Overlapping Identities under Liberalization: Gender and Caste in India

Author(s): Ashwini Deshpande

Source: Economic Development and Cultural Change , Vol. 55, No. 4 (July 2007), pp. 735-760
Published by: The University of Chicago Press
Stable URL: [Link]

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@[Link].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
[Link]

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access
to Economic Development and Cultural Change

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
Overlapping Identities under Liberalization: Gender and
Caste in India

ashwini deshpande
Delhi School of Economics

I. Introduction
There is a large literature on gender disparities in India that examines the
phenomena of the declining sex ratio, educational disparities, differential labor
market outcomes between men and women, health outcomes for women (with
a special focus on reproductive health), and so forth.1 Similarly, the caste system
has been widely researched by all of the social sciences except economics;
economic analyses of caste inequality have been few (see Deshpande [2000]
for a review of the early literature). The literature on the overlap between
caste and gender is even smaller, and within that literature, economic analyses
of this overlap are very few indeed. Elsewhere I have discussed the contribution
that economists might make to analyses of intergroup disparities; I now con-
tend that an analysis of the gender-caste overlap from an economic perspective
would be a useful contribution to the debate over identity and economic
outcomes, particularly as related to a liberalizing economy.
In previous papers (Deshpande 2001b, 2002), I have tried to quantify some
aspects of the gender-caste overlap in India. One important issue that needs
to be restated in the context of this overlap is that of the assumed trade-off
between greater material prosperity and greater immurement (of upper-caste
women), on the one hand, and lower taboos on public visibility and material
deprivation (of low-caste women), on the other hand. I have argued (Deshpande
2002) that the trade-off is vanishing; in contemporary India, low-caste women
suffer the double burden of material deprivation and immurement and thus
are the worst off in both respects.
The quantification of the intersection between caste and gender, rather than

An earlier version of this article was presented at the International Association for Feminist Eco-
nomics (IAFFE) Conference, Oxford, July 2004, and at a seminar at the South Asia Program at
Cornell University in September 2004. I thank the participants in both places and also two
anonymous referees and the editor, John Strauss, for their comments. The usual disclaimer applies.
1
See, e.g., Dreze and Sen (1995), Duraisamy and Duraisamy (1999), Probe Team (1999), Dreze
and Murthy (2001), Kingdon (2002), and Bajpai and Goyal (2004).

䊚 2007 by The University of Chicago. All rights reserved. 0013-0079/2007/5504-0001$10.00

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
736 economic development and cultur al change

being confined to labor market outcomes alone, is done more broadly using
the Gender-Caste Development Index (GCDI). I constructed this index in a
previous exercise (Deshpande 2001b) and then made calculations using Na-
tional Family and Health Survey data from 1992–93 (NFHS-1). Another round
of data is now available from the same survey for 1998–99 (NFHS-2), thus
enabling the recalculation of the index after a gap of 6–7 years. These years
are crucial in terms of Indian economic development, since India embarked
on a path of relatively full-fledged liberalization of the economy starting in
1991. Thus, a reevaluation of the index based on NFHS-2 can help us draw
some preliminary conclusions regarding the movement of the gender-caste
overlap under liberalization. I also separately discuss two components of this
index, namely, education and work/occupation. This article ends with con-
cluding comments.

II. Does Identity Matter to the Market?

While the NFHS data set does not allow us to capture economic discrimination
directly (using the standard labor market outcomes approach), there are com-
pelling reasons to believe that the material deprivation/advantage of social
groups reflects the underlying processes of discrimination/nepotism. Even at
a theoretical level, the social identities of agents have been seen as central to
the determination of their economic outcomes. The two main neoclassical
theories, “statistical discrimination” (e.g., Akerlof 1984) and “taste for dis-
crimination” (Becker 1967), approach discrimination from opposite ends. The
first one ascribes discrimination to the presence of imperfect information in
labor markets; here employers use group identity as a proxy for ability, and
discrimination occurs because of economic incentives. Becker, in contrast,
argues that individuals discriminate because they have a taste for doing so
and that they will do so even if it means a loss of income for them. Thus,
individuals will discriminate in markets despite facing economic incentives
for behaving otherwise. The strength of Becker’s theory is that it addresses
prejudice as a direct component of individuals’ utility functions. These theories,
powerful as they are, do not address “premarket” discrimination. A lot of the
empirical work on gender disparities in India focuses on premarket factors,
including education, nutrition, and even certain health outcomes.
An alternative approach to the established neoclassical theories is provided
by the work of feminist economists, whose primary concern is gender identity.
In contrast, the neoclassical theories could be used to address discrimination
based on any definition of identity (gender, caste, religion, nationality, etc.).
The feminist argument is that markets are not neutral arenas but in fact are
“gendered” institutions. One of the ways in which markets are gendered is

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Deshpande 737

the manner in which they operate at the intersection of the “productive and
the reproductive” economies and disadvantage those who work in the repro-
ductive economy, that is, women (Elson 1999).
These studies, as well as other theoretical studies and a host of case studies
across the world, firmly suggest that identity matters in the real world, even
in strongly market-oriented situations. What happens when there are over-
lapping identities? There are several contexts where identity has multiple
contours, and every individual simultaneously has overlapping identities. In-
deed, the overlap between race and gender has been studied in the context of
other societies (Sokoloff [1992], Ayers [2001], and Lovell [2003], to mention
a few). In the current article, I focus on two important definitions of identity
in the Indian context, caste and gender, while being acutely aware that there
are several other crucial ones that I am not considering simply in order to
retain expositional clarity. When overlapping is such that an individual is
disadvantaged according to one identity (gender) but not according to another
(caste), then the result is a little more complicated. However, given the van-
ishing trade-off mentioned earlier, to reiterate, the intersection of caste and
gender in India suggests that those who are disadvantaged according to both
identities, namely, low-caste women, bear the brunt of the burden of disad-
vantage (Deshpande 2002). It needs to be stated that acceptance of the van-
ishing trade-off is not critical. Even if one believes in the persistence of the
trade-off, we could hold one identity constant and vary the other–that is, look
at castes within women–and we would arrive at the same conclusion: low-
caste women are at the bottom of the hierarchy. Elsewhere (Deshpande 2004)
I have examined the relationship between “within-caste” and “between-caste”
inequality using the Theil Index, and while clearly within-caste inequality
exists, there are strong reasons to focus on “between-caste” differences.

III. The Gender-Caste Development Index

One way to assess discrimination or the impact of identity on economic
outcomes is to look at wage/income and/or employment outcomes. This pro-
cedure is important and has shown significant results whenever it has been
undertaken.2 However, because of the underestimation of women’s work (dis-
cussed briefly below) and the inadequacy of wage data in developing countries,
as well as because wage data relate to only one aspect of women’s lives, we
can think of alternative ways of examining identity and economic outcomes.
Inspired by the Human Development Index (HDI), I first constructed a Caste
2
See, e.g., Tilak (1980), Kingdon (1997), Singh and Bhattacharjee (1998), Deshpande and Desh-
pande (1999), and Duraisamy and Duraisamy (1999).

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
738 economic development and cultur al change

Development Index (CDI; see Deshpande 2001a) and later constructed the
GCDI in order to quantify intergroup disparity based on gender and caste.
Since the NFHS data have no information on income or consumption, it is
necessary to construct some measure of standard of living in order to examine
the differences in material conditions across caste groups. The index is based
on five variables: land ownership, education, occupation, livestock, and con-
sumer durables.3 However, what initially started as a necessity, construction
of an index in the absence of income data, later turned out to be valuable in
itself, in that it is a more broad-based measure than income alone.4
The NFHS-2 independently realized the necessity of having a standard of
living index (SLI), which it created, using several other variables as well (NFHS
1998–99). Of course, since the SLI was not available for the first round, it
cannot be the basis of comparison over time. Thus, the basis of comparison
for the current study is the GCDI that I have calculated for the two rounds.
Having said this, the HDI suffers from certain well-known drawbacks, for
instance, shifting goalposts. For comparisons over a long period of time, this
could be a drawback that would need to be settled. However, for this particular
two-point comparison, the maximum and minimum values do not change
(see n. 4 for the formula of the index), and thus in this specific exercise the
shifting goalposts problem does not arise.

3
The NFHS data set has rich information on health outcomes that potentially could be used in
the construction of the GCDI. However, inclusion of health variables is not as straightforward as
it appears at first glance. I was not able to find an unambiguous and precise way to blend these
health indicators into a socioeconomic index. Consider, e.g., the sex ratio. It is not clear how exactly
the sex ratio could be used an indicator of socioeconomic status. In India, richer neighborhoods
(there are studies of New Delhi) and richer states (Punjab, Haryana) have both been found to have
adverse sex ratios, but then some of the poorer states (like Uttar Pradesh) also have adverse sex
ratios. Use of the body mass index (BMI) or the infant mortality rate (IMR) has other problems.
A BMI below a critical value (and an IMR above a critical value) indicates extreme poverty, but
it is not clear that above this threshold values of BMI are indicators of economic status. I had, in
fact, considered other variables, such as the incidence of diarrhea, but these were subject to similar
conceptual problems. Thus, I decided to stick with using standard economic variables.
4
Following the UNDP methodology, the caste deprivation index is calculated as follows.
Icij is the deprivation indicator for the jth state with respect to the ith indicator for caste
group c (c can take three/four values: SC, ST, OBC, Others, as the case may be) and is given
by Icij p (max Xij ⫺ Xijc )/(max Xij ⫺ min Xij), where the maximum (minimum) values are those
that the given indicator takes from the all-India data. Thus, in the entire country, if 12 is
the maximum number of assets that an individual owns, then the max value for assets is 12,
and that will be used for calculation of the index in all states. Xcij is the mean value of the
ith indicator for the cth caste group in the jth state. The caste deprivation index is an average
of the five deprivation indicators, calculated as above. Thus, Icj p summation over i[Icij]/5, for
the cth caste group in the jth state. The CDI p (1 ⫺ Caste Deprivation Index). When this
index is calculated for men and women separately, this can be called the Gender Caste
Development Index (GCDI), since it enables a gender-based comparison of caste groups.

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Deshpande 739

Unfortunately, there are problems of comparability across the two rounds

of the NFHS. As often happens with large data sets, variables have been
dropped, added, or simply changed to address the lacunae in the first round.
Thus, while the changes in the second round are an improvement, they make
strict comparability with the first round difficult. In the context of the GCDI,
for instance, an entire set of questions based on livestock ownership is absent
from the second round: the only question that is asked is “Do you own any
livestock?” The yes/no responses to this question can hardly be compared with
“the number of animals owned” that NFHS-1 collected information on. An-
other constraint imposed by the data is the use of “total landholding,” which
typically conceals the wide variation in land quality across the country. A
broad categorization indicating the relative fertility of land would be a better
indicator of standard of living in a country where agricultural activities con-
tinue to be very important.
Regarding the occupations of respondents, NFHS-2 has a much more de-
tailed disaggregation than does NFHS-1. The total number of occupations
listed in the former was close to 100, whereas the latter had only broad
occupational groups. NHFS-2 further groups these various occupations into
broad categories, but these again are different from (and more detailed than)
the broad categories of the first round. I have rearranged the broad categories
into the same number of categories as the first round, while roughly attempting
to match the descriptions of the first round. However, it is highly likely that
the result is not strictly comparable in that, while I have made the number
of categories the same across rounds, the occupation groupings are not identical.
Finally, there is one important change in the caste categories. The first
round collected information on SC (Scheduled Castes), ST (Scheduled Tribes),
and all others. The second round has information on “Other Backward Castes”
(OBCs), and thus the erstwhile “Others” category is now divided into OBC
and Others.5 This division is characteristic not just of the NFHS but also of
all major data sets for India after the mid-1990s, for instance, the National
Sample Survey. This is linked to changes in the caste-based affirmative action
program in that affirmative action was to be extended to a group of nonun-
touchable castes that were grouped together as OBCs. The formulation of the
OBC lists, which are finalized separately for each state, has been highly con-
tentious and has been fraught with fierce lobbying. This has often resulted
5
The former untouchable castes have been identified in a government schedule as beneficiaries
of affirmative action and are hence called Scheduled Castes. In additon to the caste system,
India is home to several tribes that have been similarly designated (Scheduled Tribes). Other
Backward Castes are supposed to be castes that were not untouchable but that are very low
in social and economic position. “Others” includes everyone else.

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
740 economic development and cultur al change

in a situation where a particular caste is considered to be an OBC in one state

and not in another, rendering strict comparability across states difficult.
However, the sample size of the two rounds is comparable, as is the geo-
graphical coverage. NFHS-1 had a sample of about 89,000 women from all
states of India, while NFHS-2 has about 90,000 women respondents. However,
for reasons explained above, the absolute values of the GCDI are not strictly
comparable across the two rounds. Thus, the best we can do is examine the
shifts in gaps across groups and then examine some key variables of the index.
Figure 1 shows the GCDI for SC, ST, OBC, and Others for NFHS-2 for
the major states of India.6 Several features stand out in this figure. First, for
all the states, the women in the Others category are the best off and the SC-
ST women are the worst off in terms of the GCDI. This is not trivial, since
the “Others” category is a large, heterogeneous category that is essentially a
residual one, and it contains several castes that are not very distinct from SC
or OBC castes in terms of standard of living. In no state for any of the caste
groups does the value of the index cross 0.5 (the halfway mark). Thus, while
overall poverty continues to be a challenge (indicated by the low values of
GCDI in this exercise, as well as based on evidence from other data sets), the
fact of intercaste disparity needs to be noted in particular.
The past couple of decades have seen a rise of OBC politicians across all
political parties such that key ministers, important party functionaries, and
several chief ministers of states are OBCs. There is some evidence to suggest
that, at the village level, administrative units are dominated by OBCs. This
has given rise to the notion of a “powerful OBC section having overtaken the
upper castes,” or “OBCs as the new elite.” This belief has two consequences.
First, this leads to a corollary assumption, namely, that this increased political
influence must have translated into increased economic might. This formu-
lation could mean that the gap between OBC and Others (broadly upper-caste
Hindus) is either closing or even reversed. While this article cannot comment
on the political changes, besides being handicapped by the lack of figures on
the exact distribution of SC-OBC-Other politicians, the GCDI suggests that
the gap has certainly not reversed.7 (Since the OBC division only started with
6
The northeastern states are not shown as they are majority tribal states (i.e., majority ST)
and therefore have a distinctive social composition. I believe that they should be analyzed
separately. Also, some northern states, such as Delhi, Punjab, and Haryana, have very low
ST proportions in the state sample, and hence the ST bar for the GCDI is missing for these
states.
7
The only exception to this is the eastern state of West Bengal, where the GCDI for OBC
women is higher than that for Other women. Also, in Maharashtra, we see no gap between
OBC and upper-caste women. This requires further research, which is currently under way.

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Figure 1. Gender-Caste Development Index, India, 1998–99

741

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
742 economic development and cultur al change

the second round, we cannot comment on the changes in the gap.) In fact,
in a state like Bihar, which had an OBC woman chief minister, the gap is
substantial.
The other consequence of this notion is the argument that affirmative action
for the OBCs is misplaced: Some of these castes might have been “backward”
at some stage but that this is no longer true in contemporary India. Thus,
the OBC insistence on affirmative action is nothing but a result of strong
political lobbying. While this article does not go into issues concerning af-
firmative action, and while it is also true that specific OBC castes are indeed
rich and powerful, the GCDI suggests that the notion of the economic might
of the OBCs does not hold in the aggregate. Upper castes continue to be
economically the elite.
An argument is made—and it has a lot of merit—for unpacking the SC-
ST combine since not all STs are Hindus, and while STs have been on the
margins of development, they have not been subjected to the same kind of
“stigmatized ethnic identity” (Thorat 1979) as the SCs have. If one does such
a separation, one can see a much more regionally diverse pattern in terms of
the relative ranking of the SCs versus the STs. But that need not detain us
here. The main noteworthy feature is that, in terms of the GCDI, SCs lag
behind OBCs, who, in turn, lag behind the upper-caste women for most of
India.

IV. Health and Nutrition

The GCDI is based on indicators of material standard of living. However, the
NFHS has a large number of questions related to the reproductive health of
women. While there is no straightforward way of integrating health and
nutrition data directly into the GCDI, a separate assessment of some of the
indicators reinforces sharp intercaste divisions. For instance, infant mortality
rates (IMRs; per 1,000 live births) differ markedly by social groups: 83 for
SCs, 84.2 for STs, 76 for OBCs, and 61.8 for Others. These differences can
be seen both in rural and urban India, but rates for urban India (60.4, 57.6,
51.2l, and 43.5, respectively) are significantly lower than for rural India (88.1,
86.9, 82.2, and 69.3, respectively). Similar disparities can also be seen in
under-age-5 child mortality (119.3, 126.6, 103.1, and 82.6 [per 1,000 live
births], respectively). Regional variation in IMRs indicates that Kerala has
the lowest IMRs for all groups. States that have higher than all India IMRs
are Assam, Bihar, Rajasthan, Orissa, Madhya Pradesh, and Uttar Pradesh.
For the children who survive, immunization records vary markedly among
the social groups. Percentage of children ages 12–23 months who received
specific vaccinations at any time before the interview is the highest for Others,

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Deshpande 743

followed by OBCs, SCs, and, finally, STs. Conversely, the percentage of children
who did not receive any vaccination is the lowest for the Others and the
highest for the STs. These differences are significant for all diseases (BCG
[tuberculosis], DPT [diptheria, whooping cough/pertussis, and tetanus], mea-
sles) except polio, for which the differences might be insignificant because of
the government’s massive and vigorous program of free and compulsory im-
munization in order to eradicate the disease. Similarly, the proportion of chil-
dren (ages 12–35 months) who received at least one dose of vitamin A is 27.1
for SCs, 26 for STs, 26.8 for OBCs, and 34.8 for Others. While among children
the prevalence of acute respiratory infection, fever, and diarrhea does not differ
significantly by caste, suggesting that these diseases are ubiquitous, interest-
ingly, awareness of the symptoms and treatment of diarrhea differs significantly
among the castes, with Others being the most aware and STs the least aware.
This reflects a larger pattern of awareness or the lack of it, which, for instance,
can be seen in the percentage of respondents who had heard about AIDS (48.4,
Others; 41.9, OBCs; 32.4, SCs; 17.2, STs).
Women’s nutritional status, measured by food consumption of specific foods
at least once a week (milk/curd; pulses/beans; green, leafy vegetables; other
vegetables; fruits; eggs; chicken/mutton/fish), not surprisingly, reflects similar
disparities to material standard of living, with the proportion of women having
consumed specific foods being the highest among Others and the lowest among
STs. The proportion of women consuming fruit at least once a week is the
lowest among all the food items specified (24.5, SC; 20.9, ST; 33.5, OBCs;
39.7, Others), suggesting that women have very little access to items of
consumption that are expensive and that possibly are considered nonessential.
Consequently, the proportion of women who have a BMI of less than 18.5
(chronic energy deficiency) is much greater among SC and ST women than
among Other or even OBC women. Similarly, prevalence of anemia is much
higher among SC and ST women (56% and 64.9%, respectively) than among
OBCs (50.7%) and Others (47.6%). Disparities based on the health and nu-
trition status of women and children mirror the picture that the GCDI reveals.

V. Regional Variation in GCDI

The other distinguishing characteristic of GCDI is its regional variation. The
contemporary division of states in India is a postindependence phenomenon
that is a result of a linguistic reorganization of the provinces under British
India. Each province had a distinct history and trajectory of development—
economic, social, and cultural. It is useful to think of India as what Europe
would have been if it had been a single country. A complete understanding
of the complexities of regional diversity thus becomes a particularly challenging

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
744 economic development and cultur al change

task; this necessarily extends beyond economics and therefore is outside the
scope of the current article. However, a few comments are offered.8
The economic literature on regional differences in India mainly focuses on
the examination of whether regional inequality is increasing or decreasing or
on an account of differential growth rates between states or on a discussion
of state-level human development indicators (HDIs). (See, e.g., Jha 2002; Singh
et al. 2003; Bajpai and Goyal 2004; Milanovich 2004.) Most of the literature
seems to suggest that there has been a strong to moderate increase in regional
inequalities since the beginning of the 1990s, but the explanations for the
increase are varied. Lopsided regional development is, to a certain extent, a
legacy of the British rule in India, as industrial growth was concentrated in
the richer and more accessible regions. Agricultural growth varied due to
climatic factors in any case. Postindependence, during the heyday of the plan-
ning era, lip service was paid to the lowering of regional inequalities. Attempts
were made by organizations, such as the Finance Commission, to use the
instrument of federal transfers to states to alleviate regional imbalances. How-
ever, with the declining role of the government in the economy, conscious
interventions to reduce regional imbalances have declined, and one of the
arguments in the literature is that regional inequalities have increased after
1990.
The GCDI broadly, though not completely, corresponds to the regional
variation found in, for instance, values of the State Domestic Product (SDP).
Comparing the values of GCDI with SDP figures, we find broad consistency
in the two sets of figures. The richest five states (Delhi, Haryana, Punjab,
Maharashtra, and Gujarat) also have high GCDIs for the “Others” category.
Two noteworthy exceptions to this are Tamil Nadu and Madhya Pradesh,
where the GCDI for the “Others” category is also high. In terms of real per
capita SDP, in 1999–2000, Tamil Nadu was in the top half of the distribution,
but Madhya Pradesh ranked below the national average. Thus, while the
correspondence between GCDI and SDP cannot be expected to be exact given
8
While there is undoubtedly awareness of regional diversity, quantification of regional dif-
ferences is a recent attempt. As a survey in a popular weekly, India Today (“The State of the
States: India’s Best and Worst States,” August 16, 2004, [Link]
20040816/[Link]) revealed, how the states get ranked varies enormously by indicator,
and there is not enough clarity about the explanation of the rankings. To quote from the
survey: “India’s top-ranked states display amazing variety. Some of India’s richest people live
in Punjab, one of the most bankrupt states. Kerala is socially most advanced but has the
highest flight of human capital. Sikkim and Mizoram are among India’s most picturesque
and healthy states, but least accessible. Delhi has drawn the maximum investment when the
Central government has been shedding its powers. Some winners are being propelled by the
past, some have driven themselves to the peak by turning around in recent years.”

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Deshpande 745

that they do not measure identical variables, the broad pattern appears to be
consistent.
Since it lies between zero and one, the GCDI can easily be used to calculate
intercaste disparity. Thus, for instance, [GCDI for Others] ⫺ [GCDI for SCs]
can be taken to be a measure of upper-cast and low-caste disparity. Using this,
we can define three measures of intercaste disparity: Upper Caste ⫺ SC, OBC
⫺ SC, and Upper Caste ⫺ OBC. Figure 2 shows the regional distribution of
these three measures of disparity, as well as the Others ⫺ SC for NFHS-1 (in
fig. 2 this is referred to as Oth ⫺ SC1).
The “Oth ⫺ SC2” refers to the disparity between upper-caste and SC women
for NFHS-2. States where the “Others ⫺ SC” disparity has increased over the
two rounds of the NFHS are Delhi, Himachal Pradesh, Punjab, Uttar Pradesh,
Madhya Pradesh, Bihar, Orissa, West Bengal, Gujarat, Karnataka, and Tamil
Nadu. States where disparity has decreased are Haryana, Jammu, Rajasthan,
and Maharashtra. The five states with the highest disparity in 1992–93 were
Rajasthan, Haryana, Punjab, Tamil Nadu, and Delhi, in that order. For
1998–99, they were Tamil Nadu, Punjab, Haryana, Delhi and Gujarat, in
that order. It is noteworthy that the latter group consists of the relatively
richer states. However, if one looks at the BIMARU states, with the exception
of Rajasthan, disparity in the other three has increased.9
Thus, this division of states follows no apparent pattern, neither geographical
nor based on state domestic product (SDP) levels, as both rich and poor states
have experienced an increase in disparity. This lack of a clear association
between caste disparities and SDP levels and the rate of growth of SDP for
states suggests that caste disparity might be an independent issue that would
need to be tackled as such. Other studies confirm the fact of heterogeneity of
states and the lack of a clear correlation between SDP and human development
indicators.

VI. Liberalization and GCDI

As mentioned in the introduction, the two time periods of the NFHS are
crucial in that they mark the early years of liberalization in India. Figure 3
shows the GCDI for the first round.
Comparing figures 1 and 3 for GCDI, we see that, on the whole, the values
for the index are higher in 1998–99 than in 1992–93, indicating an im-
provement in the overall standard of living. For SC women, the top five states
9
Bihar, Madhya Pradesh, Rajasthan, and Uttar Pradesh, states in the Hindi-speaking planes,
are referred to collectively as BIMARU (literally, sick) as they share common problems of
social and economic underdevelopment.

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Figure 2. Caste disparity, 1998–99 and 1992–93

746

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Figure 3. Gender-Caste Development Index, India, 1992–93

747

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
748 economic development and cultur al change

in 1992–93 were Himachal Pradesh, Jammu, Delhi, Kerala, and Gujarat,

while for 1998–99 they were Kerala, Maharashtra, Delhi, West Bengal, and
Gujarat. The pattern of disparity discussed in the previous section can also
be seen clearly in figure 3, and, in combination with the discussion on dis-
parities, this suggests that early years of liberalization (i.e., 1992–99) have
not started to close the gaps. Once again, it needs to be pointed out that this
is not a trivial observation. Because caste and gender oppression are rooted in
an antiquated tradition, there is a serious belief that an opening of the economy
and its integration with the world economy, entry of foreign multinational
companies that are outside the framework of this tradition, and greater pen-
etration of the market can serve as catalysts of change. “Foreign” market forces,
the argument goes, are more likely to be guided by the expediency of profit
maximization than to be mired by local constraints; thus, these can actually
serve as “liberating” influences. This view cites, in its support, the period
under British rule, where mass education was for the first time made available
to untouchables, who were for centuries denied education on the grounds that
the scriptures permitted only the upper castes to be educated. Exposure to
education, to English, and to the subsequent opening of channels of employ-
ment that would have been unimaginable earlier prompted some important
Dalit leaders to endorse colonial rule or at least to see some positive aspects
of that period.10
Similar arguments are invoked for the current period as well. However,
evidence from rich and strongly market-oriented economies suggests that dis-
crimination and “modernity” can coexist. In the Indian context, both under
the British rule and in the present experience with globalization, foreign
interests may find it easier to adjust to, and hence perpetuate, preexisting
patterns of inequality rather than to break them. Certainly in the early years
of liberalization and globalization of the Indian economy, there seems to be
scant evidence of a break in patterns of caste inequalities.
In order to work out the exact linkages between processes unleashed by
liberalization and intercaste disparity, data need to be available for a bigger
time span than what the two rounds of NFHS provide. However, privatization
of education and jobs has been one of the prominent features of the 1990s.
Affirmative action directed toward SCs and STs and later OBCs, confined to
government jobs and educational institutions, has correspondingly become
weak. In addition, there is evidence of increasing casualization of jobs, with
women bearing the brunt of this change. Given the established caste hier-
archies, it is not difficult to suppose that low-caste women would be the first

10
Dalit is a term of pride for ex-untouchables.

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Deshpande 749

to bear the adverse impact of these changes. Conversely, it would not be

unreasonable to imagine that beneficiaries of the process of liberalization would
predominantly be upper-class, upper-caste sections. Thus, it is not surprising
to find that, in the early years of liberalization, the pattern of caste inequalities
has not been broken.

VII. Wealth
Of the five variables that make up the GCDI, two are individual specific and
three are household specific. The inclusion of household-specific variables in
an index used for women is based on the assumption that women can be
classified into economic strata (classes) based upon the household ownership
of resources. Indeed, we can examine this component separately in the form
of a “Household Wealth Index” (HWI), which for 1998–99 is reported in
figure 4.11
The HWI indicates a regional pattern that is very similar to that of the
GCDI, and indeed correlation coefficients between the GCDI and the HWI
for each of the caste groups are high. These are higher for SC-ST than for
OBC and Others (close to 0.8 for the SCs and 0.69 for the Others). This could
be due to the fact that, in general, household wealth is much lower in SC-
ST households than in the households of OBCs and Others (landlessness being
far greater in the former groups) and therefore much of the variation in the
GCDI is accounted for by educational and occupational attainment. However,
it needs to be noted that this is household wealth; a focus on household wealth
alone abstracts from issues related to intrahousehold distribution of wealth.
Women’s access to these resources, whether formally in the form of property
rights or more generally in their ability to draw upon these resources in times
of need, remains very weak.

VIII. Education
Given the qualifier about household wealth, it is important to examine sep-
arately the two women-specific components of the index. The literature on
educational disparities is large, and the problems with women’s education are
well known (see the references in n. 1). Tables 1 and 2 show patterns of
educational attainment for women across caste groups for NFHS-1 and NFHS-
2, respectively.
Table 1 reveals very high levels of illiteracy among women in general and
the presence of intercaste disparity within that. For reasons mentioned above,
STs have been omitted from the current discussion as have the northeastern
11
I thank John Strauss for suggesting that I include a separate section on the wealth index.

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Figure 4. Household Wealth Index, 1998–99

750

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 TABLE 1
STATEWISE EDUCATION LEVELS OF WOMEN, BY CASTE, INDIA, 1992–93

% of SCs with % of STs with % of Others with

No Primary Secondary Higher No Primary Secondary Higher No Primary Secondary Higher

All India 77.32 12.26 9.68 .74 69.69 15.4 14.03 .88 51.61 18.46 24.54 5.39
North:
Delhi 62.37 20.1 12.89 4.64 57.58 6.06 27.27 9.09 35.03 12.9 34.04 18.03
Haryana 81.96 11.89 6.01 .14 100 0 0 0 50.9 16.07 27.35 5.69
Himachal Pradesh 59.24 24.17 15.64 .95 79.82 12.84 7.34 0 34.45 24.34 32.55 8.66
Jammu 70.74 11.68 15.52 2.06 82.61 13.04 4.35 0 38.14 14.28 36.25 11.33
Punjab 68.25 15.75 15.11 .9 0 0 0 0 45.12 20.25 29.29 5.33
Rajasthan 91.24 6.04 2.63 .1 95.19 3.06 1.53 .22 75.64 10.57 11.03 2.76
Uttar Pradesh 90.96 5.12 3.4 .52 90.71 5.46 2.19 1.64 71.44 10.36 14.11 4.09
Central:
Madhya Pradesh 78.02 8.64 12.35 .99 92.06 4.65 3.29 0 64.95 16.32 14.72 4.02
East:
Bihar 90.22 2.36 7.08 .34 83.92 2.92 7.89 5.26 74.38 7.24 15.03 3.35
Orissa 76.18 15.71 8.12 0 91.25 4.26 4.49 0 53.45 23.64 19.99 2.92
West Bengal 77.25 15 7.75 0 89.08 8.3 2.62 0 49.25 26.37 20.68 3.7

751
North East:
Arunachal Pradesh 0 0 0 0 71.76 13.1 14.85 .29 58.25 9.79 27.84 4.12
Assam 47.2 28 22.4 2.4 70.38 14.67 14.4 .54 46.45 22.66 26.03 4.87
Manipur 0 0 0 0 54.62 19.68 24.9 .8 46.55 13.36 29.31 10.78
Meghalaya 0 0 50 50 53.91 24.68 19.92 1.49 30.09 28.32 33.63 7.96
Mizoram 0 0 0 0 9.27 50.54 38.34 1.85 20 40 40 0
Nagaland 0 0 0 0 43.22 23.93 31.48 1.36 38.3 12.77 40.43 8.51

This content downloaded from

Tripura 71.43 28.57 0 0 68.1 18.4 13.5 0 34.2 33.66 28.35 3.79

[Link] on Mon, 02 Mar 2026 18:33:31 UTC

West:
Goa 71.05 15.79 13.16 0 75.34 9.59 12.33 2.74 31.72 23.12 37.44 7.73

All use subject to [Link]

Gujarat 54.69 23.44 20.83 1.04 82.28 8.07 8.77 .88 48.69 18.02 28.29 5
Maharashtra 65.4 17.49 16.35 .76 73.71 14.95 10.31 1.03 42.45 22.07 30.31 5.08
South:
Andhra Pradesh 82.45 10.97 5.49 1.1 93.36 3.54 3.1 0 62.9 20.94 12.84 3.31
Karnataka 75.53 9.9 13.4 1.17 78.06 10.55 10.13 1.27 56.54 14.7 25.26 3.5
Kerala 16.79 41.22 41.98 0 41.73 36.69 21.58 0 9.32 33.83 50.63 6.21
Tamil Nadu 66.99 20.99 11.88 .14 81.82 0 18.18 0 37.84 28.81 29.62 3.74
 TABLE 2
STATEWISE EDUCATION LEVELS OF WOMEN, BY CASTE, INDIA, 1998–99

% of SCs with % of STs with % of OBCs with % of Others with

No Primary Secondary Higher No Primary Secondary Higher No Primary Secondary Higher No Primary Secondary Higher

All India 67.91 14.99 14.15 2.95 75.34 11.88 10.81 1.98 55.41 17.32 21.48 5.79 39.57 18.68 28.32 13.44
North:
Delhi 53.03 15.33 25.25 6.39 45.32 13.4 36.52 4.77 47.72 15.62 25.03 11.62 15.74 9.21 33.12 41.93
Haryana 74.76 13.81 10.75 .68 100 0 0 0 64.24 14.48 17.75 3.52 43 12.93 28.2 15.87
Himachal Pradesh 48.71 28.57 21.29 1.44 78.17 9.82 12.01 0 29.7 29.41 36.36 4.53 30.62 21.83 35.3 4.53
Jammu 68.8 12.61 16.22 2.37 67.16 13.48 17.28 2.08 76.19 8.82 12.47 2.52 60.81 12.75 18.79 7.66
Punjab 61.36 16.77 18.44 3.43 100 0 0 0 45.09 21.02 27.93 5.97 24.01 17.79 37.79 20.41
Rajasthan 86.84 6.92 5.04 1.19 92.08 4.13 2.6 1.19 78.71 10.84 8.9 1.55 60.73 14.06 17.48 7.73
Uttar Pradesh 81.62 9.7 6.79 1.89 75.89 8.33 11.98 3.81 75.19 12.11 9.48 3.22 54.32 15.88 17.51 12.28
Central:
Madhya Pradesh 73.75 16.03 7.3 2.91 80.78 13.15 5.11 .96 63.01 20.53 11.92 4.54 34.91 21.11 21.41 22.57
East:
Bihar 87.62 4.01 7.52 .85 88.23 4.34 6.72 .71 78.28 7.53 12.14 2.05 49.86 8.87 31.72 9.55
Orissa 68.99 18.47 12.01 .54 85.36 8.62 5.39 .63 51.11 25.18 21.11 2.6 30.61 26.37 34.19 8.83
West Bengal 61.53 21.52 15.99 .96 81.64 11.77 6.6 0 33.65 22.95 35.1 8.3 34.79 28.12 28.35 8.76

752
North East:
Arunachal Pradesh 50.44 14.12 25.72 9.71 50.76 18.19 28.18 2.87 57.41 9.34 29.48 3.76 36.12 18.2 30.61 15.07
Assam 50.28 21.5 25.99 2.23 60.99 16.85 20.41 1.74 27.01 14.24 44.33 14.42 52.06 18.08 24.14 5.72
Manipur 30.48 24.7 30.31 14.51 33.59 21.48 33 11.94 40.36 18.71 24.73 16.19 29.4 22.99 27.7 19.91
Meghalaya 48.92 27.2 14.39 9.49 34.77 34.64 25.68 4.91 54.41 18.23 27.35 0 60.14 20.34 16.52 3
Mizoram 53.45 22.86 0 23.7 9.52 32.43 51.96 6.36 0 0 100 0 23.15 27.72 49.13 0
Nagaland 69.35 19.43 11.21 0 29.59 23.78 41.44 5.19 56.12 12.33 28.68 2.86 58.55 19.56 12.68 9.2

This content downloaded from

Tripura 30.97 38.99 25.2 4.83 59.88 23.01 15.27 1.84 19.56 35.98 37.32 7.14 25 23.06 38.66 13.29

[Link] on Mon, 02 Mar 2026 18:33:31 UTC

Sikkim 48.1 31.25 20.66 0 45.41 19.7 26.68 8.21 53.18 19.34 24.08 3.41 38.01 23.51 30.51 7.97
West:

All use subject to [Link]

Goa 55.48 19.29 16.45 8.78 66.49 0 33.51 0 46.42 15.96 24.66 12.95 18.59 22.24 38.85 20.31
Gujarat 59 16.9 18.1 6 70.18 13.15 13.26 3.41 47.86 17.31 27.39 7.45 27.26 16.94 32.58 23.22
Maharashtra 41.29 22.76 26.76 9.19 63.8 13.45 18.83 11.97 30.99 22.36 38.01 8.64 34.47 19.38 34.18 11.97
South:
Andhra Pradesh 70.4 16.41 11.87 1.32 85.44 9.44 4.62 .51 65.95 15.7 14.82 3.53 37.42 22.01 31.2 9.37
Karnataka 65.79 8.69 20.19 5.33 78.95 9.95 8.33 2.77 49.34 13.78 28.82 8.06 36.91 16.32 32.17 14.6
Kerala 16.18 22.12 51.94 9.76 27.46 31.61 31.5 9.43 7.32 22.5 54.43 15.76 5.19 19.04 51.81 23.95
Tamil Nadu 51.87 23 20.14 4.99 69.29 13.98 16.73 0 30.29 28.3 31.52 9.89 2.74 7.6 43.03 46.63
 Deshpande 753

states. For India as a whole, in 1992–93, 77.3% of SC women and 69.7% of

ST women reported no education, as compared to 51.6% of Other women.
At the other end of the spectrum, 0.74% and 0.88% of SC and ST women,
respectively, had higher education, as compared to 5.39% of Other women.
The regional pattern indicates that the western states of Goa, Maharashtra,
and Gujarat reported outcomes better than the national average, as did the
South Indian states, with the exception of Andhra Pradesh. The North and
the East were much below the national average, with the exception of Delhi
(the capital), Himachal Pradesh, Jammu, Punjab, and West Bengal.
That the regional effect is important in explaining patterns of education
in general is suggested by the fact that, for both rounds of NFHS, the states
that have high male literacy rates are also those that have a high female literacy
rate. Median years of schooling, literacy rates, and school attendance vary
substantially by state. These vary significantly by age cohorts as well, indicating
the well-known problem of dropouts in primary school and middle school.
Table 2, presenting data for 1998–99, suggests that there has been an
improvement in women’s education level in general, which is the minimum
outcome expected from an economy in the twenty-first century. At the all-
India level, as many as 68% of SC women, 55% of OBC women, and 40%
of upper-caste women report “no education” (49% of all women and 26% of
all men above age 6 are illiterate). Male-female differences in education, which
are significant for both rural and urban areas, are reported in Deshpande (2002).
For NFHS-2, a higher percentage of men than women have completed each
stage of schooling. The median number of years of schooling is 5.5 for males
and 1.6 for women (NFHS-2, 25). Thus, while there is an improvement in
educational outcomes between the two rounds, the well-known problems with
schooling and education persist strongly, reinforcing the point that we cannot
wait for the problem to disappear on its own, that merely the growth of the
economy is not going to take care of the problem with the urgency that it
warrants.

IX. Women’s Work and Occupation

Tables 3 and 4 document the castewise distribution of women’s occupation
(in terms of broad occupational groups) for NFHS-1 and NFHS-2, respectively.
Before we examine these tables, problems with the interpretation of the num-
bers related to the first category, “not working,” need to be reviewed. In
1992–93, at the all-India level, 60.1% of SC women, 48.31% of ST women,
and 70.26% of Other women self-reported themselves to be “not working.”
These numbers do not square with the extent of poverty in the country; neither
do they square with the time-use surveys or with the numerous case studies

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 TABLE 3
STATEWISE OCCUPATION LEVELS OF WOMEN, BY CASTE, INDIA, 1992–93

% of SCs with % of STs with % of Others with

1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6

All India 60.09 .171 35.57 3.37 .77 .03 48.31 .22 45.29 4.37 1.58 .23 70.26 .23 23.72 3.62 1.91 .26
North:
Delhi 74.87 0 12.31 12.31 .51 0 78.79 0 9.09 12.12 0 0 80.66 .19 6.8 8.44 3.17 .74
Haryana 70.12 0 24.2 5.12 .55 0 0 0 0 100 0 0 61.5 .05 34.12 1.84 2.12 .38
Himachal Pradesh 33.12 .16 60.13 4.87 1.73 0 30.28 0 69.72 0 0 0 39.76 .32 50.68 4.2 4.65 .41
Jammu 59.89 0 35.74 2.59 1.77 0 56.52 0 39.13 4.35 0 0 64.98 .15 28.36 2.34 3.78 .4
Punjab 86.68 .13 10.63 2.05 .51 0 0 0 0 0 0 0 88.98 0 8.04 1.36 1.54 .09
Rajasthan 66.21 .29 32.23 1.17 .1 0 55.64 .11 43.59 .44 .11 .11 70.99 .49 26.03 .92 1.14 .43
Uttar Pradesh 74.28 .1 23.42 1.93 .26 0 42.62 0 55.74 1.09 0 .55 82.67 .06 15.54 .86 .75 .12
Central:
Madhya Pradesh 65.76 .25 31.03 2.71 .25 0 43.12 .05 54.69 1.97 .11 .05 73.3 .12 23.2 2.37 .92 .07
East:
Bihar 60.07 .51 36.68 2.37 .51 .17 57.18 .88 36.66 1.47 3.23 .59 76.07 .38 21.32 1.52 .5 .2
Orissa 69.95 0 27.46 2.59 0 0 45.52 .12 52.71 1.53 .12 0 83.28 .07 12.48 2.52 1.59 .07
West Bengal 65.67 0 24.83 9.2 .75 0 36.09 .43 60.87 2.61 0 0 79.72 .49 13.93 4.47 1.38 0

754
North East:
Arunachal Pradesh 0 0 0 0 0 0 45.93 .15 49.13 2.91 1.6 .29 77.84 0 16.49 3.61 1.55 .52
Assam 73.81 0 21.43 3.97 .79 0 76.94 0 19.3 1.34 2.14 .27 81.37 .32 10.81 4.43 2.63 .44
Manipur 0 0 0 0 0 0 31.33 0 56.63 7.63 4.42 0 48.44 .71 33.81 12.93 3.41 .71
Meghalaya 50 0 0 0 50 0 53.58 0 33.76 9.13 3.14 .39 85.34 .86 1.72 9.48 2.59 0
Mizoram 0 0 0 0 0 0 64.1 .1 19.12 13.07 3.41 .2 100 0 0 0 0 0
Nagaland 0 0 0 0 0 0 54.36 1.18 33.3 6.72 3.54 1 74.47 2.13 6.38 10.64 6.38 0

This content downloaded from

Tripura 85.71 0 0 14.29 0 0 45.73 0 46.34 6.71 1.22 0 78.26 .86 11.84 5.6 3.44 0

[Link] on Mon, 02 Mar 2026 18:33:31 UTC

West:
Goa 57.89 0 34.21 7.89 0 0 52.05 0 36.99 8.22 2.74 0 70.03 .27 17.05 7.97 4.12 .57

All use subject to [Link]

Gujarat 54.17 0 36.98 6.77 2.08 0 33.98 0 63.05 1.93 1.05 0 60.37 .36 34.61 3.17 1.4 .1
Maharashtra 42.37 0 52.29 4.2 1.15 0 25.38 .51 69.49 3.33 1.28 0 52.87 .17 40.28 4.9 1.6 .17
South:
Andhra Pradesh 25.51 .16 70.08 2.68 1.42 .16 23.89 0 73.89 1.77 .44 0 50.78 .21 41.67 5.35 1.7 .29
Karnataka 38.01 .58 57.31 2.92 1.17 0 37.39 0 61.34 .84 .42 0 54.92 .22 40.18 3.04 1.45 .19
Kerala 45.8 0 50.38 2.29 1.53 0 42.11 0 51.88 3.76 2.26 0 76.14 .05 15.7 4.3 3.26 .54
Tamil Nadu 37.73 .42 55.89 4.58 1.25 .14 45.45 0 45.45 9.09 0 0 55.78 .19 34.19 7.04 2.55 .25

Note. Occupational categories: (1) not working; (2) all others not working, including unemployed, household duties, students, retired persons, beggars, inmates of institutes, and disabled
persons; (3) farm, fish, hunt, log or production, or transport; (4) clerical plus sales workers plus service workers; (5) professional/technical low level plus administration/executive/managers;
(6) professional/technical high level.
 Deshpande 755

that document the multifarious nature of women’s work. Deshpande (2001b,

2003) discusses this issue in detail. To sum up, women’s “not working”
response reflects an undervaluation of their own work that, in turn, reflects
the low worth that society attaches to their work. This is not a problem
confined to India alone but one that applies to a large part of the developing
world, where particularly agricultural tasks are seen as a part of normal house-
hold duties and thus are not viewed as “work.” Additionally, in India, since
constraints on women’s public visibility rise with a rise in the caste status,
working for wages is seen as a mark of low status and thus that one is doing
so is often denied.
Table 4, providing data for 1998–99, indicates that there has been a decline
in the proportions reporting “not working” for SC and ST groups, such that,
at the all-India level, 53.65% of SC and 36.8% of ST women self-report
themselves to be not working. It is not clear that the same trend can be seen
in the Others category. The erstwhile “Others” are now divided into OBC
and upper castes, and the latter actually show a greater percentage of “not
working” (72.38) than the combined Others category of the previous round.
A decline in “not working” proportions for SCs is registered despite the trend
in North India, which continues to be a problem zone for women’s work.
Most of the states from this region report an increase in “not working” pro-
portions, and it is the averages in the rest of the country that compose the
national trend.
The positive trend in the distribution for SC women is that, for the first
round, proportions in the top category of occupation were practically zero,
whereas now they are minuscule but positive in most states. The highest
proportion is still under 2% (1.89 in Karnataka). The upper-caste women
show a paradox: the majority report themselves to be not working, but a
significant proportion can be found in the upper end, far greater than for the
other caste groups, reinforcing the pattern of intercaste disparity. A comparison
of the two rounds reveals that the most significant gains can be seen for the
top category of occupation for the upper-caste women (with proportions as
high as 13.63% for Tamil Nadu), suggesting that the real beneficiaries of
liberalization are the elite within the upper castes. As for the rest of “working”
women, it is category 3 (agriculture and other unskilled labor) that forms the
single largest category across all caste groups.
A much discussed phenomenon associated with globalization is the femi-
nization of the labor force. It has been argued, that, starting in the 1980s, as
labor markets were deregulated, there was a rise in female labor force partic-
ipation rates (LFPRs) and a fall in men’s employment, as well as a “transfor-
mation—or feminization—of many jobs traditionally held by men” (Standing

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 TABLE 4
STATEWISE OCCUPATION LEVELS OF WOMEN, BY CASTE, INDIA, 1998-99

% of SCs with % of STs with % of OBCs with % of Others with

1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6

All India 53.65 1.42 37.05 2.32 4.4 1.15 36.8 1.02 54.82 2.53 3.6 1.23 58.61 .09 31.36 2.17 5.48 1.47 72.38 .073 17.41 2.28 4.24 2.96
North:
Delhi 75.15 6.43 4.76 7.45 5.13 1 77.29 0 4.57 9.06 9.08 0 78.24 2.87 8.97 1.52 3.07 2.33 81.61 1.69 1.12 4.15 4.09 7.33
Haryana 85.93 2.36 8.16 2.03 1.01 .5 100 0 0 0 0 0 88.63 0 7.42 .79 2.36 .8 88 .12 4.6 .8 2.27 4.22
Himachal

756
Pradesh 90.69 1.06 12.45 1.84 3.25 .71 80.51 0 9.67 0 9.82 0 80.74 1.66 13.72 1.4 .79 1.68 78.42 .71 12.64 1.97 1.64 4.63
Jammu 71.22 1.58 20.01 2.21 4.45 .52 52.93 0 35.51 0 8.9 2.66 55.85 .72 33.1 .64 9.19 .5 66.03 .16 31.1 1.16 9.08 2.69
Punjab 90.04 2.62 3.9 1.08 1.65 .7 100 0 0 0 0 0 89.02 .7 2.52 .86 5.3 1.6 91.37 .3 .58 1.79 2.23 3.73
Rajasthan 52.89 .2 40.46 1.47 3.98 1.01 35.92 0 61.24 .24 1.82 .78 55.14 .35 37.91 .99 4.33 1.3 64.85 .16 27.46 .67 4.94 1.94
Uttar
Pradesh 70.54 .42 25.87 .91 1.53 .73 72.68 .81 21.26 1.38 2.21 1.66 72.17 .06 24.52 .89 1.72 .64 83.33 .12 10.76 1.19 2.76 1.83

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
Central:
Madhya
Pradesh 33.62 .97 52.16 1.7 10.07 1.47 23.12 .36 71.32 .78 3.59 .82 42.84 .92 47.14 1.68 6.19 1.23 71.17 .09 16.9 2.54 4.19 5.12

All use subject to [Link]

East:
Bihar 57.78 1.26 34.12 1.62 4.39 .84 55.11 0 33.39 5.88 4.7 .91 76.89 .34 18.75 1.02 2.56 .44 90.59 0 5.46 .89 1.51 1.55
Orissa 59.62 1.25 28.97 3.35 5.7 1.11 45.07 2.07 42.22 3.02 6.68 .95 76.92 .29 16.74 2.18 2.99 .88 85.39 .41 6.43 2.29 2.5 2.99
West Bengal 63.95 2.55 23.85 1.87 6.5 1.27 27.27 .43 68.03 .43 2.98 .85 74.84 1.5 16.63 1.12 5.09 .83 81.59 1.46 5.44 2.05 7.83 1.62
North East:
Arunachal
Pradesh 79.36 .64 11.83 5.5 0 2.67 24.48 0 69.28 3.93 0 2.31 75.53 0 20.25 1 .73 2.49 78.97 0 14.07 2.7 1.35 2.91
Assam 79.34 1.39 13.11 .46 4.56 1.14 70.28 .37 22.83 2.26 2.82 1.45 89.56 .37 3.86 .81 2.21 3.19 83.05 .89 10.35 1.49 2.35 1.87
Manipur 28.22 1.46 30.65 16.27 16.1 7.29 22.08 .58 47.25 13.1 8.77 8.23 43.43 1.43 14.52 20.45 18.65 1.52 34.65 .13 22.28 15.79 21.15 6
 Meghalaya 81.61 1.11 6.38 0 0 1.6 50.86 2.03 34.23 7.15 1.86 3.87 63.82 18.23 8.83 9.12 0 0 68.14 1.55 25.87 1.45 0 3
Nagaland 50.43 0 47.81 0 1.76 0 32.95 0 49.02 8.19 4.73 5.11 23.09 0 62.25 8.94 2.86 2.86 64.57 1.54 25.57 0 3.07 5.26
Sikkim 81.34 0 4.31 3.46 8.79 2.11 73.96 1.06 5.67 11.51 .98 6.62 .05 .2 5.67 7.57 1.96 4.47 78.36 .34 3.66 8.88 1.4 7.36
Tripura 80.01 1.94 7.56 2.93 3.24 4.41 57.96 0 36.63 2.36 1.18 1.87 81.4 1.47 9.38 1.84 1.78 4.14 82.02 1.07 7.75 3.36 1.02 4.78
West:
Goa 48.22 15.61 20.19 13.84 2.13 0 0 32.26 34.23 0 33.51 0 37.4 2.5 35.49 7.23 7.18 10.2 54.87 3.31 20.51 9.83 6.12 5.37
Gujarat 41.98 1.39 47.56 4.18 3.32 1.58 34.47 3.13 56.75 2.63 2.36 .66 48.11 2.39 38.65 3.04 5.95 1.85 59.85 1.43 27.15 3.65 4.29 3.64
Maharashtra 48.18 3.99 34.63 5.31 6.07 1.82 27.62 1.9 64.34 1.66 3.04 1.43 43.94 2.38 41.72 3.61 4.91 3.44 47.57 1.5 39.63 3.57 4.25 3.47
South:
Andhra
Pradesh 25.76 .63 65.4 2.8 4.79 .62 19.36 1 69.22 3.11 6.29 1.02 37.37 .75 51.3 2.03 7.53 1.01 60.8 .38 29.56 2.9 3.34 3.02
Karnataka 37.72 .98 55.43 1.55 2.43 1.89 27.77 .79 61.86 4.04 3.99 1.56 45.52 1.35 42.34 2.26 6.81 1.72 59.52 .9 27.01 3.44 5.38 3.76
Kerala 51.29 1.3 35.19 5.15 5.76 1.31 52.6 2.55 34.13 8.26 2.45 0 78.88 1.41 9.76 2.28 5.37 2.31 76.52 .68 9.99 2.53 4.56 5.72
Tamil Nadu 31.85 .85 58.48 2.98 4.12 1.7 18.38 .8 73.88 3.88 3.06 0 50.73 1.18 30.15 4.56 10.78 2.59 63.81 .4 1.92 15.41 4.83 13.63

Note. Occupational categories: (1) not working; (2) all others not working, including unemployed, household duties, students, retired persons, beggars, inmates of institutes, and disabled persons; (3) farm, fish, hunt, log
or production, or transport; (4) clerical plus sales workers plus service workers; (5) professional/technical low level plus administration/executive/managers; (6) professional/technical high level.

757

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
758 economic development and cultur al change

1989, 1077). Deshpande (2003) discusses this issue in detail. Evidence from
India does not support the feminization thesis in the aggregate: male LFPRs
have marginally risen in the last 20 years, while female LFPRs have marginally
fallen—which is exactly the opposite of what is predicted. Other evidence on
women’s employment indicates a decline in self-employment and a clear in-
crease in the casual labor category. Agriculture continues to be the main
provider; however, the share of agriculture in employment has declined much
more for men than for women. According to National Sample Survey data,
85% of rural women were engaged in agriculture in 1999–2000.

X. Concluding Comments
The GCDI for 1998–99 reveals that the material standard of living for women
continues to be low despite an improvement over the early 1990s. Within
this overall pattern, there is significant intercaste disparity as well as regional
variation. Indicators on health and nutritional status reinforce this picture. A
rough comparison of 1998–99 data with 1992–93 data reveals that the early
years of liberalization have not been able to close intercaste gaps in that these
gaps remain substantial for most states. By and large, the richer states of India
exhibit high intercaste disparity. Education outcomes for the majority of
women continue to be low, with substantial intercaste disparity, although
there is a definite reduction in illiteracy for all caste groups between the two
rounds of NFHS. Data on occupation are fraught with all kinds of uncertainties,
the most significant of which is the proportion of women that self-report
themselves to be not working. In the context of gross underreporting of
women’s work, there is a substantial increase in the proportions of upper-caste
women at the higher end of the occupational spectrum, suggesting that the
early years of liberalization have benefited sections of the upper castes.

References
Akerlof, George. 1984. “The Economics of Caste, the Rat Race, and Other Woeful
Tales.” In his An Economic Theorist’s Book of Tales, 23–44. Cambridge: Cambridge
University Press.
Ayers, Ian. 2001. Pervasive Prejudice: Unconventional Evidence of Race and Gender Dis-
crimination. Chicago: University of Chicago Press.
Bajpai, Nirupam, and Sangeeta Goyal. 2004: “India: Towards the Millennium De-
velopment Goals.” Working Paper no. 3, Center on Globalization and Sustainable
Development, Earth Institute, Columbia University.
Becker, Gary. 1967. The Economics of Discrimination. Chicago: University of Chicago
Press.
Deshpande, Ashwini. 2000. “Recasting Economic Inequality.” Review of Social Economy
58, no. 3:381–99.

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
 Deshpande 759

———. 2001a. “Caste at Birth? Redefining Disparity in India.” Review of Development

Economics 5, no. 1:130–44.
———. 2001b. “Casting Off Servitude: Assessing Caste and Gender Inequality in
India.” In Feminism and Anti-racism: International Struggles for Justice, ed. France
Winddance Twine and Kathleen Blee, 328–48. New York: New York University
Press.
———. 2002. “Assets versus Autonomy: The Changing Face of the Gender-Caste
Overlap in India.” Feminist Economics 8, no. 2:19–35.
———. 2003. “Gender Discrimination at the Workplace: India, Bangladesh and
China.” In Improving Labour Market Opportunities and Security for Workers in Developing
Countries.” U.S. Bureau of International Labour Affairs Research Symposium Pa-
pers, vol. 1, ed. William T. Kosanovich, 223–66. Washington, DC.
———. 2004. “Decomposing Inequality: Significance of Caste.” In The Dalit Ques-
tion: Reforms and Social Justice, ed. Bibek Debroy and D. Shyam Babu, 33–52.
New Delhi: Globus Books and Rajiv Gandhi Institute for Contemporary Studies.
Deshpande, Sudha, and Lalit Deshpande. 1999. “Gender-Based Discrimination in
the Urban Labour Market.” In Gender and Employment in India, ed. T. S. Papola
and Alakh N. Sharma, 223–48. New Delhi: Vikas.
Dreze, Jean, and Mamta Murthy. 2001. “Fertility, Education, and Development:
Evidence from India.” Population and Development Review 27, no. 1:33–63.
Dreze, Jean, and Amartya Sen. 1995. Economic Development and Social Opportunity. New
Delhi: Oxford University Press.
Duraisamy, Malathy, and P. Duraisamy. 1999. “Women in the Professional and
Technical Labour Market in India: Gender Discrimination in Education, Em-
ployment and Earnings.” Indian Journal of Labour Economics 42, no. 4:599–612.
Elson, Diane. 1999. “Labour Markets as Gendered Institutions: Equality, Efficiency
and Empowerment Issues.” World Development 27, no. 3:611–26.
Jha, Raghbendra. 2002. “Reducing Poverty and Inequality in India: Has Liberali-
zation Helped?” Working Paper no. 2002–04, Department of Economics, Research
School of Pacific and Asian Studies, Australian National University. http://
[Link]/archive/00001812/01/[Link].
Kingdon, Geeta Gandhi. 1997. “Labour Force Participation, Returns to Education
and Sex-Discrimination in India.” Indian Journal of Labour Economics 40, no. 3:
507–23.
———. 2002. “The Gender Gap in Educational Attainment in India: How Much
Can Be Explained?” Journal of Development Studies 39, no. 2:25–53.
Lovell, Peggy. 2003. “Race, Gender and Regional Labor Market Inequalities in
Brazil.” In Boundaries of Clan and Color: Transnational Comparisons of Inter-group
Disparity, ed. William Darity Jr. and Ashwini Deshpande, 14–26. London:
Routledge.
Milanovich, Branko. 2004. “Half the World: Regional Inequality in Five Federa-
tions.” Policy Research Working Paper no. 3699, World Bank, Washington, DC.
PROBE Team. 1999. Public Report on Basic Education in India. New Delhi: Oxford
University Press.
Singh, Manjari, and Debashish Bhattacharjee. 1998. “Pay Discrimination by Gender

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]
760 economic development and cultur al change

in the Corporate Sector: A Case Study.” Indian Journal of Labour Economics 41, no.
1:97–103.
Singh, Nirvikar, Laveesh Bhandari, Aayou Chen, and Aarti Khare. 2003. “Regional
Inequality in India: A Fresh Look.” Econ WPA Development and Comp Systems
Paper no. 0412006. [Link]
Sokoloff, Natalie J. 1992. Black Women and White Women in the Professions: Occupational
Segregation by Race and Gender, 1960–1980. London: Routledge.
Standing, Guy. 1989. “Global Feminization through Flexible Labour.” World De-
velopment 17, no. 7:1077–95.
Thorat, Sukhadeo. 1979. “Passage to Adulthood: Perception from Below.” In Identity
and Adulthood, ed. Sudhir Kakkar, 65–81. New Delhi: Oxford University Press.
Tilak, J. B. G. 1980. “Education and Labour Market Discrimination.” Indian Journal
of Industrial Relations 16, no. 2:95–114.

This content downloaded from

[Link] on Mon, 02 Mar 2026 18:33:31 UTC
All use subject to [Link]