Journal of International Development J. Int. Dev. 16, 331–353 (2004) Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jid.1081 QUANTITATIVE APPROACH TO IMPACT ANALYSIS OF MICROFINANCE PROGRAMMES IN BANGLADESH—WHAT HAVE WE LEARNED? M. A. BAQUI KHALILY* Department of Finance and Banking, University of Dhaka, Bangladesh Abstract: The quantitative impact assessment of microfinance programmes, like the pro- grammes themselves, originated in Bangladesh. This essay reflects on the significance and usefulness for present day researchers of the analytical advances made in Bangladesh since the beginning of the 1990s. Particularly in the area of selection bias, fungibility and the assessment of wider impacts, it argues, those advances are crucial, and need to be borne in mind by all practitioners; but financial sustainability remains an unresolved problem. Copyright # 2004 John Wiley & Sons, Ltd. Microfinance as a concept is nothing new. But it did not draw the attention of the professionals, international agencies, and policymakers until microfinance as a concept for poverty alleviation through institutional mechanisms was globally perceived. The Gram- een experiments and the successful operation of Grameen Bank have led to the wider acceptability of the strategy that the access to credit of the poor households through institutional arrangements matters. The institutional mechanism is based on the concept of ‘group-lending’. ‘Group-based’ production technologies are diverse in Bangladesh. The production technologies used by BRAC, ASA and PROSHIKA are different in nature. It has been over two decades that microfinance services are being provided to the poor households through microfinance institutions. As the goal of providing such services has all along been poverty alleviation, quite a large number of impact studies have been conducted in Bangladesh. However, the objectives of the studies have been diverse. Most of the studies evaluated impact at the household level (e.g., Khandker, 1998a; Zohir et al., 2001; Pitt and Khandker, 1998b; Mustafa et al., 1996; Husain, 1998). Few studies have focused on the supply-side analysis of impact of microfinance (e.g., Khandker et al., 1995; Copyright # 2004 John Wiley & Sons, Ltd. *Correspondence to: M. A. Baqui Khalily, Department of Finance and Banking, University of Dhaka, Dhaka 1000, Bangladesh. E-mail: [email protected]
23
Embed
Quantitative Approach to Measure Impact of Microfinance
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Journal of International Development
J. Int. Dev. 16, 331–353 (2004)
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jid.1081
QUANTITATIVE APPROACH TO IMPACTANALYSIS OF MICROFINANCE
PROGRAMMES IN BANGLADESH—WHATHAVE WE LEARNED?
M. A. BAQUI KHALILY*
Department of Finance and Banking, University of Dhaka, Bangladesh
Abstract: The quantitative impact assessment of microfinance programmes, like the pro-
grammes themselves, originated in Bangladesh. This essay reflects on the significance and
usefulness for present day researchers of the analytical advances made in Bangladesh
since the beginning of the 1990s. Particularly in the area of selection bias, fungibility and
the assessment of wider impacts, it argues, those advances are crucial, and need to be borne
in mind by all practitioners; but financial sustainability remains an unresolved problem.
Copyright # 2004 John Wiley & Sons, Ltd.
Microfinance as a concept is nothing new. But it did not draw the attention of the
professionals, international agencies, and policymakers until microfinance as a concept for
poverty alleviation through institutional mechanisms was globally perceived. The Gram-
een experiments and the successful operation of Grameen Bank have led to the wider
acceptability of the strategy that the access to credit of the poor households through
institutional arrangements matters. The institutional mechanism is based on the concept of
‘group-lending’. ‘Group-based’ production technologies are diverse in Bangladesh. The
production technologies used by BRAC, ASA and PROSHIKA are different in nature.
It has been over two decades that microfinance services are being provided to the poor
households through microfinance institutions. As the goal of providing such services has
all along been poverty alleviation, quite a large number of impact studies have been
conducted in Bangladesh. However, the objectives of the studies have been diverse. Most
of the studies evaluated impact at the household level (e.g., Khandker, 1998a; Zohir et al.,
2001; Pitt and Khandker, 1998b; Mustafa et al., 1996; Husain, 1998). Few studies have
focused on the supply-side analysis of impact of microfinance (e.g., Khandker et al., 1995;
Copyright # 2004 John Wiley & Sons, Ltd.
*Correspondence to: M. A. Baqui Khalily, Department of Finance and Banking, University of Dhaka, Dhaka1000, Bangladesh. E-mail: [email protected]
Khalily et al., 2002), but none have been conducted from the macroeconomic perspective.
Consequently, we have incomplete understanding of the impact of microfinance both at
the micro and macro level.
Impact analysis, using demand-side or household level data, has dominated the
microfinance research agenda during the past decade. Quite a large number of studies
have been conducted with diversified methodologies—quantitative and qualitative mod-
els. The general conclusion of the studies has been positive impact of micro credit at the
household, enterprise and individual levels. Extent of impact is the subject of the
quantitative approach to impact analysis. The approaches to and the methods of impact
assessment are diverse. They vary from descriptive to econometric; from limited use to
extensive use of econometric techniques. What lessons do we get from such analysis?
These questions are addressed in this paper.
MICROFINANCE IN PERSPECTIVE
Microfinance has been viewed from different perspectives. A very common perspective
has been poverty alleviation. Most studies have focused on determining the extent of
impact at the household level (e.g. Khandker, 1998; Husain, 1998; Zohir et al., 2001).
Micro credit impacts on poverty reduction through its impact on income, employment,
consumption, asset accumulation and savings.
But over-excitement of impact of microfinance in poverty alleviation tends to under-
mine the long-term role of micro finance in the sustainability of MFIs. The failure of
development banks and the adverse impact of distortions in rural financial markets (e.g.,
Von Pischke et al., 1993; Adams et al., 1984; Braverman and Guasch, 1989; Meyer and
Nagarajan, 2000) have caused emphasis to be placed on sustainability of microfinance
institutions as a process of developing rural financial markets and sustaining credit
facilities for the poor households (e.g. Khalily et al., 2000; Khandker et al., 1995; Yaron,
1992). However, the development sociologists view the contribution of microfinance and
microfinance institutions as the process of reducing the role of the state in financial
markets.
In the absence of ownership and/or regulation, the utility function of MFI top execu-
tives may dominate idealized patterns of organizational behaviour. In recent years, the
role of MFIs has been viewed more as a search for political influence. Cheap donor money
may contribute to such a process of promotion of political interests. In addition, cheap
money may make operation costly in the absence of any incentive for cost-effective
operations. These are likely to undermine the basic role of micro finance in poverty
alleviation.
Poor households do not have higher income level that can absorb adverse effects of
external shocks. Zaman using BRAC experience showed that micro finance contributes to
consumption-smoothening and minimizing vulnerability. Similar findings were also
derived in other studies (e.g., Sen, 2001).
Feminists and gender specialists argue that microfinance empowers women because
access to micro-credit has increased their role in family decision-making. The female
members are not only involved in economic decisions but also in family well-being.
Several empirical studies provide empirical evidence of women’s empowerment because
of microfinance (Hashemi, Schuler and Riley, 1996; Mayoux, 1999; Mahmud, 2001a,b;
Halder, 1998).
332 M. A. Baqui Khalily
Copyright # 2004 John Wiley & Sons, Ltd. J. Int. Dev. 16, 331–353 (2004)
MICROFINANCE PROGRAMMES FOR POVERTYALLEVIATION
IN BANGLADESH
In the late seventies and early eighties, two approaches were experimented for financing
poverty alleviation programmes of non-government organizations. One was the external-
finance based experiment as adopted by Grameen Bank, and the other was the self-reliant
model experimented by BRAC and ASA. The Grameen model experimentally made use of
bank credit for its expansion. The underlying thesis was that access to external credit
would accelerate the process of poverty alleviation. BRAC started its initial experiment on
self-reliance based on local resource mobilization in the early seventies. ASA did
experiment with a similar approach in the eighties. These experiments led to the general
conclusion that the pace of poverty alleviation, based on mobilization of local resources, is
very slow and negligible in most cases because of limited surplus resources. The success
of the Grameen Bank model and the limited success of local-resource based approach led
to searching for external finance. Bangladesh Bank as well as the international agencies
contributed to the early take-off of microfinance in Bangladesh. The major MFIs, like
ASA, PROSHIKA and BRAC, have been highly dependent on cheap funds.
During the past two decades there has been a wide expansion of microfinance services
in Bangladesh. There is no precise estimate of the number of MFIs and the number of
branches operating in the country. However, it is widely believed that more than 1000
MFIs with a network of over 4000 branches have been operating. It is a sector that has
created jobs for over 100 000 employees.
MFIs have grown over time with both government and non-government ownership. The
major programmes of the government are Bangladesh Rural Development Board (BRDB)
and Palli Daridro Bimochon Foundation (Rural Poverty Alleviation Foundation).
Grameen Bank is recognized as a specialized bank. The other MFIs are non-government
entities (to be referred subsequently as non-government MFIs). Until December 2001, 632
Table 1. Structure of microfinance in Bangladesh, 2000–01 (In Million)
MFIs Total active Total Cumulative Outstandingmembers net savings disbursement loan
Grameen 2.38 11 758.00 147 048.00 13 151.00
% 12.54 48.55 40.48 28.25
BRAC 3.51 4614.56 77 593.75 8537.83
% 18.50 19.06 21.36 18.34
ASA 1.98 2010.12 43 704.18 7309.52
% 10.44 8.30 12.03 15.70
Proshika 2.64 1639.97 20 611.88 5009.40
% 13.92 6.77 5.67 10.76
Top 22 except top 3 MFIs 2.66 2005.25 27 343.64 4953.67
% 14.02 8.28 7.53 10.64
BRDB 3.6 0 21 090.00 3387.00
% 18.98 0.00 5.81 7.28
PDBF 0.29 481 7746.00 716
% 1.53 1.99 2.13 1.54
Other small MFIs 1.91 1708.13 18 085.79 3486.63
% 10.07 7.05 4.98 7.49
Total 18.97 24 217.04 363 223.24 46 551.05
(100) (100) (100) (100)
Source: Credit and Development Forum (CDF), MFI Statistics, June 2002.
Microfinance Programmes in Bangladesh 333
Copyright # 2004 John Wiley & Sons, Ltd. J. Int. Dev. 16, 331–353 (2004)
MFIs including Grameen Bank and two government programmes had active members of
18.97 million and disbursed a cumulative total of Tk. 363 billion, and loans outstanding of
Tk. 46.6 billion (Table 1). The share of Grameen Bank has been around 13 per cent. Three
leading non-government MFIs (ASA, BRAC and PROSHIKA) constitute around 43
per cent of total members mobilized. Although the share in membership of Grameen Bank
is around 13 per cent, its share in loans outstanding has been around 29 per cent. This is
because of increasing loan size. Two government programmes with a share of around
22 per cent in membership mobilization had loans outstanding of Tk. 4 billion by the end of
2001 (Table 1). This suggests that non-government MFIs provide credit to the almost fifty
per cent of the aggregate members mobilized. This large amount of disbursement (annual
disbursement of over Taka 40 billion annually) is likely to have wider impact at the
household, individual and village levels.
PERSPECTIVE OF IMPACT ANALYSIS
Poor households are subject to liquidity constraints. As such, any additional flow of
external financial resources, be it from MFIs or banks or individuals, will relax the
intensity of this constraint. Hence, it will enable the households to broaden the portfolio
mix, which in turn will have positive impact outcomes at the household level. Outcomes
may be classified into intermediate-outcomes and end-outcome. Intermediate outcomes
include income, consumption expenditures, assets accumulation, savings, child education,
nutritional intake and employment. Poverty reduction is the end-outcome. All the
intermediate outcomes have a compounded impact on the end-outcome. However,
feminists, development sociologists and women professionals may consider female
empowerment an end-outcome as well. But given the historical underlying reason of
the emergence of microfinance, poverty-reduction is the end-outcome. Women’s empow-
erment may be termed a spillover impact of intermediate outcomes. At the enterprise level,
impacts will increase in terms of revenue, profit, assets, and employment, among others
(Chen and Dunn, 1996; Dunn and Arbuckle, 2001; Sinha, 1998). At the individual level,
impacts are found in women’s empowerment and increase in personal savings among
others In addition, there will always have spillover impacts—increase in wages; invest-
ment (Khandker, 1998a; Hulme, 2000).
Finance will always have positive impact at the household level or at the individual level
when such credit is more structured and targeted with strict monitoring. Because of such
obvious impact, professionals often question the need for impact assessment (as quoted in
Morduch). Nevertheless, there is still the need for assessing the extent of outcome on
several grounds. First, financial resources should be allocated efficiently so that there can
be a net positive welfare gain. Second, the microfinance programme impact can be
externally validated for continuity in intervention. Third, effectiveness of microfinance
can be compared against the rate of return on alternative uses, which in turn will contribute
to efficient allocation of resources. Fourth, there will be deeper understanding of the
process through which microfinance intervention benefits poor households and the process
through which spillover impacts occur at the village and regional level. On the top of these
arguments, donors have a special preference for impact analysis in order to justify their
investment to their taxpayers. Impact studies are donor-driven. They have little utility to
the MFIs (Hulme, 2000). Nevertheless the findings add to our knowledge. How credible is
the knowledge? We now examine this question.
334 M. A. Baqui Khalily
Copyright # 2004 John Wiley & Sons, Ltd. J. Int. Dev. 16, 331–353 (2004)
METHODOLOGICAL DEBATE IN DEMAND SIDE ANALYSIS:
WHAT DO WE LEARN?
Quite a large number of studies have been conducted on the impact of microfinance or
micro credit at the household level during the past fifteen years. Morduch and Haley
(2002) have done a comprehensive review of the impact literature. The researchers in the
AIMS project of USAID conducted a review of the selected major studies. In Bangladesh,
Rahman (1998) conducted a comprehensive survey of the impact studies carried out in
Bangladesh. This section is drawn upon those reviews.
Fundamentally, there has been debate over the methodologies of impact studies. Such
debate arises because of possible selection bias, endogeneity of program placement and
fungibility of fund. As Gaile and Foster argue, impact research focuses on precision and
accurate assessment of impact. Hulme describes the methodological issue as complex
because of selection bias and endogeneity. Hulme (2000) identifies several sources of
selection bias: (i) locational selection; (ii) difference in ‘invisible’ attributes between the
treatment group and control group; (iii) Hawthorne effect of intervention on treatment
group; (iv) contamination of control groups by treatment groups; and (v) fungibility of
micro credit. Methodologies followed in different studies are not necessarily consistent
and unique, and have not been able to control the sources of selection bias. Nevertheless,
there is a consensus that programme endogeneity and selection bias have to be tackled
appropriately for reliable estimates of the impact.
The decision of individuals to participate or not to participate in microfinance
programmes is determined by the extent of incentives provided by microfinance institu-
tions, given observable and unobservable characteristics of the family and the individuals
including financial wealth. The personal and family characteristics, behaviour and
attitudes differ from individual to individual. Hence, they too emerge as the determinants
of programme participation. The known and observable characteristics can be defined and
measured, but there will be always be some unobservable characteristics that have
influence on programme participation. The extent of impact of credit is, however, also
impacted by programme placement. It is well evident that microfinance institutions place
their branches in accessible and infrastructurally developed areas (e.g., Khandker, 1995).
Hence, programme placement does have impact on the outcome at the borrower level.
However, identifying the socio-eco-political factors and environment in assessing pro-
gramme impact at the household level can control this endogeneity. Hulme argues that
with the exception of fungibility, other sources of selection bias can be tackled through
careful selection of control groups. Most of the studies (e.g., Pitt and Khandker, 1996;
Mustafa et al., 1996; Hashemi et al., 1996; Chen and Dunn, 1996; Dunn and Arbuckle,
2001; Coleman, 1999) used quasi-experimental design (treatment group and control
group) to estimate the effect of micro credit. The difference between the parameter
estimates of treatment and control groups is the effect of intervention. However, precision
depends on the ability to control for observable and unobservable characteristics.
The issue of fungibility of credit is a critical problem in precisely determining the impact
of credit (Adams et al., 1984; David and Meyer, 1983; Von Pischke 1983). This arises from
inability to separate out the uses of micro credit and other funds between households and
enterprise. No studies have successfully controlled for fungibility. It is a matter of concern,
as argued by Hulme (2000), when impact studies ‘focus exclusively on the enterprise’. The
problem is equally present when there is inseparability of enterprises and household
activities and behaviour in most of the poor households, and inseparability between micro
Microfinance Programmes in Bangladesh 335
Copyright # 2004 John Wiley & Sons, Ltd. J. Int. Dev. 16, 331–353 (2004)
credit and informal and/or formal credit. Khandker (1998b) has candidly recognized the
problem of fungibility in impact assessment. However, he argues that ‘the close monitor-
ing of group-based credit partly resolves the problem of fungibility’. Failure to control for
fungibility may over-estimate the impacts of micro credit intervention. This is particularly
true when households have access to multiple sources of credit (Khalily et al., 2002;
Husain et al., 1998; Khandker, 1998a; Sinha and Matin, 1998).
As the issue of fungibility is not fully tractable in the impact assessment of micro credit,
several other approaches have emerged for a better understanding of the process through
which micro credit intervention influences desired outcome. They are: case study, house-
hold economic portfolio and anthropological studies. Mosley (1997) positively argues for
extensive case study in order to understand the process and influence of interventions, and
identify the difference between intended use and actual use of micro credit in order to
control for fungibility. Similarly the anthropological approach to impact studies considers
households as system and interaction among the family members and other different
agents based on their utility functions (Gaile and Foster, 1996). This minimizes degree of
the so-called unobservable factors in ‘heavy-duty’ econometric studies. More information
is available in case-based and anthropological studies. However, a good combination of
case, anthropological and econometric approaches would provide comprehensive infor-
mation on the nature and process of intervention and its impact on desired outcomes.
Khandker and Faruquee (2001) used a different approach to study the impact of
agricultural credit on desired outcomes in Pakistan. They followed a two-stage approach.
In the first stage, demand for credit was determined using household-level and village/
regional-level characteristics. The predicted demand for credit was used as a variable in
the household welfare outcome models in the second stage. The variables included
instrumental variables and predicted demand for credit.
Although somewhat conventional in concept and approach, the AIMS Team used a
Household Economic Portfolio model. In the model, total household resources—inflow
and outflow—are accounted for. Three aspects—feminist, anthropology and economic—
were the underlying constituents of the framework. Based on both the case-study and
econometric approaches, the framework provides separate estimates of the impact of
microfinance on desired outcomes. The issue of fungibility is tackled within this approach.