the poverty impact of microfinance

8/12/2019 The Poverty Impact of Microfinance

1/5

The Poverty Impact of Microfinance

A perfect impact evaluation really needs to answer a counterfacutal question: how does thestatus of participants in the program compare with how those same individuals would havefared in the absence of the program? Or, alternatively, how would non-participants havefared in the presence of a program? The problem with cross-sections of data (observationson many individuals at a given point in time is, of course, that at any given point in timeindividuals are observed to be either participants or not! "ven panels of data (observationson many individuals through time are problematic since over time many other things havehappened to the individuals in addition to program participation and it is nearly impossibleto separate out the impact of the program from all the other influences! #n reality,researchers must settle for estimates of the average impact of the program on a group ofparticipants $ the treatment group - to a credible comparison group $ a control group! Theideal control group is individuals who would have had outcomes similar to those in thetreatment group had they not participated in the program!

%ut constructing a control group comparable to the treatment group is not straightforward!&articipants in the program are usually different from non-participants in many ways:programs are usually carefully placed in specific areas, participants within those areas maybe screened for participation, and the final decision on whether or not to participate isusually voluntary! To the e'tent that these factors are nown and can be measured, theycan be controlled for in the empirical analysis, but in most cases the placement of theprogram and self-selection of participants in those areas into the program are based onunobservable factors! These unobservable factors lead to at least two inds of bias in anyempirical impact evaluation: program placement bias and self-selection bias!

)ontrolling for this bias $ determining the effects of *ust microfinance and separating outthe impact of microcredit from what would have happened to the same household withoutcredit - is often the most difficult part of careful empirical impact studies! +ellrunmicrofinance institutions do not randomie either the location of their operations or theirselection of clients! #f microfinance institutions tend to operate in areas that have relativelybetter or worse infrastructure such as access by roads, or more or less active marets, thenestimates of the impacts of the program on participants do not measure the effects *ust ofmicrofinance, but of these other factors as well! "ven within a given village, if, as studies by)oleman (.., Ale'ander (../ and 0ashemi (/112 suggest, microfinance clientsalready have initial advantages over non-clients, then the impact of microfinance will be

overestimated if these initial biases are not controlled for! 3imilarly, the impact ofmicrofinance programs that deliberately target relatively disadvantaged households in theareas they operate may find impacts underestimated if these biases are not controlled for!

4espite the importance of thining carefully about these issues, few studies have addressedthem rigorously and for good reason, as rigorous quantitative studies, among otherimitations, are costly and time consuming!85ew microfinance institutions have theresources in terms of funds or staff-time to conduct them! There is a movement in the
http://www.adbi.org/book/2005/11/28/1564.khushhali.microfinance.study/endnotes/page1.php#8http://www.adbi.org/book/2005/11/28/1564.khushhali.microfinance.study/endnotes/page1.php#8


2/5

industry to create practitioner-friendly assessment tools (for e'ample, the #mp-Act pro*ectbased at the #nstitute of 4evelopment 3tudies at 3usse', 63A#4s A#73 pro*ect andassessment tools by )8A&, but these assessments, while very useful to the institutionsthemselves in refining their targeting, products and mareting, are not rigorous

quantitative measures of impact and do not adequately address the issues of selectionbias!9

Armendari de Aghion and 7orduch (..9:pp ;-1 provide a compelling argument tomae the substantial investment required to conduct careful impact studies that control forthese potential biases:


3/5

participant households are pulled above the poverty line annually!

The accuracy of the original results as presented in &itt and >hander (/11; has beendisputed on the grounds that the eligibility criteria of low land holdings was not enforced

strictly in practice! #n a reworing of the results focusing on more directly comparablehouseholds, no impact on consumption from participation is found (7orduch /111:/@.9!This debate, which in part centers around details of econometric estimation, has not beenresolved! An unpublished paper by &itt rewors the original analysis to address theconcerns of 7orduch and is said to confirm the original results (>hander .., footnote/!

Prospective Clients as Control Group

Another approach to controlling for self-selection and placement bias, used by 0ulme and7osley (/11@ and )oleman (/111 is to include a sample of microcredit clients who haveformed solidarity groups but have not yet received loans as the control group! #n thisapproach, participating and non-participating households are again surveyed in treatmentvillages where the microcredit program is already operating and has already given loans!The control villages are villages where the micro credit program will operate andhouseholds from the village have already self-selected to participate in the program buthave not yet actually received loans!

0ulme and 7osley (/11@ employ this approach in their study of programs in a number ofcountries including the 8rameen %an in %angladesh and the %an ayat #ndonesia (%#!#n general a positive impact is found on borrower incomes of the poor with on average an

increase over the control groups ranging from /.-/D in #ndonesia, to around .D in%angladesh and #ndia! 8ains are found to be larger for non-poor borrowers, however, andwithin the poorest group gains are negatively correlated with income!

0owever, despite the breadth of the study and its use of control group techniques, 0ulmeand 7osleyCs study fails to control for program placement bias, so part of the advantage ofprogram participants relative to the control group may be due to unmeasured villageattributes that affect both the supply and demand for credit!10

)oleman (/111 advances the literature by e'panding on this concept to control forselfselection bias and introduces both observable village characteristics and village fi'ed

effects to control for program placement bias in his study of a village-baning program inThailand! 6tiliing data on E99 households, including participating and non-participatinghouseholds in treatment villages where a village ban is already offering microcredit andselected future participants and non-participants in control villages that have beenidentified to receive a village ban program but have not yet actually received funds,)oleman uses a difference-in-difference approach that compares the difference betweenincome for participants and non-participants in program villages with the same difference in


4/5

the control villages, where the programs were introduced later!

)olemanCs study measures the effects of access to rather than participation in a microcreditprogram and finds no evidence that months of access to a village ban program has an

impact on any asset or income variables and no evidence that access to village ban loansincreased productive activity! The author cautions, however, against e'trapolating theseresults to other conte'ts since Thailand is a rather wealthy developing country! One of thereasons there is a wea poverty impact is that there was a tendency for wealthierhouseholds to self-select into village bans, and the relatively small sies of loans maymean that they were largely used for consumption!

This approach as well is not perfect! >arlan (../ points out that this approach still fails tocorrect for possible attrition bias $ the fact that the control group includes potential futuredropouts (or graduates of the program, while the treatment group of older borrowers (whohave in fact remained active borrowers does not! 4epending on the reasons for attrition,attrition bias can be positive or negative! #f attrition is due to successful clients graduatingout of microfinance into the formal financial sector, then impact will be underestimated! #fattrition is due to dropouts who find the program unhelpful or whose microenterprises fail,for e'ample, then impact will be overestimated! ArmendFri de Aghion and 7orduch (..9review a number of studies that find dropout rates between [email protected] per year in variousmicrofinance programs worldwide! "ven the lower-end estimates can add up to asubstantial effect over time!11

Randomized Program esign

There are a few very recent impact studies underway that use randomied study design tocontrol for selection bias! 4uflo and >remer (.. describe the use of this type ofevaluation for an educational program in 7e'ico! %aner*ee and 4uflo (in progress willapply this approach to a microfinance impact assessment for the )enter for 7icro 5inanceesearch ()75! This approach eliminates selection bias by randomly selecting treatmentgroups (those who receive microfinance and control groups (those who do not from apotential population of participants! +ith this type of study design, the researcher can beassured that on average those who are e'posed to the program are no different than thosewho are not, and thus that a statistically significant difference between the groupsCoutcomes can be confidently attributed to the program, not to selection bias!

+ell-designed studies of this sort have the potential to rigorously address all inds ofpotential biases, although they are limited by the fact that they can only estimate partialequilibrium treatment effects, which may differ from general equilibrium treatment effects!#n the case of microfinance, this means that if, for e'ample, microfinance is introduced on aarge scale, the program could eventually affect the functioning of financial marets andthus have a different impact than the necessarily smaller scale program introduced for theimpact study!


5/5

A more practical concern in attempting to apply randomied study design is that suchstudies require tremendous cooperation from the institutions being evaluatedG they must bewilling to allow researchers to randomie implementation of their services! 3uch studiesmust also be longitudinal, maing them costly, and it can be difficult to conduct research

over a time period long enough for some impacts to show up! #n the case of %aner*ee and4ufloCs study for )75, the time frame between base line and final study is one year, whichmay not be long enough for some of the impacts of microfinance to show up quantitatively!5or these reasons randomied studies are liely to continue to constitute only a tiny fractionof all microfinance evaluations!

!"e vie#s expressed in t"is paper are t"e vie#s o$ t"e aut"or%s and do not necessarily re$lect t"e vie#s or policies o$ t"e &sianevelopment 'an( )nstitute nor t"e &sian evelopment 'an(* +ames o$ countries or economies mentioned are c"osen by t"eaut"or%s, in t"e exercise o$ "is%"er%t"eir academic $reedom, and t"e )nstitute is in no #ay responsible $or suc" usage*

the poverty impact of microfinance

Documents