the world bank and microenterprise finance: from …...the world bank and microenterprise finance:...

44
The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document of the World Bank November 15, 1999

Upload: others

Post on 17-Mar-2020

5 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

The World Bank and Microenterprise Finance:From Concept to Practice

November 15, 1999

Operations Evaluation Department

Document of the World BankNovember 15, 1999

Page 2: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

6

MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT

SUBJECT: THE WORLD BANK AND MICROENTERPRISEFINANCE: FROM CONCEPT TO PRACTICE

This is a study on The World Bank and Microenterprise Finance: From Concept toPractice prepared by the Operations Evaluation Department.

This report evaluates the design of 87 Bank-supported microfinance projects. It isa desk review of project documents which considers the World Bank's approach to thisinnovation in light of the practices of some of the most successful recent non-Bankexperiences. The focus is on showing how the structure of microfinance projects, at thedesign stage, is crucial for their subsequent success or failure. The results show that theBank's basic approach to supporting microenterprise finance is sound, and that there hasbeen significant improvement in Bank performance. However, Bank microenterpriseprojects are still performing less effectively than Bank projects generally.

The central finding of the report is that greater emphasis should be given to thecommercial viability of microfinance institutions. In the past decade microfinanceinstitutions (MFIs) have shown that they can develop lending techniques and pricingstrategies which allow them to cover costs even for very low-income clients. Focusingattention on the policy and regulatory environment and enhancing the capacity of MFIsare more important than direct financial support in improving the effectiveness of Bankinterventions in microenterprise finance.

The findings of the study lead to several recommendations. First, Bank support for MFIsshould be viewed largely as either a nonlending development service or a target ofAdaptable Lending. Second, MFI project design should be based more explicitly on afinancial system approach, and on a more detailed analysis of institutional sustainability.Third, the Bank management should report on the overall consistency of operations withO.P. 8.30 and the financial systems strategy.

Page 3: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

7

Abbreviations And Acronyms

BRI Bank Rakyat IndonesiaCAMEL Capital adequacy, Asset quality, Management, Earnings, and LiquidityCGAP Consultative Group to Assist the PoorestCIDA Canadian International Development AgencyFIL Financial intermediary loanFSD Financial Sector Development DepartmentGNP Gross national productGTZ Gesellschaft für Technische Zusammenarbeit GmbHIDB Inter-American Development BankIFAD International Fund for Agricultural DevelopmentIFC International Finance CorporationMFI Microenterprise finance institutionNGO Non-governmental organizationOECD Organization for Economic Cooperation and DevelopmentOED Operations Evaluation DepartmentQAG Quality Assurance GroupSBP Sustainable Banking with the PoorSDI Subsidy Dependency IndexUSAID United States Agency for International Development

Director-General, Operations Evaluation : Mr. Robert PicciottoDirector, Operations Evaluation Department : Ms. Elizabeth McAllisterManager, Country Evaluations & Regional Relations : Mr. Ruben LamdanyTask Manager : Mr. Robert BuckleyPeer Reviewers : Mr. Khalid Siraj

Mr. Jacob Yaron

Page 4: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

8

1. CONTENTS

Executive Summary ...................................................................................................... 3

1. Introduction ....................................................................................................... 5

Background ......................................................................................................... 5Objectives of the Study........................................................................................ 6Structure of the Study .......................................................................................... 8

2. Lessons from Microenterprise Finance Institutions......................................... 9

Background ......................................................................................................... 9Capital Adequacy................................................................................... 10Asset Quality ......................................................................................... 11Management .......................................................................................... 12Earnings................................................................................................. 14Liquidity: Funding and Supervision ...................................................... 14Scale or Social Impact............................................................................ 15

3. The World Bank and Microfinance: From Concept to Practice .................. 18

The State of the Portfolio................................................................................... 18Lessons Learned ................................................................................................ 21

Annexes

1. Summary Data on Bank Microenterprise Lending.............................................. 232. Performance Measurement ................................................................................ 253. Completed Projects............................................................................................ 284. The Evaluative Framework................................................................................ 33

Attachments

1. Management Action Record ............................................................................. 362. Informal Notes of the CODE Subcommittee meetings ....................................... 43

Bibliography .............................................................................................................. 46

Page 5: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

9

2. THIS REPORT WAS PREPARED BY ROBERT BUCKLEY (TASK MANAGER) ANDROBERT VOGEL (CONSULTANT). FREDERIC ASSELINE (PROJECT ASSISTANT),RUCHIRA BANERJEE-CORCORAN AND ASITA DE SILVA (CONSULTANTS) ALSOCONTRIBUTED TO THE STUDY. NORMA NAMISATO AND BETTY CASELY-HAYFORDPROVIDED ADMINISTRATIVE SUPPORT.

Page 6: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

10

Introduction

Background

2.1 Microenterprise finance is now recognized as one of the most promising financialinnovations available to assist the poor. This report evaluates the World Bank’s belated butrapidly increasing support for this approach. Because of the newness of this innovation, thepresent study cannot follow the traditional OED approach of analyzing the performance of WorldBank–supported projects only after the fact. It cannot do this because there have not yet beenmany Bank-supported microenterprise finance projects subjected to ex post evaluation. Only 15such reviews have been done. However, there is considerable guidance to be found in theexperiences of successful microenterprise finance institutions supported by other donor agencies,such as USAID, CIDA, IFAD, GTZ, and IDB. These institutions have been more explicitlyinvolved in microenterprise finance for a much longer time than the Bank. Furthermore, a keylesson from OED reviews of a wide variety of projects is that the quality of Bank-supportedprojects at the time of entry into the Bank’s portfolio has a dominant effect on project outcomes.One objective of the paper is to use the broader microenterprise finance institution (MFI)experience to help evaluate the design structure of the Bank’s rapidly growing MFI portfolio.

2.2 In 1995 the donor community joined together to establish and fund the ConsultativeGroup to Assist the Poorest, CGAP. This multidonor organization, operating with the guidanceof a policy advisory board of microenterprise finance lenders and 26 donors, was designed toincrease resources systematically in microfinance and to broaden and deepen the success ofpioneers in the field. Through CGAP, the donors have explicitly responded to a rapidly growingfinancial innovation.1 The Bank has supported 70 such projects since FY91; before FY89, it hadformally supported only a few such projects.2 In addition, as A Review of the World Bank’sMicrofinance Portfolio (CGAP 1997) notes, the project pipeline shows continued high interest inmicrofinance projects. While the number of projects in the pipeline is volatile, at present morethan 15 microfinance activities are planned for FY99.

2.3 A paradigm shift has taken place in the recent past. MFIs that support microenterprisesare no longer seen as charities. They are now seen as ways to provide services that the poor needand can use with high rates of return. While it is true that this shift in view is a significant changein perspectives on development, in many respects it is really a return to the past. In the early 19thcentury, and even in the 18th century, virtually all today's industrialized countries experiencedmovements that were very similar to the current microenterprise finance movement. However, asVittas (1996) shows, the earlier movements had one important distinction from the concerns oftoday’s microenterprise finance: the primary emphasis then was on encouraging savings and thriftby the poor, rather than the provision of credit for their enterprises.3 Many of the movements

1 A Donors Committee had already been established and had written the Guiding Principles before CGAPwas established.2 As Kahnert (1989) shows, the Bank had successfully supported sustainable microenterprise finance asearly as 1976 under the Second and Third Calcutta Urban Development Projects. While there are no doubtother examples of microfinance lending in the Bank’s pre-FY89 portfolio, until the CGAP initiative therewas no formalization of how these projects or interventions should be implemented until beyond the 1992implementation of the Bank’ Operational Directive 8.30 on the financial sector.3 The objectives and activities of the Self-Employed Women’s Association in India are more like those ofthe earlier microenterprise finance organizations. See Bhatt (1996) for a discussion

Page 7: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

11

prohibited funding anything but the safest investments, such as government securities. And manyof the approaches that did not include such restrictions failed, often at high cost either to the stateor the poor savers.

2.4 As in the past and regardless of the emerging evidence of success, the world ofmicroenterprise finance is fragile. Out of the perhaps thousands of institutions that have appeared,as Rosenberg (1998) estimates, only a few dozen can be termed financially sustainable bytraditional measures.4 In addition, according to Von Pischke (1991), over the 1970s and 1980s, asmuch as US$5 billion in donor funding supported micro and small enterprise lending withnegligible development impact. Despite the recent successes of MFIs, the SBP inventory ofmicroenterprise finance institutions shows that about 60 percent of the funding for theseinstitutions comes from the donor community.

2.5 In dealing with microenterprise finance, considerable caution is warranted, particularlyfor institutions whose highest priority is to address poverty in developing countries. How can theWorld Bank best help support and sustain an approach to finance that holds the potential forsubstantial gains for the poor of developing countries? The donors’ Guiding Principles providesa strategic answer to this question. It argues that “fundamental principles of finance . . . must beobserved by all institutions if they are to succeed." The World Bank and the donor communityhave made this approach to microfinance more concrete by adopting what they term a financialsystems perspective, which emphasizes the institutional sustainability of MFIs. This echoes theBank’s more general position on financial sector operations, as articulated by the Task Force onFinancial Sector Operations and as implemented by Operational Directive 8.30 (1992) and itsrecent clarification as an Operational Policy Statement in 1998. In these documents, the Bank hastaken the view that all financial sector operations should envision the attainment of full financialviability for institutions that receive support. Bank policy in this regard is continually evolvingand better measures and practices continue to emerge. However, as a minimum standard ofsound practices, as well as a standard widely agreed to and well known, the financial systemsperspective can provide a basic design standard.

Objectives of the Study

2.6 This study has three objectives. The first is to consider whether the Bank’s new approachto supporting microenterprise finance— that is, the financial systems perspective— is in fact apractical way of supporting MFIs. And, if it is, can we use this approach to evaluate, in at least apreliminary way, the design of the Bank’s rapidly growing MFI portfolio? To do this, the studyapplies what is known as CAMEL analysis, a well-known perspective used both by successfulmicroenterprise finance institutions and U.S. banking regulators, to the design of World Bank–supported microenterprise projects.5 The application of this perspective is one expedient way toorganize a desk review of the Bank’s ongoing approach to these projects. This analysis shows theconsiderable advantages in project design that can be realized from the financial systemperspective. A much greater likelihood of satisfactory outcome is a direct result of such anapproach. However, the application of this perspective indicates that, at the design stage, mostWorld Bank–supported microenterprise projects score poorly and are unlikely to be sustainable.The second objective of the study is to analyze the lessons that can be learned from successful 4 The Microbanking Bulletin (1997, vol. 1) presents evidence that 21 MFIs were “fully sustainable” as ofthat date, where this is defined as a financial self-sufficiency of at least 95 percent.5 See Annex 1 for a description of World Bank–supported microenterprise finance projects. The databaseincorporates and adds to the information in Webster, Riopelle, and Chidzero (1996), CGAP’s Review of theWorld Bank’s Microfinance Portfolio, and Siraj and Randhawa’s (1997) review of social funds.

Page 8: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

12

MFIs, Bank-supported or not. The third objective is to evaluate how the World Bank’s approachto MFIs has changed over time.

2.7 In carrying out an assessment of this kind, banking supervisors often focus on five areasof risk exposure— the so-called CAMEL components— to identify risk exposures.6 Thesecomponents are: Capital adequacy, Asset quality, Management, Earnings, and the Liquidity ofthe institution. As used here, the CAMEL measures are presented as one widely used way to helpfocus the discussion on operationalizing the financial systems perspective.7 It is also theframework used by ACCION International, a leading supporter of MFIs, to evaluate its MFIaffiliates.8

2.8 Because poverty alleviation is such an important aspect of the Bank’s involvement inmicroenterprise finance, a pure commercial analysis of MFIs is not a sufficient standard ofperformance. In our analysis of likely project outcome, we thus augment the CAMELperspective by considering other Bank policy requirements in what might be termed a“CAMELplus” or CAMELS perspective, which also takes into account the efficiency of theproject’s outreach to the poor (what ACCION International refers to as Scale or Social Impact ina CAMELS analysis). Ultimately, the CAMELS perspective has a relatively clear analogue toOED measures of relevance and efficiency, which provide two of the three underpinnings ofOED’s approach to measuring project outcome.

2.9 The standard OED definition of project relevance is that the project’s goals are consistentwith the country’s overall development strategy and the Bank’s assistance strategy for thecountry. To measure the relevance of MFI projects and their components, we determinedwhether the project sought to rely on an institution that had internalized the information-processing problems that underlie lending to poor, uncollateralized borrowers in a hostile legaland regulatory environment. In other words, did the design pay attention to capital adequacy,asset quality or management issues? Also, a judgment was made whether the legal— for example,usury laws— or regulatory environment was such that the project could eventually be presumed tobecome more than a donor-assisted enclave.

2.10 The standard OED definition of efficiency is how well do project results compare withthe inputs in terms of costs, implementation times, and economic and financial returns. Thefactors that affected our judgment of MFI efficiency were: (1) Was the MFI able to deliverresources to its clients in a cost effective way? If the MFI was able to deliver its services withoutsubsidies and continue to grow, it was assumed to meet a market test of being efficient enoughnot only to survive but to grow. (2) If the MFI was not able to deliver its services withoutexternal assistance, the case for most MFIs, efficiency was measured by how cost-effective andtargeted were the subsidies involved. Our measure of efficiency can then be said to correspond to

6 See Barltrop and McNaughton (1992) for a discussion of the CAMEL standards. CAMEL measures werefirst used in the U.S. commercial banking system in 1978 as a way to measure the financial and managerialrisk of supervised banks.7 There are other techniques that could be used to evaluate MFIs and the policy environment in which theyoperate. For example, see the Subsidy Dependency Index (SDI) developed by Yaron (1992), and widelyused in the Bank. This measure provides a “one-dimensional” snapshot of the financial status of anintermediary. In addition, a recent analysis, by van Greuning, Gallardo, and Randhawa (1998), provideswhat one might term a dynamic regulatory perspective for examining MFIs. This work focuses on thesources of funds and types of risks to which MFIs are exposed, and, in so doing, emphasizes the broaderfinancial sector environment in which MFIs operate.8 See Saltzman, Rock, and Salinger (1998) for a discussion of ACCION’s CAMEL analyses of its affiliates.

Page 9: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

13

the last three components of the CAMELS standard: MFI earnings, its liquidity and its targetingof any subsidies. If a project supported an MFI to achieve profitability by mobilizing resourceseffectively, while maintaining adequate liquidity, and provided well-targeted subsidies, it wasjudged to be efficient. As a result, we can use this perspective to discuss the attention to risks inproject design, to estimate likely project outcomes, and to assess design performance over time.9

2.11 In summary, the CAMELS perspective provides an organizing devise that allows us toreview whether a project, at its design stage, attempted to address some of the more obviousquestions embodied in the financial systems perspective: Did the MFI have any capital? Did itconsider the need for loan provisioning? Was there a management plan? Did the project considerwhere and how funds would be mobilized?

2.12 With the CAMELS methodology, the answers to these questions can usually bestructured to have unambiguous “yes/no” answers and require little judgement or interpretation.Were the risks implied by each of the CAMELS standards discussed in the SAR? If so, howfocused was this discussion? Cumulatively, when one adds up the answers to questions arisingfrom all of the elements of a widely used financial monitoring system— be it CAMELS orsomething else— one gets a sense of comprehensiveness. In short, the CAMELS standards areone way to consider whether ongoing Bank-supported projects have been paying attention to any,some, or none of the more important financial risks involved with microfinance. Clearly, thistype of analysis needs to be augmented by more detailed impact studies, and this study is nosubstitute for that kind of evaluation. Nevertheless, given the potential for poverty alleviation ofthis innovation, the rapidly expanding Bank involvement in it, and the clarity of the Bank’sfundamental strategy, a stocktaking is appropriate. How close, in fact, are Bank practices to theconceptual basis of its overarching strategy? Thus, for a sense of how the ongoing portfolio islikely to perform, the CAMELS standard provides a simple, but widely used, categorizationscheme.

Structure of the Study

1.13 The next section discusses why the financial systems perspective is particularly appositefor appraising and monitoring the performance of MFIs in a desk review such as this one. Itsuggests a clear direction for identifying the most expedient ways of supporting MFIs. Section 3details how project designs have been changing over time with the evolving perspective on donorassistance. Because a number of studies detail the rich nuances of many Bank-supported projectsand the evolving ways of measuring performance (see The Practical Guide, 1997, and thereferences cited there), the main focus of our discussion is at an aggregate level. How well is theBank’s portfolio of microenterprise finance projects doing? Using the estimates of the effectsthat project design appear to have on outcome, the last section presents estimates of how theBank’s portfolio of microfinance loans is likely to perform. It also compares and contrasts theseestimates with QAG reviews of these projects. Finally, lessons of experience showing howperformance could be improved are considered.

9 As used by banking supervisors in the United States, the CAMEL approach also provides a succinct,single summary statistic on institutional risk exposure. It does this by averaging five measures in a uniformweighting system. Because our objective is to delineate all the risks that affect MFIs, we do not attemptsuch a summary perspective. No aggregation of measures is proposed here. Rather, the CAMEL standardsare used as a way to provide perspective on the kinds of risks that should be taken into account in projectdesign. However, the use of simple summary statistics in project design, such as the SDI, can lead to asubstantial improvement in design and performance.

Page 10: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

14

3. Lessons From Microenterprise Finance Institutions

Background

3.1 The risks of microenterprise finance are complicated because loans are made in anenvironment that has few devices available to provide the information and recourse that lendershave in more formal financial systems. Microenterprise finance can be seen as a way of creatinghigh-quality financial assets that would not be provided by the banking system, and fundingborrowers, who, subject to business risk, will repay. Of course, to intermediate successfully, anMFI must also carry out two additional tasks that confront any financial intermediary but arelikely to be especially challenging in the world of informal micro-finance: (1) manage theconsiderable administrative costs of building and implementing the kind of methodology thatenables effective borrower selection and assures repayment; and (2) have access to sufficientliquidity so that promises of future loans can be fulfilled. Furthermore, it must establish acredible reputation that it will Manage its high quality Assets with the appropriate level ofLiquidity.

3.2 Establishing a credible reputation can be as important as capital adequacy narrowlydefined for an MFI that want to establish itself in the financial marketplace, so that an MFImust perform well on four of the five CAMEL measures to achieve sustainable success. Themost obvious measure, at least in the short run, that an MFI is performing well against thesestandards is provided by the fifth CAMEL measure, that of earnings. Unfortunately, besidesearnings, none of the other four elements can be measured with any degree of accuracy, and evenearnings measures can be easily distorted.10 Consequently, evaluation of MFIs will continue tobe an imprecise task, complicated further by the fact that MFIs are rarely required to subscribe tostandardized charts of accounts and accounting rules, either national or international. Rather thanrelying on straightforward financial ratios that lead to standard conclusions, MFI appraisers mustbe able to infer many intangible, but highly practical, qualities to understand the operation andgovernance of MFIs.11

3.3 The importance of this emphasis on financial sustainability is consistent with the lessonsof the Bank’s experience with the microenterprise finance projects that have been evaluated. AsAnnex 1 details, there are 15 such projects. All these projects were approved before FY93. Tenprojects did not follow a financial systems (NFS) perspective, and five did. 12 Of the formerprojects, the performance of nine was rated as unsatisfactory by Project Completion Reports. Allfive projects that followed a financial systems (FS) approach had satisfactory outcomes. Whilethis is a small sample, these results are consistent with the recent empirical findings of Mosley 10 In the application of CAMEL standards in the United States, examiners are given wide discretion ininterpretation of the various ratings. Only 70 percent of the indicators used are quantitative rather thanqualitative measures. Consequently, even in a sophisticated, widely monitored financial system, judgmentsas to risk exposure play a significant role in institutional risk assessment. In developing countries evenmore judgment is required. For instance, ACCION’s handbook on applying CAMEL ratings to its MFIaffiliates indicates that only 48 percent of the indicators are quantitative. Thus, the difficulty of creatingrobust performance indicators, particularly in less-developed financial sectors, is acute.11 In Annex 2 we discuss some of the measurement difficulties with each CAMEL element. The PracticalGuide and the donors’ Guiding Principles also discuss performance indicators.12 Annex 3 describes each of the 15 projects and discusses why we classified a project as a financialsystems (FS) approach or a nonfinancial systems (NFS) approach. It also details how we applied theevaluative standards to what were often project components rather than projects.

Page 11: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

15

and Hulme (1998) who also show that financially sustainable MFIs have a higher impact onborrowers’ income than do less sustainable institutions.

3.4 In the evaluation of Bank-supported MFIs there is no reason to have expected taskmanagers to follow a strict, comprehensive, financial systems framework, such as CAMELstandards, in designing their projects. However, for understanding MFIs, particularly in theirnascent, preprofitable stages, CAMEL ratings can serve as prospective benchmarks, rather than astools to measure precisely an MFI’s performance. They can also be used to develop some simplequestions as to whether a project’s design considered specific risk issues. In addition, onceprofitability has been achieved, these examples can serve as a standard that will permit donors toidentify and rank the problems faced by potential projects.

3.5 In short, CAMEL measures for MFIs should not be taken too precisely, nor should SARsthat do not rely on the implicit approach suggested by the standards be criticized for not doing so.Nevertheless, the perspective underlying the approach, rather than the measures themselves, canbe useful in inferring how various risks were treated. The CAMEL approach provides a simple,comprehensive, and disciplined format for understanding and comparing MFI performance.Augmenting this approach with a concern for Social Impact, as is done here to reflect the Bank’sfundamental concern with poverty alleviation in its micro-finance projects, leads to theCAMELplus or CAMELS perspective. In Section 3 we discuss how these standards can be usedto consider Bank-supported MFI projects according to more traditional Bank evaluativemeasures.13

Capital Adequacy

3.6 The Practice: The Industry Experience. The most direct source of capital for MFIs isprivate equity finance. Such private equity, however, is unlikely to reach significant levels in thenear term. Information and monitoring costs are too high, and the kinds of markets necessary forsuch investments are not well developed. Moreover, it is not equity in hand that many MFIslack, but rather the ability to raise debt on commercial terms in order to leverage themselves; this,in turn, requires credibility, including the ability to raise additional equity as it is needed.14

3.7 The Bank’s Project Design Experience. Capital adequacy has not been given muchattention in Bank-supported projects.15 The main issue is the ability of the institution tostrengthen its equity base, ideally in local financial markets. Only 42 percent of the projects hadany requirements for loan loss provisioning, a practice that assures that funds are set aside toaddress future losses and maintain capital adequacy (although this share did increase from 27percent in 1987–90 to 47 percent of the projects approved in 1994–97). In 21 percent of the 13 OECD (1996) proposes the use of SCALE, which replaces the M of CAMEL with an explicit andprimary focus on self-sufficiency— that is, the internally generated funds divided by total expenses. TheIDB’s evaluation of its microlending activities (1991) calls for lending to sustainable institutions andmonitoring the performance of these institutions with much of the information that would be collected in aCAMEL analysis. The most recent director of USAID’s active microfinance program undertook one of thefirst systematic applications of CAMEL standards to MFIs.14 The ACCION’s Handbook (1998) indicates that 9 of 11 MFI affiliates have a ratio of risk assets toequity of less than 6, about half the ratio usually associated with commercial banks. The Handbook saysthat the underleveraging of these institutions is one of their main problems.15 The percentage figures given in this chapter are for varying sample sizes. Our overall sample containsthe 87 projects approved between FY87–97 that made loans to subborrowers. In many cases, however, theinformation is not available on all 87 projects. We only note if the available sample is less than 50.

Page 12: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

16

cases, the owner of the retail-level MFI was a government agency or a public bank, and neithertype of institution has any concern with having adequate capital to continue to operate. NGOswere the intermediaries in one-fifth of the projects; private sector banks were involved in only 9percent.

3.8 The Bank’s Broader Experience. Capital adequacy requires that MFIs have theresources— both human and financial— to develop. Perhaps the most direct way for donors to dothis is to provide capital directly, as a grant. CGAP’s grant mechanism seems a particularly well-structured approach to do exactly that, and thereby addresses what is undoubtedly one of the mainconstraints to scaling up MFI impact and achieving financial sustainability. Althoughsignificantly smaller in size than the USAID or the Canadian microenterprise finance programs—US$30 million over three years, as opposed to over US$100 million yearly for AID and US$40million for CIDA— CGAP is structured much like these programs. It relies on performance-based contracts with MFIs, and its grants provide training and technical assistance to helpmicrofinance institutions develop their capital base. Its resources would be stretched further withcoordination between CGAP and IFC in project preparation, with the latter potentially acting asan equity participant.16 If grant funds were combined with investment capital, both vehicleswould be strengthened. Without coordination, the set of projects that is independently attractivewill necessarily be much smaller.

3.9 Capacity Building. A less direct but ultimately more important means of developingcapital adequacy is through building the capacity of MFIs to develop sound commercialoperations and to increase the outreach and sustainability of their operations— particularly theirability to mobilize local financial resources. In many respects, the building of this capacity is amanagement issue, i.e., a topic to be dealt with under the “M” of CAMELS. However, staffcapacity is also an essential component of institutional credibility, and hence is also an importantaspect of the valuation of the enterprise’s capital. This is the strategy adopted by both the UnitedNations (1998) and the Bank’s Africa Region. In a context of high levels of poverty and lowlevels of growth, the strategy calls for: (i) exposure to and implementation of best practicetechniques for managing risks; (ii) reducing administrative costs; and (iii) increasing revenues. Itemphasizes developing performance-based lending instruments to be used before the Bankconsiders financing MFI start-up costs and their other fixed costs of operation. The strategy callsfor a heavy emphasis on raising awareness of performance standards and focusing on the kind ofpolicy environment conducive to MFI development.

Asset Quality

3.10 The Practice: The Industry Experience. Because of the cost of information and itsimportance for borrower selection, many sustainable MFIs have adopted lending methodologiesthat start each new client with a very small short-term loan, and then escalate size and extendmaturity if repayment is timely. A small loan can often be the cheapest way to collect theinformation necessary to develop on-going client relationships. This is the approach practiced byBancoSol, which makes first loans that average US$125. It is especially useful wheremicrofinance institutions cannot accept deposits, and hence cannot use deposit histories asproxies for loan repayment records. Solidarity groups and other forms of mutual guarantees arealso widely used approaches to improving borrower selection while holding down informationcosts. Another way MFIs economize on sorting information is to avoid financing business start-ups. It is well known that most new businesses fail within a fairly short period after start-up. Yet

16 So far, IFC and CGAP have both made investments in only one firm, K-Rep of Kenya, andtheir decisions in this instance were largely independent.

Page 13: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

17

it is often tempting to support start-ups because of the readily apparent impact. Such impacts canbe especially appealing if microenterprise enhancement is viewed as a way to give work to theunemployed.

3.11 While the lending methodology is the way portfolio quality is maintained, an equallyimportant issue is the understanding and information on the quality of the existing portfolio.CGAP, for example, quite correctly focuses a great deal of attention on indicators of this quality,such as: share of portfolio at risk, the write-off policy in use, and the nature of the portfolioclassification system. For this desk review of project design we were unable to assess portfolioquality. Future projects would be well served if such indicators were included at the design stage.

3.12 The Bank’s Project Design Experience. Lending for start-ups has two dimensions. Isthe MFI a new institution? And, are the clients of the MFI often start-ups? In 11 percent ofBank-supported projects, the MFI was a new institution without any record of lending, and henceof asset quality. In another 50 percent of the projects, the MFI was encouraged to lend to start-upfirms that had no record of asset quality. Cumulatively, most projects were intended to supporteither start-up firms or start-up intermediaries, a high-risk strategy unlike that taken by successfulMFIs such as BRI’s Unit Desas or ACCION International’s affiliates. Fewer than half theprojects discussed deployment of the organizations’ assets in ways that could improve theservices offered to clients. For example, only 45 percent of projects emphasized the need toprovide easy and convenient credit access for subborrowers. However, this figure has risen from30 percent in the FY87–90 period to 60 percent in FY94–97, reflecting increased recognition ofthis important aspect of MFI success.

3.13 The Bank’s Broader Experience. The Bank’s nonlending services can help focus onenvironmental considerations— such as the lack of recourse for nonpayment, the absence of titlesto land, and inefficient judicial systems— that create information problems. In addition to havinginformation on borrower performance, lenders must be able to enforce contracts. They must haverecourse against willful defaulters or recalcitrant repayers. MFIs usually have to have creative,often indirect, ways of resolving legal difficulties, as well as skirting regulations that hold interestrates below required levels. Obviously, direct resolution of these shortcomings is a moreeffective way of dealing with the problem. For the Bank, the key issue in this regard is indiscussions of the broader financial sector dialogue, and a general emphasis on the importance ofthe policy environment. As noted repeatedly, and emphasized in both a recent UN report (1998)and the Bank’s microfinance strategy in Africa, the financial policy environment is a majorconstraint to addressing the range of poverty concerns that MFIs attempt to address. Hence, anemphasis on MFI projects as part of a broader financial sector policy dialogue is likely to beproductive.

Management

3.14 The largest single challenge to MFIs is the operational efficiency with which they processinformation about borrowers. The inherently higher costs of making smaller loans that requireintensive follow-up and continual personal contact is, therefore, a major issue for most MFIs. Tobegin to address these concerns, MFIs must focus on a number of tactical issues. First, the MFImust be decentralized, not only by locating its services close to its clients to keep transactioncosts low, but also by delegating major responsibilities to individual loan officers. Second,employee compensation programs must take two key factors into account: adequate employeecompensation17 and well-designed programs of incentives for employees based on performance 17 The study by Christen, Rhyne, and Vogel of MFI success stories necessarily has a limited number ofobservations. Because of the limited number of observations, an important, if only suggestive, empirical

Page 14: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

18

criteria carefully tied to desired results. Third, good management requires effective internalauditing; complete procedures manuals; strong external audit; and stringent, yet flexible, controlof financial, administrative, and operational functions.

3.15 The Bank’s Project Design Experience. Bank-supported projects provide little focus onthese issues. Only 26 percent of the World Bank’s projects considered the need for decentralizeddecisionmaking; slightly more than half, 59 percent, demonstrated a concern with MIS issues andintensive borrower follow-up. Of course, the fact that many of the MFIs supported by the Bankwere start-up institutions (see para 2.11 above) suggests that decentralization concerns mightinitially be of somewhat lower priority. Nevertheless, both measures improved in recent projects.However, none of the SARs discussed the salary levels of MFI staff, although after interest rates,this has been found to be the next most important explanatory variable in predictingsustainability. Finally, only 38 percent of the SARs contained a business plan that described howthe MFI would move to financial sustainability over a given time horizon, a figure that declinedto 31 percent for the FY94–97 period.

3.16 There is, however, other evidence of progress in management design issues. More thanhalf, 55 percent, endeavored to simplify appraisal procedures (e.g., minimize complex eligibilitycriteria, financial and economic analysis, and documentation requirements), as did almost 70percent of recent projects. Two thirds of the projects included an emphasis on intensive follow-up of borrowers, increasing from 45 percent in FY87–90, reflecting an incorporation of lessonsfrom lenders such as BRI, or perhaps greater awareness of the importance of relationship lending.

3.17 The Bank’s Broader Experience: Technical Assistance and Training. In more than 60percent of Bank projects, support for MFIs has taken the form of technical assistance to buildcapacity and to strengthen these institutions. Given the financial fragility of most of theseinstitutions and their general lack of a background in financial issues, this is appropriate.However, according to Rhyne (1996), one of the central lessons of the USAID microenterpriseprogram has been the likelihood of wasting resources unless there is strong evidence of the MFI’scommitment to the technical assistance. In a parallel way, the potential effectiveness of technicalassistance to subborrower clients of MFIs must also be considered. An effective way to targetsuch assistance has been implemented by the Inter-American Development Bank (IDB)and in theBank’s Africa Region Microenterprise Strategy. The IDB projects auction-off subsidy assistanceto those most willing to pay for it. This approach has been followed in two projects in Paraguayand is now being expanded to a project in Ecuador. ACCION International takes the IDBapproach one step further and charges for all technical assistance. While 44 percent of the MFIsdid not provide technical assistance to subborrowers, the 30 percent of the MFIs that providedsuch assistance did so without fee.

3.18 Difficult questions surround the issue of how long a subsidized start-up period isappropriate while management learns the techniques of low-cost, sustainable lending. The pointis not that there is a simple answer to this question, but rather that it is important that the questionbe asked. That is, what is the justification for continuing a subsidy? In only 67 percent of theBank-supported projects was financial sustainability even considered as an ultimate target, afigure that increased to 75 percent in the FY94–97 period. Lack of concern for developing a planof sustainable development will almost certainly breed indifference to such development. In halfthe cases where soundness was discussed, a financial assessment was planned at appraisal, butwas not presented. finding of their study is that employee salaries relative to GDP per capita is a statistically significantdeterminant of profitability for the MFIs studied. In particular, institutions that paid higher relative salarieswere less profitable.

Page 15: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

19

Earnings

3.19 The Bank’s Project Design Experience. Very few Bank-supported MFIs had earningsthat imply financial sustainability. Only 20 percent of the projects were lending at commercialrates where the MFI determined the on-lending rate, a figure that rose to 27 percent in the mostrecent period. Another 15 percent lent at rates that included specific mark-ups over thecommercial lending rate, a number that rose to 23 percent in the FY94–97 period. Usually thismark-up rate was such that on-lending terms were insufficient to cover costs fully. Only 20percent of Bank-supported projects were at interest rates high enough to cover all costs plusgenerating profits. Another 35 percent covered administrative costs but not the cost of funds,while 24 percent provided subsidized funding or funding on grant terms.

3.20 The Bank’s Broader Experience. Donor agreement to a set of principles on how tosupport MFIs provides a basis for evaluating the content of donors’ actions. As the coordinatingarm of donors, CGAP could be seen as well placed to monitor and report on the consistency ofdonor actions, including those of the Bank, with their declarations. For instance, there have beencases in which borrowers have rejected support for a project from one donor because of thatdonor’s insistence on charging market interest rates, while other donors have not.

3.21 In Bangladesh, for instance, the Bank provided a US$100 million IDA credit in supportof MFIs. However, the interest rate charged on these loans is not sufficient for most MFIs to beprofitable. The lower rates charged are a direct result of the low, subsidy-dependent ratescharged by Grameen Bank, one of the world’s biggest, and certainly Bangladesh’s largest, MFI.Grameen, because it receives so much support from the donor community, passes part of thatsupport on to its borrowers. However, the existence of this transfer places a constraint on therates that can be charged by the many MFIs that follow the model of Grameen in Bangladesh. Asa result, the donor community, through its desire to help Grameen, essentially limits thesustainability of other MFIs in Bangladesh. Consequently, the Bank’s support for this project isinconsistent with the announced policy of not providing subsidies to ultimate beneficiaries. And,even if all of the subsidies go to the most needy, the system cannot expand as rapidly as it could iffull sustainability rates were charged.18 The FSD Report (1996) on social funds describes theBank-supported Egyptian Social Fund as experiencing similar problems. In this case, market-rateBank funds have been slow to disburse because of the availability of subsidized interest ratelending by other donors.

Liquidity: Funding and Supervision

3.22 The Bank’s Project Design Experience. In only one third of Bank projects did MFIsrely on the deposit-raising capabilities of commercial banks; in only 20 percent of the cases didthey attempt to mobilize their own deposits, and none sold their own debt instruments. Incontrast, ACCION International has not only undertaken nondeposit debt issuance, it has done soon international markets in a number of pilot programs. Over time, the potential for deposit-based resource mobilization in Bank projects has increased only slightly. Few Bank projectshave attempted to go beyond the larger issues related to MFI liquidity, such as those that might be

18 The argument that subsidized rates reduce the rate at which microfinance can expand assumes a very lowinterest rate elasticity of demand for microcredit. Rosenberg (1998), argues that there are no cases whereMFI interest rate increases resulted in a reduction in demand. The Microbanking Bulletin (1997) makes asimilar argument. Furthermore, even if below-market interest rates for borrowers were the most efficientmeans to reduce poverty, such an approach creates problems for MFIs not participating in donor fundedpoverty alleviation.

Page 16: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

20

measured by the CAMEL liquidity standard— distribution of assets among different maturities,diversification of funding sources, and timeliness of payouts and disbursements. Only 38 percentof the projects approved in FY96 discussed savings mobilization, compared with 64 percent ofthe projects approved since FY91. The lack of concern with savings mobilization suggests a lackof understanding of the need to compete for resources so that growth potential can be realized.Only 22 percent had measures to ensure that sustainable resource mobilization would beundertaken by the MFI.

3.23 Credit Lines. Such lines may increase outreach temporarily, but ultimately they canretard sustainability by making microlenders dependent rather than self-sufficient. Even lines ofcredit without subsidized interest rates may retard sustainability because they delay the momentwhen microlenders begin to look seriously for domestic funding on a commercial basis. Unlesscredit lines induce an increasing level of self mobilization of funds, they can create non-sustainable enclaves rather than financial intermediaries which assist the poor. Internationaldonor agencies can be more supportive of sustainability by funding training and technicalassistance. Of Bank projects providing assistance to MFIs, 80 percent had credit lines. However,as noted by the CGAP Review, these lines were small and had a declining average size, falling toUS$7 million in FY96 from more than US$18 million in FY91.

3.24 The Bank’s Broader Experience. The Bank’s experience in policy dialogue with regardto the financial sector regulatory framework has, as the CGAP Review describes, focused onregional issues of supervision and basic policy distortions, such as usury ceilings. Supervisionissues in Latin America were the topic of a seminar held with ACCION; the regulation of MFIs inthe West African Common Franc Zone was the subject of another practitioner-governmentdiscussion sponsored by the Africa Region, CGAP, and EDI. In the latter case, the discussions,while inconclusive, did reveal that the supervisory and interest rate norms adopted were such thatapproaches to MFIs that had been highly successful in other countries, such as Grameen Bank,could not operate within the formal financial sector of the zone. In this kind of policyenvironment, considerably more policy change will be necessary before MFIs are likely toprosper as formal sector intermediaries. In only one Bank-supported project was the topic ofappropriate supervision and regulation discussed. But in that project, the discussion focusedexclusively on the costs of such supervision, with the project mandating that such regulatorysupervision not be enacted until the project was completed.

Scale or Social Impact

3.25 Since the ultimate purpose of donor-funded microenterprise finance initiatives is toenhance financial services for the poor, and particularly poor women, outreach is extremelyimportant. Also essential is an understanding of the poor’s needs for financial services. But sofar, little progress has been made in the development and standardization of measures of outreach,and economic sector work has paid little attention to this issue. A vital aspect of broadening theaccess of the poor to microfinance is the question of how to address the concerns of the extremelypoor.

3.26 The Practice: The Industry Experience. Perhaps the best known non-Bank experiencein addressing the social concerns of the poor is that of Grameen Bank of Bangladesh. The famineof 1974 provided the stimulus for creating this institution. Professor Muhammad Yunusdiscovered that while the credit market was the scene of brutal exploitation of the poor, it wasalso the arena where the poor could, with financial assistance, break out of the cycle of poverty.19

19 This is based upon Hashemi (1997).

Page 17: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

21

He also found, however, that the conventional banking structure would not provide access tocredit for the poor because they could not provide collateral, and the overhead costs for servicingloans were thought to be too high. Among the poor, women suffered the greatest discriminationbecause custom excluded their ownership of assets, and the work that women generally carry outwas not classified as economically productive. Grameen was to develop collateral-free credit forthe poor, and specifically for women.

3.27 Since its inception as a pilot project in 1976, and after becoming a chartered bank in1983, Grameen has spread into 35,500 villages (more than half the villages in Bangladesh) andhas provided loans to 2.1 million members, more than 94 percent of whom are women.20 Mostmembers are effectively landless, with ownership of land below half an acre. Nevertheless, evenwith these successes, Grameen has been unable to target what is called the “hard core” poor, mostof whom remain outside the Grameen net. According to Hashemi (1997), these people are sodestitute that they do not consider themselves creditworthy. They do not feel they have enoughresources to generate the incomes required to pay back loans. In effect, they “self-select”themselves out of Grameen membership. A similar pattern is seen in other MFI success stories,such as BancoSol. The conclusion from these experiences is that microcredit does not reach thevery poorest. Nor should this be their focus. Indeed, Mosley and Hulme’s (1998) results suggestthat targeting MFI support on the very poorest is, in many instances, harmful for them, suggestingthat for the “hard core” poor access to microenterprise finance is not the first priority.21

3.28 The Bank’s Project Design Experience. The Bank has considerable experience inattempting to measure the scale or social impact of its support for microenterprise finance.Indeed, in all the projects, poverty alleviation was stressed as an objective. In addition, given thedifficulties noted in reaching the core population of the poor through MFIs, an effective andcomplimentary use of social funds, as in Benin, has been the use of grants to develop thepreconditions for successful microfinance among the poorest and least served.22 In only 40percent of the projects was consideration given to the kinds of performance indicators that coulddetermine the project’s impact on poverty. Similarly, in only 40 percent of the completedprojects were gender issues raised as a concern, even though the evidence is that this form ofborrowing is particularly beneficial for poor women. Furthermore, of the limited number ofprojects that considered gender concerns, in only one third of them were the effects on genderrated as substantial. Finally, in only a handful of cases was a measure such as the SDI reported inthe SAR. The result is that, in most cases, the Bank does not know how much of its resources areused for subsidies, or finance; nor is it obvious whether the projects are likely to have any effectson gender concerns. This is an area on which CGAP is actively working. The development ofsuch impact measures and cost effective poverty measurement methodologies are areas in whichthe Bank should be able to make a substantial contribution.

3.29 The Bank’s Broader Experience. One simple approach to the measurement of outreach,as argued in Christen, Rhyne, and Vogel, is to start with a simple measure such as loan size. This

20 BancoSol also provides most of its services to women, who represent 78 percent of its clients (seeGonzalez-Vega and others, 1997).21 Their argument is the very poor tend to take out “small, subsistence-protecting loans; these are seldom invested innew technology . . . [but] in protecting consumption. As a consequence, loans to the very poor are not normally able toproduce dramatic changes in borrower income: at these lower levels of income, there is also a greater risk that unluckyor improvident borrowers may be forced by their greater exposure to debt into selling assets which will permanentlylower their income possibilities.” (p. 787).22 The Practical Guide also makes a number of suggestions for using social fund subsidies to prepare theenvironment for microfinance.

Page 18: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

22

may be an effective measure because loan size is likely to be highly correlated with borrowerincome and assets and because securing such information should not add appreciably to thetransaction costs facing lenders or borrowers.23 Most microfinance institutions can readilyprovide figures on average loan size. Nonetheless, more detail may sometimes be needed onloan size distributions, in particular detail for lenders that do not restrict themselves exclusively tomicroenterprise loans. Another approach, also suggested by Christen, Rhyne and Vogel, is toargue that the need to do impact assessments may not be as compelling as with support for socialservices. Impact assessments for microfinance are very costly and difficult to achieve, and due tothe commercial nature of the “intervention,” we could trust that, as long as loan size is small, theborrowers are likely to be relatively poor and that they are the best judges of their own well-being.

23 New loan size, as opposed to average loan size, may be a particularly effective summary statistic of thetarget audience of an MFI. While average loan size will increase with borrower exposure to MFI lending,new loan size will not. Comparing loan size to per capita GDP can help ground the figure in the localeconomic context. However, care must be taken not to “reward” MFIs for making very small loans, so thatthey simply make simultaneous multiple loans to borrowers to capture the “reward.”

Page 19: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

23

4. The World Bank And Microfinance: From Concept ToPractice

4.1 The data presented in the previous chapter on the design of Bank projects provide oneperspective on how well the projects are performing. The data do not, however, provide insightinto how they are performing as Bank projects according to OED evaluative measures. That topicis addressed in this chapter. Based on project design, we estimate how effectively projects arelikely to perform according to OED and QAG performance measures. With this sense of overallperformance and trend, we draw a number of lessons from the Bank’s experience on how toimprove performance.

The State of the Portfolio

4.2 OED rates projects according to three results-oriented measures: outcome, sustainability,and institutional development. This study can only report these measures for the 15 completedprojects discussed in Chapter 2. However, it can also use the results of the CAMELS perspectiveon project design to discuss the effectiveness of these designs in meeting the Bank’s developmentobjectives. Because project design is strongly linked to project outcome, we can then use thisanalysis of project design to consider the likely outcomes of the existing portfolio of ongoingprojects.

4.3 Project Outcome. For each completed project it evaluates, OED asks: Did this projectachieve satisfactory development results, considering the importance and relevance of its majorstated objectives and the associated costs and benefits? This answer, in turn, must take thefollowing into account:

• Relevance. Were the project’s goals consistent with the country’s overalldevelopment strategy and the Bank’s assistance strategy for the country?

• Efficiency. How do the results compare to the inputs in cost, implementation time,and economic and financial return?

• Efficacy. Did the operation achieve its stated physical, financial, and institutionalobjectives?

4.4 This last aspect of project outcome involves the ultimate accomplishments of the project.Consequently, because implementation effectiveness— rather than just project design— is animportant element of efficacy, we cannot use this measure to predict likely outcome from thedesign stage. A well-designed project that is not implemented effectively, or one that isimplemented in a hostile and/or deteriorating economic environment, can easily have anunsatisfactory outcome that has little or nothing to do with project design. Consequently,ignoring the effects of project efficacy causes our estimates to be overly optimistic. Accordingly,our estimates of likely outcome— which require that the project be both relevant and efficient—should be seen as upper-bound estimates of likely outcomes. Nevertheless, this standard appearsto be a relatively strong predictor of outcome. For instance, a review of all FY96 projectoutcomes shows that projects that achieve both moderate relevance and efficiency are about six

Page 20: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

24

times more likely to be rated satisfactory than projects that fail to achieve moderate ratings onboth measures.24

4.5 Relevance. The lessons of successful MFIs suggest that the central issue is whether thegoals of a microenterprise finance project are consistent with, as the Guiding Principles put it,fundamental principles of finance. Are policy and/or the social infrastructure consistent with theachievement of successful, sustainable MFIs? For example, can such institutions thrive in acountry that has a usury ceiling on loans that is far below the administrative costs of microloanprocessing? Can an MFI survive when it is controlled by a government-sponsored projectimplementation unit, rather than being an indigenous group that has developed a technology forassuring prompt repayment? In short, did the project design embody a financial systemperspective? Based on our CAMELS analysis, we estimate that 57 percent of the projectsimplemented had such a design and were relevant to the country and to the Bank’s overalldevelopment strategy.25 A significant portion of the nonrelevant projects were implemented ineconomic environments that offered little chance for the project to move beyond being an enclavesupported by international donors.

4.6 Efficiency. The financial systems perspective on MFIs implies that a necessary conditionfor the achievement of efficiency is that the ultimate microenterprise beneficiary not be therecipient of any credit subsidy, that any subsidies to MFIs are seen as temporary, and that theultimate institutional objectives include plans to become a sustainable intermediary. In otherwords, an MFI that charges lower than market-determined rates is unlikely to be efficient. Inorder for such transfers to be efficient, they must be transparent and targeted with someefficiency. They should not be the result of lack of follow-up on repayment, or settling for“affordable” interest rates, and they should not disrupt financial market discipline by contributingto a culture of nonrepayment. Rather, they should be small, well-targeted subsidies of the sortGrameen Bank disburses to very poor and needy borrowers, subsidies that improve the situationof the poor more significantly than other forms of in-kind transfers. In our sample, almost 60percent of Bank-supported MFIs are operating efficiently.

4.7 The 57 percent of projects that were relevant embodied the first three standards of theCAMELS perspective. They were concerned with capital and management structure, as well asasset quality. The 59 percent that were efficient fulfilled the second three CAMELS components:they focused on MFI earnings, resource mobilization capabilities, and the targeting of anysubsidies. Thus, project designs that fulfill both OED standards also satisfy all six components ofthe CAMELS standards discussed earlier. They have not only considered all of the major risksconfronting an MFI, but they have also shown a concern that resources be allocated to the poorwith some efficiency. At the design stage, what share of Bank-supported projects satisfy thisstandard?

4.8 Outcome. In the course of a decade, the Bank has achieved global reach in supportingmicroenterprise finance, including more than 60 percent the Bank’s borrowers. Nevertheless, it isunlikely to have such reach in supporting MFIs with satisfactory outcomes. According to our

24 See Annex 4 for a fuller discussion of the methodology and its empirical basis.25 In other words, a project deemed to be relevant as an MFI had to be implemented in an environment in which thebroader regulatory environment did not proscribe the possibility of a sustainable MFI. For example, it could not beimplemented in an environment which had a binding interest rate ceiling which was well below the cost-recoveringinterest rate, and still be considered relevant as an MFI project. Similarly, to be relevant, the MFI had to have a planthat recognized that over the longer term its lending activity was based on a process of leveraged finance. A projectwhich did not recognize the corresponding need for capital base and management plan may be relevant as a povertyalleviation program, but not as an MFI.

Page 21: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

25

estimates, slightly more than 50 percent of World Bank–supported projects are likely to havesatisfactory outcomes (57 percent of IBRD projects and slightly less than 50 percent of IDA-sponsored projects). This figure improves to almost 70 percent for projects implemented in thepast four years. While this latter figure is a vast improvement, it is still somewhat weaker thanthe outcome for Bank projects generally, which in FY97 had an 80 percent share of satisfactoryproject outcomes for IBRD projects (see World Bank 1997a).

4.9 While a strict comparison cannot be made, the estimate of likely project outcomes is alsomore pessimistic than the estimate of outcomes implied by the Quality Assurance Group’s (QAG)measure of projects “at risk.” QAG’s measure considers a project at risk either on the basis of thelatest Supervision Form 590 ratings or as a result of the project having 3 or more “flags” ofdistress among 12 leading indicators. According to QAG, 30 percent of the MFI projects in asubsample of our projects were at risk. This is a high proportion of projects at risk for a sector,and above the Bank sectoral average of 25 percent, but less than the 40 percent at-risk figureestimated for Financial Intermediary Loans (FILs). This difference in risk ratings for FILs andMFIs suggests that, although MFIs may be small and relatively fragile financial institutions, theWorld Bank has been able to deal with them more effectively than it has dealt with financialinstitutions generally. As described more fully in Annex 4, QAG estimates indicate that 43percent of the projects potentially at risk, and 17 percent of projects not at risk, will ultimatelyhave unsatisfactory outcomes. It estimates that almost 72 percent of the projects actually, asopposed to potentially, at risk will have an unsatisfactory outcome. Applying these coefficientsto the subsample of microenterprise finance projects for which there are QAG ratings implies that70 percent of the MFI projects are likely to have satisfactory outcomes, compared with ourestimate of 59 percent for this subsample.

4.10 One explanation for our more pessimistic estimates of likely project outcomes is that thesubsample of QAG microenterprise projects is more recent, and hence captures the improvementin performance noted in our more recent estimates. That the QAG estimate of 70 percentsatisfactory outcomes is identical to our estimate of likely outcomes for the past four years isconsistent with this age distribution of projects. However, as described in Annex 4, it is notpossible to make a strict comparison with QAG ratings. The point is not to quibble aboutdifferences in estimates of riskiness, but to stress that according to both measures MFIs appear tobe relatively risky.

4.11 Sustainability. In the past five years, about 35 MFIs in developing countries havereached a level of sustainable profitability. That is probably 25 more MFIs than just 5 yearsearlier. In addition, in the past year, a crisis in microfinance in Colombia was not only effectivelyweathered, but the method of resolution also included plans for the first issuance of MFI equityon a stock market. These changes represent major steps toward sustainability for a financialinnovation that addresses poverty in very basic ways. Nevertheless, while these achievements aresignificant, they are just a beginning. Microenterprise finance is an infant industry.

4.12 Moreover, it is an infant industry that has yet to achieve widespread sustainability. Weestimated that only 25 percent of the 40 projects providing below-market-rate lending were doingso efficiently. By efficiency, as described more fully in Annex 4, we mean that the subsidiestended to go to those relatively poor borrowers who would self-select into the program in a waythat encouraged productive borrowing. Far fewer are providing these funds in a sustainablemanner. In effect, only 6 of the 10 efficient but subsidized projects are thought to be sustainable.This is because, for sustainability, the MFI must be either fully self-financing or have clear,accessible resources to deal with any revenue shortfall. The result is that only 45 percent ofBank-supported projects are likely to be sustainable, although the figure improved to almost 70

Page 22: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

26

percent during the last 4 years. This latter figure compares quite favorably with Bank-wideperformance, which in FY97 improved to 54 percent.

4.13 It appears that the Bank’s ability to digest the conceptual lessons of almost 20 years ofmicrofinance practices is beginning to be translated into a similar revolution in Bank practices. Asubstantial improvement has occurred. The extensive training and manuals provided by SBP,FSD, and CGAP seem to be having a substantial payoff in better project design. Nevertheless, ifmicroenterprise finance lending is not to be one of the riskiest areas of Bank lending— as it isnow— much more is needed than requiring better financial data in Bank supervision reports. Tointegrate the results of the many years of effort more fully into Bank projects will require that theBank internalize in actions, as well as in words, the principles it helped to articulate in the donors’Guiding Principles. This will require much greater clarity about the importance of financialsustainability in microfinance projects.

Lessons Learned

4.14 The performance of the Bank portfolio in support of MFI is improving significantly. Butadditional improvement is needed if the sector is to perform as effectively as the typical Bank-supported project. These results indicate that the Bank may not be able to respond positively tothe challenge posed by the 1997 Microenterprise Finance Summit: to support a multiplicativeexpansion in the development of microenterprise finance.26

4.15 The availability of funding for MFIs is not the main constraint to the more rapid diffusionof this innovation. A good example of how the Bank can respond effectively is given by theAfrican Region’s Microfinance Strategy, which emphasizes increased knowledge disseminationand policy dialogue, rather than lending. But for this support to be effective for MFIs, the Bankmust adopt even more firmly the view articulated in the Bank’s new Financial Sector Strategy(1997)— that is, financial sector development is a process, not an event. The development ofsustainable MFIs is likely to be a long and costly process. In most cases, it is not a process thatwill be able to digest large amounts of funding from institutions such as the Bank.

Lesson 1. Bank support for MFIs should focus on advisory services or should take the form ofAdaptable Program Loans (APLs) or Learning and Innovation Loans (LILs). APLs and LILsseem particularly appropriate because they provide a vehicle that would permit the Bank toextend the mix of financial, analytical, and advisory services that may be needed over a longperiod of time. The staffing and supervision of these projects will require more inputs anddifferent reporting outputs than traditional intermediation loans. This kind of shift in approach isalready beginning as three LILs and three APLs have been done in the last two years.

Lesson 2. The design of MFI-Bank projects is increasingly based on the financial systemsperspective articulated by the donors’ Guiding Principles, O.P. 8.30, and CGAP’s evolvinglessons of best practices. This has resulted in a significant improvement in project design andprobable project performance. However, our findings suggest that, without further improvementsin design, it is likely that MFI projects will continue to perform less well than the typical Bankproject. Documentation of MFI projects should include the presentation of simple summarystatistics of how well a proposed project satisfies the financial systems perspective. The use ofthe Subsidy Dependence Index (SDI)— a measure that the Bank recommends for use in all

26 The Summit Declaration and Plan of Action estimates that over the next nine years, US$11.6 billionwould come from grants and concessional or low-interest loans from multilateral and bilateral donors. Thisamount is more than double the amount of such support provided by donors in the past nine years.

Page 23: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

27

financial sector projects— should be encouraged, possibly complemented by other performanceindicators such as those proposed in the CGAP-Bank-developed Practical Guide.

Lesson 3. Bank management should report on the consistency of its operations with O.P. 8.30and the general principles agreed to in the Donor’s Guiding Principles. The donors’ adoption ofthe financial systems perspective carries with it a particular approach to supporting MFIs, whichopposes conferring interest rate subsidies on ultimate borrowers. It does not appear that the Bankfollows these policies consistently.

Page 24: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document
Page 25: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

29

Annex 1

Summary Data on Bank Microenterprise Lending

This study reviewed the data provided by 10 years of Bank involvement in microfinanceactivities, from FY87 to FY97. The database incorporates and adds to the information providedby Webster, Riopelle, and Chidzero (1996); Siraj and Randhawa; and CGAP (1997).

The principal sources of information on the project pipeline were Staff Appraisal Reports(SARs) and (when available) Implementation Completion Reports (ICRs), as well as informationretrieved from the Bank's Intranet search engines and Regional Information Centers.Microfinance components were examined in all the different project modalities listed by CGAP:stand-alone microfinance projects, rural finance, financial sector loans, private sector loans, andsocial funds.

The database consists of information on 81 projects that lent for microfinance duringFY87–FY97 and was established using CGAP's definitions of microfinance institutions andmicroenterprise. A microenterprise is defined as "an informal sector business with five workersor less, and fixed assets valued at less than US$10,000"; a microfinance institution as "anorganization which takes financial deposits or makes loans in small amounts (usually less thanUS$1,000) and whose clients are, for the most part, low-income households, self-employed, orengaged in very small productive or commercial operations." Precise financial information,however, is not always available at appraisal; whereas all the projects reviewed have amicrofinance component, information on subloan size limit was present in only 59 of them. Areview of these projects shows that 41 percent had planned maximum initial loans belowUS$5,000, and the others envisaged subloans that could exceed US$5,000. The definitionpertaining to loan size is thus not to be taken as absolute, since different loan amounts relate todifferent regional realities (loans in Eastern Europe, for instance, often exceed US$1,000).

The review of the portfolio for FY87–FY97 shows total Bank support formicroenterprise finance of US$1,097.5 million in current dollars, 27 with 31 projects financed byIBRD loans and 50 projects financed by IDA loans. The total microfinance support approved forthe years 1987–90 was small, totaling US$137.8 million for 11 projects. It jumped to US$557.5million in the period 1991–93, with 34 projects in the pipeline. Six of the 87 projects did notengage in any lending and so were not included in our CAMEL S evaluation. The recent years,FY94–FY97, show a sustained interest in microfinance activities, with 36 projects representingestimated Bank support of US$402 million.

Additional findings include regional tren ds: while interest in projects in Africa remainspredominant (42 percent of all projects), a significant increase occurred in the Europe and CentralAsia Region (ECA), where projects accounted for 28 percent of the Bank portfolio for FY94–FY97 (12 percent in FY91–FY93). In contrast, projects in the East Asia and Pacific Region(EAP) and in the Latin America and Caribbean Region (LAC) have declined, representing 3percent and 14 percent, respectively, of the 1994–97 portfolio (21 percent and 24 percent,respectively, in 1991–93). However, the regional sub-samples are too small to provide anysignificant estimates of possible differences among regions in performance. Another trend is thereduction in the amount of the average microfinance component per project, down from US$16.4

27 Numbers for amounts of Bank support are estimates.

Page 26: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

30

million in FY91–FY93 to US$11.2 million in FY94–FY97. Similar patterns of regionalbreakdown and project size have emerged from the October 1997 CGAP study.

Further lessons can be derived from a CAMEL analysis of the design elemen ts of Bankmicrofinance projects, which shows the following trends: 28

• Capital adequacy: Only 42 percent of the projects reviewed have requirements forloan provisioning; 21 percent of the projects are executed by publicly owned banks(retail-level MFIs are owned by public agencies in 14 percent of the cases); andspecific reporting guidelines are not specified in 35 percent of the projects.

• Asset quality: Forty-nine percent of the projects restrict lending to existingenterprises; 52 percent emphasize short-term, small loans; and savings areincreasingly encouraged.

• Management: Only 26 percent of the projects discuss the need to delegate loandecisions to the loan-officer level.

• Earnings: Most of the lending-rate structures examined are not able to cover morethan operating costs. However, concern for the sustainability of the MFI is touchedupon in 67 percent of the projects.

• Liquidity: In only one-third of the Bank projects do MFIs rely on the deposit-raisingcapabilities of commercial banks.

Finall y, 15 projects out of 81 have been completed. While this is a limited number ofobservations, the outcomes suggest a strong correlation between sound financial design andsuccessful outcome.

28 Percentages apply to the projects for which information was available at appraisal. (See summarycharts.)

Page 27: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

31

Annex 2

Performance Measurement

This annex highlights som e of the difficult measurement problems involved in assessingthe financial situation of an MFI. Each of the CAMEL elements is briefly discussed; then, inlight of these measurement difficulties, the Subsidy Dependence Index (SDI) is presented as aconcise means of helping to sort out what the various CAMEL components imply about subsidiesand institutional sustainability.

A review of these measurement problems should make clear why the use of a CAMELstandard is far too imprecise and complicated to be a measure for MFI project design. However,these same measurement difficulties point to why some summary statistic on design structure,such as the SDI, would be helpful for quicker evaluation of project design.

Capital Adequacy

In even the most sophisticated markets, capital value is a difficult concept to measure.Hence, there is no straightforward decision rule for how much capital is enough, althoughbanking supervisors at a CGAP-ACCION conference offered a 20 percent capitalization figure asa benchmark measure. This figure is slightly higher than that of the highly successful BancoSolin Bolivia. However, it is far from clear how to calculate such a measure so that it serves as aprudential norm. To give a sense of how complicated the concept of capital adequacy can be forthe performance of MFIs, consider the case of one of the most successful projects, BRI’s UnitDesas in Indonesia. This project was undertaken by a unit of a public sector bank that wassimultaneously implementing a Bank-supported small and medium-size enterprise project.Although undertaken through the same bank, the latter project did not charge market rates ofinterest and had an unsatisfactory outcome. In contrast, the former was highly successful, lendingat rates that Yaron (1996) suggests may well be much higher than necessary to achievesustainability. Both units had the same parent, and hence the same capitalization, but onesucceeded and the other did not.

Asset Quality

There are as many ways to ensure asset quality as there are impediments to being able todiscern it. Microenterprise finance is fundamentally a means of managing contractual or verbalagreements so that the financial asset quality has the quality of the underlying asset and isaffected as little as possible by the opportunistic behavior of borrowers. In this respect, theforemost measurable aspect of asset quality is information on the likelihood of repayment, suchas the proportion of nonperforming loans, and repayment rates. The former measure indicates theproportion of loans that are not up to date in repayments, even if it is just one repayment. It is amore severe measure than the repayment rate, which measures the share of payments made as aproportion of the share due, but it provides a useful analogue to the information provided byrepayment rates— particularly in an environment in which very high repayment rates can quicklychange, and most MFIs have very high repayment rates. However, as the Practical Guide shows,the traditional repayment rate measure can underestimate the vulnerability of an MFI and lead tounderprovisioning, in a context of rapid portfolio growth or in the early stages of an installmentloan. The proportion of nonperforming loans or “portfolio at risk” indicator avoids this flaw by

Page 28: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

32

reflecting the true quality of the portfolio, although the approach might work less well in cases ofvillage banking, for instance, where group loans can be partially repaid.

Management

Management of MFIs is at the heart of successful microenterprise finance. Like themeasurement of capital adequacy, however, it is difficult to put management concerns into simpleanalytic measures or boxes, such as those of CAMEL ratings. Just as it is difficult to define andfully appreciate the operational risk facing sophisticated financial institutions operating in well-developed capital markets, it is even more difficult to evaluate management concerns at MFIs.Nonetheless, management concerns of MFIs are a key issue, and efforts to measure and evaluatethese concerns can afford a more disciplined perspective, as well as one that can highlight theinteractions of the various constraints to identifying and achieving sustainable microenterprisefinance.

Much MFI poverty-lending supported by donors is governed by the objective to expandoutreach. This scaling up may indeed be a condition for meaningful poverty alleviation. It mayalso be good management if there are economies of scale, as was experienced by BancoSol andKupedes. However, keeping growth within limits determined by managerial capacity, internalcontrol, and sensitivity to risk is essential. It is important that MFIs resist donor pressures todisburse funds more rapidly than prudence dictates.

Caution is particularl y important because MFIs are ultimately a financial innovation, andit takes time to discover the risks of such innovations. This basic uncertainty presents the largestchallenge to MFI sustainability. Risks are unlikely to be managed well in the absence of effectiverisk-oriented information systems. Risk management in response to information makes itpossible for sustainable expansion to occur more rapidly than would otherwise be possible,providing a basis for continuing innovation and responsiveness.

Earnings

Microenterprise finance cannot be formalized in an environment in which thegovernment regulates, controls and manipulates interest rates, as is the case in many financialsystems. Nevertheless, the resiliency of these institutions in hostile environments is one of thehallmarks of this innovation. A study of 28 MFIs published in the Microbanking Bulletin (1997)finds that 21 institutions had a positive real return on assets after all adjustments, although theywere operating in very different economic, institutional, and lending environments. Charginghigh interest rates remains the quickest way to achieve these returns. The study shows that MFIclients in high-inflation countries are more likely to accept high interest rates; Latin AmericanMFIs typically charge average nominal rates of interest twice as high as those charged by MFIs inthe rest of the world (but most of that higher interest yield is consumed by higher operatingcosts). Many countries still have interest rate ceilings that are incompatible with sustainability; asemphasized in the text, the policy and regulatory environment is not favorable for most MFIs.

Page 29: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

33

Liquidity

Performance measures for analyzing funding decisions and sources of funds are readilyavailable because this topic is the basis for much of the analyses of financial institutions, as wellas for recent Bank work on performance indicators for the financial sector. These measures focuson such traditional staples of financial intermediary analysis as the share of assets in loans andliquidity; the projected growth rate of the portfolio; the various funding sources of the portfolio;interest rate margins; reserves; operating and administrative costs; and the rate of return on assetsand investments. While these measures are useful, it is only recently that rigorous public financeevaluations have been applied to financial market interventions. Consequently, as was the casewith a number of the CAMEL standards, a rigorous quantitative performance standard for theliquidity measure cannot be easily developed. Nevertheless, like the management standard,performance on this score is significantly more important for MFIs than it is for typical financialintermediaries. The importance of liquidity is intensified because of the special relationship thatMFIs usually develop with their borrowers. This special relationship arises because one of thecentral motivations for borrowers to repay is the expectation that future loans will be availablewhen they are needed. If an MFI is unable to fulfill this expectation, a major source of disciplinecan quickly disappear, and failure to repay can occur quickly and widely. Hence, once again thelack of precision of the indicators in no way deprecates the importance of the concept to thesustainability of MFIs.

Subsidy Dependence Index

The SDI was developed by the Bank in 1991 to measure the explicit and implicitsubsidies (including the imputed cost of the financial institution’s net worth) and determine thechange required in the average on-lending interest rate to cover subsidies. Subsidies arecategorized in three areas: grants and concessionally priced borrowed funds, equity contributions,and in-kind donations. By pricing these three sources of subsidies at market rates, the totalamount of subsidies can be taken as a percentage of the loan portfolio, multiplied by the averageon-lending interest rate. This calculation yields the percentage change required in the average on-lending interest rate to overcome the effects of the three sources of subsidies and become“subsidy independent.” This equilibrium point, marked by the SDI, equates the return on equity,net of any subsidy received, with the opportunity cost of funds.

Page 30: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

34

Annex 3

Completed Projects

The study reviewed 15 completed projects having a microfinance component, from FY87to FY93. Three projects were stand-alone microfinance projects; the others had a microfinancecomponent representing from 1 percent to 30 percent of total investment. In three instances themicrofinance scheme was part of a social fund.

The design of each microfinance program has been assessed by considering whether theproject design conveys any of the basic information that would be used in a CAMEL analysis,and projects have been classified in two categories: a financial systems (FS) approach when thedesign provided such data, and a nonfinancial systems (NFS) approach when project design didnot.

Overall, the exercise reveals a strong correlation between sound financial design andsuccessful outcome: all 5 projects that were designed following an FS perspective had asatisfactory outcome, while only 1 of the 10 projects that followed a NFS design was satisfactoryin outcome.

Among the six projects completed in the Africa Region, all h ad an NFS design, and fivewere unsatisfactory in outcome. Five projects were completed in the East Asia and PacificRegion (EAP), of which two followed an FS design and had a successful outcome. There werethree completed projects in the Latin America and Caribbean Region (LAC), of which two wereFS in design and satisfactory in outcome. One remaining completed project was from the Europeand Central Asia Region (ECA); it had an FS design and a successful outcome.

I. Côte d'Ivoire 1987 / Third Urban

The outcome of the project overall has been rated unsatisfactory, with uncertainsustainability and negligible institutional development. The microfinance component was a pilotcredit program for microenterprises in the informal sector that represented 1 percent of the totalinvestment; its design followed a nonfinancial systems (NFS) approach. It has, however, beensuccessfully implemented and rated highly satisfactory by the Implementation Completion Report(ICR). The program has reportedly financed 354 loans, for a total of CFAF 160 million, to smallbusiness in Abidjan, with an average recovery rate of almost 90 percent (10 percentage pointsabove the initial target). According to the task manager, Jean Mazurelle, two main reasonsaccount for the success of this scheme: first, implementation was gradual, focusing initially on alimited number of neighborhoods and later expanding to Abidjan and to cities of the interior;second, local NGOs were selected and trained by an international NGO before being given fullresponsibility for assessing microfinance needs and carrying out recovery tasks. Theinternational NGO had a full-time officer who worked permanently with the government onimplementing the microfinance scheme.

Page 31: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

35

II. Indonesia 1987 / BRI-Kupedes Small Credit

This stand-alone microfinance project was designed following a financial systems (FS)approach; its outcome was rated satisfactory, with likely sustainability and substantialinstitutional development. The Performance Audit Report (PAR) prepared by the OperationsEvaluation Department (OED) in June 1995 argues that the following principal factors affectedthe outcome: (1) simple project design, with the focus given to resource generation at the Unitsrather than simply to deliver credit; (2) adequate financial policies favoring nontargeted creditexcept for maximum loan size, nonsubsidized lending, and self-sustainability through depositsmobilization; (3) staff incentives, which put the emphasis on monitoring, supervision, andaccountability; (4) government commitment and a favorable macroeconomic context, whichfavored growth of domestic savings in rural and small households.

III. Mexico 1987 / Fourth Small- and Medium-Scale Industry

The overall project outcome was rated unsatisfactory, with unlikely sustainability andmodest institutional development. The pilot microfinance component represented 10 percent ofthe total loan size, and its design did not conform to an FS approach. The ICR suggests that thecomponent achieved some outreach but produced mixed results in financial sustainability.Despite requirements to assess some PFIs' financial soundness, Bank supervision initiallyhindered disbursements and later turned out to be lax or inexistent. Contrary to what had beenplanned, NGOs did not serve as PFIs; rather, this role was assumed by credit unions. The ICRdoes not indicate recovery rates.

IV. Mozambique 1989 / Urban Rehabilitation

The overall project outcome was rated unsatisfactory, with unlikely sustainability andnegligible institutional development. The microfinance component accounted for 3 percent oftotal loan size, and its design did not follow an FS approach. According to the ICR its outcomewas unsatisfactory, with only half of the planned credit line being disbursed. Confusion betweenthe implementing agency and the commercial bank about accountability induced a poor follow-upon the loans. Although there is very little information available on the repayment of loans, therecovery of the line of credit has been estimated to be lower than 30 percent.

V. Guinea 1989 / Socioeconomic Development Support

The project overall was rated satisfactory in outcome, with uncertain sustainability andnegligible institutional development. The microfinance component accounted for less than 5percent of the total loan size and was part of a pilot program set up as a social fund. Its designfollowed an NFS approach; according to the task manager, Mark Woodward, the microfinancepilot was largely a failure. Originally, the scheme was denied the capacity to handle creditsbecause of lack of appropriate expertise. After hiring technical assistance, some credits wereallowed on an experimental basis, but without success. The ICR reports that the pilot componentlacked capacity, both on the part of NGOs and promoters, to prepare subprojects, and on the partof its management unit to appraise them and process requests for financing. It was alsorecognized that mixing credits and grants makes the management of social funds particularlyrisky and difficult; consequently, that social funds should subcontract activities instead of tryingto do everything by themselves.

Page 32: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

36

VI. São Tomé and Príncipe 1989 / Multisector

The project overall had an unsatisfactory outcome, with unlik ely sustainability andmodest institutional development. A 25 percent microfinance component had been planned atappraisal, with an NFS design. According to the ICR the objectives of the project with regard toprivate sector development were not met, and most of the line of credit for that sector wascanceled. Efforts to assist private sector growth were not successful: three out of the four privatesector projects financed were not implemented.

VII. Nigeria 1989 / Private Small and Medium Enterprise Development

The project overall was rated unsatisfactory in outcome, with unlikely sustainability andmodest institutional development. The project had a small microfinance component that wasdesigned following an NFS approach. Its outcome was unsatisfactory. According to the taskmanager, Manoucher Ashouripour, the main reasons for the failure of this component are: (1) itwas embedded in a project that was designed to stimulate medium-size and small enterprisedevelopment, not microfinance; (2) entrepreneurs lacked awareness of the existence of theprogram; (3) commercial banks were not willing to participate in the component because of lackof collateral; (4) the general macroeconomic context was not favorable, with high inflation andcurrency devaluation. The project was cut down 50 percent at midstream, with no subsequentfollow-up.

VIII. Cameroon 1990 / Social Dimensions of Adjustment

The overall project outcome was rated unsatisfactory, with unlikely sustainability andnegligible institutional development. The microfinance component accounted for about 30percent of the total investment and was designed following an NFS approach. The ICR suggeststhat the component performed poorly, although the government of Cameroon's final report claimsthat the program contributed to the development of eight microenterprises and some employmentcreation. The project was affected by financial mismanagement, and there is no documentaryevidence on the rate of success of the microenterprises.

IX. China 1991 / Fourth Rural Credit Project

The project overall has not yet been rated. The microfinance component accounted forabout 4 percent of the total Bank loan and was designed following an NFS approach. Thecomponent was not successful. According to the task manager, Ramesh Deshpande, theAgricultural Bank of China (ABC) was not interested in favoring the implementation of aprogram that the overall Bank project design was planning to take away from its jurisdiction. Afew loans were made to members of credit unions, with an average size of US$5,000–10,000 andone to two months term. The microfinance part also brought about better policy dialogue andsome institutional strengthening, but in the future microfinance projects should be managedseparately and placed under the administrative control of a newly created federation ofcooperatives, instead of ABC's tutelage.

X. Honduras 1991 / Social Investment Fund

The overall project outcome was satisfactory, with likely sustainability and substantialinstitutional development. The microfinance component accounted for about 10 percent of thetotal investment and was designed following an FS approach. This component was successful.Key factors of the program's success are given in the Bank's Case Study of the Honduras Social

Page 33: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

37

Investment Fund Project (Webb, Lee, and Sant’Anna 1995) , and include: (1) delegation oflending operations to NGOs specialized in informal sector credit, (2) market rates were chargedto final beneficiaries, (3) credit risked was transferred to intermediary organizations in order togive them the incentive to collect subloans; (4) solidarity group lending ensured loan recovery;(5) prequalification criteria applied to beneficiaries and regular supervision of loandisbursements.

According to the task manager, Anna Maria Sant'Anna, repayment rates were high andthe program evolved to include the participation of commercial banks. The program started witha relatively large number of NGOs and later shifted to a smaller number of very experiencedNGOs, a change that contributed to the success of the scheme.

XI. Indonesia 1991 / Second BRI-Kupedes Small Credit

This is a stand-alone microfinance project that was designed following an FS approachand was rated satisfactory in outcome, with likely sustainability and substantial institutionaldevelopment. The ICR lists several factors that have contributed to the success of the project: (1)economic growth and rapid growth of incomes in the rural areas, resulting in increasing demandfor credit; (2) the program was designed with the support of the government, and it was closely inline with the local environment and local conditions; (3) simple project design with an emphasison financial sustainability; (4) an incentive system setting clear targets and standards, closefollow-up and supervision, performance incentives and punishment; (5) appropriate staffing andtraining; and (6) technical assistance.

XII. Philippines 1991 / Cottage Enterprise Finance

This other stand-alone microfinance project was designed following an NFS approach,and its outcome has been rated unsatisfactory, with unlikely sustainability and negligibleinstitutional development. The PCR as well as the PAR prepared by OED identified thefollowing main factors explaining the failure of the program: (1) the program had attempted tocreate a new substitute for collateral by developing mutual guarantee associations (MGAs)governed by too expensive and burdensome regulations; (2) the project was complex and couldnot rely on prior experience in guarantee associations schemes; (3) project design was not clearand did not allow easy access to credit, in a context of poor information and limited bankingtechnology; (4) the project chose to work through accredited financial institutions instead ofmaking use of an existing network of rural banks; (5) commercial banks were uninterested inlending to the MGAs; (6) participants in MGAs should have been chosen more carefully andlocated in the same area as lenders in order to cut travel time and allow personal follow-up.

XIII. Philippines 1991 / Rural Finance

The project overall was rated satisfactory in outcome, with likely sustainability andmodest institutional development. It contained a small microfinance component that was part ofa pilot plan to finance farmer cooperatives. The component was designed following an NFSapproach and never materialized owing to interest rate differences.

Page 34: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

38

XIV. Honduras 1992 / Second Social Investment Fund

The project overall has been rated satisfactory i n outcome, with uncertain sustainabilityand modest institutional development. The microfinance component accounted for 13 percent ofthe total credit line and was designed following an FS perspective. Its outcome has beensuccessful. The audit prepared by OED indicates that, as of July 1994, the program had financed12,000 microenterprises out of a population of 250,000. The loan repayment rate has beenexcellent, higher than the private banking rate. These results are attributable principally to: (1)clear, detailed and rigidly enforced regulations; (2) a high level of supervision, and technicalassistance provided to the NGOs and through them (any NGO with more than 5 percent arrearswas eliminated from the program).

XV. Albania 1993 / Rural Poverty Alleviation Pilot

The overall project outcome was satisfactory, with likely sustainability and substantialinstitutional development. It comprised a microfinance component that accounted for 10 percentof the total credit line and was designed following an FS perspective. Its outcome was successfulaccording to the audit prepared by OED in July 1995. Repayment rates are high and usuallyahead of schedule at the village level (at the national level, however, repayment rates are zero; novillage has ever paid anything back to the Rural Development Bank, RDB, or to the government).

Page 35: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

39

Annex 4

The Evaluative Framework

The Bank rates projects according to three results-oriented measures: outcome,sustainability, and institutional development (ID). In this study we develop measures of likelyoutcome and likely sustainability. We do not measure likely ID impact because it isfundamentally affected by achievement of ID objectives rather than project design issues.Because OED’s evaluation of project outcome is based on judgments about project relevance,efficacy, and efficiency, we first discuss these measures.

Relevance

The standard OED definition of project relevance is that the project’s goals are consistentwith the country’s overall development strategy and the Bank’s assistance strategy for thecountry. To measure the relevance of MFI projects and their components, we reviewed the SARmaterial and completion reports, and in many cases talked to task managers. We used thismaterial to determine whether the project sought to rely on an institution that internalized theinformation-processing problems that underlie lending to poor, uncollateralized borrowers in ahostile legal and regulatory environment. In other words, we made a judgment whether afinancial system perspective was at least implicit in the project design. Did the design payattention to capital adequacy, asset quality or management issues? Also, a judgment was madewhether the legal— for example, usury laws— or regulatory environment was such that the projectcould eventually be presumed to become more than a donor-assisted enclave.

Our rankings compress the Bank’s five-level standard for relevance— which ranges fromhighly relevant to negligible— into a two-standard rating: either the project design was judged tobe relevant, or it was not. Forty-one of 72 projects, or 57 percent, for which there was sufficientinformation to form a judgment, were found to be relevant.

Efficiency

The standard OED definition of efficiency is how well do project results compare withthe inputs in terms of costs, implementation times, and economic and financial returns. As wasthe case with our measure of relevance, our judgment of MFI efficiency used SARs, completionreports, and discussions with task managers. The factors that affected our judgment were: (1)Was the MFI able to deliver resources to its clients in a cost effective way? If the MFI was ableto deliver its services without subsidies and continue to grow, it was assumed to meet a markettest of being efficient enough to not only survive but to grow; (2) If the MFI was not able todeliver its services without external assistance, as was the case in most MFIs, efficiency wasmeasured by how cost-effective and targeted were the subsidies involved. Were these subsidieslikely to go to the poor in amounts that appeared to be appropriate? For example, deep interestrate subsidies and very large loan size limits were taken as indicators that the subsidies were notlikely to go to those poor borrowers who would self-select into the program in a way thatencouraged productive borrowing by the poor. Similarly, loans made by MFIs that did not bearany risks were presumed to be inefficient. In other words, in some respects our measure ofrelevance relied on the first three standards of the CAMEL standard, while in many respects ourmeasure of efficiency can be said to correspond to the last three components of the CAMELSstandard: MFI earnings, its liquidity and its targeting of any subsidies. If a project supported an

Page 36: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

40

MFI to achieve profitability by mobilizing resources effectively, while maintaining adequateliquidity, and provided well-targeted subsidies, it was judged to be efficient.

As was the case with relevance, our rankings are two: effici ent and inefficient. Forty-three of the 73 projects, or 59 percent, for which there was sufficient information to form ajudgment, were found to be efficient.

Efficacy

As noted in the text, efficacy is concerned with project implementation achievements, as towhether a project achieved its objectives, be they physical, financial or institutional. Clearly, theefficacy of a project is fundamentally affected by implementation and external factors unrelatedto project design. Consequently, this aspect of outcome was not considered in our estimates ofthe likely relationship between project design and outcome. As is discussed in the text, theneglect of efficacy in our measure of expected outcome is not because of a view that designissues are not important for project efficacy. In fact, the review of the 15 completed projectsindicates that the opposite is the case: All the projects with a financial system design had asatisfactory outcome and were efficacious; and the converse is similar. Indeed, the result that 14out of 15 outcomes could have been predicted by project design led to the view that, if thestructure of a microenterprise finance project was not relevant and efficient at the design stage, itwould be very difficult for either effective implementation or simply “good luck” to overcomethese design flaws. Nevertheless, because of the limited number of completed MFI projects wealso undertook an analysis of the validity of this perspective for the cohort of all Bank-supportedprojects closed in FY96.

Outcome and Design

Because of the views on the importance of design for effective MFI operation and becauseof Mosley and Hulme’s empirical support for this view, we did not expect design to play asimportant a role in all Bank-supported projects as it does in MFIs. Nevertheless, in order for ourestimates of likely outcome to be credible, our method should be a relatively strong predictor ofthe ultimate outcomes of all Bank projects.

It was. The analysis indicates that 83 percent of projects with satisfactory outcomes andat least moderate efficacy— so that the project’s outcome was not adversely affected byimplementation— were both relevant and efficient. As a result, only one project in six would bepredicted as a “false negative” by the use of our criteria. Similarly, 86 percent of projects withunsatisfactory outcomes had either modest or negligible accomplishments on either relevance orefficiency, suggesting that only 14 percent of unsatisfactory projects would be predicted ashaving likely satisfactory outcome (that is, were “false positives”). In other words, use of ourstandards for all Bank-supported projects would predict ultimate project outcome correctly inabout six cases out of seven.

To recapitulate, our estimates of likely outcome are based on the assumption that projectswith at least moderate relevance and efficiency are likely to have satisfactory outcomes;conversely projects with lower ratings either for relevance or efficiency are likely to haveunsatisfactory outcomes. Our estimates of likely sustainability, as defined in the text in para.3.12, follows a similar approach.

Page 37: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

41

Comparison with QAG Data

A straightforward comparison of our findings and QAG data is not possible, but asuggestive comparison can be made. In our sample of microfinance projects, we rely only onthose projects which had a microfinance component exceeding 10 percent of overall project size,because the QAG figures apply to the entire project, rather than project components. There were47 such projects. Of these 47 projects, 27 had both a QAG rating and an estimated outcome byour methods. Using our approach, we estimate that 57 percent of these projects would have asatisfactory outcome. This figure is somewhat lower than a likely 70 percent satisfactoryoutcome can be inferred from a simplified use of QAG data.

The QAG implied estimate of likely outcome, everything else constant, relies on theevidence that, if the 68 percent of MFI projects in our subsample not at risk follow the pattern ofall projects reviewed by QAG, then 83 percent of these are likely to have a satisfactory outcome.Similarly, of the 16 percent of projects potentially at risk, 57 percent are likely to havesatisfactory outcomes, and 28 percent of the 16 percent of projects actually at risk are likely tohave a satisfactory outcome. Cumulatively, this means that (.68 x .83) + (.16 x .57) + (.16 x .28)= .70 satisfactory. However, one element of the QAG rating process is how risky a sector andcountry have been in the past. If either has been riskier, as MFIs have been, it is more likely to berated risky now by QAG measures. Consequently, according to QAG methodology, it would geta flag for risky sector performance. By ignoring these risk elements, we are causing our impliedQAG estimates to be higher than they would be if we relied strictly on QAG measures. We donot know the effect that the neglect of this “risk” flag would have on the comparisons. Nor canwe be sure that the QAG estimates for the entire project in which MFI projects are embeddedwould be the same as a QAG analysis of only the MFI component. Nevertheless, our findings areconsistent with what can be inferred from QAG’s data: Bank-supported MFIs tend to be riskierthan Bank-supported projects in general.

Page 38: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

42

Annex 5

Management Action Record

Page 39: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

46

Bibliography

Akerloff, George (1970). “The Market for ‘Lemons’: Qualitative Uncertainty and the MarketMechanism.” Quarterly Journal of Economics 84: 488-500.

Barltrop, Chris J. and Diana McNaughton (1992). Banking Institutions in Developing Markets—Interpreting Financial Statements, vol. 2. Washington, D.C.: World Bank.

Bennett, Lynn, and Carlos Cuevas (eds.) (1996). “Sustainable Banking with the Poor.” Journalof International Development (Special Issue) 8 (2), March-April.

Bhatt, Ela (1996). “Beyond Microcredit: Structures that Increase the Economic Power of thePoor.” India: Sewa Academy.

CGAP (1997). A Review of the World Bank's Microfinance Portfolio (FY91–FY96).

Committee of Donor Agencies for Small Enterprise Development and Donors’ Working Groupon Financial Sector Development (1995). “Micro and Small Enterprise Finance: GuidingPrinciples for Selecting and Supporting Intermediaries.”

Comptroller of the Currency (1996). Remarks by Eugene A. Ludwig before the Women’s SelfEmployment Project, March 14, 1995. News Release no. 96-33.

Dessing, Maryke (1990). Support for Microenterprises: Lessons for Sub-Saharan Africa. WorldBank Technical Paper No. 122, Africa Technical Department Series. Washington, D.C.:World Bank.

Fruman, Cécile, and Mike Goldberg (1997). “Microfinance Practical Guide for World BankStaff.” The Consultative Group to Assist the Poorest, and Sustainable Banking with thePoor.

Garson, Jose (1996). Microfinance and Anti-Poverty Strategies. A Donor Perspective. PolicySeries Working Paper. New York: UNDCF.

Gonzalez-Vega, Claudio, and others (1997). “The Challenge of Growth for MicrofinanceOrganisations: The Case of Banco Solidario in Bolivia.” In Hartmut Schneider, ed.,Microfinance for the Poor? Paris: OECD.

Gurgand, Mark, Glenn Pederson, and Jacob Yaron (1994). Outreach and Sustainability of SixRural Finance Institutions in Sub-Saharan Africa. World Bank Discussion Papers 248.Washington, D.C.

Hashemi, Syed M. (1997). “Building up Capacity for Banking with the Poor: The GrameenBank in Bangladesh.” In Hartmut Schneider, ed., Microfinance for the Poor? Paris:OECD.

Hoff, Karla, and Joseph E. Stiglitz (1996). “Imperfect Information and Rural Credit Markets:Puzzles and Policy Perspectives.” In The Economics of Rural Organization— Theory,Practice, and Policy. Washington, D.C.: World Bank.

Page 40: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

47

Hoff, Karla, Avishay Braverman, and Joseph E. Stiglitz (eds.) (1996). The Economics of RuralOrganization— Theory, Practice, and Policy. Washington, D.C.: World Bank.

IDB (1985). “Evaluation of the Program for the Financing of Small Projects.” OperationsEvaluation Office.

_____ (1991). “Evaluation report on the IDB and microenterprise: A development strategy forthe 90s.”

_____ (1991). External Review and Evaluation System. Evaluation Report on the IDB andMicroenterprise. A Development Strategy for the 90s.

_____ (1997). “Three Years of Support To Micro And Small Enterprises.”

IFAD (1997a). “IFAD Microfinance Experience and Plan of Action.” Paper presented at theMicrocredit Summit, 2-4 February 1997, Washington, D.C.

_____ (1997b). “Declaration and Plan of Action.” Paper presented at the Microcredit Summit,2-4 February 1997, Washington, D.C.

Islam, Reazul, J. D. Von Pischke, and J. M. de Waard (eds.) (1994). Small Firms InformallyFinanced. Studies from Bangladesh. World Bank Discussion Paper 253. Washington,D.C.

Johnson, Susan, and Ben Rogaly (1997). Microfinance and Poverty Reduction, OxfamDevelopment Guidelines. Oxford, U.K.: Oxfam.

Kahnert, Friedrich (1989). “The Small-Scale Enterprise Credit Program (S.S.E.P.) under theSecond and Third Calcutta Urban Development projects (CUPDII and CUPD III).”World Bank Assessment Internal Discussion Paper, Asia Regional Series, Report no.IDP-28, Washington, D.C.

Khandker, Shahidur, R., Baqui Khalily, and Zahed Khan (1995). Grameen Bank: Performanceand Sustainability. World Bank Discussion Paper 306. Washington, D.C.

Khandker, Shahidur R., Zahed Khan, and Baqui Khalily (1995 ). Sustainability of a GovernmentTargeted Credit Program: Evidence From Bangladesh. World Bank Discussion Paper316. Washington, D.C..

Levitsky, Jacob (ed.) (1989). “Microenterprise in Developing Countries.” In Papers andProceedings of an International Conference held in Washington, D.C., USA, 6-9 June1988. London: Intermediate Technology Publications.

Malhotra, Mohini (1996). “Credit Subsidies.” In Lending for Small Enterprises 1989-93, WorldBank Technical Paper no. 311. Washington, D.C.

Microbanking Bulletin. (1997). The Microfinance Bulletin. Financial Performance ofOrganizations that Provide Banking Services to the Poor, Vol. 1, No. 1. Boulder, Colo.:Microfinance Program at the Economics Institute.

Page 41: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

48

Mosley, Paul and David Hulme (1998). “Microenterprise Finance: Is There a Conflict BetweenGrowth and Poverty Alleviation?.” In World Development, Vol. 26, No. 5. UK:Pergamon.

OECD (1996). “Microcredit in Transitional Economies.” OECD/CCET website.

Pitt, Mark M., and Shahidur R. Khandker (1996). Household and Intrahousehold Impact of theGrameen Bank and Similar Targeted Credit programs in Bangladesh. World BankDiscussion Paper 320. Washington, D.C.

Robinson, Marguerite (1995). Addressing Some Key Questions on Finance and Poverty. WorldBank Sustainable Banking with the Poor Occasional Paper 2. Washington, D.C.

Sacay, Orlando J., and Bikki K. Randhawa (1995). Design Issues in Rural Finance. World BankDiscussion Paper 293. Washington, D.C.

Saltzman, Sonia B., Rachel Rock, and Darcy Salinger (1998). “Performance and Standards inMicrofinance: ACCION’s Experience with the CAMEL Instrument.” Discussion PaperSeries, document no. 7. Sommerville, Mass.: ACCION.

Schor, G., and J. P. Alberti (1995). Training Voucher Schemes for Microenterprise in Paraguay.Benefits of a Market-based Approach to Government Intervention. Montevideo: GAMA.

Sinha, Saurabh, and Imran Matin (1998). “Informal Credit Transactions of Micro-creditBorrowers in Rural Bangladesh.”

Siraj, Khalid, and Bikki Randhawa (1997). Review of Social Fund Microfinance Components.

Subbarao, K., and others (1996). “Safety Net Programs and Poverty Reduction: Lessons fromCross-Country Experience.” World Bank Poverty and Social Policy Department,Washington. D.C.

The SEEP Network and Calmeadow (1995). Financial Ratio Analysis of Micro-FinanceInstitutions. New York: Pact.

UNIFEM (1995). Annual Report.

United Nations (1998). “The Role of Microcredit in the Eradication of Poverty.” Draft Report ofthe Secretary General.

U.S. Office of The Controller, Operations Evaluation Office (1985). Evaluation of The Programfor The Financing of Small Projects. Washington, D.C.

Van Greuning, Hennie, Joselito Gallardo, and Bikki Randhawa (1998). “A Framework for theRegulation of Microfinance Institutions.” Discussion draft. World Bank FinancialSector Development Department.

Vittas, Dimitri (1997). “Thrift Deposit Institutions in Europe and the United States.” In GerardCaprio, Jr. and Dimitri Vittas (eds.) Reforming Financial Systems – HistoricalImplications for Policy. New York: Cambridge University Press.

Page 42: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

49

Webb, Anna Kathryn Vandever, Kye Woo Lee, and Anna Maria Sant'Anna (1995). TheParticipation of Nongovernmental Organizations in Poverty Alleviation, A Case Study ofthe Honduras Social Investment Fund Project, World Bank Discussion Paper 295.Washington, D.C.

Webster, Leila M., Randall Riopelle, and Anne-Marie Chidzero (1996). Lending for SmallEnterprises, 1989-93. World Bank Technical Paper 311. Washington, D.C..

Women’s World Banking (1995). “Global Policy Forum: The Missing Links: Financial SystemsThat Work For The Majority.” New York: WWB.

World Bank (1996a). “A Worldwide Inventory of Microfinance Institutions.” SustainableBanking with the Poor. Washington, D.C.

_____ (1996b). “KUPEDES: Indonesia’s Model Small Credit Program.” Operations EvaluationDepartment Précis, February. Washington, D.C.

_____ (1996c). “Portfolio Improvement Program— A Review of Social Fund MicrofinanceComponents.” Financial Sector Development Department. Washington, D.C.

_____ (1997a). “Annual Review of Development Effectiveness.” Operations EvaluationDepartment Report 17196. Washington, D.C.

_____ (1997b). “Portfolio Improvement Program: Review of Financial Intermediary LendingPortfolio.” Financial Sector Development Department. Washington, D.C.

_____ (1998). “A Framework for World Bank Group Support for Development of Micro, SmallEnterprise, and Rural Finance in Sub-Saharan Africa.” Private Sector Finance Group,Africa Region. Washington, D.C.

_____ (1998). “Financial Sector Reform. A Review of World Bank Assistance.” OperationsEvaluation Department. Report 17454. Washington, D.C.

Yaron, Jacob (1992). Assessing Development finance Institutions— A Public Interest Analysis.World Bank Discussion Paper 174. Washington, D.C.: World Bank.

Executive Summary

1. Since 1995, when the World Bank joined the donor community to establish theConsultative Group to Assist the Poorest (CGAP), the Bank has become increasingly involved insupporting microenterprise finance. The Bank is a newcomer to this field, having been onlymodestly involved before CGAP’s establishment. However, it now has an active lendingprogram and a clear strategy known as the financial systems perspective.

2. This evaluation considers the Bank’s approach to microenterprise finance. It is based ona review of staff appraisal reports for current projects and completion reports for completedprojects. As such, it cannot evaluate many of the important impact issues involved in theseprojects. It cannot, for example, analyze the effects of the projects on the poor and on genderissues. In addition, because so few projects have been completed (15), it cannot follow the usualOED methodology of culling lessons only from completed projects.

Page 43: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

50

3. Nevertheless, an evaluative study of the Bank’s activities in this area is important at thistime for three reasons. First, microenterprise finance appears to be one of the most promisinginnovations to support poverty alleviation in a sustainable way. Second, the Bank has a rapidlygrowing involvement in this area and now has a portfolio of more than 90 projects. Finally, theBank’s adoption of a financial systems approach to supporting these projects provides a basicstandard against which project designs can be evaluated.

4. Applying the CAMELS 29 approach (underlying the Bank’s new Operational PolicyStatement 8.30) to the design of the portfolio of ongoing projects indicates that there has been adramatic improvement over FY93-97. However, it also suggests that microenterprise finance isstill performing less effectively than Bank projects do in general, and that many Bank-supportedmicroenterprise projects are unlikely to be sustainable. Finally, the review of available ICRs andof the project design of ongoing microenterprise finance projects indicates that there is significantroom to enhance the consistency of Bank practices with the current Bank approach.

Specific Findings

5. The central finding is that greater emphasis should be given to the commercial viabilityof microfinance institutions. Earlier Bank microenterprise finance projects paid little attention tothe financial details of participating institutions, such as loan recovery rates or costs of grantingand administering loans. Recent Bank projects provide more information, but it is still notpossible to tell if the institutions supported are designed to ever be financially viable. Bankexperience shows that this lack of attention to financial detail weakens performance.

6. Many of the early Bank microfinance projects were desi gned on the premise thatsustainability was not possible and that subsidies would always be required. These subsidieshave traditionally been in the form of below-market interest rates for borrowers which distortresource allocation. Also, experience indicates that higher-income borrowers are better placed tocapture these subsidies. This report thus argues that it is better to deliver subsidies as incentivesto participating institutions to develop the necessary infrastructure to reach the targetedpopulation. At the same time, subsidies in Bank microfinance projects are better targeted tomicroenterprise finance institutions (MFIs) with the greatest potential for sustainability, in orderto provide appropriate incentives and discipline.

7. CGAP represents a key step in improving donor support. But stronger donorcoordination is needed to ensure that assistance to the poor is sustainable. Donor actionsconsistent with the Guiding Principles for Selecting and Supporting Microfinance Institutions(Committee of Donor Agencies for Small Enterprise Development and Donors’ Working Groupon Financial Sector Development, 1995) would go a long way toward addressing this problem.To contribute to this task, the Bank needs to be consistent in its own projects, and demonstratethat microfinance can reach the poor through sustainable institutions.

8. The Africa Region’s Strategy for Development of Micro, SME, and Rural Financeprovides an example of the appropriate form of Bank support . The paper’s emphasis on raisingawareness and policy dialogue, as well as its de-emphasis of lending, strikes an appropriatebalance between lending and other forms of Bank support. In contrast to recent calls to increase

29 CAMEL— an acronym for Capital, Assets, Management, Earnings and Liquidity — is the most widelyused regulatory rating system for financial institutions. In the microfinance case, the added S represents theefficiency of outreach to the poor.

Page 44: The World Bank and Microenterprise Finance: From …...The World Bank and Microenterprise Finance: From Concept to Practice November 15, 1999 Operations Evaluation Department Document

51

resources, the paper implies that the availability of funding for MFIs should not be seen as themain constraint to wider use of this innovation. In fact, the capacity to absorb large amounts ofdonor funding is often constrained by institutional development requirements.

Lessons Learned

9. Lesson 1. Bank support for MFIs should focus on advisory services or should take theform of Adaptable Program Loans (APLs) or Learning and Innovation Loans (LILs). APLs andLILs seem particularly appropriate because they provide a vehicle that would permit the Bank toextend the mix of financial, analytical, and advisory services that may be needed over a longperiod of time. The staffing and supervision of these projects will require more inputs anddifferent reporting outputs than traditional intermediation loans. This kind of shift in approach isalready beginning as three LILs and three APLs have been done in the last two years.

10. Lesson 2. The design of MFI-Bank projects is increasingly based on the financialsystems perspective articulated by the donors’ Guiding Principles, O.P. 8.30, and CGAP’sevolving lessons of best practices. This has resulted in a significant improvement in projectdesign and probable project performance. However, our findings suggest that, without furtherimprovements in design, it is likely that MFI projects will continue to perform less well than thetypical Bank project. Documentation of MFI projects should include the presentation of simplesummary statistics of how well a proposed project satisfies the financial systems perspective.The use of the Subsidy Dependence Index (SDI)— a measure that the Bank recommends for usein all financial sector projects— should be encouraged, possibly complemented by otherperformance indicators such as those proposed in the CGAP-Bank-developed Practical Guide.

11. Lesson 3. Bank management should report on the consistency of its operations with O.P.8.30 and the general principles agreed to in the Donor’s Guiding Principles. The donors’adoption of the financial systems perspective carries with it a particular approach to supportingMFIs, which opposes conferring interest rate subsidies on ultimate borrowers. It does not appearthat the Bank follows these policies consistently.

12. The Informal Notes of the CODE Subcommittee meeting summarizes the managementresponse to these recommendations (see Attachment 2).