microfinance repayment performance in bangladesh: how to improve the allocation of loans by mfis

WorldDevelopmentVol. 32,No. 11, pp. 1909–1926, 2004� 2004 Elsevier Ltd. All rights reserved

Printed in Great Britain
0305-750X/$ - see front matter
doi:10.1016/j.worlddev.2004.05.011www.elsevier.com/locate/worlddev

Microfinance Repayment Performance in

Bangladesh: How to Improve the Allocation

of Loans by MFIs

MARIE GODQUIN *

TEAM, Universite Paris I—Pantheon Sorbonne, and CNRS, France

Summary. — The aim of this article is to produce a comprehensive analysis of the performance ofmicrofinance institutions (MFIs) in terms of repayment. We focus the analysis on the impact ofgroup lending, nonfinancial services and dynamic incentives on repayment performance. We testfor endogeneity of loan size and use instrumental variables to correct for it. In the second sectionof the paper, we use a comparative analysis of the determinants of the repayment performance andof loan size in order to make policy recommendations on the allocation of loans by MFIs.
� 2004 Elsevier Ltd. All rights reserved.
Key words — microfinance, social ties, group homogeneity, nonfinancial services, Asia, Bangladesh

* I am grateful to Sylvie Lambert, Pr. Jean-Claude

Berthelemy and Pr. Jacqueline Pradel for helpful

suggestions and encouragements. I also want to thank

two anonymous referees for their thoughtful comments.

Final revision accepted: 26 May 2004.

1. INTRODUCTION

The primary objective of microfinance insti-tutions (MFIs) is to provide financial services(credit and saving) to the poor in order to re-lease financial constraints and help alleviatepoverty. Each MFI tries to maximize its repay-ment performance, whether or not it is profit-oriented. High repayment rates are indeed lar-gely associated with benefits both for the MFIand the borrower. 1 They enable the MFI tocut the interest rate it charges to the borrowers,thus reducing the financial cost of credit andallowing more borrowers to have access to it.Improving repayment rates might also help re-duce the dependence on subsidies of the MFIwhich would improve sustainability. It is alsoargued that high repayment rates reflect theadequacy of MFIs� services to clients� needs.They limit the incidence of crosssubvention 2

across the borrowers. Last but not least, repay-ment performance is a key variable for donorsand international funding agencies on whichmany MFIs still depend for their access tofunds.The first-best level of repayment performance

is a perfect (100%) on-time repayment rate. Ifthe maximum repayment rate the MFI canreach given its lending methodology is lowerthan the targeted 100%, the MFI will use

190

second-level strategies to increase its repaymentperformance. Such strategies include the alloca-tion of larger loans to borrowers with lower de-fault probability and attempts to reduce thedelay in repayment. The MFI will developincentive mechanisms so as to meet these objec-tives.The main factors influencing repayment are

either related to information asymmetries, toadverse shocks affecting the borrower, or tothe low performance of institutions such as jus-tice or education. Information asymmetriesarise when gaining information on the charac-teristics or on the behavior of the borrower iscostly for the MFI. Information asymmetriesgenerate problems of adverse selection––alloca-tion of loans to borrowers with undesirablecharacteristics such as a high level of risk orinability to take advantage of the loan- as wellas moral hazard—the borrower may behave inan undesirable way (make little or insufficient

9

1910 WORLD DEVELOPMENT

effort to take advantage of his loan or used itfor unproductive purposes). Adverse selectionand moral hazard increase the proportion ofborrowers who cannot repay their loans ontime. Borrowers that have enough money toreimburse their loan might also default strategi-cally. The cost of strategic default might indeedbe low if the lending institution has low collat-eral requirements and if the legal system giveslittle power to the MFI to enforce contracts.MFIs try to restrict the occurrence of thosethree types of situations in designing appropri-ate credit schemes.In this paper, we wish to contribute to the

improvement of MFI repayment performanceby examining their determinants with a partic-ular emphasis on ‘‘microfinance innovations’’such as the use of nonfinancial services, grouplending, and dynamic incentives. This paperalso questions the adequacy of loan allocations(in terms of loan size) based on the comparisonof the determinants of the repayment perform-ance to the determinants of the loan size. Thiswork uses an objective repayment variable,i.e., a repayment variable based on the declara-tion of the borrower (not the MFIs). It ad-dresses the endogeneity problem of theprincipal of the loan.The results indicate that the use of nonfinan-

cial services has a positive impact on microfi-nance repayment performance but that grouphomogeneity and social ties among groupmembers are not always associated with a bet-ter repayment performance.This article is organized as follows: after a

brief presentation of the conceptual framework(Section 2), Section 3 briefly reviews the empir-ical literature and Section 4 provides details onthe context of the case study. Section 5 presentsthe econometric methodology. The results ofthe regression models are discussed in Section6 and the article concludes with implicationsfor policy recommendations and future re-search.

2. IMPROVING REPAYMENTPERFORMANCE: THE CONCEPTUAL

FRAMEWORK

(a) The first best level of repayment performance

Credit rationing and collateral requirementare the traditional means used by banks to copewith information asymmetries in the creditmarket (Stiglitz & Weiss, 1981) but both meth-

ods lead to the exclusion of poor borrowers. Toexplain the success of microfinance in providingcredit to the poor, a large literature uses theprincipal/agent theory to demonstrate thatmicrofinance contracts lending to joint-liablegroups allow the lender to bypass moral hazard(Stiglitz, 1990) and adverse selection (Ghatak,1999) due to information asymmetries. It is alsoargued (Besley & Coates, 1995) that joint-liablelending groups help enforce repayment as so-cial interactions make strategic default morecostly. Social ties (Besley & Coates, 1995) andgroup homogeneity (Besley & Coates, 1995; Sti-glitz, 1990) are also indirectly linked to repay-ment performance as they can facilitate peermonitoring and peer pressure or result froman effective peer selection of group members.Regular repayment schedules (Armendariz deAghion & Morduch, 2000) or dynamic incen-tives 3 (Besley, 1995) are other appropriateincentive mechanisms used by MFIs to increasetheir repayment performance. The provision ofnonfinancial services as a complement to creditand saving services (Edgcomb & Barton, 1998)not only develops the economic ability of theborrower to repay but also makes the relation-ship with the MFI more valuable to him. Theprevious mechanisms are considered to befinancial innovations (Edgcomb & Barton,1998) making it financially sustainable forMFIs to lend to the poor. When the use of suchmechanisms fails to enable the MFI to reach aperfect repayment rate and when borrowers areheterogeneous in their default probability, theMFI could also allocate loans of different sizesto the borrowers in order to maximize the valueof outstanding debts repaid on time. In the fol-lowing section, we explain why borrowers areinterested in larger loans and why the MFIsshould allocate larger loans to borrowers witha lower default probability.

(b) The second-best perspective: increasing thevalue of outstanding debts repaid on time

(i) The contextWe consider a microfinance institution pro-

viding credit to joint-liable credit groups at auniform interest rate. The MFI deals with cred-it applications coming from borrowers hetero-geneous in their localization, lending group,ability and preferences. The aim of the MFI isto maximize the global net expected return ofits borrowers under a zero profit condition.Where borrowers face high credit rationing,

there is a large set of highly productive projects

Size of the

Loan (L)

Expectedprofit

L max =L*

Figure 1. Optimal size of the loan (demand size).

MICROFINANCE REPAYMENT PERFORMANCE 1911

and therefore an increasing mean capital pro-ductivity. The expected profit for the borrowerwill thus increase with the loan size for a givenduration as shown in Figure 1. 4

The borrower may acquire the informationon the cash requirement, expected return, andprobability of success of the different projectshe may manage before applying for a loan.The projects should be such that the sum ofthe return to the project and of the external re-turns is higher than each installment. This set ofprojects is possibly restricted by the local envi-ronmental factors such as the distance frommarketable activities, the economic situation,and the pre-existing competition of similar pro-jects in the area. The set of accessible projectsfor a given borrower may be further restrictedby his lending group (Madajewicz, 1999) asthe group members may incite the borrowerto undergo a project similar to the other mem-bers� projects in terms of size, activity and prob-ability of success. Such a project would indeedbe easier to monitor and would limit the devia-tion from the group�s mean default probabilityand the consequent group cross-subsidization.

Default

probability

Pmin

Figure 2. Heterogeneity in

(ii) The behavior of the borrower: demand forcredit and repaymentWe consider a borrower who is given the

opportunity of obtaining a loan from a micro-finance institution. The loan application ofthe borrower corresponds to the size and dura-tion that maximizes his expected return. 5

The optimal duration depends on the distri-bution over time of the returns of the project,on the scale of the project, on the preferencesof the borrower for the present consumption,and on the allocation of a new loan condi-tional to the total repayment of the previousloan. There is usually little flexibility in theduration of microcredit loans and the durationof the loan may be taken as an exogenous var-iable for a borrower. For a given duration,each loan size corresponds to a single projectas the borrower will undertake the project thathas the higher expected return for each loansize.As the net return is an increasing function of

the size of the loan, the borrower always pre-fers bigger loans and therefore asks for thelargest loan size he may apply for (Lmax in Fig-ure 1) given the set of projects within hisreach––defined by her own characteristics,those of her environment, and those of herlending group.For a given borrower and duration of the

loan, it is argued (Freimer & Gordon, 1965)that the repayment probability decreases withthe size of the loan as shown in Figure 2 wherePmin represents the minimum default probabil-ity explained by external factors such as illnessor accidental destruction of the borrower�s pro-ductive assets. The increase in the default prob-ability with the loan size differs acrossborrowers according to their initial endow-ments and moral hazard or strategic defaultassociated costs.

L

High risk borrower (h)

Low risk borrower (l)

the default probability.

Defaultprobability

LLl*

Ptarget

Pmin

L h*

High risk borrower (h)

Low risk borrower (l)

Figure 3. Optimal size of the loan (supply side).


(iii) The MFI: Increasing the value of outstand-ing debts repaid on timeIf the maximum repayment rate the MFI can

reach given its methodology is lower than thetargeted 100%, the MFI will have to define anew target default probability. It will then allo-cate loans to borrowers only if the defaultprobability of the loan they are asking for isinferior to the new targeted default probability.If there is observable heterogeneity in therepayment probability of borrowers, the MFIwill allocate larger loans to safer borrowers asshown in Figure 3.

3. EMPIRICAL LITERATURE REVIEW

Relying on the theoretical literature onmicrocredit we expect joint liability––especiallythrough peer selection, peer monitoring, andpeer pressure––to be associated with a betterrepayment performance. Group homogeneityand social ties are also expected to increasethe repayment performance through a greaterefficiency due to group dynamics. Grouphomogeneity as a result of effective peer selec-tion (group homogeneity in terms of risks,Ghatak, 1999) and as a means to increase peermonitoring (group homogeneity in terms ofinterest, economic power, Stiglitz, 1990) shouldimply higher repayment rates. High levels of so-cial ties should have the same impact becausethey facilitate peer monitoring and increasethe potential social sanction of peer pressure(Besley & Coates, 1995). Dynamic incentivesand the use of nonfinancial services, which aremicrofinance innovations not based on grouplending, are also expected to increase the repay-ment performance.Addressing the question of the relative per-

formance of group loans compared to individ-

ual loans and using data from Zimbabwe,Bratton (1986) states that group loans performbetter than individual loans in years of goodharvest and worse in drought years when peersare expected to default. Paxton (1996) analyzeswith a mean and covariance structural modelthe determinants of successful group loanrepayment of 140 credit groups in BurkinaFaso. She draws attention to a negative ‘‘dom-ino effect’’ 6 that can outweigh the positive ef-fects of group lending. Zeller (1998) usesinformation on 146 credit groups in Madagas-car and provides evidence in favor of grouplending. In his article, Zeller shows that thegroup generates insurance, which leads to a bet-ter repayment performance.Analyzing the potential positive effects asso-

ciated with group dynamics, some studiesexamine the impact of different levels of peerselection, peer monitoring and peer pressure.Wenner (1995) presents a methodology to testwhether the selection mechanism has an impacton the repayment performance of 25 Costa Ri-can credit groups and whether group membersuse local information for the screening of theirpeers. His study shows that lending groups useprivate information to select their peers andthat this selection mechanism increases thegroup repayment performance. 7 On the sameissue, the above-mentioned study of Zeller(1998) confirms the positive role of peer selec-tion (internal rules of conducts) on repaymentperformance. Wydick (1999) uses data from137 Guatemalan credit groups to show how so-cial cohesion affects group performance interms of repayment rates, group insuranceand moral hazard. He found that peer monitor-ing in urban groups and peer pressure in ruralones significantly affects group performance.Contrary to the conclusions of the previousthree articles on the positive impact of group


dynamics, Diagne, Chimombo, Simtowe, andMataya (2000), working on data from Malawi,found that peer monitoring, peer pressure andjoint liability had little or a negative impacton repayment performance and that peer selec-tion was found to be limited.Social ties and group homogeneity are sup-

posed to improve the power of group dynamics.Nevertheless, the empirical studies regardingthis issue produced mixed results. The studyof Sharma and Zeller (1997), based on the anal-ysis of repayment rates of 128 credit groups inBangladesh, leads to a controversial negativeimpact of preexisting social ties as well as grouphomogeneity in terms of asset and enterprisediversity. The study of Zeller (1998) investigatesthe effects of intragroup pooling of risky assetsor projects on repayment rates. While this anal-ysis supports the positive role of social cohe-sion, it also concludes that risk diversification(up to a certain level) has a significant positiveeffect on repayment performance. This could beexplained by a matching problem (Paxton,1996). The matching problem arises when cred-it terms and conditions are no longer appropri-ate to each group member�s needs as credit isrenewed. If initial group homogeneity and priorexperience of group activities were associatedwith better repayment performance of the firstloans, as time goes by, the negative impact ofmatching problems and the absence of riskdiversification which limit the possibilities ofintragroup insurance should outweigh the pos-itive effect of group homogeneity on peer mon-itoring.Along with group lending, MFIs usually use

dynamic incentives and nonfinancial services.MFIs are said to use dynamic incentives whenthey increase the amounts lent to a specific bor-rower as credit is renewed, and condition theallocation of new loans to previous repaymentbehavior. Some microfinance programs are alsoreferred to as ‘‘credit plus’’ (Edgcomb & Bar-ton, 1998; Zohir et al., 2001) as they provideservices (such as health services or adult liter-acy) or training that go beyond financial serv-ices. In contrast to group lending, these twomain features of the microfinance methodologyhave been little documented up to now.In the study of Diagne et al. (2000), the most

important factor inciting lending groups to re-pay is the relative value they attach to accessto future credit. For Sharma and Zeller(1997), credit rationing, up to a certain level,has a significant positive effect on repaymentperformances. In a study on the Grameen

Bank, Khandker, Khalily, and Khan (1995)found that the longer the branch operates inan area, the higher the loan default rate. Theyexplain this feature by the possible decreasingmarginal profitability of new projects. Thiscould also be due to a deceasing power of dy-namic incentives as credit is renewed over timeespecially if borrowers observe that credit is notsystematically denied to defaulting or late bor-rowers. Khandker et al. (1995) also found thatmembership training, which relates to nonfi-nancial services, had a positive influence onrepayment.Most of the studies on the determinants of

repayment rates also introduce control varia-bles on the characteristics of the area and ofthe borrower. Khandker et al. (1995) raise thequestion of whether default is random, influ-enced by erratic behavior, or systematicallyinfluenced by area characteristics that deter-mine local production conditions or branch-le-vel efficiency. Their empirical test on Grameenoverdue loans supports the idea of partialinfluence of area characteristics. Rural electrifi-cation, road width, primary educational infra-structure and commercial bank density arepositively correlated with a low default rate aswell as the predicted manager�s pay. Paxton(1996) shows also that access to other creditsources, market selling activities and urbanlocation were linked to a better repayment per-formance. Questioning the impact of the char-acteristics of the borrower, Zeller (1998)showed that traditional prejudices against wo-men, young borrowers or large families shouldnot influence the determination of repaymentability. Matin (1997), analyzing the determi-nants of the repayment performance of Gram-een Bank borrowers, found that multiplenongovernment organization (NGO) member-ship, which he associates to access to othersources of ‘‘cheap’’ finance, had a negative im-pact on repayment performance. He also foundthat education and the area of the operatedland, which could be proxies for wealth of theborrower, had a positive impact on repaymentperformance. The membership period was pos-itively associated with default while loan sizedid not have a significant impact on repaymentperformance.The above-mentioned studies attribute a

questionable role to mechanisms related togroup lending in the explanation of repaymentperformance of microfinance whereas the roleof nonfinancial services and dynamic incentivesis very little documented. It is thus important to

Table 1. Other source of credit

Other credit providers (353 observations)

Government 2.8%

Krishi Bank 10.2%

Commercial Bank 10.5%

Cooperative 2.3%

Other NGO 7.4%


produce further tests of the impact of thesefinancial innovations on repayment perform-ance. This article provides a test of the explana-tory power of social ties and group homogeneityas well as nonfinancial services and dynamicincentives. The impact of the main characteris-tics of the loan contract and of the borrower isalso taken into account.

Relatives 33.1%

Friends and neighbor 21.8%

Shopkeeper 3.7%

Landlord 5.1%

Other 3.8%

4. DATA

The data come from a quasi-experimentalsurvey carried out in Bangladesh in 1991–92by the BIDS (Bangladesh Institute for Develop-ment Studies) and the World Bank. This surveywas designed to assess the impact of microfi-nance on poverty outcomes in Bangladesh(Morduch, 1998; Pitt, 1999; Pitt & Khandker,1996). The survey covered 1798 households,coming from 87 villages from 29 different tha-nas (subdistricts). 8 Of these households 1,538were ‘‘eligible’’ for microfinance programs,which means that they were poor enough (pos-sessed less than 0.5 acres of cultivable land) tobenefit from microfinance services. Nine hun-dred and five of these households actually tookpart in one of the three available microfinanceprograms, namely the BRAC, the BRDB andthe Grameen Bank. For the purpose of ourwork, we concentrate our interest on house-holds that had actually borrowed from one ofthe three MFIs. We use 2,349 loan observa-tions, 485 of which corresponded to BRACloans, 430 to BRDB loans, 1,081 to GrameenBank loans and 353 to other credit providersloans (see Table 1 for the description of othercredit providers).Based on the due date of the loans and on the

date they were fully repaid, we were able toconstruct different dummy variables for indi-

Table 2. Different measure

Repayment rates MFIs BRAC

ROT 0.50 (1,700) 0.37 (388)

RTG3 0.76 (1,591) 0.65 (341)

RTG6 0.84 (1,479) 0.74 (315)

RTG12 0.94 (1,355) 0.90 (270)

ROT: Dummy = 1 if the borrower repaid his loan on time.arrears inferior to three months. RTG6: Dummy = 1 if thmonths. RTG12: Dummy = 1 if the borrower repaid his loa The number of loan observations available for the calcultheses.

vidual repayment performance. These variablesconstitute real repayment indicators since theyare based on the declaration of the borrowersand are not transformed by the microfinanceprograms based on their definition of laterepayment. As shown in Table 2, the on-timerepayment rate is significantly lower than thefigures (95% or 98%) normally used in microfi-nance programs� communications. But when a12-month grace period is given to the borrower,the repayment rate increased from 50% to94%. 9 Table 2 also shows that whatever thegrace period, microfinance programs experi-ence significantly higher repayment rates thanother credit providers (14% on time repay-ment. 10)The principal characteristics of the loan con-

tract of microfinance programs and of otherlenders are also relatively different from eachother (see Tables 3 and 4).Microfinance programs allocate significantly

smaller loans than other credit providers.Around 70% of microfinance loans were size-rationed 11 compared to only 30% of the otherloans. We have no information on the otherdimensions of rationing such as rejected or dis-couraged loan applications.

s of the repayment ratea

BRDB GB Other credit providers

0.50 (317) 0.55 (995) 0.14 (303)

0.65 (286) 0.84 (964) 0.37 (172)

0.77 (254) 0.89 (910) 0.43 (153)

0.90 (234) 0.96 (851) 0.60 (119)

RTG3: Dummy = 1 if the borrower repaid his loan withe borrower repaid his loan with arrears inferior to sixan with arrears inferior to twelve months.ation of the different repayment rates is given in paren-

Table 3. Descriptive statistics of the microcredit contract

Variables Description Predicted impact

on repayment

Number of

observations

Mean Standard deviation Min Max

PRIN Size of the loan (in taka) – 1996 2931.057 1381.296 1,000 10,000

DURATION Duration of the loan (in days) 1981 400.371 187.873 0 2542

BRAC Dummy := 1 if the loan was taken from the BRAC 1996 0.243 0.429 0 1

BRDB Dummy := 1 if the loan was taken from the BRDB 1996 0.215 0.411 0 1

SEX Dummy := 1 if the borrower is a male;

2 if the borrower is a female

1996 1.661 0.474 1 2

PASSET Value of productive assets of the

household (in taka)

1996 35756.800 57856.670 0 459,001

SELFEAGR Dummy := 1 if the borrower received income from

agricultural self-employment

1996 0.712 0.453 0 1

NBLR Number of landed relatives 1996 2.809 3.309 0 13

AGEGP Age of the borrowing group at the due

date of the loan (in months)

+ 1949 33.177 22.104 0 114

SAMEEDU Dummy := 1 if the borrower and group leader

have the same education level (more or less two years)

+ 1996 0.578 0.494 0 1

SAMEAGE Dummy := 1 if the borrower and group

leader have the same age

+ 1996 0.394 0.489 0 1

NFSL Dummy := 1 if the loan program provided

the borrower with access to basic literacy

+ 1996 0.678 0.467 0 1

NFSH Dummy := 1 if the loan program provided

the borrower with access to a primary health facility

+ 1996 0.696 0.460 0 1

CRD Dummy := 1 if the borrower would

have liked to borrow more at the same interest rate

+ 1996 0.712 0.453 0 1

PREVLOAN Size of the previous loan (in taka) 1996 1621.738 1700.985 0 8000

MIC

ROFIN

ANCEREPAYMENTPERFORMANCE

1915

Table 4. The loan contract of other credit providers

N Mean SD Min Max

PRIN 353 5753.853 17961.540 1,000 190,000

DURATION 203 356.143 350.453 0 2772

INTEREST RATE 353 36.255 53.648 0 300

SEX 353 1.102 0.303 1 2

PASSET 353 68135.710 105246.500 0 609,602

SELFEAGR 353 0.816 0.388 0 1

NBLR 354 3.500 3.579 0 15

CRD 353 0.309 0.463 0 1


We were able to reconstruct the credit historyof microfinance borrowers with their MFI. Wefound evidence of variation in the size of loansto different borrowers for a given loan cycle (seeTable 5), which seems to prove the existence ofa certain flexibility in the determination of loansize by the credit agent. This might partly re-flect the fact that the studied MFIs use differentloan size ranges for a given loan cycle depend-ing on the location of the borrower in a munic-ipality, a semi-rural or a remote area. For thethree studied MFIs, however, the variance ofattributed loan size increases with the loan cy-cle for the first three to four cycles (the varianceof the loan cycle then declines due to a signifi-cant decrease in the number of observations),

Table 5. Variation of the size of the loan with the loancycle

Mean Std. dev. Min Max N

BRAC loan cycle

1 2048.566 1147.083 1,000 6,000 274

2 2888.553 1411.682 1,000 8,000 150

3 3707.317 1600.686 1,500 7,000 41

4 3785.714 1155.493 1,500 5,000 14

5 3750 758.287 3,000 5,000 6

BRDB loan cycle

1 2036.09 727.709 1,000 6,000 266

2 2828.467 1093.83 1,000 6,000 137

3 3552.632 1289.862 2,000 6,000 19

4 5,800 2167.948 3,000 8,000 5

5 6,000 1732.051 5,000 8,000 3

Grameen Bank loan cycle

1 2253.667 947 1,000 10,000 300

2 2963.736 929.047 1,000 6,000 273

3 3727.848 1192.662 1,000 10,000 237

4 4202.532 1148.3 1,000 6,500 158

5 4777.108 1003.813 2,000 6,500 83

6 5166.667 784.464 4,000 6,500 27

7 5333.333 1154.701 4,000 6,000 3

which would imply that older groups becomeless homogeneous with time in terms of grantedsize of the loan. The mean, minimum and max-imum of the loan size increases with the loancycle, reflecting the effective use of dynamicincentives (the loan size increases at each loancycle conditioned on the repayment of the pre-vious loan).Only 0.7% of the loans provided by microfi-

nance programs had no fixed due date com-pared to 40% for the other credit providers.There is little variation in the duration of theloans of microfinance programs: 88.13% ofthe loans have a duration from 11 to 13 monthscompared to only 33% of the loans from theother credit providers.There is no variation in the interest rate of

microfinance programs, which was uniformlyfixed at 16% per annum for the three studiedMFIs (20% after 1991 for the BRAC and theGrameen Bank due to an increase in bankemployees payment).The interest rate of the other credit provid-

ers is variable and ranges from zero to 300%with 34% of interest-free loans 12 and 30% ofthe loans with an interest rate superior to20%.Microfinance programs are less flexible

regarding the purposes of loans they financewith around 85% of the loans obtained fornonagricultural/business purposes; 6% foragricultural purposes and 9% for personal use(respectively 24%, 41% and 35% for the othercredit providers). Several studies however, pointout the differences between stated purposes ofthe loan and effective ones. 13

The growth of the loan portfolio in the sam-ple was similar to the effective growth of theportfolio of the MFIs. The annual growth rateof the Grameen Bank�s loan portfolio was of32.73% during 1986–91 in our sample com-pared to 37.8% according to calculations basedon Khandker et al. (1995).

Table 6. Provision of nonfinancial services besides credit

BRAC BRDB GB

Panel A. Access to nonfinancial services

Access to primary health 56.7 48.14 80.57

Access to basic literacy 66.39 32.79 85.66

Access to marketing information 6.8 20.23 40.43

Access to occupational and skill training 32.78 72.79 8.05

N 485 915 430

Panel B. Number of nonfinancial services the borrower has access to besides credit

No access to nonfinancial services 0.82 1.4 2.13

Access to one NFS 25.36 42.09 18.41

Access to two NFS 31.75 5.12 31.54

Access to three NFS or more 42.06 51.4 47.92

N 485 915 430

Panel C. Type of NFS provided if borrower has access to only one NFS besides credit

Access to primary health 39.84 3.87 36.68

Access to basic literacy 25.2 6.63 47.24

Access to marketing information 0 1.1 0

Access to occupational and skill training 0 58.01 0


Our findings show few differences among theMFIs in terms of the characteristics of the loansor types of borrowers. 14

There were, however, differences in the provi-sion of nonfinancial services (see Tables 6, Pan-els A–C). Four types of nonfinancial serviceswere provided: primary health, basic literacy,marketing information and occupational andskill training. Almost all the loans were associ-ated with access to at least one nonfinancialservice. Less than 7% of BRAC borrowershad access to marketing information and only8% of Grameen borrowers had access to occu-pational and skill training. This is why we willfocus on basic literacy and primary health inthe empirical strategy.

5. THE ECONOMETRIC FRAMEWORK

Following the discussion on the theoreticalliterature, we describe the interaction betweenthe borrower and the MFI with the followingmodel:Stage one: The borrower applies for a loan of

a given size, which corresponds to the largestone he/she can expect given the projects withinhis/her reach, and defined by his characteristics,the characteristics of his/her environment, andthose of his/her lending group.

Stage two: Before allocating a loan to theborrower, the credit officer of the MFI assessesthe default probability of the applicant usingthe information he has on the borrower, thelending group, the environment, and the pre-dicted effectiveness of the MFI�s repaymentincentives for this borrower. If the predicted de-fault probability is inferior to the maximum de-fault probability accepted by the MFI, he willallocate the loan the borrower applied for; oth-erwise, the credit agent will allocate a smallerloan corresponding to the maximum accepteddefault probability for this borrower.Stage three: The borrower reimburses the

loan on time or not, given his/her environment,his/her characteristics/ability, the characteris-tics of his/her group and the characteristics ofthe loan contract.The breakdown in three steps of the micro-

credit relationship sheds light on the possibilityof endogeneity of the principal characteristics ofthe microcredit contract 15 in the estimation ofrepayment. The determination of the size of theloan in stage two and the determination of therepayment in stage three might indeed be basedon the same omitted variables—variables ob-served by the MFI and the borrower but notavailable in our dataset (such as characteristicsof the environment, credit investigation madeby the credit agent on the credit history of the


borrower with other lenders, and the localize ofthe borrower).We constructed an individual dummy varia-

ble for on-time repayment and use a probitmodel to estimate the probability for a bor-rower to repay his loan at the due date. 16 Weused the method of Smith and Blundell (1986)to test for exogeneity in the probit model. Endo-geneity of the size of the loan could not be re-jected and the size of the loan was instrumented.The previous discussion results in the follow-

ing estimation strategy:Step one: Estimation of the size of the loan:

P i ¼ bP i þ epi ¼ ap þX4

j¼1

bpjX ij þ

X2

j¼1

qpj Y ij þ kpZi

þXj

rpjW ij þ cpIVp þ epi ð1Þ

where Xj represents the variables associatedwith the incentive structure of group lending,social ties and group homogeneity here. Weused the age of the group at the time the loanwas due (i.e., the number of months betweenthe date the group was created and the duedate) for intragroup social ties. We postulatethat the knowledge of the other members� char-acteristics and behavior as well as the level ofintragroup social ties are likely to increase withthe age of the borrowing group (AGEGP). Thisis why we expect the ability of the members ofthe group to monitor and pressure each otherto increase with the age of the group. The var-iable AGEGP should thus have a positive im-pact on repayment performance. Thevariables of group homogeneity are based onshared characteristics (age, education level) ofthe borrower and its group leader. As discussedin the conceptual framework, we expect thevariables of group homogeneity (SAMEEDU,SAMEAGE) to have a positive impact onrepayment performance.Yj describes the variables of nonfinancial

services. We use the access to basic literacy(NFSL) and access to primary health (NFSH)as nonfinancial services provided by the MFIs.Access to such services is expected to have apositive impact on the repayment as discussedin the conceptual framework.Zi is a variable for dynamic incentives prox-

ied by credit rationing. The incentive powerof dynamic incentives (allocation of new andlarger loans conditional on previous repay-ment) is expected to be greater for credit-ra-tioned borrowers and the variable (CRD) is

expected to have a positive impact on therepayment.Wj stands for the exogenous control varia-

bles. Control variables gather characteristicsof the borrower and his/her household and ba-sic information on the loan (dummy for theMFI and duration of the loan).IVp represents the instrument used for the

size of the loan. We use the value of the previ-ous loan as an instrument for loan size. The va-lue of the previous loan is likely to be a goodinstrument for the loan size as it should not af-fect the repayment of the present loan and as itwas determined on the same unobservables asthe present loan.Step two: Smith and Blundell�s (1986) exoge-

neity test for principal in the determination ofthe repayment performance:

Ri ¼ aþ xbP i þX4

j¼1

bjX ij þX2

j¼1

qjY ij þ kZi

þXj

rjW ij þ gepi þ ei ð2Þ

where Ri, the latent variable of the model, is theborrower�s capacity to generate cash in excessto the amount (principal plus interests) he hasto repay before the initial due date.We observe the reimbursement R�

i whichtakes the value of 1 if Ri > 0 and 0 if Ri < 0.Exogeneity is rejected if the coefficient of the

error (g) of the instrumented regression of theprincipal is significant. This would indeed meanthat the structure of the error term is the fol-lowing: ei ¼ aepi þ l.The test could not reject the endogeneity of

the loan size in the determination of the repay-ment and confirms that we used an appropriateinstrumental variable for the size of the loan.Step three: Estimation of the repayment per-

formance:

Ri ¼ aþ xbP i þX4

j¼1

bjX ij þX2

j¼1

qjY ij þ kZi

þXj

rjW ij þ ei ð3Þ

Step four: Larger loans, for whom?After a comment on the regression of the

repayment probability, we compare its deter-minants to those of the size of the loan. If


we assume that the loan size reflects theMFI�sperception of the capacities of the bor-rower, this allows us to analyze the adequacyof the loan allocation by the MFI.Because, as previously noted, the loan size is

a result of both demand (stage one) and supply(stage two) factors, we cannot simply assumethat the loan size reflects the abilities the MFIattributes to a specific borrower. We must con-sider separately the following cases:

(a) If the demand of the borrower in termsof loan size is higher than the final loan size,the loan size reflects the perception of theMFI.(b) If the demand is equal to the allocatedloan size, it could be that the MFI wouldhave granted the borrower a larger loanhad he or she asked for it. In this case wedo not know if the loan size reflects the exactperception the MFI has of the abilities of theborrower.The estimation of the loan size used for this

step was thus based on a sample restrictedto those borrowers who were credit rationed. 17

(around 70% of the sample)Since the size of the loan is likely to increase

automatically with the loan cycle (use of dy-namic incentives), we completed this analysisof the determinants of the loan size by the anal-ysis of the determinants of the variation of theloan size for different loan cycles. We used thedeviation of the loan size from the mean loansize of the corresponding loan cycle and MFI(specification II) and the loan size for the sub-sample of the first (specification III) and the

second (specification IV) loan cycle. 18

6. RESULTS AND DISCUSSION

(a) The repayment behavior of the borrower

The results of the probit estimations of therepayment behavior are reported in Table7.The model was implemented with a subsetof the database that included only those loanobservations whose due date had passed atthe time of the survey and for which informa-tion on all the explanatory variables was com-pletely available. Table 7 presents fivespecifications of the empirical model of Eqn.(3). The first column provides the results ofthe restricted model in which the instrumentedsize of the loan and the exogenous control var-iables are introduced. Group level variables,nonfinancial services and dynamic incentives

are separately added to this restricted modelin columns two to four. The last column standsfor the complete model.For each of the five specifications, the Smith

and Blundell test could not reject the endogene-ity of the size of the loan in the determinationof the repayment performance and the princi-pal was instrumented with the size of the previ-ous loan and the variables of the specification.The size of the loan (PPRIN) showed the ex-

pected negative sign and is significant in the fivespecifications as in the study by Sharma andZeller (1997). This negative sign is theoreticallyexplained by the fact that the loan size increasesthe gain associated with ex ante and ex postmoral hazard. Table 2 showed however, thatmost of the unpaid loans at the due date werefully repaid one year after the due date. In thissetting, moral hazard should be understood asthe choice of a project with a longer maturity(and higher expected value) than the scheduledduration of the loan rather than a choice of ariskier project. The negative sign of the loansize of the loan could also be linked to the bor-rower�s difficulty in repaying a larger amountover a given period (usually one year). It couldbe that, for a given duration, large loans do notmeet the borrower�s needs and are not suited tothe local economy. This statement is to be re-lated to the positive and significant sign of theduration of the loan (DURATION) throughoutthe specifications.The negative and significant signs of the

dummies for the BRAC and the BRDB indicatethat the Grameen Bank has a better repaymentperformance than the other two MFIs.In our sample, female borrowers (SEX) did

not prove to have a significantly better repay-ment performance. Even though the coefficientis positive, it is not significant. My results can-not justify the priority given to women inmicrofinance programs based on better repay-ment performance of women as is sometimedone. Using the same dataset, Pitt and Khand-ker (1998) found that female borrowing had abetter impact on poverty reduction than maleborrowing. This might be a greater argumentin favor of female borrowing. The fact thaton average women have lower default probabil-ities could be partly explained by their loweraverage loan size. The value of the productiveassets of the household (PASSET), the dummyfor self-employment in agriculture (SELF-EAGR) and the number of landed relatives(NBLR) were used as control variables for thewealth of the household and wealth of its social

Table 7. Determinants of the microfinance repayment performancea

(1) (2) (3) (4) (5)

PPRIN �0.0005 �0.0005 �0.0004 �0.0005 �0.0004

(�11.87)*** (�11.94)*** (�7.56)*** (�11.87)*** (�7.72)***

DURATION 0.0018 0.0018 0.0019 0.0018 0.0018

(4.92)*** (4.90)*** (5.23)*** (4.91)*** (5.14)***

BRAC �0.8872 �0.8083 �0.7788 �0.8855 �0.7247

(�9.72)*** (�8.69)*** (�7.95)*** (�9.69)*** (�7.36)***

BRDB �0.4776 �0.3305 �0.3902 �0.4806 �0.2866

(�4.83)*** (�3.09)** (�3.71)*** (�4.84)*** (�2.58)***

SEX 0.1204 0.1363 0.0888 0.1185 0.1025

(1.60) (1.81) (1.15) (1.57) (1.32)

PASSET 0.0000 0.0000 0.0000 0.0000 0.0000

(3.37)*** (3.22)** (3.16)** (3.37)*** (3.08)**

SELEAGR 0.2636 0.2638 0.2741 0.2663 0.2805

(3.48)*** (3.44)*** (3.6)*** (3.50)*** (3.62)***

NBLR 0.0589 0.0569 0.0579 0.0585 0.0551

(5.57)*** (5.35)*** (5.42)*** (5.49)*** (5.09)***

NFSL 0.2260 0.1957

(3.01)** (2.56)**

NFSH 0.1668 0.1215

(2.05)* (1.46)

AGEGP �0.0067** �0.0057

(�3.17) (�2.65)**

SAMEEDU 0.0010 0.0048

(0.01) (0.07)

SAMEAGE 0.0860 0.0898

(1.26) (1.31)

CRD 0.0249 0.0419

(0.34) (0.57)

CONS 0.3286 0.0044 0.3149 �0.0672

(1.39) (0.02) (1.32) (�0.26)

N 1,664 1,664 1,664 1,664 1,664

Log likelihood �1022.37 �1018.94 �1015.39 �1022.29 �1013.04

Pseudo R2 0.114 0.117 0.120 0.114 0.122

a Estimates obtained using a probit, t statistics are given in parentheses.* 10% level of significance.** 5% level of significance.*** 1% level of significance.


network. Those variables showed a positive andsignificant impact on the on-time repaymentperformance. This trend can be explained bya greater ability of richer households or house-holds with richer relatives to cope with shocks.It is also likely that households with more pro-ductive assets have access to projects with high-er returns or safer projects.Social ties inside the group, proxied by the age

of the group (AGEGP), had a significant andunexpected negative impact on the repaymentrate. 19 If there is a little change in the composi-tion of the borrowing group, this result can berelated to the negative impact on repayment ofthe membership period in the study of Matin

(1997). We could explain this contrasting fea-ture in two ways. At first, we can refer to whatPaxton (1996) called the matching problem: asthe duration of membership increases, the creditneeds of the members of the group evolve differ-ently. This circumstance could result in tensionsinside the group. It is also possible think that theprovision of intragroup insurance becomesmore costly as the size of the loans increaseand especially if borrowers that are granted asmall loan are still jointly liable for borrowersthat are granted larger loans. A decreasingpower of social penalties can also explain thisfeature: as members know each other betterthey are more reluctant to control and sanction


each other. This trend seems to overcompensatethe benefits of the increasing experience in theprovision of intragroup insurance. The under-standing of the factors that lead the age of theborrowing group to have a negative impact onthe repayment performance needs further re-search. Should borrowing groups be renewedafter several years? Should individual lendingsucceed group lending after several years? 20

Group homogeneity (SAMEEDU, SAME-AGE) proved to have no significant impact onrepayment performance as in the study of Wy-dick (1999). 21 We did not address the questionof the predicted positive impact on repaymentof group homogeneity in terms of risk as a re-sult of peer selection (Ghatak, 1999). Neverthe-less, both the study by Zeller (1998) and thestudy by Sharma and Zeller (1997) give evi-dence that this kind of homogeneity has a neg-ative impact on repayment performance.Access to basic literacy services (FACL) and

to health services (FACH) had a positive andsignificant sign. Borrowers who have access tohealth services are more likely to be able to pre-vent health shocks and to cope with them. Butthis positive impact is no longer significant inthe last column where the group-related varia-bles and dynamic incentives variable wereadded. Borrowers who had access to basic liter-acy might have access to more profitable pro-jects or might be able to generate more cashout of a project. The positive impact of accessto basic literacy on the repayment performanceis even higher than the impact of access tohealth services. The innovative features of edu-cation programs (both adult and children edu-cation) run by these MFIs can explain thisfeature. The BRAC education program, forexample, emphasizes mainly the education ofgirls in rural areas, giving them the opportunityto attain higher levels of education by provid-ing flexible learning hours and scholarshipsdepending on educational performance. Accessto nonfinancial services can also increase thevalue of the relationship with the MFI and in-crease the opportunity cost of strategic default.The incentive power of dynamic incentives

(conditionality of the allocation of new and lar-ger loans conditional on the loan size) was ex-pected to be higher for credit-rationedborrowers. Credit rationing (CRD) had the ex-pected positive impact on repayment perform-ance but this impact was not significant.What we observe here are only marginal im-

pacts of microfinance innovations since each ofthe studied MFIs uses group lending, nonfinan-

cial services and dynamic incentives in its creditmethodology. This is likely to lead to an under-estimation of the impact of these financial inno-vations.The analysis of the change in the log likeli-

hood of the different specifications providessupport for the full model (column 5). Theshare of the variance explained by our variablesremains small which could pledge for missingvariables. These variables might be commu-nity-level variables such as the economic driveof the area, the degree of monetarization, andthe exposure of the area to natural disasters 22

(Khandker et al., 1995; Zeller, 1998); pro-gram-level variables like functioning costs ofthe MFIs (Khandker et al., 1995) or bor-rower-level characteristics such as idiosyncraticshocks (illness and injuries).

(b) Is the determination of the loan size efficient?

Table 8 reports the results of the estimationof the size of the loan for credit-rationed bor-rowers (step 4). We consider that MFIs allocatelarger loans for projects they anticipate to havea lower default probability. The previous anal-ysis of the repayment performance proved thatthere were factors affecting the default proba-bility and that it was rational for the MFI touse such information to lower the value of itsdefault loans.The duration of the loan (DURATION) has a

positive impact on the size of the loan. There ishowever, little variation in the duration of theloans and 88.13% of the loans� were 11–13months in duration.The coefficient of the sex of the borrower

(SEX) is negative, which means that MFIs allo-cate smaller loans to their female borrowers. IfMFIs in Bangladesh predominantly lend to wo-men, the fact that they lend smaller amounts towomen (the study of Zohir et al. (2001), usingdata collected in 1998 also found evidence thatthe microfinance programs lend significantlysmaller amounts to their female members)seems to suggest a certain bias against femaleborrowing. We have shown, however, that fe-males did not experience a lower repaymentperformance compared to their male counter-parts, which confirms Zeller�s (1998) findingthat traditional bias against female borrowingis not justified. MFIs allocate larger loans tohouseholds with bigger stocks of productive as-sets (PASSET) who also demonstrate a betterrepayment performance. The other two coeffi-cients for the wealth of the household and of

Table 8. Determinants of the size of the loana

(1) (2) (3) (4)

DURATION 1.3390 1.3194 0.8993 1.9184

(7.68)*** (6.87)*** (3.96)*** (4.31)***

BRAC �93.0995 34.8472 �202.4050 �131.1337

(�1.4) (0.5) (�2.12)** (�0.92)

BRDB �68.4574 79.6920 �175.2989 �78.1494

(�0.92) (0.98) (�1.66)* (�0.52)

SEX �108.7201 �186.7730 �199.1072 �175.7549

(�1.99)** (�3.04)** (�2.51)** (�1.45)

PASSET 0.0011 0.0010 0.0003 0.0015

(2.31)** (1.97)** (0.4) (1.38)

SELFEAGR �12.8999 14.8222 11.6931 23.6703

(�0.24) (1.8)* (1.03) (1.41)

NBLR 8.7989 �27.0551 �8.9203 �62.6599

(1.2) (�0.45) (�0.11) (�0.55)

NFSL 216.7072 281.3405 239.9620 218.4354

(4.16)*** (4.83)*** (3.18)** (1.98)**

NFSH 222.4859 238.5985 212.2208 321.3028

(3.84)*** (3.68)*** (2.56)** (2.62)**

AGEGP 2.7045 2.1302 9.1766 1.2264

(1.97)** (1.73)* (4.03)*** (0.39)

SAMEEDU �83.2923 �174.0838 �68.7365 �198.3234

(�1.71)* (�3.18)** (�0.95) (�1.9)*

SAMEAGE 100.1637 180.9802 213.1835 252.3748

(2.1)** (3.38)** (2.91)** (2.44)**

PREVLOAN 0.5394

(29.54)***

CONS 1304.6500 �692.5682 1639.9290 2027.0000

(8.63)*** (�4.09)*** (7.75)*** (5.92)***

N 1,386 1,386 593 392

ADJR2 0.6024 0.0988 0.1443 0.1026

Dependent variables, sample: Specification I: Loan size, sample: credit rationed loans. Specification II: Deviation ofthe loan size from the mean loan size of the corresponding loan cycle and MFI, sample: credit rationed loans.Specification III: Loan size, sample: credit rationed loans, first loan cycle. Specification IV: Loan size, sample: creditrationed loans, second loan cycle.a Estimates obtained using OLS. t statistics are given in parentheses.* 10% level of significance.** 5% level of significance.*** 1% level of significance.


its relatives (SELFEAGR and NBLR) show nosignificant impact on the size of the loan. But,the use of both of these wealth indicators forthe choice of the loan size would be likely to in-crease the inequalities among borrowers and itis thus preferable that microfinance creditagents do not take these variables into consid-eration when allocating loans.The coefficient of the age of the borrowing

group at the due date (AGEGP) has a positiveand significant coefficient. It is indeed expectedthat the social ties and other benefits of thegroup, such as information-sharing, increasewith the age of the group. As previously men-

tioned, however, the age of the group has a neg-ative impact on repayment performance. IfMFIs want to allocate larger loans to groupmembers with whom they have an establishedrelationship, they should also develop specificincentives for their experienced borrowers tohave a better repayment performance.Group homogeneity in terms of age (SAME-

AGE) has a positive impact on loan size andgroup homogeneity in terms of education(SAMEEDU) has a negative impact on the sizeof the loan. But, as shown in the analysis ofrepayment performance, group homogeneitybased on age or education level did not prove


to have any significant impact on the repay-ment. Group homogeneity is frequently usedas a methodological guideline for group forma-tion in many microfinance programs 23 and fur-ther research must be undertaken to understandwhat type, if any, group homogeneity has a pos-itive impact on the borrowers� repayment per-formance.Access to both of the nonfinancial services

(NFSL, NFSH) has a positive influence on thesize of loans. These servicesmight indeed increasethe borrowers� capabilities and thus increase theirprobability of success as shown in the analysis ofrepayment performance. The value of the previ-ous loan (PREVLOAN) has a positive and signif-icant impact on the loan size. This is explained bythe lending methodologies of our MFIs as thesize of the loan usually increases with the loan cy-cle. The MFIs also use different loan size rangesfor a given loan cycle depending on the locationof the borrower in a municipality, a semi-ruralor a remote area. The size of the previous loanthus incorporates this information.Since MFIs are likely to increase systemati-

cally the size of the loan with the loan cycle, ana-lyzing loan allocation through the variation ofloan size for different loan cycles might be moreappropriate. Specifications II to IV present alter-native methods of evaluating loan allocationusing the deviation to the mean loan size forthe corresponding loan cycle andMFI (specifica-tion II) and the loan size for the subsample of theloan observations corresponding to the first cy-cle (specification III) and second cycle (specifica-tion IV). These alternative specifications confirmour previous findings. The value of the previousloan (PREVLOAN) seems thus sufficient to takedynamic incentives into account into the estima-tion of loan allocation through loan size.

7. CONCLUSION

Microfinance programs are now a key elementof poverty alleviation strategies. The financialinnovations of their lending methodologies suchas the use of group lending, nonfinancial servicesand dynamic incentives have indeed raised theinterest of policy makers and researchers asmeans to alleviate poverty in a self-sustainableway. In this study we test the explanatory powerof theoretical models that attribute the perform-ances of MFIs in terms of repayment to the useof such financial innovations. We used house-hold level data, which allowed us to compute de-tailed repayment performance variables. We

also draw empirical researchers� attention onthe endogeneity problem of the principal for fu-ture analysis on repayment.Our results suggest that the provision of non-

financial services has a positive impact onrepayment performance. This provides argu-ments in favor of the integrated developmentstrategies. The provision of such services is,however, costly and further research is neededto assess the costs and benefits associated withthe different types of nonfinancial services. Re-search is also needed on the best institutionalway of providing these services: should theybe provided by the MFI or should the MFIoperate in partnership with another NGO pro-viding these services?The results also show that MFIs allocate lar-

ger loans to borrowers as the age of their bor-rowing group increases. This can be justifiedby the use of dynamic incentives, as the numberof allocated loans is likely to grow with the ageof the group. The age of the group was alsofound to have a negative impact on the repay-ment. This raises the need to develop newincentives for experienced borrowers to avoiddecreasing repayment performance and nega-tive domino effects as the clientele of the MFIbecomes more mature.Another important point that emerged from

the study is that MFIs tend to attribute largerloans to homogeneous groups in terms of age.Group homogeneity was not, however, foundto affect the repayment performance in a signif-icant way. We did not address here the questionof the predicted positive impact on repaymentof group homogeneity in terms of risk as a re-sult of peer selection (Ghatak, 1999). Neverthe-less, both the studies of Zeller (1998) and ofSharma and Zeller (1997) provide evidencefrom Madagascar and Bangladesh that showthat this kind of homogeneity has a negativeimpact on repayment performance. As grouphomogeneity is frequently used as a methodo-logical guideline for group formation in manymicrofinance programs, further research mustbe undertaken to understand what type, ifany, of group homogeneity has a positive im-pact on the borrowers� repayment performance.Microfinance programs have been success-

ful in extending credit to the poor thanksto appropriate lending methodologies. Thenegative impact on the repayment perform-ance of the size of the loan and of the ageof the borrowing group could reveal theincompleteness of these lending methodolo-


gies as the clientele of microfinance becomesmore mature.

NOTES

1. A high level of repayment performance is a prereq-

uisite for financial sustainability, but it is not a sufficient

condition of financial health as high administrative costs

or high borrower turnover could be the counterpart of

high repayment rates.

2. As borrowers have different default probabilities and

as it is difficult for the MFI to charge each borrower an

interest rate corresponding to his default probability,

borrowers who are more prone to default will be

subsidized by lower risk borrowers.

3. Dynamic incentives refer to the threat of not

refinancing a borrower who defaults on a debt obliga-

tion. The incentive power of dynamic incentive is

enhanced if the MFI allocates larger loans over time to

borrowers with a good repayment performance.

4. Figure 1 stands for a constant marginal productivity

of capital. We could also allow for an increasing

marginal productivity of capital that would strengthen

the attractiveness of larger loans for the borrowers.

5. We consider that the utility of the borrower is

increasing with the return he gets from his project.

6. The domino effect occurs when at least one member

of a credit group default due to the defaults of other

members.

7. Wenner challenges the positive effect of this feature

as further analysis indicates that costs faced by the

borrowers to get this information overcompensate the

induced benefits in terms of repayment performance.

8. These 29 thanas were randomly chosen out of the

391 thanas of Bangladesh.

9. Zohir et al. (2001), using borrower level information,

found a percentage of borrowers with current overdue of

28.7% for the Grameen Bank and 34.7% for BRAC.

These lead to higher on-time repayment rates (71.3% for

the Grameen Bank, 65.5% for BRAC) than the ones that

we found (55% and 37% respectively). These rates are

also significantly lower than 95%, but it is difficult to

compare their magnitude with the repayment rates that

we found given the nine years difference between the two

surveys.

10. Khalily & Meyer (1993) found a recovery rate on

rural loans by state-owned commercial banks of 18.8%

in 1988–1989 in Bangladesh.

11. We consider a loan to be size-rationed when the

borrower reports he would have liked to borrow more

from the same source, at the same interest rate, for the

same purpose at the time he borrowed the loan.

12. 70% of interest-free loans were provided by rela-

tives, 85% by relatives or friends and neighbors.

13. 109 of the 1996 microfinance loan observations

corresponded to loans allocated before the previous loan

was fully repaid. It is likely that some of these loans were

partly used for refinancing purposes even if refinancing

was not one of the eligible purposes for loans. Rahman

(1999) shows that 70% of the 217 Grameen loan

observations he collected were actually used for pur-

poses other than the approved loan project, including

money lending.

14. The only significant difference we found was that

the BRDB, a government program, had a lower

proportion of female borrowers (34% compared to over

70% for the other two MFIs at the time of the survey,

10% for the other credit providers), which resulted in a

higher education level of its borrowers.

15. Here the size of the loan is the only characteristic of

the loan for which correction for endogeneity is

provided. We tried to provide correction for the possible

endogeneity of the duration of the loan but we were not

able to find satisfactory instruments. This might be

explained by the limited variation of the duration of

MFI loans.

16. We also tried estimations using larger definitions of

repayment (repayment made within three, six or 12

months after the maturity date). For such estimations,

the sample had to be reduced to borrowers whose due

date was three to 12 months prior to the date of the

survey.

17. The borrowers were classified as credit rationed if

they reported they would have liked to borrow more

from the same source, at the same interest rate, for the

same purpose at the time they borrowed.


18. We did not have enough observations to extend this

analysis to �older cycles� (the adjusted R2 of the

regression of the deviation of the loan size for the third

loan cycle, 193 observations, was of 1.85% only).

19. We have no possibility of knowing whether some of

the borrowers are new members in old groups or old

members in new groups since the borrower were not

asked about the date their group was created. There is

now a lot of recomposition of borrowing groups,

especially in the case of the BRDB where members are

expected to stay within the program for 5–7 years. This

should not be an issue however here since the survey

was conducted in 1991 when microfinance programs

had not been in operation for a long in most of the

villages.

20. One of the referees pointed that the role of the

group is evolutionary. Initially, the group helps screen

applicants and increase the use of social sanctions and

mutual insurance. But as the group members interact

among themselves and with the MFI workers, critical

learning on the rules of the game takes place for all the

parties. The group then becomes more a convenient loan

collection forum and loan contracts are more individu-

alized. But program members still have to attend the

weekly meetings and repay by group to their loan

officers.

21. In Wydick�s study SAMESEX is the only variable

of social ties that affected repayment performance and it

has a negative impact only in urban groups. Here, since

the three studied microfinance programs operate with

same-sex group only, we did not include this variable in

the analysis.

22. Village fixed effects were not used because we had

only a few loan observations per village.

23. The Grameen Bank Constitution reproduced in

Rahman (1999, p. 161) states, for example, that ‘‘All

groups shall be formed with persons who are like-

minded, are in similar economic condition and enjoy

mutual trust and confidence.’’

REFERENCES

Armendariz de Aghion, B., & Morduch, J. (2000).Microfinance beyond group lending. Economics oftransition, 8(2), 401–420.

Besley, T. (1995). Nonmarket institutions for credit andrisk sharing in low-income countries. Journal ofEconomic Perspectives, 9(3), 115–127.

Besley, T., & Coates, S. (1995). Group lending, repay-ment incentives and social collateral. Journal ofDevelopment Economics, 46, 1–18.

Bratton, M. (1986). Financing smallholder production:A comparison of individual and group creditschemes in Zimbabwe. Public Administration andDevelopment, 6, 115–132.

Diagne, A., Chimombo, W., Simtowe, F., & Mataya, C.(2000). Design and sustainability issues of rural creditand saving programs for the poor in Malawi: An action-oriented research project. Washington, DC: IFPRI.

Edgcomb, E., & Barton, L. (1998). Social intermediationand microfinance programs: A literature review.Washington, DC: USAID, Microenterprise BestPractices.

Freimer, M., & Gordon, M. J. (1965). Why bankersration credit. Quarterly Journal of Economics, 79,397–416.

Ghatak, M. (1999). Group lending, local informationand peer selection. Journal of Development Econom-ics, 60, 27–50.

Khalily, B., & Meyer, R. (1993). The political economyof rural loan recovery: Evidence from Bangladesh.Savings and Development, 17(1), 23–38.

Khandker, S. R., Khalily, B., & Khan, K. (1995).Grameen Bank: performance and sustainability.World Bank Discussion Paper, 306, The World Bank,Washington, DC.

Madajewicz, M., (1999). Capital for the poor: The effectof wealth on the optimal credit contract. Mimeo,Columbia University, New York.

Matin, I. (1997). Repayment performance of GrameenBank borrowers: the unzipped� state. Savings andDeveloppment, 4, 451–473.

Morduch, J. (1998). Does microfinance really help thepoor? New evidence from flagship programs inBangladesh. Mimeo, Department of Economics,Harvard University, Cambridge, MA.

Paxton, J. A. (1996). Determinants of successful grouploan repayment: an application to Burkina Faso.Unpublished doctoral dissertation, The Ohio StateUniversity, OH.

Pitt, M. M. (1999). Reply to Jonathan Morduch�s Doesmicrofinance really help the poor? New evidencefrom flagship programs in Bangladesh. Mimeo,Department of Economics, Brown University, Prov-idence, RI.

Pitt, M. M., & Khandker, S. R. (1996). Household andintrahousehold impact of the Grameen Bank andsimilar targeted credit programs in Bangladesh,


World Bank Discussion Papers, 320, The WorldBank, Washington, DC.

Pitt, M. M., & Khandker, S. R. (1998). The impact ofgroup-based credit programs on the poor householdsin Bangladesh: does the gender of participantsmatter? Journal of Political Economy, 106(5),958–996.

Rahman, A. (1999). Women and microcredit in ruralBangladesh: An anthropological study of GrameenBank lending. Boulder, CO: Westview Press.

Sharma, M., & Zeller, M. (1997). Repayment perform-ance in group-based credit programs in Bangladesh:An empirical analysis. World Development, 25(10),1731–1742.

Smith, R. J., & Blundell, R. W. (1986). An exogeneitytest for a simultaneous equation tobit model with anapplication to labor supply. Econometrica, 54(3),679–685.

Stiglitz, J. E. (1990). Peer monitoring and credit markets.The World Bank Economic Review, 4(3), 351–366.

Stiglitz, J. E., & Weiss, A. (1981). Credit rationing inmarkets with imperfect information. American Eco-nomic Review, 17(3), 393–410.

Wenner, M. (1995). Group credit: a means to improveinformation transfer and loan repayment performance.Journal of Development Studies, 32, 263–281.

Wydick, B. (1999). Can social cohesion be harnessed torepair market failures? Evidence from group lendingin Guatemala. The Economic Journal, 109, 463–475.

Zeller, M. (1998). Determinant of repayment perform-ance in credit groups: The role of program design,intragroup risk pooling, and social cohesion. Eco-nomic Development and Cultural Change, 46(3),599–621.

Zohir, S., Mahmud, S., Sen, B., Asaduzzaman, M.,Jahirul Islam, Md., Ahmed, N., & Al Mamun, A.(2001). Monitoring and evaluation of micro-finance institutions. Final report, Dhaka:BIDS.

microfinance repayment performance in bangladesh: how to improve the allocation of loans by mfis

Documents