Empowering Women with Micro Finance: Evidence from Bangladesh

Abstract

This article examines the effects of men's and women's participation in micro credit programs on various indicators of women's empowerment using data from a special survey carried out in rural Bangladesh. These credit programs are well suited to studying how gender-specific resources alter intrahousehold allocations because they induce differential participation by gender through the requirement that only one adult member per household can participate in any micro credit program. Empowerment is formalized as an unobserved latent variable reflecting common components of qualitative responses to a large set of questions pertaining to women's autonomy and decision-making power. The empirical methods are attentive to various sources of endogeneity, and the results are consistent with the view that women's participation in micro credit programs helps to increase women's empowerment. The effects of male credit on women's empowerment were generally negative.

Full text

䊚2006 by The University of Chicago. All rights reserved. 0013-0079/2006/5404-0004$10.00
Empowering Women with Micro Finance:
Evidence from Bangladesh
mark m. pitt
Brown University
shahidur r. khandker
World Bank
jennifer cartwright
Brown University
I. Introduction
In recent years, governmental and nongovernmental organizations in many
low income countries have introduced credit programs targeted to the poor.
Many of these programs specifically target women, based on the view that
they are more likely than men to be credit constrained, have restricted access
to the wage labor market, and have an inequitable share of power in household
decision making. The Grameen Bank of Bangladesh is perhaps the best-known
example of these small-scale production credit programs for the poor, and over
90% of its clients are women. Earlier work (Pitt and Khandker 1998; Pitt
et al. 1999; Pitt 2000; Pitt et al. 2003) has found that the effects of program
participation differ importantly by the gender of program participant. For
example, Pitt and Khandker (1998) find that the flow of consumption ex-
penditure increases 18 taka for every 100 taka borrowed by women, but only
11 taka for every 100 taka borrowed by men. Pitt et al. (2003), using a totally
different approach to parameter identification, find that credit provided women
importantly improves measures of health and nutrition for both boys and girls,
while credit provided men has no significant effect.
What underlies these gender differences? There are essentially two different
mechanisms that can result in different effects of credit program participation
by gender: (i) “empowerment” effects and (ii) standard income and substitution
effects. Collective models of household decision making provide one avenue
This article benefited from the comments of seminar participants at the University of the Wit-
watersrand and from the comments of referees and the editor.

792 economic development and cultural change
of understanding empowerment. In a simple version of collective decision
making, the household’s social welfare is some function of the individual utility
functions. Browning and Chiappori (1998) have shown that if behavior in the
household is Pareto efficient, the household’s objective function takes the form
of a weighted sum of individual utilities, with weights t. The weight tcan
be thought of as representing the bargaining power of the female household
member relative to the male household member in determining the intra-
household allocation of resources. When tis zero, female preferences are given
no weight and the household’s social welfare function is identically that of
the male. In much of the literature, tis presumed to be increasing in the
relative value of female time and her money income. In addition, tmay be
altered through social pressure. The parameter t, which directly reflects
women’s power in household decision making, is one index of “women’s
empowerment.”
The differing credit effects by gender of participant reported by Pitt, Khand-
ker, and associates are not sufficient to establish that credit program partici-
pation has an empowerment effect. The results cited above can, in principle,
be simply the result of standard income and substitution effects. In an economy
in which women do not work in the wage labor market, participation in a
group-based credit program increases the shadow value of female time by
providing a complementary input for the production of goods for the market
by the self-employed. In contrast, if men still provide time to the wage labor
market, the shadow value of their time is unaffected by program participation.
Consequently, the self-employment activities of women fostered by micro credit
may generate different demand effects than the self-employment activities of
men fostered by micro credit. If the preference weight tis unaffected by male
participation, such participation does not alter the shadow price of women’s
time either. The only source of change in demand when men are the credit
program participants arises from the income effect associated with the rental
value of the capital endowment provided by the credit program. Note that
although male participation identifies the income effect conditional on t,this
information does not help disentangle the substitution effect from the bar-
gaining (empowerment) effect induced by women’s participation. Thus a find-
ing that the effect of women’s program participation on child health differs
from the effect of men’s program participation (as in Pitt et al. 2003) cannot
be taken to necessarily imply that women have gained power in the household,
even if women are assumed to prefer child quality more than their husbands.
A modeling strategy that seeks to separate out the income and substitution
effects from the empowerment effect (on t) resulting from micro credit program
participation would make difficult demands on the data and require strong

Pitt, Khandker, and Cartwright 793
restrictions on the form of preferences. An alternative approach is to collect
data on attitudes by and toward women and on women’s decision-making
autonomy. These data are necessarily self-reported and subjective, but econ-
ometric techniques, notably instrumental variables estimation, are available to
correct for the possible confounding effects of systematic variation in subjective
response. Note that self-reported measures of decision-making power, even if
experimentally elicited, do not necessarily imply that women actually have
more power (as measured by t), but they do add one more piece to the
accumulated evidence pointing in that direction. Frankenberg and Thomas
(2001) make the case that it is useful to include explicit questions about
decision making within the household in household surveys, arguing that
patterns of decision making may be the outcome of relative power within the
household. Using a special module included as part of the second Indonesia
Family Life Survey (IFLS2), they demonstrate that combining qualitative and
quantitative approaches enriches our understanding of intrahousehold decision
making.
A few studies in the recent past have attempted to establish the relationship
between credit program participation and some notion of women’s empow-
erment. But all of these studies suffer from possible bias due to endogeneity
of decisions involved in program participation and the unobserved household,
individual, and area characteristics. The unobserved heterogeneity likely to
bias the estimates includes the unobserved attitudes and characteristics of
husbands, wives, and other family members, including preexisting women’s
empowerment and autonomy. It seems quite possible, for example, that more
empowered women are more likely to be able to join a micro credit program.
For example, Hashemi, Schuler, and Riley (1996) find that membership in
Grameen Bank and BRAC has a significant positive effect on empowerment
by contrasting program villages with nonprogram villages.
1
Doing so neglects
the potential for village-level unobservable characteristics to bias the results.
Our study estimates the impact of participation in micro credit programs
on an index of empowerment and its proxy indicators using a large set of
qualitative responses to questions that characterize women’s autonomy and
gender relations within the household with due attention to heterogeneity
bias. The data come from an extensive household survey collected in rural
Bangladesh in 1998–99. We test the assertion that participating in micro
credit programs is an empowering experience for women whose life choices
1
Hashemi et al. (1996) create an “index” of empowerment through a linear weighted combination
of individual empowerment indicators. They establish an arbitrary cutoff point such that women
who score above this cutoff are labeled empowered and those who score below it are labeled
unempowered.

794 economic development and cultural change
are otherwise restricted through poverty, patriarchy, and societal or religious
norms. In addition, we examine the effect of men’s credit program participation
on these same measures of female empowerment. Unlike Hashemi et al. (1996)
and others, we measure women’s empowerment as a latent variable encom-
passing a number of indicators that proxy for a woman’s autonomy, decision-
making power, and participation in household and societal decision making.
The article is organized as follows. Section II discusses the data and their
salient features useful for the measurement of empowerment. Section III for-
malizes the concept of empowerment as an unobserved latent variable reflecting
common components of qualitative responses to a set of questions pertaining
to women’s autonomy and decision-making power. Section IV presents reduced
form determinants of various latent empowerment measures; important among
the determinants is the effect of village-level exposure to micro credit by
gender. Section V estimates the effect of micro credit programs on the em-
powerment of participating households using two-stage least squares. Section
VI summarizes the findings.
II. Data
The data used in this article come from a large household survey conducted
in 1998–99, which is a follow-up survey of an earlier survey conducted in
1991–92. Both household surveys were conducted by the Bangladesh Institute
for Development Studies (BIDS) in collaboration with the World Bank. Only
the follow-up survey (conducted in 1998–99) included a special module on
women’s empowerment.
The base household survey interviewed 1,798 households randomly drawn
from 87 villages of 29 thanas in rural Bangladesh. Of these 29 thanas, 24
were program thanas (8 from each of the three programs: Grameen Bank,
BRAC, and BRDB RD-12 project), and 5 were nonprogram thanas. Three
villages in each program thana were randomly selected from a list of program
villages in which a program had been in operation for at least 3 years. Three
villages in each nonprogram thana were also randomly selected from the village
census of the government of Bangladesh. From the village census list of house-
holds, 20 households from each village were drawn using stratified random
sampling. Out of these households, 17 were target (owned land of one-half
acre or less and, hence, qualified for program participation) and 3 nontarget
(owned land of more than one-half acre and, hence, did not qualify for program
participation). To ensure that a sufficient number of program participating
households were included in the target households in program villages, par-

Pitt, Khandker, and Cartwright 795
TABLE 1
NUMBER OF HOUSEHOLDS BORROWING FROM CREDIT PROGRAMS, BY GENDER OF BORROWER
BRAC BRDB GB ASA PROSHIKA GSS
Youth
Development Other NGO
Male 16 54 121 4 9 2 0 35
Female 273 72 545 105 29 3 1 183
ticipant households were overdrawn.
2
All regression estimates are appropriately
weighted to account for this sampling design. Of the 1,798 households se-
lected, 1,538 were target and 260 were nontarget households. Among the
target households, 905 (59%) participated in a credit program.
The program villages surveyed had either male and female credit groups
or both: 40 villages had credit groups for both men and women; 22 had
female-only groups, so that males in landless households are denied the choice
of joining a credit program; and 10 had male-only groups, so that landless
females are denied program choice. All program groups are single-sex, and
not all villages have both a male and a female group. The existence of villages
with only female or only male groups is a key feature of the parameter iden-
tification method described below. In addition, these programs require that
no more than one member of a household may join.
3
A more detailed de-
scription of this survey can be found in Khandker (1998).
These households were revisited in 1998–99. The resurvey tried to include
all households from the 1991–92 survey, including splits, plus some new
households were added.
4
A sample of 2,074 households with married couples
was administered the women’s empowerment questionnaire. Table 1 shows
the distribution of households across the eight categories of program credit,
broken down by gender. Appendix table A1 lists all of the empowerment
questions, to whom each question was asked, and the mean responses. Ap-
proximately 80% of the questions were asked only of wives.
We have grouped the survey questions into the following thematic groups:
(1) purchasing: ability to spend money independently and to make household
2
An additional 58 households were selected from 15 villages of five program thanas (covering all
three programs), because a nutrition survey was additionally conducted in those villages and a
larger number of target households was required.
3
In a few households, both men and women borrowed, and some women borrowed from more
than one program.
4
After the 1991–92 survey, one or more micro credit programs moved to some control villages
of the 1991–92 survey, making them program villages. So three new thanas (with three villages
in each thana) were added. In addition, two more villages were added to previous nonprogram
thanas. In the program thanas, six new villages were added. Altogether 104 villages from 32 thanas
were included.

796 economic development and cultural change
purchases; (2) resource: general economic power and access to funds; (3) finance:
power regarding household borrowing and ability to borrow from informal
sources; (4) transaction management: balance of power relating to decision,
implementation, and spending for household; (5) mobility and networks: free-
dom of movement, development of networks, relationships with blood kin
and in-laws; (6) activism: awareness of law and politics, autonomous action
on public and private matters; (7) household attitudes: attitudes on women’s
empowerment, dowry, and status within household; (8) husband’s behavior:
husband’s actions and opinions pertaining to women’s status; (9) fertility and
parenting: decisions and action for family planning and child rearing; and
(10) all variables: general women’s empowerment encompassing all nine of
the above thematic groups.
Economic decision making consists of questions on whether women them-
selves were involved in decisions on expenditures for house repair and con-
struction, livestock sale and purchase, borrowing money, and transactions in-
volving household equipment. For all four issues, it was very rare for women
to report either that they alone decided and implemented their decision or
that they decided alone and implemented jointly with the husband. For each
issue, less than 3% of respondents answered that they decided on these issues
alone. Similarly, for all four issues, almost all the respondents (more than 98%)
said that they themselves do not spend money in such matters; rather, it is
the husband who actually handled the money in the transaction.
Purchasing capacity involves questions in seven categories of common house-
hold purchases (food, toiletries, candies for the children, cooking utensils,
furniture, children’s clothing, and own clothing) to find out if women (rather
than someone else in the household) were able to make the purchase and, if
so, whether or not they make the purchase without their husbands’ permission.
The percentage of women who answered that they make purchases themselves
varies widely by category, from less than 5% (for furniture) to more than 60%
(for candies and household utensils). When husbands were asked about their
wives’ freedom to make purchases, 87% responded that their wives are not
able to buy assets on their own without the husband’s permission. On control
over loans, a growing literature in the field of micro credit addresses the degree
to which credit is fungible within the household.
5
Of central importance is
whether or not women retain control over their loans and management power
of the activities for which the loans are used. In cases where wives had taken
small loans, from any source, 78% of husbands reported that they use their
5
See, e.g., Goetz and Sengupta (1996), Montgomery, Bhattacharya, and Hulme (1996), and Pitt
and Khandker (1998).

Pitt, Khandker, and Cartwright 797
wives’ loan money to spend on their own income-generating projects. Among
women who had taken loans for income-generating activities, only 5% reported
having total autonomous control over the money. Fifty-six percent reported
that they share control over the loan money with their husbands, and 38%
reported that their husbands have sole control over the proceeds of the loan.
Control over income and savings reflect if women have control on these
important economic outcomes. More than 60% of men reported that their
wives have no independent source of income. Over 75% of women reported
that they do not operate any income-generating activity of their own, and
78% of women reported not having independent income that they could use
at their own discretion (without consulting their husband). A sizable number
(42%) of women reported that they do have their own independent savings,
and if they did, husbands were aware of these savings 91% of the time. Wives
expressed having a low level of control over these savings, with 85% saying
that they were not able to decide autonomously how to utilize them.
Only around 15% of women reported having received money from their
parents, siblings, or other blood relatives in the past 12 months. Of these,
95% said that their husbands knew that they had received this money. Only
17% reported that they had full control over deciding the use of that money:
62% reported partial control and 21% reported having no control at all.
More than three-quarters of women (78%) reported that they had at some
point been forced to cede money to their husbands, and 56% of women replied
that their husbands had forced them not to work outside the home. Eighty-
one percent reported that they would not be able to give their own money
away at will.
When asked if they would be able to get 500 taka in the case of an
emergency, two-thirds of women predicted that they would be able to. The
primary sources from which women predicted they would borrow such emer-
gency money were from own relatives (32%), husbands (29%), and husband’s
relatives (28%). Less than 3% of women in the sample replied that they would
borrow from moneylenders.
Mobility is very much restricted in rural Bangladesh. Traditions and family-
imposed restrictions may forbid women from leaving the family compound
or may regulate when, where, and with whom they travel. Additionally, issues
of safety often prevent women from traveling alone for even short distances.
Eighty-three percent of husbands reported that their wives never went alone
to places such as the market, bank, health clinic, and so on. Of these, over
half (55%) explained that they or their sons always accompanied the wives
when going outside the home, and another 18% explained that their wives
were accompanied by neighbors or relatives. Wives responded similarly. Fifty-

798 economic development and cultural change
three percent said that when they traveled outside the village they went with
their husbands and/or sons, and 22% traveled in the company of other women.
Almost 9% of women reported that they never left the village at all. Eighty-
two percent of women said that they had never visited their parents without
their husband’s permission.
Political awareness and activism involved a few questions that were asked
to women in the sample relating to their involvement or awareness of local
politics. Only 35% of women respondents knew the name of their member
of parliament. While an impressive majority (86%) of women reported having
voted in the last election, 74% also reported that their husbands had influenced
or compelled them to vote for a certain candidate. Less than a quarter of
women reported having ever publicly protested against an incidence of wife
beating.
The questions about networking and friendships involved asking women
about the extent of networking and friendships that they could maintain
beyond their immediate family. Marriage in Bangladesh is characterized by
patrilocal residence and village exogamy—when a woman marries, she leaves
her home, family, and village and moves into the household of her new
husband, in a new village. As a result, wives—and new wives in particular—
may not have many close relationships outside the household. In this sample,
however, women generally tended to say that they did have close friendships
and relationships (possibly with their blood relatives) outside the household.
Eighty-five percent of women stated that there were people within their bari
with whom they were close enough to share their feelings, and 73% had such
friends outside the bari.
Family planning involves questions to assess whether women exert their
power in family planning practices. Women were more likely than men to
be users of birth control. Among couples of reproductive age, over 93% of
men reported that they did not use any male birth control method. Among
these men, 65% explained that the reason was that their wives used a female
birth control method, and 16% responded that they simply did not like to
use birth control. Women’s responses were similar: over 91% of women re-
ported that they had never been able to make their husbands use a male birth
control method. Of these women, 68% explained that the responsibility of
birth control was usually given to them.
On assessing attitudes, the survey included several questions for both hus-
bands and wives regarding their opinions and attitudes on gender in society.
More than two-thirds of men (68%) replied that they believe their wives to
be less intelligent than themselves. Seventy-nine percent replied that they do
not consider their wives capable of making decisions pertaining to purchase

Pitt, Khandker, and Cartwright 799
or sale of major household assets. An overwhelming majority of women (94%)
stated that they believe that their husbands are superior to them “in qualities
and education.” When asked why, 59% of these women explained that the
husband is the earning member of the household and that this makes him
superior, and 34% stated that a woman’s lot in life is to be inferior to her
husband. When asked what kind of impact women’s empowerment would
have (or was having) on society, men were fairly evenly split between positive
and negative reactions. Roughly half (47%) responded positively by claiming
that the primary impact of women’s empowerment would either be the creation
of a better society or that it would be economic improvement for the family.
The other 53% responded negatively, saying that women’s empowerment
would cause chaos in society, problems bringing up children, or a disruption
of peace within the household.
When asked to describe what they perceived to be the greatest obstacles
to achieving women’s empowerment in Bangladeshi society, 46% of men cited
lack of education as the primary obstacle, 23% cited lack of safety, and 17%
cited religious restrictions. As secondary obstacles, men also cited religious
restrictions (30%), lack of income-generating activities (22%), lack of safety
(21%), and the social structure (18%). The main obstacles cited by women
were lack of education (47%), lack of safety (21%), and religious restrictions
(16%).
Spousal arguments and abuse have been identified as a common concern
when women get an economic influence within a household. Women were
thus asked to describe the nature of arguments that tended to arise within
the household. The most commonly cited topics of arguments were children,
money, and household chores. More than a third of women reported that when
such arguments occurred they were abused in some way: 20% reported verbal
abuse, and 16% reported physical abuse. Of those who reported physical abuse,
17% said that their injuries from the abuse had been severe enough to require
medical attention.
III. Empowerment as an Unobserved Latent Variable
Unlike many other measures of human behavior studied by economists,
women’s empowerment does not readily lend itself to direct measurement.
The large number of empowerment indicators collected in the survey suggests
not only that women’s empowerment is multifaceted but also that drawing
conclusions from a large number of regressions may be problematic. Some of
the empirical research on credit and women’s empowerment has used some
variant of an index approach to address this problem. In this approach, answers
to different questions are weighted and summed to form one universal “score”

800 economic development and cultural change
that represents empowerment. For example, a “yes” answer to each of 10
questions may be coded as one and a “no” as zero; then these ones and zeros
are added to produce an index with a minimum of zero and a maximum of
10. This approach is quite arbitrary because the researcher must choose the
weights without reference to theory or data. Some studies have used only one
scale, while others construct multiple scales for various thematic groupings
of questions.
This article treats empowerment variables as measured through a set of
observed variables or indicators. The idea is that unobserved latent variables
account for the dependencies among the indicators. The number of empow-
erment latent variables is smaller than the set of empowerment indicators,
the idea being that the number of true “underlying dimensions” that describe
a condition (such as empowerment) is smaller than the number of observed
indicators. The latent variable model estimated has two parts. The first part
links the unobserved latent variables to a set of observed indicators and is
called the measurement model. The well-known factor analysis model is a
special case of a latent variable measurement model with indicators measured
on a continuous scale. After fitting a factor analysis model, latent scores (factor
scores) are easily computed and are commonly used as dependent variables in
the second part of the analysis. The standard factor analysis measurement
model is inappropriate when the indicator variables are discrete, as they are
in our empowerment survey. Instead, we use the item response theory (IRT)
approach, in which the element of analysis is the whole response pattern of
a set of binary indicators. As demonstrated below, this approach essentially
estimates a random effects binary response model such as a random effects
probit or logit. The latent factor is an estimate of the random effect (factor)
conditional on the fitted parameters and the data.
At one extreme, we could postulate that all the variables in the study are
causally determined by only one factor, which we could call “empowerment.”
Beegle, Frankenberg, and Thomas (2001), in exploring the relationship be-
tween various indicators of power and reproductive behaviors, conclude that
women’s bargaining power is not adequately summarized by a single indicator
but spans multiple aspects of a couple’s life. We follow that view and think
it sensible to expect, for example, that those questions that pertain to political
activism measure a different type of underlying condition than do those ques-
tions that ask about reproductive control. Consequently, we estimate separate
empowerment factors for the nine thematic groups described previously. In
the one-parameter item response model, the conditional densities of the re-

Pitt, Khandker, and Cartwright 801
sponses of person jto question i, given the latent empowerment variable of
person j,l
j
, depends on the linear index given by
hpb⫹l,(1)
ij i j
where h
ij
is the (linear) index, and b
i
represents a question-specific threshold
for a positive response. The l
j
is the latent empowerment (“factor”) in the
linear index. This model has been used to estimate latent ability using data
from binary (true/false) test questions and is known as the Rasch model if the
factors l
j
are estimated as parameters rather than random effects, as in this
application. Appending a nonsystematic error ␧
ij
to equation (1) such that the
random effects l
j
are the only source of stochastic covariation between the
responses of any person, and assuming normally distributed errors, this is
essentially a random effects probit model. Estimation of this model is accom-
plished by maximum likelihood using Gauss-Hermite quadrature for numer-
ical integration. After estimating the parameters, an empirical Bayes method
is used to estimate the latent variable l
j
(the random effect) for each household
observation. This estimation was carried out with the gllamm6 package of
Rabe-Hesketh, Skrondal, and Pickles (2004). The same estimation was also
carried out using standard factor analysis methods for models with continuous
indicators. In every case, the simple correlation coefficient between the factor
estimated from the item response model (probit random effects model) and
the factor estimated from standard continuous variable factor analysis was
above 0.95. The results reported below are based on the item response model
with the exception of the “all” factor, which is estimated from continuous
factor analysis over the nine (continuous) different types of latent empowerment
factors itemized above.
The selection of variables in the 10 categories of empowerment was based
on our prior belief about which variables contain similar types of information.
Out of the 101 eligible variables, only 75 were actually used in the item
response analysis (most were used only once, but some were used to create
several different factors). The other variables were not used since it was felt
that they were not directly relevant to any of the factor themes.
IV. Reduced Form Determinants of Women’s Latent Empowerment
Having reduced the set variables of 75 individual variables to 10 empowerment
factors, including an aggregate “all” factor, we can more simply examine the
partial correlations among different dimensions of empowerment, and between
them and exogenous covariates. Table 2 presents the correlation matrix among
our 10 empowerment factors. That all the elements of this matrix are positive

802
TABLE 2
CORRELATION MATRIX OF LATENT VARIABLES
Purchasing Resources
Transaction
Management
Husband’s
Behavior
Mobility and
Networks Activism Finance
Fertility and
Parenting
Household
Attitudes
All
Factors
Purchasing 1.000
Resources .362
Transaction management .330 .172
Husband’s behavior .281 .201 .104
Mobility and networks .306 .432 .229 .010
Activism .197 .193 .114 .135 .148
Finance .293 .291 .729 .035 .382 .163
Fertility and parenting .414 .236 .256 .096 .243 .182 .215
Household attitudes .334 .147 .282 .740 .133 .152 .214 .150
All factors .658 .533 .705 .510 .514 .316 .724 .468 .649 1.000

Pitt, Khandker, and Cartwright 803
is reassuring. Only two elements are not significantly different from zero at
the 0.01 level using the Bonferroni adjustment to significance levels—the
correlation between husband’s behavior and mobility, and the correlation be-
tween husband’s behavior and finance. The highest correlations are between
transaction management and finance (0.729) and husband’s behavior and
household attitudes (0.740).
In estimating the reduced form (exogenous) determinants of these measures
of empowerment, we condition on whether the household has exogenous choice
to join a credit program. The linear-in-the-variables reduced form demand
equation for empowerment that is estimated is of the form
fm yy
ypda⫹da⫹Xb⫹Zg⫹m⫹␧,(2)
ij ij f ij m ij y j y j ij
where y
ij
is the measured empowerment factor of woman iin village j,X
ij
is
a vector of household characteristics (e.g., age and education of household
head), Z
j
is a vector of village characteristics, and are 0-1 binary indicators
fm
dd
ij ij
of the ability of women and men to choose to join a credit program (program
choice), b
y
,g
y
,a
f
,anda
m
are parameters to be estimated, is an unmeasured
y
m
j
determinant of y
ij
that is fixed within a village, and is a nonsystematic
y
␧
ij
error reflecting, in part, unmeasured determinants of y
ij
that vary over house-
holds.
A woman (man) has the choice of joining a credit program ( ) if (1)
dp1
ij
a women’s (men’s) credit program operates in village jand (2) the household
owns no more than one-half acre of average quality cultivable land and is thus
exogenously eligible to participate in the program.
6
The coefficient a
f
(a
m
)
represents the average effect of having a micro credit program for their sex
in their village on the empowerment measure y
ij
for eligible women (men).
6
The validity of the assumption that landownership is exogenous is defended at length in Pitt
and Khandker (1998). There are a number of households in our sample that were program par-
ticipants and yet had more than 0.5 acres of land at the time of program entry, raising the possibility
of mistargeting and potential bias in econometric results relying on this targeting rule. It appears
that some of this excess land is either uncultivable or marginally so. Pitt (1999) demonstrates that
the value per acre of land owned by program-participating households who also own more than
0.5 acres of cultivable land at the time of joining is a small proportion of the value per acre of
the cultivable land of program participants owning less than 0.5 acres of cultivable land at the
time of joining. This suggests that program officers are using some notion of “effective” units of
cultivable land in determining eligibility rather than of the type of mistargeting that would result
in econometric bias. Pitt (1999) discusses this issue at length and demonstrates that treating the
exogenous targeting rule to be greater than 0.5 acres provides a consistent estimator for certain
types of mistargeting. He finds that application of targeting rules greater than 0.5 acres (up to
2.0 acres) actually slightly strengthens the qualitative results on the effect of credit by gender on
household consumption. This insensitivity of results to the choice of targeting rule used in esti-
mation is further demonstrated in Pitt (2000).

804 economic development and cultural change
Our sample of households includes households in villages that do not have
access to a group-based credit program. If credit program placement across
the villages of Bangladesh is attentive to the village effects m
j
, estimating the
impact of program availability on eligible individuals by comparing individ-
uals in nonprogram villages with individuals in program villages without
controlling for the selectivity of program placement will generally result in
biased estimates of this impact. Using a village fixed effects estimation tech-
nique removes the source of this bias, the village fixed effect , but prevents
y
m
j
the estimation of the parameters g
y
.
The combination of the item response model for empowerment (eq. [1])
and the determinants of women’s response model (eq. [2]) together constitute
a structural equation model (SEM) for categorical variables. SEMs consist of
(i) a measurement model that relates a set of observed indicators to unobserved
latent variables and (ii) a “structural model” that relates the latent variables
to each other and to observed variables. Skrondal and Rabe-Hesketh (2004)
review latent variable modeling and the estimation of SEMs for a variety of
response types, including categorical variables. Our model is a MIMIC (Mul-
tiple Indicator Multiple Cause) model, a special case of a SEM. MIMIC models
are SEMs in which the relationships among the latent variables are not mod-
eled, so that the structural equation consists only of regressions of the latent
variables on a set of observed covariates. Although joint estimation of the
measurement equations and the structural equations by maximum likelihood
is possible, estimation in this study consists of three stages: (1) estimate the
item response measurement model by maximum likelihood, (2) obtain em-
pirical Bayes predictions of the latent variables from this model, and (3)
estimate various structural models by OLS, fixed effects, and, in the causal
models described in a subsequent section of this article, by two-stage least
squares with fixed effects. Joint maximum likelihood estimation of the MIMIC
model is made daunting in our case by the features of the data, binary indicators
and nonrandom sampling, and the features of the model, the need for fixed
effects and instrumental variables in the structural equation.
7
Consequently,
we use the multistep procedure for estimation.
7
Note that commercial software such as LISREL ( Jo¨ reskog and So¨ rbom 1999), EQS (Bentler 1992),
and Mplus (Muthe´ n and Muthe´n 2000) use the underlying variable approach (UVA) rather than
item response theory (IRT), the method used in this study, in modeling the measurement equations.
Very briefly, in IRT the element of analysis is the whole response pattern of an observation, whereas
UVA uses only the univariate and bivariate margins. Moustaki, Jo¨ reskog, and Mavridis (2004)
compare these two approaches on the basis of two examples. LISREL is used to estimate the UVA
model, but the authors develop their own software for the IRT model, as it is not part of LISREL.
They find that the parameter estimates of the UVA and IRT models are often close but that the

Pitt, Khandker, and Cartwright 805
TABLE 3
SUMMARY STATISTICS
Variable Mean SD
Age of household member (years) 34.95 10.59
Education of the member (years) 1.85 3.07
If parents of household head own land .161 .367
If brothers of household head own land .349 .477
If sisters of household head own land .287 .453
If parents of household head’s spouse own land .331 .471
If brothers of household head’s spouse own land .338 .473
If sisters of household head’s spouse own land .284 .451
Household land asset (decimals) 86.75 167.23
Age of household head (years) 43.69 12.14
Education of household head (years) 2.82 3.79
Highest male education in household (years) 2.83 3.77
Highest female education in household (years) 4.20 4.49
Household has male choice (yes p1) .615 .487
Household has female choice (yes p1) .685 .465
Factor: purchasing .0088 1.035
Factor: resources .00006 .500
Factor: finance ⫺.00008 .175
Factor: transaction management ⫺.0015 1.409
Factor: mobility networks .0002 .178
Factor: activism .00005 .192
Factor: household attitudes ⫺.00075 .542
Factor: husband’s behavior ⫺.00097 .188
Factor: fertility and parenting .000077 .686
Factor: all ⫺3.26#10
⫺11
.910
Log(female program loans ⫹1) 4.127 4.707
Log(male program loans ⫹1) .948 2.845
Female participation (yes p1) .443 .497
Male participation (yes p1) .106 .307
Number of observations 2,064
Table 3 presents summary statistics for the 10 factors and the 13 household-
level exogenous variables used in the analysis, as well as 0-1 indicators of
female and male credit choice and program participation, and the (log) of
cumulative actual borrowing. Table 4 presents the estimated effects of having
a micro credit program by gender on the estimated latent empowerment factors
for the (functionally landless) eligible. The column labeled “OLS” presents the
effects of female and male credit programs under the assumption of exogenous
program placement and with t-ratios adjusted for clustering in villages. The
adjacent column presents village fixed effects estimates. All regressions are
appropriately weighted to account for the sampling design. The OLS regres-
sions also included a set of 11 village-level variables and a set of 13 household-
level exogenous variables. The village fixed effects regressions include the same
IRT method fits the data better, often much better, as a consequence of its use of the whole response
pattern.

806
TABLE 4
REDUCED FORM ESTIMATES OF THE DETERMINANTS OF WOMEN’S EMPOWERMENT
Factor and Female/Male Choice OLS (Clustered) Village Fixed Effects
Factor 1: purchasing:
Female choice .199
(2.16)
.224
(3.10)
Male choice .035
(.34)
⫺.018
(⫺.28)
Factor 2: resources:
Female choice .438
(4.86)
.509
(6.84)
Male choice ⫺.601
(⫺.66)
⫺.148
(⫺2.08)
Factor 3: finance:
Female choice .411
(4.69)
.366
(4.67)
Male choice ⫺.160
(⫺1.88)
⫺.171
(⫺2.35)
Factor 4: transaction management:
Female choice .471
(4.85)
.302
(3.89)
Male choice ⫺.121
(⫺1.36)
⫺.089
(⫺1.25)
Factor 5: mobility and networks:
Female choice .376
(3.80)
.433
(5.12)
Male choice ⫺.230
(⫺2.58)
⫺.275
(⫺3.54)
Factor 6: activism:
Female choice .109
(1.12)
.188
(2.20)
Male choice .052
(.62)
⫺.068
(⫺.90)
Factor 7: household attitudes:
Female choice .181
(2.01)
.114
(1.40)
Male choice ⫺.013
(⫺.15)
.030
(.36)
Factor 8: husband’s behavior:
Female choice .144
(1.53)
.122
(1.47)
Male choice ⫺.080
(⫺.94)
⫺.043
(⫺.53)
Factor 9: fertility and parenting:
Female choice .310
(3.49)
.341
(4.28)
Male choice ⫺.057
(⫺.73)
⫺.185
(⫺2.52)
Factor 10: all variables:
Female choice .512
(6.08)
.473
(6.40)
Male choice ⫺.134
(⫺1.57)
⫺.167
(⫺2.39)
Note. Numbers in parentheses are t-statistics.

Pitt, Khandker, and Cartwright 807
set of 13 household-level exogenous variables. All regressions are appropriately
weighted to account for the sampling design. The OLS and village fixed effects
estimates are not qualitatively or quantitatively different.
8
Consequently, we
will focus our discussion on the village fixed effects estimates. The presence
of a female micro credit group in a village has a positive and highly significant
( ) effect on the factor encompassing all the questions in the ques-tp6.40
tionnaire (thus a representation of the “general level of empowerment”), and
it has a significantly ( ) positive effect on eight out of the 10 factorst12.0
(latent variables); the exceptions are factors 7 (attitudes on women’s empow-
erment, dowry, and status within household) and 8 (husband’s actions and
opinions pertaining to women’s status).
9
Male credit choice significantly re-
duces the overall empowerment factor ( ), has a statistically signif-tp⫺2.39
icant and negative effect on four of the other empowerment factors, and does
not have a statistically positive effect on any of them. The estimated empow-
erment factors have been scaled to have unit variance to aid in the interpretation
of the regression coefficients. The presence of a female group-based micro
credit program in a village increases the overall empowerment measure of
eligible women by an average of 0.473 standard deviations. The distribution
of this “all variables” factor is close to normal—a skewness and kurtosis test
cannot reject normality at the 0.05 level ( , ). This
2
x(2) p5.02 pp.081
implies that, for the median woman in the sample, the presence of a female
micro credit program increases her place in the distribution of empowerment
from the 50th percentile to the 68th percentile, and the presence of a male
micro credit program reduces her place to the 43rd percentile. Among the
nine thematic empowerment latent variables, the largest percentile gains in
8
The village-specific error accounts for approximately 30% of the variance of the total regression
error in the determinants of the composite measure of empowerment (factor 10).
9
The dependent variable in this analysis is the predicted factor derived from the estimated pa-
rameters of the item response models for empowerment. This prediction generates measurement
error of the dependent variable. Classical measurement error of the dependent variable is generally
ignorable in regression analysis, as it is absorbed into the error residual. That is, no estimation
issue results if the true empowerment factor equals the estimated factor plus a woman-specific
independently and identically distributed random error independent of the independent variables.
It may not be ignorable if there is measurement error on the individual empowerment variables
that enter into the predicted empowerment factors because the transformation into the empow-
erment factor is nonlinear in the measurement error (Abrevaya and Hausman 2004). There is no
straightforward way to correct for any bias due to the nonlinear transformation of measurement
error. However, any bias due to the nonlinearity of measurement error is likely to be slight, as our
model is very nearly linear. In standard factor analysis, the estimated factor is a linear combination
of the indicator variables. Linear regressions of the estimated factor predicted by our item response
model and the indicator variables all have and half are at least 0.99. Thus, as in factor
2
R10.95
analysis with continuous variables, the estimated factors from the item response model are almost
linear in the indicator variables.

808 economic development and cultural change
the distributions of empowerment relative to the median arising from the
availability of a female credit program in one’s village are in resources (general
economic power and access to funds) and mobility and networks (freedom of
movement, development of networks, relationships with blood kin and in-
laws), with large increases in fertility and finance factors as well. In the case
of the resource empowerment latent variable, the presence of a women’s credit
group increases the median female percentile ranking by 19 percentage points
to the 69th percentile. The largest declines arising from male credit programs
are in the mobility, fertility and parenting, finance, and resource factors. In
the case of mobility, the presence of a male micro credit program reduces the
median women’s place in the distribution of mobility empowerment to the
39th percentile.
Appendix table A2 reports the remaining regression coefficients of the
reduced form determinants of women’s empowerment using village fixed ef-
fects. These coefficients are for the 13 individual and household characteristics
in these regressions included as controls in addition to the credit choice var-
iables. Only the education variables seem to have any power (conditional on
village fixed effects) in explaining variation in the various measures of em-
powerment. In the “All factors” regression, woman’s own education is the only
statistically significant determinant of their empowerment. Household land
ownership has a marginally negative effect ( ), and the highest femaletp⫺1.54
education in the household has a marginally positive effect ( ). Thetp1.56
latter variable overlaps with “own” education in that the subject woman may
also be the woman with the highest education in the household. In this case,
the effect of increasing education on empowerment is the sum of the two
coefficients. Own education also has a positive coefficient in all nine empow-
erment thematic groups, and has t-ratios greater than 2.0 for six of these.
Household land ownership decreases transaction management and mobility/
networks but seems to increase activism. Empowerment related to husband’s
behavior in increasing in the highest male education in the household, while
empowerment in the areas of activism and fertility are increasing in the
education of the household head.
V. Estimating the Treatment Effect of Program Participation on
Empowerment
Although the reduced form estimates suggest that the presence of micro credit
programs for women increase the empowerment of landless women and that
micro credit programs for men have the opposite effect, the analysis can be
extended to the estimation of the causal effect of individual-level participation
in a micro credit program on measures of empowerment. Specifically, we

Pitt, Khandker, and Cartwright 809
estimate conditional demands for a set of empowerment indicators, conditioned
on the household’s intensity of program participation as measured by the
cumulative (price-adjusted) quantity of credit borrowed since joining the pro-
gram.
10
The econometric methods used in this analysis are essentially the same
as those presented in Pitt and Khandker (1998) and, hence, only an abbreviated
version is presented. Consider the reduced form equation (3) for the level of
participation in one of the credit programs (C
ij
), where level of participation
will be taken to be the value of program credit that household iin village j
borrows,
cc
CpXb⫹m⫹␧for dp1
ij ij c j ij ij
Cp0fordp0, (3)
ij ij
where X
ij
is a vector of household characteristics (e.g., age and education of
household head), b
c
are unknown parameters, is an unmeasured determinant
c
m
j
of C
ij
that is fixed within a village, is a nonsystematic error that reflects
c
␧
ij
unmeasured determinants that vary over households, and indicatesdp1
ij
that the household is both eligible to participate in the credit program and
resides in a village with a program, that is, that they have the choice to borrow.
The conditional demand for women’s empowerment outcome y
ij
, conditional
on the level of program participation C
ij
,is
yy
ypXb⫹Cd⫹m⫹␧,(4)
ij ij y ij j ij
where b
y
and dare unknown parameters, is an unmeasured determinant
y
m
j
of y
ij
that is fixed within a village, and is a nonsystematic error reflecting,
y
␧
ij
in part, unmeasured determinants of y
ij
that vary over households. The esti-
mation issue arises as a result of the possible correlation of with and of
cy
mm
jj
with . Econometric estimation that does not take these correlations into
cy
␧␧
ij ij
account may yield biased estimates of the parameters of equation (4) due to
the endogeneity of credit program participation C
ij
. In the model set out
above, the vector of household characteristics, X
ij
, is presumed to be the same
in both equations (3) and (4).
Using a village fixed effects estimation technique removes the correlation
10
The cumulative quantity of credit ever borrowed from these micro credit groups is thus our
measure of program exposure up until the date at which empowerment is measured. Credit is just
one facet of the multidimensional “treatment“ associated with participation in any one of the group-
based lending programs. These programs are more than just lending institutions. Nevertheless,
the quantity of credit is the most obvious and well measured of the services provided. Duration
of program participation is an alternative measure of exposure to treatment, but it is highly
correlated with cumulative borrowing, and its use makes little qualitative difference in estimation.

810 economic development and cultural change
between with as a source of estimation bias.
11
The parameter of interest,
cy
mm
jj
d, the effect of participation in a credit program on the outcome y
ij
, can be
identified if the sample also includes households in villages with treatment
choice (program villages) that are excluded from making a treatment choice
by exogenous rule. That exogenous rule is the restriction that households
owning more than 0.5 acres of cultivable land are precluded from joining any
of the three credit programs.
To illustrate the identification strategy, consider a sample drawn from two
villages—village 1 does not have the program and village 2 does; and, two
types of households, landed ( ) and landless ( ). Innocuously, wexp1xp0
ij ij
assume that landed status is the only observed household-specific determinant
of some behavior y
ij
in addition to any treatment effect from the program.
The conditional demand equation is
yy
ypCd⫹xb⫹m⫹␧.(5)
ij ij ij y j ij
The exogeneity of land ownership is the assumption that , that
y
E(x,␧)p0
ij ij
is, that land ownership is uncorrelated with the unobserved household-specific
effect. The estimator of the program effect dis a variant of the difference-in-
the-difference estimator widely applied in the general program evaluation
literature. To see this, note that an estimate of dis obtained from the following
difference-in-the-difference:
12
[E(yFjp2, xp0) ⫺E(yFjp2, xp1)]by
ij ij ij ij
⫹[E(yFjp1, xp0) ⫺E(yFjp1, xp1)] (6)
ij ij ij ij
yyyy
p(rd ⫹m)⫺(b⫹m)⫹(m)⫺(b⫹m)prd,
2y21y1
where ris the proportion of landless households in village 2 that choose to
participate in the program. If landed status is a continuous measure of land-
holding, then the credit effect dis identified from variation in landholding
within the program villages ( ), and a sample of nonprogram villages isjp2
not required.
Two-stage instrumental variable estimation of a model of this type can be
accomplished by treating as identifying instruments a dummy variable for
program choice interacted with all the exogenous variables. As noted earlier,
11
In addition, the effect of any observed village characteristics that are thought to influence y
ij
,
such as prices and community infrastructure, are not identifiable.
12
However, as Pitt (1999) points out, since this is a quasi-experiment, not an actual experiment,
the direct application of (6) would most likely result in a downward biased estimate of d. The
regression approach applied here is necessary to control for differences in other observed and
unobserved variables across the four groups for whom expectations are formulated in eq. (6).

Pitt, Khandker, and Cartwright 811
a woman has program choice if her household owns less than one-half acre of
average quality cultivable land and she lives in a program village. The idea
is that all of the exogenous variables have an effect on self-selection into the
program only for the eligible—as only they have a choice of whether to
participate—but influence empowerment outcomes for both the eligible and
ineligible. Parameter identification requires that landownership (the eligibility
criterion) not discontinuously affect the treated outcomes (contraception,
health) conditional on program participation, although it may affect outcomes
in a continuous fashion. It is important to note that the variable “landown-
ership” is not an exclusion restriction in this approach. It remains one of the
independent variables in the vector X
ijt
of equation (4).
The first-stage equation of the two-stage least squares estimation is simply
the estimation of equation (3) over the subset of women who have program
choice. For women without program choice, either because they live in a
nonprogram village or live in a household with more than one-half acre of
land, program participation (C
ij
) is deterministically zero. That is, the “re-
gression equation” for those without program choice is simply withoutCp0
ij
any error term. To see where the identifying instrumental variables arise, one
can combine the “choice” and “nonchoice” subsamples of equation (3) and
write it equivalently as
c
CpXb⫹(d⫺1)Xb⫹m
ij ij c ij ij c j
cc c
⫹(d⫺1)m⫹␧⫹(d⫺1)␧.(3

)
ij j ij ij ij
The identifying instruments available to identify C
ij
in the second-stage equa-
tion (4) are thus the elements of the vector , the interactions of(d⫺1)X
ij ij
“choice,” and the exogenous variables X
ij
, plus the variables , the
c
(d⫺1)m
ij j
interactions of the village fixed and the choice variable.
13
Standard two-stage
least squares estimation provides consistent estimates of this model.
Identification of gender-specific credit is achieved by making use of the
rule that program groups are single sex and the fact that not all villages have
both a male and a female group. As noted previously, the sample includes
some households from villages with only female credit groups, so that males
in landless households are denied the choice of joining a credit program, and
13
Note that eq. (2), the reduced form estimated in Sec. II, is obtained by substituting eq. (3

)
into eq. (4) and then dropping all of the interaction terms such as except for the(d⫺1)X
ij ij
interaction of d
ij
and the intercept. In eq. (2), the linear-in-the-variables reduced form, the coefficients
on the sex-specific choice variables d
ij
,a
f
, and a
m
represent the average effect of providing a woman
or a man with the choice of joining a micro credit program on the empowerment measure y
ij
. Also,
note that in the village random effects model, the interactions are not sources of iden-
c
(d⫺1)m
ij j
tification in the estimation of eq. (4).

812 economic development and cultural change
some households from villages with only male credit groups, so that landless
females are denied program choice. For that reason, the reduced form credit
equation is disaggregated by gender, and the instruments are the interactions
of X
ij
and gender-specific indicators of the ability of women and men to choose
to join a credit program, and , respectively.
fm
dp1dp1
ij ij
Table 5 presents the regression parameters for credit by gender estimated
by OLS (with clustered standard errors), village fixed effect regression, and
two-stage least squares with fixed effects. Appendix table A3 presents the
estimates for the noncredit independent variables. All regressions are appro-
priately weighted to account for the sampling design. Surprisingly, the esti-
mated effects of credit program participation do not vary importantly by
estimation method. If microfinance organizations are more likely to site
women’s groups where eligible women are more empowered, the ordinary least
squares estimates of the women’s credit effect should be larger than the village
fixed effects estimates. Conforming to that prediction, in only two of 10 cases
are the fixed effects estimates smaller than the OLS estimates. However, the
differences are inconsequentially small in every case. Furthermore, if there is
causation running from empowerment to joining a credit group, the village
fixed effects estimates would be expected to overestimate the women’s credit
program treatment effect as compared to the fixed effects–instrumental var-
iables estimates. This is true in six of 10 instances, but once again the dif-
ferences are quite small except for factor 7 (attitudes on women’s empower-
ment, dowry, and status within household). In this case, the fixed effects
coefficient is two-thirds larger than the fixed effects–instrumental variables
coefficient. Male credit effects are generally not statistically different from zero
no matter what the estimation method is. Only in the case of factor 5 (freedom
of movement, development of networks, relationships with blood kin and in-
laws) are men’s credit effects different from zero at the .05 level for both the
fixed effects and fixed effects–instrumental variables methods.
Female credit choice has a positive and highly significant ( ) effecttp7. 7 8
on the factor encompassing all the questions in the questionnaire (thus a
representation of the “general level of empowerment”), and it has a significantly
( ) positive effect on nine out of the 10 factors, with factor 8 (husband’st12.0
actions and opinions pertaining to women’s status) being the sole unaffected
empowerment latent variable. As noted above, male participation in a micro
credit program is not found to have a statistically positive effect on our measure
of latent overall empowerment. As before, the latent empowerment factors are
scaled to have unit variance so that the coefficients on (log) credit are the
effect of a 1% increase in credit on empowerment in units of standard de-
viations. The regression results predict that a woman who has median overall

813
TABLE 5
EFFECT OF CREDIT PROGRAM PARTICIPATION ON THE EMPOWERMENT OF WOMEN
IN PARTICIPATING HOUSEHOLDS
Factor and Female/Male Credit
OLS
(Clustered)
Village
Fixed Effects FE-IV
Factor 1: purchasing:
Female credit .032
(5.77)
.034
(7.15)
.031
(4.10)
Male credit .001
(.10)
⫺.009
(⫺1.04)
⫺.004
(⫺.25)
Factor 2: resources:
Female credit .066
(12.10)
.070
(14.63)
.062
(8.00)
Male credit .010
(.82)
.000
(.02)
⫺.022
(⫺1.35)
Factor 3: finance:
Female credit .046
(7.84)
.046
(8.49)
.051
(5.73)
Male credit ⫺.006
(⫺.46)
.001
(.05)
⫺.002
(⫺.13)
Factor 4: transaction management:
Female credit .045
(7.77)
.038
(7.56)
.043
(4.75)
Male credit ⫺.0004
(⫺.04)
.008
(.82)
.011
(.61)
Factor 5: mobility and networks:
Female credit .045
(6.84)
.045
(7.51)
.051
(4.86)
Male credit ⫺.011
(⫺.84)
⫺.022
(⫺2.43)
⫺.034
(⫺2.07)
Factor 6: activism:
Female credit .021
(3.35)
.026
(4.74)
.021
(2.56)
Male credit .005
(.60)
⫺.001
(⫺.16)
⫺.010
(⫺.76)
Factor 7: household attitudes:
Female credit .031
(5.67)
.030
(5.65)
.018
(2.07)
Male credit .006
(.49)
.009
(1.02)
.018
(1.21)
Factor 8: husband’s behavior:
Female credit .016
(2.86)
.021
(3.66)
.016
(1.77)
Male credit .005
(.70)
.011
(1.01)
.011
(.83)
Factor 9: fertility and parenting:
Female credit .032
(5.87)
.036
(6.79)
.038
(4.28)
Male credit ⫺.003
(⫺.31)
⫺.016
(⫺1.66)
⫺.020
(⫺1.27)
Note. Numbers in parentheses are t-statistics.

814 economic development and cultural change
empowerment and has never participated in a micro credit program would
be in the 72nd empowerment percentile (a 22 percentile increase) if she had
borrowed at the mean (log) level of woman participants.
It is useful to compare this calculation to the reduced form result that the
median woman would be in the 68th empowerment percentile if her village
had a women’s micro credit group and she was eligible to participate. It should
be expected that participants receive greater benefits on average from partic-
ipating than eligible nonparticipants, but this 4 percentile gap may seem
small. As Pitt and Khandker (1998) point out, it is possible that credit
programs can alter the attitudes of those who do not participate in the credit
programs as well as those who do, perhaps through demonstration or spillover
effects. These externalities may be larger for the eligible (landless) nonparti-
cipants than for the ineligible, as the former are of the same economic and
social class as participants. If these village externalities exist, they are not
captured by the estimated program effects. It is important to consider that
the reduced form equations estimate the average effect of having a village in
the program on the empowerment of eligible women compared to ineligible
women. In contrast, the conditional demand equations estimate the average
effect of program participation on the empowerment of participating women
compared to nonparticipating women irrespective of their eligibility status.
If program participants gain more from micro credit programs than eligible
nonparticipants, who in turn gain more than the ineligible, the estimated
average effect of the program on the eligible will be larger than the average
effect on participants as village spillovers/externalities are unaccounted for.
As there may be interest in the effect of female and male credit program
participation, regression analysis was performed not only on the 10 factors
but also on all of the observed variables, including some that were not included
in any factor grouping. Estimation of the determinants of the binary responses
to individual empowerment questions is complicated for some variables for
which there is little or no variation within some villages. The village fixed
effect will then perfectly predict the outcomes for the village. Instead, estimated
village fixed effects from the model with the corresponding factor are included
as independent variables to correct for any heterogeneity bias resulting from
nonrandom program placement across villages. Since the results of table 5
demonstrate that the fixed effects estimates are very close to the fixed ef-
fects–instrumental variables estimate, only the former are discussed in the
context of briefly summarizing the results for each thematic empowerment
latent variable. The results for individual observed variables are presented in
appendix table A4.
Purchasing. Female credit choice positively and significantly ( )tp4.86

Pitt, Khandker, and Cartwright 815
affects the latent empowerment factor describing women’s autonomy with
purchasing. In addition, female credit significantly augments women’s ability
to purchase all seven questionnaire items in this category as well as five of
six spending decision-making questions. Among those, female credit increases
the likelihood that a husband states both that his wife could buy assets on
her own and that she could buy them without his permission.
Resources. Female credit significantly ( ) increases the latent factortp8.00
representing a woman’s access to and control over economic resources. It also
significantly affects several individual indicators, including the likelihood that
a man says his wife has her own income, the likelihood that a wife reported
having her own income, and the likelihood of her reporting having her own
savings (it did not, incidentally, affect the likelihood that a woman had savings
that she herself could control). In addition, female credit increases the like-
lihood that a woman responds that she would be able to raise emergency funds
from any source and that she would be able to raise them specifically from
(1) selling off assets, (2) getting money from her husband, and (3) borrowing
from other people. Female credit decreases the chances that a household reports
that it fights about money.
Finance. Women’s credit significantly increases ( ) the latent em-tp5.73
powerment factor associated with finance. Women’s micro credit participation
significantly increases their role in deciding and implementing household
borrowing and in deciding how those borrowed funds are expended. Husband’s
micro credit participation increases the probability that spouses will fight over
household loans.
Transaction management. Female credit significantly ( ) increasestp4.75
the factor representing a woman’s power to oversee and conduct major house-
hold economic transactions. The individual variables describe decision making
and implementation arrangements (ranging from full power in the wife’s hands
to full power in the husband’s) and the likelihood that a wife spends money,
for four major categories: housing repair, livestock purchase, household loans,
and land/equipment transactions. In all four categories, female credit affects
women’s autonomy regarding decision making and project implementation.
The same is true for the likelihood that a woman spends money in every
category except land/equipment transactions. Male credit had a negative effect
on wives implementing housing repair projects, livestock purchase projects,
and land/equipment purchase or sale projects.
Mobility and networks. Female credit significantly ( ) affects thetp4.86
factor representing mobility/networking and also affects several individual
measures of mobility, including the odds that a husband will report that his
wife travels alone outside the house, that a woman reports traveling outside

816 economic development and cultural change
the house at all, and that she reports traveling outside alone. It also has an
effect in reducing the odds that a household will argue about the wife trav-
eling outside. Male credit reduces the mobility and network latent factor
( ), the level of a wife’s physical mobility, and the likelihood thattp⫺2.07
she ever travels outside the house (even if accompanied).
Activism. Female credit positively affects the factor relating to women’s
awareness and activism ( ). Female credit positively affects the oddstp2.56
that a woman will be informed of (meaning able to list) the ways in which
kabinnama (a premarital bridal contract) can be used to help a woman in the
event of divorce. Female credit also affects the probability that a woman knows
the name of the member of parliament in her area, the probability that she
voted in the last election, and the probability that she claims to have voted
independently (rather than under advice/pressure from her husband). Male
credit reduces the probability that his wife claims to have voted independently.
Attitudes and husband’s behavior. Female credit significantly ( )tp2.07
increases the factor relating to household attitudes but not the factor relating
to husbands’ opinions and actions ( ). This lack of effect on husband’stp1.77
opinions and actions mirrors the reduced form results that found this to be
the only factor not significantly affected by female eligibility to participate
in villages with a female micro credit group. Female credit affects the likelihood
that a man will describe his wife as intelligent and the probability that a
woman will say that she does not view her husband as superior to herself. In
the questionnaire, men were given the chance to cite positive and/or negative
impacts of women’s empowerment. Female credit increases the odds that a
man listed a positive impact of women’s empowerment and decreases the odds
that he listed a negative impact. Specifically, female credit affects the odds
that a man would cite the creation of a “better society” and “economic im-
provements for the family” as results from women’s empowerment. Male credit
has a negative effect on the odds that a husband would say that his wife is
as smart as he is.
Family planning and parenting issues. Female credit significantly (tp
) increases the fertility and parenting latent factor. Women were asked4.28
whether they initiated discussion on a range of family planning and parenting
issues and whether their husbands initiated discussion (wife initiation and
husband initiation were not mutually exclusive: answers could be one, the
other, both, or neither). Female credit increases the likelihood that a woman
initiates discussions with her husband about birth control use, birth control
methods, and birth numbers. In addition, female credit increases the likelihood
both that husbands will initiate discussion and that wives will initiate the
same discussion for issues of birth control use and children’s education (im-

Pitt, Khandker, and Cartwright 817
plying a positive effect on the total likelihood of spousal communication on
these two issues). Male credit has a negative effect on both the odds that a
wife initiated discussion regarding birth control use with her husband and on
the odds that she initiated discussion about birth control methods.
VI. Summary and Conclusion
This article examines the effects of men’s and women’s participation in group-
based micro credit programs on various indicators of women’s empowerment
using data from a special survey carried out in rural Bangladesh in 1998–99.
Credit programs are well suited to studying how gender-specific resources alter
intrahousehold allocations because they induce differential participation by
gender through the requirement that only one adult member per household
can participate in any micro credit program. The results are consistent with
the view that women’s participation in micro credit programs helps to increase
women’s empowerment. Credit programs lead to women taking a greater role
in household decision making, having greater access to financial and economic
resources, having greater social networks, having greater bargaining power
vis-a`-vis their husbands, and having greater freedom of mobility. They also
tend to increase spousal communication in general about family planning and
parenting concerns. The effects of male credit on women’s empowerment were
generally negative. The presence of male micro credit programs had a negative
effect on an overall measure of empowerment for eligible households and
specifically on women’s control of resources, finance, freedom of movement
and development of networks, and on fertility and parenting decisions.
What is the evidence that credit programs affect household welfare and
that women’s empowerment is an important causal pathway for any welfare
gains? Pitt and Khandker (1998) provide separate estimates of the influence
of borrowing by both men and women on household expenditure, nonland
assets held by women, male and female labor supply, and boys’ and girls’
schooling. They find that credit provided to women was more likely to in-
fluence these behaviors than credit provided to men. Credit provided women
was found to significantly affect all six of the behaviors studied. Credit provided
men did so in only one of six cases. Pitt et al. (2003) find that women’s credit
has a large and statistically significant impact on two of three measures of the
health of both boy and girl children. Credit provided men has no statistically
significant impact, and the null hypothesis of equal credit effects by gender
of participant is rejected. These results are consistent with an empirical lit-
erature that suggests that a mother’s relative control over resources importantly
alters the human capital of her children, specifically, that children seem to be
better off when their mothers control relatively more of their family’s resources.

818 economic development and cultural change
In contrast, the results of Pitt et al. (1999) find that women’s participation
in micro credit programs has a positive (although not always significant) effect
on fertility. However, this finding is not necessarily at odds with the finding
in this study that micro credit increases the fertility empowerment latent
variable. Income and substitution effect and empowerment effects are not
mutually exclusive. The theoretical model in Pitt et al. (1999) suggests that
the increase in the shadow price of a child arising from increased labor market
opportunities may be quite small in the case of self-employment where there
is jointness in the production of child good and the self-employment activity.
This small substitution effect may be swamped by an income effect arising
from the increased value of the women’s time endowment.
In summary, the finding that the effect of women’s program participation
on outcomes such as child health differs from the effect of men’s program
participation cannot be taken to necessarily imply that women have gained
power in the household. This result can, in principle, reflect standard income
and substitution effects. However, our analysis of the relationship of subjec-
tively measured empowerment to micro credit adds another piece of evidence
suggesting that program-induced changes in women’s empowerment may be
a powerful mechanism underlying the differential welfare impacts by gender
of participant.

819
Appendix
TABLE A1
LEGEND FOR FULL TEXT AND CODING OF INDIVIDUAL EMPOWERMENT VARIABLES
Name of Variable Full Text from Questionnaire Coding*
Asked
Of: Proportion
Thematic
Group
†
Food purchase Do you buy the family’s daily consumable food items? Yes p1, No p0 Wife 0 p.839
1p.161
1
Cosmetics purchase Do you buy toiletries and cosmetics for your own use? Yes p1, No p0 Wife 0 p.709
1p.291
1
Candy purchase Do you buy ice creams, candies, or cookies for your
children?
Yes p1, No p0 Wife 0 p.386
1p.614
1, 9
Utensils purchase Do you buy utensils, pots, and pans for the household? Yes p1, No p0 Wife 0 p.362
1p.638
1
Furniture purchase Do you buy household furniture? Yes p1, No p0 Wife 0 p.96 0
1p.040
1
Children’s clothing purchase Do you buy clothing for your children? Yes p1, No p0 Wife 0 p.837
1p.163
1, 9
Own clothing purchase Do you buy clothing for yourself? Yes p1, No p0 Wife 0 p.806
1p.194
1
Wife initiates discussion (birth control methods) Do you initiate discussion of birth control methods? Yes p1, No p0 Wife 0 p.287
1p.713
9
Wife initiates discussion (birth control use) Do you initiate discussion of birth control use? Yes p1, No p0 Wife 0 p.270
1p.730
9
Wife initiates discussion (children’s education) Do you initiate discussion of children’s education? Yes p1, No p0 Wife 0 p.172
1p.828
9
Wife initiates discussion (birth timing) Do you initiate discussion of birth timing? Yes p1, No p0 Wife 0 p.300
1p.700
9
Husband initiates discussion (birth timing) Does your husband initiate discussion of birth timing? Yes p1, No p0 Wife 0 p.368
1p.632
9
Wife initiates discussion (birth numbers) Do you initiate discussion of birth numbers? Yes p1, No p0 Wife 0 p.272
1p.728
9
House repair decision Who decides issues of repair/construction of the house? Husband alone p0,
Husband and wife together p1,
Wife alone p2
Wife 0 p.231
1p.753
2p.016
4
House repair implementation Who implements issues of repair/construction of the
house?
Husband alone p0,
Husband and wife together p1,
Wife alone p2
Wife 0 p.227
1p.769
2p.004
4
House repair spending Do you spend on repair/construction of the house? Yes p1, No p0 Wife 0 p.990
1p.010
1, 4
Livestock purchase decision Who decides issues of sale/purchase of livestock? Husband alone p0,
Husband and wife together p1,
Wife alone p2
Wife 0 p.240
1p.747
2p.013
4

820
TABLE A1 (Continued)
Name of Variable Full Text from Questionnaire Coding*
Asked
Of: Proportion
Thematic
Group
†
Livestock purchase implementation Who implements issues of sale/purchase of livestock? Husband alone p0,
Husband and wife together p1,
Wife alone p2
Wife 0 p.383
1p.611
2p.006
4
Livestock spending Do you spend on sale/purchase of livestock? Yes p1, No p0 Wife 0 p.985
1p.015
1, 4
Household loans decision Who decides issues of borrowing money? Husband alone p0,
Husband and wife together p1,
Wife alone p2
Wife 0 p.278
1p.702
2p.020
3, 4
Household loans implementation Who implements issues of borrowing money? Husband alone p0,
Husband and wife together p1,
Wife alone p2
Wife 0 p.376
1p.613
2p.011
3, 4
Household loans spending Do you spend on issues of borrowing money? Yes p1, No p0 Wife 0 p.973
1p.027
1, 3, 4
Land/equipment decision Who decides issues of sale/purchase/mortgage of land/
transport or household equipment/irrigation
equipment?
Husband alone p0,
Husband and wife together p1,
Wife alone p2
Wife 0 p.308
1p.678
2p.014
4
Land/equipment implementation Who implements issues of sale/purchase/mortgage of
land/transport or household equipment/irrigation
equipment?
Husband alone p0,
Husband and wife together p1,
Wife alone p2
Wife 0 p.466
1p.529
2p.005
4
Land/equipment spending Do you spend on issues of sale/purchase/mortgage of
land/transport or household equipment/irrigation
equipment?
Yes p1, No p0 Wife 0 p.989
1p.011
1, 4
Husband says wife is intelligent Do you think that your wife is as intelligent as you are? Less p0,
Same p1,
More p2
Husband 0 p.688
1p.241
2p.071
7, 8
Wife can buy an asset Do you think your wife can take decisions in selling/buy-
ing of major household assets?
Yes p1, No p0 Husband 0 p.807
1p.193
1, 7
Wife can buy an asset (without husband’s
permission)
Can your wife buy any asset on her own without your
permission?
Yes p1, No p0 Husband 0 p.880
1p.120
1
Wife has own income Does your wife have her own income? Yes p1, No p0 Husband 0 p.641
1p.359
2
Husband says wife travels alone Does your wife go to market/bank/doctor’s chambers,
and so on alone? If not …
Yes p1, No p0 Husband 0 p.842
1p.158
5
Reason: women not allowed outside … Why? Because women are not allowed to go outside? Ye s p0, No p1 Husband 0 p.867
1p.133
5
Reason: lack of safety … Why? Because of lack of safety? Yes p1, No p0 Husband 0 p.884
1p.116
5
Reason: wife goes with husband/son … Why? Because she goes with husband or son? Yes p1, No p0 Husband 0 p.430
1p.571
5

821
Reason: wife goes with neighbor … Why? Because she goes with a neighbor or relative? Yes p1, No p0 Husband 0 p.845
1p.155
5
Wife has independent income Do you have your own income, which you can spend
without your husband’s permission?
Yes p1, No p0 Wife 0 p.787
1p.213
2
Wife has independent savings Do you have your own savings? Yes p1, No p0 Wife 0 p.6 45
1p.355
2
Wife has independent savings which she herself
controls
Do you have your own savings which you can decide
how to utilize?
Yes p1, No p0 Wife 0 p.822
1p.178
2
Emergency funds access If you needed 500 taka in an emergency, could you get
it (from any source)?
Yes p1, No p0 Wife 0 p.342
1p.658
2
Emergency funds access (asset sale) If you needed 500 taka in an emergency, could you get
it by selling own assets?
Yes p1, No p0 Wife 0 p.977
1p.023
2
Emergency funds access (from husband) If you needed 500 taka in an emergency, could you get
it from your husband?
Yes p1, No p0 Wife 0 p.798
1p.202
8
Emergency funds access (husband’s relatives) If you needed 500 taka in an emergency, could you get
it by borrowing from your husband’s relatives?
Yes p1, No p0 Wife 0 p.830
1p.171
3, 5
Emergency funds access (own relatives) If you needed 500 taka in an emergency, could you get
it by borrowing from your own relatives?
Yes p1, No p0 Wife 0 p.788
1p.212
3, 5
Emergency funds access (moneylenders) If you needed 500 taka in an emergency, could you get
it by borrowing from moneylenders?
Yes p1, No p0 Wife 0 p.984
1p.016
3
Emergency funds access (other people) If you needed 500 taka in an emergency, could you get
it by borrowing from other people?
Yes p1, No p0 Wife 0 p.966
1p.034
3, 5
Remittance Have you received money from parents/brothers/sisters
or other relatives outside the household in the last 12
months?
Yes p1, No p0 Wife 0 p.841
1p.159
2,5
Wife can decide how to use remittance Can you decide yourself how to use that remittance? No p0,
Partially p1,
Yes p2
Wife 0 p.182
1p.634
2p.184
2
Money seizure by husband Has your husband ever compelled you to give him
money/asset if you don’t want to?
Yes p0, No p1 Wife 0 p.777
1p.223
2,8
Freedom to remit Can you give away your money/asset at will to
somebody?
Yes p1, No p0 Wife 0 p.805
1p.195
2
Husband forbids work outside home Has your husband ever forced you not to work outside
home even if you wanted to?
Yes p0, No p1 Wife 0 p.539
1p.461
8
Visits relatives (without husband’s permission) Have you ever visited your parents or other relatives
without your husband’s permission?
Yes p1, No p0 Wife 0 p.819
1p.181
5
Marriage has kabinnama Does your marriage have any kabinnama (prenuptial
bride price agreement)?
Yes p1, No p0 Wife 0 p.294
1p.706
6
Awareness of kabinnama Can kabinnama help a woman in the event of a divorce? Yes p1, No p0 Wife 0 p.056
1p.944
6
Awareness of inheritance laws Can a widow establish her legal claim over her dead
husband’s property?
Yes p1, No p0 Wife 0 p.071
1p.929
6
Has prevented husband remarrying Have you ever been successful in stopping your husband
from remarrying?
Yes p1, No p0 Wife 0 p.965
1p.035
6
Voted (at all) Did you vote in the last election? Yes p1, No p0 Wife 0 p.153
1p.845
6

822
TABLE A1 (Continued)
Name of Variable Full Text from Questionnaire Coding*
Asked
Of: Proportion
Thematic
Group
†
Voted independently Did you vote in the last election without your husband
telling you who to vote for?
Yes p1, No p0 Wife 0 p.785
1p.215
6
Protested against domestic abuse Did you ever protest against any incidents of wife
beating?
Yes p1, No p0 Wife 0 p.761
1p.239
6
Thinks dowry is good Do you think dowry is good? Yes p0, No p1 Wife 0 p.193
1p.807
6, 7
Protested against corruption Did you ever protest against any favoritism by a chair-
man or a member who distributes government relief?
Yes p1, No p0 Wife 0 p.972
1p.028
6
Confidant within bari With anybody outside your immediate family/household,
but within your bari, are you close enough to share
your feelings?
Yes p1, No p0 Wife 0 p.151
1p.849
5
Confidant outside bari With anybody outside your bari, are you close enough to
share your feelings?
Yes p1, No p0 Wife 0 p.293
1p.708
5
Severity of spousal arguments When you and your husband argue, how bad does the
argument get?
Physical abuse p0,
Verbal abuse p1,
Loud arguments p2,
Mild arguments p3
Wife 0 p.144
1p.197
2p.268
3p.391
8
Own relatives in same village Do your parents or any sibling live in the same village as
you do with your husband?
Yes p1, No p0 Wife 0 p.825
1p.175
5
Wife thinks husband is superior Is your husband superior to you in qualities and
education?
Yes p0, No p1 Wife 0 p.946
1p.054
7
Husband uses male birth control Do you yourself use any male birth control method? Yes p1, No p0 Husband 0 p.933
1p.067
9
Husband says women’s empowerment leads to bet-
ter society
Does women’s empowerment lead to a better society? Yes p1, No p0 Husband 0 p.571
1p.429
7
Husband says women’s empowerment leads to
chaos in society
Does women’s empowerment lead to chaos in society? Ye s p0, No p1 Husband 0 p.456
1p.544
7
Husband says women’s empowerment leads to
problems with kids
Does women’s empowerment lead to problems bringing
up the children?
Yes p0, No p1 Husband 0 p.208
1p.792
7
Husband says women’s empowerment leads to loss
of peace
Does women’s empowerment lead to loss of family
peace?
Yes p0, No p1 Husband 0 p.275
1p.725
7
Husband says women’s empowerment leads to the
family being better off economically
Does women’s empowerment lead to the family being
better off economically?
Yes p1, No p0 Husband 0 p.678
1p.323
7
Husband cites positive impact of women’s
empowerment
Does women’s empowerment have a good impact? Yes p1, No p0 Husband 0 p.507
1p.493
8
Husband cites negative impact of women’s
empowerment
Does women’s empowerment have a bad impact? Yes p0, No p1 Husband 0 p.586
1p.413
8
Wife has made husband use birth control Have you ever succeeded in making your husband adopt
a male birth control method?
Yes p1, No p0 Wife 0 p.903
1p.097
9

823
Wife has income-generating activity Do you have any income-generating activity? Yes p1, No p0 Wife 0 p.696
1p.304
2
Degree of mobility How do you go to banks, markets, health centers, or
places outside the village (except for your parents’
place)?
Doesn’t go at all p0,
Goes with husband or son p1,
Goes with women p2,
Goes alone p3
Wife 0 p.111
1p.549
2p.217
3p.123
5
Prevent remarriage (local government) How can a wife prevent her husband from remarrying …
by pressing charges in the local administration?
Yes p1, No p0 Wife 0 p.692
1p.308
6
Prevent remarriage (parishad) How can a wife prevent her husband from remarrying …
by pressing charges in the Union Parishad?
Yes p1, No p0 Wife 0 p.791
1p.209
6
Wife views social structure as obstacle Is the social structure an obstacle to women’s
empowerment?
Yes p1, No p0 Wife 0 p.639
1p.361
6
Wife views laws as obstacle Is inheritance law an obstacle to women’s
empowerment?
Yes p1, No p0 Wife 0 p.919
1p.081
6
Wife views religion as obstacle Is religion an obstacle to women’s empowerment? Yes p1, No p0 Wife 0 p.508
1p.492
6
*Most variables are coded with Yes p1, No p0. Variables coded differently are shaded.
†
Thematic group is coded as follows: 1 ppurchasing, 2 presources, 3 pfinance, 4 ptransaction management, 5 pmobility and networks, 6 pactivism, 7 phousehold attitudes, 8 phusband’s
behavior, 9 pfertility and parenting. All variables enter the “general empowerment” group 10 (see text).

824
TABLE A2
REDUCED FORM ESTIMATES OF THE DETERMINANT OF WOMEN’S EMPOWERMENT: NONCREDIT INDEPENDENT VARIABLES
Regressors Purchasing Resources Finance
Transaction
Management
Mobility and
Networks
Age of household member (years) .007
(1.26)
⫺.004
(⫺.57)
.006
(.63)
.001
(.10)
.006
(.45)
Education of the member (years) .029
(2.36)
.018
(1.34)
.040
(2.51)
.036
(2.75)
.045
(2.85)
If parents of household head own land ⫺.014
(⫺.25)
.036
(.52)
.006
(.06)
.071
(1.07)
⫺.028
(⫺.38)
If brothers of household head own land .032
(.62)
.096
(1.78)
⫺.011
(⫺.16)
⫺.048
(⫺.86)
.073
(1.13)
If sisters of household head own land .022
(.40)
.018
(.32)
.069
(1.13)
.021
(.34)
⫺.011
(⫺.12)
If parents of household head’s spouse own land .013
(.28)
.048
(.92)
.046
(.86)
⫺.022
(⫺.43)
.118
(1.99)
If brothers of household head’s spouse own land .014
(.28)
.068
(1.31)
⫺.017
(⫺.25)
⫺.003
(⫺.06)
⫺.079
(⫺1.30)
If sisters of household head’s spouse own land .097
(1.63)
.080
(1.32)
⫺.109
(⫺1.61)
⫺.077
(⫺1.25)
.017
(.23)
Household land asset (decimals) ⫺.022
(⫺1.39)
⫺.004
(⫺.29)
⫺.017
(⫺.94)
⫺.062
(⫺3.49)
⫺.039
(⫺2.15)
Age of household head (years) ⫺.003
(⫺.63)
.004
(.75)
⫺.006
(⫺.97)
.001
(.13)
⫺.006
(⫺.73)
Education of household head (years) .020
(2.03)
⫺.006
(⫺.57)
⫺.006
(⫺.31)
.001
(.12)
⫺.022
(⫺1.71)
Highest male education in household (years) .004
(.38)
.004
(.39)
⫺.017
(⫺1.67)
⫺.008
(⫺.78)
⫺.011
(⫺.67)
Highest female education in household (years) ⫺.011
(⫺1.26)
.024
(2.60)
.011
(1.08)
.009
(1.00)
.011
(.84)
R
2
.438 .47 .373 .456 .313
Number of observations 2,064 2,064 2,064 2,061 2,064

825
Activism
Household
Attitudes
Husband’s
Behavior
Fertility and
Parenting All Factors
Age of household member (years) .016
(2.69)
⫺.004
(⫺.56)
⫺.001
(⫺.09)
⫺.009
(⫺1.40)
.001
(.17)
Education of the member (years) .042
(2.82)
.020
(1.41)
.005
(.50)
.042
(3.18)
.048
(3.84)
If parents of household head own land ⫺.042
(⫺.62)
⫺.004
(⫺.04)
⫺.027
(⫺.36)
.098
(1.48)
.019
(.30)
If brothers of household head own land .031
(.58)
⫺.024
(⫺.41)
.005
(.06)
.001
(.03)
.014
(.26)
If sisters of household head own land .036
(.58)
.044
(.71)
.027
(.41)
⫺.004
(⫺.07)
.051
(.84)
If parents of household head’s spouse own land .047
(.92)
⫺.076
(⫺1.31)
⫺.048
(⫺.87)
⫺.003
(⫺.06)
.009
(.17)
If brothers of household head’s spouse own land .021
(.32)
.011
(.18)
.074
(1.26)
⫺.042
(⫺.77)
.009
(.16)
If sisters of household head’s spouse own land ⫺.002
(⫺.02)
⫺.096
(⫺1.61)
⫺.138
(⫺2.29)
.087
(1.50)
⫺.054
(⫺.90)
Household land asset (decimals) .031
(1.74)
.004
(.16)
.032
(1.87)
⫺.023
(⫺1.37)
⫺.024
(⫺1.54)
Age of household head (years) ⫺.010
(⫺2.12)
.002
(.28)
⫺.003
(⫺.55)
⫺.013
(⫺2.65)
⫺.004
(⫺.80)
Education of household head (years) .031
(2.58)
.018
(1.71)
.011
(1.16)
.026
(2.28)
.011
(1.07)
Highest male education in household (years) .005
(.60)
.004
(.36)
.021
(2.12)
⫺.004
(⫺.38)
⫺.002
(⫺.23)
Highest female education in household (years) ⫺.005
(⫺.71)
.011
(1.31)
.011
(.99)
.010
(1.09)
.013
(1.56)
R
2
.445 .354 .363 .367 .429
Number of observations 2,064 2,064 2,064 2,064 2,061
Note. Numbers in parentheses are t-statistics.

826
TABLE A3
EFFECT OF CREDIT PROGRAM PARTICIPATION ON THE EMPOWERMENT OF WOMEN IN PARTICIPATING HOUSEHOLDS: NONCREDIT INDEPENDENT VARIABLES
Regressors Purchasing Resources Finance
Transaction
Management
Mobility and
Networks
Age of household member (years) .007
(1.27)
⫺.002
(⫺.42)
.006
(.59)
.000
(.05)
.006
(.43)
Education of the member (years) .030
(2.37)
.018
(1.25)
.034
(2.51)
.037
(2.76)
.045
(2.83)
If parents of household head own land ⫺.016
(⫺.30)
.038
(.54)
.006
(.11)
.073
(1.10)
⫺.022
(⫺.31)
If brothers of household head own land .031
(.61)
.094
(1.74)
⫺.006
(⫺.14)
⫺.047
(⫺.84)
.079
(1.16)
If sisters of household head own land .022
(.40)
.018
(.30)
.069
(1.13)
.021
(.35)
⫺.006
(⫺.12)
If parents of household head’s spouse own land .014
(.31)
.050
(.96)
.046
(.81)
⫺.024
(⫺.48)
.112
(1.94)
If brothers of household head’s spouse own land .012
(.24)
.076
(1.43)
⫺.011
(⫺.22)
⫺.004
(⫺.06)
⫺.079
(⫺1.23)
If sisters of household head’s spouse own land .096
(1.61)
.076
(1.24)
⫺.109
(⫺1.57)
⫺.075
(⫺1.20)
.017
(.25)
Household land asset (decimals) ⫺.022
(⫺1.45)
⫺.006
(⫺.35)
⫺.017
(⫺.86)
⫺.062
(⫺3.47)
⫺.039
(⫺2.07)
Age of household head (years) ⫺.003
(⫺.61)
.002
(.60)
⫺.006
(⫺.98)
.001
(.16)
⫺.006
(⫺.75)
Education of household head (years) .020
(2.00)
⫺.006
(⫺.51)
⫺.006
(⫺.28)
.001
(.13)
⫺.022
(⫺1.67)
Highest male education in household (years) .004
(.37)
.004
(.30)
⫺.017
(⫺1.62)
⫺.008
(⫺.74)
⫺.006
(⫺.65)
Highest female education in household (years) ⫺.011
(⫺1.28)
.024
(2.52)
.011
(1.12)
.009
(1.04)
.011
(.87)
R
2
.438 .467 .372 .456 .313
Number of observations 2,064 2,064 2,064 2,061 2,064

827
Activism
Household
Attitudes
Husband’s
Behavior
Fertility and
Parenting All Factors
Age of household member (years) .016
(2.71)
⫺.002
(⫺.47)
⫺.001
(⫺.07)
⫺.009
(⫺1.40)
.001
(.19)
Education of the member (years) .042
(2.79)
.020
(1.44)
.005
(.51)
.042
(3.17)
.048
(3.82)
If parents of household head own land ⫺.042
(⫺.60)
⫺.013
(⫺.17)
⫺.027
(⫺.39)
.101
(1.51)
.019
(.30)
If brothers of household head own land .031
(.58)
⫺.028
(⫺.46)
.003
(.04)
.003
(.04)
.013
(.25)
If sisters of household head own land .036
(.57)
.044
(.70)
.027
(.41)
⫺.004
(⫺.06)
.051
(.84)
If parents of household head’s spouse own land .047
(.92)
⫺.070
(⫺1.21)
⫺.048
(⫺.84)
⫺.004
(⫺.09)
.010
(.18)
If brothers of household head’s spouse own land .021
(.37)
.006
(.09)
.069
(1.22)
⫺.041
(⫺.74)
.009
(.18)
If sisters of household head’s spouse own land ⫺.002
(⫺.04)
⫺.098
(⫺1.67)
⫺.138
(⫺2.31)
.089
(1.51)
⫺.055
(⫺.91)
Household land asset (decimals) .031
(⫺1.74)
.0004
(⫺.02)
.032
(1.80)
⫺.023
(⫺1.33)
⫺.025
(⫺1.56)
Age of household head (years) ⫺.010
(⫺2.17)
.001
(.29)
⫺.003
(⫺.55)
⫺.013
(⫺2.66)
⫺.004
(⫺.81)
Education of household head (years) .031
(2.60)
.018
(1.64)
.011
(1.14)
.026
(2.31)
.011
(1.07)
Highest male education in household (years) .005
(.58)
.004
(.29)
.021
(2.08)
⫺.004
(⫺.36)
⫺.002
(⫺.25)
Highest female education in household (years) ⫺.005
(⫺.71)
.011
(1.21)
.011
(.97)
.010
(1.11)
.013
(1.55)
R
2
.445 .351 .363 .367 .429
Number of observations 2,064 2,064 2,064 2,064 2,061
Note. Numbers in parentheses are t-statistics.

828
TABLE A4
CREDIT EFFECTS ON INDIVIDUAL EMPOWERMENT QUESTIONS (VILLAGE FIXED EFFECTS ESTIMATES)
Name of Variable
Female Credit
Coefficient
(t-Statistic)
Male Credit
Coefficient
(t-Statistic)
Food purchase .04776041 (2.2032317) .00141589 (.11540183)
Cosmetics purchase .05860924 (3.0890573) ⫺.00085887 (⫺.07842965)
Candy purchase .06361531 (3.3796502) .00238608 (.21357556)
Utensils purchase .06462877 (3.3973055) ⫺.01015248 (⫺.91474276)
Furniture purchase .11399304 (3.3408784) ⫺.04055258 (⫺1.7362839)
Children’s clothing purchase .06593455 (3.0208841) ⫺.01710887 (⫺1.3094382)
Own clothing purchase .06782688 (3.2403093) ⫺.0324938 (⫺2.541223)
Wife initiates discussion (birth control methods) .0912055 (4.3732557) ⫺.03085828 (⫺2.4979816)
Husband initiates discussion (birth control methods) .02865999 (1.5042031) .01073733 (.92167669)
Wife initiates discussion (birth control use) .12275242 (5.6980881) ⫺.04567825 (⫺3.6508717)
Husband initiates discussion (birth control use) .06501086 (3.3148577) .00328392 (.27138652)
Wife initiates discussion (children’s marriage) .0733686 (3.0038177) ⫺.01647947 (⫺1.1171852)
Husbamd initiates discussion (children’s marriage) .05491004 (2.2773254) ⫺.01561035 (⫺1.0662414)
Wife initiates discussion (children’s education) .12380831 (5.2159512) ⫺.00700765 (⫺.47441271)
Husband initiates discussion (children’s education) .14298343 (6.100405) ⫺.01911873 (⫺1.3296468)
Wife initiates discussion (birth timing) .01643459 (.83961744) .00820515 (.68634246)
Husband initiates discussion (birth timing) .03155623 (1.6648596) .02181647 (1.8607394)
Wife initiates discussion (birth numbers) .08293512 (4.0749984) ⫺.0060319 (⫺.49479264)
Husband initiates discussion (birth numbers) .08366957 (4.298626) .01287064 (1.0768375)
House repair decision .08907876 (4.2620781) ⫺.00975032 (⫺.78694676)
House repair implementation .07207265 (3.3856953) ⫺.05098127 (⫺4.2001659)
House repair spending .13213097 (2.2869854) ⫺.02265464 (⫺1.044386)
Livestock purchase decision .12567333 (5.7735449) ⫺.01387871 (⫺1.087525)
Livestock purchase implementation .06872113 (3.5365616) ⫺.03405356 (⫺2.9825091)
Livestock spending .11883032 (2.4274332) ⫺.00371256 (⫺.20287309)
Household loans decision .16210609 (7.6887465) ⫺.04490973 (⫺3.8469249)
Household loans implementation .14057259 (7.1552867) ⫺.06104301 (⫺5.4802638)
Household loans spending .12991344 (3.3276476) ⫺.08472108 (⫺2.2018748)
Land/equipment decision .12638888 (6.2372212) ⫺.02497008 (⫺2.1609634)
Land/equipment implementation .09766105 (5.1815832) ⫺.03660103 (⫺3.3099373)
Land/equipment spending .08693242 (1.5483862) ⫺.03123732 (⫺.76965112)
Husband says wife is intelligent .09782371 (5.1313304) ⫺.0359709 (⫺3.1449831)
Wife can buy an asset .07018261 (3.3838999) ⫺.0128851 (⫺1.0876201)
Wife can buy an asset (without husband’s permission) .06747569 (2.8552021) ⫺.00803558 (⫺.60671635)
Wife has own income .12659167 (6.4359697) ⫺.01102797 (⫺.99794156)
Husband spends wife’s loan money ⫺.35438 (⫺9.711638) ⫺.01802366 (⫺.68702186)
Husband says wife travels alone .12570957 (5.7274842) ⫺.01651619 (⫺1.3198174)
Wife has independent income .08544579 (4.1829839) ⫺.00230439 (⫺.20117939)
Wife has independent savings .4468341 (21.264715) ⫺.1220153 (⫺10.70441)
Wife has independent savings which she herself controls .04887458 (1.6630768) .00488506 (.29625473)
Emergency funds access .17599848 (8.8439957) ⫺.04927985 (⫺4.3592823)
Emergency funds access (asset sale) .15163642 (3.7263403) ⫺.05101663 (⫺2.0145894)
Emergency funds access (from husband) .05816154 (2.7675279) .00836456 (.69821613)
Emergency funds access (husband’s relatives) .11225464 (5.3892888) ⫺.02625655 (⫺2.1466454)
Emergency funds access (own relatives) ⫺.0035531 (⫺.17016661) ⫺.02508447 (⫺2.0300173)
Emergency funds access (moneylenders) ⫺.03163719 (⫺.65054464) .01068736 (.36908294)
Emergency funds access (other people) .08425008 (2.4472459) ⫺.00779705 (⫺.37519001)
Wife’s control over loans ⫺.01703254 (⫺.60133783) ⫺.0205718 (⫺1.1283077)
Remittance ⫺.05035626 (⫺2.1274473) .00549734 (.41368397)
Wife can decide how to use remittance ⫺.02334331 (⫺.36612289) .01616041 (.43267576)
Money seizure by husband ⫺.01399654 (⫺.68291502) .01176347 (1.0159757)
Freedom to remit .07680612 (3.554106) ⫺.01941725 (⫺1.596456)
Husband forbids work outside home ⫺.08583542 (⫺4.4739977) .03820209 (3.3884507)
Visits relatives (without husband’s permission) .01911277 (.91810268) ⫺.04657823 (⫺3.3794914)
Marriage has kabinnama .02275705 (1.1305345) ⫺.00044701 (⫺.03591271)
Awareness of kabinnama .05771666 (1.8559487) .0066104 (.33847749)
Awareness of inheritance laws .02878196 (.91363229) ⫺.01984606 (⫺1.0841944)
Has prevented husband remarrying .00728302 (.19819264) .00616823 (.29205465)
Knows MP’s name .08391138 (4.3870888) ⫺.00768854 (⫺.68322894)
Voted (at all) .13131678 (5.1633827) ⫺.00194159 (⫺.12746871)
Voted independently .0414437 (2.0471695) ⫺.02997352 (⫺2.3834331)
Protested against domestic abuse .03678351 (1.8633071) .01145688 (1.027746)
Thinks dowry is good .01067717 (.51269369) .00030867 (.02543997)
Protested against corruption .04417861 (1.2452415) ⫺.00056154 (⫺.02879464)
Confidant within bari ⫺.03205397 (⫺1.4444465) .00007672 (.00579186)

Pitt, Khandker, and Cartwright 829
TABLE A4 (Continued )
Name of Variable
Female Credit
Coefficient
(t-Statistic)
Male Credit
Coefficient
(t-Statistic)
Interval of contact within bari ⫺.03091528 (⫺1.3888953) ⫺.00121859 (⫺.09191732)
Confidant outside bari .06783766 (3.4223935) .01653874 (1.314044)
Interval of contact outside bari .07150086 (3.6184072) .01697922 (1.3522985)
Severity of spousal arguments ⫺.0394947 (⫺1.6873243) .01730603 (1.2730856)
Occurrence of physical spousal abuse ⫺.03443847 (⫺1.5268851) .00547143 (.42907311)
Own relatives in same village .08333928 (4.0508622) ⫺.06561099 (⫺4.6359715)
Wife thinks husband is superior .1450791 (4.872342) ⫺.01446311 (⫺.87795812)
Husband uses male birth control .02956166 (.93340601) ⫺.00821234 (⫺.44513926)
Reason pwomen not allowed outside .14900052 (5.0660569) ⫺.02041713 (⫺1.1715493)
Reason plack of safety ⫺.06532019 (⫺2.328394) ⫺.00085579 (⫺.05026752)
Reason pwife goes with husband/son .00564198 (.27281459) .03149786 (2.5582753)
Reason pwife goes with neighbor .14400445 (6.1232355) ⫺.06842666 (⫺4.468412)
Husband says women’s eempowerment pbetter society .05421608 (2.7797372) ⫺.01704432 (⫺1.5155241)
Husband says women’s empowerment pchaos in
society .03310243 (1.7129112) ⫺.00927146 (⫺.82634147)
Husband says women’s empowerment pproblems with
kids .02226409 (1.073948) ⫺.01075105 (⫺.88421994)
Husband says women’s empowerment ploss of peace .0175682 (.88853243) ⫺.00413204 (⫺.36055211)
Husband says women’s empowerment pbetter
economically .04666237 (2.3018161) ⫺.00193728 (⫺.16867565)
Husband cites positive impact of women’s
empowerment .04462788 (2.3051181) ⫺.02053908 (⫺1.8382332)
Husband cites negative impact of women’s
empowerment .05758083 (2.9349831) ⫺.00996125 (⫺.88881097)
Husband’s assessment of women’s empowerment .04144622 (2.1361352) ⫺.01981529 (⫺1.7648085)
Husband views lack of education as obstacle .00169411 (.09063681) .00593213 (.54870205)
Husband views lack of safety as obstacle ⫺.01163518 (⫺.6298988) ⫺.01664214 (⫺1.5188646)
Husband views lack of income-generating activity as
obstacle .06036897 (2.8999283) .02357819 (2.0845775)
Husband views social structure as obstacle ⫺.04693505 (⫺2.2465429) .00915417 (.76310031)
Husband views law as obstacle ⫺.03500877 (⫺1.0524301) ⫺.05751117 (⫺2.1869163)
Wife has made husband use birth control ⫺.00065728 (⫺.0240695) ⫺.03375585 (⫺1.7794337)
Wife has income-generating activity .15477253 (8.1828396) ⫺.04342009 (⫺3.9260772)
Wife has income-generating activity that she herself
operates .09505125 (4.7972341) ⫺.01599552 (⫺1.3898103)
Degree of mobility .31089568 (7.9629522) ⫺.07537371 (⫺4.2162371)
Wife ever travels .31089568 (7.9629522) ⫺.07537371 (⫺4.2162371)
Wife ever travels alone .11742115 (5.0836324) ⫺.02500812 (⫺1.8386287)
Prevent remarriage (threaten divorce) ⫺.06295729 (⫺2.779748) ⫺.00846561 (⫺.5652342)
Prevent remarriage (family pressure) .01169626 (.58635634) ⫺.0105242 (⫺.85967706)
Prevent remarriage (local government) .04096012 (2.0876515) .01848534 (1.5525177)
Prevent remarriage (parishad) .02593949 (1.2173839) .01867645 (1.4848455)
Prevent remarriage (deny permission) ⫺.01546954 (⫺.81803909) .007914 (.68036286)
Household fights about children .01404173 (.77187727) ⫺.01900658 (⫺1.7803394)
Household fights about money .04539589 (2.4881701) .02509254 (2.3033101)
Household fights about in-laws ⫺.04694592 (⫺1.3519341) ⫺.00885686 (⫺.47053079)
Household fights about going outside .1881154 (3.5742584) ⫺.02962608 (⫺1.0165987)
Household fights about loans ⫺.02284063 (⫺.85639037) .04246613 (2.1445597)
Household fights about chores ⫺.02625961 (⫺1.4146924) ⫺.025423 (⫺2.2501207)
Wife views lack of education as obstacle ⫺.00427501 (⫺.2370279) .01070229 (1.0109335)
Wife views lack of safety as obstacle ⫺.04081731 (⫺2.2378315) ⫺.01799485 (⫺1.6545223)
Wife views lack of jobs as obstacle .03693626 (1.9910125) ⫺.00553119 (⫺.51098451)
Wife views social structure as obstacle ⫺.03275009 (⫺1.7516397) .02392744 (2.2276055)
Wife views laws as obstacle .00952706 (.36527339) ⫺.00408264 (⫺.2639975)
Wife views religion as obstacle .02434441 (1.3408948) ⫺.04375407 (⫺4.0588575)
References
Abrevaya, Jason, and Jerry A. Hausman. 2004. “Response Error in a Transformation
Model with an Application to Earnings-Equation Estimation.” Econometrics Journal
7:366–88.

830 economic development and cultural change
Beegle, Kathleen, Elizabeth Frankenberg, and Duncan Thomas. 2001. “Bargaining
Power within Couples and Use of Prenatal Delivery and Case in Indonesia.” Studies
in Family Planning 32, no. 2:130–46.
Bentler, Peter M. 1992. EQS: Structural Equation Program Manual. Los Angeles:
BMDP Statistical Software.
Browning, M., and P. A. Chiappori. 1998. “Efficient Intra-household Allocations:
A General Characterization and Empirical Tests.” Econometrica 66:1241–78.
Frankenberg, Elizabeth, and Duncan Thomas. 2001. “Measuring Power.” FCND
Discussion Paper no. 113, International Food Policy Research Institute, Wash-
ington, DC.
Goetz, A., and R. Rina Sengupta. 1996. “Who Takes the Credit? Gender, Power,
and Control Over Loan Use in Rural Credit Programs in Bangladesh.” World
Development 24, no. 1:45–63.
Hashemi, S., Sidney R. Schuler, and Ann P. Riley. 1996. “Rural Credit Programs
and Women’s Empowerment in Bangladesh.” World Development 24, no. 4:635–53.
Jo¨reskog, K. G., and D. So¨ rbom. 1999. LISREL 8 User’s Reference Guide. Chicago:
Scientific Software International.
Khandker, Shahidur R. 1998. Fighting Poverty with Microcredit: Experience in Bangla-
desh. New York: Oxford University Press.
Montgomery, Richard, Debapriya Bhattacharya, and David Hulme. 1996. “Credit
for the Poor in Bangladesh: The BRAC Rural Development and Employment
Programme.” In Finance against Poverty, vol. 2, ed. David Hulme and Paul Mosely.
New York: Routledge.
Moustaki, Irina, Karl G. Jo¨ reskog, and Dimitris Mavridis. 2004. “Factor Models for
Ordinal Variables with Covariate Effects on the Manifest and Latent Variables: A
Comparison of LISREL and IRT Approaches.” Structural Equation Modeling 11, no.
4:487–504.
Muthe´n, Bengt O., and Linda K. Muthe´n. 2000. Mplus: The Comprehensive Modeling
Program to Applied Researchers. Bengt O. Muthe´n and Linda K. Muthe´n, 11965
Venice Boulevard, Suite 407, Los Angeles, CA.
Pitt, Mark M. 1999. “Reply to Jonathan Morduch’s ‘Does Microfinance Really Help
the Poor? New Evidence from Flagship Programs in Bangladesh,’” October 14,
1999, http://www.pstc.brown.edu/˜mp/reply.pdf.
———. 2000. “The Effect of Nonagricultural Self-Employment Credit on Con-
tractual Relations and Employment in Agriculture: The Case of Microcredit Pro-
grams in Bangladesh.” Bangladesh Development Studies 26, nos. 2–3:15–48.
Pitt, Mark, and Shahidur R. Khandker. 1998. “The Impact of Group-Based Credit
Programs on Poor Households in Bangladesh: Does the Gender of the Participant
Matter?” Journal of Political Economy 106:958–96.
Pitt, Mark, Shahidur Khandker, O. Choudhury, and D. Millimet. 2003. “Credit
Programs for the Poor and the Health Status of Children in Rural Bangladesh.”
International Economic Review 44 (February): 87–118.
Pitt, Mark M., Shahidur R. Khandker, Signe-Mary McKernan, and M. A. Latif.
1999. “Credit Programs for the Poor and Reproductive Behavior in Low Income
Countries: Are the Reported Causal Relationships the Result of Heterogeneity
Bias?” Demography 33 (February): 1–21.

Pitt, Khandker, and Cartwright 831
Rabe-Hesketh, Sophia, Anders Skrondal, and Andrew Pickles. 2004. “GLLAMM
Manual.” Working Paper 160, Division of Biostatistics Working Paper Series,
University of California, Berkeley.
Skrondal, Anders, and Sophia Rabe-Hesketh. 2004. Generalizing Latent Variable Mod-
eling: Multilevel, Longitudinal, and Structural Equation Models. Boca Raton, Fl: Chap-
man & Hall/CRC.