Can Commercially-Oriented Microfinance Help Meet MDGs (Evidence From Pakistan)

World Development Vol. 39, No. 1, pp.
87109, 2011
2010 Elsevier Ltd. All rights reserved
0305-750X/$ - see front matter
www.elsevier.com/locate/worlddev
doi:10.1016/j.worlddev.2010.09.001
Can Commercially-oriented Micronance Help Meet the

Millennium Development Goals? Evidence from Pakistan
HEATHER MONTGOMERY
International Christian University, Tokyo, Japan
and
JOHN WEISS *
University of Bradford, UK
Summary. The current emphasis in the micronance industry is a shift from donor-funded to commercially sustainable operations.
This article evaluates the impact of access to microloans from the Khushhali BankPakistans rst and largest micronance bank which
operates on commercial principles. Using primary data from a detailed household survey of nearly 3,000 borrower and non-borrower
households, a dierence in dierence approach is used to test for the impact of access to loans. Once the results are disaggregated between rural and urban areas there is a positive impact in rural areas on food expenditure and on some social indicators such as the health
of children and female empowerment. These impacts are observed even in very poor households. These ndings suggest that commercially-oriented micronance and the millennium development goals are not incompatible, given a supportive environment.
2010 Elsevier Ltd. All rights reserved.
Key words Asia, Pakistan, micronance, poverty, impact, empowerment, Millennium Development Goals
1. INTRODUCTION
examines a large sample of MFIs covering a variety of institutional forms and lending methodologies to see how far any
dilution of poverty focus (as proxied by loan size) can be associated with the MFIs lending methodology, size, or age. The
authors conclude that for larger and older MFIs the results are
consistent with the view that as institutions mature and grow
they focus increasingly on clients that can absorb larger loans
(Cull et al., 2007, p. 131). This is not necessarily mission
drift since more poor borrowers could still be served under
a commercial model, but it is a warning that even NGO MFIs
may not be focusing primarily on the very poor any more.
Conducting detailed and rigorous impact studies is dicult
and as one review paper has commented: thirty years into
the micronance movement we have little solid evidence that
it improves the lives of clients in a measurable way (Roodman & Murdoch, 2009, pp. 34).
This paper contributes to this debate by focusing on a casestudy of a relatively new commercially-oriented micronance
bank. It reports the analysis of survey data collected in Pakistan in 2005 relating to the lending activity of the Khushhali
Bank, the rst licensed micronance bank in the country
and the one designed to operate on commercial lines but with
a social mission consistent with the MDGs, which in turn form
the center-piece of the governments poverty reduction strategy (Government of Pakistan, 2003). The paper is distinct in
The Millennium Development Goals (MDGs) reect ambitious development targets encompassing not only monetary
measures of poverty, but also more comprehensive measures
of development that quantify development in areas such as
education, gender equality, and health. 1 By providing small
scale nancial services to those on low incomes, micronance
is seen by many as one of the most signicant of the instruments to support these targets. Representing this hope, the
United Nations declared 2005 the International Year of
Microcredit with the goal of Building Inclusive Financial
Sectors to Achieve the Millennium Development Goals
(International Year of Microcredit, n.d.).
None the less, within the micronance industry it is recognized that micronance institutions (MFIs) that rely on aid
funding will be subject to the vagaries of aid budgets and will
never be able to expand to the scale that will make a major
change in poverty and in related social indicators at a national
level. The response has been to shift the focus of MFI activity
from donor-driven schemes that channel subsidized credit to
borrowers to commercially-oriented operations charging interest rates that cover full costs and are thus nancially self-reliant. This commercialization of micronance has prompted
considerable debate as to how far it is compatible with the original poverty reduction mission of MFIs (Montgomery &
Weiss, 2005). One commentator has gone so far as to refer
to a battle for the soul of micronance (Harford, 2008).
This question cannot be answered a priori and empirical evidence is needed to clarify the extent to which there is indeed
a trade-o between nancial sustainability and achievement
of the MDGs and related poverty targets. The trade-o can
be examined in various ways. For example, one approach assesses the impact on borrowers of interest rate increases that
accompany the shift to a commercially-oriented micronance
sector (Dehejia et al., 2008; Karlan & Zinman, 2008). Another
* Much of this research was completed while the authors were with the
Asian Development Bank Institute in Tokyo, Japan, which provided the
funding for the eldwork. Ziyodullo Parpiev provided outstanding research assistance. The authors thank Brett Coleman for helpful discussions
on sampling and application of the approach used in his 1999 study. They
also acknowledge the comments of several anonymous referees. The authors thank Farzana Nuzhat and Asim Anwar of Khushhali Bank for
facilitating focus group discussions with clients of the Bank. Final revision
accepted: August 26, 2010.
87
88
WORLD DEVELOPMENT
a number of ways: it is based on a large national survey of

2881 households, which is a far larger sample than is normal
for this type of study; it focuses on a country in which the
micronance sector is relatively new and where little rigorous
work has been conducted on the impact of micronance; it
applies a rigorous control group approach that addresses the
main sources of bias; and it uses a wide range of outcome
measures to capture the impact of lending on alternative
dimensions of welfare.
2. KHUSHHALI BANK OPERATIONS

Micronance remains relatively little developed in Pakistan,
although there has been a rapid growth since 2000 starting
from a low base. As of March 2008 there were estimated to
be around 1.6 million active borrowers with an average loan
size of Rs. 11,000 (less than $150) as compared with Rs.
460,000 in the banking sector as a whole. However, the potential market for microcredit is likely to be much larger than this
and the micronance strategy of the State Bank of Pakistan
targeted a total of 3 million borrowers by 2010. 2
The Khushhali bank was founded in 2000 as the rst initiative of the Micronance Development Program sponsored by
the Asian Development Bank. It was an early version of a
commercially-oriented micronance intervention with its share
capital drawn from 16 commercial banks. While operating
alongside more conventional aid funded MFIs, it has become
the largest provider of micronance in Pakistan, now providing a range of loan products to over 360,000 active borrowers
(nearly 25% of the national total), while focusing on the core
objective of operational self-suciency. It is not a perfect casestudy of a commercially-oriented micronance bank since
although its objectives are commercial it has not yet achieved
full nancial self-suciency. Even though its Annual Report
for 2007 reports a prot before tax of Rs. 156 million and a
return on equity of 5%, these calculations do not net out the
eect of the substantial interest rate subsidy from the Asian
Development Bank and independent estimates suggest that
the operational self-suciency ratio is no more than 80%. 3
The most recent Annual Report for 2008 reports a prot after
tax of Rs. 103 million but again this fails to remove the eect
of any nancial subsidy. Hence what we are examining is the
impact of the lending of a bank that is aiming for nancial viability but which is not yet operating on fully commercial lines.
As part of its social mission Khushhali targets clients who
are poor and very poor, but not those who are destitute
(living o charity, or zakat) or non-poor, who receive enough income to pay income tax. In the sample drawn for this
study in 2005, more than 70% of the clients were below the
ocial poverty line of the Government of Pakistan and 20%
were at less than half of the caloric consumption of those dened as poor. 4 When data for this study were collected in
2005, the bulk of clients (60%) were in rural areas and roughly
one-third were women, although according to the banks more
recent Annual Reports both of those ratios have declined as
the bank has expanded in the past few years.
At the time of the survey the bank oered eligible clients
uncollateralized micro-loans of Rs. 3,000Rs. 30,000. 5 The
rst loan would be between Rs. 3,00010,000 and loan sizes increase 20% with each cycle to a maximum of Rs. 30,000. The
terms of the microloan vary between 3 and 12 months, to be
repaid with interest on declining balances in equal monthly
installments or in one bullet payment, depending on the purpose of the loan. Loans were oered for investments in arable
agriculture, in livestock, or in micro-enterprises to establish a
new business or to purchase assets or working capital for an

existing business, but not for consumption. At the time of
the survey in 2005 the interest rate was 20% and the average
loan per borrower was $142. This was higher than that
charged by other micronance institutions and well above
the average lending rate of 11% reported in the IMF International Financial Statistics, however it is well below earlier estimates of the cost of borrowing from informal credit sources
(Arif, 1999).
Although the bank has introduced an individual scoring
report to screen and classify clients according to the above
eligibility criteria, it uses a group lending methodology under
which clients form groups called community organizations
that can be male, female, or mixed gender groups of between
3 and 25 members (usually 35 members in urban areas and
1025 in rural) who provide personal guarantees to each other.
Loans are made directly to individuals in the group, but if any
one member of the group defaults then all members of that
group become ineligible for loans.
3. RESEARCH METHODOLOGY
The rst question to be addressed in conducting any impact
evaluation is: how to measure impact? Most micronance programs are designed to encourage borrowers to invest their
loans in their farms or microenterprises. The hope is that these
investments will lead to higher prots, which will in turn lead
to higher household income, which will gradually lift client
households out of poverty. Micronance may also aect
non-income measures of household welfare such as health,
education, or empowerment. Note that the eects of micronance on these broader measures of welfare may not be expected to be all positive. On the one-hand, households with
more protable family farms or microenterprises may be able
to aord textbooks for their school age children, or hire workers to help with duties previously required for family members.
On the other hand, the extra time required for household
members in running a newly-protable family farm or microenterprise may also lead to lower enrollment rates for school
age children who are encouraged to skip school or drop out
all-together in order to contribute to the business.
In this study, we used the Khushhali Bank mandate of
. . .providing micro-nance services to poor persons, particularly poor
women for mitigating poverty and promoting social welfare and economic
justice through community building and social mobilization with the ultimate objective of poverty alleviation (Status and Nature of Business,
from Khushhali Bank Annual Report 2004)
as a guide in designing our study. Thus, we looked at what

might be called traditional, or at least, direct, impacts of
micronance on household business prots as well as broader
measures of household welfare such as health, education, female empowerment, and nally, poverty, as measured by food
and non-food consumption-expenditure.
Table 1 reports the summary statistics for the measures of
household welfare used in this study. Business prots for the
dierent types of activities supported by Khushhali Bank
loansagriculture and livestock farming as well as microenterpriseare measured by reported prots and sales. Consumption-expenditure, the basis for measuring ocial
poverty statistics in Pakistan, measured as food expenditures,
non-food expenditures, medical expenditures and educational
expenditures per child, indicates income eects of participation in the micronance program. Non-income measures of
household welfare are also included: the probability that children are in school, absenteeism from school, the probability of
CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?
89
Table 1. Summary statisticsdependent variables

Variable label
Obs
Mean
Std.Dev.
Min
Max
Income generating activitiesmicroenterprise

Sales
Prots-reported
2,881
2,881
37,437
13,540
109,191
45,040
0
0
1,024,000
700,000
Income generating activitieslivestock

Production/sales of livestock and products
Prots-reported from livestock
2,878
2,878
67,931
61,498
278,339
273,627
0
513,000
5,549,600
5,485,000
Income generating activitiesagriculture

Value of sales to third parties
2,881
24,453
76,306
1,345,000
Consumption-expenditure
Monthly consumption-expenditure per capita: food
Monthly consumption-expenditure per capita: non-food
Monthly medical expenditure per capita
Educational expenses per child
2,859
2,859
2,859
2,881
863
772
96
630
555
1,316
274
897
0
38
0
0
8,990
4,969
8333
11,900
Education
Probability children enrolled in school
Absenteeism from school
2,881
2,881
0.44
6.25
0.44
24
0
0
1
550
Empowerment
Opinions taken into consideration in household decisions regarding
Childs schooling
Childs marriage
Whether to have another child
Type of contraception to use
Womans participation in community/political activities
Womans decision to work outside home
Repair/construction of house
Sale/purchase of livestock
Borrowing money
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
0.58
0.39
0.23
0.22
0.12
0.18
0.34
0.27
0.36
0.49
0.48
0.42
0.41
0.32
0.39
0.47
0.44
0.48
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
Health
Spending on medical care these biases
Probability seek medical treatment if ill
Probability of medical treatment from trained practitioner if ill
2,881
2,881
2,881
6,834
0.60
0.57
23,841
0.48
0.48
0
0
0
900,000
1
1
Ability to pay for medical treatment from own sources

Probability seek medical treatment if child ill
Probability of medical treatment from trained practitioner if child ill
Probability children vaccinated
2,881
2,881
2,881
2,881
0.52
0.60
0.58
0.44
0.49
0.49
0.49
0.47
0
0
0
0
1
1
1
1
Note: Financial variables are measured in rupees.
seeking medical treatment if ill, and the quality of that treatment (as measured by the degree of training of the provider).
Given the mandate to serve women particularly, we also
looked at the impacts of participation in the micronance
program on the empowerment of women, as measured by
the extent to which female respondents in each household felt
their opinions were taken into account in household decisions
on things such as the childrens upbringing (their schooling
and marriage), nancial matters (whether or not to borrow
money, sell or purchase livestock, repair the home), and the female household members decision to work outside the home
or participate in community political activities.
Once the indicator of interest has been identied, a perfect
impact evaluation needs to answer a counterfactual question:
how does the status of participants in the program as
measured by those variables compared with how those same
individuals would have fared in the absence of the program?
The problem with cross-sections of data (observations on
many individuals at a given point in time) is that at any given
point of time, individuals are observed to be either participants or not. Even panels of data (observations on many
individuals through time) are problematic since over time
many other things have happened to the individuals in
addition to program participation and it is nearly impossible

to separate out the impact of the program from all the other
inuences. In practice, researchers must settle for estimates
of the average impact of the program on a group of participantsthe treatment groupcompared to a credible comparison groupa control group. The ideal control group is
individuals who would have had outcomes similar to those
in the treatment group if the members of the treatment group
had not participated in the program.
Constructing a control group comparable to the treatment
group is not straightforward. Participants in the program
are usually dierent from non-participants in many ways: programs may be carefully placed in specic areas, participants
within those areas may be screened for participation, and
the nal decision on whether or not to participate is usually
voluntary. To the extent that these factors are known and
can be measured, they can be controlled for in the empirical
analysis, but in most cases the placement of the program
and self-selection of participants in those areas into the program are based on unobservable factors. These unobservable
factors lead to at least two kinds of biases in any empirical
impact evaluation: program placement bias and self-selection
bias.
90
WORLD DEVELOPMENT
Controlling for these biasesdetermining the eects of

micronance alone and separating out the impact of microcredit from what would have happened to the same household
without creditis often the most dicult part of empirical
impact studies. Well-run micronance institutions do not
randomize either the location of their operations or their selection of clients. If MFIs tend to operate in areas that have relatively better or worse infrastructure such as access by roads
or more or less active markets, then estimates of the impacts
of the program on participants do not measure the eects just
of micronance, but of these other factors as well. Even within
a given village, if, as studies by Hashemi (1997, chap. 11),
Alexander (2001) and Coleman (2006) suggest, micronance
clients already have initial advantages over non-clients, then
the impact of micronance will be overestimated if these initial
biases are not controlled for. Similarly, the impact of
micronance programs that deliberately target relatively
disadvantaged households in the areas they operate may be
underestimated if these biases are not controlled for. 6
Armendariz de Aghion and Morduch (2005) provide a compelling argument in favor of making the substantial investment required to conduct careful impact studies that control
for these potential biases:
Unfortunately, this is not an esoteric concern that practitioners and policymakers can safely ignore. It is not just a dierence between obtaining
very good estimates of impacts versus perfect estimates the biases
can be large. In evaluating the Grameen Bank, for example, Signe-Mary
McKernan (2002) nds that not controlling for selection bias can lead to
overestimation of the eect of participation on prots by as much as 100
percent. In other cases . . .controlling for these biases reverses conclusions
about impacts entirely.
There are three broad approaches that have been applied

elsewhere in the micronance literature. One is the use of a
randomized study design to control for selection bias (Duo
& Kremer, 2005; World Bank, 2008; McKenzie, 2009). This
approach eliminates selection bias by randomly selecting treatment (those who receive micronance) and control (those who
do not) groups from a potential population of participants.
With this type of study design, the researcher can be assured
that on average those who are exposed to the program are
no dierent than those who are not and thus that a statistically
signicant dierence between the groups outcomes can be
condently attributed to the program rather than to selection
bias.
This randomized approach follows that adopted in clinical
drugs trials and is seen by many as the best way of rigorously proving impact. A few studies applying this approach
have been conducted or are ongoing. For example, Banerjee
et al. (2009) apply the approach to the expansion of MFIs
in slum areas in Hyderabad, India, nding that in the
short-term access to microcredit helps business start-up
and to fund investment. However, it appears to have no
impact on social indicators relating to female empowerment
or family health or education. The qualication is that these
are short-term eects and through higher investment future
monetary and non-monetary benets may arise. The authors
conclude that micronance can be a useful means of helping
the entrepreneurial poor, but it is not a miracle in the sense
of transforming social conditions. Interestingly, some initial
results from randomized work question key tenets of the
micronance literature. For example, a study on Sri Lanka
nds far higher returns to male-headed as opposed to female-headed micro-enterprises (De Mel, McKenzie, &
Woodru, 2008).
Well-designed randomized studies of this sort have the potential to rigorously address all kinds of potential biases, but
their overall acknowledged weakness is external validity

the ability to learn from an evaluation about how the specic
intervention will work in other settings and at larger scales
(Ravallion, 2009, p. 3). One limitation of small-scale randomized social experiments is that they can only estimate partial
equilibrium treatment eects, which may dier from general
equilibrium treatment eects. In the case of micronance, this
means that if, for example, micronance is introduced on a
large scale, the functioning of nancial markets may eventually be aected, thus yielding a dierent impact than the necessarily smaller-scale program introduced for the impact
study. There is also the potential for spillovers that create
external benets, for example from treated villages or areas
where micronance operates to untreated ones it is yet to
reach. There is thus the possibility that non-program participants may benet indirectly from the gains of participants
which will distort the impact assessment. The biggest problem
for external validity of randomized studies is related to the
issue of unobservable characteristics that aect both program
placement and self-selection. A randomized study will include
in the treatment group individuals with and without these
unobservable characteristics to get a truly unbiased measure
of impact. But when interventions such as micronance are
introduced on a large scale, the participants tend to have those
characteristics. Thus, despite their scientic appeal, randomized studies are not always the most relevant for policy makers
(Ravallion, 2009, p. 2).
Another, more practical, concern in attempting to apply
randomized study design is that such studies require tremendous cooperation from the institutions being evaluated,
which must be willing to allow researchers to randomize
the implementation of their services. Such studies are preferably longitudinal, making them costly, and it can be dicult
to conduct research over a period long enough for some impacts to show up. Intellectual arguments aside, for these
practical reasons the randomization approach was not applied in this study.
A simpler alternative is to identify a control group through
some identiable eligibility attribute. A well-known study by
Pitt and Khandker (1998) use land ownership. The authors
sample participants and non-participants of micronance
programs in a number of treatment villages where group lending programs are operating as well as randomly selected
households from control villages without a program. They
use village xed eects to correct for the endogeneity of
program placement and take advantage of the fact that the
microcredit programs impose eligibility requirements on participants (households with land holdings of more than half
an acre are ineligible) to determine eligible and ineligible
households in the control villages. Impact is assessed using a
dierence-in-dierence approach between eligible and ineligible households and between program and non-program villages. After controlling for other factors, such as various
household characteristics, any remaining dierence is attributed to the micronance programs. The diculty with this approach is that clear eligibility criteria for access to
micronance need to be identiable and strictly applied. 7
Another approach to controlling for self-selection and
placement bias is to include a sample of microcredit clients
who have formed solidarity groups but have not yet received
loans as the control group (Hulme & Mosley, 1996; Coleman,
1999, 2006). In this approach, participating and non-participating households are again surveyed in treatment villages
where the microcredit program is already operating and has
already given loans. The controls are villages where the
microcredit program will operate and households from the
village which have already self-selected to participate in the

program but have not yet actually received loans.
Hulme and Mosley (1996) employ this approach in their
major study of programs in a number of countries. However,
arguably their study fails to control for the major source of
biasprogram placement biasso part of the advantage of
program participants relative to the control group may be
due to unmeasured village attributes that aect both the supply and the demand for credit. Using the same basic approach,
Coleman (1999) addresses the placement bias issue by introducing observable village characteristics and village dummies
in his study of a village banking program in Thailand. Utilizing data on 455 households, including participating and nonparticipating households in treatment villages where a village
bank is already oering microcredit, and selecting future participants and non-participants in control villages that have
been identied to receive a village bank program but have
not yet actually received funds, Coleman (1999) uses a dierence-in-dierence approach that compares the dierence
between income for participants and non-participants in program villages with the same dierence in the control villages,
where the programs were introduced later. The rationale is
that the unobservable characteristics of bank members (such
as entrepreneurial drive) will be shared by members who have
taken out a loan and those who have applied but not yet
received their loan. Then, if other measurable household and
village characteristics can be controlled for, the remaining
dierence can be attributed to the receipt of the microcredit.
Chowdhury et al. (2005) use a similar approach in an analysis
of Bangladesh.
The nature of the Khushhali Banks operations lent itself to
an impact assessment using the latter approach, of taking
clients who had not yet accessed loans as the control group.
In 2005 the bank was expanding rapidly into new villages
and the number of active clients was increasing at a rate of
approximately 20,000 clients every 3 months. These were
mostly rural areas of Pakistan in which there were previously
no formal nancial services provided. Bank management and
sta were willing to cooperate with surveyors in identifying
new villages that had just received the service and within those
villages identifying new clients, allowing them to be surveyed
in the interim between their application and the approval to
get a microloan and the actual disbursement of the money.
Concerns about this approach to impact study design center
around program placement, selection, wealth eects, and attrition bias. Each of these is addressed in our study design. From
discussions with bank management, the sequence in which the
bank moved into villages did not appear to be driven systematically by an economic rationale, so that there was no deliberate attempt to provide services rst to villages with an
advantageous (or disadvantageous) location due to either
climate, geography or infrastructure links. Hence placement
bias does not appear to be a factor. However, as a check,
village dummies are included in the regression models to
account for unobserved village eects. 8 Selection criteria of
the bank were under review, but no changes had been implemented at the time of the survey and since the control group
of members who had not yet borrowed was drawn from
villages just receiving the service, there is no reason to think
that the self-selection criteria on the part of clients changed
systematically over time, either. Thus, we have no reason to
think the latent characteristics of our control group of borrowers are fundamentally dierent from those of the treatment
group. However, signicant for the external validity of this
study, we do admit that these unobservable characteristics in
both the treatment and the control group of bank members
91
may dier from the non-member population at large. There

was only a short time periodtypically just a few weeksbetween loan approval and disbursement and the survey was carried out during that brief interval. Thus, while we recognize
that a wealth eectthe fact that notication of approval
for a loan may lead to increased household consumption even
before actual receipt of the loanmay bias downward any impact ndings using this approach, the short time period in
question should limit this eect. Also, overall, there is no signicant relationship between bank membership and aggregate
expenditure, which also suggests that a wealth eect may not
present any serious problems in interpreting the results. Attrition biasthe fact that the control group of approved future
borrowers includes potential future dropouts or graduates of
the program, whereas the treatment group of older borrowers,
who have remained active, may not (Karlan, 2001) 9can be
ruled out here since the sample of clients analyzed here is unusual in that it includes clients from not only active groups, but
also from groups that are in default, currently inactive for
other reasons, or have completely dropped-out of the program. 10 Hence, while the control group of future borrowers
contains potential drop-outs and defaulters, so does the treatment group, eliminating attrition bias.
In summary, at the time of survey the Khushhali Bank was
expanding rapidly into new areas with many new members
who had self-selected to join the Bank, but not yet taken out
a loan. These circumstances provided a natural experiment
in the form of a control group for whom outcomes could be
compared with members who had already taken out and spent
a Khushhali loan. While the approach involves some implicit
assumptions, for example on the absence of a wealth eect and
on the unchanging nature of the unobservable characteristics
of the treatment and control groups, the short time periods
involved lessen their signicance, thus allowing a relatively
simple and practical means of assessing impact.
4. DATA AND REGRESSION FRAMEWORK
To conduct the empirical analysis, primary data were collected from 2,881 householdsmore than any other rigorous
impact study on micronance to date. A stratied random
sample of 1,454 Khushhali Bank clients and future clients
was drawn from 139 rural villages and 3 urban cities where
the bank operates. A roughly equal number (1,427) of
randomly selected non-clients from the same villages or
settlements were also surveyed. The survey covered 11 districts
across all provinces in Pakistan, including Kashmir.
At the time the sample was drawn, the bank was operating
in approximately 42 districts in Pakistan and had about
175,000 active clients, 37,000 of which were in the 11 districts
nally sampled. Thus, the sample represented more than a
quarter of the districts served by the bank and about 4% of
the clients in the selected districts, but less than 1% of the total
number of clients at that time. Roughly 40% of the sample of
client households were female clients, meaning that the female
in the household was the Khushhali Bank program member.
This slightly over-represents female clients in the sample since
at the time of the survey roughly 30% of Khushhali clients
were women. But the advantage to this slight oversampling
is that it may yield more robust estimates of gender issues
an important policy question for poverty reduction in Pakistanin the empirical analysis.
One-quarter (732 or 25%) of the total sample were from
urban areas. At the time of sampling, roughly 35% of the
surveyed areas and approximately 15% of the total population
92
WORLD DEVELOPMENT
of clients at that time were from urban areas. Loans for microenterprises were roughly one-third of the total, with the
remainder divided between livestock rearing and arable farming, with some overlap between these two categories. In each
household, both the male and the female head of household
were interviewed separately (regardless of the gender of the
borrower). Thus, data collection involved nearly 6,000 individual surveys. As stated above, client households surveyed, both
borrowers and those who had not yet borrowed were predominantly poor: 70% were living below the poverty line. Average
household consumption expenditure per capita in the sample
(including expenditure on food, consumer durables, heating,
rent, transport, health, and education) was less than $40 per
month in nominal terms (see Table 1) and average years of
education for the most well-educated household member were
less than six years for males and less than two and a half years
for females (see Table 2). The Appendix gives more details on
survey design and implementation.
As described above, the methodology involves a form of difference-in-dierence analysis comparing members of the bank
and non-members and, within the group of members, borrowers, and non-borrowers. Two sets of villages are compared
those where the Khushhali bank has already been in operation
and those where it has decided to operate but not yet commenced lending. Within these villages two sets of households
are compared, those who are not members of the bank and
those who are. Members in the old villages where the bank
is already in operation will have taken out loans (and hence
are borrowers), while members in the new villages, where
bank operations have not yet commenced will have been
approved for a loan, but not yet received it (hence they are
members but still non-borrowers). Members are assumed to
share unobservable characteristics that will inuence outcomes
in the absence of microcredit. However, by separating those
members who have received a loan from those who have
not, and controlling for household characteristics (like age,

education, literacy, family size, and so forth) and village xed
eects, the impact of microcredit alone on various outcome
indicators can be isolated.
Following this approach, the control group for existing borrowers is members in new villages, who have not yet actually
received their loan, but who are taken to share the unobservable characteristics of current borrowers. Impact is estimated
with a single equation:
Y ij b1 X ij b2 V j b3 M ij b4 M ij T ij eij
where Yij, is a vector of impact or outcome variables, Xij is a

vector of household characteristics, Vj represents village xed
eects, which control for observable and unobservable variables that may inuence program placement, Mij is a membership dummy variable equal to 1 for any household that is a
member of the bank and thus able to borrow under the program and Tij is a dummy variable for treatment, which reects
bank members who have already received a loan.
Thus, Tij takes a value of 1 for borrower households, who
have already taken out a loan, either currently or in the past.
This excludes the control group of new bank members who
have applied for and been approved to receive a loan, but have
not yet received the funds (for those households Tij is zero).
The treatment group of bank member households who have
already participated in the program would have received their
loan sometime between one and ve years ago. The most
recent entrants to the treatment group would have received
their loans one year from the survey date and successful clients
would therefore have fully repaid the loan. The oldest borrowers in the treatment group would have received their rst loan
as long as ve years before the survey date, when the bank
opened its rst branches. Tij does not capture the degree of
borrowing, so clients who borrowed the smallest loan size only
once (and perhaps dropped out after that) are included in the
Table 2. Summary statisticsindividual household characteristics

Variable label
Obs
Mean
Std.Dev.
Min
Max
Education of highest educated male (years)

Literacy of male
Numeracy of male
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
Total number males in HH
Education of highest educated female (years)
Literacy of female
Numeracy of female
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Total number females in HH
Children age 04
Children age 59
Children age 1015
Generations family in village
Number of relatives in village
Household member holding oce
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
2,881
5.76
1.38
1.68
0.52
0.51
0.43
0.31
0.20
0.19
2.18
2.41
0.64
0.82
0.51
0.47
0.43
0.30
0.15
0.33
3.78
0.98
1.19
1.10
1.70
43.71
0.16
5.09
1.41
1.45
0.76
0.76
0.61
0.48
0.40
0.40
1.38
3.94
1.04
1.15
0.75
0.68
0.57
0.47
0.36
0.56
1.93
1.11
1.23
1.21
1.40
59.73
0.36
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
20
9
9
4
5
4
3
2
2
11
16
8
8
5
4
3
3
2
4
14.5
9
8
9
3
600
1
treatment group in the same way as clients who joined the program ve years earlier and who took and successfully repaid
several loan cycles. This aggregation of borrowers does not inform us of the incremental impacts of borrowinghow much
more impact comes from another loan cycle or a larger loan
but since unobservable factors that inuence the degree of a
clients participation are also likely to inuence the outcome
variables of interest, measuring treatment with simple participation rather than the degree of participation provides the preferred estimate of impact purely from the micro-lending.
The hypothesis tested is whether access to the micronance
program of the Khushhali Bank has a positive eect on the
various outcome measures of welfare described above. Support for the hypothesis requires that the estimated coecient
b4 on the interaction term M T in Eqn. (1) is statistically
signicant (and, in most cases, positive).
In terms of outcome variables under vector Y our concern is
to test for the contribution of micronance to a broad set of
indicators that roughly correspond to the millennium development goals of eradicating extreme poverty and hunger
(MDG 1), achieving universal primary education (MDG 2),
promoting gender equality and empowering women (MDG
3) and reducing child mortality (MDG 4). This broad agenda
also corresponds to the goals set by the government of Pakistan for the bank. To this end we test for the programs impact on income variables, consumption-expenditure
variables, education variables, variables reecting aspects of
gender empowerment and healthespecially child health
variables.
Table 1 presents summary statistics on the dependent variables in vector Y. Income variables measured include the sales
and the prots reported for the households income generating
activities: microenterprise, livestock raising, or other agricultural activities.
The consumption-expenditure variables include food consumption, which is the basis for calculating the ocial poverty
line in Pakistan, and thus perhaps the most direct indicator of
the programs impact on poverty. Note that the poverty measure is thus an objective indicator. The survey did not attempt
to ascertain subjective or perception measures of poverty from
respondents. Educational objectives are measured by the
probability of school age children being enrolled in school
and absenteeism from school. Including measures of gender
empowerment is unusual in a quantitative study of this type.
The measures are derived from yes/no answers to questions
on whether the womans opinion is taken into account in a series of household decisions ranging from childrens schooling,
their work outside the home, use of contraception to the economic activity of the family. Health impacts are measured by
the probability of household membersadults and children
receiving treatment for illnesses and the quality of the treatment, as measured by the training of the provider. The survey
also collected information on whether children in the household were vaccinated and received treatment (ORS) for diarrheal disease, which kills more than 50,000 children under
the age of ve in Pakistan each year.
Table 2 sets out the summary statistics on the set of household characteristics used under vector X in Eqn. (1).
For most of the empirical analysis, ordinary least squares
analysis (OLS) is applied. Regressions in which the outcome
variable of interest was a yes/no dummy variable on qualitative information, logit estimation technique is used. In addition to the aggregate results on the entire sample,
heterogeneity is explored by splitting the sample into urban
and rural households and exploring dierential eects on
households in which the borrower is the male or female head
93
of household. Given the potential for the very poor or core

poor to be left behind by micronance, particularly where,
as in this case, commercial interest rates are charged, we also
test for whether the relationship between access to funds and
the various outcome indicators diers for borrowers in the
bottom quintile as compared with the sample average.
5. RESULTS
(a) Monetary indicators
The rst millennium development goal is to eradicate
extreme poverty and hunger, with some of the specic targets
being to reduce the proportion of the population living on less
than $1 a day and increasing the proportion of own-account
and contributing family workers in total employment. The
immediate objective of most micro-lendingto provide a
source of nancial intermediation for micro-enterprise and
small farmersis perhaps most directly related to these targets. Thus, we begin our discussion with the programs impact
on monetary indicators of welfare.
Table 3a reports the impact of program participation on
sales or prots of the income generating activity run by the
householdlivestock/animal raising, arable agricultural activities, or micro-enterprise. For the full sample the impact of
borrowing under the program is not evident in either sales
or prots from livestock/animal raising activities or from arable farming. At the time the sample was drawn, around twothirds of loans were for this type of activity. Neither is it evident for the sales or prots of microenterprises. Additional
variables on number of loan cycles, size of loans, and number
of months since taking a loannot reported hereinare also
insignicant in most cases. This result suggests no income effect from taking out loans.
In addition to the analysis of income eects above, the impact on expenditure of drawing on Khushhali bank loans is
also examined. The results for four separate expenditure variables are reported in Table 4a. The rst outcome variable,
monthly consumption per capita, looks at the impact of the
program on caloric consumption as measured by consumption-expenditure on food items. The items used in calculating
this variable correspond as closely as possible to the items used
by the government of Pakistan in calculating the ocial poverty line. Separately, three other expenditure outcome variables are used; monthly per capita consumption of non-food
items (heating, household durables, clothing, transport, and
so forth), monthly per capita expenditure on health care,
and annual educational expenditure per child in the household.
In general, there is no evidence of a quantitatively signicant
impact of program participation on poverty as indicated by
consumption-expenditure. There is no statistically signicant
relationship between program participation (M T, or treatment, in Eqn. (1)) and aggregate consumption expenditure on
either food or non-food expenditures. Participation in the program also has no statistically signicant inuence on household educational expenditures per child. Not surprisingly,
educational expenditures per child are determined principally
by the education level of the highest-educated male and female
in the household. Nonetheless in one specic area there is evidence that borrowing under the program has had an impact
on expenditure. Table 4a reports a positive impact on health
expenditures, as indicated by the statistically signicant positive coecient estimate in column 3. Relative to the control
group households borrowing under the program tend to have
94
WORLD DEVELOPMENT
Table 3A. Income generating activities
Education of highest educated

Literacy of female
Numeracy of female
Literacy of male
Numeracy of male
Family generations in village
Household member holding o
Children age 04
Children age 59
Children age 1015
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
Kushaili Bank client
1
OLS
Sales of livestock
and products
2
OLS
Prots from livestock
and products
3
OLS
Agricultural sales
to third parties
4
OLS
Sales of
microenterprise
5
OLS
Prots from
microenterprise
8037.24
[5299.45]
16079.04
[26518.75]
373.87
[21493.53]
374.88
[3703.89]
34912.72
[23431.75]
49729.04**
[22995.09]
22621.43
[14050.53]
97140.27**
[39007.15]
34.9
[248.32]
42691.59**
[14218.26]
4114.45
[12497.39]
5942.89
[12224.27]
9111.58
[21251.12]
14139.35
[24816.33]
18058.68
[31075.80]
33096.96
[36120.24]
11723.19
[41069.01]
17252.13
[26736.64]
23761.67
[24705.39]
20641.02
[25029.46]
29463.27
[29283.60]
67116.44*
[35461.86]
60074.16
[40515.53]
6357.66
[37895.99]
11360.84
[51252.12]
7648.64
[5287.66]
16898.63
[26459.77]
1089.66
[21445.73]
495.33
[3695.66]
33577.21
[23379.64]
46850.80**
[22943.95]
22881.38
[14019.28]
93757.29**
[38920.40]
15.41
[247.76]
41480.78***
[14186.64]
4195.07
[12469.60]
5542.95
[12197.09]
9604.08
[21203.86]
12870.64
[24761.14]
16462.14
[31006.69]
32329.28
[36039.90]
10546.64
[40977.68]
16814.30
[26677.18]
23372.43
[24650.44]
20019.49
[24973.80]
28459.80
[29218.47]
66535.77*
[35383.00]
60115.56
[40425.43]
8204.87
[37811.71]
12001.04
[51138.14]
1715.33***
[541.60]
8389.67***
[2710.19]
9329.90***
[2196.62]
54.1
[378.53]
529.82
[2394.70]
7033.08***
[2350.07]
3523.30**
[1435.95]
18761.91***
[3986.49]
47.19
[25.38]*
4978.10***
[1453.09]
1984.22
[1277.22]
849.36
[1249.31]
3503.41
[2171.84]
5186.28**
[2536.20]
9256.75***
[3175.91]
11653.38***
[3691.45]
6223.90
[4197.21]
1265.34
[2732.46]
4436.85*
[2524.86]
5482.66**
[2557.98]
2954.73
[2992.75]
5290.36
[3624.16]
6978.30*
[4140.64]
765.51
[3872.93]
14929.48***
[5237.91]
1895.70**
[792.70]
1920.28
[3931.67]
2206.16
[3222.45]
186.35
[550.67]
2893.08
[3473.62]
7433.76**
[3418.93]
6366.44***
[2087.08]
12410.47**
[5782.82]
54.37
[37.34]
1881.59
[2121.00]
1821.97
[1852.22]
5903.80***
[1810.53]
5987.14*
[3151.55]
5620.76
[3674.96]
1274.31
[4601.20]
4197.08
[5348.43]
3166.86
[6081.74]
3884.98
[3958.82]
5749.95
[3657.79]
1837.71
[3706.75]
1646.90
[4343.33]
1717.20
[5270.72]
77.44
[6011.15]
15562.28***
[5612.45]
12113.51
[7596.04]
12746.88
[26814.68]
990.95***
[327.72]
1556.98
[1625.43]
706.91
[1332.23]
467.08**
[227.66]
1789.02
[1436.07]
777.11
[1413.46]
820.91
[862.84]
2972.30
[2390.74]
5.09
[15.44]
1012.85
[876.87]
159.04
[765.74]
2371.32***
[748.51]
1652.97
[1302.91]
2402.94
[1519.30]
304.92
[1902.23]
165.93
[2211.15]
178.67
[2514.32]
1587.88
[1636.66]
265.05
[1512.20]
1287.39
[1532.45]
937.47
[1795.62]
1342.02
[2179.02]
4585.35*
[2485.13]
4130.90*
[2320.30]
2026.50
[3140.36]
15131.12
[11085.74]
39248.24
[52020.57]
37989.37
[51904.88]
4076.68
[5316.45]
17568.54
[53186.66]
1,678
12041.60
[53068.37]
1,678
30411.01***
[5435.62]
1,226
9926.26
[8110.92]
49282.35***
[9192.36]
1091.67
[35655.47]
837
2104.70
[3353.22]
14862.33***
[3800.31]
9090.50
[14740.70]
837
Urban client
Accessed loans
Urban Accessed loans
Constant
Observations
Notes: Standard errors in brackets.
*
Signicant at 10%.
**
Signicant at 5%.
***
Signicant at 1 %.
95
Table 3B. Income generating activitiesonly URBAN

Literacy of female
Numeracy of female
Literacy of male
Numeracy of male
Children age 04
Children age 59
Children age 1015
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
Accessed loans
Constant
Observations
1
OLS
Sales of livestock
and products
2
OLS
Prots from livestock
and products
3
OLS
Sales of
microenterprise
4
OLS
Prots from
microenterprise
87.28
[113.385]
976.16
[807.720]
545.88
[755.515]
50.18
[94.560]
1671.75***
[620.496]
668.30
[727.819]
141.11
[419.064]
231.61
[363.960]
262.64
[336.089]
624.34
[583.554]
225.39
[695.763]
226.14
[905.684]
78.84
[1003.365]
950.24
[1062.244]
63.81
[752.587]
804.29
[740.746]
869.09
[739.961]
790.25
[902.532]
690.77
[1062.751]
1781.97
[1136.122]
383.07
[1100.593]
92.90
[1233.405]
89.90
[100.635]
743.18
[716.896]
23.72
[670.561]
9.75
[83.927]
1068.55*
[550.724]
588.43
[645.979]
143.74
[371.942]
417.33
[323.035]
84.60
[298.297]
208.35
[517.936]
597.99
[617.528]
576.71
[803.844]
372.92
[890.541]
1141.98
[942.799]
207.11
[667.962]
365.40
[657.453]
128.75
[656.756]
288.04
[801.046]
438.18
[943.249]
1579.80
[1008.370]
319.67
[976.836]
134.99
[1094.715]
174.20
[1766.442]
5258.45
[12583.648]
1452.80
[11770.331]
996.76
[1473.175]
4007.14
[9666.849]
22060.76*
[11338.845]
2431.93
[6528.681]
9579.41*
[5670.218]
7324.89
[5236.003]
1554.65
[9091.317]
1081.98
[10839.445]
17178.14
[14109.844]
2115.29
[15631.642]
4061.57
[16548.921]
11796.64
[11724.719]
11674.70
[11540.248]
8376.52
[11528.011]
4802.94
[14060.740]
17317.28
[16556.821]
10833.39
[17699.892]
42437.31**
[17146.371]
19930.09
[19215.490]
1198.42
[776.705]
2483.58
[5,533.032]
5591.14
[5175.416]
1596.90**
[647.755]
1680.80
[4250.515]
727.96
[4985.692]
3553.47
[2870.662]
2393.44
[2493.196]
5623.96**
[2302.271]
1410.51
[3997.453]
4713.34
[4766.105]
5915.01
[6204.100]
1824.06
[6873.235]
4628.76
[7276.563]
2764.26
[5155.361]
973.93
[5074.249]
2790.76
[5068.868]
1472.38
[6182.509]
3288.98
[7280.037]
236.10
[7782.645]
4450.65
[7539.262]
5028.76
[8449.054]
1255.26
[1255.775]
492.13
[2424.923]
732
1083.91
[1114.569]
193.66
[2152.251]
732
34326.88*
[19563.991]
31796.02
[37778.393]
732
10621.15
[8602.290]
2103.95
[16611.165]
732

Exclude sales to third parties b/c urban.
*
Signicant at 10%.
**
Signicant at 5%.
***
Signicant at 1%.
on average Rs. 41 per month higher per capita expenditures on

health care. Although this is small in absolute terms, since the
banks program does nothing explicit to encourage health
awareness per se (and loans are supposed to be used for investment purposes), this suggests that poor households either
prefer to or need to use any extra income generated by

micro-credit on health care rather than on education or food
consumption.
These results indicate a clear lack of impact of borrowing
on income and most aspects of expenditure. The survey
96
WORLD DEVELOPMENT
Table 3C. Income generating activitiesonly RURAL

Literacy of female
Numeracy of female
Literacy of male
Numeracy of male
Household member holding o
Children age 04
Children age 59
Children age 1015
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
Accessed loans
Constant
Observations
*
Signicant at 10%.
**
Signicant at 5%.
***
Signicant at 1%.
1
OLS
Sales of livestock
and products
2
OLS
Prots from
livestock and products
3
OLS
Agricultural sales
to third parties
4
OLS
Sales of
microenterprise
5
OLS
Prots from
microenterprise
9419.95
[7942.303]
32134.68
[35415.703]
14547.74
[27311.915]
1741.18
[5057.976]
38656.89
[31348.861]
54141.21*
[29890.054]
27053.32
[16466.208]
90081.52**
[45644.387]
102.29
[292.503]
46307.70**
[18577.592]
7599.28
[16194.742]
6628.74
[16197.718]
8826.23
[28408.265]
20370.87
[32736.816]
17602.55
[40536.339]
40446.30
[47565.889]
9471.83
[55276.697]
22939.30
[34981.050]
31707.26
[32399.999]
18642.52
[32923.789]
26077.97
[37750.868]
73013.59
[46212.358]
70594.37
[53721.688]
8884.02
[49510.294]
16341.25
[70785.996]
9113.81
[7927.350]
32067.00
[35349.024]
15166.33
[27260.493]
1841.57
[5048.453]
37632.09
[31289.839]
51239.00*
[29833.778]
26893.84
[16435.206]
87473.48*
[45558.450]
109.81
[291.952]
45385.14**
[18542.615]
7463.37
[16164.251]
6341.14
[16167.222]
9633.36
[28354.779]
18753.25
[32675.181]
16111.77
[40460.019]
39727.06
[47476.334]
8412.80
[55172.625]
22421.52
[34915.189]
31312.91
[32338.997]
18120.90
[32861.801]
25225.66
[37679.792]
73181.26
[46125.351]
71653.79
[53620.543]
11253.32
[49417.078]
17679.95
[70652.723]
1800.99**
[794.324]
5320.53
[3541.988]
6628.24**
[2731.513]
379.34
[505.857]
1625.59
[3135.255]
5954.78**
[2989.358]
5053.44***
[1646.815]
15802.90***
[4564.976]
0.73
[29.254]
3611.66*
[1857.978]
1610.28
[1619.665]
291.13
[1619.963]
5275.96*
[2841.161]
6943.17**
[3274.067]
9847.48**
[4054.111]
13488.12***
[4757.149]
4951.83
[5528.321]
1308.88
[3498.517]
5136.59
[3240.382]
5953.34*
[3292.767]
5230.14
[3775.532]
10128.99**
[4621.780]
12801.33**
[5372.802]
972.77
[4951.613]
21518.26***
[7079.434]
3266.46***
[858.810]
133.33
[3829.540]
3061.67
[2953.268]
203.03
[546.925]
3867.75
[3389.788]
5690.36*
[3232.045]
5940.59***
[1780.510]
12616.81**
[4935.579]
54.21*
[31.629]
1883.26
[2008.816]
88.25
[1751.156]
5139.78***
[1751.478]
6308.55**
[3071.818]
6784.02*
[3539.869]
5056.57
[4383.240]
4213.81
[5143.353]
1220.85
[5977.132]
1300.51
[3782.540]
5314.17
[3503.448]
3428.13
[3560.086]
2137.60
[4082.044]
1703.29
[4996.995]
3324.26
[5808.987]
9102.23*
[5353.604]
10971.24
[7654.170]
1094.18***
[333.490]
1517.94
[1487.074]
166.64
[1146.803]
81.72
[212.380]
2556.28*
[1316.311]
896.23
[1255.057]
875.52
[691.401]
3058.17
[1916.567]
8.20
[12.282]
520.77
[780.056]
283.86
[680.003]
1373.41**
[680.128]
1310.39
[1192.838]
3944.55***
[1374.590]
835.40
[1702.085]
1891.61
[1997.249]
1248.59
[2321.019]
2458.63*
[1468.823]
89.79
[1360.447]
741.39
[1382.440]
974.62
[1585.125]
855.08
[1940.416]
4673.32**
[2255.726]
3460.19*
[2078.893]
2348.89
[2972.241]
47312.25
[71763.927]
722392.60**
[307437.703]
2,149
46522.68
[71628.813]
661284.70**
[306858.873]
2,149
8781.92
[7177.238]
218669.85***
[30747.392]
2,149
8791.19
[7759.915]
31034.11
[33243.586]
2,149
2477.09
[3013.303]
5641.70
[12909.034]
2,149

Table 4A. Indicators of poverty: consumptionexpenditure
Education of highest educated female

Literacy of female
Numeracy of female
Education of highest educated male
Literacy of male
Numeracy of male
Children age 04
Children age 59
Children age 1015
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
Kushaili Bank client (0/1)
Accessed loans
Constant
Observations
*
Signicant at 10%.
**
Signicant at 5%.
***
Signicant at 1%.
1
OLS
Monthly expenditure
per capitafood
2
OLS
Monthly expenditure
per capitanonfood
3
OLS
Monthly expenditure
per capitahealth care
4
OLS
Expenditure on
education (per child)
0.52
[3.86]
35.54*
[19.14]
23.1
[15.50]
3.9
[2.69]
23.36
[16.93]
35.2**
[16.59]
19.39*
[10.17]
45.08
[28.11]
1.2***
[0.18]
58.37***
[10.25]
60.11***
[9.01]
39.89***
[8.82]
65.82***
[15.36]
30.77*
[17.90]
10.07
[22.46]
69.83***
[26.15]
96.3***
[29.66]
77.5***
[19.27]
96.54***
[17.82]
29.67
[18.07]
16.85
[21.20]
12.01
[25.71]
34.22
[29.34]
-61.09**
[27.35]
30.66
[37.07]
13.55
[9.73]
50.72
[48.30]
39.84
[39.12]
27.97***
[6.78]
33.06
[42.71]
16.6
[41.87]
2.17
[25.66]
238.12***
[70.93]
0.9**
[0.46]
34.44
[25.86]
52.77**
[22.74]
17.86
[22.27]
73.7*
[38.76]
61.04
[45.16]
69.38
[56.68]
79.48
[65.99]
7.57
[74.84]
89.4*
[48.64]
136.14***
[44.97]
67.68
[45.59]
23.66
[53.51]
16.86
[64.89]
112.63
[74.05]
67.12
[69.01]
95.43
[93.54]
0.34
[2.07]
15.23
[10.27]
20.1**
[8.32]
2.82*
[1.44]
3.01
[9.08]
6.26
[8.90]
0.53
[5.46]
11.71
[15.09]
0.26***
[0.10]
2.45
[5.50]
7.92
[4.84]
4.21
[4.74]
1.88
[8.24]
11.63
[9.60]
14.95
[12.05]
17.76
[14.04]
31.74**
[15.92]
14.62
[10.34]
9.03
[9.56]
5.01
[9.70]
3.49
[11.38]
0.8
[13.80]
21.29
[15.75]
3.29
[14.68]
48.64**
[19.89]
44.61***
[6.08]
43.94
[30.43]
31.52
[24.66]
35.54***
[4.25]
38.31
[26.89]
17.57
[26.38]
13.5
[16.12]
12.64
[44.76]
0.55*
[0.28]
18.42
[16.31]
43.47***
[14.34]
134.53***
[14.03]
18.27
[24.38]
42.82
[28.47]
28.58
[35.66]
111.73***
[41.44]
17.61
[47.12]
21.99
[30.68]
79.49***
[28.35]
78.86***
[28.72]
26.93
[33.60]
50.41
[40.69]
15
[46.49]
85.61**
[43.48]
43.81
[58.81]
52.14
[37.63]
1191.18***
[38.67]
2,876
25.2
[94.96]
1062.19***
[97.59]
2,876
40.7**
[20.20]
111.43***
[20.75]
2,876
4.8
[59.69]
199.09***
[61.03]
2,876
97
98
WORLD DEVELOPMENT
Table 4B. Indicators of poverty: consumption- expenditureonly URBAN

Literacy of female
Numeracy of female
Literacy of male
Numeracy of male
Children age 04
Children age 59
Children age 1015
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
Accessed loans
Constant
Observations
1
OLS
Monthly expenditure
per capitafood
2
OLS
Monthly expenditure
per capitanon-food
3
OLS
Monthly expenditure
per capitahealth care
4
OLS
Expenditure on
5.58
[5.234]
34.12
[36.555]
4.94
[34.151]
5.02
[4.363]
65.85**
[28.424]
24.24
[33.082]
59.89***
[18.999]
47.34***
[16.464]
37.75**
[15.304]
75.88***
[26.676]
32.05
[31.543]
1.20
[41.216]
56.50
[45.733]
70.97
[48.471]
68.53**
[34.109]
139.02***
[33.600]
71.24**
[33.571]
51.72
[41.262]
6.00
[48.777]
17.21
[52.073]
170.47***
[50.149]
16.23
[56.553]
25.03**
[12.669]
35.03
[88.478]
40.81
[82.661]
31.98***
[10.559]
77.62
[68.798]
83.63
[80.074]
50.64
[45.985]
48.01
[39.849]
19.36
[37.043]
54.03
[64.566]
57.36
[76.348]
60.68
[99.760]
64.42
[110.693]
65.54
[117.321]
124.43
[82.558]
152.98*
[81.326]
82.63
[81.256]
92.40
[99.873]
99.24
[118.061]
77.99
[126.038]
30.18
[121.381]
96.51
[136.883]
1.97
[3.674]
54.75**
[25.658]
45.88*
[23.971]
4.04
[3.062]
29.14
[19.951]
12.64
[23.220]
7.82
[13.335]
7.29
[11.556]
9.15
[10.742]
12.48
[18.723]
8.61
[22.140]
28.72
[28.929]
62.52*
[32.100]
68.58**
[34.022]
8.21
[23.941]
14.05
[23.583]
5.33
[23.563]
6.26
[28.962]
14.79
[34.236]
47.61
[36.550]
41.73
[35.199]
40.57
[39.694]
38.79***
[12.313]
8.89
[87.714]
17.70
[82.045]
16.70
[10.269]
60.15
[67.382]
21.35
[79.037]
57.30
[45.508]
118.78***
[39.524]
139.85***
[36.497]
48.93
[63.371]
62.74
[75.556]
96.71
[98.352]
142.14
[108.960]
86.75
[115.354]
28.43
[81.727]
195.96**
[80.441]
81.05
[80.356]
52.37
[98.010]
122.61
[115.409]
18.17
[123.376]
274.43**
[119.518]
48.50
[133.941]
17.15
[57.598]
969.72***
[111.001]
720
99.95
[139.411]
1631.97***
[268.670]
720
36.28
[40.427]
224.83***
[77.911]
720
196.63
[136.370]
35.60
[263.333]
720

*
Signicant at 10%.
**
Signicant at 5%.
***
Signicant at 1%.
distinguishes between households (in both treatment and

control groups) in rural and urban locations (on the basis of
the location of the Khushhali bank oces) with about onequarter of borrowers from urban areas and the remainder
from rural. Since borrowers may face dierent constraints
and use the loans in dierent ways in rural and urban settings,
the analysis sub-divides the sample into urban and rural

households. Tables 3b and c report the urban and rural results,
respectively, for income generating activities and Tables 4b
and c do the same for expenditure. In the analysis of sales
and prots, there is a very little change with the exception that
there is statistically weak (at the 10% level) positive eect of

Table 4C. Indicators of poverty: consumption-expenditureonly RURAL

Literacy of female
Numeracy of female
Literacy of male
Numeracy of male
Children age 04
Children age 59
Children age 1015
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
Accessed loans
Constant
Observations
*
Signicant at 10%.
**
Signicant at 5%.
***
Signicant at 1%.
1
OLS
Monthly expenditure
per capitafood
2
OLS
Monthly expenditure
per capitanon-food
3
OLS
Monthly expenditure
per capitaHealth care
4
OLS
Expenditure on
0.79
[5.198]
19.60
[23.133]
10.71
[17.855]
0.99
[3.308]
4.40
[20.477]
20.67
[19.512]
13.31
[10.800]
31.80
[29.805]
1.40***
[0.193]
46.14***
[12.128]
60.10***
[10.579]
35.98***
[10.583]
66.65***
[18.547]
35.58*
[21.380]
9.66
[26.511]
70.15**
[31.186]
96.12***
[36.087]
81.93***
[22.839]
91.01***
[21.167]
24.63
[21.519]
21.17
[24.727]
9.83
[30.293]
66.00*
[35.200]
37.35
[32.331]
59.91
[46.198]
8.43
[13.335]
39.59
[59.351]
55.84
[45.809]
25.42***
[8.488]
79.43
[52.537]
17.37
[50.060]
1.15
[27.709]
232.15***
[76.469]
0.90*
[0.494]
32.15
[31.117]
53.98**
[27.141]
25.28
[27.152]
72.91
[47.585]
48.67
[54.852]
67.00
[68.018]
82.12
[80.010]
40.46
[92.585]
73.95
[58.596]
125.56**
[54.307]
58.22
[55.210]
4.59
[63.441]
5.19
[77.721]
142.82
[90.308]
73.20
[82.949]
195.97*
[118.527]
0.12
[2.603]
28.54**
[11.586]
26.69***
[8.943]
3.01*
[1.657]
6.15
[10.256]
0.16
[9.772]
0.15
[5.409]
13.48
[14.928]
0.22**
[0.096]
0.36
[6.074]
8.54
[5.298]
9.25*
[5.300]
1.21
[9.289]
16.67
[10.708]
12.14
[13.278]
5.54
[15.619]
22.22
[18.074]
21.30*
[11.439]
13.64
[10.602]
5.86
[10.778]
3.97
[12.385]
1.22
[15.172]
21.90
[17.630]
0.37
[16.193]
40.65*
[23.138]
44.95***
[7.119]
52.36*
[31.746]
15.80
[24.482]
38.29***
[4.534]
28.36
[28.100]
15.53
[26.792]
13.96
[14.760]
2.20
[40.914]
0.76***
[0.262]
1.67
[16.652]
27.58*
[14.516]
128.51***
[14.519]
7.55
[25.464]
35.19
[29.344]
13.28
[36.335]
121.94***
[42.637]
32.90
[49.548]
18.98
[31.356]
44.77
[29.042]
88.87***
[29.512]
15.44
[33.839]
13.39
[41.423]
21.01
[48.154]
45.81
[44.379]
90.11
[63.450]
94.56**
[46.843]
22.13
[201.073]
2,139
100.30
[120.180]
924.80*
[515.876]
2,139
30.37
[23.461]
315.25***
[100.707]
2,139
83.36
[64.327]
615.56**
[275.577]
2,149
99
100
Table 5. Indicators of poverty: education and health
Numeracy of
male
Family
generations in
village
Household
member holding
oce
Number of
relatives in
village
Children age
04
Children age
59
Children age
1015
Female age
1621
Female age
2229
Female age
3039
2
OLS
Education: Days
Children Absent
from School
3
LOGIT
Health: Probability
seek medical
treatment if ill
4
LOGIT
Health: Probability
medical treatment from
trained professional
5
LOGIT
Health: Able to pay
for medical treatment
from own assets
6
LOGIT
Health:
Probability
seek medical
treatment
if child ill
7
LOGIT
Health:
Probability
medical
treatment from
trained
professional
if child ill
8
LOGIT
Health:
Probability
take ORS
to treat diarrhea
9
LOGIT
Health:
Probability
children
vaccinated
0.06***
[0.02]
0.26
[0.17]
0.1***
[0.02]
0.08***
[0.02]
0.08***
[0.02]
0.05**
[0.02]
0.04**
[0.02]
0.07**
[0.03]
0.02
[0.02]
0.04
[0.10]
0.08
[0.08]
0.09***
[0.01]
0.37
[0.87]
1.68**
[0.71]
0.19
[0.12]
0.44***
[0.11]
0.75***
[0.10]
0
[0.01]
0.12
[0.09]
0.4***
[0.08]
0.01
[0.01]
0.31***
[0.09]
0.57***
[0.08]
0
[0.01]
0.17*
[0.10]
0.15*
[0.08]
0.02
[0.01]
0.04
[0.10]
0.02
[0.08]
0.02
[0.01]
0.04
[0.11]
0.01
[0.09]
0.02
[0.02]
0.14
[0.12]
0.02
[0.09]
0.02*
[0.01]
0.25***
[0.08]
0.13
[0.08]
0.05
[0.05]
3.4***
[0.77]
3.66***
[0.75]
0.49
[0.46]
0.34***
[0.08]
0.41***
[0.08]
0.05
[0.04]
0.27***
[0.08]
0.32***
[0.07]
0.09**
[0.04]
0.14*
[0.07]
0.23***
[0.07]
0.1**
[0.04]
0.08
[0.09]
0.07
[0.08]
0.05
[0.05]
0.08
[0.09]
0.04
[0.08]
0.1**
[0.05]
0
[0.10]
0.2*
[0.11]
0.1
[0.06]
0.3***
[0.10]
0.23**
[0.10]
0.04
[0.06]
0.19
[0.13]
2.52**
[1.28]
0.31**
[0.12]
0.3**
[0.12]
0.27**
[0.12]
0.09
[0.14]
0.1
[0.14]
0.31*
[0.17]
0.07
[0.15]
0***
[0.00]
0.02**
[0.01]
0***
[0.00]
0***
[0.00]
0***
[0.00]
0***
[0.00]
0***
[0.00]
0
[0.00]
0***
[0.00]
0.07
[0.05]
0.64***
[0.05]
0.77***
[0.05]
0.04
[0.07]
0.19**
[0.09]
0.06
[0.11]
0.1
[0.47]
3.23***
[0.41]
1.41***
[0.40]
0.97
[0.70]
1.54*
[0.81]
1.26
[1.02]
0.11**
[0.04]
0.18***
[0.04]
0.1**
[0.04]
0.03
[0.07]
0.02
[0.08]
0.02
[0.10]
0.08*
[0.04]
0.17***
[0.04]
0.08**
[0.04]
0.03
[0.06]
0.02
[0.07]
0.01
[0.09]
0.05
[0.04]
0.15***
[0.04]
0.08**
[0.04]
0.1
[0.06]
0
[0.07]
0.04
[0.09]
1.22***
[0.06]
0.29***
[0.04]
0.05
[0.04]
0.09
[0.08]
0.11
[0.09]
0.02
[0.11]
1.17***
[0.06]
0.29***
[0.04]
0.08*
[0.04]
0.09
[0.08]
0.14
[0.09]
0.05
[0.11]
0.54***
[0.06]
0.25***
[0.05]
0.18***
[0.06]
0.08
[0.10]
0.11
[0.12]
0.15
[0.14]
1.92***
[0.08]
0.37***
[0.05]
0.01
[0.05]
0.2**
[0.09]
0
[0.10]
0.25**
[0.12]
WORLD DEVELOPMENT
Education of
highest educated
female
Literacy of
female
Numeracy of
female
Education of
highest educated
male
Literacy of male
1
LOGIT
Education:
Probability
Children School
0.09
[0.13]
0.29**
[0.14]
0.07
[0.09]
0.16*
[0.09]
0.36***
[0.09]
0.03
[0.11]
0.04
[0.13]
0.12
[0.14]
0.16
[0.13]
0.4**
[0.18]
Implied odds
ratio
Accessed
loans
Implied
odds ratio
Constant
1.5
Observations
2.84**
[1.19]
1.64
[1.35]
0.77
[0.88]
0.42
[0.81]
0.35
[0.82]
1.8*
[0.96]
1.98*
[1.16]
1.04
[1.33]
0.4
[1.24]
1.68
[1.68]
0.24**
[0.11]
0.41***
[0.13]
0.31***
[0.08]
0.13*
[0.07]
0.03
[0.08]
0.14
[0.09]
0.28**
[0.11]
0.1
[0.12]
0.13
[0.12]
0.15
[0.16]
0.23**
[0.11]
0.41***
[0.13]
0.29***
[0.08]
0.13*
[0.07]
0
[0.07]
0.12
[0.09]
0.28***
[0.11]
0.05
[0.12]
0.13
[0.11]
0.14
[0.15]
0.16
[0.11]
0.26**
[0.12]
0.18**
[0.08]
0.14*
[0.07]
0.03
[0.07]
0.13
[0.09]
0.26**
[0.11]
0.05
[0.12]
0.06
[0.11]
0.24
[0.15]
0.35***
[0.13]
0.18
[0.15]
0.02
[0.09]
0.2**
[0.09]
0.06
[0.09]
0.29***
[0.11]
0.26**
[0.13]
0.13
[0.15]
0.25*
[0.14]
0.24
[0.18]
0.32**
[0.13]
0.2
[0.15]
0.09
[0.09]
0.17*
[0.09]
0.02
[0.09]
0.22**
[0.11]
0.18
[0.13]
0.05
[0.15]
0.11
[0.14]
0.21
[0.18]
0.03
[0.18]
0.06
[0.20]
0.2*
[0.11]
0.24*
[0.13]
0.21
[0.13]
0.3**
[0.15]
0.31*
[0.17]
0.23
[0.21]
0.16
[0.18]
0.16
[0.24]
0.33**
[0.14]
0.03
[0.16]
0.08
[0.11]
0.27***
[0.10]
0.07
[0.10]
0.02
[0.11]
0.27*
[0.14]
0.16
[0.16]
0.24
[0.15]
0.03
[0.20]
1.16
1.14
1.27
0.79
0.81
1.17
0.98
0.21
[0.18]
1.5
1.06
[1.71]
0.01
[0.16]
1.07
0
[0.15]
1.06
0.15
[0.15]
0.94
0.45**
[0.19]
1.47
0.45**
[0.19]
1.44
0.08
[0.25]
0.92
0.14
[0.20]
0.98
1.81***
[0.19]
5.43***
[1.75]
0.21
[0.16]
0.13
[0.16]
0.44***
[0.16]
1.87***
[0.20]
1.86***
[0.20]
3.14***
[0.27]
1.7***
[0.22]
2,876
2,876
2,876
2,876
2,876
2,876
2,876
2,876
2,876

*
Signicant at 10%.
**
Signicant at 5%.
***
Signicant at 1%.
Female age
4049
Female age
5059
Female age
over 60
Male age
1621
Male age
2229
Male age
3039
Male age
4049
Male age
5059
Male age
over 60
Kushaili Bank
client (0/1)
101
102
WORLD DEVELOPMENT
Table 6. Female empowerment
1
2
LOGIT LOGIT
Childs
Childs
schooling marriage
Education of highest
educated female
Literacy of female
Numeracy of female
Education of highest
educated male
Literacy of male
Numeracy of male
Family generations in
village
Household member
holding oce
Number of relatives in
village
Children age 04
Children age 59
Children age 1015
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
Khushaili Bank client
(0/1)
Implied odds ratio
Female Khushhali Bank
client (0/1)
0.02
[0.02]
0.27***
[0.08]
0.3***
[0.07]
0.02*
[0.01]
0.08
[0.07]
0.08
[0.07]
0.05
[0.04]
0.26**
[0.12]
0**
[0.00]
0.05
[0.04]
0.2***
[0.04]
0.1***
[0.04]
0.23***
[0.06]
0.28***
[0.08]
0.32***
[0.09]
0.11
[0.11]
0.67***
[0.13]
0.22***
[0.08]
0.19**
[0.08]
0.25***
[0.08]
0.22**
[0.09]
0.21**
[0.11]
0.09
[0.12]
0.15
[0.11]
0.37*
[0.19]
0.69
0.22
0.01
[0.02]
0.03
[0.08]
0.09
[0.07]
0.01
[0.01]
0
[0.08]
0
[0.08]
0.05
[0.05]
0.32**
[0.12]
0***
[0.00]
0.01
[0.05]
0.09**
[0.04]
0.05
[0.04]
0.39***
[0.07]
0.35***
[0.08]
0.35***
[0.10]
0.42***
[0.11]
0.11
[0.14]
0.21**
[0.09]
0.03
[0.08]
0.1
[0.08]
0.19*
[0.10]
0.29**
[0.12]
0.02
[0.13]
0.04
[0.12]
0.38*
[0.23]
0.69
0.45
3
4
5
6
7
8
LOGIT
LOGIT
LOGIT
LOGIT
LOGIT
LOGIT
Whether
RepairSale-purchase Borrowing
Womans
Womans decision
to have
construction of livestock
money
participation in
to work
another child
of house
community-political
outside home
activities
0.01
[0.02]
0.39***
[0.12]
0.32***
[0.11]
0.04***
[0.01]
0.08
[0.11]
0.09
[0.10]
0.06
[0.06]
0.3*
[0.16]
0***
[0.00]
0.27***
[0.06]
0.05
[0.05]
0.02
[0.05]
0.15*
[0.09]
0.05
[0.11]
0.34***
[0.13]
0.09
[0.15]
0.73***
[0.20]
0.27**
[0.11]
0.17
[0.11]
0.15
[0.12]
0.19
[0.13]
0.32**
[0.15]
0.06
[0.17]
0.23
[0.17]
0.57*
[0.32]
0.57
0.01
0.07***
[0.02]
0.14
[0.09]
0.08
[0.07]
0.02*
[0.01]
0.1
[0.08]
0.38***
[0.09]
0.05
[0.05]
0.51***
[0.13]
0**
[0.00]
0.1*
[0.05]
0.12***
[0.04]
0
[0.04]
0.12
[0.08]
0.04
[0.09]
0.16
[0.11]
0.14
[0.13]
0.43***
[0.16]
0.27***
[0.10]
0.25**
[0.10]
0.25**
[0.10]
0.48***
[0.12]
0.52***
[0.14]
0.32**
[0.15]
0.38***
[0.15]
0.46*
[0.27]
0.63
0.36
taking out a loan on the sales of urban micro-enterprises

(Table 3b, column 3). On the expenditure side, however, there
is one signicant notable eect, which is that in rural areas,
taking out a loan is positively associated with higher food
0.12***
[0.02]
0.32***
[0.10]
0.05
[0.08]
0.02
[0.01]
0.36***
[0.09]
0.47***
[0.09]
0.14***
[0.05]
0.46***
[0.14]
0
[0.00]
0.2***
[0.05]
0.13***
[0.05]
0.02
[0.05]
0.15*
[0.08]
0.17*
[0.10]
0.35***
[0.12]
0.08
[0.14]
0.44**
[0.18]
0.34***
[0.11]
0.12
[0.10]
0.12
[0.11]
0.51***
[0.13]
0.48***
[0.15]
0.52***
[0.17]
0.44***
[0.16]
0.79**
[0.31]
0.45
0.12
0.11***
[0.02]
0.24***
[0.09]
0.08
[0.07]
0.02
[0.01]
0.28***
[0.08]
0.45***
[0.08]
0.07
[0.05]
0.57***
[0.13]
0
[0.00]
0.15***
[0.05]
0.13***
[0.04]
0.01
[0.04]
0.15**
[0.07]
0.14
[0.09]
0.24**
[0.11]
0.04
[0.13]
0.49***
[0.16]
0.24**
[0.10]
0.25***
[0.10]
0.23**
[0.10]
0.48***
[0.12]
0.64***
[0.13]
0.52***
[0.15]
0.46***
[0.14]
0.55**
[0.27]
0.58
0.96**
0.1***
[0.03]
0.65***
[0.16]
0.26*
[0.15]
0.01
[0.02]
0.43***
[0.16]
0.41**
[0.16]
0.01
[0.07]
0.06
[0.21]
0.01***
[0.00]
0.13*
[0.08]
0.16**
[0.07]
0.17**
[0.07]
0.12
[0.12]
0.18
[0.15]
0.08
[0.17]
0.4*
[0.21]
0.29
[0.25]
0.38**
[0.16]
0.12
[0.14]
0.08
[0.15]
0.07
[0.17]
0.32
[0.20]
0.09
[0.23]
0.4*
[0.24]
0.3
[0.40]
0.74
1.16
0.01
[0.02]
0.39***
[0.13]
0.21*
[0.12]
0.01
[0.02]
0.37***
[0.13]
0.31**
[0.13]
0.06
[0.07]
0.11
[0.18]
0***
[0.00]
0.13**
[0.06]
0.1*
[0.06]
0.07
[0.06]
0.04
[0.10]
0.02
[0.12]
0
[0.15]
0.22
[0.17]
0.51**
[0.21]
0.41***
[0.13]
0.04
[0.12]
0.13
[0.12]
0.16
[0.14]
0.12
[0.17]
0.11
[0.18]
0.31
[0.19]
0.17
[0.32]
0.85
0.6
expenditure (Table 4c, column 1). In contrast, the positive

impact on household health expenditure found in the full
sample does not show up in the analysis of the rural and urban
sub-samples.
103
Table 6Continued
1
2
LOGIT LOGIT
Childs
Childs
schooling marriage
Implied odds ratio

Accessed loans
Implied odds ratio
Female accessed loans
Implied odds ratio
Constant
Observations
3
4
5
6
7
8
LOGIT
LOGIT
LOGIT
LOGIT
LOGIT
LOGIT
Whether
RepairSale-purchase Borrowing
Womans
Womans decision
to have
construction of livestock
money
participation in
to work
another child
of house
community-political
outside home
activities
[0.30]
1.25
[0.34]
1.57
[0.51]
0.99
[0.40]
1.43
[0.57]
0.88
[0.38]
2.62
[1.08]
0.31
[0.60]
0.55
0.49**
[0.20]
1.63
0.13
[0.33]
1.14
0.98***
[0.16]
0.64***
[0.24]
1.90
0.18
[0.36]
0.83
1.5***
[0.17]
0.62*
[0.33]
1.86
0.4
[0.53]
1.49
1.7***
[0.22]
0.62**
[0.28]
1.85
0.05
[0.43]
1.05
1.48***
[0.20]
0.84***
[0.32]
2.31
0.8
[0.59]
2.22
3.25***
[0.27]
0.67**
[0.28]
1.95
0.07
[0.40]
0.93
1.82***
[0.20]
0.19
[0.42]
1.21
1.94*
[1.10]
6.96
3.54***
[0.34]
0.25
[0.33]
1.29
1.25**
[0.63]
3.47
1.7***
[0.24]
2,876
2,876
2,876
2,876
2,876
2,876
2,876
2,876

* Signicant at 10%.
** Signicant at 5%.
*** Signicant at 1%.
These results are disappointing for those seeking evidence of

a positive impact of lending by the Khushhali bank on household income. There is little evidence of a direct impact on
income generation, the main intention of the bank. Considering broader monetary indicators, positive impacts are found
for health and food expenditures, although the former result
is not robust and the latter applies only to rural areas. These
results may trouble purists who hold that micronance should
be granted for investment purposes only, not for consumption
smoothing to meet short-term household expenditures on food
and health. A more exible view of the role of micronance
would allow these expenditures to be considered as part of
household nancial planning where fungible resources are
allocated in line with household needs.
Of those households who seek treatment for their childrens

illnesses, participating households are one and a half times
more likely than a not yet borrowing household to turn to a
trained medical professional for that treatment.
As with the analysis of the monetary indicators, the sample
is further divided by urban and rural households. The positive
impact of the program on health indicators seems to be primarily enjoyed by rural households. For urban households
no signicant results are obtained for the key education and
health indicators, while for rural households the results from
the general sample on the greater probability of seeking medical treatment for children and on the use of a trained health
professional associated with taking out a loan are conrmed. 11
(b) Social indicators
(c) Female empowerment
Expenditure-based poverty measures can omit important

non-monetary dimensions of welfare, as reected in the millennium development goals to achieve universal primary education, empower women, and reduce child mortality. Tables 5
and 6 look at the impact of microcredit on non-monetary
social indicators relating to education, health, and female
empowerment.
Turning rst to Table 5, there is little evidence of a positive
impact of the program overall on education. Although children in Khushhali Bank member households are more likely
to be enrolled in school than children in non-member households, children in households already participating by taking
out loans are no more likely to be enrolled in school and have
the same rates of absenteeism as children in households that
are members of the bank but have not yet taken out loans.
For health indicators, however, in line with the expenditure
results there is evidence that households already participating
in the program are more likely to get medical treatment for
their childrens illnesses, and that the treatment is more likely
to come from a trained medical professional, as compared
with non-borrowing members (see Table 5, columns 6 and
7). The odds ratios from the logit regressions suggest that
borrowing households are about one and a half times more
likely to seek medical treatment when their children are ill.
An important aspect of the study design is to test whether

borrowing under the program has an impact on female
empowerment, as gauged through responses to questions on
female involvement in family decision-making, political activity, and work outside the home. The survey asks married
females between the ages of 15 and 40 to grade answers to
questions on their involvement in the various family decisions
on a scale of 16 ranging from always to never and
answers are grouped into yes/no categories to allow the application of logit analysis. In analyzing female empowerment, we
explore heterogeneity in the sample across households in
which the bank member and borrower was male or female.
These results are reported in Table 6.
Table 6 conrms that, as expected, female Khushhali bank
members are already more empowered than non-members,
as evidenced by the highly signicant positive parameter
estimates on the Female Khushhali Bank client dummy in
specications looking at the role of females in decisions on
number of children, repair of property and borrowing money.
However, unexpectedly, wives of male bank members tend to
be less empowered than women in non-member households in
a few areasthe decision on number of children and on
property repairas evidenced by the signicant negative
coecient on the variable Khushhali Bank client. After
104
WORLD DEVELOPMENT
Table 7. Indicators of poverty: consumption-expenditure impacts on the core poor

Literacy of female
Numeracy of female
Literacy of male
Numeracy of male
Children age 04
Children age 59
Children age 1015
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
villcity
Khushhali Bank Client (0/1)
Core Poor
Accessed Loans
Core Poor*Accessed Loans
Constant
Observations

*
Signicant at 10%.
**
Signicant at 5%.
***
Signicant at 1%.
1
OLS
Monthly
Expenditure per
Capita - Food
2
OLS
Monthly
Expenditure per
Capita NonFood
3
OLS
Monthly
Expenditure per
Capita Health
Care
4
OLS
Expenditure on
Education
(per child)
2.05
[3.57]
15.43
[17.72]
7.06
[14.35]
2.09
[2.49]
17.2
[15.65]
25.24
[15.35]
12.8
[9.41]
44.79*
[25.99]
0.93***
[0.17]
36.12***
[9.54]
44.21***
[8.36]
34.45***
[8.17]
52.82***
[14.21]
19.64
[16.56]
1.89
[20.78]
41.62
[24.21]
61.37
[27.47]
71.73***
[17.83]
64.88***
[16.56]
9.7
[16.73]
19.1
[19.61]
13.89
[23.78]
37.23
[27.13]
42.84
[25.30]*
0.02
[0.06]
25.05
[34.27]
491.44***
[29.08]
48.79
[36.13]
16.71
[43.73]
1,197.83***
[36.13]
2876
12.35
[9.71]
60.21
[48.25]
47.18
[39.08]
27.25***
[6.77]
30.22
[42.62]
20.79
[41.79]
0.69
[25.62]
237.65***
[70.78]
0.76*
[0.46]
24.83
[25.98]
44.93**
[22.77]
15.79
[22.24]
67.89
[38.70]
55.19
[45.09]
62.72
[56.58]
66
[65.93]
24.72
[74.80]
85.52
[48.56]
119.6***
[45.09]
58.39
[45.56]
22.13
[53.39]
17.82
[64.74]
111.38
[73.88]
57.77
[68.90]
0.57***
[0.17]
97.55
[93.33]
198.42**
[79.19]
8.48
[98.38]
90.23
[119.07]
1,055.61***
[98.38]
2876
0.29
[2.07]
14.83
[10.29]
19.8**
[8.33]
2.84**
[1.44]
2.89
[9.09]
6.11
[8.91]
0.42
[5.46]
11.66
[15.09]
0.26***
[0.10]
2.09
[5.54]
8.28
[4.86]
4.27
[4.74]
2.11
[8.25]
11.91
[9.61]
15.3
[12.06]
18.34
[14.06]
32.52
[15.95]
14.38
[10.35]
9.83
[9.61]
5.4
[9.71]
3.39
[11.38]
0.84
[13.80]
21.33
[15.75]
2.85
[14.69]
0.02
[0.04]
48.7**
[19.90]
6.45
[16.89]
38.89
[20.98]*
8.78
[25.39]
112.29***
[20.98]
2876
44.09***
[6.08]
39.81
[30.43]
27.92
[24.66]
34.97***
[4.25]
36.91
[26.85]
14.74
[26.37]
15.16
[16.11]
13.7
[44.70]
0.51*
[0.29]
12.5
[16.41]
46.17***
[14.38]
136.39***
[14.02]
15.11
[24.38]
41.51
[28.45]
29.23
[35.63]
117.02***
[41.45]
11.61
[47.14]
22.61
[30.66]
75.53***
[28.45]
74.51***
[28.73]
27.4
[33.56]
50.96
[40.64]
14.39
[46.43]
82.97*
[43.45]
0.09
[0.11]
45.68
[58.73]
153.24***
[50.03]
34.39
[61.86]
127.98
[75.18]*
215.59***
[61.59]
2876
member households actually receive a loan, the wives of male

bank clients report being more empowered in several areas (see
the signicantly positive coecients on the variable Accessed
Loans), and these positive eects outweigh the negative bias
105
that apparently exists in member households. Thus overall,

accessing a loan does appear to make a dierence to the sense
of empowerment: the negative coecients on Khushhali Bank
client in Table 6 columns 3 and 4 are smaller and less
Table 8. Indicators of poverty: education and health: impacts on the core poor
1
2
3
4
5
6
LOGIT
OLS
LOGIT
LOGIT
LOGIT
LOGIT
Education:
Education:
Health:
Health: Probability
Probability take Probability
probability
days children
Probability seek
medical
ORS to treat
children
children enrolled absent from school
medical
treatment from trained
diarrhea
vaccinated
in school
treatment
professional
if child ill
if child ill
Literacy of female
Numeracy of female
Literacy of male
Numeracy of male
Children age 04
Children age 59
Children age 1015
Female age 1621
Female age 2229
Female age 3039
Female age 4049
Female age 5059
Female age over 60
Male age 1621
Male age 2229
Male age 3039
Male age 4049
Male age 5059
Male age over 60
Khushhali Bank Client (0/1)
Implied Odds Ratio
0.05***
[0.02]
0.02
[0.10]
0.1
[0.08]
0.09***
[0.01]
0.25***
[0.08]
0.12
[0.08]
0.06
[0.05]
0.2
[0.14]
0***
[0.00]
0.09*
[0.05]
0.66***
[0.05]
0.78***
[0.05]
0.05
[0.07]
0.18**
[0.09]
0.05
[0.11]
0.11
[0.13]
0.27*
[0.14]
0.07
[0.09]
0.15*
[0.09]
0.34***
[0.09]
0.03
[0.11]
0.04
[0.13]
0.12
[0.14]
0.15
[0.13]
0.41**
[0.18]
1.5
0.27
[0.17]
0.32
[0.87]
1.72**
[0.71]
0.19
[0.12]
3.41***
[0.77]
3.69***
[0.76]
0.51
[0.46]
2.51*
[1.28]
0.02**
[0.01]
0.17
[0.47]
3.2***
[0.41]
1.39***
[0.40]
0.93
[0.70]
1.52*
[0.82]
1.25
[1.02]
2.77**
[1.19]
1.56
[1.35]
0.78
[0.88]
0.36
[0.82]
0.3
[0.82]
1.8*
[0.96]
1.97*
[1.16]
1.05
[1.33]
0.36
[1.25]
1.66
[1.68]
0.05**
[0.02]
0.17*
[0.10]
0.15*
[0.08]
0.02
[0.01]
0.08
[0.09]
0.07
[0.08]
0.05
[0.05]
0.09
[0.14]
0***
[0.00]
1.22***
[0.06]
0.29***
[0.04]
0.05
[0.04]
0.09
[0.08]
0.11
[0.09]
0.01
[0.11]
0.35***
[0.13]
0.18
[0.15]
0.03
[0.09]
0.21**
[0.09]
0.06
[0.09]
0.29***
[0.11]
0.26**
[0.13]
0.13
[0.15]
0.25*
[0.14]
0.24
[0.18]
0.79
0.04**
[0.02]
0.05
[0.10]
0.02
[0.08]
0.02
[0.01]
0.08
[0.09]
0.04
[0.08]
0.1*
[0.05]
0.11
[0.14]
0***
[0.00]
1.18***
[0.06]
0.29***
[0.04]
0.07*
[0.04]
0.09
[0.08]
0.14
[0.09]
0.05
[0.11]
0.31**
[0.13]
0.2
[0.15]
0.1
[0.09]
0.17*
[0.09]
0.02
[0.09]
0.22**
[0.11]
0.18
[0.13]
0.05
[0.15]
0.11
[0.14]
0.21
[0.18]
0.81
0.07**
0.02
[0.03]
[0.02]
0.03
0.13
[0.11]
[0.12]
0
0.02
[0.09]
[0.10]
0.02
0.02*
[0.02]
[0.01]
0
0.3***
[0.10]
[0.10]
0.21*
0.22**
[0.11]
[0.10]
0.1
0.04
[0.06]
[0.06]
0.32*
0.08
[0.17]
[0.16]
0
0***
[0.00]
[0.00]
0.54***
1.94***
[0.06]
[0.08]
0.24***
0.37***
[0.05]
[0.05]
0.18***
0
[0.06]
[0.05]
0.08
0.19**
[0.11]
[0.09]
0.1
0
[0.12]
[0.10]
0.14
0.26
[0.14]
[0.12]
0.05
0.32**
[0.18]
[0.14]
0.04
0.03
[0.20]
[0.16]
0.19*
0.07
[0.11]
[0.11]
0.26**
0.28***
[0.13]
[0.10]
0.22*
0.06
[0.13]
[0.10]
0.3**
0.03
[0.15]
[0.11]
0.32*
0.27*
[0.18]
[0.14]
0.24
0.16
[0.21]
[0.16]
0.18
0.23
[0.18]
[0.15]
0.16
0.02
[0.24]
[0.20]
1.17
0.98
(continued on next page)
106
WORLD DEVELOPMENT
Table 8Continued
1
LOGIT
Education:
probability
children enrolled
in school
2
OLS
Education:
days children
absent
from school
3
LOGIT
Health:
Probability seek
medical
treatment
if child ill
4
LOGIT
Health: Probability
medical
treatment
from trained
professional
if child ill
5
LOGIT
Probability take
ORS to treat
diarrhea
6
LOGIT
Probability
children
vaccinated
0.56***
[0.15]
0.57*
0.33*
1.88
[1.43]
0.07
[0.16]
0.94**
0.39**
[0.19]
1.47
0.28
[0.23]
1.32
1.84***
[0.20]
2876
0.24
[0.16]
0.78
0.36*
[0.19]
1.44
0.35
[0.23]
1.42
1.83***
[0.20]
2876
0.09
[0.20]
0.91
0.09
[0.26]
0.92
0.59**
[0.29]
1.81
3.06***
[0.27]
2876
0.45**
[0.18]
0.64
0.02
[0.21]
0.98
0.78***
[0.26]
2.18
1.63***
[0.22]
2876
Core Poor
Implied Odds Ratio
Accessed Loans
Implied Odds Ratio
Core Poor*Accessed Loans
Implied Odds Ratio
Constant
Observations
[0.19]
0.72
0.55**
[0.23]
1.74
1.76***
[0.20]
2876
0.71
[1.77]
1.5
[2.15]
5.62***
[1.76]
2876

* Signicant at 10%.
** Signicant at 5%.
*** Signicant at 1%.
signicantly signicant than the positive coecients on the

variable Accessed Loans. Female borrowers, however, do
not report any major improvement in their sense of empowerment after taking out a loan. In most instances the interaction
term between a female client and accessing loans is insignicant. 12 This lack of impact on empowerment for female borrowers as opposed to the wives of male borrowers may be
due to their higher initial empowerment.
Exploring heterogeneity in the sample, it again appears that
rural households are driving the overall results (results for the
urban and rural sub-sample, respectively are not reported but
are available from the authors). Urban households do not
report much impact of program participation on female
empowerment, with the exception of female borrowers, who
report that accessing loans brings signicantly more inuence
in household decisions as to whether or not to borrow money.
These results are interesting and may merit further examination since they counter conclusions of some earlier studies,
which found that even when borrowers are women, they do
not necessarily exert inuence over key family decisions. 13
The results herein suggest that access to micronance empowers women in the household, even if the woman herself is not
the borrower.
(d) Impact on the core poor
The millennium development goals are particularly concerned with the poorest quintile of the populationthe core
poor. But a common argument in the debate on commercially-oriented micronance is that it cannot reach the core
poor either because they are excluded by social networks or
because they are too risk averse to participate to the same
degree as other borrowers. To inform this debate, we look
for heterogeneous impacts of the program on particularly
poor households by introducing a dummy variable that takes
the value of 1 for households in the bottom quintile of
consumption-expenditure for the sample. This dummy is
interacted with M T in Eqn. (1) and the interaction term is
introduced as an additional explanatory variable.
The results are reported in Tables 7 and 8 for the analysis of

expenditure-consumption and education and health impacts,
respectively. Table 7 reports that, as expected, the expenditure
of this group of core poor is signicantly lower for three out of
the four (all except health care) measures of expenditure.
Including the interaction term for the bottom quintile
indicates no statistically signicant dierences in impact of
borrowing to this group of very poor borrowers for expenditure, with the exception that borrowing under the program
has a statistically weak positive impact on educational expenditures per child for the very poor. It should be noted that this
is a dierential eect relative to non-core borrowers and is not
strong enough to oset the disadvantage the core poor have
due to their lower income.
In relation to the education and health variables, as
expected, the poorest households are less likely than the average to have their children enrolled in school or to have them
vaccinated. However for the core poor who borrow under
the program there is a positive eect on both of these probabilities that more than osets this disadvantage. Table 8 shows
that for columns 1 and 6 where the core poor access loans
there is a signicant positive coecient estimate on the interaction between the core poor dummy and loan access, which
is highly statistically signicant and quantitatively larger than
the negative coecient estimates on the core poor dummy.
Borrowing under the program also statistically signicantly increases the probability that children in core poor households
receive ORS treatment for diarrheal disease, a signicant killer
of children under 5 in Pakistan.
There is no evidence that for income generating activities in
livestock rearing, arable farming, or micro-enterprises there is
any dierential eect between core-poor borrowers and other
borrowers. 14 Thus we nd no evidence the very poor borrowers are dierentially aected in any major way relative to
better-o borrowers in terms of monetary measures, but we
do nd access to loans appears to help in terms of their educational and health status. This result challenges the applicability of some earlier research ndings where it has been found
that it is the better-o poor who have most to gain from
micronance 15to the case of Pakistan.
6. CONCLUSIONS
The form of micronance practiced by the Khushhali bank
involves small loans oered over short periods and at interest
rates that are high relative to those in the formal banking sector.
It is a clear example of second generation micronance institutions that aim for nancial sustainability to avoid reliance on
donor funding. A key policy question is how far this commercial
orientation dilutes the poverty reducing and welfare improving
goals that are associated with the micronance movement. This
study addresses this issue empirically by linking the banks lending activity with progress on a range of indicators that are consistent with progress toward the MDGs that form the centerpiece of the governments poverty reduction strategy.
The empirical analysis here provides little evidence that
access to the Khushhali Banks microcredit program has had
positive impacts on monetary measures of welfare on aggregate, although there is evidence that rural households participating in the program have higher food expenditure. However
there is no impact on measures of income generating activities,
either external sales or prots, for microenterprises or agricultural activities.
In relation to non-monetary measures, access to loans
does appear associated with a higher probability of medical
treatment for children in the household and a higher probability that the provider of that treatment is trained. This
appears to be an explicit household choice not inuenced
by training or advice associated with the loan, but one
which is clearly in line with the Millennium Development
Goal to reduce child mortality.
In terms of family dynamics, wives of male bank members
who take out loans report greater empowerment in family
decisions than wives of members who have not yet taken out
a loan. The Millennium Development Goal to promote gender
equality and female empowerment is particularly important in
South Asia and this is evidence that micronance in Pakistan
is contributing toward this goal.
107
For the very poor householdsthose in the bottom quintile of the sample, or at less than half the ocial poverty
linethere is no special impact on monetary indicators,
but there is evidence that the poorest households share in
the improvements in childrens health care and even get
some extra impacts. Among the very poor, borrowing increases the likelihood that children in the household are
vaccinated and receive treatment (ORS) for diarrheal
disease, a major killer of children under ve in Pakistan.
It also brings educational benets, since the poorest borrowers report higher than expected educational expenditures per
child and higher rates of childrens enrollment in school.
Here the loan or any additional income it creates appears
to allow the poorest households to invest in childrens
health and education, again helping toward important
MDG targets.
Overall, we nd no evidence that at the time of the survey
the Khushhali bank has made much progress toward the goal
of stimulating self-sustaining household income growth
through either urban microenterprises or rural agricultural
activities. These results are of concern for all who view the
goal of micronance as stimulating small scale income-generating activity. The survey was conducted in 2005 only ve
years after the rst loans were made and it is possible that
cumulative income eects may build up over time through frequent repeat borrowing and repayment.
However, we do nd evidence thatat least in rural areas
the bank is contributing to the achievement of some of the
MDGs by reducing poverty, empowering women, and improving the health of children. The poorest households are also
making progress on the goal of achieving universal primary
education. Thus, these ndings suggest that, given a supportive regulatory environment, it is possible for commercially-oriented micronance banks to meet a double bottom line of
simultaneously pursuing prots and a social mission to promote development.
NOTES
1. These goals are to (1) eradicate extreme poverty and hunger, (2)
achieve universal primary education, (3) promote gender equality and
empower women, (4) reduce child mortality, (5) improve maternal healthy,
(6) combat HIV/AIDS, malaria, and other diseases, (7) ensure environmental stability, and (8) develop a global partnership for development
United National Development Programme (n.d.).
2. This information on the micronance sector in Pakistan comes from

Haq (2008). In 2009, there were problems in the sector with one MFI
failing and the extent of outreach shrinking.
3. See Haq (2008); this estimate is higher than the operational selfsuciency ratios for other MFIs in Pakistan.
4. The ocial poverty line is based on caloric intake and translates into
approximately 1,000 rupees per capita per month of food consumption.
Very rough calculations indicate that this level of caloric consumption
corresponds to about 87 cents per day on total consumption. The authors
thank Talat Anwar for raising this issue and providing an updated poverty
line estimate.
5. This is a range of $50$500 at the exchange rate of approximately
R60/US dollar prevailing in mid-2005.
6. In Pakistan, Zafar and Abid (1999) is an example of a qualitative

approach to micronance impact using focus groups and survey data from
55 households with no control group identied.
7. Morduch (1999, p. 1605) pointed out that the eligibility criterion of low
land holdings was not strictly enforced in practice. In a reworking of the
results focusing on more directly comparable households, he found no
impact on consumption from participation. The most recent critique of the
original analysis and re-working of the data is Roodman and Morduch
(2009).
8. An F test that coecient estimates on all village dummies are jointly
insignicant is rejected at the 1% condence level for many of the
specications, but re-running the regressions without village dummies
yielded no qualitative dierence in the results. The basic results in the
model without the village dummies are not reported but are available from
the authors.
9. Coleman (1999) does not in fact does not suer from this problem.
10. Seven hundred and six, or roughly half of the treatment group, were
currently active clients: 13% were inactive, 18% were members of a group
where at least one in the group was in defaultthus making all members
ineligible to receive loansand 16% had dropped out of the program.
108
WORLD DEVELOPMENT
11. To avoid excessive detail these results are not given but are available
from the authors.
12. Where is it weakly signicant (in the case of child-rearing decisions) it
has an unexpected negative sign.
13. See Goetz and Gupta (1996) for an interpretation along these lines
for the use of microcredit in Bangladesh.
14. The results for these income generating activities are not reported
here but further details of this aspect of the study are in Montgomery
(2006). It should be noted that in the study design the bottom quintile had
to be identied ex post rather than ex ante in relation to program

participation. This means that there is the possibility of bias where the
program itself inuenced who was in the bottom quintile. Given the way in
which the study was organized it is not possible to control for this so that
the results from this part of the analysis should be interpreted with some
caution. However the fact that overall there seems to have been no
participation eect on total or income consumption weakens the risk of
this bias.
15. See the surveys of Meyer (2002), Montgomery and Weiss (2005),
Goldberg (2007), and Karlan and Murdoch (2010, chap. 2).
REFERENCES
Alexander, G. (2001). An empirical analysis of micronance: Who are the
clients? Paper presented at 2001 Northeastern Universities Development Consortium Conference.
Arif, G. M. (1999). The structure of informal credit market in Pakistan.
Pakistan Institute of Development Economics Research Report No.
16, Islamabad.
Armendariz de Aghion, B., & Morduch, J. (2005). The economics of
micronance. Boston, MA: MIT Press.
Banerjee, A. Duo, Glennester, E. R., & Kinnan, C. (2009). The miracle of
micronance: Evidence from randomized evaluations. MIT Department
of Economics Working Paper, Boston, MA.
Coleman, B. E. (1999). The impact of group lending in Northeast
Thailand. Journal of Development Economics, 60, 105141.
Coleman, B. (2006). Micronance in Northeast Thailand: Who benets
and how much?. World Development, 34(9), 16121638.
Chowdhury, M., Ghosh, D., & Wright, R. (2005). The impact of
microcredit on poverty: Evidence from Bangladesh. Progress in
Development Studies, 5(4), 112.
Cull, R., Demirguc-Kunt, A., & Morduch, J. (2007). Financial performance and outreach: A global analysis of leading microbanks.
Economic Journal, 117, 107133.
De Mel, S., McKenzie, D., & Woodru, C. (2008). Returns to capital in
microenterprise: Evidence from a eld experiment. Quarterly Journal of
Economics, CXXIII(4), 13291372.
Dehejia, R., Montgomery, H., & Morduch, J. (2008). Do interest rates
matter? Credit demand in the Dhaka slums. mimeo.
Duo, E., & Kremer, M. (2005). Use of randomization in the evaluation of
development eectiveness. In O. Feldstein,, G. Ingram, & G. Pitman
(Eds.). Evaluating development eectiveness. New Brunswick, NJ:
Transaction Publishers.
Goetz, A. M., & Gupta, R. S. (1996). Who takes the credit: Gender, power
and control over loan use in Rural Credit Programs in Bangladesh.
World Development, 24(1), 4563.
Goldberg, N. (2007). Measuring the impact of Micronance: Taking stock
of what we know. Grameen Bank Foundation, Washington, DC.
Government of Pakistan. (2003). Accelerating economic growth and
reducing poverty: The road ahead (Poverty Reduction Strategy Paper).
Ministry of Finance, Government of Pakistan.
Government of Pakistan. (2004). Between hope and despair. Pakistan
Participatory Poverty Assessment National Report. Islamabad, Planning Commission, Government of Pakistan.
Grosh, M., & Glewwe, P. (2000). Designing household survey questionnaires for developing countries: Lessons from 15 years of the living
standards measurement study. Volumes 1, 2, and 3, Washington DC,
World Bank.
Haq, A. (2008). Micronance industry assessment: A report on Pakistan.
Pakistan Micronance Network, Islamabad, downloaded from
<www.mfn.org.pk>.
Harford, T. (2008). The battle for the soul of micronance. Financial Times
December 6th.
Hashemi, S. (1997). Those left behind: A note on targeting the hardcore
poor. In G. Wood, & I. Sharif (Eds.). Who needs credit? Poverty and
nance in Bangladesh, Dhaka. Dhaka: University Press.
Hulme, D., & Mosley, P. (1996). Finance against poverty (Vols. 1 and 2).
London: Routledge.
International Year of Microcredit (n.d.). Building inclusive nancial sectors
to achieve the millennium development goals. Downloaded from
<www.yearofmicrocredit.org/>.
Karlan, D. (2001). Micronance impact assessments: The perils of using

new members as a control group. Journal of Micronance, 12.
Karlan, D., & Morduch, J. (2010). Access to nance. In D. Rodrik, & M.
Rosenzweig (Eds.). Handbook of development economics (Vol. 5).
York, UK: North Holland.
Karlan, D., & Zinman, J. (2008). Credit elasticities in Less Developed
Countries: Implications for micronance. American Economic Review,
98(3), 10401068.
McKenzie, D. (2009). Impact assessments in nance and private sector
development: What have we learnt and what should we learn?
Policy Research Working Paper 4944, World Bank, Washington,
DC.
McKernan, S.-M. (2002). The impact of microcredit programs on selfemployment prots: Do noncredit program aspects matter?. Review of
Economics and Statistics, 84(1), 93115.
Meyer, R. (2002). Track record of nancial institution in assisting the poor
in Asia. ADB Institute Research Paper 46, Tokyo.
Montgomery, H. (2006). Serving the poorest of the poor: The impact of
the Khushhali Banks micronance lending in Pakistan. In J. Weiss, &
H. A. Khan (Eds.), Poverty strategies in Asia. Cheltenham: Edward
Elgar.
Montgomery, H. (2005). Meeting the double bottom line: The impact of the
khushhali banks micronance program in pakistan. berlin: ADB
Institute.
Montgomery, H. and J. Weiss (2005) Great Expectations: Micronance
and Poverty Reduction in Asia and Latin America Oxford Development Studies, 33, 3-4, 391-416.
Morduch, J. (1999). The micronance promise. Journal of Economic
Literature, 37, 15691614.
Pitt, M. M., & Khandker, S. (1998). The impact of group-based
credit programs on poor households in Bangladesh: Does the
gender of participants matter?. Journal of Political Economy, 2,
958977.
Ravallion, M. (2009). Should the randomistas rule? Economists Voice.
Berkeley Electronic Press, 2, 15.
Roodman, D., & Morduch, J. (2009). The impact of microcredit on the poor
in Bangladesh: Revisiting the evidence. Centre for Global Development
Working Paper, p. 174.
United Nations Development Programme (n.d.). What are the Millennium
Development Goals? Downloaded from <http://www.undp.org/mdg/
basics.shtml>.
World Bank. (2008). Impact evaluation for micronance. Doing Impact
Evaluation 7. Poverty Reduction and Economic Management, World
Bank, Washington, DC.
Zafar, R., & Abid, S. (1999). Impact assessment as a management and
policy tool: The social and economic outcomes of Kashfs micronance
series. Kashf Foundation Discussion Paper Series, Discussion Paper
No. 3.
APPENDIX A. DETAILS OF DATA COLLECTION

Design of the survey followed international guidelines, in
particular those laid out in the three volume series by Grosh
and Glewwe (2000) on the Living Standards and Measurement
Survey (LSMS). Full details of the questionnaire are in Montgomery (2005).
A.1 Survey instrumentquestionnaire

Design of the survey instrument, the questionnaire to be
used in gathering data for the study, was primarily guided
by the research question: what has been the impact of the
micronance program on household welfare? It was decided
to include a relatively wide denition of welfare that includes
non-economic measures of welfare such as education, health,
or empowerment.
Core components of the LSMS were incorporated, and the
nal questionnaire also drew upon the AIMS-SEEP Impact
Survey Tools (Assessing the Impacts of Microenterprise ServicesSmall Enterprise Education and Promotion Network),
impact assessment tools designed specically for assessment of
micronance institutions, as well as several carefully designed
questionnaires used in previous studies in Pakistan. The ndings of a nationwide participatory poverty assessment (Government of Pakistan (2004)) were also consulted and results
of focus group discussions with Khushhali Bank clients were
incorporated.
Length of the questionnaire was limited to what could be
reasonably delivered in a maximum of one hour if all components were asked. In the nal administration, most questionnaires took substantially less than one hour since very few
households would actually respond to all sections. The sequence of the questionnaire was guided by the LSMS, and
accordingly sensitive questions on nances or empowerment
issues were administered last.
To increase the accuracy of the information gathered and
to enable the survey to address gender issues such as
empowerment, both the male and the female head of
household were interviewed separately for each household.
The suitability of dierent components of the questionnaire
for the male or female version was decided based on the
previous questionnaires listed above and conrmed in pretesting.
The questionnaire was prepared simultaneously in English
and Urdu and then translated into two regional languages:
109
Pushto and Sindhi. Accuracy of the translations was checked

by back translation into the original language.
A.2 Implementation
The survey was pre-tested on both client and non-client
households and the nal survey was implemented over an
8 week period after a major harvest when there would be
many new villages and clients just getting access to Khushhali
Bank services for the rst time, making it easier to collect data
on a suitable control group.
The survey was carried out by an independent multinational
survey company. Male surveys were conducted by male surveyors and female surveys by female surveyors. Surveyors
and supervisors for each team were recruited from local areas
and interviews were conducted in local languages. Since many
of the surveyors were new, one week of classroom training on
administration of surveys, and eld testing of the surveyors
skill in both rural and urban areas were conducted. Extra surveyors were trained in the event that any surveyor had to be
replaced during the training, eld-testing, or once the survey
was underway, but that was not necessary.
A.3 Quality control
Survey teams spent 34 days in each village included in the
survey to allow time for the team supervisor to edit all completed
questionnaires and back-check 15% of the eldwork. If any
problems were discovered during back-checking, then 100% of
that individual surveyors work was checked. An independent
quality control department similarly carried out back-checking
of each supervisors work. Data processing was not able to be
conducted on-site due to cost considerations, and was instead
done on edited questionnaires in a centralized location. A data
program was designed to automatically check the consistency
of answers and in addition 10% of the data entry and coding
was randomly back-checked.
Available online at www.sciencedirect.com

Can Commercially-Oriented Microfinance Help Meet MDGs (Evidence From Pakistan)

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Can Commercially-Oriented Microfinance Help Meet MDGs (Evidence From Pakistan)

Hochgeladen von

Copyright:

Verfügbare Formate

World Development Vol. 39, No. 1, pp.

Can Commercially-oriented Micronance Help Meet the

a number of ways: it is based on a large national survey of

2. KHUSHHALI BANK OPERATIONS

new business or to purchase assets or working capital for an

as a guide in designing our study. Thus, we looked at what

CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?

Table 1. Summary statisticsdependent variables

Income generating activitiesmicroenterprise

Income generating activitieslivestock

Income generating activitiesagriculture

Ability to pay for medical treatment from own sources

Note: Financial variables are measured in rupees.

addition to program participation and it is nearly impossible

Controlling for these biasesdetermining the eects of

There are three broad approaches that have been applied

their overall acknowledged weakness is external validity

CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?

village which have already self-selected to participate in the

may dier from the non-member population at large. There

not, and controlling for household characteristics (like age,

where Yij, is a vector of impact or outcome variables, Xij is a

Table 2. Summary statisticsindividual household characteristics

Education of highest educated male (years)

CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?

of household. Given the potential for the very poor or core

Education of highest educated

CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?

Table 3B. Income generating activitiesonly URBAN

Education of highest educated

Notes: Standard errors in brackets.

on average Rs. 41 per month higher per capita expenditures on

prefer to or need to use any extra income generated by

Education of highest educated

CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?

Education of highest educated female

Education of highest educated female

Notes: Standard errors in brackets.

distinguishes between households (in both treatment and

the analysis sub-divides the sample into urban and rural

CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?

Education of highest educated female

Table 5. Indicators of poverty: education and health

Notes: Standard errors in brackets.

CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?

taking out a loan on the sales of urban micro-enterprises

expenditure (Table 4c, column 1). In contrast, the positive

CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?

Implied odds ratio

Notes: Standard errors in brackets.

These results are disappointing for those seeking evidence of

Of those households who seek treatment for their childrens

(b) Social indicators

(c) Female empowerment

Expenditure-based poverty measures can omit important

An important aspect of the study design is to test whether

Education of highest educated female

Notes: Standard errors in brackets.

CAN COMMERCIALLY-ORIENTED MICROFINANCE HELP MEET THE MILLENNIUM DEVELOPMENT GOALS?

member households actually receive a loan, the wives of male

that apparently exists in member households. Thus overall,

Notes: Standard errors in brackets.

signicantly signicant than the positive coecients on the

The results are reported in Tables 7 and 8 for the analysis of