male versus female in their * ?- P’
;^si
.
societies, these civilisations
might *■ have
during this period, as revealed bv the SRS
hn i
went census, which can also realised that given equal nutrition and care v
data, h is surprising that in thhs context the £ u! X’”' ? or-"’c "’"'-“mg poor
V-’
at every age from conception onward, the ,
sex ratio has declined between 1981 and '' greater demir"’ “ “
inVCS,1fa,ed in
i
male is the biologically weaker sex and
ihcre
factor th'tfc been advanced by
needs relatively belter treatment and attention ' v.’ .
for equalising the chances of survival. Onc *‘»her censuses since 1951. This widespread
........................
ilppcn al
X ''''d
F':
expects
this to "
happen
at lhc
the ,amil
family
level
, large decline in the Sex ratios in the country.’of female foeticide in the
g 'ncidence where the relatively wcakerchild gets better
v - mla»t decade in the context ofan increasing theu7e«of the '
,h™Ugr nU,ri,ion and a"cnlio"
'he others^ To
<
rend m female Itfc expectancy (rising faster V.^llrason^raphyandaSL^"’?!''*^ haVca balan‘-cd sex ratio in the population. -4.5 fe?
in3? °rm?ICS)’and S'milar pllcnom^r-a ’ ’,o identify the 4x of the babv at wrJ,hrtP ?n"larovcr;’pro«c«ivc mechanisms seem to ■ <
'
■
tnta number ol states, raises the possibility^:'stages of pregnancy WteS
haveopera,cda"hes<>cietallevclwiihregard
i »-■
. < of large-scale ■,nd..r.------------. 8
pregnancy. When these procedures lomalcchildren.Howcver.ovcrthecenturies
‘
0%
lhc additional care required for lhc male
child bwamc institutionalised as his having .
.
o
intrinsic social and religious value not
country bccause’of the issue
Ponsc
in «*« la‘ci-80s posrscss?d bX »hc female child, and
-------of
T.-T^v^’^Mina.e^aiiucany.yijsoniv rrs
intlv*
____________
recommendatiom»^ciiiesofB<>mbay.CalcuUa.MadRisandD^IH S ^ICrcnl,al nutri<ion practices might have
t>l jobs and scats in At,____ . . . .
bi become strong gender biases that became fe. co,IcScs on lhe basis of caste) and
l<^n,qucs
" ininstitutionalised.
«»«i<ionaliscd. Without such‘ an
A s yatras^ormassivcprocessionsorganS^ birth^
increase the sex ratio at
explanation it is difficult to underatand how
, the Hindu fundamentalist urouns ov^hl "
? ,lnd,an
b* onc P°'m. for
the sex ratio in many of these countries has
been unfavourable to females for centuries
in the past.
'
/ ■ -iXSSfiE^
an empirical basis, the
•
rSiS4 P°in,s-j c- rn>"> >06 to 110
computed separately lor two grouns of .T'10 b,r,hs ,o 100 fc,,lalc binhs-even then
districts, i c. those more disturb^Hh m rh ■ '
f“Pula"*,n
«<<• in the 1991 Census
- ssssrat:
.■ <
■
had aconsistcmiy lower sex-rmio thS'.Z “ ±„C™
“V*
Child duri"8
others: 909 versus 919 in Bihar 974 vf*rci*c
P
• nCy
CVCn ,f donc for clinical or
950 in Gujarat.915 versus 925 in M-idh^ ^cnchc masons, to prevent divulging of the
Prtulcsh^veSmnM^
.nfonnal.onto.hemothers.Fcmmefitieide
X74 versus 892 in Uttar Pradesh ISrinivks-m ‘
I -W11. s.milarlv the nrniJc, ? " r:
P<|pulationwith 1981 asbase.andtaking'the
observed levels of fertility and mortality/
between 1981 and 1991 under, differentV
assumptions of sex ratios at birth, revealed
that the 1991 Census is likely to have missed /
about 3.7 million more females than males
in the count (Parasuraman and Roy 1991]
Thus the possibility of increasing under
enumeration of females in India in the more ’
recent censuses needs further investigation. •
There is no doubt that women’s lower >
status m Indian society has contributed,
historically, to low age at marriage forgirls
lower literacy and educational attainment’
ann01ad™nccd “ a major contributory
bch,nd ,he dcclinc in ll>c «x ratio of
the
' C Indian
ndl‘^ population
p',pula'‘on between
bclwcen 1981 and
.
!
suh^±
■o onaufhonsid SpX
|L
‘
B
•
'A
*
11OU1 . . ... ... ..... . < ' hr;:.
Ke,erences
Srinivasan, K (1991).. 'The Demographic
Scenario Revealed by the 1991 Census’.
Journal af Fainiiy Wcljart. Vol XXXVII*
No 3. September, pp 3-9. .
if;. --'v’Vr
' • ■■
«"> »
-bserip.ion agen.s
,nUS,
india .hat all fomign
“S
"o'.
■
We take no responsibility whatsoever in respect of subscriptions not registered with us.' ’
J;:
*4;
"
'-A
■ ‘
Manager
3234
Economic and Political Weekly
?
F't;
"u,nber or suh“rip,ions ,o ,hc FPW rroi"
h^ fo^a^o T
r y<
,n,<hO"R“janS.USMishniandKNavancelhaiii
(1991); Dcclinc in Sex Ratio: An
An
Alternative Explanation’. Erononur and
■
Political Weekly. December 21.
- (1992): ’Decline in Sex Ratio: Alternative
Explanation Revisited’. Economic and
«
Poltucu! Weekly. November 14. '
-suraman. S and T K Roy (1991): ’Sonw
Obreryotionsonthe 1991 CcnxusPopuialton
g®
of India -. Journal "f
of ^""'v
Ftimilv Wef/nre ^.5’-'
Y”1 XXXVII. No 3. September, pp 62-68.
v >£-"
General of 1ndia(l992): 799/ Crnsus
?£;^
For the Attention of Subscribers and !
Subscription Agencies Outside India ;
in india £-e
‘
• ». fe.’i
December 17-24. 1994
f •'
Missing^inales: A Disaggregated Analysis ft
.-•Af/kW- ..
.■j..-J
.'
7he problem of sex ratio imbalance in India needs a disaggregated analysis. The absence of such analysis masks
the seriousness of the problem among certain groups and in certain areas. This paper presents data on the
female/male ratio for scheduled castes, scheduled tribes and the rest of the population. While further disaggregatio.
m
among various subgroups is necessary, the data presented here help to identify some major problem areas.
THE adverse female-to-malc sex ratios in A literature search between 1981 and 1991
South Asia has attracted considerable also indicates that this line of analysis has
academic and policy concern in the recent not been pursued.
years. Among different regions of the world,
This omission is surprising because
South Asia stands out both for low sex ratios anyone familiar with the tribal Indian
and lower lifeexpectancy at birth for females society
J cannot miss out its similarities with
(Sen 1987 T-l]. The ndian subcontinent Boserup’s ‘female farmingsystcm’/These
represents extreme manifestation of this arc high
- ■ - female labour participation,
pattern. .
•
•>
> " , Prcvalencc of shifting cultivation, late
Census reports in India' have'/vbici
liced entry or even absence of plough and low
concern over the declining trends in the sex level of monetisation of the economy,
ratios [Nath 1991]. This decline has taken Boserup (1976) has herself pointed this
place steadily since the turn of the century out. She has also mentioned the scheduled
[Bose 1991]. The seeming reversal in this castes as being major suppliers of casual
trend in the 1981 Census had given rise to labour and being ’unprepertied’ class in
much optimism. This has turned out to.be Miller’s terms [Miler 1981].The supply of
short-lived with the 1991 Census'figures’/ casual labour covers both agricultural and
indicating a decline in this ratio<to^927; > non-agricultural sectors. These are
females per thousand males. Available data important respects in terms of which the
reveal that increased urbanisationor? ST and the SC populations differ from the
improved economic prosperity.ihas^not; /general’ population.
necessarily resulted in correcting ithese;- One would prima facie expect, therefore,
adverse trends (Kynch and Sen1983; a difference in the sex ratio pattern among
Knshnaji 1987]. As to the understanding of- these three groups. An analysis of the
the phenomenon, the official position is that census date of three decades 1961 to 1991
it is ”... difficult to pinpoint any particular indicates this to be indeed the case
reason for the declining sex ratios which (Table 2). The female-male ratio (fmr)
require a detailed analysis...” [Nath opcit].'*.’ values expressed as nunitjfer of females per
The available literature on this subject.: thousand male populatib^. are significantly
is dominated by two themes -and, .an; . higher for the tribalipqtfulation and closer
omission - that will be the subject matter? to the African ratios.’This is followed by
o. some detailed analysis .below/Thc'
.below.^Thci fmr values for the Scheduled Caste
I/’S • F r*
.—_
•••
.•
a
omission refers to
the absence .of. an.-• —population,
while
the fmr values
for the
analysis of the sex ratio patterns amongp
among! ' non.SC/STpopu
’ktionha’vet^l^'low^t
non-SC/STpopuIaiionhavcbeenthelowest
three distinct categories of the Indian} until recently. The difference between fmr
population-the
population-the tribals,
tribals, the
the scheduled!scheduled. - values
values for
for SC
SC and
and non-SC/ST
non-SC/ST or
or‘‘general
general’’
castes
(oner.
cnn’iirirreH ”loucjhablcs)311(11 population has narrowed. down
.
. . the
ih^1 n* h°
nCCu
during
the others - normally referred to as the;
‘general' category. The analyses currently:
The fmr pattern presented in Table 2, is
done
oonc do look at different categories - urban j' both interesting and worrying. It also brings
f!rC!11’./C81i°ns or, different-; out the merit in disaggregation of the data
occupational.groups like;cultivators,^ in more than one ways/lt
allows> one to
,
agricultural labourers and the ■non-'’ compare
c:...r
2._ fmr data across the three
the
agricultural workers. The analysis in terms r socioeconomic groups, across different
ofthc scheduled tribes, the scheduled castes} - states and over the four decades, viz, 1961
.u----.
• - - - l0 ; |99I
and the
general
or the non-schcdulcd
components of the total population is absent. ’
The first feature to be noticed is that the
Miller (1981) in her ' otherwisej problem of declining fmrs is more serious
authoritative analysis of the juvenile sex •' than envisaged so far for the ‘general’
ratios
r?“0|S. in
,n India
lndia has
has explicitly
CXP*>cit|y opted
opted out
out of
of category in a number of states and for the
this line of analysis and has instead taken Scheduled Castes as a whole. For the
up individual castcwise analysis.’ Two of scheduled castes the overall fmr declined
the most recent and detailed reviews on . further by three points during 1971-1981
1 IS|I/ioai\IZ|'ChatlerjCC^"0)and Bcn“Ct when the overall fmr for the total population
ct al (l )) 1) do not cover this aspect either. went up. even if temporarily. This decline
t
■ 5
!
is seen in almost all states including
lamil Nadu and Andhra. Only Himachal
Pradesh and Punjab in the north. Kerala in
the south. West Bengal, Assam and
Sikkim in the cast were significant
exceptions. The decline was pronounced
inthecowbelt Bihar, UP, MP and Haryana.
Il was also pronounced in Orissa as well.
During 1981-1991 decade, the decline in
fmr has been significant for ST population
as well - from 983 in 1981 to 972 in 1991.
It needs to be stressed that the absence
of such disaggregation had masked some
of th- drastic reductions in fmr that took
place in the past. During 1961-1971 the
declines in fmr values among some of the
groups were as sharp as 50 females per
1000 males in Bihar. 32 females per 1000
males in MP/23 females per thousand
males in Orissa and 45 females per thousand
males in UP. Hardly any attention has been
focused on these sharp declines. Similar.
sharp declines hive taken place during
1981-1991 also. Interestingly, wherever
the fmr values for SC population have
gone up. the ’general’ category fmr values
have gone up. But the converse is not
necessarily true.
**
•
«
-----------------------------------------
i__
. Cm”’ ‘
-• • .
M
‘'f
* Y
V
■•r - v
&
>-<•:*!
Table 1: Sex Ratio and Lufe Expectancy Ratto
(Fcmalc/Male)
Region
Sex Ratio
Life
(1980) Expectancy
Ratio
(1980-85)
World
Western Europe
Eastern Europe
United Stales
Latin America
Asia •
India
Pakistan
Bangladesh
Western Asia
Eastern and
south-eastern Asia
China
0.990
1.064
1.056
1.054
0.999
0.953
0.931
0.929
0.939
0.940
1.047
1.104
1.098
1.106
1.047
1.022
0.993
0.961
0.979
1.052
1.008
Africa
1.015
0.986
1.024
1.065
1.404
1.065
Northern Africa
Non-Norihem Africa
0.941
1.060
1.021
Source: Sen, A K (1987b): ’Africa and India What Do We Have to Learn from
Each Other?’ WIDER Discussion
Paper No 19.
■I
2074
" .
. ‘
Economic and Political Weekly
August 19. 1995
* 0
•
* r •
> V..; .
Missing Females
. Another interesting way of looking at
.
the same information is to calculate the
number of missing females. If we take
say, the 1961 fmr (941) as the base fmh
■ then the number of missing females in,
.•.^■say, 1991 would be the number of
I
<
women required to be added to the 1991
I •
female population, so as to have the same
'?•. fmr as in 1961. This number can be
• • expressed as:
,
dw = [ fmr (61) fmr (91) ] * M (91)
..3.1
where M(91) is the male population of 1991
Census.
This is straightforward, as,
population. Similalry SCs account for 29 per
cent of the missing females although they
make 14.6 per cent of the population as
their share. The two figures for general
fmr (91) = W(91) / M(91). and,
5
fmr (61)-fmr (91) = (W(91) + dWJ/ :
category are 68 per cent and 78.5 per cent
respectively.
,
M(91) - W(9!)/M(9!)
•
The choice of an adhoc base, however,
Readjusting this we get the expression 3.1
above.
. j creates, a problem in further disaggre. gation of.* the data. This is becauscTthc
An exercise on above lines for'1981
• total fmr*j:.'.*-2
shows the number of total missing females
’
•
to be 31.24 lakh. The ST population F = .W/M.= ZW(i)/M = L{W(i)/M(i))»
accounts for only 3 per cent of this
number. STs form about 7 per cent of the
or. F = Ep(i) * f(i)
34
■ 'Wrs;J-'—■ :
/•
Table 2: FMR Table
(No of women per thousand
,
7
■ .v j. Statc/Union
'
•' • Territories
Year
Total
FMR
1961
1971
1981
1991
,•* x'..* ; £*<-q '■ Andhra Pradesh
1961
1971
1981
1991
^Vj^AntnacMRradesh
1961
1971
1981
1991
;
' <•
< Assam
1961
1971
1981
*■,
5^ "j j.!.
».'-y • ; .
1991
<’ ■<. '7/^ *'? Bihar
’
1961
1971
1981
1991
\Goa
1961
1971
1981
1991
Guj«*rat
1961
1971
1981
1991
Haryana
1961
1971
1981
1991
Himachal Pradesh
1961
1971
1981
1991
..
Jammu and Kashmir 1961
1971
1981
1991
a*.'/.
‘ Karnataka
1961
1971
1981
1991
Kerala
1961
1971
1981
1991
. .. Madhya Pradesh
1961
1971
1981
1991
941
• 930
. India -
z.
;;
1£
I
•
.■
:
••
’
.
t
934
927
981
977
’ 975
972
894
861
862
859
869
896
901
923
994
954
946
9il
1,066
981
975
967
940
934
942
934
868
867
870
865
938
958
973
976
878
878
892
959
957
963
960
1.022
1.016
1.032
1.036
953
941
943 '
931
Economic and Political Weekly
SC
FMR
6T
Non
FMR ’ SC-ST
FMR .
SuieAJruon.
Ten-,tones
SC
--^FMR^FMR
. ST . .
'.fmr'. ■
Non
.
FMR
987
934 .
Maharashtra
962 . ... 978 . • 932
982
983
930 <: AT” •-v'972 'S’, on
976
973.
f
962 \
960
1.013
299
. Meghalaya
‘
‘
904
1.007
460
592
1.005
599 *,
627
998
658
883
918
Miroram
862
. 7 ■ ■ - 1961 .•:1.009r'.!<>«; 0 7-:. -1.026 ■>. <• 390
917
1971;<-^.946Jpxy 38 ; -1.0214;: 208
960
886
901
1’81 -^919';125 ■ • 997 !
227
919
967
916 '
• ■■■■ ^.^-.'1991 •^j;921«»*j.i157.
982 >
243
Nagaland -.. j .7,
1961 ^933g@*575 .'r-'5.007
.287
1.031 • 1.014
985
981
1.003
943
966
993
937
.
1’81
.
955 :?;= -495
914
971
• 905
.’.v
1991
888j.—. ’»946:;>, 558
Orissa
1,066
’ 1961 ^1.001S<1.015 •-1.016991
936
742
982
1971 O988J^:”3 - J i.007 -I ' 979 .
956
845 .
976
969
967 .
889
--967‘
959
972
970
Punjab
933
950
968
927
868
942
976
936
- ■’ 1981
883
925
967
929
. - 1991 Z'
885
894
363
Rajasthan
..
1961 ■"'-y,908 ^.yt4r923 • • 926'j
902
871
r.
1971
^911^914
'
—
•
866
' 930 r
907
864
872
z, 1981.^919^913
•' 945 < ■_ 916
860
866
i99i :?A910
- ; >
^199!
899
930'!: ’ 909
934
983
937
Sikkim
;
196! >^ 904
■■
w- • 904
950
1.000
959
1971
A;
863
^842
V •
. .<m
864
J
959
978
> -.1981 :<835
:
h
;
977
•-rf «35*-. ^913
»• 801
927
-^.
967
981
978
< */1991 <
•. 914 ; <.' 862
890
876
/ Tamil Nadu
..1961 ’
. oni
. • 951
924
874
1971 ■ 978 ■
. 984
951 .
977
922
889
. 1981
■ • 968 •
976
1991 ..<974
V97S
960
973
?
965
953
Tripura
958
921
1961 ■,• & 932 •
955
921
957
957
957
1971 < ■.^^-.\:^--.94O
943:
J'- 954 ■ • 938
968
971
961
^•1981 <'946
■.^:.942
962
‘ ' 940
^9«^949
962
961
959
A
1991 ^945AV€/949
965
931
1.013
1.006
1.022
Uttar Pradesh
1961 ’ fx^-909 ’941
941
■i 901
1.0'2
995
1,017
1971 %^879??Va 896
. 880
i 874
1,022
992
1981 . <885?
’
1,033
jy 892
915
. 883
1,029
996
1,038
• 1991 .^879.;-d;,:877
914..- : 879
---...
973
1.003
934
West Bengal ;
1961 . 878
916
.969 ■. . 861
941
998
925
1971 ■'>;
927
955.
877
932
997
926
51981 ,;;%<9ii ^1926
: 969.
902
915
985
916
1991 '.•.<§.? 9I7/.
931
' 844
- ; 964
957
935
932 l
922 S*
980
973 •_
971
969
sswtes’.wi ■"
c
•4?977 O 980
Aucust 19. 199S
2075
I
Table?: Missing Females 1971-1991
o
C-J
(in 000s)
Stite/UT
Year
Male
Female
Total_________________sc_
Missing Incremental Male
Female Missing
Female
Increase
Female
India
1961
2262.93
2129.42
133.52
329.37
315.12
14.25
65.87
1971
2840.49
2641.10
199.39
1981
3439.30
3213.57
225.73
1991
4352.16
4033.68
318.49
1961
181.62
178.22
3.40
413.43
386.62
26.81
542.11
505.44
36.67
719.29
662.94
56.35
214.94
5.15
25.12
24.61
0.51
271.09
264.41
6.68
1991
337.25
327.83
9.41
29.2?
28.47
0.80
1961
1.78
1.59
0.19
1.93
40.39
39.22
1.17
53.80
52.12
0.00
0.00
191.81
188.34
3.48
260.39
255.90
4.48
2.51
2.16
0.35
1981
3.39
2.93
0.47
1991
4.65
4.00
343.63
333.95
9.68
1961
57.98
1971
Non SC-ST____________
Missing Incremental
Female
Increase
1781.94
1664.61
117.33
2235.25
2066.14
169.10
2636.81
2452.23
184.58
3289.24
3036.78
252.46
6.70
6.5*1
0.16
149.79
147.06
2.72
51.77
15.48
5.20 •
67.88
8.40
8.18
0.22
182.41
173.29
4.12
214.51
209.61
4.90
262.02
255.14
6.88
0.30
0.09
0.21
0.67
0.31
0.36
1.17
0.70
0.47'
0.39
15.57
0.61
1.67
21.43
20.57
0.86
0.00
1.48
130
-0.02
0.50
0.78
0.25
0.00
0.00
0.02
0.01
0.01
0.65
0.02
0.02
50.39
7.59
3.89
77.14
69.11
8.03
1981
104.67
94.29
10.38
1991
116.58
107.56
1961
233.01
231.54
0.15
1.84
1.85
-0.01
2.20
2.21
-0.01
0.01
2.75
2.75
0.00
1.87
1.23
0.64
3.43
0.46
6.07
5.57
0.50
48.03
41.39
6.64
4.76
4.36
0.40
8.20
7.87
0.33
64.18
56.88
7.31
0.00
0.00
0.00
0.00
0.00
0.00
104.67
94.29
10.38
9.02
8.65
7.95
0.70
14.62
14.13
0.49
93.32
85.49
7.83
1.47
32.19
33.18
-0.99
20.88
21.17
-0.29
179.95
177.19
2.75
223.71
210.99
12.71
278.57
261.05
17.52
0.01
0.19
.0.00
0.00
0.44
2.35
0.67
-0.33
0.70
p
0-17
-0.17
-0.40
-1.36
0.11
0.02
-0.06.
i
1.98
0.01
0.00
I
1.39
0.06
16.19
0.12
Assam
Female
1.01
0.16
1971
Male
1.55
0.36
2.73
Arunachal Pradesh
149.69
0.29
1.54
1981
151.62
19.68
1.75
220.09
Female
9.86
92.76
197)
Male
12.56
26.34
Andhra ?radesh
Incremental
Increase
ST_______________
Missing Incremental
Female
Increase
I
3.07
0.49
-2.55
1.69
i
Bihar
11.93
1971
288.47
275.06
13.41
1981
359.31
339.84
19.46
1991
452.02
411.72
40.30
1961
2.86
3.04
-0.19
%
1971
4.01
3.94
0.08
£
1981
5.10
4.98
0.13
i
0.21
40.14
39.37
0.77
51.58
49.84
1.75
65.69
60.02
5.67
0.00
0.07
0.11
24.71
-0.08
29.15
28.95
0.20
33.58
32.59
0.98
352.75
319.11
33.64
0.00
0.00
0.00
0.00
2.86
3.04
-0.19
0.07
0.00
0.00
0.00
0.00
3.94
3.87
0.07
0.10
0.00
0.00
0.00
0.00
6.06
0.97
20.83
0.28
3.92
16.13
0.78
■t-
0.00
0.26
-0.00
0.05
’991
j
1
i
1
V
5.95
5.75
0.20
0.12
0.12
0.00
0.26
-0.00
-0.00
0.07
i
4.80
-5.67
Goa
.•
9.96
24.62
0.05
4.99
4.87
0.12
5.82
5.63
0.19
. -0.00
0.00
0.00
0.00
j
0.07
(Contd)
1
I
h
I
I
I
I
!
Table 3: (Contd)
|-----Male
Total_______________ sc______~
Femafe Missing ” Incremental
Female Missing'Incrcmental Male
Female
• Increase
Female
Increase
State/UT
Year
Gujarat
1961 . 106.34
99.99
1971
138.02
128.95
9.08
1981
175.53
165.33
10.19
1991
213.55
199.54
14.01
1961
40.63
35.28
5.35
6.34
6.93
6.74
9.36
8.89
0.47
12.56
11.83
15.90
7.20
0.20 ’
1971
53.77
46.60
7.18
§
1981
69.10
60.13
8.97
jo
1991
88.27
76.36
11.91
1961
14.51
13.61
0.90
1971
17.67
16.93
0.73
1981
21.70
21.11
0.59
1991
26.17
25.53
0.64
Jammu and Kashmir 1961
18.97
16.64
2.32
1971
24.58
21.58
3.00
1981
31.65
28.23
3.42
1991
0.00
0.00
0.00
1961
120.41
115.46
4.95
-0.14
1981
189.23
182.13
7.10
1991
229.52
220.25
9.27
5.72
109.69
101.68
8.01
2.29
25.95
0.59
138.44
129.55
8.88
14.71
1.19
31.32
30.30
1.02
166.34
154.54
11.80
6.44
0.76
0.00
0.00
0.00
33.43
28.84
4.59
43.64
37.77
5.87
55.88
48.71
7.17
70.80
61.33
9.47
10.13
8.82
1.31
13.22
11.42
1.80
-0.01
17.48
15.03
2.45
3.33
3.11
0.22
3.95
3.75
0.20
5.38
5.16
0.22
6.66
6.44
0.22
1.50
1.34
0.17
1.98
1.83
0.15
2.59
2.39
0.20
0.00
0.00
0.00
15.87
15.31
0.56
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.62
0.61
0.01
0.71
0.71
0.00
1.00
0.98
0.02
1.10
1.08
0.02
0.00
0.00
0.00
0.00
-0.02
-0.00
28.43
27.52
0.91
0.00
0.00
0.00
37.56
36.13
-0.00
10.57
9.90
0.67
13.01
12.48
0.54
15.32
14.97
0.35
18.41
18.01
0.40
17.46
15.31
2.16
22.60
T9.75
2.85
29.06
25.84
* 3.22
0.00
0.00
0.00
0.98
103.56
99.21
4.35
128.87
123.31
5.56
151.53
145.62
5.91
0.05
0.94
182.19
174.13
7.46
77.13
. -1.70
-0.19
0.05
0.69
0.01
0.28
1.18
1.13
0.05
9.26
8.99
0.27
0.22
0.11
0.51
1.43
9.77
-0.13
0.37
0.00
0.00
0.08
2.17
2.29
0.00
0.00
0.05
0.00
0.00
0.65
-0.01
0.00
-0.02
0.84
1.30
0.02
0.02
18.83
1.28
0.00.
0.65
19.67
0.87
0.00
0.49
1.50
6.45
79.70
24.54
-3.42
143.27
85.42
0.26
0.68
149.72
0.43
0.55
0.05
. 1971
13.56
0.73
0.42
Karnataka
13.98
0.60
1.80
-0.17
Female
18.37
2.94
Himachal Pradesh
Male
18.97
1.12
1.83
>
ST________________
Missing Incremental
Female
Increase
0.17
3.81
Haryana
Female
0.27
2.73
Non SC-ST___________
Missing Incremental
Female
Increase
Male
0.38
9.39
1.21
0.35
1.55
1-4 A-.X . •
Kerala
1961
83.62
85.42
0.08
1971
105.88
107 >0
-1.72
1981
125.28
129.26
-3.98
7.22
7.13
-1.80
-0.10
8.81
8.91
-0.10
12.61
12.89
-0.28
14.23
14.64
-0.41
21.56
20.97
0.59
-0.17 .
-2.26
-0.14
-1.22
Madhya Pradesh
1991
142.89
148.10
-5.21
1961
165.78
157.94
7.84
i
214.55
.201.99
12.57
1981
268.26
252.93
15.34
1991
342.67
319.14
28.10
26.’44
1.67
38.08
35.50
2.58 „
•1.35
••
1.31
1.34.
*
' • ■
.1.30
'■
‘»'v
•
o.oi
"
1.61
1.60
0.01
33.34
33.44
-0.10
41.99
41.89
0.10
60.03
59.84
0.20
45.99
4.29
77.58
76.41
1.17
75.43
■ 95.72/
97.34 ■ ^1.62
111.36
115.07
-3.72
127.06
131.85
-^.80
110.88
103.53
7.35
144.47
133.66
10.80
170.15
157.59 '
12.56 .
214.81
196.74
18.07
0.00
-o.oo
0.19
• 0.10
1.71
50.28
0.01 .
0.01
w*
0.91
2.77
8.20
23.53
1.07< -0.01
1.08
4.73
' 1971
1.06.
-0.01 .
• ■< I
0.08
-2.10
-1.08
3.45
0.98
1.76
5.51
(Ceil id)
e
,
4!W^
Table 3 (Could)
Staic/UT
Year
Male
Maharahstra
1961
204.29
_____ Total____________________
Female Missing Incremental
Female
Increase
Male
Female
SC
Missing
Female
13.04
11.35
10.92
0.43
191.25
Incremental
Increase
5.16
1971
261.16
242.96
18.20
1981
324.15
303.69
20.46
■ 1991
408.26
381.12
27.14
Manipur
1961
3.87
.3.93
■,-0.06 .
.,5.31
0.11
1971
5.42
’
15.54
14.71
0.83-
22.99
21.81
1.18
7.21
7.00
0.21
1991
9.38
8.99
0.40
45.05
42.52
2.53
1961
3.97
3.72
0.25
.0.07
0.06
0.17
0.08
0.09
0.00
■
:•
o.oi
5.21
4.91
0.30
1981
6.84
6.52
0.32
1991
9.08
8.67
0741
1961
1.32
1.34
—0.01 '
0.09
0.09
0.00
0.19
0.18
0.01
0.01
0.01
0.00
1.62
0.09
1981
2.57
2.37
0.21
1991
3.59
3.31
0.28
0.02
0.02
0.00
0.03
0.02
0.01
1961
1.91
1.78
0.13
0.05
0.04
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.00
0.00
0.00
S°
o
36.00
1.17.
1.23
1.26
-0.03
0.01 ..
x 1.68
-0.01
2.76
4.16
3.59
0.57
1991
6.41
5.69
0.72
0.00
1961
87.71
87.78
-0.08
13.71
13.92
-0.21
1971
110.41
109.04
1.38
16.61
16.50
0.11
1981
133.10
130.60
2.49
19.44
19.21
0.23
1991
160.64
155.96
4.69
25.96
25.33
0.64
1961
60.08
51.28
8.80
13.38
11.49
1.89
1971
72.67
0.00
0.00
0.00
1.96
•’ 1.92
0.05
3.23
3.09
0.13
0.00
3.19
'3.20
-0.00
Onssa
0.00
4.08
4.06
O.Of
5.38
5.39
-0.01
7.60
7.58
0.03 ■
1.29
1.32 . -0.03
1.58
-0.03
78.52
10.86
2.31
2.31
0.01
3.30
3.24
0.06
1.71
1.72
-0.01
107.78
95.04
12.74
• '2.57
• . ■ • •
■a
3.67
, 2.61 .
-0.04.
3.55
0.11
■ .
o.i5
•
0.04
5.15
5.00
0.16
5.97
5.71
0.26
0.77
0.52
0.26
1.11
0.83
0.29
1.43
1.11
0.32
1.42
1.05
0.37
0.04
0.01
0.02
0.10
0.03
0.03
0.05
0.10
'
0.16
0.03
0.12
0.08
0.26
0.06
0.20
0.29
0.07
0.22
0.06
18.04
15.45
2.59
0.20
0.06
0.14
0.44
0.15
0.29
0.83
0.41
0.42
0.54
0.43
0.48
3.18
0.15
5.45
5.16
0.00
21.29*
0.29
0.96
-0.34
53.04
52.57
25.27
25.45*
-0.19
68.53
67.08
1.45
29.40
29.75*
-0.35
84.26
81.64
2.62
35.13
35.19*
-0.06
99.55
95.43
4.11
0.00
0.00
0.00
46.69
39.79
6.91
54.63
47.40
7.23
24.16
20.96
3.20
65.21
57.56
7.65
-0.17
3.89
0.00
0.00.
0.00
0.32
0.00
0.42
0.00
0.00
\
1.49
0.00
0.69
26.77
1.17
0.29
0.61
30.66
0.97
0.16
0.00
0.00
V1
«
I
0.01
1.20
77.12
68.27
8.85
(Con id)
I
u,.......
i
I'
0.12
3.33
0.00
ii
0.15
0.09
0.00
■ I
0.02
0.07
2.26
-•
.4.92
0.05
0.69
1.89
1991
1.55
0.04
0.41
1.04
89.37
18.52
0.00
0.11
1.02
1981
253.40
0.04
20.95
2.19
9.82
271.92
4.62
-0.03
0.32
1.12
62.85
16.97
0.14
■
1.45
Punjab
213.68
0.02
2.32
0.00
230.64
-
0.08
0.00
0.00
12.35
0.06
-0.00
1971
168.48
0.41
0.15
I
>
£
§
37.18
0.00
1981
180.82
0.35
1.55
0.21
I
0.76
0.00
0.23
1
28.48
0.00
0.00
0.36
29.24
0.00
0.07
2.40
0.41
0.00
0.11
Nagaland
14.57
0.00
0.10
1.71
14.97
0.00
0.09
1971
0.26
-0.00
0.01
Mizoram
o.oo /.
Non SC-ST
Female Missing Incremental
Female
Increase
Male
0.15
1.66
0.05
1971
11.86
1.34
0.19
Meghalaya
12.12
ST_________________
Missing Incremental
Female ■ Increase
0.35
0.10
1981
Female
0.40
2.26
6.68
'3
5 • ■
Male
I
I
i
I
■
■.
■
\
_.
.' ' Table 3: (Contd)
Slate/UT
Year
Male
Female
Rajasthan
1961
105.64
95.92
L
I
17.47
9.73
1971
134.84
122.81
12.03
1981
178.54
164.08
14.46
1.35
16.12
21.29
19.46
1.83
30.52
27.87 .
2.66
Sikkim
209.63
20.80
1961
0.85
0.77
0.08
0.97
40.07
36.01
1981
1.44
1.72
0.00
0.00
0.05
0.04 - . 0.01
000
0.15
0.28
i;
0.01
Tamil Nadu
2.16
1.90.
0.26
1961
169.11
167.76
1.35
1971
208.28
203.71
4.57
0.10
0.09
0.01
:
0.12
0.01
30.45
30.23
’ 0.22
36,86
36.29
0.57
44.85
---- 0.8J'
43.96
■
; 0.12
.
3.22
0.35
1.10
1981
244.88
239.20
5.67
0.32
1.72
Tripura
1991
282.99
275.60
7.39
1961
5.91
5.51
0.40
1
2 1971 :
-
1981
Uttar Pradesh
8.01 .4 JJS'
10.55
. ..
14-18
IQfil <
-----
0.28
;.
54.15
7 .; o.63
. 0.05,
0.46
x H O.H.
v.i 14
9.98
0.57
. **■ "
••• 0.22 ;
13,39
;0.79
S
a 0.99
0.99
'si
52.98
1.17
0.58 7 '..0.05 ‘
0.93 ?
c v&
1.60
1.51
•.» 4. -> ■> s •
A i. :i ■
V. -7
LI'
I
. °-01 j
0.06
i’ ■
0.09
?•»
- iS ■ ikslgBSM ■ OM® a B s.:-
7 ■ a'
0.88
16.19
15.06*
1.13
21.51
20.32*
1.18
-
------- • •
88.17
68.69
19.48
113.55
88.29
25.57
148.02
115.89
32.13
28.37
26.38*
1.99
190.36
147.25
43.11
0.00
1.50
-1.50.
. 0.85
. 0.77
5.79
6.86
10.98
0.00
.
West Bengal
o
588.19 ~ 520.43 - -. 61.17 '
0.08
0.00
0.06
1.08
0.93
0.15
1.63
. 1.00
0.63
0.03
0.38
'A-
0.03
0.48
•
0.0 r
0.06
0.48 '
0.43*
2.04
1.35
0.69
1.29
1.23*
138.66
136.31
2.35
171.42
165.90
5.51
200.02
192.69
7.34
228.84
' 5.29
5.29
219.81
3.17
3.17
9.03
2.11
Q-04
J. • ‘
0.06 ■'
•? \ 0.02
1.60
1.52*
0.08;
::
0.00
2.64
2.56*
0.08
„ ... S 0.03
0.12, *'V; . »<•
2.93
2.81*
0.12;.
■ l.sa'1.76* ; 0.08' 5
3.16
1.82
1.69
■ ...0.02
c
; .i. .. .
1-: 5
u 7.02
7.02 • <14.42•2.31 S'2J0*; 0.1
____
£ •< ;• >?
3
;'4.42- :: 2.60
0.03 ‘ • <
K 0.01a ,
2.98 2.86*
0.11
8.95
5.61
3.34
0.03. t
S a .i ? ?; ?,• 0.04
■ •
. 434
4-^ts
-J
-<,11.86
.7.01
4.86
s i | J .Apo
0.00 ”■ . 0.00- J ; A 306.99
276.47 ■ 30.52't'
0.49-
0.74
1.52
17.14,
- 10.85 £ •
'
- f 123.97' - 110.56 ” 13,41' “
21.85
‘
::
’ 7 ;~ 5.81 r650.75 ' - 89.62 :
1991 \ 740.37
155.99 136.77
19.22 z
1961 . 185.99
163.27 ’ 22.72 .
?
35.97 : 32.94
. 3.03
■ ••
M
-■ .
’ 2.88 *
0.32 •
208.76 ■' 25.60'
1971 . 234.36
: - 45.75'
•'
42.41
3.35 ‘
.
-0.19
1.28
1981
285.61 ■ 260.20
25.41
62.32
57.69 ; 4.63
4.02
1.10
199!
355.1 1
325.67
29.43
83.27
77.54
5.73
O o .1981
.
’
1.50
j: 1971 5 470.16-, ’413.25 .
. - V.
Non SC-ST____________
Missing Incremental
Female
Increase
0.05
' 0.00
0.00 ’ :
y •
-0.00 |
•0.02
1991
11.IP
Female
0.81
4.07
0.13
3
11.99
•
1*
1.13
ST_______________
Missing Incremental Male
Female
Increase
1.41
0.07
1971 ■
Female
0.25
0.83
6.33
230.43
Male
0.47
2.43
1991
.
. Incremental
Increase
2.30
I
>
Total_________________
______ SC
Missing Incremental Male
Female Missing
Female
Increase
Female
*Z---
. f-
47.66
584.38
512.61
71.77
150.03
120.22
29.80
”. 188.61’
153.98
34.6j
223.29
187.39
35.90
271.84
229.44
42.40
”A,(.
0i03; ..'
7:
:■
1.50
10.43
324.67
i> -’ -
1 Tl*
1.37*
10.11*
. o.o3'
A
IT
‘
0.13?®
0.32.
0.26 .
12.95
12.38*
0.58
1539
15.11*
0.48 .
15.59
18.70*
-3.11
»■
■
7.8f
16.31
4.83
-0.10
1.27
-3.58
6.50
J
■W'l
Where M(i) and W(i) are men and women
in ith category and P(i) is the proportion of
male population.
It is easy to see that further disaggregation
would render
■ r-
F = Xp(i)*f(i) = X £p(i)* p(i, j).f(i, j)] ...3.5
and so on where f(i, j) is the fmr for the
jth subgroup of ith category and p(i, j) the
proportion of male population of the jth
subgroup to that of the ith category. ,
If we use the above expression into the
expression for missing females, i e, in 3.3
above the disaggregation would become
messy indeed. A simple way out of this
would be to take the desirable or ‘base’
fmr to be unity or 1000 females per
thousand men. In that event, the number
of missing females dW = M - W; i e, the
difference between the male and-the
female population would give the number
of missing females. This expression lends
itself immediately to further disaggregation,
for,
r”
•-
and declining fmr. Some others may be
improving in all three respects. For
scheduled castes, UP provides the example
of the former while Kerala provides an
example in latter category. West Bengal
on the other hand gives a mixed picture
of high contribution to the number of
missing SC females but steadily improving
fmr.
The number of missing females has gone
up from 133.52 lakh in 1961 to 318.49 lakh
in 1991. However, the tribals who
contribute about 8 per cent to country’s
population contribute about 3 per cent to
the missing female popuk .ion. The nonSC/ST category contributes disproportion
ately to it - about 79.3 per cent while its
share of population is 75.4 per cent The
scheduled castes contributed 17.69 per cent
to the missing female population as against
their share of 16.5 percent of the overall
population.
)
’
If we notice thb trend across time, the
incremental increase in the number of
missing females has gone up sharply in
1991. Further, the percentage contribution
of the SC population to the missing female
population is increasing while that of the
‘general’ category coming down. In fact,
the picture becomes sharper in locus if
we look at the addition io the number of
missing females across 1971-1991
period. The contribution of ST, SC and
general streams to this addition (66 lakhs)
in 1971 was 2.3 per cent, 19 per cent and
78.8 per cent which changed to 5.6 per
cent, 21.2 per cent and 73.2 per cent,
respectively.
If we disaggregate the data at the state
level, few more useful insights can be
obtained. Different stales contribute
differently to the national aggregate and one
can identify the major contributors or
‘culprits’ in a given category. The trend in
fmr pattern in such states is also worth
noting.
'The
T^- ST
'
. .
—
population:
Three categories ofstates can be mainly discerned on the basis
of the fmr - the tribal majority states, states
with significant tribal population, i e, Orissa
•
t
dW = ZdW(i j,k...) = Z [M(i j,k....).
-W(ij,k...)J
.
a .13.6
4
4
I
I■
\
where i,j,K... represent the progressive
disaggregation into different relevant
subgroups. One could now really play around
with disaggregation as the reference fmr has
been ‘normalised’ to unity,
.:
State/UT
Table 3 gives the estimate of the missing
females statewise and categorywise arrived _______________
at in the above manner. It also shows the .. India '
incremental increase in" the number of Andhra Pradesh
missing females in successive decades.
Arunachal Pradesh
Tables 4.1, 4.2, 4.3 and 4.4 estimate this ~ Assam
Bihar
number for the 1991 census data. It also
indicates the percentage contribution made Goa
by a particular state to the number of missing Gujarai
Haryana
females in a given category to that at all
Himachal Pradesh
India level. This is further compared to the
Jammu and Kashmir
share of population of that state to the all
Karnaiaka
India population in that category.This helps
Kerala
one in getting some idea whether the state
Madhya Pradesh
contributes disproportionately or
Maharashtra
otherwise to the number of missing females
Manipur
in a given category. Andhra Pradesh, for Meghalaya
example, has 7.93 per cent of the country’s
Mizoram
population but its contribution to the Nagaland
o____
number of missing females is hardly 2.95
Orissa
.i
ihandiihas i16.59
/r m
Punjab
'
per cent. U.P. on ♦the other
per cent of the country’s population but
Rajasthan
contributes 28.14 per cent of the number
Sikkim
of missing females..
Z
Tamil Nadu
Tripura
In addition to this we also have to look
Uttar Pradesh
at whether the fmr is decl ini ng or increasing
West Bengal
in a. given category and in a given state.
A and N Island
Some states may be doing badly on all the
Chandigarh
three scores, e g, larger contribution to the
Dadra and Nagar Haveli
number of missing females on all India
Daman and Diu
basis in a given category compared to its
Delhi
share of population, larger contribution to
Lakshadweep
overall missing female population in the
Pondicherry
slate compared to its share in population
Table 4.1: Missing Females-Statewise 1991
(Total Population)
Male
Female
Missing
Female
Percentage
Contribution
All India
Total
Percentage
Share of
Population
(in 000s)
4352.16
337.25
4.65
116.58
452.02
5.95
213.55 •
88.27
26.17
4033.68
327.83
4.00 \
107.56
411.72
5.75
199 54
76.36 .
2553
318.49
9.41
0.65
9.02
40.30
0.20
14.01
11.91
0.64
100.00
2.96
0.21
2.83
12.65
0.06
4.40
3.74
0.20
100.00
7.93 0.10
2.67
10.30
0.14
4.93
;
i,96
0.62
5
229.52
142.89
342.67
408.26
9.38
9.08
359
6.41
160.64
107.78
230.43
2.16
282.99
14.18
740.37
355.11
154
359
0.71
0.52
51.56
0.27
4.08
220.25
148.10
319.14
381.12
8.99
8.67
3.31
5.69
155.96
95.04
209.63
1.90
275.60
13.39
650.75
325.67
1.26
2.83
0.68
0.50
42.65
0.25
4.00
9.27
-5.21
2353
27.14
0.40
0.41
0.28
0.72
4.69
12.74
20.80
0.26
7.39
0.79
89.62
29.43
0.28
0.75
0.03
0.02
8.90
0.02
0.08
2.91
(163)
7.39
8.52
0.12
0.13
0.09
0.23
1.47
4.00
6.53
.0.08
2.32
0.25
28.14
9.24
0.09
0.24
0.01
0.01
2.80
5.36
3.47
7.89
9.41
0.22
0.21
0.08
0.14
3.78
242
5.25
. 0.05
6.66 «
0.33
.16.59
8.12
0.03
0.08
0.02
0.01
1.12
0.01
0.10
0.03
.
>
I
!■
2080
Economic and Political Weekly
August 19, 1995
.
><•
I
I
and MP and the ‘mainland’ states of
numbers but not disproportionately
Rajasthan again contribute disproportion
Rajasthan, Gujarat, Maharashtra. Bihar,
compared to their share of the population
ately to the number of missing females
Andhra Pradesh and West Bengal. The
on all India basis. The southern states and
compared to their share of population.
former two categories have high fmr values
Orissa contribute much less to the number The share of West Bengal is expected to
but these are coming down at a rate which
of missing females comnarcd to their share steadily come down with improving fmrs
should cause concern particularly in MP
in the country’s SC population. But as the
and as indicated below, the figures would
and Orissa. The ‘mainland’ states show a
fmr tables reveal, the fmr values for the SC
improve further if migration correction is
sharper decline still and contribute
population are going down in all these stales
made.
signiticantly to the number of missing
except Kerala.
Icma es. Rajasthan deserves particular
The 1991 Census data reveals a sharp
The Migration Effects
■ attention in (hiscontext. Withan fmrdecline
increase in the number of missing females
. of 15 women per thousand men between
in this category. Their number has gone up
The above analysis of the sex ratios has
1981 and 1991 its contribution to (he number
from 184.6 lakh in 1981 to 252.5 lakh in
to incorporate the effects of migration
of missing ST females has gone up to 20.57
1991. The incremental increase which
on sex ratios. FMR values will be higher
percent while its share of the country’s tribal
had come down in 1981 (15.5 lakh) has
if there is an outmigration of males or
population remains only 8 per cent. If we jumped nearly four-fold to 67.88 lakh.
inmigration of females. Similarly,
look at the incremental increase in the number
Bihar and UP which share about 28 per outmigration of females or inmigration of
of ‘missing females’ between 81 and 91, we
cent of the country’s non-SC/ST population
males will depress the FMR values. All the
find that the four states of Bihar. MP. Gujrat
contribute about 41 per cent to the number four types of migration occur in any region.
and Rajasthan contribute to about 60 per of missing females in 1991. Their As such-one has to work out the net
cent to this increase.
contribution to the incremental increase is outmigration of both male and the female
The decline in the ST fmr values across
higher still and is over 47 per cent. These population for each state and add it to
’the 1981-1991 period has virtually occurred
two states are singled out not just for the the enumerated population to arrive at the
in all the states and is in a sharp contrast
magnitude of their contribution alone but corrected FMRs. It is quite possible that
; with the trends between 196! to 1981. This
also for the declining trends in fmr. The different, social groups exhibit different
. is a matter of worry and warrants further decline in Bihar by about 32 females per patterns^of;migration. This would
y-' scrutiny.
thousand men across 1981-1991 period is necessitate?a disaggregated analysis of
The scheduled caste population has
rather serious. Haryana. Punjab and the migration figures as well. Some
witnessed a steady deterioration in the fmr
values across all the regions. Notable
Tablt. 4.2: Missing Fcmaijf-S. 1991 ,
exceptions are Kerala, Himachal Pradesh.
(Non-SC/ST) , ( \
Punjab, West Bengal. Sikkim and Manipur.
Percentage
In general wherever the SC fmrs have gone
Contribution
.
up during a given decade, the non-SC/ST
Statc/UT
Female
All India Percentage
Male
Missing
Slate
fmr values have also gone up but not vice
Female
. Total
Total
Share of
versa.
(in OOOs)
Population
The decline of the fmr in Bihar from 966
3289.24
India
252.46’ 6 ' 79.27
!00.(X)
100.00
3036.78
in 1981 to 914 in 1991 shows that there
Andhra Pradesh
255.14
6.88 73.08
2.72
8.18
262.02
is something seriously wrong. In fact, the
Anmacha! Pradesh
1.87
0.64
97.90
1.23
0.25
0.05
x trend across 1961 to 199i shows similar Assam
85.49
7.83 v -1. 86.86 ., . . 3.10 .if 2.83
93.32
decline: in 1961 Bihar had fmr of 1031
Bihar
3.3.64
83.49
10.62
352.75
319.1!
13.33 .
which declined to 981 in 1971 and further
Goa
5.82
0.19 >
97.80
0.08
0.18
5.63
11.80; .
. 84.21 *
To 966 in 1981 .This merits further attention
166.34
4.67 .
Gujarat
154.54
5.07
9.47 ‘^79.46
ns well as policy intervention.
Haryana
70.80
2.09
6133
3.75
0.40 ■■ 7 62.62
18.41
18.0!
0.16
0.58
As indicated earlier, the contribution of Himachal Pradesh
_
Jammu and Kashmir
the SC population to the number of missing
5.64
182.19
Karnataka
174.73
.7.46
80.50
2.95
females is growing over the years. In absolute
Kerala
127.06
(1.90) '
4.09
131.85
—4.80
92.19
numbers it has moved from 14.25 lakh in
Madhya Pradesh
214.81
196.74
7.16
6.51
18.07
76.79
.. 1961 to 56.35 lakh in 1991. In 1991 its
‘ 302.59
Maharashtra
23.44 ....... . 8636
9.94
326.02
9.28
percentage contribution to the total number
Manipur
5.71
0.10
0.26;,.: , 65.19
5.97
0.18
of missing females was 17.7 while its
Meghalaya
0.04
1.42
o.i5 ;
1.05
contribution to the incremental increase was
Mizoram
0.29
0.09
0.01
1 0.07
Nagaland
096
0.17
hr o
higher still, i e, 21.2 per cent. Of this
0.02
. 0.54
Orissa
4.11.
87.80
’ 1.63
3.08
99.55
95.43
incremental increase of 19.68 lakh, UP. Bihar
8.85 ' T-A ■ 69.48
3.51
Punjab
68.27
77.12
2.30
and MP contribute about 58 per cent while
5.84 ;
161.99
14.74 ^7 70.87
4.89
Rajasthan
147.25
their share of the country’s scheduled caste
0.22 '
‘ 81.55
Sikkim
0.09
0.05
1.56
1.35
population is 37 per cent .
b.lO/Vv.Vf ’ 82.58
Tamil Nadu
225.91
219.81
2.42
7.05
If we look at the absolute figures for Tripura
0.20 ■
0.23
7.52
7.01
0.52.
. 65.65
1991, we find that UP alone contributes
17.32
Uttar Pradesh
582.87
512.61
70.27^. .:, 78.41
27.83
34.11 per cent to the total number of missing
. 7.62
229.44
23.01..;. 78.19
9.12
West Bengal
252.45
females on all India basis. This is about 13
0.04
0.11
A and N Island
* '1.13
0.27
.97.40
<^(11
0.64 ’.
85.20
0.08
('•3.00
0.25
per cent in excess of its population share
Chandigarh
236
0.04 . • ’
0.02
0.11
Dadra and Nagar Havelii/l 0.15
(21.2 per cent). Haryana, Punjab and
0.01
: 0.44
0.01
■ 81.6!
0.01
Daman and Diu
0.42
Rajasthan contribute another 18.5 per cent
'’81.74
2.88
1.21
34.49
Delhi
41.77
7.28
while sharing only about 12 per cent of
0.0!
Lakshadweep
002
0.01
0.01
• 89.73
the country’s SC population. West Bengal
0.1 I
86.82
0.03 •
Pondicherry
3.42
0.07 ••
335
and Madhya Pradesh contribute in large
°
■
fe
E
I
L
Economic uncl Political Weekly
<’■
<■
«■<>■
August 19. 1995
^: 5
20NI
S
. J
'J
h
(
■
--------------------------
■r‘<. ■ * -
i
Table 5.1 : Percentage of Total Migrants to
useful observations could nevertheless outmigration Iror' Kerala. While a
disaggregated analysis of the migration Total Population in Each State according to
be made, t
y
■ h,
Place of Birth
figures and necessary’ migration corrections
The migration effects will be" more
arc needed to be made one can expect that
pronounced as we go down from the
Total Within Outside
Stale
the picture would change only marginally
national ’evcl to the regional^ slate, district
the
Migrants the
in most eases. These changes would
State
Slates
and the village level. At the state level the
certainly riot account for the trends in the
effects are still insignificant as revealed by
1.63
31.57 29.86
Andhra Pradesh
fmr and the number of missing females
(he 1971 census figures (Table 5.1). Inter
3.58
33.21 23.03
Assam
observed above.
state migrations hardly involve 3 to 3.5 per
1.66
27.96 25.87
Bihar
This analysis establish the usefulness of Gujarat
2.87
cent of the total and 10 to 15 per cent of
32.06 28.53
8.54
18.6
disaggregation while analysing the
32.61
Haryana
the migrant population. Major migration
3.85
Himachal Pradesh
33.17 27.96
observed sex ratio imbalances in India. Il
takes place within the states.
Jammu and Kashmir 18.79 16.38
1.5
also presents a simple and useful method
Among di fferent slates, Maharashtra and
1.17
Kerala
21.83 20.5b
of
analysing
the
problem
by
unpacking
West Bengal have traditionally attracted
29.1
4.27
Ma<’ lya Pradesh
34.16
the missing females issue through
migrants fromolherstatcs for occupational
6.51
38.7 31.51
Maharashtra
equation 3.6
reasons. In these two stales the migrant
2.34
Manipur
18.73 15.23
The above disaggregation prioritises the
5.39
population accounts for 6.41 and 4.72 per
29.98 18.41
Meghalaya
3.95
areas needing remedial intervention and
31.26 27.21
Mysore
cent of the population of. the Slate,
17 91
7 27
832
Nagaland
suggests that it has to be designed
respectively. Haryana (8.54 per cent) and
3062 27.95
Orissa
differently in different parts of the society.
Punjab (4.32 per cent) are the other two
36.34 24.04
Punjab
states with high migrant population. This Tribal Orissa for example, would require
3.32
29.76 25.42
Rajasthan
different intervention compared to. say,
is mainly on account of the agricultural
1.94
Tamil Nadu
28.56 26.21
the
‘
developed
’
Haryana.
In
areas
where
prosperity following the green revolution.
1.62
Tripura
47.15 I 1.46
fmrs arc already favourable, the intervention
Madhya Pradesh is the only olher^non25 01 23.04
1.42
Uttar Pradesh
4.72
Wesi Bengal
30.22 18.19
should aim at consolidating the position. On
border slate where lhe migrant population
3.4
30.42 25.31
India
the other hand, where the fmrs arc low and
accounts above 4 per cent of the population
deteriorating, immediate corrective inter
of lhe slate.
(Migrants from outside India 1.66)
A preliminary study of the effect of ventions, c g. improvement in girl child's
Si>ur< r: Census«// India 1971. Paper No 2 of 1979
access to nutrition and female literacy
S C Shrivaslavu: Mixixilian in India
migration on FMR for different states was
done for the total rural population
(Table 5.2).Thetotal population was taken
Table 4.3: Missing Females 1991
for lhe analysis as the disaggregated data
(Scheduled Caste)
for migration in terms of the ST, lhe SC
Percentage
and the general categories is not readily
Contribution
available. The population of the state was
All India Percentage
Slate
Male
Female
Missing
State/UT
corrected for by adding lhe figure for the
Total
Share of
Female
Total
net outmigration separately for the male
Population
(in (XXK)
tmd the female population and the corrected
719.29
662.94
17.69
l00.(X)
100.00
56.15
India
FMRs were worked out. The expression
1.67
7.66
•2.97
17.77
53.80
52.12
Andhra Pradesh
for the corrected population is thus:
0.01
0.02
0.02
1.42
0.02
Arunachal Pradesh
7.74
1 24
0.70
1.20
7 95
8.65
Assam
CP = EP + Out - In
1006
9.10
5.67
14.07
65.69
60.02
Bihar
where CP is the corrected population, EP
2.09
0.02
0.01
0 12
0 12
Goa
2.21
14 71
8.50
2.11
is the enumerated population, ‘out*, is the
15.90
I 19
Gujarat
4.34
20 54
2.35
17.48
2.45
Haryana
15.03
outmigrating population and'‘in* the
0.95
644
0.39
0 22
34.05
6X6
Himachal Pradesh
inmigrating population.
• '
A comparison of the fmrs for different Jammu and Kashmir
5.33
15.42
36.13
1.43
Karnataka
37 56
slates with lhe corrected fmrs shows
(0.73)
2.09
14 64
-O.4I
7.93
14.23
Kerala
marginal changes in the fmr. values.
6.96
7.61
4 29
50.28
45 99
18.23
Madhya Pradesh
6.34
However, there is a significant improve
4.49
42 52
9.32
Maharashtra
45.05
2.53
0.01
0.01
0.03
0 19
0.18
1.28
Manipur
ment in the fmr for both Maharashtra
2.19
0.04
0.01
0.02
0.01
Meghalaya
0.05
(965 to 993) and West Bengal (942 io
0.00
1.78
0.01
0.01
0.01
Mizoram
978). respectively. For Madhya Pradesh
Nagaland
lhe improvement is marginal (956 to 961).
0.64
13.58
1.13
3.71
25.96
25.33
Orissa
All lhe northern states of UP, Rajasthan,
30.52
6.90
4.15
26.77
3.89
30.66
Punjab
Punjab, Haryana and Himachal Pradesh
7.22
5.50
40.07
4.07
19.55
36.01
Rajasthan
0.02
2.89
0 01
0.12
0.01
0.12
Sikkim
show a decrease in lhe fmr. The decrease
7.75
54.15
1.17
15 82
2.08
52.98
Tamil Nadu
for Bihar (971 to 952) is quite significant
0.33
0.21
0.12
15 15
2.32
2.20
Tripura
and it would be interesting to study lhe
21.18
34.1 I
136.77
19.22
21.45
Uttar Pradesh
155.99
calcgorywisc breakup. There is a likeli
10.17
11.63
77.54
19.47
5 73
West Bengal
83 27
hood of significant outmigration of tribal
A and N Island
males from Bihar which may account for Chandigarh
0.08
0.20
0.47
0.11
14.80
0.59
0 01
0.01
3.09
the high fmr for the tribal population.
Dadra and Nagar Haxeli
002
(7.92)
Daman and Diu
0.02
Kerala is the only southern slate where
1.30
18.26
2.88
9 79
8 16
1.63
Delhi
lhe corrccicd fmr drops appreciably
Lakshadweep
from 1020 to 1005. But this is consistent
0.09
13.18
0 02
0 01
0 66
0 65
Pondicherry
with the known and traditional male
■'*
I,
I
I
6
8
i
h
t
•
■
■.
11
2082
<2
Economic and Political Weekly
August 19, 1995
Table 5.2: Nter-State Migration Corrected FMRS
Migration (Total POP) Correction (0.1) Corrected PDP
Statc/UT
Total(M)
Total(F)
India
284049
225320
58729
22009
17698
4311
7714
6975
739
28847
25729
3118
13802
9842
3960
5377
4420
957
1767
1629
138
2458
1997
461
10588
8852
1736
21455
17823
3632
26116
17842
8274
542
470
72
521
444
77
14972
11249
3723
276
241
35
11041
10041
10<M)
7267
5533
1734
13484
I 1061
2423
20828
14439
6389
80.
707
84
47016
40214
6802
23435
17174
6261
264116
930
213716
949
50390
858
21494 .
977
17402
983
4092
949
6911
896
6361
912
550
744
27506
954
24990
971
2516
807
12895
934
7358
951
3537
893
4660
867
3844
870
816
853
1693
958
1590976
103
746
2158
878
1762
882
396
859
10759
1016
9029
1020
1730
997
-0198
941
17046
956
3152
868
24296
930
17219
965
7077
855
531
980
461
981
70
972
491
942
423
953
638
883
14327
957
10928
971
3399
913
240
870
224
929
16
457
10903
988
10056
1002
845
845
6285
865
4801
868
1484
856
12281
911
10161
919
2120
875
20371
978
14296
990
6075
951
755
943
677
944
78
929
41325
879
35739
889
*586
821
20876
891
16171
942
4705
751
R
U
Andhra
T
R
U
Assam
T
R
U
Bihar
T
R
U
Gujarat
T
R
U
Haryana
T
R
U
Himachal
T
R
U
J and K
T
R
U
Kerala
T
R
U
M P
T
R
U
Maharashtra T
R
U
Manipur
T
R
U
Meghalaya T
R
U
Karnataka
T
R
U
Nagaland
T
R
U
Onssa
T
R
U
Punjab
T
R
U
Rajasthan
T
R
U
Tamil Nadu T
R
U
Tripura
T
R
U
UP
T
R
U
West Bengal T
R
U
Arunachal
Pradesh
T
R
U
Goa, Daman T
Diu
R
U
251
231
12
431
311
120
216
211
5
427
320
107
FMR
86!
883
417
991
1029
< <-892
OUT(M)
OUT(F)
1N(M)
(IND
NET(M) NET (F)
Corrected
POP+W(M)
POP+W(F)
FMR
(Correction in the total category not tflected on account of the
Unspecified* Category included in the Census Figures
Figui
326.06
1753
364.81
182.22
165.9
144.7
248.7
149.3.
160.16
30.6
116.11
32.92.
.z
17858.16
4341.6
17518.11
412-1.92
981
95/.
69.6
336
52.3
29
•300 8
46.6
131.6
31.3
231.2
• 0.13
-
-
79.3
2.3
6743.8
726
6281.7
547.7
931
754
249^.:..^ 26525.2
8.16 : - ; 3186.86
25239.2
2507.84
952
7X7
' ’ 984 1.61
9408.96
3559.72
956
894
3813.3
818.9
863
849
o.r'y.148.3
1618.24
103.1
964
695
1.414;^ 1989.63
731
1763.41
. 403.51
886
859
9086.7
1946.5
9136.4
. 1853.1
1005
952
16818.9
•.-■..2044.2
961
872
.
1045.7
184.06
686.2 249.5 ? 437
796.2
119.14 . ll£2
1273 .-£68.86
25651
182.26
240.56
171.92
2$6*.9
159.6
189.6 ;;: ft 039
189.6
1492
C 22.66
225.8
103
374.8
127.5
225.5
95.7
4053
124.6
& 03
44 73
89.14
37.7
65.54
28.6
39.7
27.4
37.3
28.5
49.44
10.3
19.13
22.88
16.01
20.91
26.5
143
.14.6 -/U 737
13.4 r : 838
•
“
•
••“ . --'J, ■
50.96
22.72 .&fe3982'66
1678.44
28.24
.
'
^’Wx69-58
317.9
254.6
184.3
164.5
83.2
44.1
76.9
41.4
215.9
126.7
448.3
174.9
540.9
268
675.4
282.7 .
325
227.1 y
17498 .
141.3 - 107.8^L49C17
249.4
242.8
362.4
280.2
1172.8
719.4
779.5
567
923.4
476.6
417.1
286.8
16918.6
7797.4
16801.9
6790.2
993
871
4.86
2.11
2.91
1.14
12.8
2.9
7
1.22 .
7.94
0.79
4.09.
0.08
462.06
71.2'1
■ 456.91
69 92
9S9
9X2
8.34
6.84
7.48
6.38
8.34
6.84
7.48
6.38
452.34
83.84
430.48
74.38
954
887
268.5
238
366.1
240.4
-42.9
-5.1
11206.1
10922.9
3420.6
975
920
2.5
1.3!
1.7
1.15
225.7
17.15
927
472
194.3
44 4
198.8
34
166
62.2
207.2
54 8
28.3
-17.8
-8.4
-20.8
10069.3
982.2
10049.6
824.2
998
839
386.6
225.2
338.8
215.6
163.7
104.4
202.1
112.4
222.7
120.8
136.7
103.2 ?
5755.7
1854.8
4937.7
1587.2
858
856
438 6
189.5
5209
184.1
186
109
410
146
252.6
‘ 80.5
110.9
11313.6
2503.5
10271.9
2158.1
908
862
288
277.8
269
244.2
178.7
218
183.4
216.6
109.3
59.8
85.6^ 145423
27.6:
.6448.8
14381.6
6102.6
989
946
12.3
4.1
7.22
3.1
103
4.29
6.4
3.6
2
-0.19
0.82 ;X.
-0.5. 83.81
677.82
77.5
943
925
1414.9
555.6
982
427.6
268.9
172.9
550.4
249.9
1146
382.7
431.6 ■ > . ‘ 41360
7184 7
36170.6
5763.7 .
875
802
164.6
168.3
262.4
• 177.4
1009.9
224 9
463
123
-845.5
-56.6 *
-200.6 ?•?•? 16328.5
54.4 .. . 6204.4
15970.4
4759.4
978
769
239
12
211
5
883
417
26.09
31.2
27.85
33.95
311.4
243.6
371.2
218.8
234.7
.■ 210.5
123.1
<
' -
2.5
1.31
26.09
31.2
‘
'
38.1
27.85
33.95
’
M719
337.09
151.2
347.85 • 1032
932
140.95
(- Indicates insignific-'^t value-'
Economic and Political Weekly
y
August 19, 1995
20X3
1 .
r
campaign would be called for. Areas where
the fmrs are low but improving the measures
already taken should be accelerated. A
While a detailed discussion of the
interventions and their design would be
outside the scope of this presentation; two
observations can be made. The author has
elsewhere analysed lhe 71 census data in
more details and has examined the link
between fmr and flp. The significant
differences between lhe three sociocultural
groups, i e, the ST the SC and the ‘general’
population arc in conformity with above
•analysis. However, lhe lip emerges as a
dominant variable in determining the fmr.
The implications of these findings arc
that creation of opportunities for female
labour participation in productivcaclivities
and ensuring the female’s control over
the income thus generated will be the
immediate need in areas of low and
deteriorating fmrs.
The second issue relates to further
disaggregation of lhe missing female
patterns. The analysis of 1971 census data
referred to above indicates that agricultural
labourers, cultivators and non-agricultural
workers also exhibit important differences
between themselves within each of lhe
above categories. The magnitude of (he
problem of sex ratio imbalance among
these subgroups will warrant further
scrutiny.
Conclusion
It will be ambitious to draw detailed
conclusions at this stage of lhe analysis.
But it is clear that lhe problem of sex ratio
imbalance in India needs a disaggregated
analysis. Il is also clear that lhe absence
of such analysis masks lhe seriousness of
lhe problem among certain groups and also
in certain areas. This has perhaps been the
reason why fmr deteriorations as sharp as
50 women pei thousand men during
different decades in some of the stales,
failed to attract policy or academic attention.
The decline in fmrs revealed in 1991
necessitates that such attention is focused
on (he major problem areas. Such problem
areas can be identified with the help of lhe
Table 4.4: Missing Females. 1991
... (Scheduled Tribe)
Siate/UT
Male
India
Andhra Pradesh
Arunachal Pradesh
Assam
Bihar
Goa
Gujarat
Haryana
Himachal Pradesh
Jammu and Kashmir
Karnataka
Kerala
Madhya Pradesh
Maharashtra
343.63
Female
Missing
Female
(inOOOs)
*
Manipur
Meghalaya
Mizoram
Nagaland
Orissa
Punjab
Rajasthan
Sikkim
Tamil Nadu
Tnnura
Ultar Pradesh
West Bengal
A and N Island
Chandigarh
Dadra and Nagar Haveli
1 iainan and I)iu
Delhi
I .akshadweep
I'oikIk licuy
20X4
21.43
2.75
14.62
33.58
0.00
31.32
0.00
1.10
333.95
20.57
2.75
14.13
32.59
0.00
30.30
9.6X
0.B6
0.49
0.98
1.02
1.08
0.02
939
0.38
0.01
1.17
1.17
0.13
Percentage
_ _ Contribution. _
Stale
All India Percentage
Total
Tola!
Share of
Pupuk.on
3.04
100.00
9.15
8.90
0.68
0.05
5.40
5.03
2.44
10.14
0.11
0.00
7.29
10.55
O.IX) , f
O.tXI
0.22
0.00
4.08
3.90
(0.13)
0.07
4.98
12.12
4.32
12.11
33.53
1.37
6.26
0.26
21.54
0.63
40.95
3.05
(0 66)
(137)
O.(X)
0.00
9.58
20.57
15.56
0.42
1.60
1.22
19.20
1.56
9.77
1.61
77.58
37.18
3.23
7.60
3.30
5.45
35.13
76.41
36.00
3.09
7.58
3.24
5.16
35.19
28.37
0.48
2.93
4.34
1.50
1939
0.14
26.38
. 0.43
2.81
4.19
1.37
18.70
0.13
1.99
0.04
0.12
0.15
0.13
0.69
0.01
0.14
1.34
235
2.60
7.15
0.08
0.54
0.55
-o.oi
0.06
0.06
(34.30)
26.31
0.24
0.24
10 27
(0.12)
0 04
0.G0
0 02
1.60
0.03
0.06
0.29
-0.06
100.00
6.20
0.81
4.24
9.77
9.09
OCX)
0.32
0.00
2.83
0.47
22.73
10.80
0.93
2.24
0.96
157
10.38
0.00
8.08
0.13
0.85
126
0.42
5.62
0.04
O.(JO
0.16
0 02
0.00
0 04
(l (Ml
data presented here and through further
disaggregation among various subgroups.
Wc already notice that the northwestern
stales of India exhibit an adverse trend
towards female survival. It can further
be seen that such u trend is making
inroads in other parts of lhe country, as
well. Certain regions also stand out for not
succumbing to this trend. The strategy for
improving lhe declining fmr trend will
differ between these two regions. That
would be the objective of the next stage
of this analysis.
Note
1 Miller opines, in her analysis of the social
variations that ""...Given the choice between
using the modem (census) data for just three
categories (i c ST. SC and General) or using
somewhat old data (1931 census) on castes.
I opted for lhe latter.
References
Agnihotri. S B (1992): ‘Sex Ratio AnalysisThe Indian Context’. Proceedings of lhe
Annual Conference of the Development
Studies Association (USA) at Nottingham.
September.
Bennet. L (J99|); (’tenderand Pavertv in India:
A Cauntrv Study, The World Bank.
Washington.
Bose. A (1991): Papulation of India-i9VI
Census Results and Melhodolo^w BR
Publishing Corporation. New Delhi.
Boscrup. E (1970): Women's Role tn E< anomic
Development, Allen and Unwin. London.
Chailerjee. M (1990): "Indian Women: Their
Health and Productivity*. World Bank
Discussion Pa|x.-r No 109.
Krishnaji. N (1987): "Poverty and Sex Ratio*,
Economic and Political Weekly, 22. 23.
892-97
Kynch. J and A K Sen (1983 ): "Indian Women:
Well Being and Survival*. Cambridge Journal
of Economics, 83. 7. 363-80
Miller. B D (1981): The Endangered Sex,
Cornell University Press. Ithaca. N V.
Nath V( 1991): "Official Paper on 1991 Census*.
Economic and Political Weeklev. 24(22),
1229-36.
Sen A K. (1987b): "Africa and India: What do
wc have to Learn from Each Other?. United
Nations University. WIDER Discussion
Paper No 19.
Economic and Political Weekly
available from:
Churchgate Book Stall
Churchgate Station
Opp Indian Merchant Chambers
Churchgate
Bombay-400 020
Economic and Political Week!}
August 19. 1995
PERSPECTIVES
Demographic Outcomes, Economic
Development and Women’s Agency
■
i
I
variations in poverty are small. This is not
implausible, since the NSS regions aie
meant to be relatively homogeneous in terms
of agro-climatic and socio-economic
features?
Jean Dreze
Two further remarks are due concerning
Anne-Catherine Guio
the poverty variable. First, the poverty
Mamta Murthi
estimates calculated by Jain ct al (1988)
relate to rural areas specifically. For want
This paper examines the determinants offertility, child mortality and
of information on the level of poverty in rural
and urban areas combined, we have used
gender bias in child mortality in India using district-level data from the
these estimates, and also included a separate
1981 Census. The findings highlight the powerful effects of variables
independent variable indicating the level of
relating to women's agency (e g. female literacy and female labour force
urbanisation (Table la). Second, the reference
participation) on mortality and fertility. Further, higher levels of female
yeai for this variable is 1972-73 (rather than
1981, as with the other variables). This is
literacy and female labour force participation are associated with
the only year, prior to 1981, for which the
significantly lower levels of female disadvantage in child survival. In
relevant estimates are available. The use of
contrast, variables relating to the general level of development and
1972-73 as the reference year for the poverty
modernisation have relatively weak effects on demographic outcomes.
variable seems legitimate, since the 1981
mortality estimates are based on birth and
death information pertaining to the late 1970s,
INDIA is a country of striking demographic
of this analysis can be found in Guio (1994)
and since poverty levels in different regions
and Murthi. Guio and Dreze (1995)?
diversity. Even broad comparisons between
during that period must have been quite close
different stales within the country bring out
The reference year for this analysis is
to those observed in 1972-73. As a matter
1981. For that year, a fair amount of district
enormous variations tn basic demographic
of fact, the relative position of different
level information is available from the 1981
indicators. At one end of the scale, Kerala
regions in terms of poverty levels does seem
Census and related sources. Table 1 a presents
has demographic features that are more
to be reasonably stable in the short run.
a list of the variables used along with their
typ' al of a middle-income country than of
Replacing the Sen index for 1972-73 with
definitions. The relevant information is
a poor developing economy, including a
the Sen index for 1987-88 (also available for
available for 296 districts, all located in 14
life expectancy at birth of 72 years, an
NSS regions) has little effect on the results
of India’s 15 largest states (these 14 states
infant mortality of only 17 per 1,000 live
presented in this paper?
had a total of 326 districts in 1981, and
births, a total fertility rate below the re
The regression equations were initially
accounted for 94 per cent of the total
placement level (1.8 in 1991), and a female
estimated by ordinary least squares. However,
population of India). The sample averages
male ratio well above unity (1.04 in 1991).
statistical tests indicate that the error terms
of the variables used in the analysis are
At the other end, the large north Indian
are spatially correlated, i e, the error terms
presented inTable i a. while the state averages
stales find themselves in the same league
of adjacent districts are correlated. We
are In Table lb.
as the least developed countries of the world
therefore present estimates based on a more
The regressions presented below may be
in terms of the same indicators. In Uttar
general model in which the errors may be
interpreted as the ‘reduced form’ of a system
Pradesh, for instance, the infant mortality
spatially correlated.7 In this model, the error
of simultaneous equations which detennines
rate is six times as high as in Kerala, the
terms are assumed to have the following
three endogenous variables: the total fertility
total fertility rate is as high as 5.1, and the
structure:
rate (TFR), the level of child mortality for
female-male ratio (879 in 1991) is lower
u = l.W.ti + e
both sexes combined (Q5). and the extent
than that of any country in the world.1
where
Is lhe'/error term for the (th
of female disadvantage inchild survival (FD).
India is alsoacountry of rapiddemographic
observation, k is a scalar measure of the
as measured by the proportionate difference
change. As in many other developing
intensity Of spatial correlation, and W is a
between female and male child mortality (or.
countries, mortality rates in India have
matrix of spatial weights with entry ‘1’ in
significantly declined in recent decades, e
more precisely, by FD = |.Q5J - Q5 )/Q5r
row i and column j if districts i and j arc
g, the infant mortality rate has been reduced
where Q5 and Q5m are the levels of female
adjacent, and ‘O’ otherwise. Estimation is
and male child mortality,respectively). The
by about 50 per cent since 1961. The same
based on the principle of maximum
period has seen a sustained decline in fertility,
other variables listed in Table la are the
likelihood. Interestingly, the choice of
particularly in the south Indian states (in
independent variables.4
model (with or without spatial correlation)
The variable wc have used to measure
Tamil Nadu, for instance, the total fertility
does not affect the broad conclusions of the
’poverty’ requires comment. Poverty
rate declined from 3.5 to 2.2 during the
analysis.
indicators at the district level are not available
1980s). There have also been significant
We found no evidence of non-linearity in
in India. However, Jain el al (1988) have
changes in the relative survival chances of
the relationships, except in the equation for
men and women?
computed 1972-73 poverty measures for the
female disadvantage. Visual examination
National Sample Survey ‘regions’, which
Apart from being of much interest in
and non-paramctric methods suggested that
themselves, these inter-regional and inter
are intermediate units between (he state and
the relationship between female disadvan
temporal variations provide useful •the district. The 296 districts included in
tage and the individual independent variables
opportunities to study the determinants of .our analysis belong to 51 different regions,
follows a logistic pattern. We therefore used
demographic outcomes in India. This paper
and the poverty indicator used for each
a logistic transformation of female dis
district is the 'Sen index’ of poverty for the
is an attempt to examine some of the relevant
advantage (FD) as defined earlier as the
region in which the district in question is
relationships based on a cross-section
dependent variable in that regression
situated. This procedure involves the
analysis of district-level data lor 1981. A
implicit assumption that intraregional
equation.
more detailed presentation and discussion
Economic and Political Weekly
July 6. 1996
1739
u
Basic Rt.sut.Ts; Child Mortality and
Glndlr Bias
Table 2 presents the main regression results.
Apart from indicating the signs of different
coefficients, and whether they arc statistically
significant at the 5 per cent level. Table 2
makes it possible to assess the quantitative
effects of different variables on child
mortality and fertility (by combining the
information given in Table 2 with the mean
values presented in Table la). As far as child
mortality is concerned, the following
observations deserve explicit mention:
(1) Female literacy: Female literacy has
a negative, large and statistically significant
effect on child mortality. Female literacy
also has a negative (and statistically
significant) effect on FD, the extent of
female disadvantage in child survival. The
last result com: asts with the position, taken
by several oil r researchers, according to
which higher i -male literacy may be a tool
of intensified discrimination against female
children."
Il is worth noting that higher female literacy
reduces child mortality, and anti-female bias
in child survival, independently of male
literacy. Male literacy also has a negative
effect on child mortality (independently of
female literacy), but thccffcctof male literacy
is much smaller than that of female literacy,
and is not statistically significant. Male
literacy has a significant effect on the extent
of gender bias in child survival, in the
direction ofcnhancing female disadvantage.9
(2) Female labour force participation:
Female labour force participation has no
statistically significant effect on the absolute
level of child mortality.10 This is plausible,
given the opposite effects of different links
betv/een female labour force participation
and child mortality. For instance, greater
involvement in remunerative employment
gi ves women greater control over household
resources, and this may be expected to have
a positive influence on child health. On the
other side, to the extent that female employ
ment outside the home imposes a ‘double
burden’ on women, and reduces the time
available for child care, it may also have
some negative effects on child survival.
Higher female labour force participation
unambiguously reduces the extent of gender
bias in child survival, and this effect is
statistically significant. There are a number
of possible reasons for this link, including
that higher female labour force participation:
(i) raises the status of women in society, and
therefore the value attached to young girls;
(ii) raises the returns to ‘investment’ in girls;
(iii) lowers dowry levels, and therefore
reduces the costs of bringing up daughters;
(iv) makes women less dependent on adult
sons for security in old age, and therefore
reduces boy preference; (v) raises the
bargaining power of adult women, and their
ability to resist male pressure to discriminate
in favour of boys."
1740
(3) Urbanisation: Urbanisation has a
negative and statistically significant clfcct
on child mortality. The effect on male
mortality is larger than the effect on female
mortality, so that urbanisation is also
associated with higher levels of female
disadvantage in child survival, but this effect
is not statistically significant.
(4) Medical fat ilitits: Medical facilities
have essentially the same effects as
urbanisation: they reduce child mortality,
and amplify the female disadvantage in child
survival, but the last effect is not statistically
significant.
(5) Poverty: As expected, higher levels of
poverty arc associated with higher levels of
child Inortality. Less evidently, there is a
negative and statistically significant
association between poverty and FD, i e,
higher levels of poverty go with lower levels
of female disadvantage in child survival.
This is consistent withtheobservation. made
in a number of studies, that anli-fcmalc
discrimination may be particularly strong
among privileged classes.11
(6) Scheduled tribes: A higher proportion
of ‘scheduled tribes’ in the population reduces
theextent ofanti-female bias in child survival,
and this effect is statistically significant. It
is interesting that this variable has a significant
effect even after controlling for female labour
force participation (which is generally higher
among scheduled tribes than in the population
as a whole).This suggests that tribal societies
have other features that enhance the relative
survival chances of female children.
Examples of possibly relevant features are
kinship systems and property rights.
It is also worth noting that the absolute
level of child mortality seems to be relatively
low in districts with a high proportion of
scheduled tribes, even after controlling for
poverty and literacy. This is consistent with
the common notion that tribal lifestyles have
some healthy aspects (e g. relatively low
levels of crowding and pollution). But the
precise basis of this statistical association
requires further investigation.
(7) Scheduledcasirs:'V\\c\c is no significant
association between the proportion of
scheduled castes in the population and the
Definition
TER
Q5
FD
Female literacy
Male literacy
Female labour
force participation
Urbanisation
Poverty
Medical facilities
Scheduled castes
Scheduled tribes
South
r
Total fertility rate. 1981
Under-five mortality rate. 1981: probability that
a child will die before iIk fifth birthday (x I.(XX))
Female disadvantage in child survival.
1981. defined as FD=(Q5( - Q5bi)/Q5( (per cent)
Crude female literacy rate. 1981 (per cent)
Crude male literacy rate. 1981 (percent)
Proportion of “main workers” in the
female population. 1981 (percent)
Proportion of tire population living in
urban areas. 1981 (per cent)
Sen index of rural poverty. 1972-73. for lire
"region” in which the district is situated (x I(X))
Proportion of villages with some medical
facilities (per cent)
Proportion of scheduled<a.sle persons in
the population. 1981 (per cent)
Proportion of scheduled-tribe persons in
the population. 1981 (percent)
Dummy variable, with value 1 for
Mean
Standard
Deviation
5.02
0.95
156.91
42.84
5 36
22 08
44 77
10 74
1371
12.20
14 S4
10 49
19.81
12 02
i 7.60
8.50
21.36
20.50
16.01
6.95
8.04
1351
0.23
0.42
West
0.16
0.37
0.14
0 35
Kerala and Tamil Nadu
Dummy variable, with value I for
distm in Bihar. Orissa and West Bengal
Dummy variable, uith value I for
districts m Gujarat and Maharashtra
Sources. See Drvze and Munhi (forthcoming) Most of the information is derived from 1981 Census
data TN? Sen index of rural poverty is taken from Jam et al (1988).
Economic and Political Weekly
bet\
ca
fen
pa
in ti
vicpre
cco
an
ink
IM
n
th'(1.
lor
st
Fci
sc
prt
ot‘
sh;
ar
let
g*ra
d
(s
I
st
i
k
IV
g
^•'•ricts in Andhra Pradesh. Karnataka.
East
i(
extent of female disadvantage in child
survival. This is in line with the fact that
female-male ratios among scheduled castes,
which used to be higher than average, arc
now very similar to those of the population
as a whole.*' In contrast with the
corresponding finding for scheduled tribes,
child mortality levels among scheduled castes j
appear to be comparatively high, even after
controlling for poverty and literacy (but this
association is not statistically significant).
(8) Regional 'dummies':1* Even alter
controlling for the other variables, the .
southern region has considerably lower levels
of child mortality. This is particularly the
case for git Is. so much so that female children
have a survival advantage over boys in that
region (Tabic lb). In both respects (child
mortality and gender bias), the contrast
between the southern region and (he rest of
the country’ is statistically significant.
The particular demographic features of
south India, including the relatively
favourable survival chances of female
children, have been much discussed in the
Txble la: Variable Dehmtions and Samh e Summary Statistics
Variable
lik”
2 i
July 6. 1996
ir
r
b
d
o
o
I
between the level of development and redu
ced gender bias in survival, it seems to work
through variables that are directly ^elated to
women’s agency, such as female literacy and
female labour force participation."’
Similarly, w hile indicators df development
such as male literacy, reduced poverty.urbani
sation and the spread of medical facilities
do have positive effects on absolute levels
of child survival, these effects are relatively
small compared with the powerful effect of
female literacy. This point is illustrated in
Table 3, which indicates how the predicted
values of Q5 and FD respond to changes in
female literacy when the other variables are
kept at their mean value (and similarly with
male literacy and poverty). It can be seen
that the influence of female literacy on child
mortality is quite large, in comparison with
literature. The findings presented in Table
2 suggest that the demographic contrast
between south India and the rest of the country
cannot be explained entirely in terms ot
female literacy, female labour force
participation, and other variables included
in the regression. This is consistent with the
view that differences in kinship systems,
property rights, and related features of the
economy and society not captured in this
analysis (for lack of adequate statistical
information), play an important role in this
north-south contrast."
Table 2 includes further results relating to
the determinants of the total fertility rate
(TFR). Female literacy and female labour
force participation have a negative and
statistically significant effect on TFR.
Fertility is also significantly lower in the
southern region, and in districts with a high
proportion of scheduled tribes. None of the
other variables is statistically significant.
Discussion
The findings summarised in this note
sharply bring out the role of women's agency
and empowerment in reducing mortality,
fertility and gender inequality.
Consider, for instance, the determinants of
gender bias in child mortality rales. Il is
rather striking that, while the variables
directly relating to women's agency
(specifically, the female literacy rate and
female labour force participation) have a
strong md statistically significant negative
impact on FD, those relating to the general
level of economic development and
modernisation in the society as a whole (e
g, poverty, urbanisation, male literacy and
medical facilities) do nothing to improve the
relative survival chances of girls vis-a-vis
boys. In fact, to the extent that these variables
do have a statistically significant influence
on female disadvantage, this influence turns
out to go in the ‘wrong’ direction in each
case, i c. higher levels of male literacy and
lower levels of poverty are both associated
with a larger female disadvantage. Insofar
as a positive connection docs exist in India
that of male literacy or poverty.
The same point also emerges in connection
with the determinants of fertility. In fact, in
this case, none of the variables relating to
the general level of development and
modernisation is statistically significant. By
contrast, female literacy and female labour
force participation appear to be crucial
determinants of the total fertility rate. As
shown inTablc 3. for instance, female literacy
alone is a considerable force in reducing
fertility. Here again, the message seems to
be that some variables relating to women’s
agency (in this case, female literacy) often
play a much more important role in
demographic outcomes than variables
relating to the general level of development.
1 The figures cited in this paragraph are taken
from Drcze and Sen (1995), statistical
appendix, and are based on census and sample
registration system data. A few countries ot
west Asia (e g, Kuwait and the United Arab
Emirates) actually have a lower female-male
ratio than Uttar Pradesh, but this is due to
exceptionally high levels of male in-migration.
2 On the latter, see e g, Dyson (1988).
3 For related analyses based on Indian district
data, see Rosenzwcig and Schultz (1982);
Gulati (I992); Kishor (1993); Khemani (1994).
4 We have also examined the effects of other
independent variables, c g. relating to the
structure of economic activity. But (Im: variables
Table 2: Maximum Likelihood Estimates
Dependent Variable
Independent •_ .*■
. FD
Q5
TFR
Variable
Constant
Female
literacy
Male literacy
Female
labour force
participation
Urbanisation
Medical
facilities
Poverty
Scheduled
castes
Scheduled
tribes
South
East
Notes
(This paper is based on a more extensive analysis
presented in Guio (1994) and Murthi. Guio and
Drcze (1995) We arc grateful to Sansh Agnihotri.
Sudhir Anand. Peter Boone. Jean-Marie Baland.
Monica Das Gupta. Angus Deaton. Tim Dyson.
Haris Gazdar. Stuti Khemani. Sunita Kishor. P
N Mari Bhat. Jean-Philippe Platteau. Rohini
Somanathan and P V Srinivasan for helpful
discussions and comments. Th is collaborative work
was Completed under the Economic Security
Programme of the Centre for Development
Economics.]
West
0.857
(3.00)*
-0.036
(-4.46)*
0.015
(1.97)*
-0.020
(-3.85)*
0.005
(1.73)
0.005
(1.84)
-0.021
(-3.13)*
-0.007
(-1.13)
-0.014
(-3.96)*
’ -0.820
(-4.91)*
, 0.154
(0.81)
-0.148
(-0.87)
205.822
(14.37)*
-0.873
(-2.45)*
-0.489
(-1.40)
6.594
(23.10)’
-0.031
(—4.28)*
-0.005
(-0.70)
-0.017
0.440
(-3.57)’
(1.82)
-0.310 -3.9E-O4
(-2.40)* (-0.15)
-O.(X)2
-0.246
(-2.23)* (-1.04)
0.007
0.535
(1.14)
(1.76)
-0.007
0.548
(1.89)
(-1.23)
-0.01 I
-0.598
(-3.57)* (-3.40)*
-0.548
-41.504
(2.60)*
(-3.85)*
-0.254
-38.0X0
(-2.91)* (-0.99)
-0.379
-12.245
(-2.06)*
(-1.32)
0.610
(11.00)*
0.836
(28.O7)*
0.821
(25.95)*
Mean
0.39
squared error
0.81
Adjusted R:
Log likelihood -190.80
Sample size
296
15.15
0.87
-1310.26
296
0.31
0.89
-155.95
296
X
Notes: Asymptotic t—ratios in brackets..
* Significant at 5 per cent level.
Tabic lb: State-Lesu Average of the Regression Variables
Andhra Pradesh
Bihar
Gujarat
I iaryana
Karnataka
Kerala
Madhya Pradesh
Maharashtra
Orissa
l'un|ah
Rajasthan
Tamil Nadu
llitai I’nulcsh
West Ben cal
Tl-R
Q5
H)
Female
Literacy
Male
Literacy
4.35
5 24
4.80
5 40
4 6X
3.40
5 57
4.34
4 81
3 26
6 05
192
5 HU
4 57
138.6
141.1
126 I
139 0
142 3
81.2
202.9
155.7
175 7
110 6
174 6
126 X
IK'S 6
1210
-6.?
14.4
62
17 5
-3 4
-10 5
4 4
194
13 4
30 9
21 5
27 I
66 0
14 5
31 8
18 9
33 4
10 5
35 7
14 7
28 2
38.4
37 6
53.1
4X 0
4X 0
754
3X 5
56 4
44 9
47 4
34 4
5X s
50 2
46 6
Economic and Political Week I v
10 6
9X
2x
15 <
1.0
Julv 6. 1996
Female Urbanisation Medical
Facilities
Labour
Force Parti
cipation
275
8.6
107
4 5
19 9
13 I
20 3
26 2
11 8
24
96
22 7
KO
7.1
22 8
11.6
28 2
214
24 5
17.9
19 6
26 2
11 6
26 7
19 2
12 3
17 3
23 3
25.9
18.1
28.2
58.2
13.4
95.8
5.8
18.3
10.8
26 8
16.7
32 6
11 X
15.2
Poverty
15.8
24.8
15.5
3.7
14.5
20.9
19.3
25.1
37.8
3.8
13.2
17.6
13.0.
28.4
Scheduled Scheduled
Caste
Tribe
15.0
14.9
7.4
18.9
14.2
10.4
14.9
7.3
14.2
26.7
16.7
17.6
20 X
22.9
6.4
1.8
11.0
0.0
5.1
0.9
21.1
10.1
24.9
0.0
14 2
1.1
0.5
7.2
1741
T|
for which data were available, other than those
included in Tabic la. were found to have no
significant effect on mortality, fertility or
gender bias; nor does their inclusion affect the
basic results presented in this appendix.
5 An alternative approach is to carry out the
entire analysis at the level of ‘regions' rather
than that of districts. This approach has tiie
advantage that it involves an accurate poverty
indicator for each observation, but reducing
the number of observations from 296 to 51
also entails a major loss of information. As
it turns out, the main results obtained under
this alternative approach are similar to those
obtained on the basis of district-level analysis.
In this appendix, we present the district-level
results; the region-level results can be found
in Murthi. Guio and Dreze (1995).
6 To our knowledge, 1972-73 and 1987-88 are
the only two years for which poveny indicators
have been calculated for the NSS regions.
7 For further details of this approach, see Anselin
(1988). The method of estimation is fully
described in Murthi. Guio and Dreze (1995).
For a similar application of this method see
Kishor (1993).
8 Sec c g, Das Gupta (1987); Amin (1990): Basu
(1992), Gupta ct al (1993); for the opposite
view, see Caldwell ct al (1989) and Bourne
and Walker (1991).
9 Interestingly, tlie last statement remains true
even if female literacy is dropped from the
regression.
10 In an earlier analysis of 1981 district data,
Sunita Kishor (1993) found that female labour
force participation has a positive and
statistically significant effect on both female
and male child mortality. The contrast bet
ween that result and our own may be due to
the fact ’hat, in the analysis presented here,
the levels of poverty and female literacy are
included as explanatory variables, indeed,
when examining the effects of female labour
force participation on child mortality, it is
important to control for the economic and
social disadvantages that motivate many
women to seek employment. For further
discussion of these issues, see Guio (1994) and
Murthi. Guio and Dreze (1995).
11 Some of these hypotheses have been discussed
by Miller (1981); Roscnzweig and Schultz
(1982); Drezc and Sen (1989); Kishor (1993),
among many others. For reviews of these and
other studies, see Guio (1994) and Kishor
(1994).
12 For some relevant studies, see Miller (1981
1993); Das Gupta (1987); Krishnaji (1987/
Basu (1992); Dasgupta (1993).
13 Sec Agnihotri (1994) and Dreze and Sen (1995).
chapter 7.
14 In the regressions presented in Table 2. the
control region is northern India, consisting
of Haryana. Madhya Pradesh. Punjab.
Rajasthan and Uttar Pradesh.
15 On these different influences, sec the studies
cited in Guio (1994) and Drczc and Sen (1995)also Alaka Basu (1992); Sunita Kishor (1993)’
Satish Agnihotri (1994) and Bina Agarwal
(1994) among other recent contributions. The
persistence of regional influences on relative
survival chances, even after controlling for a
wide range of district characteristics on which
quantitative dara are available, has been noted
earlier by Sunita Kishor (1993).
16 In the light of these findings, the decline of
India’s female-male ratio since 1901 (on which
sec Drezc and Sen 1995. and the literature
cited there) may not be much of a mystery.
There has been much progress, in the
intervening years, in terms of general
development, but comparatively little
expansion of u omen’s agency. There is little
evidence, for instance, of asubstantial increase
in female labour force participation over time,
and while female literacy has slowly increased,
the crude female literacy rate remained as low
as 22 per cent in 1981. The fact that, taken
together, these different developments have
gone hand in hand with a decline in the female
male ratio is quite consistent with the cross
section findings summarised in this paper.
References
Agarwal. Bina (1994): A Field of One’s Own:
Gender and Land Rights in South Asia,
Cambridge University Press. Cambridge
Agmhotn. Satish (1994): ‘Missing Females: A
Disaggregated Analysis’ (mimeo) University
of East Anglia; (forthcoming in Economic and
Political Weekly).
Amin. S (1990): ’The Effect of Women’s Status
on Sex Differentials in Infant and Child
Mortality in South Asia’. Genus. 46.
Ansclin, L (1988): Spatial Econometrics: Methods
and Models Kluwer Academic Publishers
Netherlands.
]Basu. Alaka Mai wade (1992): Culture, the Status
of Women and Demographic Behaviour.
Clarendon Press. Oxford.
Table
""r 3:
': EFr;CCTS 0F Selected Independent Vaxubles (Female LrTE«ACv. Male Litehacy and
______ 2 hRTY) 0N Child Mortality (Q5). Female Disadvantage (FD) and Fertiuty (TFR)
Assume!
Predicted Values
L^evel oi
of Q5, FD and
independent
1FR, when the Female
Variable
- Literacy Rate Takes the
(Percentage)
Value Indicated in the
First Column
Q5
FD
TFR
10
•20
30
40
50
60
70
80
Note:
166.4
157.7
149.0
140.2
131.5
122.8
114.0
105.3
lol
5.9
1.1
-3.34
-7.1
-10.3
-12.8
-14.8 '
5.38
5.07
4.76
4.45
4.15
3.84
3.53
3.22
Predicted Values
Predicted Values of
of Q5, FD and
Q5. FD and TFR, when the
TFR, when the Male
Proportion of the Population
Literacy Rate Takes thq
Below the Poverty Line
Value Indicated in the*
Takes the Value Indicated
First Column
in the First Column@
Q5
FD
TFR
Q5
FD
TFR
172.9
168.0
163.1
158.2
153.3
148 4
143.5
138.7
-0.1
1.8
3.9
5.9
8.0
10J
12.2
5.18
5.13
5 08
5.03
498
4 93
4.88
4.83
751 5
152.7
153.8
1549
1560
157.2
158.3
159.5
Ts
8.5
7.1
5.8
4.4
3.1
18
0.5
4.79
4.85
4.91
4.97
5 03
509
5.15
5.21
a For convenience of interpretation, the "Sen index” has been
i replaced, in this table, by the ’’headcount ratio” (■
~ ,u
'
(i e.
the----------proportion
of the population below the poxerty line) The hgi
gures presented
in these columns
are hncrdnnlfvrnrn»r.r.~,,:
based on the same regression?
inin.surf
_______
-r- . . as in Table 3. w ith the Sen index replaced by
the head-count ratio.
Bourne. K and G M Walker (1991): 'The
Differential Effect of Mothers’ Education on
Mortality of Boys and Girls in India’.
Population Studies, 45.
Caldwell, J C. R H Reddy and P Caldwell (1989)
The Causes of Demoftraphit Change,
University of Wisconsin Press. Madison.
Das Gupta. Monica (1987): •Selective
Discrimination against Female Children in
Rural Punjab . Population arid Development
Review, 13.
Das Gupta. Monica. T N Krishnan, and Lincoln
Chen (cds) (1994): Women‘s Health in India:
RHkandVulnerabiliry.OxfordUmvcrilypKss.
Mumbai.
Dasgupla. Partha (1993): An Inquiry into WellBeinx and Destitution, Clarendon Press.
Oxford.
Drczc. Jean and Amartya Sent 1989):. lunger and
Public Action. Clarendon Press. Oxford.
- (1995): India: Economic Development and
Social Opportunity. Oxford University Press.
New Delhi and Oxford.
Dyson. Tim (1988): ’Excess Female Mortality in
India. Uncertain Evidence on a Narrowing
Differential’ in K Srinivasan and S Mukeiji
(eds). Dynamics of Population and Family
Welfare 1987, Himalaya, Mumbai.
Dyson.Tim and Mick Moore (1983): On Kinship
Structure. Female Autonomy
and
Demographic Behaviour in India’. Population
and Development Review. 9.
Guio. Anne-Catherine (1994): ‘Aspects du Sex
Ratio cn Inde’. (unpublished MSc thesis).
Univcrsite de Namur. Belgium.
Gulati.SC( 1992): •Developmental Determinants
of Demographic Variables in India: A District
Level Analysis . Journal of Quantitative
Economics, 8(1). pp 157-72.
Gupta. D B. A Basu, and R Asthana (1993);
•PopulationChange, Women's RoleandStatus.
and Development in India: A Review1 (mimeo).
Institute of Economic Growth. Delhi
University.
Jain. L R. K Sundaram, and S DTendulkar (1988):
•Dimensions of Rural Poverty: An InterRegional Profile’. Economic and Political
Weekly, November (special issue); reprinted
in Krishnaswamy (1990).
Khcmani. Stuti (1994): -Neoclassical vs Nashbargained Model of Household Fertility:
Evidence from Rural India’ (undergraduate
thesis). Department of Economics. Mount
Holyoke College. US.
Kishor. Sunita (1993): * ’May God Give Sons to
AH’: Gender and Child Mortality in India’.
American Sociotof’ical Review. 58
-(1994):’Gender Differentials in Child Mortality:
A Review of (he Evidence’ in Das Gupta et
al.
Knshnaji.N(l987): ’Povertyand Sex Ratio: Some
Data and Speculations . Economic and
Political Weekly. June 6.
Miller. Barbara (1981): The Endangered Sex.
Cornell University Press. Ithaca.
Miller. Barbara D (1993): 'On Poverty. Child
Survival and Gender: Models and
Misperceptions’. Third World Planning
Review. 15.
Murthi. Mamta. Anne-Catherine Guio. and Jean
Drczc (1995): Mortality, Fertility and Gender
Bias in India’. Discussion Paper No 61.
Development
Economics
Research
Programme. STICERD. London School of
Economics.
Roscnzweig. Mark R. and T Paul Schultz (1982):
Market Opportunities. Genetic Endowments.
and Intrafamily Resource Distribution: Child
Survival in Rural India'. Ameiican Economic
Review. 72.
1742
Economic and Political Weekly
July 6. 1996
!
I
I
/
:<■
i
SPECIAL ARTICLES
Juvenile Sex Ratios in India
A Disaggregated Analysis
S /ignihotri
Tins paper studies the regional variations in sex ratio patterns in India in the juvenile age group (0-9 years)
d'e ero^
7aj°r adVa'"aSeS'he^ni,e
^e not affected by migration and
efore. (b) their analysts provides useful insights into patterns of differential mortality among children by sev
A disaggregation of these sex ratios into the 0-4 and 5-9 age groups brings into sharper focus the pattern of
excess female mortality beyond the age of one year. This is a socially driven phenomenon as against excess infant
ma e morta Uy »• uch is essentially a biological phenomenon. The female to male ratio (FMR) in the 5-9 age
s^ZmK
4ffaS an appro!’r>a,e parame,erf°r ana,y^
ratio variations across the country. It reveals
cas es and theZ^f ‘th
Z
grOUpS- Viz’ ",e ,ribal’ ,he Muled
regions
f ,hyOpula,,On- " d,sPlays a remarkable spatial contiguity across different geophysical
regions of the country. These regions turn out to be a more suitable unit for analysis of spatial variations in
the sex ratios than the administrative units of different Indian states. Certain regions, cutting across the state
boundaries, stand out for their alarmingly low FMRs. The observed patterns raise important questions about
I
■
SEX ratios in India are highly masculine
compared to most other regions in the world.
The proportion of females in its population
continues todecline and stands at 927 females
per thousand men in the 1991 population
Census. The female male ratios (FMRs)?
however, are neither uniformly low nor
uniformly declining across different regions
in the country and show a considerable
variation across these regions. These
variations have attracted considerable
attention in literature, dominated by one
major feature; the ‘north south’ divide?This
refers to the highly masculine sex ratios in
the north-western states and more favourable
FMRs in the south-eastern states of India.
Concern has also been expressed at the
growing masculinisation of the sex ratios in
the south-eastern states over t he years (M il ler
1989; Heyer 1992].
Apart from recognition of this ‘divide’,
different correlates of these sex ratios have
also been analysed; economic as well as
socio-cultural. The economic factors have
mainly been analysed in terms of female
labour participation asa detenninant of female
‘worth’ [Bardhan 1974; Miller 1981;
Rosenzweig and Schultz 1982 and Meis
1988], Recently, the role of capital in terms
of dowry has also been examined in greater
detail [Heyer 1992; Rao 1993; Wadley 1993;“
Kapadia 1994],
Studies of the role of socio-cultural factors
has mainly focused upon the status of women
shaped by culture [Dyson and Moore 1983;
Economic and Political Weekly
Dasgupta 1987-. Berreman 1993; Madan the rest of the population have also not been
1993]. Different kinship systems in the north systematically analysed. Scparatcanalysis of
and the south and the process of assimilation the 0-4 and 5-9 age group FMRs has also
of a woman into the family of her marriage not been done earlier.
have informed the bulk of this analysis which
We use the 1981 Census data, which has
has by and large been qualitative. In the for the first time provided the five year ace
absence of suitable quantitative data and group break up for the scheduled tribe and
analysis, the debate on the cultural aspects the scheduled caste population. We use the
has not moved much beyond highlighting data in respect of 355 districts based on the
the north-south divide. Given India’s ?ultural Indian District Development Database6 and
diversity, this is not adequate?
the special tables for scheduled castes and
This paper elaborates upon this regional scheduled tribes from the 1981 Census
diversity. It departs from the conventional reports. As random fluctuations in sex ratios
analyses on four counts. It uses district level are large for small population sizes? we
data. Further, it usesjuvenile sex ratios (JSRs) have not considered districts with very lowinstead of overall population sqx ratios to population of tribal or scheduled castes in
cut down the migration ‘noise’. It then the analysis of the SC or ST FMRs. Districts
disaggregates these among three major social in the north-eastern region have also been
groups - the tribal, the scheduled castes and excluded from the analysis? Data from the
the rest of the population. The sex ratio 1991 Census for the five year age groups
patterns among these three groups differ is yet to become available.
significantly - a point which has been
Using these data, we (i) show that the
consistently overlooked in the literature.4 JFMRs arc free from migration noise’which
Finally, the juvenile EMRs are broken down affects the all age group FMRs, (ii) bring
into 0-4 age group and^-9'age group FMRs out significant differences in the FMR
to capture the differences in the mortality patterns among the ST, SC and the rest of
patterns in thejuvenilfe-age group. The effect the population and (iii) show that the 0-4 and
of excess male mortality during infancy gets 5-9age group FMRs differ significantly from
rcOccted in the 0-4 age group FMRs while each other. The 0-4 age group FMRs capture
the 5-9 age group FMRs capture the excess the excess male infant mortality, essentially
girl child mortality in later years of the a biological phenomenon, whEe the 5-9 age
childhood.
group FMRs capture the patterns of excess
The received literature uses state-level sex female mortality that set in during latcryears
ratio data for much of its analysis? It also •of childhood.’These reflect the different
uses the all age groupsex ratios.not corrected extents to which the biological advantage of
for migration. Differences in sex ratio patterns the female infant is ‘reversed’ beyond the
among the tribal, the scheduled castes and age of one year in different regions mainly
December 28, 1996
3369
."T
Figure 1: Comparwo FMR Variances
AH age group and juvenile age group
s
A
1 , 80 4
a
AimI'
All
Districts
. •
■’ <
n
4=71 —
d
a 60 Districts with Tribals above 5 per cent
r
d
J'
r
Kerala
D 40 ~
e
v ’
a - 20 ■ t A t-
n
i
,
.
ok_
.J A11
gM EnB
NT
ST^^. All
ggp
^(T “
All
7. Different social groups
■|0^ age group
: O All age
^Not-Tribal
ST-Tribal
and below (t-value of 7.93) as revealed by>;; S
l-test for paired samples. That JFMRs are’? ’
fc
free from migration ‘noise’ can also be
H
mfetredIbyanalysing themtanand thereral
1
FMRs. The low mean for the urban FMRs;
■ rW females pcr thousand males compared
■ L
to a mean of 946 females in rural areas is
consistent with net excess male inmigration
in urban areas. The difference between these
sex ratios; urban and rural, is highly
significant at 1 per cent level and below (tvalue of 16.3).
k If sex selecti ve migration between urban
and rural areas is and
insignificant
in the 0-9
age
^uMh^rural
urban JFMRS
"shoX
group, the rural and urhan iehdc __ , . ,
not differ significantly.
This is indeed the
—
"J ' -••••* «w ■■BWWM U1C
case.
The mean for
the JFMR for the overall ': ■ < Tcase.Thcmean
fortheJFMR
population
___■ ..
. the
.
■';!
population in
in nir~>I
rural area nrr
955 and
that. in
urban areas.
uirt>an
arc'*. 958.
958, are close to each other and S
7 ''
• ' L|
mediircrencebetween
_ ____ 7
J: v |I
^cdiffcrence
between them is not significant
at 5 per cent level (t-value of-1.76).
NT
<
err,
Mipratinn riafo
. ’ ■V.77’5-Ktt
»k_ i_ j;_
iccount of marriage, most of the
provide
composition
of the
1
The FMRs in the 5-9 age group'show a Sinard r*™
011 “— hCavUy male
nU,C poPula
^n intheterms
ofYocfal
k migrant
~ '
--------------1
groups.
A
remarkable homogeneity within different reasons’^High ^Ll'e1’ Of.CCO.nOnVc ^P^nof all age FMRs and ^cJFMlQ
-----------------_
' -?LX
ecological or geophysical rerrion< Th„. ,'^>“ns. .High net male inmigration m r^
-------8-,l?CSC
groups’ c
tribalenvucs
or the, 7 ‘
metropolitan
areas,
net
male
outmigration
scheduled
casS^m^sl/h^
regions sometimes cur across the boundaries frem Ke^aT5’-- — °Ulmieralion ^hedulcd
s castrS
can throw
interestingamong
light .-tte£
’- .'K'
> or !the
dominated
on the
the - nature
nature
of
of different states and a state often contains inmiL,
,,c ,male
"a,e oon
>'nalei
on
of migration
migration among
these ’7 ®t^eX^ns^o^
ofsoCTrh^^Z^
communit.es.
of some of the no^iere^K
here further.
’
■
satisfactory basis of spatial analysis than^he out-Thisresultsinconsiderablevariationin
the
FMRs
at
the
district
level,
e
g,
772
for
IH
states. One immediate outcome of focus on
O^-ig-tioneffectsareeliminated, sex '•
E;'
these regions is the identification of a triOb^SOrKCraIaan<1-558fornonBermuda Triangle’ for the female child;’ -
s
western UP, three districts of Rajasthan and’
the ra^nes ofMP with alarmingly low FMRs.
This has important implications for the
woman and child welfare policies and the
design of interventions.
The paper is organised in five sections.
JFMR11 1 cxamincs lhc suitability.of the
ttassss
for Bombay; 934, compared to the all age
distncts of Kerala (Table 1) with (he all age
FMRs not corrected for migration. Both
Sopher (1980) and Miller (1981) have used
the JSRs in their analyses of the
Of J?Rs has nevertheless not gained
subsc<lucnl analyses and some
com^nut^
varialions sliu
migS-3
™
°le played
P'ayCd by
continues
around
the rrole
migration.’‘
3
TablelestablishesthisfeaturcoftheJFMR
in further detail in the next section, primaril^ a^grou^FM^"
0^
through comparison
of variance for the all
;n respect of the 5.9 age
agegroup
groupFMRs
FMRs_and
andjuvenile
juvenileFMRs
FMRsatatall
all
Usefulness of grouping by geophysical India level. Variance of the JFMRs is
regions over the states as units of spatial considerably less compared to that in
analysis is established. Final Section the all age group FMRs for the total
summonses the implications and the scope population as well as its tribal and nontnbal segments. Districts with significant
tor further research.
tnbal population, i e, above 5 per cent of
H
.4^7. '
the overall [xipulaiion yield similar result,
The Migration Factor ?!•.
inc reduction is much sharper for the
Analysis of regional variations in sex ratio state of Kerala known for its pattern of
patterns in
in India
India’ has
has often
often’relied
reli^d‘upon
u^n data
data
^a.,effoulmiSralion (Figure 1).
patterns
not corrected for
,hc 3,1 a8c
d°'
f°r migration
miErati0''and continues
continues too tn™ qia7?°°
into 0-4 and 5-9 age group FMRs a i
examines their broad spatial variation acro«
different regions. Differences in the FMR
aPndrhrnSam0?g?etribal’,hescheduled“s'e
'
I
In India a
i
■ 'l I
In India, a substantial portion; nearly 60
per cent, of the deaths in the 0-9 age group
occur during infancy, i e. in the 0-1 age
group. Deaths in the 1-4 age group account
c e abovc 30 Pcr ccnt whereas deaths
.r», .
»
Different Mortality Patterns
There «
is however,
however, an
an important
important difference
difference
‘“^TOofthesedeathUJuringSy
- <•
a
District
All Age FMR 0-9 Age FMR
1034 '
974 T*
949
979
1020
976
1052
964
1056
986
1100
97]
998
967
there is always
''
■-•J
Tabu: 1: ^'“son or Au. Ace and 0-9 Ace
Grovr FMRs „ ^7° ™
Cannanore
Wayanad
Kozhikode
Malappuram
Palghal
Trichur
Emakulam
Idukki
Kottayam
-Quilon
Trivandrum
963
1001
99]
969
1050
1026
1030
977
97]
977
r...
c.
•
3370
Economic and Political Weekly
•/
December 28, 1996
■
Figure 2a: 0-4 FMRs for SC Popitation
11
t- ■
' -{/
J
'
■
■
_
...
....................■
■‘v.
1 •• -
Ji
•
;
■
•
-
■
-v‘i
1
J
f, -.-v.
SCD4FMR
■■■
.
<■
no data
701-960
■
>960
J :'7.:
■'
■
'
'
i-
r ■'
I71.
J •.
mortality.This is due to inherent vulnerability
onhe malejnfant compared to the female
infant in a health neutral environment [see
among others Miller 1989; Caldwells 1990;
Mukherjee 1986; Government of India
1988b). This physiological advantage ofthe
female child (gets
’rtc rpvprtiM
reversed from the age of
one onwards and by the age of 5 excess
female mortality becomes the norm. It is
marginal in some areas, significant
some
___ __in___
others and substantial in yet others. District
level estimates of the probability of death
based on the 1981 Census data [Government
of India 1988b) bear this out. The ql and
q2 tables, which indicate the probability of
a child dying
by “
the
first
and nil,
the 3U.UUU
second year
J
•V in
JI auu
yc<u
respectively, indicate excess male mortality.
At q3, i e, by the age of three years, the
survival chances for both the sexes are nearly
balanced while at q5 level the girl children
• show a higher mortality.
The strong reversal ofthe mortality pattern
after infancy in south Asia is well recognised
and attributed to the differential care of the
girl child or to put it more plainly to her
access inequality to food, nutrition and care
including health care compared to the male
Economic and Political Weekly
child?4 This is essentially a socio-cultural
process linked with the perceptions about the
members of the household whereas excess
female mortality is more often than not
attributed to such a discrimination. To that
extent one can term the excess male child
mortality as natural and the excess female
child mortality as socio-cultural. The former
is driven by exposure to health risks and
possibly the absence of effective health care,
whilethelatter is driven by access inequalities
described above.
Juvenile sex ratio aggregates both these
mortality patterns It is a combined ratio of
10 single year age cohorts. The first two
cohorts among these will reflect the pattern
ofexcess male mortality during infancy while
the subsequent eight cohorts will reflect the
reversal of this pattern to various degrees.
If we were able to disaggregate the JFMRs
into two age groups; the 0-2 age’group and
lhc
aSc £rouP- some significant
differences could be anticipated between the
two..'.-;.;;
'
The FMR in the 0-2 age group (0-2FMR
henceforth), would show preponderance of
female children compared to those in the 39 age group (3-9 FMRs). Further, as the excess
male mortality during infancy is a biological
phenomenon, it will be distributed randomly
across different regions and there should not
be any significant spatial variation.’6
The3-9FMRs on the other hand will reflect
the effects of discrimination against the girl
children driven by the socio-cultural
practices. These practices vary from region
to region in their extent and nature. As such
the 3-9FMRs can be expected to display a
less random regional distribution. We can
in fact expect these FMRs to be significantly
contiguous within different socio-cultural
regions and significantly different across
these, regions.
Such differences will also be observed
across different social groups, e g, the tribals,
the scheduled castes and the rest of the
population which differ in terms of the
position of women in their society. It will
role of the sons and the daughters, and reflects
the ‘son preference’?’ its extent and its
_operational
r
|
consequences.
As a matter of
fact excess male mortality is never ascribed
__ :_ :_____ ____
• ...
>
to any discrimination
against
the male
.
;
TablsI: Analysis of Variance of MFRs. 1981
.
(Femalepgr ^ooo malg)
Variable
All FMR
NTFMR
All Age Qroup FMRs
Mean Value Variable
SldDev-
936
934
4095
4034
’ ‘’
‘
. j . .
Juvenile Age Group FMRs
Mean Value Variance
Std Dev
64 -W*' 958
64
’ 957
’31 ■ Cs-
Districts with tribal population above I per cent of the total population
ST FMR
963
1751
42
y: 988
Districts with tribal population above 5 per cent of the total population
All FMR
949
.2140
46
981 ..
NTFMR
942
2024
45979
3,^:'
ST
FMRR
STFM
974
1548
39
993
Kerala
All FMR
1023
1730
42
975
NTFMR
1022
1920
44
975
1678
1835
355 districts
41
43
1068
. 189 districts
33
125 districts
976 5
31
1534 .
39
903
’ 30
12 districts
8
8
Note: Variable description: All FMR - FMR for total population; NTFMR - FMR for non-tribal
population; and STFMR - FMR for all age group and tribal population. ;
;•
December 28. 1996
-
59
62
3371
/
riwBamiiiw^
Figure 2b: 04FMRs for Non-ST/SC Population
I
0TH04FMR
I no data
700-960
>960
be worthwhile to explore therefore, the
differences in the FMR patterns between
these two age groups; 0-2 and 3-9 years,
across different regions and social groups’
Unfortunately, the suitable census data
available lor this purpose is the 5-ycar age
group data at the district level. Even this
break-up has become available separately
for the first time for the tribals and the
scheduled castes
castes in
in me
the iy»i
1981 Census
Census As
As a
result one can only use the 0-4 and 5 9 vear
age group data for working out the FMRs
among these age groups
Wc could nonctheicss examine th. \ '
ratio patterns among the 0-1 age group and
the 5-9 age group. The differences in the
04FMRs and the 59FMRs may not be'as
sharp as those between the 02FMRs and the
39FMRs. Yet, valuable insights may be
obtained through such an exploration. The
04FMRs would contain two additional single
age group cohorts in addition to the three
in the 02FMR. Both these will carry the
effects of excess female mortality that sets-
?Se °f lhree years- As such the
04FMR values will be lower than those for
the 0-2 age group and the randomness in the
spatial variation may also be less.” But this
blurring’ may get compensated in the
patterns that 59FMRs reveal. The five single
year age group cohorts that contribute to the
59FMR will reflect the mortality pattern
from the age of 5 and above; a stage where
the pattern of excess female mortality has
already‘stabilised’. Asaresultwemayexpcct
tO
signif,candy lowcr
d ff
l° refleCt lhc rcgionaI
real^a^rWewillsceif,his
n r
’
BefQrcanalysmg these differences in detail,
FMR^fnrVff diSlriclJC'xl n,aPs of lhesc
*
• r;-d,^ercnl1
groups. These
of l^varialions in
the sex ratio patterns across different regions
for the two age groups and the three social
groups. Figures 2a and 2b show the 04FMRs
for the SC and the non-ST/SC or the general
population separately (in this paper the terms
non-ST/SC, general and others will be used
interchangeably). Figures 3a, 3b. 3c display
the 59FMRs for the SC, general and the
tribal population.
Setting aside the niceties of sequencing
we first examine the 59FMR map (3b) for
non-SC/ST population. Four broad EMR
ranges which are more or less spatially
contiguous can be noted. The region with
lowest FMR range; below 900, spreads
across (he plains north of Narmada. It
contains a ‘hard’ core area of alarmingly
low FMRs; below 850; spanning 21
districts in Haryana, western UP, north
eastern Rajasthan and the ravines of MP.
The next range of FMRs between 900 and
950 covers most of the remaining districts
in the plains in the north, districts of West
Bengal, districts in the central regions and
descends across Narmada into south
through Jalgaon, Dhule and Nasik districts
of Maharashtra. The gradual ‘recovery’ of
the FMRs and the contiguous contours
make the description of ‘pit with sloping
sides very apt for the region with very
low FMRs [Oldenbcrg 1992).
The remain i ng regions in the south-eastern
states have 59FMRs above 950. The hilly
state of Himachal Pradesh, Jammu and
Kashmir and hill districts of UP also have
FM Rs above 950. There are some 52 districts
in the south eastern region, in different
contiguous patches where the 59FMRs arc
above 1000. One significant block covers
Telangana region of Andhra, Chandrapur in
Maharashtra, Chhattisgadh in MP and the
tnbal tract of Orissa. Most of these distnets are known to be economically poor and
backward.
The 59FMRs for scheduled caste popu
lation follow a similar pattern (Figure 3a).
but the FMRs arc disturbingly low in the
northern region. There are 24 districts with '
59FMRs below 800. This situation, to put
it mildly, is scandalous. Oidenberg’s
(1992:2658) use of the term ‘Bermuda
triangle for the girl children’ will be applicable
for this region.
The range of 59FMRs below 900 for the
SC population spreads over a larger area
covering 107 districts, all confined to regions
north of Nannada. The next FMR range of
900 to 950; in 79 districts, does make inroads
in the Rayalseema region of Andhra and in
Salem region of Tamil Nadu. Rest of the
districts, 110 in number, mostly in south-east
have FM R range between 950 and 1000. The
contiguous patches of FMR above 1000 in
43 districts in the south eastern region more
or less coincide with the districts where
59FMR for the general population arc also
high.
For tribal population the 59FMR rarely
goes below 900 (Figure 3c) except in 11
districts;of Rajasthan.” The next set of about
45 districts in the FM R range of 900 to 950,
he mainly in the plain regions of Rajasthan
and northern M P. Anotherconliguous cluster
covers the Rayalseema region of Andhra and
its adjoining regions in Tamil Nadu. Most
other districts; 133 in numbers, have FMRs
abovc950 with a sizeable number; 45 among
3372
Economic and Political Weekly
December 28, 1996
J
.■
4
;
■
!'
I
j
Iiriifli limn■Mfimi
Figure 3a: 59FNIRs (SC)
—f
. ■[ y.
.■
V). > X- -
■
.
7 •-
r
/!
1
SC59FMR
| | no data
HH 700-800
ssa 801-900
901-960
961-1010
-
ES
■■
them, having FMRs above I (XX). Most of the
districts of north-eastern states have FMRs
in range of 950 to 1000. We have, however,
not included these in our analysis as indicated
earlier.
The 04FMRs for the scheduled caste
population (Figure 2a) reveal a contiguous
patch of about 80 districts with 04FMR
values below 950. There are only 12 districts
where the 04FMRs go below 900. In rest of
the districts the 04FMRs are above 950; in
96 of them 04FMRs exceed 1000. The
contiguous patch of low FMRs is confined
to the north-western belt. It starts from the
plains of Punjab and Haryana, and through
plains of western UP and ravines of Madhya
Pradesh descends down to East Nimar and
West Nimar districts of Madhya Pradesh.
Anotherpatch starts with Ajmer in Rajasthan
and descends down the Pali, Sirohi route to
the plains of Gujarat down to Vadodara. But
for Nagaur and Sikar districts it could have
joined the Haryana region in a contiguous
chain.
The pattern of 04FMRs for the non-ST/
SC population (Figure 2b) is similar but
more intriguing. The number of districts
Economic and Political Weekly
■ -4
■
•
>1010
with FMR below 900 is only 8, and those
with 59FMRs below 950 is 85. These form
a contiguous tract which again descends
down the plains of Punjab, Haryana and
western UP, to Hoshangabad and Betul
districts of MP. But it makes further ingress
into the south across Khandesh region of
Maharashtra, into its western ghat region
and goes as far as inland Karnataka.
(Permitting a minor liberty taken with FMR
in East Nimar (969) and Bijapur (965)
districts.) The belt joins up with the Sirohi,
Vadodara belt through Khandesh region of
Maharashtra and, except a break in Ajmer
district,joins the Haryana plains through the
Pali. Nagaur and Sikar.
There are other stray cases of di stricts with
low 04 FMR. Two of these deserve attention;
Lahaul and Spiti and Kinnaur in Himachal
Pradesh and Salem in Tamil Nadu. Lahaul
and Spiti and Kinnaur districts stand out for
their low FMR values for both the 0-4 and
5-9 age groups. Being situated in a zone with
otherwise favourable sex ratios these present
an anomaly that needs to be investigated.
Salem district has already been in focus
for the practice of sex selective infanticide
December 28. 1996
[George el al 1992:1153-57). Such practice
in large numbers will automatically show up
in low 04FMRs and this is precisely the case
with Salem. It is the only district in south
where 04FMR value goes below 900 (876
for the non-ST/SC). Il is pertinent to suggest
that a focused search for extremely low values
of 04FMRs among different regions and
groups at block levels (for which 1991 Census
data is available), may be a very useful method
of detecting areas where sex selective
infanticide or foeticide may have assumed
serious proportions.
:
Low^04FMRs are in any.case.indicative
of the regions where the excess female child
mortality has’made ingress into early years
of childhood. As such lhese.. indicate
intensification of anti-female bias. In view
of the concerns expressed by Miller (1989)
and others about the ingress of northerni sex
ratio patterns into the south, the route of low
04FMRs: described above assumes
significance. If such ingress is taking place
J1 *s most likely to take place along the route
of cultural circulation between the north and
south. The low 04FMR route appears to
coincide more or less with this route of
cultural circulation.,y This aspect needs
attention from a combined perspective of
cultural geography, gender and demography.
This issues intended to be pursued in further
details once the five year age group break
up for different social groups in the 1991
Census.becomes available.
Among the tribals, 04 FMRs rarely go below
950 (8.districts) and are above 1000 in
majority ofdistricts (about 116). In 16 districts
out of these the 04FMR exceeds 1050. There
are no particular spatial patterns. As such a
district level map is not presented here.
We thus find that the disaggregation of the
juvenile age group FMRs into 59FMR and
the 04FMR is a useful one and these FMRs
have significantly different patterns. We can
therefore proceed to examine in some detail,
the differences between these FMRs for
different regions and different social groups.
We first analyse the differences between the
09 FMRs, 59FMRs and 04FMRs for different
groups and examine the hypothesis that the
04FMRs will be significantly higher than the
59FMRs. This indeed happens to be the case.
Table 3.1 provides the data for the overall
population, as well as its three segments the
tribal, the scheduled caste and the rest. For
all these groups the 04 FMRs are significantly
higher and 59FMRs significantly lower. The
gap between 04FMRs and 59FMRs is
pronounced for the SC population and
becomes even more so if we take the 94
districts where the SC population accounts
forabove 20 per cent of thedistrict population.
This gap is less pronounced for the tribal and
tends to narrow down as we go to the 50
. districts which have 20 per cent or more
tribal population.
3373
•/
.
-..y
■.
—-——-------------------------
-
-
•
Figure 3b: 5-9F Aoe-Group FMR Non-SC/ST(198I)
V v '
I
I
I
I;
r
I
OTH59FMR
[•
[7 J no data
Ea 700-850
. ElSl 851-910
"
911-960
' S 961-1010
>1010
Vie next examine the differences in the
59FMRs across different social groups
( Fable 3.2). This is done for the 355 districts
as well as for districts with different
concentration of scheduled caste and
scheduled tribe population. The59FMRs for
the overall population, the non-tribal
population and the non-ST/SC population
do not differ significantly from each other.
But the 59FMRs for the SC population (339
districts) and the tribal population (189
districts) differ from these significantly: The
difference between the 59FMRs for the SC
and the non-ST/SC population remains
significant when we lake sub-samples, of
districts with increasing concentration of the
SC population. The 59FMRs for the nonST/SC and the tribal population also differ
significantly in these sub-samples and so do
the 59FMRs among the tribal and the
scheduled castes. FMRs for tribals arc high,
FMRs for the scheduled castes arc low while
those for the non-ST/SC population occupy
an intermediate position.
When we group the districts by the
percentage oftribal popu 1 ation, the di fference
between the tribal and the non-ST/SC
population FMRs becomes insignificant
while the 59FMRs for the SC population
remains significantly different and low. The
difference narrows down considerably in the
50 districts where tribal population is above
20 per cent of the district population.
When wc compare the 04FMRs between
the three social groups (Tabic 3.3) we find
that these do not differ significantly among
the SC and the non-ST/SC groups. 04FMRs
for the tribal population are, however, signi
ficantly higher in all the groups of districts
and have a mean value above 1000. As the
percentage of tribal population goes from 1
per cent to 20 per cent, these FMRs show
more evenly spread across different regions
• and do not differ significantly within the
non-tribal population, i c. between the SC
1 and the non-ST/SC group. The 04FMRs for
the tribals are different and have quite high
values, typically above 1000.
The 59FMRs on the other hand, assume
much lower values, differ between the three
social groups significantly across different
sub-samples of the districts. They show a
rcmarkablecontiguityacrossdiffcrent regions'
and vary over a much larger range than the
04FMRs arc. In certain regions they have
'J
• alarmingly low values.
The relatively even range of the 04FMRs
compared to the 59FMRs can be seen through
' an analysis of variance in the two FMRs over
same set of districts and social groups
(Table 4). The variance or the standard
deviation in the04FMR values is considerably
I
lower than that for the 59FMR values..The
pattern persists in the sub-sample of districts
I
where the SCor the ST population percentage
exceeds 20 per cent.
Thus we find that 59FMR satisfactorily
f
captures the socio-cultural patterns of excess
female mortality in thejuvenilegroup. Lower * • .
59FMRs indicateadversesurvival conditions
faced by the girl children compared to the
I
male children while higher 59FMRs indicate
less discrimination between the two.
We can notice that the districts with low- :
04FMR also have low 59FMRs. They will
rarely, if at all, have high 59FMRs. Districts
with high 04FMRs, on the oilier hand, can
have lew 59FMRs; typically in the north- western region, or high 59FMRs; typically
in the south eastern region.
We thus get four categories of districts:
(a) those with low 04FM Rs and low 59FMRs,
(b) thoscwithhigh04FMRsbullow59FMRs,’
(c) those with high 04FMRs and high
59FMRs, and (d) those with low 04FMRs <
and high 59FMRs.
The first category covers districts where
the pattern of excess female mortality gets
established during infancy, or in extreme
cases even before it through infanticide or
sex selective abortions. The second category
will be indicative of the districts where the
excess female mortality sets in during later
years of childhood. The third category of
districts will be the ones where the
But. when the SC population
r04 .kMR
sc ,O“
morta,i,inyihai
wK
h’1s3fo; n,ng
bC reversed
lalcr
i «Sf°rlhctnbalP*°n
We thus see a clear difference between the
two ‘layers’ within the juvenile age group
FMRs: the 04FMRs and the 59FMRs. The
04FMRs L.e significantly higher then the
59FMRs for the three social groups and
different sub-samples of districts. They are
in of ehildho^
j;One possib.e way of grouping different
districts is the percentage of the scheduled
caste or scheduled tribe population. This has
been used in some of the recent analyses
[Kishorc 1993; Murthy 1995] and we have
used it above. The concentration of the sche
duled caste or the tribal population by itself
may not, however, be a very useful indicator
3374
Economic and Political Weekly
December 28, 1996
Figure 3c: 59FMRS (Tribal Population)
| : A
■
■
•
k
I■■■ * frv
.-'J
1
•
) ■
•
J’;-.-’’
! '
of the differences in the survival patterns by
sex. The 59FMRs can, as we have seen
above, be more useful indicators of such
differences and so can the 04FMRs; at least
at the lower end of the range. It will be
relevant therefore to examine the differences
in FMR patterns by different FMR ranges.
This analysis is presented in Table 5. We
examine the 04FMR and 59FMR patterns
for di fferent sub-samples ofdistricts grouped
according to different 59FMR ranges. Four
ranges of 59FMRs are chosen: (i) below 900,
(ii) between 900 and 950, (iii) between 950
and 1,000, and (iv) above 1,000. For SC
59FMRs an additional range of very low
FMRs (below 800) is also considered.
We analyse both 04FMRs and 59FMRs
in these ranges for the three social groups.
We also look at the drop from the mean
11
Even though the 59FMR for tribal in these
districts arc higher compared to those for non
tribal, these are not high in terms of the
’ FMRs for the tribal population.20
In the next range of 59FMRs between 900
and950, one finds that the04 FMRs improve
sharply for both the SC and the general
categories.The04FMRs forthetribal remain
significantly higher, though the gap between
these and the 04FMRs for the non-tribal
groups is quite narrow now. But the 59FMRs
for the SC population remain significantly
low while those for the tribal population
improve considerably. The 59FMRs for the
general category occupy intermediate
position.
• As the 59FMR for SC population rises
above 950, the 04FMRs for the SC and ST
population converge and become
significantly higher than those for non-ST/
‘ SC groups. This is seen more sharply in the
next range where the 59FMRs for the SC
' - population exceed 1,000 in 43 districts. The
difference in the 59FMRs for the three groups
is insignificant.
•
>7 We now examine the sub-sample of 189
districts with tribal population above 1 per
. cent Slightly different results are obtained
:• at the low FMR end. In the 11 districts of
Rajasthan with 59FMR forthe tribal is below
900, the 04FMRs are comparable for all the
thred groups and fall in the range of 955 to
970. But the drop to 59FMRs is sharp for
| | Districts
ST59FMR
both the tribal and the SC population, lliese
I
| no data
are comparable and are lower than those for
@1 700-900
the general population. This pattern remains
BSS 901-950
more or less the same for the 56 districts
SSSS 951-1000
>1000
(mostly in Rajasthan and Madhya Pradesh)
where the 59FMRs for the tribal population
districts are low (mean 901) as well. .The are. below 950. The FMRs for the tribal
non-ST/SC population fares relatively better population, in both 0-4 and 5-9 age group,
(mean 04FMR 940), but only relatively, become significantly higher than those for
Clearly, the adversities in survival impihge the SC population in the FMR range of 950
upon the girl children among the scheduled to 1,000 (88 districts), and higher than those
castes more sharply. The drop from the mean'• for both SC and non-SC/ST group in the
04FMR to the mean 59FMR is 138 points 1,000 plus range.
for the scheduled castes and 92 points for
When we take the 21 districts where the
the others. There are only three districts in 59FMRs for the non-ST/SC population are
this group with significant tribal population, below 850, the 04FMRs for the SC and the
viz, Bharatpur and Sawai Madhopur of general category are comparable (there are
Rajasthan and Morena of MP.
only six districts with significant tribal
In the 107 districts where 59FMRs for the population). But the drop between mean
SC population are below 900, the pattern of 04FMR and mean 59FMR is far sharper for
adverse survival foflhfc girl children of the ■ the SC population and their 59FMRs are still
scheduled castes persists. Whilethe04FMRs significantly lower, than those of the nonfor the SC population are higher compared ST/SC population.
to the previous group of 24 districts and are
It thus appears that the scheduled castes
comparable to those for the general category,
are worse off at the lower end of the 59FMR
the 59FMRs remain significantly lower. The range no matter what the selection criterion
drop between mean 04FMR and mean is, i c, 59FMR for the SC, ST or the non59FMR is 105 points for the scheduled caste ST/SC group. Further, at this end, the girl
population compared to a drop of 64 points children among tribal and the SC groups fare
for the general category. The 04FMR for the worse than those in the non-ST/SC
tribal population in 30 districts are high population.The situation in the tribal pockets
(mean of 993) but the drop from 04FMRs improves quite sharply but that does not
to 59FMRs (mean of 925) is quite large. happen for the scheduled.castes. This may
04FMR to the mean 59FMR for the given
set of districts. The extent of this drop
indicates the survival adversity faced by the
girl children. We find out how the three
social groups fare within different sub
samples relative to each other.
There are 24 districts with 59FMR for SC
population below 800. The 04FMRs in these
■
';''
Economic and Political Weekly
December 28, 1996
’ 3375
B
-7
Figure 4: Regional Clusters
s
P
)
5
10
12
18
19
20
16
have something to do with the pattern of
‘assimilation’ of these two groups in the
‘mainstream’ of the society - tribals have
been relatively isolated from this
‘mainstream’ even where their population
percentage is moderate [Raza and Ahmad
1990]. But we would not pursue this
discussion here.
While the patterns observed at the lower
FMR end provide certain insights into the
differences, those observed at the higher end
also raises certain questions. In almost all
the cases where 59FMRs arc above 1,000,
two trends can be noticed; the 04FMRs are
also very high and, more importantly, the
drop from 04FMR to 59FMR is insignificant
and even negative. This would indicate that
the pattern of excess male mortality persists
beyond infancy. While this may indicate an
absence of discrimination against the girl
children beyond the age of one year, there
could be other more worrying possibilities.
It could also mean higher health risks and
poor health infrastructure resulting in
unusually high male infant deaths.
Continuation of the excess male child
mortality beyond infancy may not be an
indicator of the girl children faring well but
of the male children faring badly. The
subsequent absence ofdiscrimination agai nst
the girl child may also be indicative of the
poverty in the region, i c, the material
wherewithals for such discrimination may
themselves be absent. Both these possibilities
should cause concern. It is not a mere
coincidence that the districts with 59FMRs
above 1,000 are by and large the poorer and
backward districts. This point needs further
investigation.
IV
We saw above that the sex ratio patterns,
especially forlhe59FMRs, vary’considerably
within the boundaries of a stale. On the other
hand, they show remarkable contiguity across
certain groups of districts. Some of these
clusters have been noted for discernible
ecological or geophysical boundaries?1
The 1981. population census has
demarcated different geophysical regions and
subregions across the country [Government
of India 1988; 1981 Census Atlas 192-28],
1961 Census had laidconsidcrablccmphasis
3376
on such a classification and had initialed a
number of studies related to it [Bose 1994].
There have also been other classifications of
regions, apart from the one done by the
census. A comprehensive compilation of
these has been done by Bose (1994). Tnese
regions considerably overlap. While regional
studies have not been a new phenomenon
in India, their application in the field of
demography, at least sex ratio analysis, has
not been very frequent.22
We use the census classification here to
identify 19 different contiguous regions
within which the 59FMRs display
homogeneity (Figure 4). We then examine '
the significance of such a classification in
terms of analysis and policy implication. A
list of these regions and corresponding census
regions and subregions is given in
Appendix 1. Departure from the census
scheme, when done, is indicated separately.
Table 7 gives the mean of the 04FMRs and
59FMRs values of the districts in these
regions.
The first region consists of the stale of
Himachal Pradesh, Jammu and Kashmir and
the hilly region of UP, representing the‘south’
within the north. Mean 59FMR for the 32
out of 34 districts in this region is 961
(barring the districts of L and S and Kinnaur
which merit separate.scrutiny for their low
FMRs). FMRs for the scheduled castes are
comparable (mean 957). This region, largely
above 300 metres from the sea level, marks
an important ecological boundary between
the northern mountains and the plains in
northern India.
Adjoining this region arc the plains of
Punjab -and Haryana marked by highly
masculincscx ratios; the mean 59FMR being
887 for the non-ST/SC population and 842
for the scheduled caste population. There is
no tribal population in this belt.
But region 3 is the more alarming region,
liic ‘pit’. It comprises of the upper Ganga
plain of western UP, the three districts of
Alwar, Bharatpur and Sawai Madhopur of
Rajasthan, the ravines of Chambal in MP f
and the Zhansi uplands of UP. These 32
districts have a mean 59FMR of 850 for the
non-ST/SC population and 797 for the
scheduled caste population.
Low FMRs in these two regions, given
that these are relatively prosperous regions
of India, should be a matter of concern in
both policy and academic realm. This also
warrants a special coverage of at least some
of the districts in this region during the 2001
Census to set at rest some of the optimistic
speculation that undcrenumeration of females
may be the cause of low FMRs in India.
The 23 districts of the middle Ganga plain
represent the eastern side of the sloping
region around the ‘pit’ with mean 59FMR
for the non-ST/SC group being 892 and
those for the SC population being 883. The
Economic and Political Weekly
•/
;>
December 28. 1996
<
■d
r
. f?.i
a
nf
31
as
to
p**
A
in
nd
'■J
riv
ren
inn
I"
: is
on.
32
the
ven
r--
...
11 so
001
ales
lam
y '
h
?nd
■•••
996
1 ’A
■
.
i’/
t/-
FMRs continue to look up along the lower
up with the south Bihar hills and plateau
region spreading from Valsad in Gujarat
Ganga plain of north and south Bihar (region
region. The mean 59FMRs here are 1,000 or
(Dang included), to Kerala through Maha
5 with mean 59FMR for the 22 districts
above for all three social groups. South of
rashtra Konkan, Goa and Karnataka Konkan.
being 920 for non-ST/SC and 908 for the
the dividing belt, we have the west coast
It has high FMRs for all three groups.
SC population). In fact Pumia and Katihar
Table 3. i: Comparison of FMRs in the Age-Groups 0-9, 0-4, and 5-9
districts of Bihar show a closer pattern to
Variable
Mean
________ Cl
Per Cent CI
the lower Gangetic plain districts of West
Remarks
Lower
Upper
Bengal and have been grouped with these
as such. These districts have high and
355 districts
comparable 59FMRs for al I the three groups
AH04FMR
976
973
980
95
AI109FMR
and a sizeable presence of tribal population.
958
954
962
95
Overall population
AII59FMR
942
936
948
The sou’ icrn Bihar hills and plateaus mark
95
Oth04FMR
974
970
978
95
Non
ST/SC
an ecological transition from plains to hills
population
and also mark one end of the north-south
Oth09FMR
959
954
963
95
divide. These seven districts with sizeable
Oth59FMR
946
940
952
95
tribal population have high 59FMRs for the
Districts
with
percentage
of
scheduled
caste
population,
SCPCT
>1
per cent 339 districts
tribals and others and relatively lower
SC04FMR
978
973
982
95
59FMRs (mean 956) for the SC population.
SC09FMR
949
943
.955
95
The semi arid plains of Rajasthan and
SC59FMR
925
917
932
95
Gujarat, the Kachchh and Kathiawar
Districts with percentage of scheduled caste population, SCPCT > 20 per cent 94 districts
peninsula and some areas of semi arid
SC04FMR
962
952
972
95
Rajasthan, form a block of 34 districts. These
SC09FMR
923
911
936
95
adjoin the low FMR regions 2 and 3 on one
SC59FMR
890
i 874
906
95
side and the central belt dividing the north
Districts with percentage of trjbal population STPCT > 1 per cent
189 districts
and the south on the other. The 59FMRs for
ST04FMR
1008
10I4&^
1008 -« , r 1003
. the non-ST/SC population are marginally
SC09FMR
988
‘* ->983
>983 993 '.
95
better (mean of 920) but continues to be low
SC59FMR
970
964
976<rr
970
95
for the SC population (mean of 886). The
Districts with percentage of tribal population STPCT > 20 per cent
50 districts
59FMRs for the tribal population in 21 of
ST04FMR
|0l8
"•
-----1011
1024
90
its districts are low by the standards of tribal
SC09FMR
1002
994
1009
90
SC59FMR
population (mean of937). As a matterof fact
988
978
998 K
90
there is a low FMR track starting from Ajmer
in Rajasthan descending down through Pali
Table 3.2: Analysis of FMRs tn Age'-Group 5-9 Years
and Sirohi districts to the Mehasana,
Variable
'^Mean
_____ Cl
_____ W: Per Cent
Remarks
Ahmedabad, Bhavnagar route in Gujarat.
Lower
Upper . ?
CI
< <____
This track is flanked by relatively higher
/•“ ■
FMR regions on its western side and regions
355 districts
AI159FMR
942
936
948
95
Total population
of high FMRs on its south-eastern side. But
NT59FMR
957
952
961
95
Non-tribal population
we have not separated this region on the
Oth59FMR
940 ‘
946
952 dT/v , 95
Non-ST/SC population
basis of FMRs as this track cuts across
Districts with percentage of scheduled caste population, SCPCT >1 per cent
different geophysical regions.
339 districts
The districts of Udaipur, Chittaurgard,
SC59FMR
925
917
933
95
Scheduled caste population
Dungarpur, Banswara and Bhilwara in the
Oth59FMR
945
951-A;
95
939
Non-ST/SC population
relatively difficult terrain of the Aravali range
ST59FMR
970
964
976 - ; ’95
Tribal population
of Rajasthan form a contiguous block with
Districts with percentage of scheduled caste population, SCPCT > 10 per cent
districts of Malwa plateau in MP extending’
272 Districts
to the three districts of Narmada Valley, viz,
SC59FMR
917
908
926
95
Jabalpur, Narsimhpur and Hoshangabad
Oth59FMR
939
.932
946
95
through West Nimar in the Satpura hills. It
ST59FMR
966
958
974
95
has high FMRs for the non-ST/SC and the
Districts with percentage of scheduled caste population, SCPCT > 20 per cent
tribal population but low FMRs still for the
94 districts
scheduled castes.
SC59FMR
890
876
904
90
The other remaining region north of
Oth59FMR
915
906
924 .
90
946
928
Narmada covers the northern uplands of MP, • ST59FMR
963
90
the Sagar and Bhopal plateau and through
Districts with percentage of tribal population STPCT > 1 per cent
J 89 districts
East Nimar, descends into the Khandesh
ST59FMR
970
964
976
95
Tribal population
Oth59FMR
969
962
region of Maharashtra and the Nasik Basin.
976
95
SC59FMR
951
943
960
95
It has mean 59FMR values of 925 for the
non-ST/SC, 886 for the SC, and 955 for the
Districts with percentage of tribal population STPCT > 10 per cent 88 districts
9?499J
ST59FMR
983
tribal population.
95
Oth59FMR
983
971
995
The transition to high FMR zones begins
95
SC59FMR
957
947
968
95
with the central and eastern Satpura hill
range of MP, the Baghelkhand plateau, the
Districts with percentage of tribal population STPCT > 20 per cent 50 districts
----- --------98g
ST59FMR
Chhattisgarh region Dandakaranya and
978
998
90
Oth59FMR
989
973
1005
Orissa highlands. This block of 23 districts
90
SC59FMR
966
954
977
90
also lies on the central tribal belt and joins
Economic and Political Weekly
... .
December 28, 1996
3377
. ... •
■”T
- ------- --
.....
---------- n.T.
I
>
f
I
!•
t
i
On the inland side of this coastal belt, we
ratios, however, is outside (he scope of this
ratios. But even this has to be disaggregated
have the western ghats of Maharashtra, north
paper.
•
into the 0-4 and 5-9 age group FMRs because
Karnataka plateau and the central Maidan
Such change is also not significant for the
of the different mortality patterns in the
forming one block of 13 districts and the 0-4 age group FMRs (Table 6). Although
0-1 and 1-9 age group.
Vidarbha, Marathwada and Mahakoshal
regions provide a more homogeneous
The regionalisation of the 59FMRs has
region if Maharashtra forming another grouping, the improvement in F-values is
important implications forlheanalysisofthe
contiguous block of 13 districts with high
small and not as sharp as in the case of correlates of sex ratio variations. I: has of
FMRs for all three groups (mean 59FMR
5-9FMRs. This should not be surprising in
late been recognised that both cultural and
typically in the 975 to 990 range). 17 districts
.... of the random spatial spread of the
view
economic
factors affect female survival
ofTamilNadufollowsimilar59FMRpattcm, 04FMRs notwithstandVng rcgio^ with low
together [Kishore 1993; Murthy 1995]. It is
except that the tribal 59FMRs arc low. 1'
04FMRs. It will be instructive to compare
quite likely that the regional patterns will
There arc two contiguous blocks of very the situation with the one revealed by 1991
throw a better light on the role of cultural
high 59FMR (mean 59FMR above 1,000) Census data,
variables. Even the economic factors like
which need attention. One block covers 11
female labour participation would show
districts of the central south and the southern
V
considerable variation among different
Karnataka plateau and Chittoor district of
regions and social groups. The ravines of
Andhra. 'Die other block covers 10 districts
What do the patterns above signify? Four
Madhya Pradesh, for example, will have a
ofTclanganaregion in Andhra Pradesh.This r
„.
pertinent r
points
emerge. First, the sex ratio
very different pattern of female labour
block adjoins the Dandakaranya region and analysis has► to move away from the state
participation than, say, Chhattisgarh region.
the Chandrapur, Bastar, Koraput tribal belt level and lake into account the regional
The reversal of mortality patterns within
known for its backwardness. The eastern diversities. Evenaffieregl^dwthedlluia
the 0-9 age group and differences in the 0coastal region of Andhra including the district level
level the
the differeryres
differeirces in
in the
the patterns
patterns for
for thi
the
4 and 5-9 age group FMR patterns among
of Ganjam in Orissa also has uniformly high tribal, the schbduled caste and the rest of the
the three social groups raise two important
FMRs for the three groups.
population
population have
have to
to be
be taken
taken into
methodological points. One relates to the use
A group of live districts in Rayalseema consideration. For these analyses, the JSRs
of undcr-5 mortality as a composite variable.
region of Andhra stands out for its low FMRs or the JFMRs uni I be more appropriate
The other relates to the use of the population
by southern standards (mean 59FMRs of variables compared to the all age group sex
percentage of the SC or the tribal population.
961 for non-ST/SC, 936 for the SC, and 931
for the tribal population). This covers the
Tablk 3.3: Analysis of FMRs in Age-Grovp 0-4 Years
districts of Prakasam, Nellore, Cudappah,
Variable
CI
Per Cent
Anantpur and Kumool.
Remarks
Lower
Upper
CI
We now examine, through analysis of
variance if these 19 regions provide a more
AI104FMR
976
973
980
95
Total population
homogeneous grouping compared to the 20 NT04FMR x
975
971
978
95
Non-tribal population
974
970
978
states involved. The within group variance Oth04FMR
95
Non-ST/SC population *
would be significantly less ifa given grouping Districts with percentage of scheduled caste population. SCPCT >1 percent
ic
mnrr* homogeneous. 11.™
—»:__ ______
»j
*
is more
The c
F-ratios
would
339 districts
as a result be higher. Table 6 indicates the SC04FMR
978
973
982
95
Scheduled caste population
974
970
978
within group variance and the total variance OthCMFMR
95
Non-ST/SC population
ST04FMR
1008
1003
1014
95
for both 04FMRs and the 59FMRs by region
Tribal population
(193 districts)
and state for the three social groups separately.
Districts with pcrccntgagc of schedule caste population, SCPCT > 10 per cent
Corresponding F-ratios are also indicated.
The regions provide much more
272 districts
978
{ 972
984
95
homogeneous grouping than the states do for SC04FMR
Oth04FMR
973
969
977
95
the 59FMRs for the general and thescheduled
ST04FMR
1009
1002
1015
95
143 districts
caste population. In the case of grouping by SC04FMR
988
982
995
95
143 districts
states, the within group variance accounts Oth04FMR
983
978
98
95
143 districts
for nearly half of the total variance whereas
Districts with percentage of scheduled caste population, SCPCT > 20 per cent
if we group by regions it reduces to less than
94 districts
one-fourth of the total variance. The
SC04FMR
962
952
972
95
corresponding jump in the F-ratios is also Oth04FMR
964
957
972
95
significant. (It is possible to further ‘fine ST04FMR
1004
986
1022
95
28 districts
973
955
tune’ the regional grouping taking other SC04FMR
991
95
28 districts
Oth04FMR
981
968
994
factors into account and improve the F-ratios
95
28 districts
further. Such an exercise ofcluster formation
Districts with percentage of tribal population STPCT > 1 per cent
189 districts
1009
*
with minimum internal varianceis not within CTn<cnn
^(MFMR
1004
1014.
95
Tribal population
Oth04FMR
983
978
the scope of present paper. We only intended
988
95
985
979
991
95
to demonstrate that these regions, comparable SC04FMR
in number to the number of states involved,
Districts with percentage of tribal population STPCT > 10 per cent 88 districts
1013
STOIFMR
1013
1007
achieve substantial reduction in the variance ST04FMR
1019
95
Oth04FMR
993
984
within the groups.)
1002
95
SC04FMR
987
977
997
95
While the grouping by regions appears
much appropriate for the spatial analysis for
Districts with percentage of tribal population STPCT > 20 per cent 50 districts
ST04FMR
ST01FMR
JOIS
1009
1018
the not tribal population segments, such is
1026
95
Oth04FMR
993
979
l(>07
95
not the case with the tribal population. A
SC04FMR
983
969
997
95
study of the regionalisation of the tribal sex
3378
Economic and Political Weekly
December 28, 1996
I
L
......p----- ——....... .......... . .............. ..
may have to be given suitable incentives.
Chittaurgarh. Dungarpur, Banswara and
Chhattisgarh, Baslar and all districts of
Such provisions can be built into the planning
Jhalawar.
Orissa except three coastal districts
process itself. . ■
y.
, ..... .... Malwa plateau and Narmada valley
■ V’ (region 7) and Ganjam clubbed with
[Bose, 1994:46] in MP; the districts of
region 19.
Appendix-I
. ,
Mandsaur, Ratlam, Ujjain, Shajapur,
Region 14: Western Ghats in Mahaiashtra;
Dcwas, Jhabua, Dhar, Indore, West
Ahmednagar, Pune, Satara, Sangli,
Regions with Homogeneous Sex Ratio
Nimar, Rajgarh, Hoshangabad, Jabalpur,
’ Solapurand Kolhapur, Inland Karnataka,
Patterns
Narsimh’apur
ie, Bel gaum and Dharwar, North Maidan,
(West, and East Nimar show a sharp
ie, Bidar, Gulbarga and Bijapur and
Region 1: All districts of Himachal Pradesh,
divide which may be worth analysing at
Central Maidan, ie, Bellary and Raichur
Jammu and Kashmir and hilly districts
the block level).
[Bose 1994: 47; 1961 Census
of UP, viz, Chamoli, Pithoragarh,
Region 11: Western coastal districts; Starting
classification]
Uttarkashi, Dehradun, Garhwal, Tehri
from districts of Valsad and Dang at its
Region 15: Marathwada, Vidarbha and
Garhwal, Almora and Nainitai. These
northern end, goi ng do wn through Thane,
Mahakosal regions of Maharashtra
form part of the Northern Himalayas.
Raigarh and Ratnagiri in Maharashtra,
covering rest of its districts
Region 2: Part of the ‘Great Plains’; All
......................
Goa, Uttar and Dakshin Kannada in
Region 16: Rest of the districts
of Karnataka
districts of Punjab and Haryana.
Karnataka to all districts of Kerala.
covering South Maidan and Malnad and
(Although couple ofdistricts in Haryana
Region 12: North Malwa uplands i.e. Guna
Chiltoor of Andhra which shows
could go in the region 3) ;
and Shivpuri; north central MP, ie.
different characteristics from Rayalseema
Region 3: Districts of western UP in the
Chhatrapur, Tlkamgarh, Vindhya range
where it is included in the census
i
“ •
upper Ganga plain; Saharanpur,
. andRewaplateauVidisha,Raiscn,Sagar
classification.
'
• Muzaffarnagar, Bijnor, ?. Meerut,
Meerut,.
- -Panna, Rewa, Satna,
Damoh, Bhopal,
Region 17: Eastern Coastal region I; all
Ghaziabad, Bulandshahar, Moradabad,
East Nimar and: Jalgaon. Dhule and
districts
districts of
of Tamil
Tamil Nadu
Nadu and
and Pondichery
Pondichery
Rampur, Budaun, Bareilly, Pilibhit,
y-Nasik districts of Maharashtra.(This Region 18: Telangana region of AndhraShahjahanpur, Aligarh, Mathura/Agra,
™L
d^^
nlhel961rCg
’OnS
Mahboobnagar.Rangareddy,
Hyderabad,
ETOUDine draws
linon
thr- 1061 rvninne
Mnkk
__ I______ ______
Etah, Mainpuri, Farrukhabad, Etawah,
\ Bose (1994).)
/ JOO, X
Medak, Nizamabad, Adilabad, Karim-’
Jalaun, Jhansi, Lalitpur,; Hamirpur, Region 13: Remaining districts of MP
nagar, Warangal, Khammam and
Banda and Kheri, Bharatpur and Sawai
. covering Satpuras. Bagherkhand,
Nalgonda
Madhopur which constitute the
Cllb»r#*rTtr»r, r\f
—1__ 1 1__ •
subregion
of Banas Chambal
basin and
Table 6: Analysis of FMRs by Regions and by States
Alwar (Alwar defies the subregional
Region (DF= 18)
State (DF+ 19)
classification; it has been included here
.. — —
,
I
i
■
i:
I
r
although it belongs to jhq subregidn of
Aravalli range and associated uplands of
Semi-arid Rajasthan).
,
Chambal ravines of B undelkhand in M P;
the districts of Bhind, Morena, Gwalior
and Dalia (the 1981 classification
includes Guna and Shivpuri.-We have
used the 1961 classification quoted in
Bose (1994:44-48). These classify Guna
and Shivpuri under northern Malwa
uplands.
Region 4: Middle Ganga plain; Remaining
districts of UP
- •'
Region 5: Districts of Bihar in lower Ganga
plain; All districts of Bihar, except
Katihar and Pumia clubbed with region
7 and the districts in region 6. '
Region 6: South Bihar Hills and plateau;
Districts ofPalamau, Ranchi, Hazaribagh,
Singhbhum, Dhanbad, Santhal Parganas.
Region 7: Katihar and Purnia of Bihar, all
districts of West Bengal and Cuttack,
Puri and Balasore districts of Orissa.'
(Although Puruliacouldbeclubbed with
region 6, and the hilly districts could
form a separate group.)
•
Region 8: Semiarid Rajasthan and plains of
Gujrat; All districts of Rajasthan except
those
musein
in region
region 10
1 u (hilly
(hilly region)
region) and
andall
all
districts of Gujarat except Valsad and
~ clubbed
.................in region 11 covering
Dang
districts on west coast.
Region 10: HillydislrictsofRajasthan[Bosc,
1994:45], viz, Bhilwara, Udaipur,
3380
w
I
Variable , . Within
Sum of Squares (in 000s)
Between
Total
• F-Ratio
Oth59FMR
SC59FMR
ST59FMR
245
369
187
835
1422
166
1080
1790
326
62.48
67.73
8.9
526
811
219
554
980
134
1080
1790
326
20.53
22.53
8.99
Oth04FMR
SC04FMR •_
ST04FMR
242
422
66
186
284
179
428
705
245
14.06
11.8
3.69
297
528
205
131
177
41
428
705
245
8.58
6.26
2.92
(F Probability in all the cases is 0.0000)
Table 7: Mean FMR Vautes in Diffekent Geophysical Regions
Region No
1
2
.
ST
1015
Others
ST
988
921
932
992
1012
989
996
958
988
974
948
1019
978
980
1004
991
1013
1001
996
972
924
947
979
993
1022
997
958
1010
975
962
995
952
975
984
971
1007
995
980
977
59FMRs
SC
.A
’ -J
y'
I?
■ y;
.'V
in
\ •
w
Sum of Squares (in 000s)
Within Between
Total F-Ratio
04FMRs
SC
..
■'
'
;
'7.$ fe?
BU
:i •;T'§
ir..^'7
yy b
ys tel
.7 I,
■y. 'I■
J
•'
-
Others
957
842
792
882
908
956
981
886 .
934
975
886
1000
982
983
1016
981
998
971
937
956
887
854
850
4
996
892
5
949
920
6
989
972
7
975
980
8
937
920
10
961
980
11
964
979
12
955
925
13
1017
1019
14
Q77
972
15
992
982
16
1001
1009
17
945
973
18
968
1014
19
979
993
20
931
961
Nute'. Mean values here are mean of the FMR value of the districts in the region and NOT the mean
FMR values for the region.
972
1028
1056
1033
1003
988
1009
7 1001
7,' 1008
• < • 1030
989
998
1008
1004
1020
1027
987
• •/
Economic and Political Weekly
December 28, 1996
i 'J
-
5
s ■
....i jmC,
1
■■ ■
may have to be given suitable incentives.
S uch provisions can be built into the planning
process itself.
Appendix -1
I
J
I
■
i
I.
i-
Regions with Homogeneous Sex Ratio
Patterns
-lu,.
■’ ■ f
3
Chittaurgarh, Dungarpur. Banswara and
Jhalawar.
Malwa plateau and Nannada valley
[Bose, 1994:46] in MP; the districts of
Mandsaur. Ratlam, Ujjain, Shajapur,
Dewas, Jhabua, Dhar, Indore, West
Nimar, Rajgarh, Hoshangabad, Jabalpur,
Narsimhapur
(West, and* East Nimar show a sharp
divide which may be worth analysing al
the block level).
Region 11: Western coastal districts; Starting
from districts of Valsad and Dang al its
northern end, goi ng down through Thane,
Raigarh and Ratnagiri in Maharashtra,
Goa, Uttar and Dakshin Kannada in
Karnataka to all districts of Kerala.
Region 12: North Malwa uplands i.e. Guna
and Shivpuri; north central MP, i e,
Chhatrapur. Tikamgarh, Vindhya range
and Rewa plateau Vidisha,Raiscn, Sagar,
Damoh, Bhopal, Panna, Rewa, Satna,
East Nimar and; Jalgaon, Dhule and
Nasik districts of Maharashtra.(This
grouping draws upon the 1961 regions
Bose (1994).)
Region 13: Remaining districts of MP
. covering Salpuras. Baghcrkhand.
Chhattisgarh, Bastar and all districts of
Orissa except three coastal districts
. (region 7) and Ganjam clubbed with
region 19.
Region 14: Western Ghats in Maharashtra;
Ahmednagar, Pune, Satara, Sangli,
Solapurand Kolhapur; Inland Karnataka,
ie, Bclgaumand Dharwar. North Maidan,
ic, Bidar, Gulbarga and Bijapur and
Central Maidan, ie, Bellary and Raichur
[Bose 1994: 47; 1961 Census
classification]
Region 15: Marathwada. Vidarbha and
Mahakosal regions of Maharashtra
covering rest of its districts
Region 16: Rest of the districts of Karnataka
covering South Maidan and Malnad and
Chittoor of Andhra which shows
different characteristics from Rayalscema
where it is included in the census
classification.
Region 17: Eastern Coastal region I; all
districts of Tamil Nadu and Pondichcry
Region 18: Telangana region of Andhra;
Mahboobnagar.Rangareddy, Hyderabad,
Medak, Nizamabad, Adilabad, Karimnagar, Warangal, Khammam and
Nalgonda
Region 1: All districts of Himachal Pradesh,
Jammu and Kashmir and hilly districts
of UP, viz, Chamoli, Pithoragarh,
Uttarkashi, Dehradun, Garhwal, Tehri
Garhwal, Almora and Nainital. These
form part of the Northern Himalayas.
Region 2: Part of the ‘Great Plains’; All
districts of Punjab and Haryana.
(Although coupleofdistricts in Haryana
could go in the region 3)
Region 3: Districts of western UP in the
upper Ganga plain, Saharanpur,
Muzaffarnagar, Bijnor, Meerut,
Ghaziabad, Bulandshahar, Moradabad,*
Rampur, Budaun. Bareilly, Pilibhit,
Shahjahanpur, Aligarh, Mathura, Agra^
Etah, Mainpuri, Farrukhabad, Etawah,
Jalaun, Jhansi, Lalitpur, Hamirpur,
Banda and Kheri, Bharatpur and Sawai
Madhopur which constitute the
subregion of Banas Chambal basin and
________ 6: Anal«is of FM Rs by Regions and by States
Alwar (Alwar defies the subregional
Region (DF = 18)
State (DF + 19)
classification; it has been included hei
here
although it belongs to the subregidn of
Sum of Squares (in 000s)
Sum of Squares (in 000s)
Ara valli range and associated uplands of Variable .
Within Between
Total
F-Ratio
Within
Between
Total
F-Ratio
Semi-arid Rajasthan).
Oth59FMR
245
835
1080
62.48
526
Chambal ravines of Bundelkhand in MP;
554
1080
20.53
SC59FMR
369
1422
1790
67.73
811
980
1790
the districts of Bhind, Morcna, Gwalior ST59FMR
22.53
187
166
326
8.9
219
134
326
8.99
and Datia (the 1981 classification
includes Guna and Shivpuri. We have Oth04FMR
242
186
428
14.06
297
13!
428
8.58
SC04FMR
used the 1961 classification quoted in
422
284
705
11.8
528
177
705
6.26
ST04FMR
66
Bose (1994:44-48). These classify Guna
179
245
3.69
205
41
245
2.92
and Shivpuri under northern Malwa
(F Probability in all the cases is 0.0000)
uplands.
Region 4: Middle Ganga plain; Remaining
Table 7: Mean FMR Values in Different Geophysical Regions
districts of UP
Region 5. Districts of Bihar in lower Ganga
—■
04FMRs
59FMRs
Region No
ST
SC
Others
ST
plain; All districts of Bihar, except
SC
Others
Katihar and Pumia clubbed with region 1
~ Toi5
988
972
977
957
956
2
7 and the districts in region 6.
921
924
842
887
3
972
Region 6. South Bihar Hills and plateau;
932
947
854
792
850
4
1028
Districts ofPalamau, Ranchi,Hazaribagh’
992
979
996
882
892
5
1056
1012
Singhbhum, Dhanbad, Santhal Parganas.’
993
949
908
920
6
1033
989
1022
989
Region 7. Katihar and Purnia of Bihar, all
956
972
1003
7
996
997
975
981
980
districts of West Bengal and Cuttack,
8
988
958
958
937
886
920
Puri and Balasore districts of Orissa.
10
1009
988
1010
961
934
980
(Although Puruliacould beclubbed with
H
1001
974
975
964
975
979
12
1008
region 6, and the hilly districts could
948
962
955
886
925
13 ‘
1030
form a separate group.)
1019
995
1017
1000
1019
14
989
978
Region 8: Semiarid Rajasthan and plains of
952
977
982
972
15
998
980
975
992
Gujrat; All districts of Rajasthan except
983
982
-------- r16
1008
1004
984
1001
1016
those in region 10 (hilly region) and all
17
1004
1009
991
971
945
981
973
districts of Gujarat except Valsad and
18
1020
1013
1007
968
998
1014
Dang clubbed in region 11 covering
19
1027
1001
995
979
971
993
20
987
districts on west coast.
996
980
931
937
961
Region 10:1 lillydistrictsof Rajasthan [Bose,
Note: Mean values here arc mean of
< (he FMR value of the districts in (he region and NOT (he
1994:45], viz, Bhilwara, Udaipur,
mean
FMR values for the region.
K
3380
I
•••
I
•/
Economic and Political Weekly
December 28, 1996
Region 19: East coastal Andhra and south
Orissa; Ganjam district of Orissa and
rest of the districts of Andhra except
those in region 20.
Region 20: Anantpur.Cudappahand Kumool
of Rayalseema and the two southern
coastal districts of Prakasam and Nellore
classified as a separate subregion.
Remarks: (1) Delhi, Chandigarh and Bombay
not included in the regional analysis. (2)
The central zone which represents the
north-south transition is difficult to
*
classify and indicates tracks which further
cut across these regions. These could be
analysed at a more detailed level with
smaller regions.
I ■
1*
1
s
Guio and Dreze (1995) by using district level
except observing that parity in one aspect
data, yet most of the analysis continues to
does not necessarily mean parity in other
draw upon state level figures not corrected for
aspects and the combined impact of
migration.
•
different factors in terms of mortality wul
6 Vannemann and Barnes (1992) provide
always be guided by the factor in which
extremely useful data on various aspects at
the gap is larger (and critical). There may
district level including the population Census
beequality in consumption of calories, for
data from 1961 to 1981. They too have not
example, but a critical gap in access to
included the detailed data on the tribal and (
health care, between two groups. The
SC population in some respects. For these the
mortality differences will be driven by the
special tables from 1981 Census have been
. latter and if the situation is reverse, then
used. This database tends to aggregate some
by, the gap in calorie consumption.
of the data at the state levd/br the small states,
15 See Kieimann (198?: 185) quoted in Caldwells
particularly in the nort^easL
(1990:17)
7 Visaria (1971:26) has /calculated the effect
16 This will not be the case where excess female
of random errors op.the sex ratios for
child mortality is very sharp in the toddler (1 different sample sizes of live births. He
2) age group or has made inroads in the 0indicates that one requires a surprisingly
1 age group or even prior to it through
high number of births. 10,000 and above,
infanticide or sex selective abortions. Also see
• to get the sex ratios at birth within a narrow
Notes
note. 17.,
confidence range,
17 This will particularly be the case in districts
8 Present analysis does not cover the tribal
where excess female child mortality has
[This work is part of my ongoing research on
districts in the north-easi. This may appear
. become
ucvuuic significant
atigiuncani between
oetween the
me age
agi of one
Sex-Ratio Imbalances in India at the Eshool of
unusual, but in our opinion the tribal in the
and two years itself or even earlier■ as some
Development Studies University of East Anglia,
north-east and in central India belt differ
- -- - data
■ - from
..................
- Health
Norwich. UK, NR4 7TJ.
the
National-Family
,, recent
considerably,
a point
intended to be dealt with
, A
Survey (NFHS. 1995: Table 8.5) shows.
I wish to gratefully acknowledge the help and
elsewhere. Inclusion of these districts could rO; ’,'-i.
; Caldwells (1990) have
such
guidance of Richard Palmer-Jones in preparation
uu»v also
<uav discussed »uvil
have resulted in aggregation of nonpossibility. Government of India (1988b) in
of maps 2A to 3C. He and Cecile Jackson also
compMabte groups.
•
facL’ lists out 142 districts, ,,
(20
in Bihar,
gave useful comments and suggestions on the
______
______,9_ in
..i
9 Oldcnberg (1992:2658) has used this term
“ *
Gujarat.
12 in Haryana, 9 in Punjab. 14 each
draft version of this paper. I have also benefited
although not on the basis of the juvenile sex
, in Rajasthan and MP, 46 in UP being main
from discussions with Bina Agarwal. Ashish
ratios. We will later see that this is an apt .
Bose, K S Natarajan and Ravi Verma.]
contributors) where the q2 values for the girl
description for a group of 24 districts with
children are high compared to those for the
very low 59FMRs. This region resembles, to
__ „ .
? 1 We use the term FMR defined as the number '
use Oldcnberg’s graphical lean, a ‘pit’ with '18 Thx?
These~districts are:
a Alwar, Bharatpur, Sawai
of females per thousand male population to
sloping sides when we look al the spatial
Madhopur, Jhunjhunu, Sikar, Jaipur, Tonk,
nvo’d confusion in use of the term sex ratios..
distribution
of--the-FMRs
------ —
. ..— across districts.1
uuiiui
Bundi,, jujvi
Jalor tuiu
and Dimitri.
Banner.
In India the term sex ratios is traditionally used
10 Desai (1969:Ch7 and pp206-207);Srivastava 19 See Bharadwaj <19??: Ch D and V), Sopher
to mean number of women per thousand male
(1979:58-59
1961
/ktto.co and71) for details based on =«,.
(1980:.-Ch 10), Spate (1971, Ch 6) and
population; exactly opposite to the
and 1971 Census data.
~j
international convention.
Schwartzberg (1992: IIIA and related maps)
11 See, for example, Kundu and Sahu (1991).
2 While this was noted in the earlier literature
on the point of cultural circulation between
who
opine
that
“
at
the
state
or
district
level,
the north and the south across Narmada:
[Bardhan 1974; Visaria 1971], this pattern
was <
” ’ ‘
‘ ‘
‘
migration, is the single most important factor
Chhotanagpur belt
was explicitly brought into focus by Sopher
explaining the temporal and cross-sectional
20 It will be very interesting to see how the
(1980), Miller (1981) and later by Dyson and
_______________
_ WUIK
.,a
variations in sex ratio” In our opinion, once
Moore
(1983) among
others.
undercnumeration optimists*, will like to
^^tivesurvivalofthefcmalesisrecogniscd
3 Dyson and Moore haviralso recognised the .
explain away this pattern except by invoking
■_______________________________________
as the central issue of analysis, the debate on
an underenumeration in the 5-9 age group and
difficulty
in placing the eastern region into
migration becomes unnecessary if (he sex
its absence in the 0-4 age group. The sharp
the northern or southern stereotype. Caldwells
! (1990) in fact consider this ‘dichotomy’ as
ratios for the juvenile population are used.
decline in the FMRs from 0-4 age group to
12 We assume here that the sex ratio at birth is
, 5-9 age group among the tribal is
exaggerated. In a recent paper, the author has
---- . different
... regions; 104
nearly--------constant-across
. hardl° to explain ° away b/ Invoking
attempted to extend the two-fold; north-south,
to
per thousand
rn 109
ino male children
---------- ------ A cfemale
—..
underenumeration.Thetribalarefarlesslikely
classification into a five-fold classification
children. We also assume that enumeration
using the sex ratio data for different language
. to practice underreporting of the girl children •
errors are not significant enough. We do not
groups in India [Agnihotri Forthcoming].
given, their.social structure. (• Those who
rule
out
enumeration
errors,
but
do
not
share
. expect the relative underenumeration of the
4 There have been occasional references to the
the view that the low sex ratios can be explained
. girl children to be the main cause of the low
> generally high FMRs among the tribals. Dangc
away by sex selective undcrenumeration of
, •>
(1972:282) has raised this point explicitly in
FMRs.) ci
.
female children. Visaria (1971) has adequately
t s the context of Madhya Pradesh but this has
21 Libby (1980:94) has noted, for example, the
dealt with this issue but this view persists
not been pursued further. Miller (1981:74)
- ecological distinction between eastern and
nevertheless. Also see Kishore (1993) and
has in fact considered the classification of
western UP. Bardhan (1974) has also talked
Murthy
(1995).
.
;
tribals, scheduled castes and the rest as toe
; about the wheat and the rice divide in the sex
13
(a)
Government
of
India
(1991).
;
gross. It is only recently, [Agnihotri 1995; and
ratios.arguingthatthelowfemaleparticipation ■
(b) Deaths in 0-9 age group account for about
in wheat regions vis-a-vis high participation
Agnihotri Forthcoming) that the differences
one-fourth of the total deaths and little less
between the sex ratio patterns among these
in rice regions affects the sex ratio patterns
than half of the deaths below the age of
in uiv>v
these regions.
three social groups has been examined
60 (Government of India 1991);?,-. , - j, . 22 Even
studies like
(1991) who has ;
| • systematically on a all-India basis leading
" recent
.
___Raju
____________
14 See Hariss (1987),-Miller (1981. 1989), . dealt
. -.rwith ‘Gender and Deprivation-A Theme
•
to the present analysis. The possibility of
Basu (1989). Caldwells (1990), Mukherjee
Revisited with Geographical Perspective’ or
significant differences between the scheduled
(1986). Bhatia (1983) (quoted in Caldwells
I caste and the non-SC/ST population has also
Agarwal (1994: Ch 8,316-419) while‘tracing
(1990) ). An elaborate discussion of this
not been seriously explored so far.
cross-cultural diversities’ in respect of
‘neglect’ has been done in Miller (1981).
■ > 5 The trend normally is to use the state level
women’s position haveconfined their attention
There is some debate about the relative
at state level. Murthy, Guio and Dreze (1995)
data. Valuable contributions have been made
•l
importance
of
different
factors
in
survival,
by Sopher (1980), Libby (1980), Miller (1981)
are perhaps the first to take up regional
e
g,
nutrition
versus
health
care,
etc.
We
and more recently Kishore (1993) and Murthy.
t classification, systematically in the gender
do not intend going into it in this paper .. context. '
■
4
•
■ r3 "•'<
a ■
i■
■<
•,
'i.
■
t -■6r
■
• j Mr'
Economic and Political Weekly
December 2b, 19^0
■
3381
1
References
Agarwal, B (1994): A FieldofHer Own, Cambridge
University Press, Cambridge.
Agnihotri, S B (1995): ’Missing Female - A
Disaggregated Analysis’. Economic and
Political Weekly, August 19, pp 2074-84.
Agnihotri, S B (Forthcoming): ‘Sex Ratio
Variations in India: What do Languages Tell
I
Bardhan, P K (1974): ’On Life and Death
Questions’, Economic and Political Weekly
Special Issue No 9,' pp 1293-1304.
Bcrreman, G D (1993): ’Sanskritisation as Female
Oppression in India’ in B D Miller (cd). Sex
and Gender Hierarchies, Cambridge
University Press, Cambridge, pp 366-92.
Basu, A M (1989): Ts Discrimination in Food
Really Necessary for Explaining Sex
Differentials in Childhood Mortality^’
Population Studies, 43.
Bharadwaj, S M (1973): Hindu
I ilgnmage in India, University of California
Press, London.
Bose, A (1994): Demographic Zones in India, BR
Publishers, New Delhi.
Caldwell, P and J Caldwell (1990): ’Gender
Implications for Survival in South Asia’, Health
Transition Working Paper No 7, NCEPH
Canberra.
’
DanSe. A S (1972): ’An Analysis of the Sex Ratio
Differentials by Regions in Madhya Pradesh’
Artha Vijnana, 14:273-86
Dasgupta, M (1987): ’Selective Discrimination
Against Female Children in Rural Punjab,
13d77 ’ iy£Ulat,t,n
DeveloPm£nt Review,
Desai. P B (1969): Siz^e and Sex Composition of
Population in India, 1901-1961, Asia
Publishing House, London. Dyson, T and M Moore (1983): ‘On Kinship
Structure,
Female Autonomy and
Demographic Balance’, Population and
Development Review, 9:35-60
George, S, R Abel, B D Miller (1992): ’Female
Infanticide in Rural South India’. Economic
and Political Weekly, Vol 27', pp 1153-56
Government of India (1988a): Census Atlas'.
National Volume Census ofIndia 1981 Scries
I Part XII, 1988, Office of the Registrar
General and Census Commissioner. New
Demi, India.
■
- (1988b): ’Child Mortality Estimates of India’
I981' Occasi0^ Paper No
5 of 1988. Office of the Registrar General and
,.^US(.Comniissioncr- New Delhi, India.
(1)91): Survey of Causes of Death, Annual
Report 1991, Office of the Registrar General
and Census Commissioner, New Delhi
India.
Harriss B (1987): Differential Child Mortality
and Health Care in South Asia1, Journal of
Social Studies, 44:2-123
Heyer, J (1992): ’The Role of Dowries and
Daughtcrs Marriages in the Accumulation
and Distribution of Capital in a South Indian
Community’, Journal of International
Development 4(4)
Kapadia. K (1994): ’Bonded by Blood: Ma tn lateral
Kin inTamil Kinship’, Economic and Political
Weekly, April 9, pp 855-61.
Kishor, S (1993): ’May God Give Sons to All
Gender and Child Mortality in India’,
American Sociological Review. 58:247-65.’
Kundu. A and M K Sahu (1991): ’Variation
in Sex Ratio: Development Implications’,
Economic and Political Weekly, October 12
pp 2341-42,
Libbee, M (1980): *Tcrritorial Endogamy and the
Spatial Stnjcture of Marriage’ in D E Sopher
(cd).
Eiploration of India, Longman,
London, pp 65-104.
Madan. T N (1993): ’Structural Implications of
Mamage/Mliancein North India: Wife-Givers
and Wife-Takers among the Pandits of
Kashmir’ in P Ubcroi (cd). Family. Kinship
and Marriage in India, Oxford University
Press. New Delhi, pp 287-306.
.Mies, M (1988): Women: The Last Colony, Zed
Books. London.
Miller. B D (1981): The Endangered Sex, Cornell
University Press. Ithaca. NY.
Mukherjee. S (1986): Establishment of a Child
Mortality System in India, UPS, Bombay.
Murthi, M. AGuio and J Drcze( 1995):’Mortality
Fertility and Gender Bias in India: A District
Level Analysis’, DERC discussion paper No
61. LSE, London.
NFHS (1995): National Family Health Survey
Oldenberg. P (1992): ’Sex Ratio. Son Preference
and Violence in India’. Economic and Political
Weekly. December 5-12, p 2658
Rao. V (1993): ’The Rising Price of Hus
bands: A Hedonic Ana! )-sis of Dowry Increases
I01(4^*a " Jt>Urnal
Economy,
Raza, M and A Ahmad (1990): An Atlas ofTribal
India, Conccpt/ICSSR, New Delhi.
Raju, S (1991): ‘Gender and Dcprivaion — A
Theme Revisited with a Geographical
Perspective’, Economic and Political Weekly
December 7. pp 2827-39.
Roscnzwcig. M and T P Schultz (1982): ‘Market
Opportunities, Genetic Endowment and Intra
-Family Resource Distribution: Child Survival
in Rural India’, American Economic Review
Vol 72. pp 803-15.
Sopher, D E (cd) (1980): An Exploration ofIndia,
Longman. London.
Spate. O H K and ATA Lcannonth (1967): India
and Pakistan: Land, People and Economy.
Methuen. London.
Schwanzbcrg. J E (cd) (1992): A Historical Atlas
ofSouth .4.na, Oxford University Press Oxford,
S.ivastava. S C (1979): ’Migration in India’.
Census of India 1971, Paper 2 of 1979, Office
of RGI. New Delhi.
Vanncmann. R and D Barnes (1992): Appendix
V - 6, Economic Activities, Indian
Development District Database, Centre for
Population. Gender and Social Inequality.
University of Maryland.
..
Visaria, P (1971): The Sex Ratio ofthe Population
of India: Monograph No JO. Census of India
1961. Office of the Registrar General, New
Delhi. India.
Wadley. S( 1993): ’FamilyComposition Strategies
in Rural North India’, Social Science and
Medicine, 37(11), pp 1367-76.
RECONFIGURATION OF INDIAN POLITICS:
F-’------ STATE
ASSEMBLY ELECTIONS, I993I95
January 13-20, 1996
Reconfiguration in Indian Politics:
State Assembly Elections,
1993-95
Yogendra Yadav
HP: Political Necessity vs Lost Possibilities
Javeed Alam
UP I: Sectional Politics in an Urban Constituency
UP II. Grass Roots Political Process: Atraulia
Madhya Pradesh I; Setback to BJP
V B Singh
Christophe Jaffrelot
Sanjay Kumar
■ Madhya PcStdesh 11: Muslims in Electoral Politics
Andhra Pradesh: Elections and Fiscal Reform
Goa: A Democratic Verdict?
R K Srivastava
Sudha Pai
Peter Ronald deSouza
Karnataka: Emergence of Third Force1
Sandeep Shastri
Orissa: Tribal-Dalit Conflict: Phulbani
Bishnu N Mohapatra, Dwinpayan Bhattacharyya
GJjarat: BJP’s Rise to Power Ghanshyam Shah
Maharashtra I: Shift of Power from Rural to Urban Sector
Maharashtra II: Capturing the Moment of Realignment
Rajendra Vora
Suhas Palshikar
For Copies, Write to:
Circulation Manager
Economic and Political Weekly
Hitkari House, 284 Shahid Bhagat Singh Road,
Mumbai 400 001
3382
Economic and Political Weekly
December 28, 1996
A Less Valued Life: Population Policy
and Sex Selection in India
By Rupsa MaHik
October2002
Uj
a
<o
ns
Demographic research over the past two decades has
confirmed that a preference for sons over daughters
remains entrenched in many countries throughout the
W'orld. In such se tings, religious traditions and social
norms coupled with economic discrimination against
women and girls conspire lo ensure that young boys
have greater access to education, health care, and even
food than do lheir sisters. Such neglect leads lo
markedly' higher rates of illiteracy', malnutrition, and
poor health among girls.
In the w'orst cases,
discrimination against girls lakes the form of female
infanticide, in which girl children are killed outright
immediately after birth These practices have evolved
in recent years to include the use of modern
technologies to determine the sex of children in the
womb and the subsequent use of sex-selection
abortion to avoid the birth of a girl child altogether.
The result of such practices is evident in the growing
imbalance in the survival of girls relative to boys in
some countries today.
Such is the case in India, where the combined effects of
historical discrimination against girl children and the
use of advanced technology for sex selection are now
clear. Data collected in the ZOOI Census of India1
reveal that the juvenile sex ratio has declined steadily
over the past decade, from 945 girls per 1,000 boys
ages 0-6 years old in 1991 to 927 girls per 1,000 boys in
2001. This decline has been attributed both to excess
neo-natal female mortality due to ihe spread of female
infanticide, and to the rapidly expanding use of pre
natal diagnostic technology- for the purposes of sex
determination (SD) followed by use of sex-selection
abortion (SSA).
It is now- indisputable that, as India enters the 2b>
century, SD and SSA have been integrated into the
range of family building strategies used by couples to
ensure, a desired "imbalance" in the number of male
and female offspring. What is less well understood are
the ways in which population policies supported bv
both the government and international donor agencies
have fueled the insidious use of modem technology to
eliminate girl children even before they' are born.
The Roots of Gender Bias
The roots ol son preference in India lie in deeply
entrenched
social,
cultural,
and
economic
discrimination against women and girls. The
predominant system of patrilineal descent and
inheritance legitimizes and propels the desire for sons.
Sons, for example, traditionally perform the last rites
after the death of a parent.
Indeed, a strict
interpretation of Hindu tradition holds that salvation in
the afterlife can only be achieved if a son lights his
parent's funeral pyre (Mutharayappa, et al.; 1997). As a
result, many religious Hindus strive to ensure they have
at least one son.
Economic calculations are increasingly' a factor in the
perpetuation of son preference.
In much of the
country, men and boys aie more likely to work for
cash wages than are women and girls. Although
women often work longer hours than men, lhev are
more likely to be engaged in impaid subsistence and
domestic work that, while critical to family survival, is
ironically perceived to be less valuable. Al marriage,
daughters leave their natal homes and must bring a
dowry’ to lheir husband's family, lo which lhey are
also expected lo contribute economically, whether in
the form of paid or unpaid work. Sons are expected lo
support lheir parents in old age, and therefore are
viewed as a source of social security.
In fact, the desire to accumulate wealth has become an
increasingly important factor in son preference in
recent years, in part as a result of the desire among the
growing middle class for upward mobility. The spread
of consumerism and the associated increase in the cost
of down- and marriage, plus the desire to maintain
landholdings within a family all have contributed to an
environment that is extremely hostile to women and girl
children, even among the educated middle and upper
classes. Indeed, contrary to what might lie expected, the
most dramatic declines in the sex ratio over the past
decade were found in I’unjab, Haryana and
Maharashtra, among the richest stales in India (Census
of India; 2(X)1).
Sex Discrimination and the Small Family
Norm
Average family size in India has been declining over the
past two decades, in response to a number of economic
and social changes, including rising aspirations for
children coupled with the increased costs of rearing
them, and die entr\' of large numbers of women into the
formal labor force. Such changes have taken root more
quickly among some segments of the population than
6930 Carroll Avenue, Suite 910, Takoma Park, Maryland, USA
www. genderhea Ith. org
A Less Valued Life: Population Policy and Sex Selection in India
Rupsa Mallik
others, and families of three children
remain lhe nomi in a number of stales,
including Andhra IVadesh, Bihar, and
Ullar Pradesh. The Govemmenl of India
has attempted to hasten the transition to
small families among every segment of
lhe population through population
policies and programs implemented
largely through the Indian Family
Welfare program. These strategics have
ranged
from
the
heavy-handed
approaches of the seventies and
eighties—which relied on social pressure
and outright coercion to increase
contraceptive use and reduce family
size—lo the "Target-Free Approach"
adopted in lhe mid-nineties, which was
intended lo eliminate lhe coercive tactics
that had become commonplace in lhe
rush to raise contraceptive prevalence
rates.
Over the past three years, however,
political pressure has orce again been
mounting for lhe government lo redouble
its efforts on "population control."
Today, national and state population
policies focus variously on building
volunlary support for small families
through a variety of strategies lo lhe
outright imposition of twochild families
Ihrough the use of social and economic
incentives and disincentives. In Andhra
Pradesh and Rajasthan, for example,
preferred access to housing, education,
and other needed social resources is now
given to couples thiit have no more than
two children.
In Andhra Pradesh,
Rajasthan, Madhya Pradesh, Haryana
and several other stales, laws also
prohibit individuals with more than two
children
from
contesting
local
government elections.
The shift to smaller families now evident
in India has not, however, been
accompanied by a concurrent shift in the
social and economic pressures that
underlie the preference for sons over
daughters (George; 1997). Indeed, if
anything, the pressure to have sons has
intensified
as
couples
strive
simultaneously to reduce family size
and ensure the birth of the desired
number of sons, leading lo increased
acceptance of and reliance on the use of
sex-selection strategies to achieve these
results.
Evidence of these trends has been clear
for a number of years, but neither the
national nor the state governments in
India have effectively addressed the root
causes of pervasive son preference.
Population and health policies have
focused on building pressure for smaller
families through a variety of means, but
largely have failed to address the sor il
norms that simultaneously privilege
sons over daughters, and tacitly support
the epidemic of gender violence that
afflicts women and girls throughout
their lifecycle.
The government has
failed lo effectively address persistent
gender gaps in education, employment
and access to productive resources such
as land and property. Even existing
laws, such as the Child Marriage
Restraint Act and the Dowry Prevention
Act, have been poorlv implemented, if at
all.
With the exception of UNICEF and
UNFPA, international donors also have
largely ignored the issues surrounding
the stark decline in the sex ratio.
USAID, for example, has played an
active role in the planning and
formulation of state population policies
in several slates—including Andhra
Pradesh and Uttar Pradesh—none cf
which
address
the
issues
of
discrimination,
violence,
and
sex
selection in any but the most superficial
manner. Instead, these policies take the
same simplistic approaches lo reducing
fertility in lhe short run which
exacerbate son preference over the long
run.
Civil Society Responses
Official
neglect
notwithstanding,
numerous civil society organizations
have been working on Uiis issue since the
eighties. In 1986, for example, the Forum
Against Sex Determination and Sex PreSelection (FASDSP) began a campaign to
enact legislation lo regulate llie misuse of
technologies, and subsequently played a
critical role in focusing national attention
on the issue of sex-selection abortion.
A direct outcome of this effort was the
passage ol a national law lo regulate pre
natal diagnostic technologies as well as
their misuse - the Pre-Natal Diagnostic
Techniques (Regulation and Prevention
of Misuse) Act passed in 1994. The Act
was meant to establish institutional
mechanisms at all levels of the health
system to register users of technologies,
and record complaints of violation of the
law by doctors. The law has Lwn largely
ineffective, however, as many of the
national and stale-level instilulional
mechanisms were never put in place or
have not been effectively implemenled.
Renewed efforts focused on belter
implementation are now underway In
2(XX),
several
individuals
and
organic tions- including
long-time
activist
Sabu
George,
and
two
Maharashtra-based advocacy groups,
CEHAT and MASUM—filed public
interest litigation in the Supreme Court of
India seeking lo ensure effective
implementation of the existing PNDT
Act. In response, the Court recently
ordered the national and stales level
health secretaries lo impound ultrasound
machines in unregistered clinics, and lo
file comprehensive affidavits wilh the
court detailing all other actions taken lo
effectively ensure implementation of the
law. The lawsuit, court decisions, and
release of the 2001 census data showing
further declines in the juvenile sex ratio
together have generated further media
interest in SD and SSA, raising these
issues once again to the level of national
concern.
The national government also established
a technical committee within the Ministrv
of Health and Family Welfare to review
and make recommendations for better
implementation of the existing PNDT
Act,
including
those
aspects
of
unregulated use of pre-conception
techniques for sex selection that remain
outside the ambit of the existing law. This
committee proposed an amendment,
titled Pre-Conception and Pre-Natal Sex
Selection/ Determination (Prohibition and
Regulation) Act, 200T2 preventing use of
pre-natal diagnostic techniques for sex
determination, banning use of pre-
2
Center for Health & Gender Equity 301-270-1182 fax 301-270-2052
6930 Carroll Avenue, Suite 910, Takoma Park, Maryland,. USA
www.genderhealth.org
A Less Valued Life: Population Policy and Sex Selection in India
Rupsa Mallik
conception techniques for sex selection,
and selling sl ndards for lhe use of
ultrasound lo monitor pregnancies (as is
lhe norm in most countries).
The
amendment, however, has yet lo be
pa.ssed, in part because of resistance bv
lhe medical community itself, a portion af
which profits substantially from the
increased use of these technologies.
Profiting From Bias
While the government has as yet largety
failed lo effectively address these issues,
lhe private seclor has sought lo exploit
them for profit. The use of genetics
Lesling
and
olher
reproductive
technologies for lhe purpose of sex
selection has become a thriving industry
in many parts of lhe country, one lhal is
direclly implicated in lhe rapid and
unregulated spread of reproductive
teclinologies used for sex determination
and selective abortion.
Doctors and
quacks alike have cleverly if insidiously
used advertisemenls and other means of
communication
lo
market
these
technologies as a means of expanding
reproductive choices for women, playing
simultaneously on lhe negative norms
and beliefs thal underlie gender bias
within Indian s<xnely and on lhe broader
movement lo secure reproductive rights
for Indian women.
In pari as a result, public debales
regarding sex selection have been linked
with the right lo access to safe abortion in
ways that actually threaten women’s
access to pregnancy termination services
over the longer term. On one hand, some
associated with lhe private seclor have
argued lhat offering women the option of
abortion for lhe purpose of sex selection
needs io be viewed within lhe framework
of women's autonomy and right to safe
abortion services. On lhe other, some
opponents of sex selection have
attributed its spread to India's 'liberal
abortion
laws,
proposing
greater
restrictions on access to early and safe
abortions as lhe remedy. Neither of these
positions addresses lhe issues in a way
lhe ultimately safeguards women's rights
while simultaneously addressing the root
causes of this phenomenon.
Meanwhite, lhe altitude of lhe larger
medical community with regard lo
banning
sex
selection
remains
ambiguous.
Widespread protests
erupted, for example, when lhe
Government of India proposed the 2001
amendment lo lhe existing PNDT law.
Some of the proposed changes in lhe
existing PNDT Act include compulsory
maintenance of written records by
providers of pre-natal diagnosis, a
requirement lhat has been severely
criticized by the medical community.
This and other regulations have been
contested in a lawsuit filed by lhe Delhi
Medical Association.
Changing Norms at the Local
Level
on lhe issues, seeking belter enforcement
of lhe PNDT Act but also conducting
public education on lhe issues.
The state-level Campaign against SexSelection Abortion in Tamil Nadu and
Voluntary Health Association of Punjab
are other notable examples of efforts that
have used diverse strategies, including
meeting
with
religious
leaders,
organizing
protest
marches,
and
reporting unregistered clinics and
practitioners lo authorities at lhe district
and stale levels. The Indian Medical
Association and National (.Commission
on Women have collaborated in some of
these efforts largely as part of a
UNICEF-funded initiative.
Addressing Bias at the Source
While lhe legislative and policy issues are
debated, efforts are in fact being made in
some communilies lo change llae
altitudes and behaviors underlying son
preference and violence against women
and girls. Numerous conununity-based
organizations
(CEOs),
and
non
governmental organizations (NGOs)
have sought to address these issues on
lhe ground in both urban and rural
areas. In Tamil Nadu, for example, a
coalition of organizations (including lhe
Indian Council for Child Welfare, lhe
Community
Services
Guild
and
Alternatives for India Development) is
working lo change both norms and
behaviors in a number of districts that
show both a high prevalence of female
infanticide and an increase in relian -e on
sex-selection abortion. The efforts focus
on mobilizing conununity leaders lo
counter the practice of female infanticide
and feticide, often using integrated
women’s development strategies as a
way lo address lhe scxdo-cuHural and
economic roots of the problem. Their
strategies have included a focus on
educating adolescent girls and women;
forming self-help groups lo. increase
women's access to credit and paid
employment; building solidarity among
women within these communities; and
changing the altitudes of youth toward
social practices like down’ and
discrimination against girls. In addition,
they have formed networks to campaign
As is evident, there are numerous efforts
underway at the national and state level
within India intended to influence
opinion and take action against sex
selection abortion. With the exception of
local organizations, however, much of
what
is
happening
focuses
on
addressing the symptoms, rather than
the longer-term steps needed to attack
gender bias al its roots. Moreover, the
disconnect between the problem and the
government's own response is no where
more evident than in the wav
contemporary population policies seek
to enforce a two-child norm, in spile of
growing evidence that doing so in the
absence of concerted efforts to address
such bias often leads to an increase in
practices like female infanticide and
feticide.
Yet there is much tliat could be done to
combat lhe spread of discriminalion
against
girl
children,
including
infanticide and sex selection. Among
those sleps that should be taken are the
following:
Establishment of a permanent and
autonomous
commission
on
reproductive and genetic technology,
including
representation
from
government,
medical
associations,
research institutes, and civil society
organizations
with
long
3
Center for Health & Gender Equity 301-270-1182 fax 301-270-2052
6930 Carroll Avenue, Suite 910, Takoma Park, Maryland, USA
www.genderhealth.org
A Less Valued Life: Population Policy and Sex Selection in India
Rupsa MalliL
established work in this area: Such a
commission should establish regulations
governing standards of care and
monitoring
of
clinics
providing
reproductive technologies, including
newer technologies as they become
available. Similar commissions have
been
established
to
regulate
reproductive technologies in both
Canada and UK with positive results.
Effective implementation of laws and
policies: The government must act to
ensure
adequate
and
effective
implementation of the PNDT Act, as
well as a wide-range of laws and
regulations that address gender inequity
at
different
levels
of
socieh-.
Enforcement of related laws is essential,
and should include enforcement of the
Child Marriage Restraint Act, the Dowry
Prevention Act, and the various
provisions in family law guaranteeing
equal rights to property and inheritance
for
daughters.
Efforts
must
simultaneously be made to establish
effective laws and policies regarding
gender violence, including domestic
violence and coercion.
Creation of gender and rights-based
population policies and programs and
multi-sectoral strategies to address
gender bias:
Current population
policies ignore the gender dimensions of
reproductive
decision-making,
and
thereby actually exacerbate practices like
sex selection. To date, for example, the
only national and state programs
intended
to
directly
address
discrimination against girls have been
those providing cash incentives to
families that have girl children. These
include lump stun deposits made by the
government in the name of a girl child to
be made available to her when she
reaches age 18. This strategy has been
criticized for many reasons, including
because it appears to sanction dowry by
providing a cash savings used by
parents to subsidize their dowry
payments. Moreover, the program has
been dropped in some stales where
governments were unwilling to allocate
the resources necessary to sustain it.
In another innovation,' the stale ol
Tamil Nadu pul out cradles in health
centers intended to enable parents to
leave unwanted girls instead of killing
them.
In the absence of efforts to
address the deeply rooted economic and
social biases against women and girls,
however, these steps have had little if
any effect on the practices of female
infanticide and feticide in those states
where the problems are greatest. The
national and slate governments need to
focus
instead
on
simultaneous
implementation of a range of programs,
including mid-day meals for school
children, community level childcare, and
educational opportunities for girls
forced to drop out of school to care for
younger siblings or work in the field.
Moreover, high priority must be placed
on increasing access to primary
education, and increased access for
women and girls to wage employment,
land and other productive resources,
issues that have received much
rhetorical but little practical attention.
Land reform and redistribution policies
intended
to
increase
women's
inheritance and ownership of land are
on the books in many stales, for
example, but are not implemented
despite the fact that the desire to retain
undivided control of land through sons
has been directly linked to an increase in
sex determination and sex selection in
several states, including Haryana and
Punjab.
Only by undertaking these and other
concerted strategies can the government,
donors, and civil society can begin to
address the issue of female infanticide
and
sex-selection abortion
in a
meaningful way. Given what is al slake,
there is no lime to lose.
Rupsa Mallik is a Program Associate al the Center lor Health and Gender Equity. Correspondence
about the paper should be directed to Rupsa Mallik <rinallik@genderheaith.org> or Jodi Jacobson
<iiacobson’3genderhcalth.org>. For additional copies, send ait email to <info-@gcndcrhcaltlt.org>.
All rights reserved by the Center for Health and Gentler Equity. No part of tins document may lx*
rcproduceit disscmiiwtcd, published, or transferred, except with prior permission and appropriate
acknowledgment of the Center for Health and Gender Equity. Suggested citation: Mallik, Rupsa.
A ls.ss Valued Life.: Population Polia/ and Sex Selection m India. (Takotna Park, MD: Center for Health
.....i
i.... c...onnot
A Less Valued Life: Population Policy and Sex Selection in India
Rupsa Mallik
Bibliography and Web-Site Resources
1 http;// www.censusindia.net/
2 http://mohfw.nic.in/PNDT%20Amendments.htm
Forum Against Sex Determination arid Sex Pre Selection (FASDSP). 1992. 'Using Technology, Choosing Sex.’
Development Dialogue. Uppsala: Dag Hammarskjold Foundation. 1&2;91-102.
Cieorge, S.M. 1997. 'Female Infanticide in Tamil Nadu: From Recognition Back to Denial?' ^productive Hi illh Mailers.
November. 10:124-32.
Ckiodkind, D. 1996. 'On Substituting Sex Preference Strategies in Fast Asia; Does Pre-Natal Sex Selection Reduce PostNatal Discrimination?' Population and Development Rcvieio. 22( l):lll-25.
Kundu, A. and Sahu. M.K. 1991. 'Variation in Sex Ratio: Development Implications.' Economic and Political Weekly.
MumbakSameeksha Publications Trust. October 12.XXVI(41):2341-42.
Menon, N. 1995. 'The Impossibility of "Justice": Female Foeticide and Feminist Discourse on Abortion.' Contributions
to Indian Sociology. New Delhi; Sage Publications. 29(l&2):369-92.
Mutharayappa, R., Choe, M.K., Arnold, F. and Roy, T.K. 1997. 'Son Preference and Ils Effect on Fertility in India.'
National Painily Health Survey Subject Reports. No. 3, March. Mumbai: International Institute for Population Sciences.
Oomman, N, and Ganatra, B; et al. 2(X)2. 'Roundtable on .Sex Selection.' Rqnvductive Health Mailers. May. Volume
10(19);184-197. UK: Elsevier Science.
Ravindra, R.P. 1987. 'Struggle Against Sex Determination Techniques - Unfinished Battle.' Economic and Political Weekly.
Mumbai: Sameeksha Trust I’ublication. March 12. XXII(12):490-92.
1991. 'Fighting Female Foeticide - A Long Way To (kx' The lawyers. Mumbai: SNDT. August 6(8);4 -Tl.
1993. 'The Campaign Against Sex Determination Tests,' pp 51-99, in Datar, C (ed.) Struggle Against Violence.
Calcutta: Stree Publications.
1995. 'Myths About Sex Determination Tests.' Pacts Against Myths. Mumbai: Vikas Adhyayan Kendra. June.
2(3):l-6.
Sudha, S. and Rajan I.S. 1999. 'Female Demographic Disadvantage m India 1981-1991: Sex Selective Abortions and
Female Infanticide.' Development, and Giange. Oxford: Blackwell Publishers. July. 30:585-618.
Weiss.C. 1995. 'Sex-Selective Abortion: A Relational Approach.' Hypatia. Bloomington: Indiana Universih' Press. Winter.
12(3).
C'omprehensive compilation of articles on sex selection in India:
http:/ / www. hsph.harvard.edu/ grhf/SAsia / library
5
Center for Health & Gender Equity 301-270-1182 fax 301-270-2052
6930 Carroll Avenue, Suite 910, Takoma Park, Maryland, USA www.genderhealth.org
Position: 4642 (1 views)