Actually how Empowering is Microcredit?

Simeen Mahmud

ABSTRACT

This article re-assesses the effect of microcredit programme participation on

women’s empowerment by applying an analytical framework that recognizes
the conceptual shift in emphasis in the definition of empowerment, from
notions of greater well-being of women to notions of women’s choice and
active agency in the attainment of greater well-being. The author finds that
microcredit programme participation has only a limited direct effect in
increasing women’s access to choice-enhancing resources, but has a much
stronger effect in increasing women’s ability to exercise agency in intra-
household processes. Consequently, programme participation is able to increase
women’s welfare and possibly to reduce male bias in welfare outcomes,
particularly in poor households.

INTRODUCTION

Women’s empowerment in the development process has, up to now, been

viewed as the achievement of a ‘better deal’ for women, with the ‘concen-
tration . . . mainly on women’s well-being’ (Sen, 2000: 189). Consequently,
and quite justifiably, the reduction and ultimate removal of gender inequal-
ity in well-being, both economic and social, has been the articulated goal in
all the internationally agreed development targets. National development
agendas, too, have primarily addressed the relative deprivations experienced
by women, namely excess mortality, greater undernutrition, higher illiter-
acy, and so on.
However, the empirical association between aggregate economic develop-
ment (reflected in GDP growth, levels of urbanization and industrialization,
aggregate literacy, availability of public services) and reduction in gender
bias in well-being was never very strong (Sen, 2000). In fact, any positive
association was actually found to work mainly through variables that
directly influenced the ability of women to exercise agency, namely formal
schooling and participation in the labour force. Thus, there came a gradual

The author wishes to thank Prof. Naila Kabeer for her extremely thoughtful and detailed
comments, and an anonymous referee of this journal for very useful feedback on an earlier
draft.
Development and Change 34(4): 577–605 (2003). # Institute of Social Studies 2003. Published
by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St.,
Malden, MA 02148, USA
578 Simeen Mahmud

realization that improvements in women’s well-being were closely connected

to women’s capabilities as agents of change and, subsequently, that these
capabilities of women must also strongly influence the well-being of other
family members, particularly of children but also of the elderly and of adult
men.
Over time the stated development goals of women’s empowerment have
evolved and broadened, from a concentration on the relative well-being of
women vis-à-vis men to the incorporation of notions of women’s agency in
the attainment of greater well-being for all, and particularly for women. It is
important to distinguish this shift in emphasis as one from passive accept-
ance of well-being enhancing assistance to one of active participation in the
attainment of those well-being objectives. In fact, women’s agency is being
increasingly posed as the pathway of social change that ‘can alter the lives of
both women and men’ (Sen, 2000: 189). However, this realization and the
accompanying shift in emphasis have not yet permeated into the inter-
national and national development agendas. There is a need to re-define
women’s empowerment so that the choice and agency aspects of this process
are seen as both distinct from and overlapping with the well-being aspects.
It is also important to recognize women’s agency as an important develop-
ment goal in itself, and not just as a means to achieving other development
objectives.
Some of the most compelling ideas about the relationship between socio-
economic development and women’s empowerment have emerged from
grassroots development experience in Bangladesh, an experience that has
been both unique and varied.1 It is hardly surprising that the outcomes of
this unique and varied experience have also been quite remarkable. First,
there have been significant development successes in Bangladesh even in the
face of severe adversities like political instability, inefficient aid utilization
and frequent natural disasters. These ‘successes’ are well known: a rapid fall
in birth rates, improvements in child survival and nutrition, increase in
school enrolment, reduction in gender gaps in survival and schooling, and
women’s increased labour force participation in the modern sectors. Second,
in all of these ‘success’ stories, women have emerged as the major players: as
factory workers in the manufacturing export sector, as borrowers of micro-
credit and as users of modern contraceptives. Clearly, the Bangladesh
experience has much to contribute towards a greater understanding of the
relationship between empowerment and socio-economic development.

1. The Bangladesh experience has been unique because it has been the ‘laboratory’ for some
of the most challenging and innovative development experiments in the contemporary
world. It has been varied because development interventions have ranged from vertical to
horizontal, conventional to radical, and imported to domestically evolved; and because
development actors have been many: the state, multilateral agencies, non-governmental
organizations and private entrepreneurs.
How Empowering is Microcredit? 579

Empirical evidence on this relationship comes from evaluation studies

of microcredit programmes, the largest development programme in
Bangladesh today and one that is believed to have strong conceptual links to
women’s empowerment. Unfortunately, however, assessments of the impact
of microcredit programme participation on women’s empowerment have
yielded contrary results, which can be difficult to reconcile with each other,
even when evaluating the same programme. Some studies find that partici-
pation in microcredit programmes results in women’s economic and social
empowerment, while others conclude that participation leads to greater
subordination of women by reinforcing patriarchal norms of behaviour
(see Kabeer, 2001 for a comprehensive review of evaluation studies).
Given the lack of consensus, this article re-examines the effect of micro-
credit programme participation on the process of women’s empowerment
by taking into account the conceptual shift and drawing upon the rich
grassroots development experience in Bangladesh. The next two sections
critically review theoretical frameworks of women’s empowerment and
empirical efforts so far to measure women’s empowerment in Bangladesh,
and then propose an operational framework for measuring women’s
empowerment. This is followed by a discussion of data and methods, and
the presentation and examination of results, before some brief concluding
remarks.

THE MEASUREMENT OF WOMEN’S EMPOWERMENT IN BANGLADESH

One of the most visible recent changes in the lives of rural women in
Bangladesh is the significant increase in their access to credit. Loans are
provided to women by microcredit programmes of non-governmental organ-
izations (NGOs) and the Grameen Bank2 through informal groups mobil-
ized as part of a programme strategy to reach the poor.3 Although access to
credit and women’s empowerment are inextricably linked in much of the
international discourse on the impact of microcredit, within Bangladesh the
major policy thrust of microfinance institutions has not been women’s
empowerment but the broader macroeconomic imperative of employment
generation and poverty reduction. However, given the gender and class

2. By June 2001 such programmes covered 9.8 million women in rural areas. The expansion
in coverage has been increasing since the mid-1980s with the fastest growth occurring after
the mid-1990s (CDF, 2002). Membership in microcredit programmes is overwhelmingly
female (over 85 per cent) and programme inputs, particularly credit, are targeted to the
poorer segment of the population.
3. Group formation is organized initially around savings and awareness-raising activities.
Later, collateral-free credit is provided to group members for various income-generating
activities, undertaken either individually by households or, less frequently, by the group as
a whole. The collateral is in the form of peer monitoring by group members, which creates
a pressure to maintain loan repayment discipline.
580 Simeen Mahmud

based nature of rural Bangladesh society, it is hardly surprising that the

tremendous capacity of microcredit programmes to reach poor women
should have important consequences for women’s empowerment.4
Recently, concern with the ‘empowering’ effect of credit programme
participation has arisen as a result of pressure, largely coming from outside
Bangladesh, to make programmes financially sustainable (Murdoch, 2002).
This in turn has led to a growing tendency by programmes to adopt a
‘minimalist’ approach which de-emphasizes the non-credit inputs (such as
adult literacy, legal awareness, health and hygiene awareness, leadership
training, skills training, management training, and so on) that support
women’s programme participation.5 Such an approach to service provision
and women’s mobilization accompanied, as it is, by rapid scaling-up of
credit operations has direct implications for women’s empowerment. For
example, programme approaches that emphasized household welfare or
general poverty reduction were found to be less effective in achieving
women’s direct investment of loans in irrigation equipment and their con-
trol over income from loan use, compared to programmes that emphasized
women’s welfare specifically (van Koppen and Mahmud, 1996). Given that
women represent the major conduit for credit transfer to poor rural house-
holds it is quite pertinent to ask whether there are indeed any significant
payoffs for women from participation in these programmes.
In evaluating the impact of microcredit programmes, research studies
have frequently addressed this question from different disciplinary stances
and from a variety of theoretical perspectives on women’s empowerment.6
There is general consensus that empowerment is a process that is manifested
in more than one dimension. The most frequently identified dimension is
women’s absolute well-being, where empowerment is seen as the process of
improving welfare of women and girls, indicated by outcomes that measure
current status with respect to literacy and schooling, health and nutrition,

4. It is well recognized that socio-economic constraints on women’s control over productive

resources are major reasons for their low status generally and their subordinate position
relative to men. Hence, it is argued that credit to women, by increasing access to resources,
can play an effective role in improving their socio-economic status and in reducing their
economic and social subordination. Moreover, since the main mechanism for credit
delivery is through the process of group formation, which mobilizes poor women for
collective action at the village level, it is also likely that group membership promotes an
improvement in gender relations within the home and in the larger community (Mahmud,
2002). On both these grounds participation in microcredit programmes may be an
‘empowering’ experience for poor women, irrespective of whether there are any direct
poverty reducing effects or not.
5. ASA, the third largest microcredit programme in Bangladesh, with 1.33 million active
members in December 2001, adopts the minimalist approach and provides only credit
services (CDF, 2002).
6. See Amin and Pebley (1994); Goetz and Sen Gupta (1996); Hashemi et al. (1996); Huda
and Mahmud (1998); Kabeer (1998); Mahmud (1991, 1994); Pitt and Khandker (1995);
Rahman (1986); Steele et al. (1998); Zaman (1996).
How Empowering is Microcredit? 581

labour force participation, contraceptive use, mobility and ownership of

clothing and assets. Another dimension is women’s relative well-being,
where empowerment is seen as the process of improving the position of
women relative to men within the household, and is indicated by women’s
involvement in intra-household processes like decision-making, control over
household income and assets, control over loans taken, and so on.
However, while the multi-dimensional nature of the empowerment pro-
cess is recognized, the conceptual links between women’s relative position
and absolute welfare are usually not adequately explicated. In addition,
indicators used to measure empowerment at the different dimensions suffer
from a lack of grounding in reality and absence of empirical validation.

Conceptual Frameworks

An influential hypothesis regarding women’s empowerment that has domin-

ated the rationale of development programmes targeted at women, espe-
cially microfinance, views the effect of access to financial services like credit
on women’s empowerment as a set of mutually-reinforcing ‘virtuous spirals’
of increasing economic empowerment, improved well-being and social/
political/legal empowerment for women (see Mayoux, 1999). Mayoux suggests
that this dominant view among practitioners, and many researchers, has
emerged as the ‘confluence of three rather distinct paradigms’ of microcredit
programmes for women that come from different theoretical underpinnings
but are believed to lead to similar virtuous spirals of empowerment. These are
the financial self-sustainability paradigm, the poverty alleviation paradigm
and the feminist empowerment paradigm.
The financial self-sustainability paradigm preferred by economists, which
advocates large-scale, minimalist financial services for the poor, assumes
that financial services that stimulate women’s microenterprise development
will lead to increased income under women’s control which will in turn lead
to poverty reduction and increased well-being (consumption, health status,
nutrition, literacy, housing) for women and their children (see for example
Pitt and Khandker, 1995). In the poverty alleviation paradigm financial and
other services targeted at women are seen as fulfilling women’s practical
needs for income and employment, and are believed to be the best way of
enabling women to address gender inequality and to lead to women’s
empowerment. It is assumed that poverty alleviation benefits women par-
ticularly because of higher levels of poverty among women, and because of
women’s greater responsibility for family welfare. The feminist paradigm
regards women’s empowerment as an end in itself, requiring both a process
of internal change at the individual level and organization at the macro
level. Under all three paradigms women’s individual economic empower-
ment is believed to lead to wider social, political and legal empowerment
582 Simeen Mahmud

and to contribute to building ‘social capital’ through developing and

strengthening women’s networks.
Mayoux (1999) critiques this framework on the grounds that the assump-
tion that access to microfinance automatically contributes to increased
incomes by stimulating women’s own economic activity is frequently unten-
able. There are a number of reasons for this: there are barriers and con-
straints to access to participation; the impact on women’s incomes is small;
control over income is not always evident; norms regarding intra-household
responsibilities and rights vary; and the increase in women’s access to other
more formal networks and services is not well established. Moreover,
although the transformation of social and economic structures of women’s
subordination or wider social empowerment is seen as an eventual outcome,
the causal mechanism is assumed rather than explained and the role of
women’s agency in contributing to the virtuous spirals of empowerment is
not sufficiently spelled out.
A contrasting but increasingly influential thesis views the effect of
women’s access to credit as reinforcing patriarchal norms of women’s
subordination, leading to worsening gender relationships and disempower-
ment of women (Goetz and Sen Gupta, 1996; Montgomery et al., 1996;
Rahman, 1999). Drawing on case studies of microcredit borrowers this
thesis questions how far women have been ‘able to convert access to credit,
and membership in credit organizations, into a process of empowerment’,
where the implicit definition of empowerment is a ‘demonstrated capacity
to invest loans profitably’ (Goetz and Sen Gupta, 1996: 47). It is argued that
credit imposes a burden upon women as debt collectors for microfinance
organizations. The intense pressure of timely loan repayment produces new
forms of social and institutional dominance over women by families and by
microfinance organizations, increasing tensions within the family and in the
relationship between women clients of microcredit and programme person-
nel. There is also the hypothesis that microcredit diverts attention and
resources from other more effective strategies for women’s empowerment.
The assumption is that new cash inputs to the household through women
are likely to be identified by household males as a resource for their use. It is
also believed that women’s labour burden is increased without any signifi-
cant concomitant increase in control over the productive process. As a
result, it is suggested that women do not benefit from access to credit in
terms of welfare outcomes. Moreover, it has been argued that the ‘hidden
transcript’ of programmes which target women to generate ‘social collateral’
in order to ensure high repayment rates, may escalate violence toward
women borrowers (Rahman, 1999).
However, a case can also be made that women use their relatively greater
access to credit to gain some degree of leverage in unequal gender relation-
ships, and may actually negotiate a better bargaining position within mar-
riage by handing over loans to husbands. Loan use by male relatives also
makes greater economic sense and is more likely to ensure financial security
How Empowering is Microcredit? 583

through higher returns. Thus, the assumption that simple access to credit
without significant or full control over loans improves women’s relative
position in the household is also quite plausible. The question also remains
whether ‘demonstrated capacity to invest loans profitably’ can be a realistic
definition of empowerment in the context of rural Bangladesh, where there
are significant barriers to the expansion of women’s productive investments
and women’s access to markets and public services is restricted (Hashemi
et al., 1996; Kabeer, 2001). In fact, capacity to invest loans actually represents
the different ways in which women respond to credit provision rather than
the impact of access to credit, and is determined very much by women’s
pre-existing capacities and the pre-existing situation of their households.

Measurement Indicators

The weakness of concepts means that the relationships between indicators

used to measure empowerment at different dimensions are assumed without
adequate causal explanations and without sufficient empirical support. In
one study access to credit was seen as empowering because the outcome
‘increased income earning time spent in credit-based activities’ was assumed
to automatically lead to greater influence in household decision-making, but
without providing any evidence of a positive relationship between women’s
role in household decisions and time spent in market production (Pitt and
Khandker, 1995). By contrast, the outcome ‘use of loans by male relatives’
was interpreted in another study as loss of women’s direct control over
loans and seen as reinforcing unequal gender relationships, so that access to
credit was interpreted as disempowering (Goetz and Sen Gupta, 1996). In
this case, too, no evidence was given to support the assumption that male
loan use was equivalent to women’s loss of control.
A related problem is that indicators used are not well grounded in women’s
lived reality. For example, contrary to popular notion the empirical relation-
ship between women’s role in decision-making and welfare outcomes is weak.
Cases in which women make decisions independently — that is, women
considered to be most empowered — do not necessarily have the best welfare
outcomes simply because women’s independence in household decision-
making usually takes place in female-headed households that are generally
poorer and have lower levels of aggregate welfare compared to male-headed
households where women’s decision-making is less likely to be independent.
Moreover, since the validity of an indicator as a measure of empower-
ment is not always empirically established, commonly used indicators
(regardless of the theoretical underpinning of the concept of women’s
empowerment) frequently over- or under-estimate the extent of empower-
ment. As Kabeer (1999) rightly points out, women’s perspectives are also
rarely taken into account when assigning value to indicators. For example,
an independent role in household decision-making is commonly assigned
584 Simeen Mahmud

a high value on the scale of empowerment. In reality, however, co-operation

and joint decision-making with husbands or male relatives is actually more
empowering in some situations, such as in matters that require access to
public institutions like markets, legal and financial agencies and public
service providers, because of women’s restricted access to the public
domain. Hence, from the perspective of women themselves, an independent
role in household decisions is not always more empowering, and often a
joint or co-operative role rather than an independent role reflects active
participation and exercise of agency more accurately.
Women also face difficult trade-offs between the exercise of choice and
adherence to existing social norms, many of which perpetuate women’s
subordination.7 These trade-offs impose costs upon women that can influ-
ence outcomes and behaviours. In other words, observed outcomes are
highly context dependent, and cannot on their own indicate the role played
by women’s enhanced choice and agency, or that played by realistic cost-
minimizing strategies. Therefore, using an indicator in isolation of the
underlying household process not only poses a validity problem but can
also give misleading assessments of empowerment. Using the variable ‘who
uses the loan’ to measure loan control is a clear case of such a pitfall (Goetz
and Sen Gupta, 1996). Because institutions and norms are generally not
supportive of women’s independent use of NGO loans, the variable ‘who
uses the loan’ cannot meaningfully assess women’s control over loans unless
it is also known whether women have a say in how the loan will be used and
how income from loan use will be spent. In Kabeer’s 2001 study of women
borrowers, the percentage of women in female loanee households in her
sample who had some say in decision-making varied between 40 per cent as
primary decision-maker and 90 per cent as joint decision-maker.
Finally, there is an inclination to assume that the empowering process is
triggered only by exogenous events, overlooking the importance of pre-
conditions like socio-economic status, life-cycle situation and intra-
household relationships, as well as local social norms and practices, in shaping
the environment in which new experiences can or cannot be empowering.
Thus, the initial situation of individual women mediates the empowering effect
of subsequent experiences, including joining a microcredit programme.

A CONCEPTUAL FRAMEWORK FOR ASSESSING WOMEN’S

EMPOWERMENT IN BANGLADESH

The enormous challenge in assessing the process of empowerment is

captured by Kabeer’s observation that ‘Empowerment is seen to occur at

7. One such trade-off is between the pressure to maintain the norms of purdah (seclusion)
that limit women’s access to markets and other public institutions and the need to
maintain control over the use of loans taken by them.
How Empowering is Microcredit? 585

a number of different levels, to cover a range of different dimensions and

to materialize through a variety of different processes’ (Kabeer, 1999: 2).
Kabeer’s concept of empowerment rests upon the notion of power as
determining choice and ability to choose, and with how lack of power and choice
can be ‘disempowering’. She refers to empowerment as ‘processes by which
those who have been denied the ability to make choices acquire such an
ability’ (ibid.). Kabeer further points out that a distinction has to be made
about the type of choice, and the focus necessarily has to be on strategic
life choices, that is, choices that shape livelihoods or are ‘critical for
people to live the lives they want’ (ibid.: 3). In her framework for assessing
empowerment Kabeer suggests tracing women’s ability to make choices
through three dimensions: the pre-conditions of choice or resources; the
exercise of choice or agency; and the consequences of choice or achieve-
ments that reflect increased capacity to transform the structures of
women’s subordination.
In a complimentary approach Chen and Mahmud (1995) distinguish
different dimensions of empowerment — material, cognitive, perceptual
and relational — on the basis of pathways of positive change, the conse-
quence of which is women’s improved fall-back position and greater bar-
gaining power. Material empowerment occurs through expansion in the
material resource base of women. Cognitive empowerment occurs from
women’s recognition of their own abilities and skills, indicated by greater
self-esteem and self-confidence. Perceptual empowerment occurs through
changes in how others perceive them, indicated by increased social prestige
and value. Relational empowerment takes place through changes in gender
relations within the family and in the broader society, indicated by reduc-
tion in gender inequality in relationships. These empowering changes are
triggered by specific events in women’s lives like schooling, paid work and
participation in development programmes, and by secular life cycle events
like marriage, birth of children, setting up of separate household, marriage
of children and divorce or widowhood.
In both these frameworks women’s empowerment is evident not only
from the process — women’s agency — but also from the outcome —
women’s greater welfare. However, while Kabeer emphasizes empowerment
as a process of internal change, Chen and Mahmud identify external
changes as also being important for empowerment, namely through the
changed perceptions of others and through changed gender relations.
Drawing upon them, an operational framework for assessing women’s
empowerment in Bangladesh is presented as follows. In Bangladesh social,
economic and political structures generate norms and practices of women’s
subordination to men, the consequences of which are women’s relative lack
of choice and agency in decisions that shape their welfare. In this context
empowerment is the process of gaining power over those strategic life
decisions through expansion of women’s choice and increase in women’s
agency in order to increase welfare, both absolute and relative, and reduce
586 Simeen Mahmud

Table 1. Complementarity of Frameworks

Kabeer framework Chen and Mahmud framework Operational framework

Choice: resources Resource base: Condition for empowerment:

material resources access to material and
non-material resources
Prestige and value:
non-material resources
Exercise of choice: Self esteem and self confidence: Route to empowerment:
sense of agency cognitive skills and abilities active participation
in household processes
Improved position
vis-à-vis men: gender relations
Achievements: Improved fall-back Achievements:
increased capacity position and greater increased welfare
to transform bargaining strength and reduced subordination
structures

subordination to men. This process of change evolves over three dimen-

sions: the condition for empowerment, the route to empowerment and the
achievements of empowerment. In each dimension there are several path-
ways through which change occurs. Table 1 illustrates the complementarity
among these frameworks.
The condition for empowerment is defined by the set of choices and
options initially available to women for increasing welfare and reducing
subordination to men. Expansion of choice renders this condition more
favourable, regardless of whether those choices are eventually exercised or
not. In practice absence of choice is a consequence of the gendered nature of
access to and control over material and non-material resources. Therefore,
in this dimension empowerment results from an increase in access to and
control over choice-enhancing resources under male control,8 which both
strengthens women’s resource base and increases their social prestige and
value. The pathway of change is both material, through increase in access to
and control over material resources like land and assets, employment and
human capital like education and skills; and perceptual through increase in
access to non-material resources such as non-kin networks, public services
and markets, and favourable household and community attitudes.
The route to empowerment is the translation of choice-enhancing
resources into greater welfare for women and reduced subordination to
men. In this dimension empowerment takes place through women’s agency
in intra-household processes that determine welfare, such as strategic
decision-making, allocation of resources and income control. Women’s active
participation in these processes is a more direct and sustained means of
increasing welfare and reducing subordination than passive acceptance of

8. It is recognized that resource access and control will both increase choice, but in different
ways and in different degrees.
How Empowering is Microcredit? 587

the actions of others because it facilitates the incorporation of women’s own

preferences into household decisions and reduces their dependence on male
family members. The transition from passivity to agency takes the cognitive
and relational pathways. Cognitive empowerment consists of increase in
women’s self esteem and self confidence that comes from greater cognitive
skills and capabilities to influence household processes, such as ability to
judge the welfare implications of different actions, ability to make informed
decisions, and ability to be self-reliant. Relational empowerment consists of
improvement in women’s position vis-à-vis men within the household and
comes from more egalitarian intra-household relationships.
The immediate achievement of empowerment is increased welfare of
women, which encompasses the welfare of other family members particu-
larly children. Immediate achievements are outcomes that reflect improve-
ment in both women’s absolute welfare and in welfare vis-à-vis men reflected
in reduced male bias in welfare outcomes such as literacy, health, nutrition,
mortality and morbidity, consumption, and so on. The long-term achieve-
ment is reduced subordination to men resulting from the transformation of
social, economic and political structures that constrain women’s welfare and
reinforce women’s subordination. Thus, long-term achievements are out-
comes reflecting changed norms and practices with respect to institutions
such as inheritance, marriage, resource ownership and control, division of
labour, purdah, and political power.
Empowerment is thus an evolving process, with empowering changes at
one dimension leading to empowering changes in subsequent dimensions.
The framework allows for the possibility that there may be varied entry
points of impact of development interventions on this process and different
pathways of empowering change. It recognizes the reality that empowering
effects of an experience like participation in a microcredit programme can
be mediated, for example, by where a woman is situated in her life cycle. By
identifying and distinguishing women according to the initial condition for
empowerment this framework is also useful for tackling the selectivity
question, which contends that the empowering effect of development pro-
grammes is over-estimated because women who join programmes have a
relatively more favourable initial condition.

DATA AND METHODS

The above operational framework is used to re-assess the relationship

between participation in microcredit programmes and women’s empower-
ment in rural Bangladesh using new data from a large representative survey
of rural households in programme and non-programme villages. The meth-
odological problem of using survey data to assess a dynamic process like
empowerment lies in the inadequacy of quantitative indicators to capture
difficult-to-measure concepts; nevertheless the power of numbers in
588 Simeen Mahmud

establishing patterns cannot be ignored. A large survey also allows us to

attribute effects more confidently since statistical tests are permitted. How-
ever, in order to minimize the grossness of quantitative measurement,
indicators have been fine-tuned by breaking down a dynamic process, like
income control or household decision-making, into discrete and more read-
ily quantifiable sub-processes. This increases validity and reduces subjectiv-
ity of indicators used by limiting the scope for interpretation. As discussed
above, the validity problem is not limited to quantitative data alone and can
even be a constraint for more qualitative ethnographic data.
Data used in this paper come from the first round household survey held
in early 1998 under a longitudinal study on the evaluation and monitoring
of microcredit programmes in rural Bangladesh (BIDS, 1998). The
study covered nearly 2000 households in ninety-one villages of twenty-two
rural thanas (sub-districts) of Bangladesh, of which eighty villages were
programme and eleven were non-programme villages. The villages were
randomly selected through multi-stage sampling from the universe of
villages covered by all the active partner organizations (POs) of the Palli
Karma Shahayak Foundation (PKSF), which is an apex body that provides
institutional and financial support to its POs for running microcredit
programmes.9 Non-programme villages were selected from among
neighbouring villages in order to maintain comparability in geographical
characteristics and infrastructure. In each selected village, households for
in-depth interviews were chosen to cover participating and non-participating
households from both poor and non-poor categories. The poor/non-poor
status of households was on the basis of household ‘eligibility’, the criterion
used by programmes to target poor households.10

9. Under the study, a complete enumeration of all households in 91 villages of 22 rural

thanas of Bangladesh was conducted at the end of 1997. The first stage was the selection
of 13 partner organizations (POs) on the basis of criteria set out by the PKSF, followed by
the selection of one to three thanas under each selected PO on the basis of NGO
programme diversity. The geographic coverage of the survey was spread fairly evenly
over Bangladesh, and thana-level comparisons revealed that selected thanas were not
different from the average (BIDS, 1998). The next stage was the selection of villages in
each selected thana, stratified according to presence of NGO activity (programme village)
or absence (non-programme village). In each PO area six programme and two non-
programme villages were selected after stratification according to the level of
infrastructure development. However, since non-programme villages could not be found
under some of the POs, only 11 non-programme villages could be included for the study.
In order to over sample participating and poor households and to ensure a minimum
number of households from each group, a non-proportionate random sampling technique
was followed. Proper weights were used to arrive at means for the sample categories and
for programme and non-programme villages.
10. As practised by most NGOs, households owning less than half an acre of arable land were
defined as ‘eligible’ for membership in an NGO programme. Participation was defined as
membership in an organization, which, at the least, undertook such activities as group
formation, savings mobilization, awareness building and supervised lending.
How Empowering is Microcredit? 589

The effect of participation in microcredit programmes on women’s

empowerment is assessed by comparing observed differences in the condi-
tion for empowerment, the route to empowerment and the achievements of
empowerment at the household level in each of the four categories accord-
ing to participation and poor/non-poor status, and at the aggregate level for
programme and non-programme villages. The comparison by participation
status is the primary focus of this analysis. Both household and individual
level indicators are used, where individual indicators are restricted to ever
married women aged fifteen years or more, either the wife of the head or the
female head. This restriction is imposed as a control for life cycle effects on
the process of empowerment, since senior women or wives of heads and
junior women or daughters and daughters-in-law in the same household will
have unequal power in intra-household processes.
The condition for empowerment is assessed from women’s access to
choice-enhancing resources. These are both material resources (education
and employment) and non-material resources (public domain and a favour-
able household attitude). While the effect of education and employment in
increasing women’s choice is fairly clear, the effect of access to the public
domain in expanding choice is more ambiguous (Kabeer, 2001). In Bangladesh
women’s mobility in the public domain, and consequently their access to
markets, information, services and employment opportunities, is strongly
circumscribed by norms of purdah.11 Although the public domain is male-
dominated, there are differential degrees of male dominance which deter-
mine the social acceptability of women’s mobility into specific locations.
Thus, the bazaar or the bank is less acceptable than the health centre or the
NGO office. This is different from the actual observance of purdah, the use
of a head and body cover when going outside the home, which is viewed by
some as a resource in terms of allowing women mobility outside the home.
The use of such a cover, however, does not relax the restrictions imposed
on women’s mobility into the male-dominated public domain. Therefore,
access to the public domain enhances choice for women regardless of
whether they actually use purdah or not. A household attitude that is
favourable to women also expands choice since this increases the likelihood
that women’s welfare is incorporated into household livelihood strategies.
Husband’s approval of family planning represents a favourable household
attitude because this increases welfare of women and children by reducing
the costs to women of using modern birth control methods.12

11. In this analysis purdah refers to the norms of women’s seclusion, which prescribe the
boundaries of women’s legitimate mobility into the public sphere comprising almost all
locations outside the homestead. Purdah is also commonly used to mean the actual veil or
covering some women use when moving in the public domain.
12. Husband’s approval of the use of modern birth control methods, although not essential,
reduces women’s ‘costs’ of using contraception to regulate fertility, especially in the case of
young women living in extended families.
590 Simeen Mahmud

The route to empowerment or the exercise of women’s agency in strategic

intra-household processes is more difficult to assess empirically, especially
with quantitative survey data. In this study, exercise of women’s agency is
indicated by an active role vis-à-vis men in two intra-household processes:
gaining access to household income and participating in decisions about
household expenditures. A distinction is made between an active role and an
independent role. Women’s active role is not necessarily an independent
one, and is possible even in a mutually dependent relationship with male
family members through joint control over income and co-operation in
decision-making on expenditures. Women’s independent role, on the other
hand, is evident when women make decisions alone and control income
entirely on their own.
Conventionally, women’s access to household income is greater with
respect to small incomes that women themselves earn (although sometimes
women may not have access even to their own incomes) compared to
income from male production, which is relatively much larger. In order to
assess women’s income control household income was disaggregated by
source into income from the sale of crops and income from the sale of
poultry and eggs, roughly corresponding to ‘male’ and ‘female’ incomes
respectively. Women’s agency in income control is greater when women
acquire access to male incomes.
Since access to household income raises the likelihood of influencing how
that income is spent, women have a smaller role in decisions about how
male incomes will be spent — that is, decisions in the male domain (produc-
tion outside the home) — and have a relatively greater role in decisions in
the female domain (production inside the home). Women’s access to male
incomes increases their role in ‘male’ decisions, which generally have greater
impact on family members’ well-being because male incomes are larger.
Consequently, women’s agency in household decision-making is greater
when women actively participate in decisions falling within the male
domain. For assessing women’s role in decision-making this process was
also disaggregated into male and female domains of decision-making by
distinguishing the type of expenditure. Crop production expenditure deci-
sions were located in the male domain, food consumption expenditure
decisions in the female domain, with decisions on children’s schooling fall-
ing in the common domain.
Given data limitations, this article only examines immediate achievements
of empowerment, that is, outcomes and behaviours that indicate increased
current welfare and/or reduced gender inequality in welfare. Current welfare
is indicated by the behaviours ‘time allocation’ and ‘fertility regulation’, and
by the outcomes ‘treatment for illness’ and ‘child immunization’. Since
women in rural areas have rigid patterns of time use any flexibility in their
time allocation, particularly a decline in household chores and childcare and
increase in time spent on income earning work, represents behaviour with
the potential to increase welfare. Since fertility regulation using modern
How Empowering is Microcredit? 591

birth control leads to improved survival prospects of children and better health
of women, contraceptive use also represents welfare augmenting behaviour.
Receiving treatment when ill and, in particular, receiving modern treatment
increases women’s absolute welfare, while immunization increases child welfare,
and consequently women’s welfare as well. Moreover, the sex ratios of pro-
portions of children immunized and proportions of household members
receiving any type of health care and modern health care when ill indicate
gender inequality in welfare. The variables used in the analysis to assess
empowerment at the different dimensions and their indicators are described in
Table 2.

Table 2. Variables
Dimension Variable Indicator

Condition for empowerment: Initial choice

Material resources Education Primary schooling (1–5 years)
Secondary schooling (6þ
years)
Paid employment Currently self employed
Currently wage employed
Non-material resources Male-dominated public Visited bank/bazaar
domain
Less male-dominated Visited NGO office/
public domain health centre
Favourable household Husband approves family
attitude planning
Route to empowerment: Agency in household processes
Access to Access to ‘female’ income Women keep income from
household income poultry rearing
Access to ‘male’ income Women keep income from
crop sale
Participation in household Active role in ‘female’ Joint or independent decision
decision-making decision maker for food expenditure
Active role in ‘male’ Joint or independent
decision decision-maker
for crop expenditure
Active role in ‘common’ Joint or independent decision-
decision maker for school expenditure
Achievement of empowerment: Improved relative and absolute welfare
Welfare-augmenting Fertility regulation Current use of contraceptive
behaviours
Time allocation in Time spent in household work,
home-based work childcare and income earning
Welfare outcomes Family member’s access Received any treatment
to health care by sex when ill
Received modern
treatment when ill
Children’s access to Children under 7 immunized
immunization by sex
592 Simeen Mahmud

RESULTS

Before comparing indicators of empowerment it is important to ascertain

whether the groups are indeed comparable with respect to basic socio-
economic characteristics. Table 3 presents the socio-economic profile of
ever-married women, wives or female heads, and their households by partici-
pation status, household poverty status and programme placement. Indi-
vidual variables used in the profile were: age, whether currently married,
number of children born alive, age of the youngest child and whether
literate (able to read and write). Socio-economic profile variables at the
household level were head’s literacy status, occupation and sex. Compari-
sons at the individual level always refer to ever-married wives of heads or
female heads.
Participant women were generally older and had more children compared
to non-participant women. In non-poor households participation was asso-
ciated with greater literacy of respondents and heads, but this was not the
case among poor households. Socio-economic differentials were, in fact,
more pronounced according to household poverty status than according
to participation status. For example, participant women in poor households
were younger and less likely to be literate, as well as more likely to belong to
labour-selling households and less likely to belong to farming households
compared to participant women in non-poor households. Socio-economic
profiles did not vary significantly by the presence of a programme in the
village.
Table 4 presents the distribution of respondents according to school
attendance, participation in paid employment, visit to the health centre or
NGO office, visit to the bank or bazaar, and husband’s approval of family
planning. It was found that about one third of rural ever-married women
had attended school at some time, but were far less likely to be employed,
particularly in wage work, or to have visited the bank or bazaar. Women’s
mobility in the public domain was not only very restricted but also nega-
tively associated with the degree of male dominance of the public space.13
The only exception was husband’s approval of family planning, which was
fairly common among all respondents, a reflection of the wide acceptance
of the use of modern birth control methods. Women in programme villages
were more likely to have visited the NGO office or health centre in the last
year and to have a husband who approved of family planning, but not more
likely to have visited the more restricted public domain of the bank and the
bazaar.
Women in poor households, regardless of participation status, were less
likely to have gone to school and more likely to be wage employed com-
pared to women in non-poor households. On the other hand, participant

13. Within the public sphere male dominance is greater in the bank and the bazaar compared
to the health centre and NGO office.
 How Empowering is Microcredit?
Table 3. Indicators of Socio-economic Status of Women Respondents (wife of household head or female head) and Sample
Households (averages and mean proportions)
Number of women/ Socio-economic status of women Socio-economic status of households
households
Age Prop. currently Parity Age of Prop. who Prop. head Prop. head Prop. head
(in years) married last child can read who can farmer manual
in months & write read & write labour selling

Programme Villages
Poor Households 1442
Participant 956 31.7* (8.01) 0.94 (.24) 3.3* (1.95) 29 (40.9) 0.18 (.38) 0.27 (.44) 0.15 (.36) 0.43 (.49)
Non-participant 486 29.5 (8.13) 0.92 (.27) 2.9 (2.04) 27 (34.6) 0.20 (.40) 0.29 (.45) 0.16 (.36) 0.41 (.49)
Non-poor 889
households
Participant 571 34.5* (8.54) 0.93 (.26) 3.8* (2.21) 35 (38.1) 0.35* (.48) 0.50* (.50) 0.58 (.49) 0.07 (.25)
Non-participant 318 32.9 (8.97) 0.92 (.29) 3.4 (2.23) 38 (73.5) 0.50 (.50) 0.59 (.49) 0.57 (.59) 0.06 (.23)
All households 2331 31.7 0.92 3.3 31 0.24 0.34 0.26 0.33
Non-programme Villages
All households 310 31.8 0.92 3.3 29 0.27 0.35 0.33 0.32

Notes:
Figures in brackets are standard deviations.
Means for all households in programme and control villages are weighted means.
*Significantly different from the mean value for non-participants in the same household category.

593
 594
Table 4. Distribution of Women Respondents according to Formal Schooling, Paid Employment, Visit to Locations in the Public
Domain and Husband’s Approval of Family Planning (%)
Per cent of women respondents Husband approves
use of contraceptives
With Formal Schooling In Paid Employment Visited in the last year

Primary Secondary Wage Employed Self Employed Thana or Bazar NGO office/Health Centre

Programme Villages
Poor Households 2331 2331 1889 2331
Participant .14 (.35) .15 (.36) .06 (.24) .20* (.40) .09 (.29) .75* (.43) .75* (.44)
Non-participant .12 (.32) .15 (.36) .07 (.25) .13 (.34) .12 (.33) .27 (.45) .67 (.47)
Non-poor households
Participant .21 (.41) .33* (.47) .02 (.14) .07* (.25) .10 (.30) .71* (.45) .73 (.44)
Non-participant .20 (.40) .43 (.50) .02 (.13) .03 (.16) .10 (.30) .26 (.44) .66 (.47)
All households .15 .21 .05 .15 .10 .60 .72
Non-programme Villages 310 310 255 310
All households .16 .20 .07 .14 .09 .22 .66

Notes:
Figures in brackets are standard deviations.

Simeen Mahmud
Figures in bold are the number of valid cases.
Means for all households in programme and control villages are weighted means.
*Significantly different from the mean value for non-participants in the same household category.
How Empowering is Microcredit? 595

women were more likely than non-participant women to be self-employed,

to have visited the health centre or NGO office and to have husband’s
approval of family planning in both poor and non-poor households.
Generally speaking, household poverty status was more strongly associated
with access to resources than programme participation.
Table 5 presents the distribution of households according to who keeps
income from different sources14 and confirms that men were more likely to
keep crop income while women were more likely to keep poultry income.
However, although poultry income was predominantly kept by women
(men kept poultry income in only 7 to 12 per cent of households), crop
income was kept by both men and women. In fact, the proportion of
households where crop income was kept by women was surprisingly large
across all households and villages, ranging from 36 per cent in programme
villages to 43 per cent in non-programme villages. Women were more likely
to keep crop income and less likely to keep poultry income in non-
programme compared to programme villages. Women in poor households,
irrespective of participation status, were more likely to keep household
income from all sources compared to women in non-poor households. In
the poor households participant and non-participant women were equally
likely to keep crop income and poultry income. In non-poor households
participant women were more likely than non-participant women to keep
crop income but not poultry income.
Table 5a presents the distribution of households according to gender roles
in household expenditure decisions. Participation of women in household
decisions conformed to the usual pattern of male and female domains of
decision-making.15 Women actively participated (joint or independent decision-
maker) most frequently in decisions on food expenditure, least frequently
in decisions on crop production expenditure and fairly commonly
in decisions on children’s school expenditure. There was joint decision-
making on food expenditure in nearly three-fourths of the households; in
crop production expenditure in almost one third of households; and in

14. Access to income was determined on the basis of the variable ‘who kept cash income from
that source’ and the response was categorized as ‘both’ when income was shared equally
by the head and spouse, as ‘men’ when the male head and/or other male family members
kept more than 50 per cent, and as ‘women’ when the head’s wife or the female head and/
or other female family members kept more than one half of the income.
15. In each case the decision-maker was categorized as ‘men’ when the male head and/or male
family members were the primary decision-maker, as ‘both’ when the decision was made
jointly by the male head and spouse, and as ‘women’ when the primary decision-maker
was the head’s wife or female head and/or other female family members. Women were
reported as having no role in household decision-making when the primary decision-
maker was ‘men’ and to have an active role when the primary decision-maker was
‘both’ or ‘women’. Independent decision-making by women was not very common, seen
in 8 per cent or less of households in programme villages and in 13 per cent or less of
households in non-programme villages.
596 Simeen Mahmud

Table 5. Distribution of Households according to Who Keeps Household

Income from Different Sources (%)
Who keeps household income from different sources

Crop sale Poultry or egg sale

Both Men Women Both Men Women

Programme Villages 1136 1384

Poor Households 390 830
Participant .10 (.30) .50 (.50) .40 (.49) .04 (.21) .06 (.23) .90 (.30)
Non-participant .06 (.50) .53 (.23) .41 (.49) .04 (.20) .09 (.28) .87 (.33)
Non-poor households 746 554
Participant .11 (.31) .53* (.50) .36* (.48) .08 (.27) .09 (.29) .83 (.30)
Non-participant .11 (.32) .61 (49) .28 (.45) .09 (.28) .08 (.26) .84 (.37)
All households .10 .54 .36 .06 .07 .88
Non-programme Villages 103 119
All households .07 .51 .43 .06 .12 .82

Notes:
Figures in brackets are standard deviations.
Figures in bold are the number of valid cases.
Means for all households in programme and control villages are weighted means.
*Significantly different from the mean value for non-participants in the same household category.

Table 5a. Distribution of Households according to Who Participates in

Household Expenditure Decisions (%)
Crop production Children’s schooling Food consumption

Both Men Women Both Men Women Both Men Women

Programme 1541 1573 2684

Villages
Poor Households 652 814 1580
Participant .30 .66 .04 .60* .32 .08 .77* .17 .06
(.46) (.47) (.20) (.49) (.47) (.26) (.42) (.38) (.24)
Non-participant .33 .62 .05 .50 .38 .11 .70 .21 .10
(.47) (.49) (.22) (.50) (.49) (.32) (.46) (.40) (.30)
Non-poor 889 759 1104
households
Participant .29 .68 .04 .49 .44 .07 .69* .25 .05*
(.45) (.47) (.19) (.50) (.50) (.26) (.46) (.44) (.23)
Non-participant .29 .65 .06 .50 .42 .08 .64 .27 .09
(.45) (.48) (.24) (.50) (.49) (.27) (.48) (.44) (.28)
All households .30 .66 .05 .55 .37 .08 .73 .20 .07
Non-programme 143 164 287
Villages
All households .27 .68 .05 .49 .38 .13 .70 .23 .07

Notes:
Figures in brackets are standard deviations.
Figures in bold are the number of valid cases.
Means for all households in programme and control villages are weighted means.
*Significantly different from the mean value for non-participants in the same household category.
How Empowering is Microcredit? 597

children’s school expenditure in a little more than half of households. Moreover,

women in programme villages had a joint role in decision-making more fre-
quently compared to women in non-programme villages for all three types of
decisions.
With the exception of decisions on children’s school expenditure,
women’s active participation in household decision-making was different
in poor and non-poor households. In decisions on crop production expend-
iture women in non-poor households were more likely to have an active
role compared to women in poor households, possibly because poor house-
holds engaged in crop production less often. The opposite was observed in
decisions on food expenditure. With respect to differences by programme
participation, participant women in poor households were more likely to
have an active role in decisions on children’s schooling and food consump-
tion expenditures compared to non-participant women. In non-poor house-
holds participant women were more likely to have an active role compared
to non-participant women only in food expenditure decisions.
Table 6 shows the distribution of respondents according to contraceptive
use and time allocation in home-based work. Current contraceptive use was
restricted to currently married women only. About half of all currently
married women were using modern contraceptives, and the extent of use
was similar in poor and non-poor households. Women in non-programme
villages were far less likely to be using contraceptives compared to women in
programme villages. Additionally, participants were significantly more
likely to be using contraceptives compared to non-participants in all house-
holds.
Women in programme villages were found to spend nearly nine hours per
day working inside the home, with three-fourths of the time spent in house-
hold work, about 16 per cent in childcare and 8 per cent in income-earning
work. Women residing in non-programme villages spent slightly more
time working inside the home compared to women residing in programme
villages, as well as proportionately less time in income-earning work and
proportionately more time in childcare. The pattern of women’s time use
varied according to both household poverty status and women’s programme
participation. On average, better-off women spent less time in income-earning
work inside the home, and consequently in total work. Participant
women spent proportionately more time in income-earning work regardless
of poverty status of household, while in poor households participants spent
less time in childcare compared to non-participants. Thus, participation was
associated with significant re-allocation of women’s home-based work.
Table 6a shows the distribution of households according to proportion of
sick members receiving any treatment and modern treatment and the pro-
portion of children aged seven years or less who were immunized, disag-
gregated by sex of family member. The possibility of receiving treatment,
either modern or any type, when sick depended more upon sex than house-
hold economic status. In all household categories male family members were
 598
Table 6. Distribution of Respondents according to Current Contraceptive Use and Time Allocation in Home-based Work
Per cent of current contraceptive user Time use pattern in home-based work (mean minutes/day)

All work Income earning work Household chores Child care

Min Min % Min % Min %

Programme Villages 2331 2257 2257 2257 2257

Poor Households
Participant .51* (.49) 555 56* 10 (124.4) 410 74 (170.6) 89* 16 (102.6)
Non-participant .43 (.45) 541 27 5 (100.9) 410 76 (170.9) 104 19 (101.6)
Non-poor households
Participant .52* (.50) 495 29* 8 (90.5) 395 78 (168.8) 71 14 (94.5)
Non-participant .42 (.49) 499 7 1 (47.9) 419 84 (176.6) 73 15 (89.6)
All households .48 536 39 7 409 76 88 16
Non-programme Villages 310 294 294 294 294
All households .38 569 20 4 426 75 123 21

Notes:
Figures in brackets are standard deviations.
Figures in bold are the number of valid cases.

Simeen Mahmud
Means for all households in programme and control villages are weighted means.
*Significantly different from the mean value for non-participants in the same household category.
 How Empowering is Microcredit?
Table 6a. Distribution of Households according to Sick Family Members Receiving Treatment and Children Immunized by Sex (%)
Per cent of sick members Per cent of sick members Per cent of households
receiving modern treatment receiving any treatment with children under 7 immunized1

female male f/m*100 female male f/m*100 female male f/m*100

Programme Villages 759 741 759 741 729 728

Poor Households
Participant .67 (.47) .76 (.43) 86 .78 (.41) .84 (.37) 93 .86 (.34) .88 (.32) 97
Non-participant .66 (.47) .81 (.39) 83 .76 (.43) .87 (.34) 87 .87 (.33) .90 (.29) 96
Non-poor households
Participant .72 (.45) .80* (.40) 92 .82 (.39) .87 (.34) 95 .88 (.31) .95 (.21) 92
Non-participant .64 (.48) .67 (.47) 100 .77 (.42) .86 (.34) 96 .88 (.32) .92 (.27) 96
All households .67 .76 88 .79 .85 93 .87 .90 96
Non-programme Villages 103 93 103 93 87 89
All households .61 .60 102 .74 .77 96 .76 .84 90

Notes:
Figures in brackets are standard deviations.
Figures in bold are the number of valid cases.
Means for all households in programme and control villages are weighted means.
1
Only DPT vaccination
*Significantly different from the mean value for non-participants in the same household category.

599
600 Simeen Mahmud

more likely to receive treatment than female family members when ill
(indicated by female/male ratios of less than 100). The male bias in receiving
treatment was slightly greater in poor households compared to non-poor
households; and in poor households the male bias was greater with respect
to modern treatment. This suggests that male bias in treatment is intensified
by household resource constraints, being greater in poor households and
with respect to costlier modern treatment.
In general, programme participation did not increase the likelihood of
receiving treatment when ill, except in non-poor households for male family
members receiving modern treatment. In poor households programme par-
ticipation was associated with higher female/male ratios or lower male bias
in the proportions receiving any treatment and modern treatment. The
likelihood of receiving treatment was generally lower for all family members
in non-programme villages compared to programme villages, while the male
bias in treatment received was also less pronounced.
The pattern of child immunization also varied more by sex of the child
than according to household economic status. There was a persistent small
male bias in the likelihood of children being immunized. This male bias was
slightly lower for programme participants in poor households, but could be
greater for participants in non-poor households. The overall extent of child
immunization, both girls and boys, was higher in programme villages com-
pared to non-programme villages, and the male bias was also lower.

DISCUSSION

Socio-economic status differentials among women respondents were more

strongly related to household poverty than to participation in a microcredit
programme. This was not unexpected because targeting of participants was
not strictly adhered to and women from both poor and non-poor house-
holds joined microcredit programmes. However, variation in the demo-
graphic status variables, age and parity, by participation status suggested
that programmes preferred slightly older women.16
The issue of selectivity bias or non-random differences between pro-
gramme participants and non-participants has long plagued impact assess-
ment studies, particularly when assessing the impact on women’s
empowerment (Pitt and Khandker, 1995; Zaman, 1996). An examination
of the condition for empowerment reveals whether there is any systematic
inclusion into microcredit programmes, to the extent this is observable, on
the basis of access to resources. Selectivity bias is indicated if participants
have relatively greater access to resources whose access is unlikely to be
influenced by programme participation, in this case formal schooling, wage

16. Respondent’s age was, in any case, not strongly correlated to the indicators used in the
analysis.
How Empowering is Microcredit? 601

work, ability to visit the bank or bazaar and husband’s approval of family
planning. A positive effect of participation on the condition for empower-
ment is indicated if participants have greater access to those resources whose
access is influenced by participation, namely self employment and visit to
the NGO office or health centre.
In non-poor households women with less schooling were more likely to join
a programme, while in poor households women having husbands who
approved of family planning were more likely to join a programme. In non-
poor households this does not pose a problem for assessing the effect of
programme participation on empowerment because the bias will, if anything,
tend to depress the small improvement in the condition for empowerment
associated with participation. In poor households, the bias may exaggerate
improvement in the condition for empowerment associated with participa-
tion. However, because husband’s approval of family planning is a very
common experience (observed for more than two thirds of respondent
women) the magnitude of its effect on women’s agency and on welfare
outcomes is unlikely to be very differentiated by participation status.
On balance, women in poor and non-poor households had comparable
initial conditions for empowerment. The evidence shows that microcredit
programme participation expands women’s access to resources like self-
employment and mobility into certain public spaces, like the NGO office
and health centre. This is hardly surprising given that programmes require
visits to the NGO office, encourage women’s self-employment and often
motivate participants to avail themselves of public health services. The
positive effect of participation on the condition for empowerment is, how-
ever, small since women’s access to more remunerative wage employment
and mobility into the male dominated public sphere is not increased.
In general, women’s access to household income was negatively related to
the degree of male involvement in income earning and positively related to
household poverty status. A positive programme placement effect on women’s
access to female household income was evident. Programme participation was
associated with women’s increased access to male income in poor households,
and to both male and female incomes in non-poor households. Thus, partici-
pation effect on women’s access to household income was greater in situations
where women were traditionally less likely to have access to household income,
that is, in non-poor households and in the case of male incomes.
The pattern of women’s participation in household expenditure decision-
making was less consistent than the pattern of access to household incomes.
Poor women were more likely to have an active role in some decisions (food
expenditure) while non-poor women were more likely to have an active role
in other types of decisions (crop production expenditure). Programme par-
ticipation was associated with women’s active role in household decision-
making, particularly in poor households and primarily with respect
to decisions in the ‘female’ or common domains. Programme placement
had a generally positive association with women’s active involvement in
602 Simeen Mahmud

household decision-making. Hence, programme participation enabled women

to exercise agency in household processes, and the effect of participation in
promoting women’s active role in household decision-making processes was
greater in situations where women are traditionally most subordinate to men
and least likely to exercise agency.
With regard to welfare-augmenting behaviour, programme participation
directly and presence of a programme in the village indirectly both had a
positive effect on women’s welfare by raising the probability of modern contra-
ceptive use. Participation also increased women’s welfare by allowing them to
re-allocate work time within the home, while the presence of a microcredit
programme in the village probably had a favourable effect on women’s time
use as well. With regard to welfare outcomes, although participation in micro-
credit programmes did not increase women’s access to health care generally, it
was likely to reduce male bias in access to health care and possibly in access to
child immunization as well. However, the effect of participation in reducing
male bias in immunization was less obvious than its effect in reducing male
bias in access to health care. Hence, the positive effect of participation in
microcredit programmes on women’s absolute and relative welfare is quite
strong. The positive links are most visible in the case of behaviours that imply
improved intra-household gender relationships, for example fertility regula-
tion and women’s time allocation. The links are less visible in the case of
outcomes in which a re-allocation of household resources is implied, for
example in women family members’ access to health care.

CONCLUSION

Improvement in women’s material resource base due to participation in micro-

credit programmes is small because increase in women’s resource access was
limited to resources that do not expand women’s choice a great deal. In rural
areas women’s self-employment has lower returns than wage employment and
mobility to the NGO office or health centre is less choice-enhancing than
mobility to the bank or bazaar. The positive effect of participation on women’s
agency is larger since there is a relatively greater increase in the likelihood that
women will actively engage in intra-household processes that determine wel-
fare. Both these empowering changes can be causally linked to the observed
achievements of empowerment. Thus, improvement in women’s access to
health care is relatively small because expansion in women’s material resource
base is small; on the other hand, improvement in fertility regulation and time
allocation is significant because increase in women’s agency is large.
Hence, the effect of participation in microcredit programmes on women’s
empowerment is not uniform across all three dimensions. Programme par-
ticipation has limited potential for improving the condition for empower-
ment since participation has little impact in altering gender-based access to
resources within the household and outside; participation has relatively
How Empowering is Microcredit? 603

greater potential for enhancing women’s ability to exercise agency in intra-

household processes. Programme participation is most likely to increase
women’s absolute and relative welfare through more egalitarian intra-
household gender relations. Thus, norms governing intra-household gender
relationships appear more responsive to programme participation, while
norms governing gender-based access to resources within the household
and outside are less responsive.
This is not surprising, because households have strong internal incentives
to improve intra-household gender relations as this produces outcome
whose benefits to the household outweigh the costs, for example lower
fertility through contraceptive use and greater time spent by women on
income earning through a changed pattern of women’s time use. On the
other hand, there is less incentive for re-allocation of resources in favour of
women, say with respect to investment in health and human capital, and it is
more costly because benefits to the household are limited by the gendered
nature of the external opportunity structure, which is determined by resili-
ent socio-economic institutions like marriage, purdah and inheritance.
It has been argued that measurement of programme effect on women’s
empowerment is distorted by self selection of relatively more empowered
women into microcredit programmes. Unfortunately, factors that may
influence women’s empowerment but not participation are not readily
observed, so that disproving this contention is not easy. However, this
analysis indicates that selectivity bias in programme membership with
respect to women’s empowerment is not as generally valid as it is made
out to be. Indeed, if loan use by male family members can be taken as an
indicator of ‘loss of control’ by women, as some have done, then case
studies of women borrowers in the first year would suggest that a very
high proportion of women who join microcredit programmes are actually
disempowered (Goetz and Sen Gupta, 1996). In fact, at a conceptual level it
is equally plausible that women who are more empowered are less likely to
agree to join a microcredit programme, which is usually part of a house-
hold’s livelihood strategy where men decide whether their wives will join a
programme or not. Recent research which has examined the household
decision-making process of whether to join a programme or not has
shown that women in unequal gender relationships within the household
are more likely to become members (Mahmud, 2000).17
Finally, there are some general observations that emerge from the map-
ping of rural women’s access to resources, exercise of agency and actual
outcomes. It is clear that empowering changes may not be manifest at all
dimensions and, even when they are, empowering effects may not be of
similar magnitude. This suggests that the process of empowerment is more

17. This research found that the household decision to join a programme is determined by the
interplay of the household’s demand for microcredit, the opportunity costs of membership
activities and the natures of intra-household gender relationships (Mahmud, 2000).
604 Simeen Mahmud

readily influenced at some dimensions than at others. This has program-

matic implications depending upon the objectives that a programme wishes
to achieve. If the programme objective is to raise absolute levels of women’s
welfare then increasing women’s access to male-dominated resources does
not appear to have immediate benefits. In that case expanding access to
resources may be less urgent than increasing women’s ability to exercise
choice within the existing structures of resource access and control. If, on
the other hand, the programme wishes to reduce male bias in welfare out-
comes then household resource allocation must be altered in favour of
women. In this case strengthening women’s resource base becomes import-
ant and programmes will have to address the issue of increasing women’s
access to male dominated resources and be willing to tackle the costs
associated with the erosion of long-standing male dominance of social and
economic institutions. For the sake of social sustainability microcredit and
other development programmes may very well be averse to imposing such
costs on their clientele, especially women.
The redeeming feature is that even without a significant increase in
women’s access to resources whose allocation is structurally determined, it
is possible to increase access to other choice-expanding but less restricted
resources and to enhance the exercise of women’s agency, both of which can
eventually be effective in transforming structures that restrict women’s
access to resources. In this respect a long-term and sustainable programme
strategy would be to promote the expansion of women’s access to household
resources, particularly income generated from loan investment, even when
these are used by men in family based enterprises. This is possible by
providing services that support women’s active involvement in decisions
about loan use and in the control over incomes from loan use.

REFERENCES

Amin, S. and A. Pebley (1994) ‘Gender Inequality within Households: The Impact of a
Women’s Development Programme in 36 Bangladeshi Villages’, Bangladesh Development
Studies 22(2&3): 121–54.
Bangladesh Institute of Development Studies (BIDS) (1998) ‘The Census Report of the BIDS
Study on PKSF’s Monitoring and Evaluation system’. Dhaka: Bangladesh Institute of
Development Studies.
Credit and Development Forum (CDF) (2002) ‘CDF Statistics: Microfinance Statistics of
NGOs and other MFIs’, Vol. 12. Dhaka: Credit and Development Forum.
Chen, M. and S. Mahmud (1995) ‘Assessing Change in Women’s Lives: A Conceptual Frame-
work’. Working Paper No 2, BRAC-ICDDR,B Joint Research Project at Matlab, Dhaka.
Goetz, A. M. and R. Sen Gupta (1996) ‘Who Takes the Credit? Gender, Power and Control over
Loan Use in Rural Credit Programmes in Bangladesh’, World Development 24(1): 45–63.
Hashemi, S. M., S. R. Schuler and A. P. Riley (1996) ‘Rural Credit Programmes and Women’s
Empowerment in Bangladesh’, World Development 24(4): 635–53.
Huda, S. and S. Mahmud (1998) ‘Women’s Control over Productive Assets: The Role of
Credit-based Development Interventions’. Working Paper No 28, BRAC-ICDDR,B Joint
Research Project at Matlab, Dhaka.
How Empowering is Microcredit? 605

Kabeer, N. (1998) ‘‘‘Money Can’t Buy Me Love’’? Re-evaluating Gender, Credit and Empow-
erment in Rural Bangladesh’. Discussion Paper 363, Institute of Development Studies.
Brighton: University of Sussex.
Kabeer, N. (1999) ‘The Conditions and Consequences of Choice: Reflections on the Measure-
ment of Women’s Empowerment’. UNRISD Discussion Paper DP108. Geneva: United
Nations Research Institute for Social Development.
Kabeer, N. (2001) ‘Conflicts over Credit: Re-evaluating the Empowerment Potential of Loans
to Women in Bangladesh’, World Development 29(1): 63–84.
van Koppen, B. and S. Mahmud (1996) Women and Water Pumps in Bangladesh. London:
Intermediate Technology Publications.
Mahmud, S. (1991) ‘Current Contraception among Programme Beneficiaries’, Bangladesh
Development Studies 18(3): 35–62.
Mahmud, S. (1994) ‘The Role of Women’s Employment Programmes in Influencing Fertility
Regulation in Rural Bangladesh’, Bangladesh Development Studies 22(2&3): 93–120.
Mahmud, S. (2000) ‘The Gender Dimensions of Programme Participation: Who joins a Micro-
credit Programme and Why?’, Bangladesh Development Studies 26(2&3): 79–102.
Mahmud, S. (2002) ‘Informal Women’s Groups in Rural Bangladesh: Group Operation and
Outcomes’, in F. Stewart, J. Heyer and R. Thorpe (eds) Group Behaviour and Development,
pp. 209–225. Oxford: Oxford University Press.
Mayoux, L. (1999) ‘Questioning Virtuous Spirals: Micro-Finance and Women’s Empowerment
in Africa’, Journal of International Development 1(7): 957–54.
Montgomery, R., D. Bhattacharya and D. Hulme (1996) ‘Credit for the Poor in Bangladesh’, in
D. Hulme and P. Mosley (eds) Finance against Poverty, pp. 94–176. London: Routledge.
Murdoch, J. (2002) ‘Micro-finance without Trade-offs’. Revised version of keynote address
delivered at the Third International Conference on Finance for Growth and Poverty
Reduction, University of Manchester, UK (10 April).
Pitt, M. and S. Khandker (1995) ‘Household and Intrahousehold Impacts of the Grameen
Bank and Similar Targeted Credit Programmes in Bangladesh’. Paper presented at work-
shop on ‘Credit Programmes for the Poor: Household and Intrahousehold Impacts and
Programme Sustainability’. World Bank and Bangladesh Institute of Development Studies,
Dhaka (19–21 May).
Rahman, A. (1999) ‘Microcredit Initiatives for Equitable and Sustainable Development: Who
Pays?’, World Development 27(1): 67–82.
Rahman, R. I. (1986) ‘Impact of Grameen Bank on the Situation of Poor Rural Women’.
Working Paper No 1, Grameen Bank Evaluation Project. Dhaka: Bangladesh Institute of
Development Studies.
Sen, A. K. (2000) Development as Freedom. Oxford and New York: Oxford University Press.
Steele, F., S. Amin and R. T. Naved (1998) ‘The Impact of an Integrated Microcredit Pro-
gramme on Women’s Empowerment and Fertility Behavior in Rural Bangladesh’. Policy
Research Division Working Paper No 115. New York: The Population Council.
Zaman, H. (1996) ‘Microcredit Programmes: Who Participates and to what Extent?’. Working
Paper No 12, BRAC-ICDDR,B Joint Research Project at Matlab, Dhaka.

Simeen Mahmud is Senior Research Fellow at the Bangladesh Institute of

Development Studies (E 17 Agargoan, Sher-e-Bangla Nagar, GPO Box
3854, Dhaka 1207, Bangladesh; e-mail simeen@sdnbd.org). She studied
Statistics at Dhaka University and Medical Demography at the London
School of Hygiene and Tropical Medicine. Her current research is on
children’s work and schooling, the transition of adolescents to adulthood
and on citizenship and community participation in the health sector.