Does Micro Finance Benefit The Poor
Does Micro Finance Benefit The Poor
Does Micro Finance Benefit The Poor
or policies of the
Asian Development Bank. The Asian Development Bank does not guarantee the accuracy of the data presented.
Shahid Khandker is a senior economist in the Rural Development Group, Research Department of the World Bank. This paper
is to be delivered at the Asia and Pacific Forum on Poverty: Reforming Policies and Institutions for Poverty Reduction,
to be held at the Asian Development Bank, Manila, 5-9 February 2001. The paper is based on the authors past research
and an ongoing research project on micro-finance with Mark Pitt of Brown University and M.Abdul Latif, Binayak Sen and
others from Bangladesh Institute of Development Studies. The author would like to express his thanks to Mark Pitt for
comments and suggestions, and to Nidhiya Menon and Hussain Samad for excellent research assistance. Views expressed
in this paper are entirely his and do not reflect those of the World Bank or its affiliated organizations, nor of his research
collaborators.
I. Introduction
Micro-finance means transactions in small amounts of both credit and saving, involving
mainly small-scale and medium-scale businesses and producers. The poor, who cannot
run a small business because they lack capital, may also benefit from micro-finance
organization. The poor, especially poor women, are the prime reason for micro-finance
intervention in many countries, including Bangladesh. The question of whether micro-
finance really benefits the poor depends on how we define poverty and what kind of
help micro-finance offers to the poor to combat poverty. The face of poverty varies from
country to country.
Poverty may mean a lack of some or all of the following:
(i) Entitlement to food and other basic necessities.
(ii) Access to public provision of economic, social, and human infrastructures.
(iii) Credit, opportunities for income generation, and consumption stabilization.
(iv) Empowerment in both private and public resource allocations.
(vi) Access to a social safety net and other resources that help households
withstand natural and other shocks, thus safeguarding the very existence of life
and families.1
Given such a multifaceted definition of poverty, it is clear that what micro-
finance can do for the poor depends on the poor’s ability to utilize what micro-finance
offers them. In many countries, micro-finance provides a window of opportunity for the
poor to access a borrowing and saving facility. In other countries, these facilities also
provide organizational help, training, safety nets, empowerment, and financial and
other help during crises. Micro-finance organizations can alleviate liquidity constraints,
stabilize consumption, and enhance both income and consumption for the poor,
thereby augmenting the poor’s welfare. Borrowing from a micro-finance facility to run a
business is a self-help activity. Moreover, micro-finance satisfies the derived demand
for capital in the resource-poor households. Success in such a business depends on
skill and entrepreneurship, as well as local economic conditions. Many poor participants
may lack such skills and may benefit very little from micro-finance. On the other hand,
since a lack of economic growth is the cause of poverty in many economies, the
success of micro-finance borrowers is very much limited by their local economies.
Since micro-finance transactions with the poor are very small, the accrued
benefits from micro-finance may also be small. In such a case, the important policy
question is whether micro-finance is indeed a “real” help to the poor.2 To understand
whether micro-finance really benefits the poor, we must understand who participates in
micro-finance programs, how the participants can benefit from these organizations, and
whether such benefits are sustainable. When there is not much growth in an economy,
borrowing can redistribute income. Micro-finance may help the poor through income
redistribution, but benefits accrued to participants are unsustainable through income
redistribution alone.
Income redistribution also requires net income generation in the economy.
Therefore, we must know whether micro-finance helps sustain borrowers' accrued
benefits. To know this, we must assess both the participant-level and aggregate-level
impacts of micro-finance on both the short-term and the long-term welfare indicators in
an economy. Income, consumption, and employment impacts are the immediate effects
of micro-finance interventions. Assessing program impacts on such outcomes is
necessary, in order to establish whether micro-finance helps the participants. However,
it is difficult to establish whether micro-finance really helps the poor participants. Micro-
finance’s impacts are made through borrowing, without assessing the borrower-level
net worth and other long-term welfare impacts of micro-finance. Similarly, in order to
assess whether micro-finance impacts are substantial and go beyond the program
participants, we need to assess the impacts on aggregate welfare, both in the short
term and long term.
In this paper, I address both issues. First, I identify the participants of a micro-
finance program and their characteristics. Are the poor being attracted to such
programs? The poverty of a household can be characterized by two general factors in
many countries: physical and human capital endowments. A person is poor because he
or she is poor in physical and human capital resources.3 Thus, I differentiate program
participation by these two factors of poverty.
2. According to some critics of the micro-finance programs, who see this as a failure, argue that it cannot be an
effective tool for helping the poorest people (Adams and von Pischke 1992). Some impact studies seem to support
their opinion (e.g, Coleman 1999).
3. Of course, these are two proximate factors of poverty that are easier to measure.
finance programs did not benefit the poor in a sustainable way, then efforts to channel
donor resources into micro-finance would be misplaced.
4. Those who are interested in Mark Pitt’s analysis of Morduch’s re-analysis of Pitt and Khadker (1998) results,
may see the paper in Mark’s Pitt’s website: http://pstc3.pstc.brown.edu/~mp/
The outstanding research issues are: Can micro-finance reduce poverty on a sustained
basis at the participant level, if not at the community level? Does micro-finance really
benefit those who need the most help? What are micro-finance’s long-term impacts on
household welfare? Are the estimated impacts compelling, given that impact estimates
are sensitive to impact assessment methodologies? These issues are topics in an
ongoing research project supported by the World Bank.5 In 1998/1999, researchers
from BIDS, Brown University, and the World Bank revisited the households that were
surveyed in 1991/1992.
Program impact assessments are subject to two sources of bias: (i) program
placement endogeneity and (ii) program participation endogeneity. Earlier results show
that endogeneity is a serious issue and that assessment results could be misleading if
the endogeneity/heterogeneity were overlooked by the impact assessment (Pitt and
Khandker 1998). The methodology used in the BIDS-WB study is based on cross-
section data and uses a quasi-experimental survey design to resolve the problems of
endogeneity at the village and household levels.
The survey design has three components: (i) households are sampled in
villages with and without programs, (ii) eligible and ineligible households alike are
sampled in both types of village, and (iii) program participants and nonparticipants
within eligible households are sampled. Program impact identification has two
underlying conditions: (i) exogenous landholding and (ii) gender-based program design.
Households with less than half an acre of land are eligible—this eligibility requirement
makes the program’s impact more apparent to researchers. Additionally, researchers
identify women-only and men-only groups within villages to highlight the program
impacts by gender.
Three compelling reasons for using a repeat survey and panel data analysis are
the following:
(i) Results may otherwise be less robust, as some studies show that the
measurement of program impact depends on the treatment of program
endogeneity (Lalonde 1986). An alternative method such as the
household/individual fixed-effects method takes care of program
endogeneity, without relying on controversial assumptions.
(ii) Results may be sensitive to the land-based exclusion restriction that was
imposed in the analysis. It is important to see the robustness of the
results with a method that does not rely on the exogeneity of landholding
in identifying the program’s effects. With panel data, the fixed-effects
method is less reliant on the exact application of the landholding rule by
the micro-credit programs.
(iii) Cross-section data provides short-term program effects; however,
programs take a long time to influence outcomes such as net worth. A
panel survey analysis will measure the long-term program effects.
BIDS and the World Bank surveyed 1,798 households selected from 87 villages
in 29 thanas in 1991/1992. Eight thanas were drawn randomly from each of BRAC,
Grameen Bank, and BRDB's RD-12 project thanas. Five thanas without a micro-credit
program were also drawn randomly. Three villages were then drawn randomly from
5. Besides, the ongoing research project addresses other issues such as the role of credit versus non-credit
inputs, the role of group in group-based lending, and the role of micro-finance in empowering women.
each thana, where the programs had operated for at least three years. The household
distribution by program participation status according to 1991/1992 survey is: (i) 8.5
percent Grameen Bank members; (ii) 11.2 percent BRAC members; (iii) 6.3 percent
RD-12 project members; (iii) 40.9 percent eligible non-participants, and (iv) 33.1 percent
nontarget households. A follow-up survey in 1998/1999 included the households from
the previous survey, new households from the same villages, new households from
new villages in old thanas, and three more thanas, increasing the number of sample
households to 2,599.
According to the 1998/1999 re-survey with 2,599 households, household
distribution by program participation status is: (i) 14.3 percent Grameen Bank members,
(ii) 9.3 percent BRAC members, (iii) 3.6 percent RD-12 project members, (iv) 11.1
percent other NGO members, (v) 7.4 percent multiple program members, (vi) 32.0
percent eligible nonparticipants, and (vii) 22.3 percent non-target households. Program
participation has increased from 26.2 percent in 1991/1992 to 45.6 percent in
1998/1999.
6. The landholding is taken prior to program participation for micro-finance program members.
Both the short-term and long-term impacts of micro-finance are important for program
participants. Short-term behavior such as per capita consumption or the probability of
being above the poverty line determines whether the program makes any significant
impact in alleviating poverty among program participants. On the other hand, micro-
finance’s impact on long-term behavioral outcomes (such as household net worth)
determines whether the program’s impacts are sustainable, make headway over a
longer period of time, and allow households to accumulate assets over and above the
amount of their outstanding loans.
The conditional demand for outcomes (Yijt) (such as per capita consumption)
conditional on the level of credit demand (Cij) is
where δf and δm are the effects of female and male credit respectively.
With cross-sectional data (where t=1), the estimation issue arises as a result of
the possible correlation of µ cjf , µ cjm , and µ jy , and of ∈cij with ∈ijy (Pitt and
Khandker, 1998). To resolve this endogeneity, Pitt and Khandker used the village fixed-
effects method along with the exogenous gender and landholding-based exclusion
restrictions to estimate the impacts female and male credit. With panel data where
households have more than one observation, such complicated identification
restrictions are not required. This estimation is done by differencing out the unobserved
village and household attributes, which are the sources of correlation between the
credit demand and household outcome equations. Differencing equation (3) at two
points of time yields the following outcome equation
Consistent estimates of the credit effect δf and δm can be obtained from equation (4)
using the household fixed-effects method without recourse to instrumental variables
estimation such as the one used in the quasi-experimental survey design.
The above fixed-effects method consistently estimates credit impact. However,
if credit were measured with error, this error would be amplified when doing fixed-
effects estimation, particularly with only two time periods. This measurement error
would impart "attentuation bias" on the credit coefficients; they would be biased
towards zero. A standard correction for this bias is the use of instrumental variables (IV)
estimation. In our earlier paper (Pitt and Khandker 1998), IV method was called for
because of the endogeneity of credit and it also took care of any measurement error in
credit. In this household fixed-effects model, IV is not required because of endogeneity,
but may still be required because of measurement error. These conditions lead us to
estimate the above equation in two stages. First we regress changes in the amount of
borrowing between 1991/1992 and 1998/1999 at the 1991/1992 level of household and
village attributes, to get the predicted change in the amount of borrowing. Second, we
use the fitted credit amount in the changes in outcomes equations with the household
fixed-effects estimation.
Table 3 presents the summary statistics of all endogenous variables. Micro-
finance participants do better than non-participants in both 1991/1992 and 1998/1999,
in per capita income, per capita expenditure, and household net worth. The incidence
of poverty among participating households is lower in 1998/1999 than in 1991/1992,
and lower than among nonparticipating households in both periods.
Table 4 shows the two-stage estimates of program impacts on three short-term
outcomes: (i) per capita expenditure, (ii) per capita income, and (iii) the incidence of
poverty. Table 4 also shows the impact estimates on a long-term outcome: household
net worth. Results show that borrowing by women has a positive significant impact (at a
1 percent level of significance) on per capita expenditure, per capita income and net
worth, and a negative effect on the incidence of poverty. The response elasticity for
male borrowing is 0.05 for per capita consumption and 0.11 for net worth. In contrast,
the response elasticity for female borrowing is 0.03 for per capita consumption, and
0.06 for per capita income. A 10 percent increase in borrowing from a micro-finance
program reduces the probability of being above poverty line by 0.3 percent for male
borrowing and by 0.2 percent for female borrowing. The results support the assertion
that programs help poor borrowers over time, and meet their immediate needs.
There are four major programs (Grameen Bank, BRAC, ASA, and Proshika) that
account for about 65 percent of borrowers (68 percent of membership) and 72 percent
of all outstanding loans in 1998/1999. Grameen Bank accounted for one fourth of the
market share of the micro-finance industry in Bangladesh. The total loan outstanding
was about US$600 million in 1998/1999. This figure shows a large inflow of micro-
finance funds into rural Bangladesh and is expected to make an aggregate impact on
the local economy. That is, the impacts must extend beyond the program participants.
where Cjft and Cjmt is the village-aggregate amount of borrowing by female and male
borrowers in micro-credit programs. Unlike the earlier equation, every household in the
village in this equation receives the same female and male credit irrespective of
whether they participate in a micro-finance program. We use a village-level, fixed-
effects method to estimate the micro-finance effect on the welfare of all households in
a village.
Table 5 presents the aggregate micro-finance impact on the local economy.
One set of results includes no village-level attributes such as prices and infrastructures,
while the other set includes them. Without controlling for village-level attributes, we find
that micro-finance affects the aggregate village-level welfare indicators such as per
capita income, expenditure and net worth, but with an insignificant effect on poverty.
The effects either decrease or disappear when variations in village-level attributes are
controlled for. This trend also appears with the results for the nonparticipant
households in the economy (Table 6). Results may indicate no substantive program
impacts on village-level welfare indicators or among nonparticipating households,
although household net worth increases with female borrowing and income increases
with male borrowing for both cases.
The decline in impact due to the inclusion of village-level variables may imply
that micro-finance has an impact on all households or nonparticipants, and via changes
in village-level attributes such as prices and wages. In order to determine whether this
correlation exists, we fit a regression by relating changes in price, wage, and other
village attributes to changes in female and male borrowing between 1991/1992 and
1998/1999, after controlling for the 1991/1992 level of prices, wages, and other village
attributes. The estimates appear in Table 7. The results show that micro-finance affects
the local economic environment through price, wage, and village accessibility, thereby
affecting the welfare of the local economy.
VII. Conclusion
References
Adams, D.W., and J.D. Von Pischke 1992, “Microenterprise Credit Programs: Déjà vu” World
Development, 20 (October): 1463-70.
Coleman, B.E., 1999, “The Impact of group lending in Northeast Thailand” Journal of Development
Economics, 60 (October): 105-141.
Hashemi, Syed M., Sidney R. Schuler, and Ann P. Riley, 1996, “Rural Credit Programs and
Women’s Empowerment in Bangladesh” World Bank 24 (April): 635-53.
Hossain, Mahabub, 1988, “Credit for Alleviation of Rural Poverty: The Grameen Bank in
Bangladesh.” IFPRI Research Report 65, IFPRI, Washington, D.C.
Khalily, M. B., M. O. Imam, and S. A. Khan, 2000, "Efficiency and Sustainability of Formal and Quasi-
formal Micro-finance Programs-- An Analysis of Grameen Bank and ASA," The Bangladesh
Development Studies, A Special Issue on Microfinance and Development: Emerging Issues,
26 (June/September): 103-146, edited by Rushidan I. Rahman and Shahidur R. Khandker.
Khandker, Shahidur R., 1988, Fighting Poverty with Microcredit: Experience in Bangladesh, Oxford
University Press for the World Bank.
Lalonde, R., 1986, "Evaluating the Econometric Evaluations of Training Programs with Experimental
Data", American Economic Review, 76: 604-620.
Morduch, J., 1998, “Does Microfinance Really Help the Poor? New Evidence from Flagship Programs
in Bangladesh” Unpublished mimeo.
Morduch, J., 1999, “The Role of Subsidies in Microfinance: Evidence from the Grameen Bank” Journal
of Development Economics, vol. 60 (October): 229-248.
Pitt, Mark M., 1999, "Reply to Jonathan Morduch's 'Does Microfinance Really Help the Poor? New
Evidence from Flagship Programs in Bangladesh'" Unpublished Manuscript.
Pitt, Mark M. and Shahidur R. Khandker, 1998, “The Impact of Group-Based Credit Programs on
Poor Households in Bangladesh: Does the Gender of Participants Matter?” Journal of Political
Economy, 106 (October): 958-96.
Rahman, Rushidan I., 1996, “Impact of Grameen Krishi Foundation on the Socioeconomic Condition
of Rural Households,” Working paper No. 7, BIDS.
Ravallion, Martin, and Binayak Sen, 1995, “When Methods Matter: Towards a Resolution of the
Debate about Bangladesh’s Poverty Measures,” Policy working paper 1259, World Bank,
Washington, D.C.
Schuler, Sidney R., and Syed M. Hashemi, 1994, “Credit Programs, Women’s Empowerment, and
Contraceptive Use in Rural Bangladesh,” Studies Family Planning, 25 (March/April):65-76.
Yaron, Jacob, 1994, “What Makes Rural Finance Institutions Successful?” The World Bank Research
Observer 9 (1): 49-70.