Microcredit and Women’s Empowerment: Through the

Lens of Time-Use Data from Rural India

Supriya Garikipati

ABSTRACT

This study examines the impact of microcredit on male and female time
use, and draws on this analysis to explore the linkages between credit and
women’s empowerment. A study of time use can help understand these
linkages, because if credit is intended to improve women’s livelihoods, it
can also be expected to influence the way women allocate their time. Its
other advantages are that it does not suffer from much time lag and can
be objectively measured. Using household survey data from rural India, the
findings show that while microcredit has little impact on women’s time use,
it helps their husbands move away from wage work (associated with bad
pay and low status) to self-employment. This is because women’s loans are
typically used to enhance male ownership of the household’s productive
assets. Further, it is found that it is only women who use loans in self-
managed enterprises who are able to allocate more time to self-employment.
If credit is intended to increase the value of women’s work time, it follows
that it is not access to loans but use of loans that matters. Ensuring women’s
control over loan-created assets must therefore be a critical policy objective.

INTRODUCTION

Supporting poor women via microcredit has become central to poverty re-
duction strategies across the developing world. Underpinning this trend is an
implicit model of the empowered woman who invests money in a successful
enterprise, uses the income to enhance the nutritional status of her family,
educates her children and begins to participate in major family decisions
(Ackerly, 1995). This ideological construct has been so powerful that many
microcredit institutions lend to female clients only and nearly 85 per cent
of all clients worldwide are women (Daley-Harris, 2007). The relationship

The author gratefully acknowledges financial support received from the Department for Interna-
tional Development (award number R7617) and the Newton Trust (award number INT 2.05[d]).
The author also acknowledges the contribution of the field research team: Achari, Chandrasekhar,
Lakshmamma, Narsimhulu, Lakshmi, Padma, Ravi, Rathish and Sridevi, and is grateful to the
two anonymous referees for their insightful and constructive comments on the manuscript. She
also thanks Marek Hudon, David Hulme, Uma Kembampathi, Mark Pitt, Gita Sen, Kunal Sen,
Ariane Szafarz, Chris Udry and Jana Vyrestakova for comments on earlier versions of the paper.

Development and Change 43(3): 719–750. DOI: 10.1111/j.1467-7660.2012.01780.x


C 2012 International Institute of Social Studies.
Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and
350 Main St., Malden, MA 02148, USA
720 Supriya Garikipati

between microcredit and empowerment, however, is intensely debated in

the literature (Garikipati, 2008a; Kabeer, 2001). This study contributes to
the debate by examining one specific outcome of credit intervention, namely
changes in the allocation of households’ work time, through the use of survey
data from rural India.
India’s microcredit programme was launched in 1992 and, like several
other programmes around the world, mainly targets poor rural women. Credit
groups, referred to as Self-Help Groups (SHGs), consist of ten to fifteen
women who come together from a similar socio-economic background.
Group formation is facilitated by NGOs but the primary focus of the scheme
is credit with little attempt at capacity building. After six months of regular
savings, groups become eligible for bank credit. The National Bank for
Rural and Agricultural Development (NABARD), India’s apex rural bank,
provides 100 per cent re-finance to the lending institutions. Repayment rates
are consistently over 95 per cent compared to other rural modalities which
are around 40 to 65 per cent (World Bank, 2003, 2005, 2006). With around 40
million clients and an average annual growth rate of 112.2 per cent between
1999 and 2007, it is the biggest and the fastest growing microcredit scheme
in the world (Garikipati, 2008a).1
With respect to the relationship between microcredit and women’s em-
powerment, the evidence emerging from India is extremely conflicting. In
a series of studies that use data from SHG clients only and cover eleven
states, Puhazhendi (2000), Puhazhendi and Badatya (2002) and Puhazhendi
and Satyassi (2000) conclude that women have experienced significant ex-
ternalities in terms of better personal relations and enhanced social status.
Swain and Wallentin (2009) compare SHG clients with non-clients from
five states and find that while both groups became more empowered over
time, the change for SHG women was significant. Reporting evidence from
a randomized evaluation, Banerjee et al. (2010) find that women’s access
to microcredit has no effect on their decision-making power. Examining the
impact of NGO-led institutions, Hunt and Kasynathan (2001) find that if
credit programmes are to support women, strategies that transform gender
relations must receive more attention. Similar results are reported by EDA
(2005), who find that patriarchal cultural norms could restrict the poten-
tial for women’s empowerment. Leach and Sitaram (2002) find that, due
to the resentment it can create, the exclusion of male relatives from credit
programmes could have a detrimental effect on the women concerned. In
her investigation of the impact of different lending technologies, Holvoet
(2005) finds that credit alone is insufficient to have a positive impact on
women’s decision-making patterns; to be beneficial it needs to be channelled
through groups and combined with training. Garikipati (2008a) compares
SHG clients with non-clients from Andhra Pradesh (AP), India, and finds

1. For a brief history of India’s rural financial sector see Garikipati (2008a).
Microcredit and Women’s Empowerment in Rural India 721

that access to credit alone does not have an effect on empowerment; it is the
way in which credit is used that counts.
This study contributes to the existing literature by focusing on one specific
outcome of credit intervention. It examines the impact of microcredit on male
and female time allocation decisions. The focus is on how both men and
women respond to credit intervention, hence how they are impacted upon.
Detailed time-use data were collected from men and women belonging
to poor households in rural AP. In around 40 per cent of the households
surveyed, the female spouse was participating in the microcredit programme.
Since the economic theory of the household was pioneered by Becker
(1965) and Gronau (1973), a good deal of research has been conducted into
the study of time allocation behaviour in developing country households.
The determinants of women’s market participation, in particular, have re-
ceived much attention (Fafchamps and Quisumbing, 1998; Khandker, 1988;
Mueller, 1984; Rose, 2000; Skoufias, 1993). Studies have also specifically
examined how male and female time allocations respond to new economic
opportunities in rural areas (von Braun and Webb, 1989; Jacoby, 1993;
Newman, 2002; Paolisso et al., 2002). Given the level of interest, it is sur-
prising that hardly any research has been conducted to understand the impact
of microcredit on households’ time allocation. One possible reason for this
is the difficulty in obtaining rigorous time-use data from clients and compa-
rable non-clients. The only exception to this is Pitt (2000), who examines the
effect of microcredit on the household’s mix of agricultural contracts and
the supply of male agricultural labour, using survey data from the landless
poor in rural Bangladesh. Pitt, however, focuses on male time use only.
Cultural restrictions imposed on women in rural Bangladesh imply that their
involvement in income-generating activities is rather negligible (ibid.).
If rigorously collected, time-use data can be valuable in understanding
the impact of microcredit. Of the various indicators that can be affected by
an inflow of credit, time-use decisions are likely to change almost immedi-
ately. The changes are also likely to become more noticeable over time, as
the duration of participation in the credit programme increases. Moreover,
where credit affects long-term livelihood decisions, the changes in time use
are likely to be permanent. The direction and magnitude of these changes
can shed light on whether credit is likely to have the desired impact. The
other advantage of time use is that it is an objectively measurable indica-
tor — as opposed to several other impact indicators that require subjective
judgement — either on the part of the respondent or the researcher(s). Fur-
thermore, time use can be viewed both as a process that facilitates empow-
erment as well as an outcome that indicates empowerment. It is a process
in that it traces women’s journey from credit to better livelihoods, but also
an outcome as it indicates the value of their time after they have accessed
credit. This dual role of time use goes right to the heart of the debate on
whether processes or outcomes are better indicators of women’s empower-
ment (Garikipati, forthcoming; Kabeer, 2001; Malhotra and Schuler, 2005).
722 Supriya Garikipati

In this study, the idea that credit may influence the value of women’s work
time is captured by dividing the types of market work done by individuals
into two broad categories: self-employment and wage employment (or wage
work).2 Market work is defined as any work that brings income (in the
form of cash or kind) into the household; self-employment is defined as
work on productive assets owned by the household like land and cattle or
family business; and wage employment or wage work is defined as work
done for daily wages. Classifying market work in this way for poor rural
communities is important because wage work is associated with bad pay and
arduous working conditions and is also considered socially demeaning.3 It is
done mainly by people from lower economic classes who don’t have access
to sufficient productive assets that can help them earn a living.
Historically, female participation rates in rural India have been high (Cen-
tral Statistical Organization, 1995). Most of the rural women workers, how-
ever, are compelled to take up badly paid and socially demeaning agricultural
wage work. According to the India Census, 43.4 per cent of rural women
workers are agricultural labourers compared to 27.4 per cent of men (Gov-
ernment of India, 2001). Furthermore, in several states of India, like Andhra
Pradesh, Karnataka, Maharashtra and Tamil Nadu, women have overtaken
men in worker numbers. Census data show that women are taking over wage
work as men move into self-employment.
This trend is mainly because, despite favourable inheritance laws that give
women a right to family land, the patriarchal norms that exist across much
of the Indian sub-continent prevent women from exercising this right. With
some slight regional variations, this is more or less true across much of rural
South Asia (Agarwal, 1994). This is a source of concern especially since
it is well established that land as a productive asset is a critical factor in
determining women’s economic well-being, social status and empowerment
(Agarwal, 1994; Allendorf, 2007; Rao, 2007). Meagre land holdings mean
that it is usually men who, as owners, retain working rights in family land

2. Studies examining women’s time-use patterns tend to focus on the determinants of their
market participation only. Little attempt is made to differentiate the types of market work
done by women. This approach is appropriate for situations where women’s market par-
ticipation is rare because of exogenous constraints, and studying factors that may help
overcome these obstacles is critical for policy. For instance, Khandker (1988) rejects dif-
ferentiating between the types of market work done by women in rural Bangladesh. He
argues that once women start work, it is fairly easy to switch between different types of
work. This is possible, given that in Bangladesh women face substantial cultural barriers to
working outside the domestic sphere. The idea is that once women transcend these barriers,
the type of work they do is not difficult to choose, provided they can access such work.
In the case of rural India, as in several other developing countries, where women’s market
work is not stigmatized and significant numbers work outside the household, it is important
to go beyond participation and examine the factors that help improve the value of their
work time.
3. Pitt (2000) classifies market work similarly and for similar reasons but, as mentioned
earlier, his analysis does not include women’s time use.
Microcredit and Women’s Empowerment in Rural India 723

while women are left to take up what work they can find. This is usually
agricultural wage work. The conditions surrounding this type of work can be
detrimental to their welfare (da Corta and Venkateshwarlu, 1999; Garikipati,
2008b).
In this context, lending to women is expected to help them invest in non-
farm enterprises, which in turn allows them to shift their work time from
wage work to self-employment which is considerably better remunerated
and enjoys higher social status. This is likely to enhance the value of their
incomes, improve their self-esteem and empower them. The linkages that
take women from accessing credit to spending more time in self-employment
are especially important for poor rural women who have limited livelihood
options. As their time becomes more valuable, women may spend less time
in wage work and housework. The impact that lending to women is likely to
have on male time use is somewhat more ambiguous. Straightforward income
and substitution effects suggest that men should reduce time allocated to
market work and spend more time in housework. However, cultural norms
may dictate otherwise. Male time in self-employment may also increase as
a result of their involvement in the investments made using women’s loans.
This study examines some of these linkages.
For the purpose of this enquiry, all members of the household are assumed
to make time-use decisions simultaneously. Estimating such a system of de-
cisions is complicated by the large differences found across households, both
in the number of household members and the types of relations. The effect of
lending to women on household time allocation is captured indirectly, in two
ways. First, time allocation outcomes for men and women from SHG house-
holds are compared with outcomes for men and women from households
in the control group. Second, to focus on the issue of whether microcredit
increases women’s time in self-employment, reduced-form determinants of
the type of market work done by women are estimated for those receiving
SHG loans.
The rest of this article is structured as follows. The next section describes
the questionnaires used in the fieldwork and the resulting datasets. This is
followed by an overview of the time-use outcomes by gender and participa-
tion in the credit programme. Next, the empirical models used to investigate
the impact of microcredit on male and female time use are presented. The
descriptive statistics of the variables are also presented and results are dis-
cussed. This is followed by an exploration of the reasons for the observed
results and concluding remarks.

FIELDWORK AND DATA

Fieldwork was carried out in villages of the Mahabubnagar district, a

drought-prone district in the southern state of AP. Over 45 per cent of
its rural households live below the poverty line (Government of Andhra
724 Supriya Garikipati

Pradesh, 1996). The state government has resolutely pursued the SHG pro-
gramme as part of its poverty-alleviation strategy with the twin objectives
of promoting livelihood diversification and women’s empowerment. As a
result, Mahabubnagar has one of the oldest, biggest and fastest growing SHG
programmes in India (NABARD, 2003).
The data were collected between 2001 and 2003 in three separate rounds
as part of a larger study that investigated female labour market participation
and issues surrounding pro-poor growth. All the survey villages have an
active SHG programme. Interviews were carried out by six enumerators,
three men and three women, all of whom were graduates and trained in basic
survey techniques. Interviews were always carried out by two enumerators:
one male and one female. The author participated in over one-third of all
interviews and also carried out all the focus group interviews.4
During the Kharif season in 2001 and again during the Rabi season in
2002,5 data were collected from 302 households that were randomly selected
from a population of economically active, married-couple households. Ten
de facto male or female-headed households and one income outlier were
dropped from the analysis. Of the remaining 291 households, the female
spouses from 117 households participated in the SHG programme (com-
pleted at least one loan cycle) and the remaining 174 were not in the
programme. From each household, either the male or female spouse was
randomly selected for interview such that an approximately equal number
of men and women were consulted. In the remainder of this article, these
surveys are referred to as ‘the household surveys’.
Data from the household surveys include detailed modules on demo-
graphic characteristics of household members, their economic activities,
asset holdings, credit programme participation, household decision making
and respondent’s time use. Time-use data were collected using the 24-hour
recall method. This method is considered more accurate compared to others
because it is more detailed and it is easier for respondents to recall what
they did the day before.6 Around 4 per cent of the respondents reported the
previous day to have been ‘unusual’ (for instance, they visited a doctor or
relatives living outside the village). In these cases, respondents were asked
to recall details from the last ‘routine’ day. In all cases, this was no more
than four days prior to the interview. The aim was to capture the activities
that were routine to the particular agricultural season.
Detailed time-use data were obtained for 145 women and 146 men across
two typical days and seasons. The average household consists of 6.2 mem-
bers with a landholding of 2.5 acres. Of the survey households, 60.3 per cent

4. For more details on survey protocol, see Horrell et al. (2008).

5. Kharif and Rabi are the two main agricultural seasons in South Asia. The Kharif crop is
the monsoon harvest and is usually sown with the beginning of the first rains in July. The
Rabi crop is the spring harvest and is usually completed by mid-June.
6. For a review of time-use measurement methods and issues surrounding these, see Juster
and Stafford (1991).
Microcredit and Women’s Empowerment in Rural India 725

earn an average monthly net per capita income below the poverty threshold
of Rs 262.9 for rural AP (Planning Commission, 2001).7 While there is
some evidence of livelihood diversification, households mainly rely on agri-
cultural incomes. A total of fifty-eight women (40 per cent) and 104 men
(71.2 per cent) spend most of their work time in self-employment, while
seventy-seven women (53.1 per cent) and thirty-eight men (26 per cent)
spend most of it in wage labour. While the majority of men and women were
either self-employed or worked for wages, a small proportion of men (18.69
per cent) and women (10.37 per cent) did a bit of both. Self-employment
mainly comprised working on own farms and with livestock. In addition,
some men and women also had their own businesses such as petty trading,
tailoring or making taddi (a local alcoholic beverage). Wage labour was
mainly on-farm but some men also worked off-farm (construction, transport
or as an employee in a small shop). The remaining ten women (6.9 per cent)
and four men (2.7 per cent) reported no market work and have been excluded
from further analysis. Data from the household surveys are used to compare
time allocation decisions of men and women from SHG households with
those from non-SHG households.
During 2002, an additional survey was conducted with members of all
the SHGs that had completed at least one loan cycle. A total of 397 married
women belonging to twenty-seven SHGs were interviewed as part of this
survey. This survey was conducted mainly with the objective of investigating
the paradoxical findings that emerged from the household survey. For the
rest of the article, this is referred to as ‘the SHG survey’.
The SHG survey data included modules on household characteristics,
primary market work of the SHG woman and her husband, and use and
repayment of SHG loans. Given various constraints, detailed time-use data
could not be collected in this survey; instead the focus was on collecting in-
formation regarding the types of market work that men and women did. This
is likely to give an accurate idea of how men and women spend most of their
work time, given that most men (81.69 per cent) and women (89.63 per cent)
report only self-employment or only wage labour in the household surveys.
On average, groups comprised 14.7 members and had completed an average
of 3.78 loan cycles. The average loan amount was Rs 26,138.2 per group
or Rs 1,777.7 per woman and this only occasionally varied from cycle to
cycle. Loans were usually divided equally among group members, although
members of two SHGs pooled their loans for joint projects. Loans were
mainly used to meet households’ productive and consumption requirements
and in some cases to finance self-managed enterprises. The repayment rate
was reported to be 100 per cent. SHG households had average landholdings
of 2.5 acres and 52 per cent of them fell below the poverty threshold. In this
sample, sixty-three women (15.9 per cent) reported self-employment as their

7. There is an intense debate surrounding these official poverty figures (for details see Deaton
and Drèze, 2002). Here, income is net of costs but not of loan repayments.
726 Supriya Garikipati

primary market work and 296 women (74.6 per cent) reported wage work
as their primary work. The remaining thirty-eight (9.6 per cent) reported no
market work and were dropped from further analysis. Data from the SHG
survey are used to study the determinants of the type of market work done
by SHG women.
During 2002 and 2003, we also conducted a complementary series of
participatory focus group and individual interviews with thirty-eight respon-
dents of the above surveys. Stratified sampling was used to select respondents
for these interviews. The SHG survey respondents were grouped into four
categories based on their loan use. About ten women were selected from each
of these four groups for the focus groups. The unstructured interviews were
designed to focus on experiences that could not be captured by conventional
survey techniques. They are used to further understand the findings of this
study.

CREDIT PROGRAMME PARTICIPATION AND TIME USE BY GENDER

As mentioned earlier, detailed time-use data were collected in the house-

hold surveys from the male or the female respondent from married house-
holds. For the purpose of this study, routine activities carried out by men
and women were broadly divided into four categories: Self-employment,
Wage-work, Housework and Leisure. Self-employment includes work on the
household’s agricultural and non-agricultural assets. Wage-work includes
both farm and off-farm work for which wages were paid on a daily basis.8
Housework includes time spent in all reproductive work, including childcare,
care of elderly and house repairs. Leisure includes time allocated to sleeping,
recreational activities like visiting friends and relatives in the evenings, but
does not include time allocated to routine activities like personal care and
having meals.
Table 1a shows the time spent on these activities by the surveyed men and
women. The time use is in hours averaged over two typical days, one from
each of the two main agricultural periods (Kharif and Rabi). A gendered
pattern of time use is apparent immediately. The t-statistic of comparing
the mean of male time-use variables with female time-use variables differs
significantly. Men work about two hours less than women and consequently
enjoy more leisure time. Women also did most of the housework, clocking up
over four hours when compared to thirty minutes of male time in similar jobs.
These differences are comparable to other studies for developing countries
(Khandker, 1988; Newman, 2002; World Bank, 2001).

8. Note that nearly all the wage work reported was of this type. Two of the respondents
reported that they were working for a government office, but they were not included in the
main analysis.
Microcredit and Women’s Empowerment in Rural India 727

Table 1a. Male and Female Time Use (hours per day)
Activity Male respondents, n = 142 Female respondents, n = 135 t-statistic

Self-employment 6.00 (4.05) 2.66 (3.25) 7.60∗∗∗

Wage-work 2.33 (3.19) 3.48 (3.09) −3.51∗∗∗
Housework 0.50 (0.92) 4.30 (1.84) −21.64∗∗∗
Leisure 9.94 (2.04) 8.82 (1.32) 5.45∗∗∗

∗∗∗
Significant at the 1% level.
Notes: Standard deviations are given between parentheses.
t-statistic compares mean values of variables for men and women in the sample.
Note that the total hours do not add up to 24 because time spent on routine tasks like eating and personal
care have not been included.
Source: Author’s calculations based on household surveys conducted in Kharif 2001 and Rabi 2002.

The striking observation from our time-use data is that although women
spend less time in market work than men, they spend nearly 60 per cent of
this in wage work which is significantly more than men. Furthermore, they
spend far less time in self-employment than men, who spend nearly 75 per
cent of their market time in this type of work. The suggestion here is that
women work mainly for wages, while men work mainly on own assets. This
‘gender-based division’ of market work for rural India has been observed
in other studies as well (Chaudhry, 1994; da Corta and Venkateshwarlu,
1999). The population census data for India also corroborates this finding.
According to census figures, by 2001, 43.4 per cent of female workers were
classified as agricultural labourers compared to 27.4 per cent of male workers
(Government of India, 2001). This trend of feminization of agricultural wage
work is stronger in the south. For instance, according to the census for AP,
by 2001, 60.7 per cent of female rural workers were agricultural labourers
compared to 37.8 per cent of male workers.
Lending to women may help challenge the gendered work patterns ob-
served in the time-use data. Women could use their loans to buy productive
assets, which may help improve the value of their market time and also
enable them to bargain for a reduction in their domestic burdens. As a first
step towards investigating whether lending to women has resulted in such a
shift, we compare the time use by men and women from SHG households
with those from non-SHG households.
Table 1b shows male and female time use by credit programme partici-
pation of the female spouse of household. With respect to the 135 female
respondents, the table shows that the time-use variables do not differ much
when comparing SHG women with non-SHG women. For the 142 male
respondents, however, the amount of time spent in self-employment, wage
work and leisure differ significantly when comparing men whose wives are
SHG members with those whose wives are not. The t-statistic for time spent
in self-employment is positive but for wage work and leisure it is negative.
This suggests that men with SHG wives spend more time working, specif-
ically in self-employment and less so in wage work, and enjoy less leisure
728 Supriya Garikipati

Table 1b. Time Use, by Gender and Credit Programme Participation

(hours per day)
Male respondents, n = 142 Female respondents, n = 135
SHG Non-SHG SHG Non-SHG
Activity n = 61 n = 81 t-statistic n = 50 n = 85 t-statistic

Self-employment 6.97 (4.35) 5.27 (3.66) 2.47∗∗ 2.70 (3.05) 2.64 (3.37) 0.10
Wage-work 1.77 (3.13) 2.77 (3.40) −1.91∗ 3.58 (3.31) 3.42 (2.96) 0.29
Housework 0.47 (0.83) 0.52 (0.99) −0.32 4.50 (1.92) 4.19 (1.79) 0.93
Leisure 9.55 (2.06) 10.23 (1.98) −1.99∗∗ 8.73 (1.61) 8.87 (1.17) −0.59

∗
Significant at the 10% level,∗∗ Significant at the 5% level,∗∗∗ Significant at the 1% level.
Notes: Standard deviations are given between parentheses.
t-statistic compares mean values of variables for men and women in the sample.
Source: As for Table 1a.

time as compared to the averages of these variables for the men with non-
member wives. Note that these differences become shaper when we compare
households that have been in the SHG programme for over three years with
non-participating households.
The time-use outcomes presented in Tables 1a and 1b suggest two things.
First, there is a gendered division of market work — women are mainly
engaged in wage work and men mainly in self-employment. Second, par-
ticipation in the credit programme does not help women move away from
wage work, but it helps their husbands spend more time in self-employment.
In other words, lending to women seems to deepen the gendered division of
market work by helping men consolidate the amount of time they spend in
self-employment. Clearly, these results require further investigation. What
follows is a detailed econometric analysis of all factors that determine male
and female time use to see if these preliminary results hold in a multivariate
world.

DETERMINANTS OF MALE AND FEMALE TIME USE

The Empirical Models and Description of the Variables

In this section, data from the household surveys are used to investigate
the determinants of male and female time use. The focus is on testing
whether microcredit helps women spend more time in self-employment,
which is associated with better pay and higher social status compared to
wage work. This is a particularly relevant enquiry for rural India because,
as discussed above, women here are more heavily involved in agricultural
wage work when compared to men, who spend most of their work time in
self-employment.
One aim of the analysis is to separate the substitution effects of asset
incomes on time use from the bargaining effects. In this analysis, substitution
Microcredit and Women’s Empowerment in Rural India 729

effects are the direct outcome of an increase in women’s income from assets
that result from access to credit, while bargaining effects are the outcome of
an increase in women’s agency. To the extent that asset incomes influence a
woman’s behaviour in a pure economic sense, we expect to observe her doing
more self-employment and less wage work or/and enjoying less leisure. To
the extent that asset incomes influence behaviour through a bargaining effect
via a change in her agency, we expect to observe a fall in the time she
spends on housework. Whether or not this translates into more housework
for her husband depends on the (cultural) norms that determine how domestic
responsibilities are distributed.
The determinants of time use are estimated separately by gender, because
the aim is to understand how the male and female time-use decisions from
SHG households differ not from each other but from members of the same
gender from non-SHG households. The linear-in-the-variables equation used
to estimate the determinants of respondent’s time allocation is of the form
Ti j = αT Hi + βT Vi + εi j (1)
where Ti j is the time that respondent i allocates to task j. We consider a
total of four tasks: Self-employment, Wage-work, Housework and Leisure,
and hence four time-use models are estimated for men and women sepa-
rately. Hi is a vector of household characteristics (e.g., age and education
of respondent), Vi is a vector of village characteristics, αT and βT are scalar
parameters to be estimated and εi j is an unobserved error term reflecting,
in part, unmeasured determinants of Ti j that vary over households. The
household and village characteristics used in the models are given below.9
Duration: in the female time-use models, the number of years the woman
has been a member of the credit programme is indicated. In male models,
the length of his wife’s membership is indicated. Non-members in all cases
are coded as 0.
Age2 : respondent’s age measured in years. Squared term in age is included
to examine the non-linear effect of age.
Education: the educational background of the respondent. It takes the values
0 ( = illiterate), 1 ( = high school) and 2 ( = beyond school).
Sons: number of sons the respondent has.
Daughters: number of daughters the respondent has.
H-land: the amount of wetland owned by the husband in acres.
W-fallback: coded as 1 if wife’s maternal family owned more than 3 acres
of wetland. This variable is used as a proxy for the wife’s fallback position.

9. Wages could not be included as a determinant because nearly all wage work was agricul-
tural, and there was little variation within a gender category. Returns on self-employment
had to be computed and this required subjective assumptions regarding land and labour
productivity. Computed returns were found to be highly correlated with the amount of wet-
land owned by the husband (r = 0.209, p = 0.000). Rather than include wages and returns
on self-employment directly, factors that could influence these, such as age, education,
landownership and women’s participation in the credit programme are considered.
730 Supriya Garikipati

It may also indicate whether the woman was given dowry at the time of
marriage.10
F-head: coded as 1 if the head of the household is reported to be female.
O-loans: coded as 1 if the household received credit from sources other than
the SHG-programme in the last three years.
Dependency: the proportion of household members aged 13 or over divided
by household size. This number indicates the dependency burden on the
working members of the household. A lower number denotes greater depen-
dency.
Dwelling: coded as 1 if the external walls of the house are made of concrete
and the house has a durable roof (tiles or similar materials) and 0 otherwise.
This variable indicates the relative economic status of the household.
Caste: coded as 1 if the household is from the lower castes (Scheduled Castes
or Scheduled Tribes).
Market: a village-level dummy indicating the distance from the main market.
Coded as 1 if the household is from a village which is 10 km or more from
the market, and 0 if less.
For Duration, a positive sign of the coefficient is expected in the Self-
employment model and a negative sign is expected in the Wage-work model
for women: participation in the credit programme is likely to increase
women’s time in self-employment and decrease the time spent in wage
work. For control variables, Age2 and Education, a negative sign of the co-
efficient is expected in the Wage-work model for men and women: increases
in an individual’s age and education are likely to decrease the time spent in
physically demanding and low-skilled wage work. In addition, for Educa-
tion and Dwelling, a positive sign is expected in the Self-employment model
for men and women: increases in education and wealth are likely to increase
time spent in self-employment. For H-land, a positive sign of coefficient is
expected in the Self-employment model and a negative sign in the Wage-work
model for men: increase in access to wetland is likely to increase male time
in self-employment and decrease the time in wage work. Wage work is less
likely to be carried out by landowners because of low pay and social status
associated with it. There are no explicit expectations on the signs of the
remaining variables.

Endogeneity of Credit Programme Participation

Studies examining the impact of programme participation routinely suffer

from possible bias due to endogeneity of the decision to participate in the

10. Although the practice of giving and receiving dowry is common in the survey area, col-
lecting direct data on dowry proved to be problematic. The practice is illegal and there was
a general perception among the respondents that the survey enumerators being ‘young and
educated’ would be critical of those who admit to have taken or received dowry.
Microcredit and Women’s Empowerment in Rural India 731

programme and the unobserved household characteristics. The unobserved

heterogeneity between the members of the credit group and non-members
includes the unobserved attitudes and characteristics of the husbands, wives
and other family members, including pre-existing women’s autonomy. It
seems quite likely that more autonomous women are more likely to be able
to join a credit programme but these women are also more likely to spend
their time in activities that are better remunerated and of higher status. If
this unobserved heterogeneity is not accounted for then its effects will be
captured by the variable measuring programme participation and cause its
statistical effects to be exaggerated.
In econometric terms, this means that in equation (1), the covariate mea-
suring programme participation (Duration) and other unmeasured casual
variables collapsed into the error term εi j are correlated. Hence, using
Duration as an independent variable is problematic. Consider the follow-
ing reduced form equation for the level of participation in the credit pro-
gramme, where level of participation will be taken to be the duration of
credit programme participation,
Durationi = α D Hi + β D Vi + μi for Criterioni = 1
(2)
Durationi = 0 for Criterioni = 0,
where Hi is a vector of household characteristics and Vi is a vector of village
characteristics, α D and β D are unknown parameters, μi is a non-systematic
error that reflects unmeasured determinants that vary over households and
Criterioni = 1 indicates that the household meets the criterion for partici-
pating in the credit programme.
The time-use outcome conditional on the duration of programme partici-
pation Durationi is given as:
Ti j = αT Hi + βT Vi + χT Durationi + εi j (3)
where Hi is a vector of household characteristics and Vi is a vector of village
characteristics as described above, αT , βT and χT are unknown parameters
and εi j is a non-systematic error that reflects unmeasured determinants of
Ti j that vary over households. The estimation issue arises as a result of
the possible correlation of εi j from equation (1) with μi from equation (2).
Econometric estimation that does not take these correlations into account
may yield biased estimates of the parameters of equation (1) due to the
endogeneity of credit programme participation Durationi . In such a case, it
is valid to use the estimates to predict values of Ti j given values of Durationi ,
but the estimate does not recover the causal effect of Durationi on Ti j .
Several econometric techniques, notably instrumental variables estima-
tion, are available to correct for the possible confounding effects of
systematic variation between participants and non-participants.11 An

11. The IV technique and other methods are discussed in Heckman (2008). For studies that
have used the IV technique, among others, see Khandker (1998) and Pitt et al. (2006). The
732 Supriya Garikipati

instrumental variable z is one that is correlated with the independent variable

Durationi but not with the error term εi j . The instrument is then used to ob-
tain a consistent IV estimator for Durationi . To construct a valid instrument
— such that it is correlated with Durationi but not to the error term — the
official rules governing the formation of SHGs were used. These rules and
their implementation in practice are described below.
The formation of SHGs is governed by explicit policy directives from
NABARD. The prevailing rules require that groups consist of members who
are from a similar socio-economic background, which in rural India means
sharing cultural affinity in terms of caste grouping and belonging to a similar
wealth or income stratum (NABARD, 2003, 2007). NABARD proactively
encourages the adoption of these rules by bank officials and NGOs involved
in group formation via its training material (see NABARD, 2007).
Various studies suggest that serious consideration is given to these rules
during the formation of credit groups. In his study of SHGs in Gulbarga
district of Karnataka, Harper (2002) describes how bank managers are anx-
ious to ensure that women forming groups share close caste ties. Badatya
et al. (2006) also mention such practices in their study of SHGs in three
districts of AP, viz. Chitoor, Nizamabad and Warangal. Our survey data also
confirm that women from a neighbourhood have an increased probability of
forming a group if they belong to a similar socio-economic background. Ex-
plaining the need for this policy by way of outlining the ‘ills’ of groups that
have women from different socio-economic backgrounds, the manager of
Sanghameshwara Grameen Bank (the bank serving the survey area) wrote:
‘These groups are difficult to manage and are likely to split up because of
lack of unity among the women’.12
Focus group interviews suggest that women preferred to form SHGs with
others from their caste and wealth group due to reasons of trust and cultural
affinity, but also to minimize the transaction costs associated with screening
and monitoring group members as encumbrances related to caste and class
hierarchies did not have to be maintained when members belonged to a
similar background. Describing her relationship with others from her group,
here is what one SHG woman says:

Caste is very important when you are in a group. Look at our group — we are all from the
same caste and look how well we work . . . because we already know and understand each
other. I just know that they [members of the group] will not cheat me. (G9W1)
Her [another member of the same SHG] house is directly in front of mine. We are like
sisters. . .if you don’t find me in my house, you will find me in hers. . . So when there was
the talk of forming an SHG in our neighbourhood, we both decided to be in the same group.
(G9W1)

econometric methods used here are similar to those used in Pitt et al. (2006), without the
village fixed effects.
12. Mr Pradhaman Achari, bank manager of Sanghameshwara Grameen Bank, e-mail corre-
spondence (19 November 2002).
Microcredit and Women’s Empowerment in Rural India 733

Harper (2002) reports similar behaviour among women forming groups in

Angul district of Orissa. In their report which assesses SHG–bank linkage
programmes, Kropp and Suran (2002) also observe neighbourhood and caste
affiliations among groups they interviewed. MYRADA (2002) also makes
similar observations for SHGs in Boodhikote district of Karnataka.
In our sample, all SHGs were neighbourhood groups which meant that
women from the same neighbourhood came together to form groups; the
average group size was around fifteen members. Furthermore, 72.79 per
cent (n = 397) of SHG members belonged to the same sub-caste as the
majority in their group. We used these two pieces of data to construct a
relevant instrument. This exercise needed utmost care as caste can also
influence time use. We had to come up with an instrument which did not
measure women’s caste category directly but instead gave us a measure of
the probability of a woman joining a neighbourhood credit group with others
from a similar social background.
With this in mind, we wanted to identify the number of households within
a neighbourhood cluster that belonged to similar social background. We be-
gan by identifying the different neighbourhood clusters within each village.
For this we used the electoral roles which list addresses associated with
specific neighbourhoods. The electoral roles were also used to identify the
caste category of all the households within each neighbourhood. Using only
the broader constitutionally recognized caste classifications gave us little
variation within a neighbourhood, so sub-castes were used for this exercise.
For instance, under the caste category of Scheduled Tribes, in the survey
villages, we have three sub-castes: Chenchu, Lambada and Yerukali. Table
A1 in the Appendix lists all the sub-castes from the villages surveyed.
Using this information we constructed the following instrument,
InstCaste — a dummy variable coded as 1 if the respondents belonged
to a sub-caste that had at least fifteen households in his/her neighbourhood.
Note that this instrument does not capture an individual’s caste category
(and hence the ability or inability to use time in a certain way) but only the
number of households within an individual’s neighbourhood who belonged
to a similar social background.
To illustrate the identification strategy used by this study, consider Figure
1 which depicts a neighbourhood cluster. Coloured-in houses belong to sub-
caste x and uncoloured houses belong to sub-caste y. In this neighbourhood
there are a total of seventeen houses that belong to sub-caste x and four that
belong to sub-caste y. For the instrumental variable InstCaste, a household
from sub-caste x takes the value 1 and a household from sub-caste y takes
the value 0. Ceteris paribus, households from x sub-caste have a relatively
higher probability of forming a neighbourhood credit group with fifteen
other households from a similar background when compared to households
from y sub-caste.
Two-stage instrumental variable estimation of the determinants of male
and female time use can be achieved by treating as identifying instruments
734 Supriya Garikipati

Figure 1. Illustrating the Identification Strategy: A Neighbourhood Cluster

with Two Sub-castes

a dummy variable for programme choice interacted with all the exogenous
variables. The idea is that all of the exogenous variables have an effect on
self-selection into the programme only for those with a sufficient number
of within-caste households in their neighbourhood — as only they are able
to effectively decide whether or not to participate — but influence time-use
outcomes for all. Parameter identification requires that living in a neighbour-
hood with a sufficient number of within-caste households (the eligibility
criterion) does not affect the time-use outcomes conditional on programme
participation, although a person’s caste itself may affect time-use outcomes.
It is important to note that the variable Caste is not an exclusion restriction
here. It is one of the independent variables in the vector Hi of equation (3).
In the first stage, the endogenous covariate Duration is regressed on all
exogenous variables, including the identifying instrument, as follows:

Durationi = α D Hi + β D Vi + δ D InstCastei + μi (2’)

Note that for women who live in neighbourhoods with insufficient number
of within-caste households (InstCastei = 0), duration of programme partic-
ipation is deterministically 0 (Durationi = 0). The predicted values from
these regressions are obtained. We run the above regression separately for
men and women in the sample and report the results in Appendix Table
A2. The second stage in the two-stage least squares estimation is simply the
estimation of equation (3), but after replacing Durationi with the predicted
p
Durationi . The predicted values for men are used in the male time-use mod-
els and the predicted values for women in the female time-use models. The
time-use equation for male and female respondents can be written as:

Ti j = αT Hi + βT Vi + χT Durationpi + εi j (3’)

Standard two-stage least squares estimation provides consistent estimates of

this model. We use the ivregress (2SLS) command in STATA to compute
the time-use equations.
Microcredit and Women’s Empowerment in Rural India 735

Table 2. Descriptive Statistics of the Independent Variables Used in the

Time-Use Models
Male respondents, n = 142 Female respondents, n = 135
SHG Non-SHG SHG Non-SHG
Variable n = 61 n = 81 t-statistic n = 50 n = 85 t-statistic

INDIVIDUAL CHARACTERISTICS
Age 38.46 (8.96) 36.99 (9.41) 0.95 35.84 (6.62) 37.89 (8.88) −1.42
Education 0.77 (0.92) 0.53 (0.88) 1.57 0.14 (0.45) 0.06 (0.28) 1.28
Sons 1.23 (1.19) 0.99 (0.99) 1.32 1.78 (1.22) 1.56 (1.15) 1.03
Daughters 1.08 (1.14) 1.33 (1.39) −1.15 1.00 (1.07) 0.92 (0.99) 0.45
H-land 2.74 (4.84) 2.27 (3.05) 0.66 1.92 (2.57) 2.01 (2.19) −0.23
W-fallback 0.38 (0.49) 0.30 (0.46) 1.01 0.32 (0.47) 0.27 (0.45) 0.61
HOUSEHOLD CHARACTERISTICS
F-head 0.00 (0.00) 0.01 (0.11) −0.87 0.10 (0.30) 0.16 (0.37) −1.04
O-loans 0.21 (0.41) 0.25 (0.43) −0.47 0.30 (0.46) 0.20 (0.40) 1.32
Dependency 17.18 (12.27) 17.03 (9.33) 0.08 14.95 (7.70) 19.25 (11.06) −2.42∗∗
Dwelling 0.21 (0.41) 0.16 (0.37) 0.80 0.12 (0.33) 0.15 (0.36) −0.53
Caste 0.31 (0.47) 0.26 (0.44) 0.68 0.36 (0.48) 0.29 (0.44) 1.24
Market 0.59 (0.50) 0.47 (0.50) 1.43 0.68 (0.47) 0.37 (0.48) 3.69∗∗∗

∗
Significant at the 10% level,∗∗ Significant at the 5% level,∗∗∗ Significant at the 1% level.
Notes: Standard deviations are given between parentheses.
t-statistic compares mean values of variables for men and women in the sample.
Source: Author’s calculations based on household surveys conducted in Kharif 2001 and Rabi 2002.

Data Description and Empirical Results

Table 2 provides descriptive statistics of all the variables used in the em-
pirical analysis by gender and SHG membership. The table shows that the
statistics for most of the variables included in the empirical model do not
differ much when comparing SHG members with non-members, suggesting
that the treatment and control groups are comparable. The only significant
differences are in the case of female respondents with respect to the variables
Dependency and Market. For these variables, the t-statistic of comparing the
mean of the members with non-members differs significantly. SHG women
are more likely to come from large households with fewer working mem-
bers and are more likely to live in villages further away from the nearest
market town as compared to the averages of these two variables for the
non-members.
Tables 3a and 3b provide the results of the second-stage time-use regres-
sions for male and female respondents respectively. Each column represents
a separate time-use model, which examines whether membership of the
credit programme affects the time used by the respondent in that particu-
lar activity. The central result from the time-use models is that lending to
women affects their husbands’ time use by helping them spend significantly
more time in better remunerated and socially respectable self-employment
and less time in wage work. However, it fails to help women challenge
the conventional demands surrounding work and leisure. In particular, the
coefficient for Duration is found to be statistically significant in three of the
four male time-use models: (3–1), (3–2) and (3–4). It has a positive sign in
736 Supriya Garikipati

Table 3a. Determinants of Male Time Use, n = 142 (Second-stage

Regressions)
3–1 3–2 3–3 3–4
Self-employment Wage-work Housework Leisure

PROGRAMME RELATED VARIABLE

Duration (est) 0.494 (3.30)∗∗∗ −0.214 (−2.24)∗∗ 0.007 (0.22) −0.192 ( − 2.53)∗∗
INDIVIDUAL CHARACTERISTICS
Age2 0.0001 (1.75)∗ −0.001 (−3.23)∗∗∗ −0.0001 (−0.99) 0.0001 (1.52)
Education −0.237 (−0.59) −0.258 (−1.91)∗∗ −0.129 (−1.36) 0.531 (2.60)∗∗∗
Sons −0.148 (−0.44) 0.017 (0.05) 0.008 (0.11) 0.270 (1.56)
Daughters −0.011 (−0.04) 0.055 (0.24) 0.076 (1.12) −0.024 (−0.17)
H-land −0.007 (−0.08) −0.075 (−1.71)∗ 0.008 (0.36) 0.004 (0.09)
W-fallback −0.797 (−1.08) 0.297 (0.50) −0.114 (−0.65) 0.138 (0.37)
HOUSEHOLD CHARACTERISTICS
O-loans −0.512 (−0.64) 0.021 (0.03) −0.158 (−0.84) 0.086 (0.21)
Dependency −0.071 (−1.93)∗ 0.053 (1.79)∗ 0.004 (0.46) 0.016 (0.88)
Dwelling 0.739 (0.81) 0.282 (0.38) −0.204 (−0.94) −0.351 (−0.75)
Caste 0.061 (0.08) −0.391 (−0.62) 0.048 (0.26) −0.176 (−0.44)
Market 0.720 (1.07) −0.662 (−1.22) 0.107 (0.67) 0.346 (1.01)
Constant 5.078 (3.84)∗∗∗ 4.234 (3.97)∗∗∗ 0.596 (1.91)∗ 8.738 (13.01)∗∗∗
Wald chi-squared (12) 26.37 25.47 5.68 16.09
Root MSE 3.799 3.066 0.900 1.932

∗
Significant at the 10% level,∗∗ Significant at the 5% level,∗∗∗ Significant at the 1% level.
Note: t-statistics are given between parentheses.
The variable F-head was dropped from the male models due to lack of variation.

(3–1), but a negative sign in (3–2) and (3–4). Of the control variables, the
coefficient for Age2 is statistically significant in (3–1), (3–2), (3–5) and (3–
6); Education in (3–2), (3–4), (3–5), (3–6) and (3–8); Sons in (3–7); H-land
in (3–2) and (3–7); Dependency in (3–1), (3–2) and (3–7) and Market in
(3–7).
With respect to duration of credit programme participation, the results
suggest that men whose wives are SHG members spend more time in self-
employment and less time in wage work even when measured against the
impact of other variables. They also spend less time in leisure when compared
to men whose wives are not SHG members. These relationships are likely
to be strengthened as duration of membership increases. It is likely that, as a
result of their wives’ SHG membership, their returns from self-employment
have increased when compared to those of wage work and hence they have
shifted their work time from wage work to self-employment. And given
that the opportunity cost of leisure time is higher for them when compared
to men who work mainly for wages, they seem to be substituting leisure
time for work, due to it being more expensive. It is worth pointing out
here that husbands of SHG women still have significantly more leisure time
than their wives. Programme membership does not impact on women’s
self-employment and there is little evidence to suggest that it may actually
increase their involvement in wage work.
Microcredit and Women’s Empowerment in Rural India 737

Table 3b. Determinants of Female Time Use, n = 135 (Second-stage

Regressions)
3–5 3–6 3–7 3–8
Self-employment Wage-work Housework Leisure

PROGRAMME RELATED VARIABLE

Duration (est) 0.037 (0.27) 0.042 (1.16) 0.011 (0.14) −0.046 (0.82)
INDIVIDUAL CHARACTERISTICS
Age2 0.001 (2.28)∗∗ −0.001 (−2.18)∗∗ −0.000 (−0.13) 0.000 (0.42)
Education 1.23 (1.75)∗ −1.945 (−2.59∗∗ 0.251 (0.58) 0.858 (2.68)∗∗
Sons −0.287 (−1.12) 0.073 (0.30) 0.258 (1.83)∗ 0.033 (0.32)
Daughters 0.290 (0.93) −0.337 (−1.13) 0.015 (0.09) 0.013 (0.11)
H-land −0.060 (−0.48) 0.023 (0.20) 0.134 (1.93)∗ −0.055 (−1.09)
W-fallback 0.345 (0.56) −0.216 (−0.36) 0.287 (0.83) 0.0003 (0.00)
HOUSEHOLD CHARACTERISTICS
F-head 0.694 (0.87) −0.215 (−0.28) −0.197 (−0.45) 0.039 (0.12)
O-loans −0.165 (−0.25) −0.191 (−0.30) 0.246 (0.67) 0.272 (1.00)
Dependency −0.010 (−0.31) −0.016 (−0.53) 0.029 (1.73)∗ 0.013 (1.00)
Dwelling 0.932 (1.12) −0.688 (−0.87) 0.277 (0.60) −0.340 (−1.00)
Caste −0.796 (−1.31) 0.312 (0.54) 0.135 (0.40) 0.255 (1.03)
Market −0.143 (−0.25) −0.193 (−0.35) 0.985 (3.09)∗∗ 0.087 (0.37)
Constant 1.381 (0.89) 5.724 (3.88)∗∗∗ 2.426 (2.84)∗∗ 8.373 (13.34)∗∗∗
Wald chi-squared (13) 15.03 15.03 23.32 16.01
Root MSE 3.060 2.917 1.690 1.242

∗
Significant at the 10% level,∗∗ Significant at the 5% level,∗∗∗ Significant at the 1% level.
Note: t-statistics are given between parentheses.

The other determinants of male and female time use are more or less in line
with theoretical expectations. Age affects men and women similarly. It has a
positive effect on time spent in self-employment but a negative effect on time
spent in wage work. Education has a negative effect on the amount of wage
work that men and women do, suggesting that better-educated individuals
may have other work opportunities available to them or that better-educated
people may avoid wage work. It has a positive effect on the time women
spend in self-employment, suggesting that education may enhance women’s
ability to seek self-employment opportunities. It also has a positive effect on
the amount of leisure men and women enjoy. This may be an income effect,
reflected in the ability of better-educated individuals to earn a higher income.
It may also indicate that better-educated individuals place a higher value on
leisure relative to consumption. Land ownership has a negative effect on
male time in wage work. Given the socially debasing nature of wage work,
landowning men are expected to shun this type of work. Having sons and the
amount of land owned by the household have a positive relation with the time
women spend in housework, suggesting that the cultural asymmetries around
sharing of housework by men and women worsen with having sons and as
a household’s wealth improves. Having a greater proportion of working
members in the household is negatively related to the time men spend in
self-employment, but correlates positively with the time spent in wage work.
738 Supriya Garikipati

This is expected as the household’s productive assets are likely to be fixed.

Living in a village further away from the main market has a positive effect
on the women’s time in housework. This is at least partly because a large
number of households in this village rely on traditional fuels like cow pats
and women spend substantial amounts of time preparing these.
Sensitivity analyses were carried out to test the robustness of the results.
The time-use models re-estimated with Age and H-land increased by 5 per
cent and different cut-off points for maternal landownership were used to
construct the variable W-fallback. The coefficients for Duration remained
relatively unaffected by these changes.
Our results from the time-use models compare with the findings of other
studies. Pitt (2000) examines the impact of microcredit on male time use
only. He finds strong evidence to suggest that participation in microcredit
substantially increases male own-cultivation through sharecropping, coupled
with a significant increase in male hours spent in self-employment and a
reduction in male hours spent in wage work. Moreover, Pitt measures the
impact of credit accessed by men separately from that accessed by women.
He finds that female credit effects are greater than male credit effects, both
in increasing sharecropping and male self-employment and reducing male
wage labour. While Todd (2001) does not examine the impact on time use
directly, she finds a noticeable shift in employment patterns of microcredit
households from irregular, low-paid daily labour to family business, with
livestock being the most widely acquired productive asset.
Overall, our estimates indicate that women work significantly longer hours
compared to their husbands, and spend most of their work time in low status
activities like wage work and housework, and microcredit does not allow
them to break this pattern. Men, on the other hand, work mainly in self-
employment and their wives’ participation in credit programmes helps them
to deepen this commitment. This indicates that although SHG programmes
target women, the actual beneficiaries are their husbands.13 The next section
investigates this paradoxical result further.

WOMEN’S LOAN USE AND THEIR TIME USE

Some Preliminary Observations

In this section the data from the SHG survey are used to understand why
lending to women helps their husbands spend more time in self-employment

13. One of the reasons why households may divert credit to men is because the returns earned
by men may be higher than those earned by women. This has been an ongoing theme in the
microcredit literature. For instance Berger (1989: 1021) notes that ‘On average, women’s
microenterprises have lower sales revenues, fewer assets, and smaller profit margins than
men’s’. Unfortunately, the survey data do not allow further exploration of this issue.
Microcredit and Women’s Empowerment in Rural India 739

Table 4. SHG Women’s Primary Market-Work by Loan Use

Family farm Household All Household Total
Loan-use Own enterprise or business assets Consumption Activities cases
(1) (2) (3) (4) (2+3+4)

Market-work
Self-employment 48 (13.37) 11 (3.06) 2 (0.56) 2 (0.56) 15 (4.18) 63 (17.55)
Wage-work 7 (1.95) 209 (58.22) 34 (9.47) 46 (12.81) 289 (80.5) 296 (82.45)
Total numbers 55 (15.32) 220 (61.28) 36 (10.03) 48 (13.37) 304 (84.68) 359 (100)

Notes: In case of multiple loan uses (9.32%), the primary use is recorded.
Percentages are given between parentheses.
Source: Author’s calculations based on household surveys conducted in Kharif 2001 and Rabi 2002.

and less time in wage work but does not help them do the same. The focus is
on investigating whether the way women’s loans are used affects their time
use, where loan procurement alone failed to do so.
Women’s loans were mainly used to meet household’s production and
consumption needs (84.7%) and only a small proportion of the SHG women
used their loans in businesses they managed or helped manage (15.3 per
cent). The loans used on household requirements were primarily used as
working capital in household farms or businesses (61.3 per cent), or used
to buy or improve assets like land and livestock (10 per cent) and towards
consumption (13.4 per cent). These figures indicate that the demand for
credit within the household is high (see also Mahajan and Ramola, 1996).
Typical examples of working capital are to buy a motor for irrigating land
or a pair of plough bullocks or equipment for a small family business.
The noteworthy point here is that loans used in the household’s productive
activities ultimately result in enhancing male assets. This is because land is
almost always owned by men (less than 3 per cent of household land was
owned by women) and most family businesses are also controlled by men.
In this respect, loans procured by women are contributing to widening the
existing resource divide between men and women.
Of loans used on women-controlled businesses, given the small loan size
(Rs 2000 to Rs 3500) women tend to invest in similar businesses which
meant severe competition and low profitability (also see Berger, 1989).
Women typically bought livestock like chickens and goats or started petty
vendor shops selling tea, groceries or a tailoring business. Just two SHGs
had pooled their loan together in order to start a group business.
As mentioned above, the SHG women were asked about their and their
husband’s primary market work. Table 4 provides an overview of the rela-
tionship between loan use and main market work for SHG women. The table
shows that 48 of the 55 women (87.3%) who used their loans on their own
enterprise report their main market work to be self-employment, while only
7 (12.7%) report this to be wage work. This, combined with results from
the previous analysis, suggests that loan use and not loan procurement is an
740 Supriya Garikipati

important determinant of female time use. Table 4 also shows that just 15 of
the 304 women (4.9%) using loans in household activities report their main
market work to be self-employment, while a staggering 289 of them (95.1%)
report this to be wage work. The suggestion here is that although women’s
loans are used to enhance household assets, they lack co-ownership in these
assets, in that they are unable to spend their time working on these assets.

An Econometric Analysis

This section presents an econometric analysis of the factors that may de-
termine SHG women’s market work. The conditional probability of women
being self-employed, conditional on the duration of programme participation
Durationi is:
Selfemploymenti = α S Hi + β S Vi + χ S Durationi + ϕi (4)
where Selfemploymenti is a dummy variable that takes the value 1 if the
woman reports self-employment as her main market work and 0 if she re-
ports this to be wage work. Hi is a vector of household characteristics and Vi
is a vector of village characteristics, α S , β S and χ S are unknown parameters,
ϕi is a non-systematic error that reflects unmeasured determinants that vary
over households. Of the household and village characteristics used in the
equation, Duration, Age2 , Education, H-land, W-fallback, F-head, Depen-
dency, Dwelling, Caste and Market are as described earlier. Three additional
variables described below are also included:
Own use: coded as 1 if the woman used a loan in an enterprise managed by
her solely or in partnership with others and 0 otherwise.
Peer effect: indicates the proportion of women in the respondent’s SHG
who report self-employment as their main work. This is a proxy for the
influence exerted by the respondent’s peer group in encouraging her to take
up self-employment.
Sons: coded as 1 if the woman has one or more sons.
We are once again confounded with the endogeneity problem described
earlier that precludes the use of the variable Durationi directly. The econo-
metric methods used to tackle the endogeneity of credit programme partic-
ipation are essentially the same as before with the difference that to obtain
the IV estimates we use IV probit, an estimation procedure that fits mod-
els with dichotomous dependent variables where the regressor is endoge-
nously determined. In the first-stage equation of the two-stage IV probit
estimation, the endogenous regressor is instrumented using ordinary least
squares,
Durationi = α D Hi + β D Vi + δ D InstCastei + μi (2’)
The first-stage regression is presented in Table A2 of the Appendix
(column 3). The second stage, like before, is simply the estimation of
Microcredit and Women’s Empowerment in Rural India 741

Table 5. Descriptive Statistics of the Variables Used in the Women’s Primary

Market-work Model
SHG women (n = 359)
Own-use (n = 55) Household-use (n = 304) t-statistic

DEPENDENT VARIABLE
Self-employment 0.87 (0.34) 0.05 (0.22) 23.53∗∗∗
PROGRAMME RELATED VARIABLE
Duration 3.61 (1.32) 5.29 (3.73) −6.05∗∗∗
Peer effect 35.64 (20.61) 10.08 (16.56) 8.70∗∗∗
INDIVIDUAL CHARACTERISTICS
Age 34.49 (10.67) 30.08 (9.19) 2.87∗∗∗
Education 0.15 (0.40) 0.14 (0.37) 0.07
Son 0.89 (0.31) 0.87 (0.34) 0.46
H-land 0.97 (1.62) 0.44 (0.95) 2.34∗∗
W-fallback 0.47 (0.504) 0.24 (0.426) 3.27∗∗∗
HOUSEHOLD CHARACTERISTICS
F-head 0.13 (0.34) 0.04 (0.20) 2.53∗∗
O-loans 0.11 (0.315) 0.09 (0.290) 0.39
Dependency 15.90 (9.89) 12.69 (6.70) 2.31∗∗
Dwelling 0.31 (0.466) 0.13 (0.311) 2.80∗∗
Caste 0.27 (0.45) 0.28 (0.45) −0.06
Market 0.62 (0.49) 0.57 (0.50) 0.68

∗∗
Significant at the 5% level,∗∗∗ Significant at the 1% level.
Notes: Standard deviations are given between parentheses.
t-statistic compares mean values of variables for men and women in the sample.
Source: Author’s calculations based on household surveys conducted in Kharif 2001 and Rabi 2002.

p
equation (4), but after replacing Durationi with the predicted Durationi .
It is written as:
Selfemploymenti = α S Hi + β S Vi + χ S Durationpi + ϕi (4’)
We use the ivprobit command in STATA to compute this equation. For
Duration, Own use and Peer effect a positive sign of the coefficient is ex-
pected: the probability of women in self-employment is likely to increase
if the woman participates in the credit programme, uses her loan in a self-
managed enterprise, and as the proportion of women in self-employment in
her SHG increase. The expectations for the signs of Age2 , Education and
H-land are as before. There are no expectations for the signs of the remaining
variables.
Table 5 provides descriptive statistics of all the variables used in the
empirical investigation for SHG women by their loan use. The table shows
that the statistics for Self-employment, Duration, Peer effect, Age2 , H-land,
W-fallback, F-head, Dependency and Dwelling differ significantly when
comparing women who use loans in own enterprises with those who do
not. Women who invest in own enterprises spend significantly more time
in self-employment, have been SHG members for fewer years, have other
members in their credit group doing mainly self-employment, are somewhat
742 Supriya Garikipati

Table 6. Determinants of SHG Women’s Primary Market Work, n =

359 (Second-stage regression)
DEPENDENT VARIABLE: Self-employment
6–1 SHG women (n = 359)

PROGRAMME RELATED VARIABLE

Duration 0.029 (0.32)a
Own use 2.640 (7.51)∗∗∗
Peer effect 0.039 (4.79)∗∗∗
INDIVIDUAL CHARACTERISTICS
Age2 0.0000 (0.33)
Education 0.609 (1.53)
Son 0.706 (1.41)
H-land 0.165 (1.43)
W-fallback −0.061 (−0.16)
HOUSEHOLD CHARACTERISTICS
F-head 0.027 (0.05)
O-loans −0.403 (−0.86)
Dependency 0.015 (0.79)
Dwelling 0.412 (1.20)
Caste −0.156 (−0.49)
Market −0.424 (−1.29)
Constant −3.625 (−4.28)∗∗∗
Wald chi-squared (14) 88.93

∗
Significant at the 10% level,∗∗∗ Significant at the 1% level.
Note: t-statistics are given between parentheses.
Source: Author’s calculations based on household surveys conducted in Kharif 2001 and Rabi 2002.

older, enjoy better fallback positions, have husbands who own relatively
more wetland, are more likely to be considered as head of the household,
live in households with lower dependency burdens and which are relatively
better-off when compared to the averages of these six variables for the
women who use loans on household needs. Taken together, the suggestion
here is that it is mainly women from better-off households with somewhat
better household statuses who are able to invest in their own businesses.
Table 6 provides the results of SHG women’s main market-work model.
Overall, the results indicate that loan use plays a significant role in deter-
mining women’s market work. In particular, statistically significant coef-
ficients are found for Own use and Peer effect. Both variables have the
expected sign. None of the control variables have statistically significant
coefficients — although some like Education, Son and H-land come very
close.
With respect to loan use, the results show that if women use their loans in
an enterprise they manage, the probability of self-employment being their
main market work increases. With respect to the influence of women’s peer
group, the results show that if others in the woman’s credit group are mainly
self-employed, the probability of the woman spending her market time in
a similar way increases. Taken together with the earlier results from time-
use models, the suggestion here is that although loan procurement alone is
unlikely to influence woman’s time use, how she uses her loan influences
Microcredit and Women’s Empowerment in Rural India 743

Table 7. Male and Female Time Use by Loan Use (hours per day)
Male (n = 39) Female (n = 34)
Own-use Household-use Own-use Household-use
Activity (n = 16) (n = 23) t-statistic (n = 11) (n = 23) t-statistic

Self-employment 8.10 (3.59) 7.61 (4.15) 0.39 4.98 (1.74) 2.64 (3.22) 2.25∗∗
Wage-work 1.03 (2.56) 1.22 (2.67) −0.23 0.64 (1.95) 3.93 (3.36) −3.01∗∗∗
Housework 0.46 (0.85) 0.55 (0.89) −0.32 5.30 (0.93) 4.32 (1.45) 2.03∗∗
Leisure 9.02 (1.35) 9.54 (2.02) −0.90 9.03 (1.84) 8.62 (1.61) 0.66

∗∗
Significant at the 5% level,∗∗∗ Significant at the 1% level.
Notes: Standard deviations are given between parentheses.
t-statistic compares mean values of variables for men and women in the sample.
Source: Author’s calculations based on household surveys conducted in Kharif 2001 and Rabi 2002.

her time use significantly. If her loan is used in a business that she manages
or helps manage then this is likely to have a significant beneficial impact
on the way her market time is spent — specifically, this will help her spend
more time in better remunerated and socially superior self-employment.
To test for the robustness of the relationship between women’s loan use
and time use, advantage was taken of the fact that some households were
surveyed in both the ‘household surveys’ and the ‘SHG survey’. A new
sample was created by matching households included in both surveys. A
‘matched household’ had either the husband or the wife interviewed in the
household surveys, and the wife interviewed in the SHG survey. Although
106 households were included in both of the surveys, only seventy-three
definite matches were found — with thirty-nine men and thirty-four women
respondents in the household surveys. For these men and women, we have
the 24-hour recall time-use data (from the ‘household surveys’) and detailed
loan use data (from the ‘SHG survey’). All respondents in this sample
identified themselves as participating in market work. This sample is used
to check if the results regarding primary market work hold when detailed
time-use data are considered.14
Table 7 provides the detailed time use for men and women using the
matched sample by loan use. The table shows that when comparing women
who use loans in their own businesses with women who do not, the t-statistic
for the variables Self-employment, Wage-work and Housework differ signif-
icantly. The findings show that women who use their loans in their own
businesses spend more time in self-employment and housework, and less
time in wage work when compared to the averages of these three vari-
ables for the women who use loans on family requirements. The findings
on self-employment and wage work suggest a direct substitution effect and
support our earlier findings, but the result on housework requires some

14. Despite the small size of this sample, the time-use data for these men and women closely
resemble the time-use data for men and women from the full sample. Details are available
from the author upon request.
744 Supriya Garikipati

deliberation. Standard economic theory suggests that the time a woman al-
locates to housework declines as a result of the indirect bargaining effect of
an increase in her income as she increases time spent in self-employment.
In our sample, women engaged in self-employment save nearly an hour
when compared to those working for wages, but instead of enjoying more
leisure time, they spend this extra hour on housework. If greater involve-
ment in housework is a free choice then such work produces greater utility
for women than that gained from a similar amount of extra leisure time. A
more likely explanation is that self-employment offers women greater flex-
ibility in their work schedules which in turn compels them to take on extra
housework. Whatever the reason for the extra hour spent on housework, it is
apparent that women are unable to use the leverage of microcredit to nego-
tiate for less of it. The division of housework is so entrenched in patriarchal
norms that exogenous factors like credit are ineffective in helping women to
overcome it.
With respect to male time use, the table shows that, statistically speaking,
the different ways men use their time are not influenced by their wives’ loan
use. Although these differences are not statistically significant, men whose
wives invest in their own businesses are seen to spend more time in self-
employment and less time in wage work than others. This may reflect the pos-
sibility that men whose wives use loans to start their own businesses may not
be excluded from accessing their wives’ assets and are able to work on these.
Note that an earlier result suggested that women lacked co-ownership of
male-owned assets and were excluded from working on these even when they
invested their loans in these assets (see Table 4). The reverse, however, does
not seem to hold. Men, it seems, are not excluded from working on female-
owned assets in the same way that women are excluded from male assets.
The focus group interviews were useful for further understanding the
diverse experiences of those SHG women who used their loans for household
needs and those who used it for their own businesses. G3W7, G3W11,
V2W6 and V11W3 are women whose loans were used as working capital
in family farms and G5W12, G7W2 and V4W9 had voluntarily used their
loans to avert a household crisis. The experiences of these women indicate
that they retained little influence over the assets and incomes created from
their loans. Before obtaining loans, some of these women worked on family
farms or within their households, but now found themselves working as
wage labourers mainly to meet repayments. Women’s wages were, in fact,
the main source of repayment in around 70 per cent of all cases. Interviews
indicate that these women felt resentful about this perceived decline in status.
Here is what some of the women said:

I was happy working in my home and on the farm [before joining the group], but now I have
to go for kulie pani [wage work] every day. Sometimes I think I should leave the group, so I
can stop this chakiri [drudgery]. . .I know I have to clear all my debts before I can talk like
this. (G3W11)
Microcredit and Women’s Empowerment in Rural India 745

I gave my [loan] money to my husband to buy a motor for the farm, but this was a mistake.
He does not discuss anything with me nowadays. . .he feels that if he tells me anything, I will
pester him for money. Instead of helping me with repayments, he asked me to go for kulie
pani [to repay the loan]. (V11W3)
My husband is actually a good man, but lack of money is an evil thing — it can change
people. He used to hand over every paisa [penny] to me. . . But after he got TB and I had to
use the SHG loan to pay his hospital bills — things have changed. He does not give me all
his money – only what I need for the house — he thinks I will use it to repay the loan. He
even goes to the zamindar to collect my wages. (G5W12)

On the other hand, women like G6W14, G7W7, V7W1 and members of
G9 who used their loans in their own businesses perceived a positive change
in their statuses. They spoke about their newly found confidence in their
ability to earn incomes independent of their husbands and without recourse
to wage labouring. They felt a change not only in the attitudes of their family
members but also the society at large.15 Here are a couple of examples:
Mahalaxmi [name of V7W1’s group, but also that of the Hindu Goddess of Wealth] has made
me what I am today. I manage not one but two businesses now. I cook meals for the aganwadi
[pre-school group] and in the afternoons I manage a tea-shop. . .. I could not imagine even
two years back that one day I will own a mobile phone — but I need it for my business.
The aganwadi teacher has to call me every morning to tell me how much food to prepare.
(V7W1)
I have changed. Everyone around me has changed. We are now business women. We have
to deal with all sorts of people in our fertilizer business. . . At first, people only saw us as
women — but now they are used to us. They know we run a good business — they respect
us for this. . . My son wants to start a business of his own. He comes to me for advice.
(G9W1)

Where women’s loans are diverted into household needs, these mainly
enhance their husbands’ asset holdings and help them spend more time
in self-employment. Not only are women unable to access these assets to
improve the value of their work time, they have little claim on the income
from these assets. Several either continue to work for wages or have taken up
wage labour mainly to repay loans. Such work is not only physically arduous
and badly paid, but it is also associated with a low social status. Women
who use their loans to start or enhance their own businesses have entirely
opposite experiences. They have improved their asset holdings, are able to
spend more time on better remunerated self-employment and are gaining
in confidence that is likely to enhance their agency and empower them.
These findings suggest that if credit interventions are aimed at forwarding
the empowerment agenda, then women retaining control over loan created
assets is a critical condition.

15. Note that very few women who managed their own enterprise reported any actual profits
after repayments. Of the two SHGs that had started joint ventures, one was profitable (a
fertilizer shop), but the other reported a small loss (a rental business that catered to special
occasions like weddings, etc.).
746 Supriya Garikipati

CONCLUDING REMARKS

Poor rural women in many developing countries are pushed into marginal
employment opportunities mainly because of poverty but also because of
patriarchy which constrains their rights to own land and other significant
household assets. Such employment is usually associated with bad work and
pay conditions and is often also deemed to be socially debasing. In such
situations, lending to women can have a beneficial impact by enabling them
to invest in productive assets and improve the value of their work time. This
may also lead to improvements in their status and empower them in other
ways. This expectation drives microcredit to the forefront of the fight against
women’s poverty and empowerment. This study provides a rare empirical
insight into whether credit improves the value of women’s work time. It uses
survey data from villages in India to examine the impact of microcredit on
male and female time use. Our findings suggest that access to credit alone is
unlikely to improve the value of women’s work time; it is the way in which
women use their credit that matters. Improving the value of women’s work
time requires that women use their credit in ways that will help enhance their
control over productive assets.
We find that women’s loans are mainly used to improve households’
productive assets and because these are typically owned by men, women’s
loans ultimately benefit them. We find that men with wives in the credit
programme spend significantly more time in self-employment and less time
in wage work when compared to others. It is only when the focus shifts
to how women use their loans (as opposed to procurement alone) that
the impact of microcredit on their time use becomes apparent. In partic-
ular, women who use their loans to start or enhance businesses that they
operate — in other words, enhance their ownership of productive assets —
are seen to spend significantly more time in self-employment.
From a policy perspective, our findings suggest that, if improving women’s
situation is a policy concern for microcredit programmes, then it is important
to explore mechanisms that influence loan usage rather than focus solely on
disbursement of credit. While changing the patriarchal notions surrounding
ownership of a household’s productive assets is likely to remain a long-term
objective, a more achievable paradigm that provides the first steps towards
such change may be to focus on assets bought with women’s loan money.
Where women’s credit is used to procure or enhance productive assets,
it is conceivable that providers can persuade households to accept greater
female control over such assets. This is likely to give them greater rights to
access such assets for work and repayments and challenge their sale. In the
longer term, this may also influence women’s bargaining positions within
the household and their status.
While these findings provide an initial step towards better understanding
of the potential of credit on time use with the available data, much fur-
ther research is required. Importantly, the crucial aspect that needs to be
Microcredit and Women’s Empowerment in Rural India 747

incorporated into time-use studies is the returns on male and female enter-
prises. If the profitability of women’s businesses is indeed lower than that of
men there is a need to reassess the obsessive targeting of women clients.

APPENDIX

Table A1. List of Sub-Castes in the Survey Villages

Scheduled Tribe Scheduled Caste Other Backward Caste Other Castes

Chenchu, Lambada, Madiga, Mala Bagham, Baliga, Chakali, Kappu, Komitee,

Yerukali Golla, Gouda, Housula, Reddy
Jangam, Kammari, Katika,
Mangali, Medari, Mudhiraj,
Muslims, Padmasali, Tamballi,
Veddera

Table A2. Determinants of Duration of Credit Programme Participation

(First-stage regressions)
DEPENDENT VARIABLE: Duration
A1–1 A1–2 A1–3
Variable (For 3–1 to 3–4) (For 3–5 to 3–8) (For 6–1)

INSTRUMENTAL VARIABLE
InstCaste 4.638 (16.20)∗∗∗ 4.399 (17.44)∗∗∗ 0.237 (20.78)∗∗∗
PROGRAMME RELATED VARIABLES
Own use – – −0.107 (−0.27)
Peer effect – – −0.046 (−6.57)∗∗∗
INDIVIDUAL CHARACTERISTICS
Age2 0.0003 (1.28) 0.0003 (1.34) 0.00006 (0.37)
Education 0.203 (1.24) 0.139 (0.43) 0.312 (0.75)
Sons −0.033 (−0.23) −0.180 (−1.71)∗ −0.305 (−0.86)
Daughters −0.318 (−2.71)∗ −0.019 (−0.15) –
H-land 0.022 (0.60) −0.040 (−0.79) 0.223 (1.97)∗∗
W-fallback 0.199 (0.66) 0.347 (1.36) −0.367 (−1.03)
HOUSEHOLD CHARACTERISTICS
F-head − 0.128 (0.39) −0.484 (−0.91)
O-Loan 0.217 (0.66) 0.404 (1.47) −0.514 (−1.27)
Dependency 0.019 (1.26) −0.030 (−2.29)∗∗ 0.003 (0.17)
Dwelling 0.436 (1.13) −0.365 (−1.07) −0.707 (−2.04)∗∗
Caste −0.106 (−0.33) 0.067 (0.27) −0.188 (−0.71)
Market 0.548 (1.99)∗ 0.394 (1.65)∗ −0.287 (−1.15)
Constant −0.410 (−0.75) 0.556 (0.89) 0.114 (0.22)
No. of observations 142 135 359
Adjusted R-squared 0.7016 0.7355 0.6038
Root MSE 1.5638 1.2605 2.2159

∗
Significant at the 10% level,∗ Significant at the 5% level,∗∗∗ Significant at the 1% level.
Notes: t-statistics are given between parentheses.
The variable F-head was dropped from the male models due to lack of variation.
748 Supriya Garikipati

REFERENCES

Ackerly, B.A. (1995) ‘Testing the Tools of Development: Credit Programmes, Loan Involvement
and Women’s Empowerment’, IDS Bulletin 26(3): 56–68.
Agarwal, B. (1994) A Field of One’s Own: Gender and Land Rights in South Asia. Cambridge:
Cambridge University Press.
Allendorf, K. (2007) ‘Do Women’s Land Rights Promote Empowerment and Child Health in
Nepal?’, World Development 35(11): 1975–88.
Badatya, K.C., B.B. Wadavi and S. Ananthi (2006) ‘Microcredit for Microenterprise: An Impact
Evaluation Study of Self Help Groups’. Evaluation Study Series No. 13. Andhra Pradesh.
Mumbai: NABARD.
Banerjee, A., E. Duflo, R. Glennerster and C. Kinnan (2010) ‘The Miracle of Microfinance?
Evidence from a Randomized Evaluation’. Cambridge, MA and Chennai: Abdul Latif
Jameel Poverty Action Lab (J-PAL), MIT, and Center for Microfinance, IFMR. http://econ-
www.mit.edu/files/4162 (accessed 1 July 2011).
Becker, G.S. (1965) ‘A Theory of Allocation of Time’, Economic Journal 75(299): 493–
517.
Berger, M. (1989) ‘Giving Women Credit: The Strengths and Limitations of Credit as a Tool
for Alleviating Poverty’, World Development 17(7): 1017–32.
von Braun, J. and P. Webb (1989) ‘The Impact of New Crop Technology on the Agricultural
Division of Labor in a West African Setting’, Economic Development and Cultural Change
37(3): 513–34.
Central Statistical Organization (1995) Men and Women in India. New Delhi: Government of
India.
Chaudhry, P. (1994) The Veiled Woman: Shifting Gender Equations in Rural Haryana, 1880–
1990. New Delhi: Oxford University Press.
da Corta, L. and D. Venkateshwarlu (1999) ‘Unfree Relations and Feminization of Agricultural
Labor in AP 1970–95’, The Journal of Peasant Studies 26(2): 71–139.
Daley-Harris, S. (2007) Report of the Microcredit Summit Campaign. Washington, DC: Micro-
credit Summit.
Deaton, A. and J. Drèze (2002) ‘Poverty and Inequality in India: A Re-examination’, Economic
and Political Weekly 37(36): 3729–48.
EDA (2005) The Maturing of Indian Microcredit: Findings and Policy Implications from a
National Study. Gurgaon: EDA Rural Systems Private Limited.
Fafchamps, M. and A. Quisumbing (1998) Social Roles, Human Capital and the Intrahousehold
Division of Labor: Evidence from Pakistan. Washington, DC: International Food Policy
Research Institute.
Garikipati, S. (2008a) ‘The Impact of Lending to Women on Household Vulnerability and
Women’s Empowerment: Evidence from India’, World Development 36(12): 2620–42.
Garikipati, S. (2008b) ‘Agricultural Wage-work, Seasonal Migration and the Widening Gender
Gap: Evidence from a Semi-arid Region of Andhra Pradesh’, The European Journal of
Development Research 20(4): 629–48.
Garikipati, S. (forthcoming) ‘Microcredit and Women’s Empowerment: Have We Been Looking
at the Wrong Indicators?’, Oxford Development Studies.
Government of Andhra Pradesh (1996) Below Poverty Line Survey for IX Five-Year Plan: 1997–
2002. Hyderabad: Government of Andhra Pradesh.
Government of India (2001) Census of India (data taken from Table 3, Statement 2). New Delhi:
Government of India.
Gronau, R. (1973) ‘The Intrafamily Allocation of Time: The Value of Housewives’ Time’,
American Economic Review 63(4): 634–51.
Harper, M. (2002) ‘Promotion of Self Help Groups under the SHG Bank Linkage Programme in
India’. Paper presented at the Seminar on SHG-Bank Linkage Programme, New Delhi (25–26
November).
Microcredit and Women’s Empowerment in Rural India 749

Heckman, J. (2008) ‘Econometric Causality’. NBER Working Paper 13934. Cambridge, MA:
National Bureau of Economic Research.
Holvoet, N. (2005) ‘The Impact of Microcredit on Decision-Making Agency: Evidence from
South India’, Development and Change 36(1): 75–102.
Horrell, S., M. Johnson and P. Mosley (eds) (2008) Economic Work, Female Empowerment and
Economic Development. London: Routledge.
Hunt, J. and N. Kasynathan (2001) ‘Pathways to Empowerment? Reflections on Microcredit
and Transformation in Gender Relations in South Asia’, Gender and Development 9(1):
42–52.
Jacoby, H. (1993) ‘Shadow Wages and Peasant Family Labour Supply: An Econometric Ap-
proach to the Peruvian Sierra’, Review of Economic Studies 60(4): 903–21.
Juster, T.F. and F.P. Stafford (1991) ‘The Allocation of Time: Empirical Findings, Behavioural
Models and Problems of Measurement’, Journal of Economic Literature 29(2): 471–522.
Kabeer, N. (2001) ‘Conflicts over Credit: Revaluating the Empowerment Potential of Loans to
Women in Rural Bangladesh’, World Development 29(1): 63–84.
Khandker, S.R. (1988) ‘Determinants of Women’s Time Allocation in Rural Bangladesh’, Eco-
nomic Development and Cultural Change 37(1): 111–26.
Kropp, E.W. and B.S. Suran (2002) ‘Linking Banks and (Financial) Self Help Groups in India
– An Assessment’. Paper presented at the Seminar on SHG-Bank Linkage Programme, New
Delhi (25–26 November).
Leach, F. and S. Sitaram (2002) ‘Microcredit and Women’s Empowerment: Lessons from India’,
Development in Practice 12(5): 575–88.
Mahajan, V. and B.G Ramola (1996) ‘Financial Services for the Rural Poor and Women in India:
Access and Sustainability’, Journal of International Development 8(2): 211–24.
Malhotra, A. and S.R. Schuler (2005) ‘Women’s Empowerment as a Variable in International
Development’, in D. Narayan (ed.) Measuring Empowerment: Cross-disciplinary Perspec-
tives, pp. 71–88. Washington, DC: The World Bank.
Morduch, J. (1999) ‘The Microcredit Promise’, Journal of Economic Literature 37(4): 1569–
614.
Mueller, E. (1984) ‘The Value and Allocation of Time in Rural Botswana’, Journal of Develop-
ment Economics 15(3): 329–60.
MYRADA (2002) ‘Impact of Self Help Groups on the Social/Empowerment Status of Women
Members in Southern India’. Paper presented at the Seminar on SHG-Bank Linkage Pro-
gramme, New Delhi (25–26 November).
NABARD (2003) ‘NABARD’s Microcredit Initiatives’. Mumbai: NABARD.
NABARD (2007) ‘Assessing a Self Help Group: Note for Guidance’. Mumbai: NABARD.
http://www.nabard.org/microcredit/assessmentmethods.asp (accessed 8 May 2010).
Newman, C. (2002) ‘Gender, Time Use and Change: The Impact of the Cut Flower Industry in
Ecuador’, The World Bank Economic Review 16(3): 375–96.
Paolisso, M.J., K. Hallman, L. Haddad and S. Regmi (2002) ‘Does Cash Crop Adoption De-
tract from Child Care Provision? Evidence from Rural Nepal’, Economic Development and
Cultural Change 50(2): 313–37.
Pitt, M.M. (2000) ‘The Effect of Nonagricultural Self-Employment Credit on Contractual Re-
lations and Employment in Agriculture: The Case of Microcredit Programs in Bangladesh’,
Bangladesh Development Studies 2/3:15–48.
Pitt, M.M., S.R. Khandker and J. Cartwright (2006) ‘Empowering Women with Micro Finance:
Evidence from Bangladesh’, Economic Development and Cultural Change 54(4): 791–831.
Planning Commission (2001) ‘Press Release’, 22 February.
Puhazhendi, V. (2000) ‘Evaluation Study of Self Help Groups in Tamil Nadu’. Mumbai:
NABARD.
Puhazhendi, V. and K.C. Badatya (2002) ‘SHG-Bank Linkage Program for Rural Poor: An
Impact Assessment’. Paper presented at the Seminar on SHG-Bank linkage Programme,
New Delhi (25–26 November).
750 Supriya Garikipati

Puhazhendi, V. and K.J.S. Satyassi (2000) ‘Microcredit for Rural People: An Impact Study’.
Mumbai: NABARD.
Rao, N. (2007) ‘Custom and the Courts: Ensuring Women’s Rights to Land, Jharkhand, India’,
Development and Change 38(2): 299–319.
Rose, E. (2000) ‘Gender Bias, Credit Constraints and Time Allocation in Rural India’, The
Economic Journal 110(465): 738–58.
Skoufias, E. (1993) ‘Labour Market Opportunities and Intrafamily Time Allocation in Rural
Households in South Asia’, Journal of Development Economics 40(2): 277–310.
Swain, R.B. and F.Y. Wallentin (2009) ‘Does Microcredit Empower Women? Evidence from
Self Help Groups in India’, International Review of Applied Economics 23(5): 541–556.
Todd, H. (2001) ‘Paths out of Poverty: The Impact of SHARE Microfin Limited in Andhra
Pradesh, India’. Unpublished report. Brighton: Imp-Act.
World Bank (2001) Engendering Development through Gender Equality in Rights, Resources
and Voice. New York: Oxford University Press.
World Bank (2003) Microcredit in India: Issues, Challenges and Policy Options. Washington,
DC: The World Bank.
World Bank (2005) ‘Scaling-up Microcredit for India’s Rural Poor’. World Bank Policy Research
Working Paper 3646. Washington, DC: The World Bank.
World Bank (2006) ‘Microcredit in South Asia: Towards Financial Inclusion for the Poor’.
Washington, DC: The World Bank.

Supriya Garikipati is a Senior Lecturer in Development Stud-

ies at the Management School, University of Liverpool, UK (e-mail:
[email protected]). Her research examines the impact of public policy
interventions in India with a focus on gender and poverty. She has worked
on India’s microcredit sector, its rural labour markets and on implications of
India’s economic liberalization. One of her current projects examines the in-
terplay of colour, caste and class in modern India, especially its implications
for the employment and marital markets.