Micro Finance - Emerging Trends and Challenges

Download Micro Finance - Emerging Trends and Challenges

Post on 21-Feb-2015




0 download


MicronanceMicronanceEmerging Trends and ChallengesEdited bySuresh SundaresanProfessor of Economics and Finance, Columbia Business School, USAEdward ElgarCheltenham, UK Northampton, MA, USA Suresh Sundaresan 2008 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical or photocopying, recording, or otherwise without the prior permission of the publisher. Published by Edward Elgar Publishing Limited The Lypiatts 15 Lansdown Road Cheltenham Glos GL50 2JA UK Edward Elgar Publishing, Inc. William Pratt House 9 Dewey Court Northampton Massachusetts 01060 USAA catalogue record for this book is available from the British Library Library of Congress Control Number: 2008935924ISBN 978 1 84720 920 7 Printed and bound in Great Britain by MPG Books Ltd, Bodmin, CornwallContentsList of contributors Preface 1 The changing landscape of micronance Suresh Sundaresan 2 The role of international capital markets in micronance Brad Swanson 3 Securitization and micro-credit backed securities (MCBS) Ray Rahman and Saif Shah Mohammed 4 Cell phones for delivering micro-loans Anand Shrivastav 5 How should governments regulate micronance? Richard Rosenberg 6 Gender empowerment in micronance Beatriz Armendriz and Nigel Roome Index vi xi 1 25 46 71 85 108123vContributorsBeatriz Armendriz is a Lecturer in Economics at Harvard University, and a Senior Lecturer at University College London. She is also a research aliate at the David Rockefeller Center for Latin American Studies at Harvard University. She has taught at the London School of Economics and has worked as a Visiting Associate Professor at MIT, and as a Visiting Professor at the Toulouse School of Economics. Her research focuses on economic development and nance. Having published numerous articles on micronance, notably with Christian Gollier and Jonathan Morduch, she recently co-authored The Economics of Micronance, a book published by MIT Press in 2005. Her current research includes eldwork on micronance and gender empowerment, with researchers from the Innovations for Poverty Action (Yale and Harvard), and the Financial Access Initiative (Harvard, Yale, NYU). She is also co-editing a handbook on micronance, and a book, The Contemporary Latin American Economy, also for MIT Press. Lecturer Armendriz grew up in southern Mexico where she founded AlSol and Grameen Trust, Chiapas, the rst Grameen-style micronance organizations in the region. Saif Shah Mohammed is co-founder of MR Analytics. In 200506, he worked extensively on the BRAC micro-credit securitization, having moved to Bangladesh to complete and implement the transaction. Prior to joining MF Analytics, he worked as an analyst at Cornerstone Research, assisting industry and faculty experts in developing economic and nancial analyses in litigation contexts. Saif graduated from Harvard College in 2002 with a B.A. magna cum laude in economics. He currently attends the Columbia University School of Law. Ray Rahman is the founder and CEO of MF Analytics and one of the chief architects of the BRAC micro-credit securitization. Prior to starting MR Analytics, Ray was at Lehman Brothers, working on the cutting edge of asset securitization in the edgling commericial mortgage-backed security (CMBS) industry in the 1990s. He has also working in private equity, restructuring multinational companies with sales of over $1 billion. Rahman is a cofounder of Potenco, a green energy company that is creating the worlds most ecient hand-held power generator, targeted at the needsviContributorsviiof developing countries. The generator was featured in Wired magazine in the summer of 2007. Rahman is also founder of Fast Forward India, a nonprot organization that links graduate students from MIT, Stanford and IIT with local mentors and NGOs, to identify and solve key problems that directly aect the life of disadvantaged populations in India. FFI helps teams pilot test proposed solutions on the eld with local NGOs, obtain funding to grow and widen the scale and scope of impact. Rahman graduated with a BS degree in mechanical engineering (high distinction) and a BA in economics, (cum laude) from the University of Rochester. He has an MBA from MIT Sloan School of Management. Nigel Roome holds the Daniel Janssen Chair in Corporate Social Responsibility at Solvay Business School, Free University of Brussels, and the University Chair in Corporate Global Responsibility and Governance at TiasNimbas Business School, Tilburg University, Netherlands. He previously held chairs in The Netherlands and Canada, and academic positions in Britain. He is widely published on topics that relate business strategies, innovation, and technology to issues of corporate responsibility, sustainable development, and global change. His books include Management Education for Sustainability and Corporate Environmental Management (1994); Sustainability Strategies for Industry (1998); and The Ecology of Information and Communications Technologies (2002). Professor Roomes career has involved innovations in both education and research. His achievements as a European Faculty Pioneer over 20 years were acknowledged by The Aspen Institute in 2006. Roome is currently Chair of the Academic Board of the European Academy of Business in Society and is a member of the academys management board. In addition to these responsibilities Roome has served as an environment councilor, an advisor to the president of Ontario Hydro and the Oce of the Canadian Commissioner for Environment and Sustainable Development, a member of ABBs global stakeholder advisory group, an expert to the European Commissions Futures project, and Chair of the European Commissions expert group on competitive and sustainable production systems to the year 2020. He has been invited to contribute his views on corporate responsibility and sustainability to meetings of the European Union, under successive presidencies of Sweden, The Netherlands, Finland, and Germany. Richard Rosenberg holds a law degree from Harvard University. He has worked in international development since 1984 with USAID and the World Bank, concentrating on development nance, especially micronance. He has authored or co-authored two dozen publications on micronance, including several on issues associated with regulation andviiiContributorssupervision of micronance, and teaches regularly on that topic at the Boulder Institute of Micronance Training. Anand Shrivastav is the chief software architect and founder of Suvidha. He has over 25 years of experience in marketing and distribution of consumer, nutrition, and health care products besides telecom and banking services. He has set up projects, implemented marketing strategies, and managed large distribution channels. Shrivastav currently serves on the boards of the following organizations: Suvidha Starnet (mobile transaction services), as Chairman; and Genesis Biogen (stem cell technologies), as Chairman. In the past Shrivastav has served Intercorp Biotech (biotechnology, nutrition) as managing director, Hiperworld (information technology/ software) as managing director, The Home Store India (retail chain) as director, Intercorp Industries (hybrid seeds) as director, IEPCL (telecom, chemicals) as executive director, as marketing consultant for public sector units Kerala State Drugs & Pharmaceuticals and Kerala Soaps & Oils (consumer products), and as resident manager with RG Soft Drinks. Shrivastav attended Harvard University Graduate School of Engineering Design. Suresh Sundaresan is the Chase Manhattan Bank Professor of Economics and Finance at Columbia University. He is currently the Chair of the Finance subdivision. He has published in the areas of Treasury auctions, bidding, default risk, habit formation, term structure of interest rates, asset pricing, pension asset allocation, swaps, options, forwards, futures, xedincome securities markets, and risk management. His research papers have appeared in major journals such as the Journal of Finance, Review of Financial Studies, Journal of Business, Journal of Financial and Quantitative Analysis, European Economic Review, Journal of Banking and Finance, and Journal of Political Economy, among others. He has also contributed articles in the Financial Times and to World Bank conferences. He is an associate editor of Journal of Finance and Review of Derivatives Research. His current research focus is on default risk and how it aects asset pricing and sovereign debt securities. Sundaresan has consulted for Morgan Stanley Asset Management and Ernst and Young. His consulting work focuses on term structure models, swap pricing models, credit risk models, valuation, and risk management. He is the author of the book Fixed-income Markets and their Derivatives (1996). He has served on the Treasury Bond Markets Advisory Committee. He has served as a resident scholar at the Federal Reserve Bank of New York. He has worked with the Center for Micronance Research (CMFR) on issues relating to interest rates on micro-loans, the eects of repeated borrowings, and contract design.ContributorsixBrad Swanson is a partner in Developing World Markets, a socially responsible investment bank and fund manager, which he joined in 2003. As a banker and a diplomat, he has worked in emerging markets for 25 years and has done business in more than 50 countries. He began his career in the US State Department and was posted to West Africa. He then worked for Donaldson, Lufkin & Jenrette Securities Corp. in New York, Bankers Trust and Banque Nationale de Paris in London, and Global Environment Fund in Washington, DC. During spring and summer of 2004, he was on assignment with the US Department of Defense in Iraq running private-sector nance programs for the occupation government. He is a member of the Investment Committee of Partners for the Common Good, a non-prot community development nance organization, and has served on the boards of investment banks in Poland and the Ivory Coast. He has a BA from Princeton University and an MBA from Columbia University.PrefaceThis bookMicronance: Emerging Trends and Challengesbrings together recent practical innovations and policy questions in the eld of micronance, and is largely based on the contributions made by dierent scholars and practitioners to the conference hosted by the Social Enterprise Program of Columbia Business School on 20 April 2007 at Columbia University. This conference, entitled Credit Markets for the Poor: Focus on Micro-Finance, examined recent developments in the eld, research ndings, and the challenges that lie ahead. The contributions in this book, after the introductory chapter, focus on (1) integration of capital markets with micronance, (2) securitization and micro-credit backed securities, (3) technological innovations such as the delivery of banking services to the unbanked sector using mobile phone technology, (4) the regulatory challenges and opportunities as the landscape of micronance undergoes a sea change, and (5) the consequences of gender empowerment, where women, especially among the poorest, predominate micro-loan borrowings.xi1. The changing landscape of micronanceSuresh SundaresanINTRODUCTIONThe ability of households to save, access capital, and manage risk exposures of various kinds, such as life, property, and health through insurance is a prerequisite for their economic and social development. Access to basic nancial services (such as credit, savings, and insurance) is most likely to develop the entrepreneurial skills and opportunities among those poor who are currently outside the perimeter of such nancial markets and services. Furthermore, over time, such access will promote better risk management capabilities and promote the economic aspirations of the poor. The World Bank uses two reference benchmark levels of consumption/ income to measure poverty: a consumption level of (US) $1.08 per day and a consumption level of $2.15 per day.1 These levels are measured in 1993 purchasing parity terms. As of 2001, the World Bank estimated 1.1 billion people had consumption levels below $1 a day, and 2.7 billion lived on less than $2 a day. While these gures are very stark, it is also true that the proportion of people living under $1 a day has fallen from 28 percent in 1990 to 21 percent in 2001. This progress not withstanding, it remains clear that poverty alleviation should be a major priority, especially for countries where a great proportion of people live under $12 a day. It is dicult to visualize how countries such as Brazil, China, and India could truly emerge as developing economies until their numerous poor citizens nd easy access to essential nancial services, which is critical to them in climbing out of poverty. Micronance, which has emerged and evolved over the past 35 years, is one such mechanism that has attempted to deliver the core nancial services to the poor. This mechanism along with some recent developments in this eld is the focus of this book. It should be made clear that while the provision of low-cost access to nancial services is an important ingredient in alleviating poverty, there are a number of other basic services that are unavailable to the poor. These constraints include (1) absence of primary education, (2) absence of12Micronanceprimary health care, and (3) relatively primitive technologies used by the households. These constraints need to be tackled in parallel as the momentum to deliver nancial services gathers speed. There are signicant complementarities between access to nancial services and the ability of the poor to access education, health care, and better technologies.FORMAL AND INFORMAL MARKETSCapital markets and nancial institutions help provide mechanisms for thrifty households to save and provide access to households and institutions seeking capital for funding their consumption and investment plans. In addition, they oer nancial services such as life insurance, property insurance, and health insurance. The successful development of economies in Europe and the United States is in no small measure due to the services provided by such institutions and markets, which allocate savings from one part of the economy to nance the capital requirements in other parts. Organized stock exchanges, bond markets, dealer markets, commercial banks, insurance companies and other institutions, constitute a vital part of nancial architecture of most economies, whereby households and institutions are able to allocate their savings and have access to capital. These markets and institutions are known as formal markets for nancial services. Formal credit markets consist of institutions such as commercial banks, credit unions, rural banks, and other nancial institutions, which are subject to public and private oversight and regulation. They are governed by bankruptcy codes, investor protection laws, and disclosure requirements. In many developing economies, however, a signicant percentage of the population is simply unable to access these organized nancial institutions and markets due to the fact that they are extremely poor, ill-educated, and have no sustained opportunities for gainful employment. From the perspective of the lenders in the formal credit markets, delivery of small loans or provision of insurance to cover very small individual exposures, or accepting tiny savings deposits is simply not an economically viable proposition. This is due in no small measure to the transactions costs associated with extending such tiny scale of services to the poor, who are numerous. Yet another important reason for lack of access of markets to the poor is the fact that lenders and service providers in the formal markets have very little information about the potential consumers of their services. As a result, such poor households are unable to tap meaningfully into the formal (organized) markets for nancial services or capital markets for nancing their consumption and investment needs.The changing landscape of micronance3The properties of nancial markets where there are signicant dierences between the information possessed by the lender and the information possessed by the borrower have been well documented in a number of inuential papers.2 In particular, the following insights emerge from this strand of research: rst, with asymmetric information, there will be credit rationing, leading to households being shut out of credit markets. Second, contracts will have to be designed so that the lender (who has limited information about the borrowers ability or willingness to pay), can assure him or herself a fair rate of return on loans. In fact, micronance is an area where the concept of group lending or self-help groups (SHGs) is widely practiced: loans are extended to a group of borrowers who are jointly liable for each of the loans extended to the members of the group. Even the formation of the group itself is a matter of considerable importance: since the borrowers have better information about each other, it is recognized that the borrowers should be allowed to form the groups themselves. The presence of joint liability will then imply that high-risk borrowers will not be accepted in the group, and the pool of low-risk groups will then also be conservative in the choice of projects. The loan contract itself may be structured to reect ground realities: loan contracts may tend to have short maturity in order to set strong incentives for the group to service the debt payments, where the income stream is steady. Payments of initial rounds of loans may qualify borrowers to obtain a loan of higher size in the subsequent rounds: this provides an additional incentive to borrowers to acquire good credit history. Interest payments may be collected on a weekly basis, thereby assuring the presence of loan ocers at site levels at that time. In seasonal projects, loan payments may be indexed to reect the revenue patterns. Loans may be bundled with insurance programs to mitigate aggregate risk exposure such as monsoon or droughts. In the past and to a signicant extent even now, poor households have tended to rely extensively on informal markets for their capital needs and other nancial services such as insurance or savings. Indeed, these other nancial services have only recently become available to poor households through micronance. Informal credit markets typically operate outside the perimeter of regulators and are often not subject to monitoring and supervision by governments or agencies of governments. The rights and responsibilities of lenders and borrowers in such markets do not come under a formal bankruptcy code. Players living in close proximity to such markets often have detailed knowledge of each other. Examples of such informal markets and participants might include the following: (1) family members, (2) friends, (3) trade credit from local shops, (4) professional moneylenders in the region, (5) pawnbrokers, (6) local landowners, and so on. These markets4Micronancehave existed for a long time, and indeed have preceded the development of formal banking systems and capital markets. Because they provide a very useful function in poorer sectors of world economies in keeping the savings ow from lenders to borrowers who are unable to access formal credit markets such markets will continue to exist. In this context, it is worth pointing out that even in a well-developed economy such as the United States, pawnbrokers and payday lending institutions exist and serve a clientele of borrowers.3 Without necessarily detracting from the useful functions provided by such informal credit markets, it is also important to examine more carefully the interest rates that prevail in them. Roughly, interest rates in such informal markets provide an upper bound on the interest rates that the borrowers would be willing to pay in micronance markets. While detailed and reliable micro-level data on such markets are usually not available, several studies have documented that the eective borrowing costs in such informal credit markets are rather high. The eective annualized interest rates in payday lending runs well into three digits, often in excess of 200 percent on an annualized basis! For example, interest rates estimated in pawnbroking and local moneylenders run into triple digits, on an annualized basis. In the United States, interest rates charged by pawnbrokers ranged from 36 percent (in New Jersey and Pennsylvania) to 240 percent (in Oklahoma) during 198788.4 In addition, the supply of capital from informal credit markets tends to be rather limited. These observations suggest that a prevalence of high interest rates in informal credit markets where the private sector individuals possessing asymmetric information play the role of lender of last resort to liquidity-constrained households.EVOLUTION OF MICROFINANCEInformal credit markets described in the previous section have preceded micronance. Institutions such as credit unions and specialized lending programs targeted to agricultural sectors have also been in existence since the early 1900s.5 The seeds for micronance in its current form were planted during the period 195080, when small loans were extended to poor borrowers who could not post meaningful collateral. Major organizations, which pioneered this initiative, were ACCION International in Latin America, SEWA Bank in India and the Grameen Bank founded by Muhammad Yunus in Bangladesh. These initiatives demonstrated for the rst time that poor borrowers, especially women, were not only willing to take on smallscale projects funded by loans, but were also capable of chalking up excellent payment records.The changing landscape of micronance5Over the period 198090s many micronance institutions began to develop and found sustaining models of lending to the poor: nongovernmental organizations (NGOs), non-bank nancial institutions (NBFI), rural banks of nationalized banks, and village banks began to develop.6 In the early stages of evolution, micronance was largely restricted to loans and was funded by either governments or aid agencies and was thus based on soft capital. Since the 1990s, micronance has branched out both in terms of the range of nancial and economic services extended, as well as in terms of how capital is raised. Banks began to access this market in a more signicant way than ever before. Financial services ranging from savings deposits, loans, insurance to cover life, health, crop, and properties are currently oered. Many micronance institutions access capital markets either by issuing equity or debt capital in order to raise capital. There are others who have been able to securitize their loans and thus attract capital by issuing micro-credit backed securities. Technological innovations have also paced the evolution of micronance: widespread availability of mobile phones, access to community-level kiosks of computer terminals with access to the Internet, biometric technology to obtain loan approval and credit history, and correspondent banking have dramatically changed the landscape of micronance. The year 2005 was declared as the Year of Micronance, and a number of private sector enterprises and foundations have now dedicated pools of capital for exclusive investments in the area of micronance: in November 2005, Pierre Omidyar (founder of eBay) announced a $100 million micronance fund in partnership with Tufts University for exclusive investments in the micronance initiatives. TIAA-CREF created in September 2006 a $100 million Global Micronance Investment Program (GMIP) to invest in selected micronance institutions worldwide. TIAA-CREF made a $43 million private equity investment in ProCredit Holding AG, a micronance company.7 These are high prole, large-scale investment initiatives. Given the current size and status of micronance institutions it will be interesting to see how these initiatives pan out. As Swanson and Rahman and Mohammed point out respectively in Chapters 2 and 3, concepts such as securitization are rapidly integrating micronance with the capital markets of the developed world, thereby altering the landscape of micronance. As we will demonstrate later in this chapter, traditional institutions such as NGOs still remain an important force, especially for the poorest of micro-borrowers. In the upper tier of the micronance, capital markets are actively helping micronance institutions to tap debt and equity capital. New market institutions have developed, which are likely to improve the transparency and potentially reduce the cost6Micronanceof accessing nancial services: for example, a recent study has reported that the micronance institutions (MFIs) in countries with credit bureaus tend to have a 5 percent lower operating expense ratio than the ones in countries without credit bureaus.8 Credit rating agencies have developed in countries that rate institutions in micronance. Two rating agencies, MicroRate and M-CRIL are well established in the eld.QUESTIONS ADDRESSED IN THIS BOOK AND A WORD ABOUT DATAWhat are the costs of accessing credit in micronance? Does it depend on the nature of contracting? Does the nature of organization of the micronance institution aect the interest rates, and the characteristics of the borrowers? How have the capital markets and nancial intermediaries inuenced the character of micronance? What are the risks and returns to institutions in micronance? What are the implications of technological innovations such as biometrics, the Internet, mobile phones, and so on, on micronance practice? Is there a strong gender bias? What are the positive and adverse consequences of such a bias? These questions form the intellectual basis for this book. To motivate these issues, I would like to examine some of the studies that have used MIX data, in order to shed some light on the issues studied here. Before delving into these questions, it is useful to note the following properties of MIX data. MIX data is based on voluntary reporting by 704 institutions to a detailed survey conducted by MIX. As of 2006, the survey covered more than 52 million borrowers with $23 billion in loans. The survey also covered 56 million depositors with over $32 million in deposits. Clearly, these gures are a downward-biased estimate of the true size of the micronance market. It is also conceivable that the participating institutions are qualitatively dierent from the non-participating institutions, which are likely to be smaller and more donor-nanced. With these caveats, let us examine the questions in turn in the following sub-sections. Cost of Access to Credit Lacking micro-level data, it is dicult to estimate the costs of obtaining nancial services to the borrower in the micronance area with great precision. Still, we have some anecdotal evidence about the levels of interest rates that micro-loan borrowers encounter. In a recent paper, CGAP (Consultative Group to Assist the Poor) reports that the interest yield for Compartamos (a lender in this market) stood at 86.3 percent!9 CGAPThe changing landscape of micronance7estimates that the cost to the micro-loan borrower is about 100 percent when taxes are taken into consideration. This data corresponded to the 2005 period when the median interest rates charged by village banking MFIs stood at 47.2 percent, and the interest rates charged by low-end MFIs stood at 35.4 percent. Gonzalez (2007) provides an analysis of the MIX database, which enables us to shed some light on this issue at a more aggregate level,10 and using this data, we can estimate the interest costs faced by the borrower. Our rst estimate is the yield on gross loan portfolio. This measure takes the ratio of adjusted nancial revenue from the loan portfolio to the adjusted average gross loan portfolio. Financial revenue includes the revenue from the loan portfolio and other assets plus revenue from other nancial services. Such services may include insurance, passbooks, smart cards, and so on. For our purposes these expenses are relevant to the measurement of the costs of nancial services. Gross loan portfolio excludes write-os. In Table 1.1, I provide an estimate of the costs for dierent lending institutions for the period 200305. Note that the costs range from a low of about 20 percent to a high of over 42 percent. The estimated costs are the highest for NGOs, which service the poorest of the borrowers, and lowest for credit unions. Transactions costs associated with delivery of loans, monitoring, collection, and the risk of default are the primary reasons for such high interest rates. Gonzalez (2007) in his analysis also provides another way to look at this measure: across dierent target group of borrowers,11 as shown in Table 1.2. Table 1.2 conrms our nding that the low-end borrowers with a loan balance of less than $150 are the ones who face the highest cost of obtaining nancial services. Their costs range from 35 percent to 38 percent. This raises the question of whether at these rates it is reasonable to think that micronance can play an eective role in alleviating poverty? There are several reasons to harbor such a hope. First, it is safe to assume that the Table 1.1 Costs of obtaining nancial servicesvariations across institutions (in %)Institution Banks Credit unions NBFI NGOs Rural banksSource: Gonzalez (2007).2003 33.7 22.0 32.6 42.5 27.32004 36.9 20.2 32.1 40.2 26.82005 28.3 20.4 29.3 38.6 23.28MicronanceTable 1.2 Costs of obtaining nancial servicesvariations across target groups (in %)Target Group Low end Broad High end Small businessSource: Gonzalez (2007).2003 37.8 34.5 24.1 24.82004 37.8 33.7 24.1 22.72005 35.4 31.1 22.7 22.6alternative sources of credit and other nancial services are even more expensive.12 Since the participation in micronance is purely voluntary, it is reasonable to conclude that for the participants this avenue must be cost-eective. Second, the costs appear to decline for the broad and higher-end borrowers. Presumably these are seasoned borrowers, who by repeated borrowing have established a good credit reputation and honed their entrepreneurial skills. This has in turn dramatically declined the costs of access. Finally, there are reasons to think that the average costs of participating in this market is expected to go down signicantly due to the advent of technology such as biometric screening, smart cards, and the delivery of loans by mobile phones. This said, it is clear that the borrowing costs must come down signicantly in order for micronance to be a credible and sustaining avenue for poverty alleviation. Contracting and cost of access As noted earlier, lenders utilize dierent contracting methods to extend loans. Some borrowers get individual loans and others are part of a solidarity group where the group is jointly liable for the loans taken by the members. Table 1.3 reports the cost of accessing loans during the period 200305 across dierent contracting methods. Table 1.3 shows that the interest rates are highest for village banking groups which consist of poorest borrowers formed into solidarity groups. We note that individual borrowers face borrowing costs ranging from 28 percent to 30 percent, among the lowest. Capital structure and default risk The cost of extending nancial services such as credit and insurance reects the costs of accessing capital for dierent lending institutions and the risksThe changing landscape of micronance9Table 1.3 Costs of obtaining nancial servicesvariations across contracting (in %)Contracting Method Individual Individual/Solidaritya Solidarity Village bankingNote: a. Group lending. Source: Gonzalez (2007).2003 30.1 33.8 35.3 52.02004 30.7 33.4 40.4 49.62005 28.5 30.7 35.9 47.2of the consumers of nancial services. We now turn to these issues. In the analysis of trend lines, Gonzalez (2007see note 10) reports that during the period 200305, the debt/equity ratios of banks varied from 3.6 to 5.6. The corresponding range for credit unions was 3.4 to 4.4. For the rural banks, the debt/equity ratios ranged from 4.6 to 5.2. In contrast, non-bank nancial institutions and NGOs had much smaller debt/equity ratios. The low debt/equity ratios of NGOs (ranging from 0.9 to 1.6) may reect the pervasive use of donor capital by NGOs. High debt/equity ratios of banks, credit unions, and rural banks may reect their inability to get sucient equity capital and the relative ease of access to debt capital. In the next subsection, we provide some evidence concerning the ability of micronance institutions to tap into debt and equity capital. Role of Capital Markets in Financing and Investment in Micronance Chapters 2 and 3 explore the role of capital markets in the eld of micronance. Micronance began largely as a philanthropic eort or a quasi-philanthropic eort. Government-mandated programs such as rural banks, and branch expansion by nationalized banks into rural areas, are examples of such eort. Non-governmental organizations (NGOs), whose supply of capital is largely donor-based, dominated the scene. In fact, an analysis of the MIX database shows that there are over 400 NGOs and over 100 rural banks that extend nancial services as of 2006. Indeed, this evidence demonstrates the very key roles that these groups continue to play in micronance. But we have seen a very active interplay between capital markets and micronance: increasingly, capital markets are being used to source capital for providing micronance services. Indeed, we tend to see a tiered market: some of the top-tier micronance institutions are able to access capital markets for fairly large chunks of capital.10MicronanceThere are other micronance institutions that appear to rely on domestic and local markets for their funding needs. In Chapter 2, Swanson explores this development and the challenges and opportunities that lie ahead. In Chapter 3, Rahman and Mohammed explain in detail the rst micro-credit backed securitization, which tapped capital from dierent nancial institutions to the micronance sector in Bangladesh.13 These developments show that micronance is starting to establish itself as a major sector in the eyes of big nancial institutions. In due course, this will mean that the assets backed by micro-loans may potentially establish themselves as a separate asset class. In addition to the securitization deals that are described by Swanson and Rahman and Mohammed, we have also seen other major developments in the interaction between capital markets and micronance institutions. In 2006, shares of Equity Bank in Kenya were listed on the Nairobi Stock Exchange. This is reputed to be the rst micronance IPO in Africa, signaling the ability of some institutions in micronance to access equity (risk) capital.14 On 20 April 2007, Banco Compartamos, a micronance institution launched in 1990, made the rst initial public oering (IPO).15 This implies the ow of equity capital into micronance from the capital markets. Major international nancial institutions (IFIs) such as EBRD, IFC, and KfW raise signicant amounts of capital and invest in the eld of micronance. These IFIs are private-sector arms of public nancial institutions. A recent report suggests that IFIs have a portfolio of $2.4 billion in micronance, with the average investment size at $4 million. Micronance investment vehicles (MIVs) have been created to channel private sector funds into micronance. The MIVs have a portfolio of $2 billion, with the average investment size of $1 million. This is a topic analyzed by Swanson in Chapter 2. Securitization techniques are used to create debt and equity nancing possibilities for micronance institutions. For example, Citigroup recently created $165 million of collateralized debt obligations (CDOs), which will be initially backed by 30 MFI loan portfolios in 13 countries. Private investors are expected to invest $69 million of senior tranches rated AA by Fitch. IFIs and Citigroup will invest in the subordinated tranches and equity.16 Swanson analyzes the CDO structures and the benets that accrue from such innovative nancing methods. Rahman and Mohammed in Chapter 3 analyze the BRAC securitization deal in sucient detail so that the reader can appreciate the benets of such innovative ways to tap into capital markets. A migration has also started to occur, whereby NGOs are starting to shed their not-for-prot status and become regulated providers of nancial services.17 Stephens (2007Note 17) reports the growth of deposit-takingThe changing landscape of micronance11services and increased commercialization, with the median commercial funding of loan portfolios increasing from a level of 40 percent in 2003 to 60 percent in 2005. This also presents interesting regulatory challenges on issues such as governance, investor protection, disclosure requirements, and so on.18 Technology and MicronanceOpportunities and Challenges In an earlier sub-section, we noted that the borrowing rates in micronance markets can be rather high. In addition, we noted that the delivery of nancial services such as tiny loans and insurance payments, and accepting tiny savings deposits require that the service providers overcome (1) signicant transactions costs, (2) adverse selection problems, and (3) moral hazard issues. In fact, a skeptic could reasonably ask how micronance could be the mechanism for alleviation of poverty, if the interest rates are so high, and if scalability is simply unattainable due to the factors outlined above. Technological developments over the last decade hold much hope in overcoming these obstacles. I will review some of these developments in order to set the stage for Chapter 4, in which Shrivastav explains how mobile phone technology is used in India to potentially deliver loans that are as low as $20. Electronic matching of borrowers and lenders: search and delivery costs One of the factors that contributes to the relatively high interest rates on loans is the cost associated with search and delivery of loans, and the subsequent eorts that are needed to manage the default risk of the loan portfolio. The Internet has opened up new vistas for matching borrowers and lenders electronically, and the developments that we have seen in the last decade have the potential to lower the costs of very small loans. In Table 1.4 we highlight some of the recent innovations that have occurred in the electronic matching of borrowers and lenders.19 There are rms such as Prosper.com in the United States, and Zopa.com in the United Kingdom, which permit peer-to-peer micro-lending and essentially avoiding any intermediaries.20 These options are alternatives for the small borrower to piling up credit-card-type debt. On these websites potential lenders are able to view and evaluate loan requests, ranging from about $1000 upwards, and bid on them. Most of the loans are of short duration, not exceeding three years. Legally binding contracts are entered into. Information about credit scores is used in contracting. Monthly payment schedules are enforced. Unsurprisingly, interest rates on these websites are well above what one sees in the formal credit markets, ranging anywhere around 15 percent to 25 percent or more.12MicronanceTable 1.4 Matching borrowers and lenderssome innovationsOrganization KIVA Services Provided Kiva enters into a partnership with existing micronance institutions worldwide to link lenders and borrowers Zopa partners with credit unions thus, it oers deposits as investments, and lending opportunities Provides an open market place to match lenders and borrowers. Diversication of loans is encouraged Borrowers complete a personal loan request and are screened based on a minimum FICO score, and so on. Interest rate is xed for three years Some Attributes Lenders are able to select their borrowers and their loans are electronically transmitted to the local MFIs Zopa is able to oer the attractive features of credit unions through the Internet to its investors, lenders, and borrowers Has several specialized lending programs targeted at dierent communities Accessible to members in the networkZopaProsper market placeLending ClubSource: Authors compilation.Table 1.4 shows a group of organizations that have exploited the Internet to reduce dramatically the time it takes to match borrowers and lenders, and to arrange the ow of loans between them as well as to promote social lending at a prot.21 This trend may well expand more in to the eld of micronance, although there are barriers such as illiteracy, and inability to access community networks of computers to access the Internet. These barriers may well yield due to other technological advances, which we sketch next. Correspondent banking and biometric authorization of credit22 A development that has accelerated the access of nancial services to the poor is the concept of correspondent banking or branchless banking. This has been especially successful in Brazil, and the concept is bound to have a major eect in other regions.23 Under this concept, local post oces and shops are equipped with barcode-reading point-of-sale (POS) terminals. These local entities then actThe changing landscape of micronance13as agents for banks. It is estimated that currently there are nearly 100 000 such correspondent entities in Brazil alone. CGAP estimates that nearly 13 million customers have been brought into the fold of the banking system through these correspondent networks. It is well recognized that the literacy levels of micronance borrowers are rather low. This presents unique challenges in reducing the costs of lending to them. One technological development that has entered the eld of micronance is the application of ngerprints for the purpose of identifying and validating nancial transactions, through ATM networks. Biometrics and smart cards are already in use in micronance in India and Indonesia. Biometric teller machines (BTMs) reduce the administrative costs of extending small loans in communities where the literacy levels are low. The smart card contains the credit history of the borrower. Banks such as ICICI in India and Danamon in Indonesia are employing such technologies. Village Internet centers and eChoupals One of the developments that has occurred in the delivery of services to poor citizens in remote corners of the world is the proliferation of community Internet centers, which deliver both the infrastructure and information that may be utilized by small agricultural producers to get the best possible price and thus eliminate intermediaries.24 Some see this development as using the technology to empower as well as signicantly increase the range of services to the poorer sections of society. One such innovation is the concept of eChoupals, pioneered by India Tobacco Company (ITC), which deliver price information and an infrastructure for buying the agricultural production from small farmers. eChoupals are village Internet kiosks run by local entrepreneurs who provide pricing information for dierent delivery dates to local farmers and enable the farmers to sell their produce directly to ITC, bypassing the intermediaries. Farmers benet from a known price schedule, and ITC benets from the elimination of commissions and transactions costs that intermediaries would have charged. Mobile phone delivery of nancial services25 It is estimated that there are over 2 billion mobile phone users who not have bank accounts.26 In recent times, many companies have been able to utilize mobile phone technology to provide transfer of cash, make loans, and extend basic nancial services. This promises to bridge the unbanked sector with the market for nancial services at a speed that could not have been imagined a decade ago. In South Africa, it is reported that banks have been able to link a debit card and a bank account to a cell phone. This14Micronanceenables the owners of cell phones to make a deposit at a bank or any post oce, and the deposits are then credited to an account and conrmed via text message. USAID has pioneered mobile-phone-based access to nancial services for the poor in the Philippines. Micronance customers make loan payments by a text messaging system, dramatically lowering the transactions costs and eliminating intermediaries in the process. A concept known as Gcash has been implemented in the Philippines, which eectively allows the users of cell phones to send and receive cash via text messages. As of March 2006, over 1.3 million customers are using the G-cash system, which handles $100 million a day. This avenue promises to cut the transactions costs, time, and eort for both borrowers and lenders. Since the growth of cell phone customer base has been exponential in the last decade, the potential for a steep drop in the cost of delivery of nancial services through mobile phones is promising. This particular technological breakthrough is the topic explored by Anand Shrivastav in Chapter 4. Shrivastav examines the growth of mobile phone technology in India and its potential for delivery for nancial services. He also examines the dierent players in the market. He then describes a technology that he has developed for the delivery of microloans. Regulating Micronance We have traced some of the major changes that have occurred in the landscape of micronance. These changes present regulatory challenges, which form the focus of Chapter 5 in which Rosenberg explores a framework for regulating micronance. One such challenge is the question of how to integrate mobile-phone-based delivery of credit into the overall banking system with appropriate safeguards. In addition, other challenges arise from lenderborrower relationships. We have noted that the costs of accessing credit in informal credit markets in general and in micronance in particular can be in the range of 20 percent to in excess of 50 percent. Our analysis also showed that the poorest of the borrowers are often subject to the highest of the interest rates, and they are often in group lending (selfhelp groupsSHGs), obliging them to expend greater resources in peer monitoring. A question that naturally arises is whether there must be some oversight on the interest rates charged by micronance lending organizations. Since the borrowers are often not well educated, another question is whether they fully understand the eective interest rates that are being charged by the lending institutions. Such eective interest rates will have to reect aThe changing landscape of micronance15number of factors such as: (1) frequency of compounding and payments, (2) transactions costs charged, (3) costs of any mandated insurance policies, and (4) eorts associated with compliancepeer-monitoring eorts, time spent with loan ocers, and so on. Full disclosure of these factors may not always take place. A related issue is the social pressure that may be placed upon the borrowers to enforce timely payments: a borrower in a group-lending contract faces considerable pressure from the community to service the loan payments, as there is (1) joint liability, and (2) he or she lives in the community, and default can qualitatively and adversely aect his or her social life. Absence of credit bureaus and other institutions may also contribute to a situation whereby the same borrower may end up obtaining multiple loans well beyond his or her ability to service the loan payments. This may result in defaults, which may adversely aect otherwise nancially healthy micronance lenders. Defaults may arise due to many reasons. A borrower may default because he or she is unable to service the loan contract due to insucient income. Such poor realizations of income may be the result of factors outside the control of the borrower (such as drought, monsoon, and so on) or owing to poor incentives or insucient eort. Defaults may also be strategic and may be coordinated by the entire borrowing group. The risk of default is managed by the lenders through various methods: (1) forming groups that are less risky, (2) contracting with built-in incentives for peer monitoring, (3) frequent (often weekly) payments where feasible, (4) extensive lender monitoring, and so on. Some recent developments, related in the next subsection, point to some underlying problems that call for prudent oversight of micronance practice. Defaults in 2006 on micro-loans in Andhra Pradesh27 The state government of Andhra Pradesh (in India) shut down 50 branches of two major MFIs in the Krishna District in March 2006. This extreme action was precipitated when many micro-loan borrowers complained to the state government about the usurious interest rates and forced loan recovery practices. There was an allegation that ten borrowers of MFIs in the Krishna District committed suicide because of their inability to repay the loans taken from the MFIs. MFIs operating in Andhra Pradesh reached an agreement with the state government on MFI interest rates, product portfolio, inter-MFI competition, credit disbursement, and loan recovery methodologies. As per the terms of the agreement, MFIs have agreed to an interest rate ceiling of 15 percent. They have agreed to desist from providing multiple credit to an existing borrower and recovery of loans at a pace compatible with the borrowers income level.16MicronanceMFIs are also to remain strictly within the micro-credit domain, avoiding micro-insurance products.28 This episode raises several important regulatory questions: should the government set a ceiling on interest rates charged by micronance lenders? How often should such ceilings be reviewed and reset to reect credit market conditions? Should the government set standards and enforce such standards on acceptable loan recovery practices? This episode is a watershed in focusing attention on the need to have appropriate institutions and lending and loan recovery standards in place in order to promote private capital ow into underserved communities.29 Savings and intermediation Many micronance institutions also accept savings deposits and provide other services such as insurance. The direct and indirect costs associated with the provision of these services should be transparent to the consumers of these services.30 Perhaps more importantly, this raises the question of deposit insurance. Should there be a deposit insurance program? What will happen to the deposits if a micronance institution were to default? To get a sense of the magnitude of this problem, let us take a look at the question of how pervasive deposits in micronance are. Gonzalez (2007see note 8) reports that the ratio of deposits to loans ranges from 30 percent to 46 percent for individual loans. For solidarity groups and village banking, deposits are not reported, and may be presumed to be negligible. Deposits are a big part of banks, credit unions, and rural banks. Obviously, they are not an issue for non-bank nancial institutions and NGOs. How should the deposits of the poor be protected? Technological innovations Mobile phone delivery of nancial services, community Internet portals that enable farmers to sell their crops and so avoid intermediaries, and other such innovations also raise important regulatory questions. Should mobile phone deliverers of loans be given a banking license? We noted that in correspondent banking, third parties interact with customers to deliver banking services. Some of the issues that arise from a regulatory standpoint include the following: 1. Should third parties (correspondents, for example) be permitted to execute banking transactions and interact directly with customers? In this context, how should electronic money transfers and mobile phone transfer of cash be treated? After 9/11, the regulatory practices in organized credit/banking markets have brought into play risk-based anti-money laundering2.The changing landscape of micronance17(AML) rules, as well as rules for combating the nancing of terrorism (CFT) adapted to the realities of remote transactions conducted through agents.31 How should they be adapted in micronance? Finally, issues of consumer protection also have to be addressed. Regulation is costly from the perspective of both formulation and enforcement from the regulators point of view. If poorly framed, it can also be costly to comply with from the standpoint of the lending institutions in micronance. Rosenberg addresses this very important issue in Chapter 5 and articulates how we must evaluate the challenges in this regard. Gender Empowerment and its Characterization One of the striking facts about the eld of micronance is that an overwhelming number of the borrowers are women. In Chapter 6, Armendriz and Roome examine this strategy of targeting women or the so-called issue of gender empowerment in micronance and note the salient fact that indeed, in the aggregate, seven out of ten micronance clients are women. This predominant empowerment in lending has been examined in the literature, and many benecial eects that arise from such an empowerment have been documented. More recent evidence, though anecdotal, points to some potential dysfunctional consequences of such an empowerment. In their chapter, Armendriz and Roome observe that we have no reliable empirical evidence to examine the results of this gender strategy on the extent and quality of economic and social development, and argue for future research to examine this issue in greater detail. They also explore the potential benets of women bringing their male partners into the fold of micronance on a voluntary basis. Such eorts may reduce domestic friction, reduce default, and potentially increase overall welfare. These and other important issues raised by Armendriz and Roome warrant additional research. Rather than summarizing the potentially important issues that Armendriz and Roome raise in Chapter 6, it is useful to characterize the nature of gender empowerment in greater detail: Is this empowerment related to the nature of the lending institution? If so, what are the potential causal factors? Do we tend to see greater gender empowerment when the average pool of borrowers is poorer? Is it the case that poorer women borrowers tend to gravitate to certain types of lending institutions? Are there geographical variations in gender empowerment? What are the cultural factors that might explain such geographical variations?18MicronanceTable 1.5 Gender empowermentpercentage of women borrowers across lending institutions and regionsRegion Banks Cooperatives and Credit Unions 39 N/A 52 99 Non-bank Financial Institutions 62 86 52 65 NGOsAfrica East Asia Latin America South Asia52 61 52 9781 98 71 100Source: MIX data of 2006 and authors calculations.Evidence on gender empowerment Establishing such stylized facts may enable us to further expand on the questions that are analyzed by Armendriz and Roome. To this end, we analyze the MIX data of 2006 to shed some light on these questions.32 Table 1.5 documents the distribution of the fraction of women borrowers in the sample, split across geographical regions and the types of lending institutions.33 While this is admittedly a very aggregated picture, which may potentially mask country-specic sociocultural dimensions, it still provides an interesting breakdown. The data clearly show that the gender empowerment is extremely strong in South Asia, and to a lesser extent in East Asia. It is much less so in Africa and Latin America. Indeed, for micro-loans extended by banks, cooperatives, and credit unions, and non-bank nancial institutions, we nd that the borrowers are very nearly evenly split across the gender in Latin America, and roughly so in Africa. Sorting the results by the type of lending institution leads to a very interesting stylized fact: NGO borrowers are predominantly women across all regions. We will see later, unsurprisingly, is that NGO borrowers are typically the poorest of micro-borrowers and this is where we nd the greatest of gender empowerment. Gonzalez (2007see note 8) in his trend line analysis has also documented a very similar picture for the period 200305, which is summarized on a more aggregated level in Table 1.6. Micronance borrowers are targeted for credit either on an individual (stand-alone) basis or as a part of a solidarity group (group-lending) basis. The benets of a solidarity group and the associated peer monitoring eects have been addressed in the literature.34 In addition, there is the concept of village banking, which is another self-help group (SHG) consisting of very poor borrowers, with low per capita income. SHG-basedThe changing landscape of micronance19Table 1.6 Percentage of women borrowers across dierent institutions 200305Year 2003 2004 2005 Banks 52.8 50.0 52.5 Credit Unions 73.0 56.5 60.0 NBFI 53.8 60.8 56.1 NGO 79.0 82.1 79.7Source: Gonzalez (2007).Table 1.7 Percentage of women borrowers across dierent contracting arrangements 200305Year 2003 2004 2005 Individual 47.9 53.9 51.8 Individual/Solidarity 67.9 66.2 62.0 Solidarity 82.0 92.0 100.0 Village Banking 90.3 94.5 90.2Source: Gonzalez (2007).lending has the benecial impact of peer monitoring and assortative matching, which leads to lower default risk and hence should result in lower borrowing rates. What we see in the data as documented in this chapter, however, is that the SHG borrowers are among the poorest, and this causes the interest rates to actually go up. Table 1.7 clearly shows the concentration of women borrowers in the SHG category and much less so in individual loan programs. Median loan size Table 1.8 shows that the median loan size of the NGO borrowers is, by an order of magnitude, smaller in each region when compared with other borrowers. South Asian borrowers are all uniformly poor, with a median loan balance ranging from $100130 across dierent lending institutions. In light of this it is interesting that the gender empowerment is so dierent for non-bank nancial institutions in South Asia. The median borrower in Latin America has a loan balance, which is four to ten times the median loan balance of borrowers in South Asia: this appears to suggest that the gender empowerment declines with increases in average loan balance, which may be an instrumental variable for seasoned (proven) borrowers, lower default risk, and greater entrepreneurial skills. These assertions remain to be tested.20MicronanceTable 1.8 Median loan size in US$ across lending institutions and regionsRegion Banks Cooperatives and Credit Unions 373 N/A 1510 124 Non-bank Financial Institutions 215 235 1039 126 NGOsAfrica East Asia Latin America South Asia529 562 1445 127129 90 437 107Source: MIX data (2006) and authors analysis.Median total assets Table 1.9 is provided to give the reader a sense of the relative dominance of dierent lending institutions in the micronance space. A caveat is in order in this context: only four banks reported from South Asia, and Grameen Bank, which operates out of Bangladesh, reported assets worth $819 830 340 and is by far the biggest player. But this is an exception and not the rule, as the median total assets of banks in South Asia is only $33 709 260. In general, there are fewer reporting banks in the market, but they dominate the assets (loans, for the most part) in the market. In South Asia, the median assets of banks are nearly ve times the median assets of NGOs. In Latin America the corresponding multiple is 46, and in East Asia, the corresponding multiple is 550! These numbers point to the increasingly dominant role played by banks in extending nancial services to poor borrowers. As we noted in the section on Technology and Micronance, with the advent of Internet and mobile phones, we will begin to see an increasing presence of banks and new players in this market. This has far-reaching regulatory implications: the median loan sizes are far higher for banks, and the gender empowerment is far lower for banks, with the exception of South Asia. The issue of gender empowerment in micronance requires a careful study at two levels: rst, at a suitable level of aggregation (perhaps at a country or a regional level) in order to better understand how it comes about and what the evidence is from a development perspective. Several interesting questions arise at such a broad level of aggregation: perhaps certain types of lending institutions prefer women borrowers. Or, it could be the case that women borrowers gravitate towards certain lending institutions. Are there dierences at the level of countries? If so, what are the determinants of such empowerment? For example, with the exception of Bangladesh, in countries with a majority Islamic population, we tend to seeThe changing landscape of micronance21Table 1.9 Median total assets in US$ across lending institutions and regionsRegion Banks Cooperatives and Credit Unions 1 915 690 N/A 11 394 423 1 143 689 Non-bank Financial Institutions 3 006 357 6 237 324 29 777 549 8 516 418 NGOsAfrica East Asia Latin America South Asia19 980 000 742 857 915 166 225 000 33 709 2601 809 247 1 350 415 3 560 909 6 098 412Source: MIX data (2006) and authors analysis.fewer women borrowers in the MIX database, but NGOs seem to have succeeded in attracting women borrowers in these countries.35 At a second level, we need to address this issue at a very micro level: do we see group-based lending and greater gender empowerment going hand in hand? Is there a signicant positive association between rst-time borrowers (who are likely to be among the poorest) and gender empowerment? I nd the issues studied in the chapter by Armendriz and Roome to be very important in many respects. First, we want to document clearly the developmental and social benets arising from targeting women in micronance to guide future eorts. Second, we need to better understand why this empowerment does not appear to be present with some lending institutions in some countries, at least in the context of the evidence presented from the MIX database. The potential benets of voluntary introduction of male partners by women into a micronance program as a way to alleviate the number of important issues highlighted by Armendriz and Roome (in Chapter 6) seems to be well worthy of more detailed investigation in dierent regions of the world.CONCLUSIONThe landscape of micronance has changed dramatically, especially at the upper end, where capital markets are rapidly becoming integrated with the nancing and investment needs of the micronance markets. This rapid growth and inows of comparatively large amounts of capital presents its own set of challenges. On the other hand, at the lower end, NGOs remain the mainstay for the poorest of borrowers, where the loan sizes are very small and the interest rates remain high. Technological innovations may22Micronancewell hold the key to reducing the costs of delivering small loans and accepting very small savings deposits. They may in turn pave the way to making the micronance approach more scalable. The regulatory challenges associated with the rapid change in the landscape deserve greater attention from policy-makers and researchers. An issue that the book does not consider but is perhaps extremely important is the fact that the poor lack very basic services that the rest of society takes for granted. Such services include: (1) access to primary education, (2) access to primary health care, and (3) access to and training in operating more modern tools and technology in day-to-day activities. Access to these is as important if not more so than access to nancial services. On the other hand, it is clear that the access to nancial services will enhance the ability of poor households to access these important basic services and skills as were. A number of practitioners have already recognized the need to deliver in parallel both these basic services and the nancial services.NOTES1. 2. World Bank, PovertyNet website: http://go.worldbank.org/33CTPSVDC0. Akerlof, G.A. (1970), The market for lemons: quality uncertainty and the market mechanism, The Quarterly Journal of Economics, 84 (3), 488500; Stiglitz, J. and A. Weiss (1981), Credit rationing in markets with imperfect information, American Economic Review, 71 (3), 35166. Flannery, M. and K. Samolyk (2005), Payday lending: do the costs justify the price?, FDIC Center for Financial Research working paper, Chicago University, Chicago. Caskey, John P. (1991), Pawnbroking in America: the economics of a forgotten credit market, Journal of Money, Credit and Banking, 23 (1), 8599. See Global Envision (14 April 2006), The history of micronance, www.globalen vision.org/library/4/1051/. Many nationalized banks in India extended credit to priority sectors such as agriculture. In Indonesia, Bank Rakyat Indonesia (BRI) extended micro-savings and credit products. The motivation for this initiative underscores the view of many nancial institutions: We subscribe to the view that micronance investing can contribute to a double bottom line, said Ed Grzybowski, TIAA-CREFs Chief Investment Ocer, GMIP, and this investment in ProCredit, gives us an opportunity to seek competitive returns through socially responsible investments that we believe have a low correlation to traditional equity and xed income markets, quoted in TIAA-CREF launches $100m micronance program, in SustainableBusiness.com (2006), accessed 18 October at www.sustainablebusiness.com./index/cfm/go/news.feature/id/1384. Gonzalez, A. (2007), Eciency drivers of micronance institutions: the case of operating costs, MicroBanking Bulletin, 15 (Autumn). CGAP (2007), CGAP reections on the Compartamos IPO: a case study on micronance interest rates and prots, Focus Note, 42 (June). Gonzalez, A. (2007), Resilience of micronance to national macroeconomic events: a look at MFI asset quality, MicroBanking Bulletin, 14 (Spring). The criteria used for the target group classication and the participating institutions are detailed in Gonzalez (2007) (see Note 10). Median values are reported in the tables.3. 4. 5. 6. 7.8. 9. 10. 11.The changing landscape of micronance12.2313. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.28. 29.30.31. 32.Local moneylenders, who operate with limited capital, provide the outside borrowing options for the poor, besides friends, relatives, and trade credit from local shops. Anecdotal and scattered evidence suggests that the interest rates charged by the moneylenders are signicantly higher than the rates under the micronance alternative. See also Zaman, S. and S.N. Kairy (2007), Building domestic capital markets: BRACs AAA securitization, MicroBanking Bulletin, 14 (Spring). Reddy, R. (2007), Micronance cracking the capital markets II, Insight, 22 (May). See Note 9. CGAP (2008) Micronance capital markets update, 23 (January), www.cgap.org/ mcm/archives/V23_0108.html. Stephens, B. (2007), Commercialization continues apace, MicroBanking Bulletin, 14 (Spring). Hishigsuren, G. (2006), Transformation of micronance operations from NGO to regulated MFI, IDEAS, www.micronancegateway.org/content/article/detail/36733. Pearlstine, J. Boon (2006), Lenders, borrowers hook up over the web: Prosper.com and other sites provide forum for individual bidders willing to oer small loans, Wall Street Journal, 20 May; also see Credit Union Magazine, January 2008. Zopa now has operations in the United States. Current websites of these organizations may be found at www.kiva.org/; www.prosper.com; www.lendingclub.com/home.action; https://us.zopa.com/. Milbrandt, J. (2008), Biometrics and smart cards serve as successful micronance innovations in Asia, Micronance Report, 9 January. www.cgap.org/policy/branchlessbanking. Kumar, R. (2004), eChoupals: a study on the nancial sustainability of village Internet centers in rural Madhya Pradesh, Information Technology and International Development, 2 (Fall), 4573. Milbrandt, J. (2008), The rise of mobile phone banking, Micronance Report, 17 January; Chemonics (2006), Mobile-phone banking expands into rural Philippines, 24 May, www.chemonics.com. Global Envision (2007), Micronance goes mobile: cell phone banking revolutionizes nancial services for the poor, 3 August, www.globalenvision.org/library/4/1708. Shylendra, H.S. (2006), Micronance institutions in Andhra Pradesh: crisis and diagnosis, Economic and Political Weekly, 20 May; Microcapital (2006), Micronance institutions reach crucial agreement with government in Andhra Pradesh, India, 11 October, www.apmas.org/pdf%5Cn.pdf; Microcapital (2006), Indian Bank ICICI reaches arrangement with provincial government on micro-loan interest rates, www.microcapital.org/?p=580. Shylendra, H.S. (2006) as Note 27. This episode also called attention to the potential for increased default risk, and the political-economic nature of micronance. The Andhra Pradesh state government had taken a tough stance following allegations that usurious interest rates and heavy-handed loan recovery procedures contributed to farmer suicides. Four lending institutions Spandana, Asmita, Umduma Poddu Pedatha, and SHARE Micron came in for extensive scrutiny. In addition, ICICI Bank, which writes most of these loans using these MFIs as disbursement and collection agents, faced potential write-os worth $100 million. Some micronance institutions require savings as a condition for lending, and impose costs for providing such services. Products such as savings, insurance, and loans are bundled in a way that the overall costs to the consumer can become prohibitive. See Shylendra (2006), note 27. Branchless banking: rapid growth poses regulatory challenges, CGAP Focus Note, accessed 31 January, 2008 at www.cgap.org/p/site/c/template.rc/1.26.2154. The analysis is based on MIX database, which is in the public domain. There is no entry in Table 1.5 for credit unions and cooperatives for East Asia as there was only one such institution there during the sample period considered. MIX data only included institutions that voluntarily report their nancial and outreach data, and hence is necessarily a biased and under-represented sample of the population of the lending institutions and24Micronanceborrowers in reality. We nevertheless believe that the MIX data can provide very useful stylized facts about the issues studied in this book. The sample size of reporting institutions is as follows: the data have 37 reporting banks, 124 cooperatives and credit unions, 158 non-bank nancial institutions, and 295 NGOs. The data are not necessarily contemporaneous and may have a time lag of up to one year. Stiglitz, J.E. (1990), Peer monitoring in credit markets, World Bank Economic Review, 4 (3), 35166. In Pakistan, of the six non-bank nancial institutions reporting, the percentage of women borrowers ranged from 4.1 percent to 34 percent. On the other hand, two NGOs from Pakistan report a near 100 percent empowerment in favor of women! In Afghanistan, of the three non-bank nancial institutions reporting, the percentage of women borrowers ranged from 45 percent to 65 percent. One NGO reports 100 percent women borrowers and the other two report 25 percent and 40 percent participation by women.33. 34. 35.2. The role of international capital markets in micronanceBrad SwansonINTRODUCTIONIn 2004, international capital markets awoke to the attractiveness of investing in micronance. Since then, debt and equity security issues for micronance have raised an estimated US$1 billion from private sector nancial institutions seeking commercial returns.1 The deals have taken forms that are familiar in developed markets such as initial public oerings, bond issues, collateralized debt obligations (CDOs), and securitizations of the underlying micro-loans. In addition, private sector debt and equity micronance funds have sprung upfor investors who prefer to give discretion to professional managersand are now thought to control more than $2 billion, of which more than $300 million is mainstream commercial investment.2 Overall, cross-border investment in micronance surged to $1.4 billion in 2006, triple the rate only two years previously.3 While traditional suppliers of micronance capitalnon-prot organizations, governmental development agencies, and individualsare contributing to this surge,4 the novelty since 2004 is the participation by private sector institutional investors seeking full market returns. These mainstream commercial investors, most located in Western Europe and the USA,5 are driving the opening of capital markets to micronance. How and why commercial mainstream investors have come into micronance and the likely evolution of capital markets funding for micronance is the topic of this chapter.6THE NEED FOR CAPITAL MARKETS FUNDING IN MICROFINANCEWhen properly conducted, micronance is a protable, low-risk, and expanding nancial activity. For example, from January to June 2007, the2526Micronance26 widely dispersed micronance institutions (MFIs) in Micronance Securities XXEB, a $60 million CDO sponsored by Developing World Markets (DWM), had an aggregate annualized return on equity of more than 25 percent and were growing their loan portfolios by more than 50 percent on an annual basis, while their PAR-30 (total amount of portfolio at risk, or loans with payment delays, beyond 30 days) was only 2.9 percent.7 This is a performance that any commercial bank would be proud to announce. Already, the number of borrowers served by MFIs is globally estimated at 100 million.8 With an average loan size of $170, the total market size is estimated at $17 billion. Yet the potential demand is 15 times the current marketestimated at 1.5 billion, or half the 3 billion global working poor. Thus, micronance represents a total commercial market of more than $250 billion. Currently more than three-quarters of the $17 billion funding total is raised from domestic markets. However, this number is skewed by the amountalmost $8 billioncoming from deposits in the few countries where MFIs are allowed to take deposits. Most of the estimated 10 000 existing MFIs are not deposit-taking institutionsand are unlikely to become so, given the cost and complexity of complying with regulations typically applied to institutions taking deposits from the public. Future funding for MFIs is thus unlikely to be sourced mainly from deposits. Domestic emerging country commercial banks, which should be major funding sources for MFIs, are typically averse to lending to them (see the section on Local Currency below). Moreover, capital markets in most developing countries are thin and the major institutional players are averse to or legally constrained from signicant investment in micronance. For these reasons, it is unlikely that domestic sources in emerging countries will generate more than a fraction of the more than $200 billion that will need to be raised to satisfy potential demand. Moreover, while non-commercial investors account for 80 percent of the $4 billion in funding now sourced internationally, this is a legacy of the origin of micronance in charitable and ocially sponsored development activity. As MFIs appetite for capital grows exponentially, it is unlikely that government agencies and non-prot organizations will increase their ow of funding proportionately: rst, they will be faced with competing demands for assistance; and, second, they will begin to question whether their mission is best served by funding nancial enterprises that are protable and are increasingly transforming into privately owned companies able to attract commercial investment. (However, this realization may not have begun to sink in yetsee discussion of role reversal below in the section on The Contribution of Non-commercial Investors.)The role of international capital markets27The only available source of funding for commercial lending of this magnitude is the international capital markets. Already, micronance investment vehicles, which typically include private sector institutional investors, are growing their investment portfolios at 233 percent per year, while ocial development agencies are lagging at 150 percent.9 For the international capital markets, funding a $200 billion industry is routine.FROM FUND TO CDOThe rst10 micronance fund to reach beyond socially responsible investors was established in 1998. The Dexia Microcredit Fund, sponsored by Dexia, a Franco-Belgian bank, and advised by BlueOrchard Finance SA, based in Geneva, oers investors a return above their cost of funds (typically 12 percent over a benchmark rate11) and an ability to redeem their investments. In November 2007, funds under management were $233 million.12 As a fund (a Luxembourg-based SICAV13) oering redemption rights to investors, Dexia needs to keep its maturities to MFIs relatively short and a large portion of its assets in cash (typically 20 percent or more). This limits returns to investors and the attractiveness of Dexias funding to MFIs, many of which need longer-term maturities on a portion of their liabilities to better manage risk. In 2004, after six years of operations and with $45 million under management, BlueOrchard wanted to provide longer-term funding to MFIs and more attractive rates to investors. It partnered with DWM, an emerging markets fund manager and advisor based in Connecticut, to create the rst CDO in the micronance industry. In this transaction, loans were made to MFIs for seven years from the proceeds of issuing xed rate bonds. As the bond investors were not entitled to their principal until the bonds maturity, there was no need to keep large quantities of cash on hand to deal with redemptions. Furthermore, MFIs had use of the funds for the full period with no interest rate uncertainty. The CDO was named BlueOrchard Micronance Securities I (BOMSI). The rst closing of $40 million occurred in July 2004 and a subsequent closing of $47 million was held in April 2005. This transaction looked very dierent from any existing micronance investment vehicle and it marked the beginning of mainstream capital markets investment in micronance. The major innovations in micronance funding pioneered by BOMSI include the following: 1. BOMSI is not a fundinvestment decisions are not handed o to a professional manager. There is no asset substitution or active28Micronance2.3.management. Investors in BOMSI have a single source of repayment, a static pool of 14 loans to MFIs taken on at closing. When investors came into BOMSI, they did so on the basis of their own assessment of the credit risk of the underlying MFIsand they have to live with this decision for seven years. Legally, BOMSI is a special-purpose (legal) vehicle (SPV)a limited liability corporationregistered in the business-friendly state of Delaware. The vehicle is limited by its constitutional documents solely to servicing its loans to MFIs and repaying its creditors. Cashows from debtors to creditors pass transparently through the vehicle. When the loans pay o and the liabilities mature, BOMSI will make its nal payments to investors and be liquidated (see Figure 2.1). BOMSIs funding is stratied in ve levels of risksenior, three classes of subordinated, and, at the bottom, equity. (Both BlueOrchard and DWM are equity investors in BOMSI.) The cashow from BOMSIs loans to MFIs is applied according to a strict order of precedence, known in structured nance as the cash waterfall. Senior investors are paid completely rst, then the other classes in order of precedence. Equity investors do not get a current return on their investment but if, after all MFI loans have reached maturity and all other investors have been repaid, there is residual cash left in the BOMSI SPV it will be allocated to the equity investors. BOMSIs investors do not hold units in a fund and have not made loans to BOMSI. Rather, they have purchased securitiesbonds and equity interests. As we will see later, this distinction was important in attracting institutional investment.InvestorsDebt serviceBOMSISpecialpurpose legal vehicle1st priority 2nd prioritySenior Subordinated (three classes A,B,C) EquityMFIs3rd priorityTrusteeServicerAdvisorService providersFigure 2.1 Cashows from loan repaymentsThe role of international capital markets29These elements are common to CDOs and other forms of securitization in more developed asset classes such as mortgages, corporate loans, auto loans, or student loans. But these are asset classes with substantial data going back a number of years describing default performance under a number of economic scenarios. In the micronance industry, by contrast, MFI write-o policies vary widely and data on micro-loan defaults typically are not recorded consistently by dierent MFIs. Moreover, these data typically are neither independently audited nor rigorously modeled to determine likely performance under varying circumstances. (Although recently, a non-prot research rm, Center for the Development of Social Finance, did a static pool analysis of more than 600 000 micro-loans from two MFIsSKS in India and IMON in Tajikistanusing developed world methodology, in order to demonstrate that at least some MFIs are rigorous enough in their record-keeping to permit this style of analysis.)14 Moreover, BOMSI securitized loans to only 14 institutions in nine countriesmuch less diversication than typical CDOs or other securitization transactions in developed markets, where the asset pool may comprise many hundreds or thousands of loans. Given these factors, implementing a CDO for the micronance industry required changing the way investors viewed both micronance and the CDO product.INTRODUCING COMMERCIAL INVESTORS TO THE MICROFINANCE CDODespite the relative paucity of data and diversication, DWM, which took primary responsibility for structuring the transaction, encouraged investors to compare BOMSI to mainstream commercial investments. DWM held the view that to attract sucient investor interest, BOMSI had to reach beyond the circle of funders primarily motivated by social, not nancial, returns. To distinguish BOMSI as a commercial investmentdierent from investment funds, donations to NGOs, or other means then available to support micronanceDWM highlighted the following:Low default rate in MFI loan portfolios. All participating MFIs reported default rates below 1 percent. Although reporting systems were not consistent or their results independently veried, the professionalism and the track record of the MFIs themselves added credibility to their ndings. Favorable risk/return ratios. The tiered capital structure enabled BOMSI to oer high returns to the higher-risk tranche investors,30Micronancewhile providing the lower-risk investors with a substantial degree of collateralization, enabling them to feel satised with a low credit spread over the benchmark Treasury bond because their notes had the highest priority of repayment. Investors were not asked to discount their return expectations in view of the presumed social value of micronance. With a variety of securities oering dierent risk and return parameters, DWM was able to segment the international investor base and thus appeal to a wide spectrum of potential investors. Familiar investment instruments. BOMSI debt investors purchased bonds drafted in their language, and carrying features common to commercial bonds. They beneted from the appointment of a trustee to safeguard their interests, as is the case in most bond issues. The bonds are transferable and each series is endowed with a unique CUSIP15 number that facilitates record-keeping, valuation, and permitted transfers. (However, the bonds were privately placed, are not listed, and are not intended to be actively traded.) These features helped to ensure that investors had a high comfort level with the form of the investment and could focus clearly on the underlying risk and return.In one important respect BOMSI was dierently structured from other commercial transactions: Overseas Private Investors Corporation (OPIC), a US government development agency, purchased the most senior tranche of securities. Note that OPICs ownership of the senior tranche conveyed no protection to more junior investorsby virtue of the cash waterfall, they were exposed to risk in the MFI loan portfolio ahead of OPIC. However, the participation by a large and well-respected development agencyoften referred to as the halo eectencouraged investors who otherwise might have been unwilling to consider the transaction.GROWING PARTICIPATION BY COMMERCIAL INVESTORSIn the event, the rst closing of BOMSI attracted only $1.5 million, or 4 percent of the capital raised, from private sector investors seeking a full market return (see Table 2.1 below). However, by the time of the second closing, nine months later in April 2006, interest in the transaction had spread and commercially motivated institutional investors accounted for 41 percent of the amount invested. Moreover, the commercial investment came from a wider spread of investor types.The role of international capital markets31Table 2.1 Commercially motivated CDO investors by type and risk categoryInvestor type Bank Money manager Insurance company Pension fund University endowment Total % of total investment Risk category Equity Juniora Mezzanineb Senior Total BOMS 1 USD 500 000 1 000 000 BOMS 2 USD 500 000 500 000 18 000 000 100 000 19 100 000 41% MFS USD 9 139 640 3 036 000 1 000 000 20 500 000 33 675 640 56% 1 500 000 125 320 1 125 320 30 925 000 33 675 640 Total USD 9 639 640 4 536 000 1 500 000 38 500 000 100 000 54 275 640 37% 1 500 000 1 225 320 20 625 320 30 925 000 54 275 6401 500 000 4%500 000 1 000 000 1 500 000600 000 18 500 000 19 100 000Notes: a. For BOMS 1 and 2, subordinated notes C and B. b. For BOMS 1 and 2, subordinated notes A. Source: DWM.A little over a year later, in June 2006, DWM closed its third CDO transaction, Micronance Securities XXEB (MFS), for which it was sole sponsor. This $60 million securitization of loans to 26 MFIs had more investment primarily commercially motivated than primarily socially motivated. Moreover, for the rst time commercial investors (besides the sponsor) purchased equity. By this time, not only had market familiarity with micronance grown, but DWM had also obtained an investment grade ratingAon the MFS senior notes from MicroRate, a specialized micronance rating agency. This heightened commercial investors comfort with the senior tranche. In addition, DWM had sponsored a study indicating that micronance is less correlated to economic downturn than other emerging markets assets, making portfolios including micronance, in theory, less volatile (see the section below, On the Path to an Asset Class). This development was of interest to commercially motivated investors. Table 2.1 shows the amount of investment in three CDO transactions contributed by institutional investors seeking full market returns, with socially positive impact a desirable additional benet. The remainder of the32Micronanceinvestment came from investors whose primary motivation was social thus, for these, nancial return was of secondary importance. High net worth individuals (HNWIs) constituted 10 percent of the investment amount in the rst BOMSI close. (They are not shown in Table 2.1 as we do not characterize them as commercial investors.) This percentage fell in the second close and by the closing of MFS, HNWIs as a group were down to under 5 percent of the total capital invested. While there is doubtless a signicant potential market among HNWIs, and among retail investors generally, for micronance risk, nancial institutions are the bellwethers as they have greater sophistication, more resources, and stronger tolerance for volatility and illiquidity. The market for micronance CDOs has continued to grow. Notably, BlueOrchard has sponsored two more CDOs, with Morgan Stanley as placement agent, in April 2006 and May 2007, raising $99 million and $110 million respectively. The entrance of Morgan Stanley, a bulge bracket investment bank, is another signal that micronance funding is gaining credibility as a capital markets activity. However, in the latter half of 2007, and continuing into 2008, the failure of a number of CDOs based on sub-prime mortgages in the USA made the very term CDO suspect in the eyes of many institutional investors and slowed the pace of growth in micronance CDOs, even though the two types of assets are unrelated. As MFIs in micronance CDOs have continued to perform well, capital markets intermediaries believe that receptivity among investors to this asset class will improve with increasing recognition of the inherent robustness of micronance credit risk. The relatively small volumes outstanding to date and the legal restrictions on trading privately placed securities mean that secondary markets have not developed. Secondary markets should not be anticipated until the number of participants and amounts outstanding increase signicantly, including several listed issues to act as price indicators for the markets.CDOs VS. FUNDSCDOs were the rst non-fund capital markets products in micronance for several reasons:MFIs typically have balance sheets that are too small to justify transactions of the scale required to access international capital markets aggregating MFI loans is necessary. On the other hand, MFIs are used to borrowing internationallycreating loans to international standards and packaging them into theThe role of international capital markets33asset side of a special-purpose nancing vehicle does not present insuperable challenges. Capital markets investors predominantly demand instruments denominated in USD or euros (although the decline of the dollar and relative stability of many emerging market currencies in recent years are persuading investors to become more open to local currency risk). On the other hand, MFIs typically (but not always) lend in local currency; therefore, they are used to borrowing in hard currency and passing the risk on to their clients, whose demand for loans is relatively interest rate inelastic. Top-quality MFIs are found throughout the emerging markets so geographic diversication can be achieved. MFIs typically have very few loan products, so their risk on the asset side is relatively easy to analyze and can serve as a proxy for the underlying risk of the micro-borrowerhighly diversied, highly granularthat strongly draws the typical capital markets investor.Despite these factors, the relative scarcity of top-quality MFIs may act to brake the growth of this asset category. Of an estimated 10 000 MFIs worldwide, fewer than 100 have qualied for inclusion in a CDO to date. As market demand for CDOs grows, CDO arrangers will need to push farther down the pyramid to tap MFIs of lesser size and credit quality to generate assets. But, given the absence of data in the micronance industry, as noted above, the analysis of risk in CDOs is not a function of statistics but rather of individual assessment of MFIs. Investors nd it dicult to make the time necessary to take individual credit decisions on numerous MFIs, especially given that the investment represents only a very small part of the investors portfolio responsibility. Up to now, the presence in CDOs of MFIs that are mostly top-rankeddemonstrated either through ratings or performance over timehas served to ease these credit decisions. But with the top tier of MFIs growing overbanked (see the section below, Is Micronance Riding for a Fall?), CDO arrangers will need to persuade investors to take risks on MFIs that are less known or appear nancially weaker. Part of this persuasion may come through education some smaller MFIs may be as creditworthy as their larger peersbut structural features such as credit guarantees or higher collateralization levels may become necessary in some deals to assuage investor concern. While the CDO has broken new ground as an investment instrument in micronance, investment funds have also been growing, and, as previously noted, are thought today to control more than $2 billion of capital. Of course, investment funds in micronance are not new. Traditionally, lowreturn or no-return funds sponsored by non-prot organizations have been34Micronancea major source of funding for micronance. What is new is an emphasis on funds that actually oer a return to investors. Even as of January 2008, of the 89 micronance funds listed by MicroCapital, a micronance news and research service, only 26 are characterized as actually seeking a nancial return.16 Would-be institutional micronance fund managers have several hurdles to overcome in persuading clients to invest:Investment fee cascades. Institutions that manage funds make it a practice for their funds not to invest in other funds in order to avoid a build-up of fees that erodes the ultimate returns to their investors. Also, ceding investment discretion to others may appear to weaken their own standing as managers. Lack of transparency. Investors may nd it dicult to understand the pricing, volatility, and performance of assets that exist primarily in funds, as the portfolio eects and the managers screening activities could mask the underlying data. Liquidity. Funds typically trade o liquidity (that is, redemption) for return. If they provide an easy exit for investors, funds that invest in illiquid assets like micronance will nd it necessary to keep a relatively large percentage of their portfolio in low-yielding cash. Investors with long time horizons, as many institutions have, may prefer to invest directly in the underlying assets and run the liquidity risk.However, funds do play an important role in the growth of capital markets access for micronance. For example, many institutions will choose a fund as their rst investment in a new asset category, relying on the managers experience and knowledge of the market to enhance the investors comfort level, as well as to gain familiarity with a multiplicity of MFIs through a single investment.MICRO-LOAN SECURITIZATIONSCDOs and funds that specialize in senior loans to MFIs are not the only capital markets instruments in micronance. Direct securitization of microloans has attracted a great deal of interest, as micro-loans are relatively homogeneous and vastly diversied. As the spectacular growth in recent years of asset-backed securities in international markets makes clear, investors welcome a pure play risk on granular nancial assets. However, several important constraints are slowing the emergence of a true assetbacked notes product in micronance:The role of international capital markets35Short maturity of micro-loans. As opposed to 30-year mortgages, most micro-loans mature in less than a year and feature frequent amortization, so that all but the shortest-term micro-loan securitizations will need to incorporate a mechanism to roll over or substitute the underlying assets, which greatly increases the structuring complexity and administrative cost. Origination risk. Because the portfolio of underlying micro-loans needs constant replenishment, the ability of the MFIs to originate continually a sucient volume of micro-loans is a signicant additional risk. Important role of servicer. Successful MFIs cultivate intimate relationships with borrowers. Thus, the MFI role in servicing securitized micro-loans is a critical element in the performance of the securitized portfolio. This makes it dicult to portray micro-loan securitizations as pure borrower risk. In eect, the performance risk of the MFI servicer is a key component in the overall risk proleand a dicult one to quantify, much less hedge against. Government regulation. Many emerging market jurisdictions have non-existent, rudimentary, or inexible regulatory structures that pose daunting obstacles to the legal structuring necessary to set up securitization vehicles, execute true sales of micro-loans, and transfer payments transparently to oshore investors.Given these constraints, there have been only two cases of micro-loan securitization in international capital markets (as opposed to CDOs that securitize loans to MFIs), and both of them have featured substantial credit enhancement by non-commercial investors. First in May 2006, ProCredit Bank Bulgaria, a subsidiary of ProCredit Holding AG, sold 48 million of its loan portfolio to institutional investors in a deal rated BBB by Fitch Ratings. The European Investment Bank and KfW, the German development agency, provided partial guarantees.17 Four months later, BRAC, a large Bangladesh MFI, held the rst close of a program, backed by microloans, which will issue $15 million (local currency equivalent) of six-month maturity notes twice a year for six years. The issue was rated AAA by a local rating agency. The partial guarantors were KfW and the Dutch development agency FMO.18EQUITYAs MFIs mature and transform from non-prot organizations into companies, including in some cases regulated institutions, their need for equity36Micronancegrows. With the high return on equity and fast growth of the industry, the internal rate of return of MFI equity investment looks compelling on paper. Consequently, at least 15 private equity funds mobilizing $620 million (much of it from non-commercial sources, however), have been set up to address this need.19 The major uncertainty in commercial equity investment in MFIs is the small number of exits, that is, portfolio investment liquidations, to date. Most private equity investors look more to capital gains upon sale of their stakes and less to dividends as the principal component of their return. This is appropriate in micronance as MFIs need to retain earnings in the business to nance further growth if they are to escape an endless cycle of sourcing fresh equity. But without a deep track record of successful exits, the private equity investor is entitled to puzzlement if not skepticism regarding the prospective return on MFI equity investment. The only private equity fund that has gone through a complete cycle of investment and liquidation is ProFund Internacional SA, which from 19952005 invested approximately $20 million total in ten Latin American MFIs for an annual average return of 6 percent. ProFund is of interest here not for its nancial returnsit was sponsored by socially motivated investors and did not set out to maximize protsbut for its success in realizing all ten exits within its allotted ten-year life.20 All but one of ProFunds exits came from sales to shareholders or sponsors of portfolio MFIs, several of them pursuant to a put (that is, a contract requiring one counterparty to purchase an asset at a specied price from another party at the sellers option) or pursuant to an agreement among existing shareholders. While eective in the case of ProFund, exits to insiders (management, major shareholders, and sponsors) are worrisome to private equity investors if they are the only feasible means of liquidating investments. Investors prefer a mix of mechanisms including those that bring in third-party buyers, such as initial public oerings (IPOs), mergers, and acquisitions, in order to set arms-length pricing and foster competition. Moreover, puts to insiders expose the put-holders (that is, investors) to the credit risk of put-writers (that is, insiders), and expose the put-writers to substantial future liabilities they may not be willing to take on, or may accept only at very conservative valuations. If a put can be agreed, and the credit risk of the counterparty is acceptable, the risk-adjusted return is not likely to excite the private equity investor. Acquisitions by nancial or strategic investors are more welcome pathways to exit, but there have been very few examples of this in micronance. Micronance networks might seem to be likely acquirers but most, whether for-prot or non-prot, prefer to build their own operations in new countries from the ground up or to partner with smaller, non-corporate MFIs.The role of international capital markets37No substantial organization has attempted a roll-up, or a growth strategy through acquisition to date. IPOs have been used to provide exits to investors in two signicant casesEquity Bank in Kenya and Compartamos in Mexico. The latter transaction, in April 2007, garnered much publicity, some of it unfavorable, for putting $450 million into the hands of existing investors who had paid approximately $2 million for these shares originally.21 Equity Bank also rewarded its early investors, but on a smaller scale. These examples have given MFI owners and private equity investors hope that IPOs will provide lucrative exit opportunities. However, few emerging country stock markets have sucient liquidity to provide assurance of full valuation. In addition, both Compartamos and Equity Bank are relative giants in their jurisdictions. As market leaders and rst-movers, they represented unique investment opportunities that by denition later entrants to these domestic public markets will not provide. It is more likely that MFI acquisitions will provide consistent exit paths for private equity investors. Strategic investors such as commercial banks, leasing companies, and insurance companies will see the value in MFIs not just as lenders but as delivery vehicles for other nancial services to a proprietary and loyal customer base. These potential investors in many cases will come cross-border, recognizing that the fundamentals of micro-lending are roughly similar in most countries, as shown by the success of networks that apply a common methodology across the developing world. Already some Western European banks have purchased Eastern European banks that specialize in small and micro-enterprise lending in order to extend their footprint into the European Union hinterland. Another likely source of acquisition is by a competitor. Already some countries, including Bolivia, Ecuador, India, Nicaragua, and Peru, are seeing competition among MFIs that previously relied for growth on an under-penetrated market.THE CONTRIBUTION OF NON-COMMERCIAL INVESTORSAs we saw in the BOMSI case, an ocial development agency can provide credibility and ease market acceptance of a product even without direct enhancement of risk. But as international capital markets grow more familiar with micronance, the value of the halo eect is diminishing. Yet noncommercial investors are not superuous in micronance. They can play a valuable role in taking on risks that commercial investors dont understand or are uncomfortable with, and in so doing, leverage this investment. For38Micronanceexample, the Global Commercial Micronance Facility (GCMF), sponsored and managed by Deutsche Bank, is a $75 million fund whose investors include socially responsible HNWIs, ocial development agencies (from the USA, the UK, and France), foundations, and also a number of commercially motivated investors such as banks, insurance companies, and pension funds. The facility is designed to make it easier for MFIs to obtain local currency loans from local banks.22 While we deal with local currency issues later in this chapter, the importance of the facility for this section is to recognize that most institutional investors are uncomfortable taking local currency risk, especially inasmuch as many currencies in emerging markets either cannot be hedged or can only be hedged at unacceptable cost. In essence, the GCMF takes advantage of the ability of non-commercial investors to shield commercial investors from risks they are unwilling to take on, thus leveraging the risk capital of the non-commercial investors to the benet of both. Another, less obvious, example of the catalytic role of non-commercial investors is the $11.4 million bond issued by Micronance Bank of Azerbaijan (MFBA) in August 2007, managed by DWM. This was the rst case of a bond issued in international capital markets by an MFI without credit enhancement. While MFBA had a growing and protable business, the issues attractiveness was bolstered by investors perception that the AAA-rated development agencies owning a majority of MFBA shares European Bank for Reconstruction and Development, International Finance Corporation, and European Investment Bankwould step in if the issuer faced nancial diculty rather than face the embarrassment of a default in a portfolio investment. Whether providing a halo to comfort commercial investors or actually taking on risk that commercial investors feel uncomfortable with, noncommercial investors can signicantly speed up access to capital markets investment for MFIs. But it appears that bilateral and multilateral development agencies are going beyond this role and are actually crowding out private sector investors in commercially credible deals. MicroRate, a Washington, DC-based MFI rating agency, has published a study23 claiming that development agencies are today heavily concentrating their funding on the largest and most successful MFIs, exactly the target investment market of private investors. The study posits that development agencies tend to make easy choices and are squeezing private investors out of the market with their subsidized nance rates. In 2005 (last full year of data), the study found that the development agencies increased their direct funding to top-rated MFIs by 88 percent. At the bottom of the pyramid, where MFIs are most in need of the patientThe role of international capital markets39capital and technical assistance that these agencies provide at taxpayer expense, the development agencies actually cut their funding to the lowestrated MFIs by 25 percent. Shortly after the appearance of the publication, a number of private sector micronance funders joined together to appeal to the development agencies to change this practice, but the results have been inconclusive. Clearly, if development agencies see their roles as competing with private sector investors, they will slow the access of micronance to capital markets.LOCAL CURRENCYOne of the largest constraints to growth of micronance funding is the illiquidity and volatility of many local currencies in the developing world. Of course, if MFIs were able to rely on local funding sources, this would not be a problem. But, as we noted earlier, the bond markets of most developing countries are thin and poorly regulated. Moreover, institutional investors, the largest capital sources in these countries, are often highly restricted in their permitted range of investments. Paradoxically, local commercial banks, which should be a major source of funding for MFIs, in many countries are less likely to accept MFI risk than foreign banks. This is symptomatic of the larger problem of riskaversion among these banks. In many countries, capital-hungry governments crowd out private lender borrowers. In some countries, banks are content to lend to large corporations, state-owned entities, and foreign businesses and are under no pressure to expand their presence into smaller indigenous businesses. In some countries, banks have simply not made the eort to understand and analyze MFI risk, assuming that banking the unbankable, whether directly or indirectly through MFIs, cannot be prudent. Foreign investors typically are uncomfortable with local currency risk that cannot be hedged. This means that many MFIs must borrow in dollars or euros and push the risk onto their borrowers. Fortunately for the MFIs the short maturities of their loans gives them exibility to eectively reprice their assets to account for currency uctuations. Even more fortunately for the MFIs, most borrowers are unable to access capital from other sources and so accept interest rate hikes that a more auent and competitive market would challenge. Nevertheless, adjusting constantly to unforeseeable shifts in exchange rates is a strain on MFI operations and imposes additional risk on borrowers. On occasion, MFIs and oshore lenders hedge by depositing the hard currency loan in a local commercial bank, which then lends to the MFI in40Micronancelocal currency, secured by the deposit. (In a variant of this technique, the deposit-taking bank is dierent from the local bank but issues the local bank a standby letter of credit to secure the risk of the MFI local currency loan.) Although the local banks loan to the MFI is eectively risk-free, the local bank frequently will not reduce the interest rate to the MFI by a large enough quantum so that the combination of the local currency interest rate plus the guarantee fee paid to the oshore lender for taking the risk works out as a feasible nancing cost for the MFI. A number of initiatives are underway to provide unorthodox hedging facilities for capital markets investors in thinly traded currencies. The Dutch development agency FMO, for example, is putting together a swap vehicle capitalized with $350 million in equity that would support $1.5 billion outstanding in currency swaps that are beyond the maturity available commercially. By acting as swap counterparty for a basket of emerging market currencies, the facility aims to achieve risk mitigation through diversication while providing a substantial return to equity investors.24 Ultimately, local currency markets will mature and provide ecient and exible hedging tools. In addition, by that time, local capital markets may have suciently matured to lessen the strain put on foreign investment to meet MFIs growing capital needs.ON THE PATH TO AN ASSET CLASSThe term asset class has a number of denitions. From an institutional investors standpoint, an asset class is a kind of asset that is suitable for inclusion in an investment portfolio. In order to be suitable, the asset class must fulll certain requirements. Fundamentally, it must be recognizable as a distinct kind of asset, such that dierent investments in the same asset class can be analyzed together, can substitute for each other, and can be relied upon to perform similarly in similar circumstances. Crucially, the asset must be liquid, so that portfolio managers can trade into and out of the asset easily according to their changing viewpoint and their portfolios cashow. Liquidity is a function of several factors including volume, exchange listings, ratings, research, and so on. Additionally, it is important that the asset has a track record, data that can be analyzed to make predictions about price changes in response to market conditions. If the asset is relatively less correlated to other assets in the portfolio, that is, of course, a positive as the overall volatility of the portfolio will be reduced by including the new asset in the mix. Overall, micronance funding is a long way from meeting these requirements. It approaches the denition most closely in its distinctiveness andThe role of international capital markets41relative homogeneity. But it is extremely illiquid and likely to remain so for an extended period of time while volumes build up. Secondary markets are not likely to develop until there is a critical mass of exposure among a large number of investors so that willing buyers can be matched with willing sellers. Interestingly, a case can be made that micronance is largely uncorrelated to other emerging market assets and so would reduce portfolio volatility, or beta. In a study sponsored by DWM and carried out by New York University25 the operating performance under dierent economic scenarios of 283 MFIs in 65 developing countries was compared with that of 112 commercial banks from 33 developing countries. The ndings were that MFI nancial results are less sensitive to economic downturn than that of emerging market commercial banks. While the authors concede that the study is based on somewhat inconsistent and incomplete data, it nevertheless serves as a useful indicator and will likely lead to further useful investigation of the characteristics of micronance as a prospective asset class.IS MICROFINANCE RIDING FOR A FALL?Looking ahead, some micronance investors see events on the horizon that worry them: How will micronance perform in turbulent economic conditions? The concern is that the risk of default in the event of global or even localized recession is unknowable, and may be substantial, for MFIs that have only operated during periods of prosperity. This fear overlooks the fact that micronance as a nancial service segment is not nascent, even though capital markets only recently discovered the asset. Many MFIs have been in business for 1020 years and have weathered signicant economic and political instability in countries such as Indonesia and Bolivia. Experience and research, such as the correlation study noted earlier, indicate that MFIs are inherently less vulnerable to economic shocks than other nance providers. (Of course, a sovereign event such as rescheduling or capital controls, or a breakdown of law and order, could force default on even the strongest and most liquid MFI, as well as any other debtor to external markets.) The top tier of MFIs shortly may be overbanked The fear is that too much investment is chasing too little opportunity and that returns are falling to the extent that investors will lend imprudently to lower-quality MFIs in order to meet return expectations. The current compression of42Micronanceemerging market spreads relative to higher-rated paper, while cyclical, highlights this concern. However, while many of the best-known and largest MFIs are attractive candidates for investment, many smaller and more obscure MFIs also have high-quality credit risk. This stems from the underlying robustness of the micronance business model. Most micro-enterprises operate under the radar of the formal economy. The level of economic activity they engage in is so basic as to be immune from the normal ebb and ow of the economic and political systems they operate in. Their operating margins are commonly quite high (although, of course, small in absolute terms). Their employees are family members or close associates whose terms of employment are informal and exible. Their owners liability for business debts is not limited by a legal formmicro-borrowers take personal responsibility for the loans made to them, and they know their ability to continue to make a living, and often to maintain the respect of their community, is intrinsically tied to their punctual payment of all amounts due. For the MFI, administering the loan book is time-consuming and laborintensive, but once the procedures are carefully designed, inculcated, and tested in practice, operations are usually stable, and extending the customer base of the MFI by opening new branches becomes almost routine. Financial controls need to be strict and minutely observed, however. In fact, it is dicult to nd instances of default by MFIs that seek selfsuciency (that is, do not view themselves as charitable operations) and have been in business several years. Certainly some MFIs may have sought support from the international networks they belong to in order to shore up a weakened balance sheet or improve faulty operations. In addition, some MFIs are believed to understate their portfolio at risk numbers by routinely extending the maturity of overdue loans. But MFIs that practice this usually do end up collecting close to 100 percent of the principal and interest from the overdue borrowers. Are MFIs abandoning their core constituency? A third concern is the move of some MFIs up market along with their more successful clients. While the vast bulk of MFI activity currently consists of small loans to individual micro-entrepreneurs some MFIs have begun to oer more sophisticated services to larger clients involving more substantial riskssmall business lending, mortgages, factoring, leasing, insurance, and so onand also enhanced revenue. Generally speaking, a larger loan is more protable to a nancial institution than a smaller one, as administration costs do not increase proportionately with loan size. This is a controversial development. Some observers denounce MFI mission drift and worry that MFIsThe role of international capital markets43will abandon their low-income clients as they progress upstream. Others believe MFIs can continue to remain committed to poverty alleviation and still retain their more successful clients as they accumulate wealth. As these products take on more importance on MFIs balance sheets, the analysis of the MFIs nancial strength will grow more complicated, and their performance vis--vis other emerging markets assets may grow more highly correlated, reducing their value in lowering portfolio beta. On the other hand, as these MFIs grow to more resemble mainstream nancial institutions, both in terms of size and structure, they may attract the attention of some mainstream analysts, traders, and investors, further enhancing investment sources and liquidity. Ultimately, while some MFIs may turn their backs on their origins, most will keep their focus on micro-loans even while providing higher-level services, both because micronance is good business in itself and because it will provide the breeding ground for the higher-value customers.CONCLUSIONThe rush of capital markets investment in micronance is unprecedented and it is wise to question its sustainability. Certainly, risks to continued growth abound, and we have noted a number of them, including: eventual exhaustion of investment opportunities at the well-known and accessible tip of the MFI pyramid; structural obstacles to providing investors with direct exposure to micro-loans via securitization; scarce track record of equity exits; lack of clarity regarding the role of non-commercial investors; underdeveloped local capital markets, coupled with insucient hedging tools for foreign currency investment; illiquidity, sparse data, and small volumes slowing the journey toward achievement of asset class status; mission drift eroding MFIs distinctive risks and returns, and lessening their value in reducing portfolio volatility.Many of these risks reect the fact that micronance has only recently been introduced to capital markets. They should ease over time as investors accumulate exposure to this asset. Extrapolating current trends points to nancial products that are more numerous, more standardized, and more tted to capital markets norms. At the same time, secondary markets will make their appearance, and ratings agencies and researchers (both commercial and academic) will focus more attention on the sector. Specialized44Micronancehedging tools will ease the distortions of too much lending in foreign currency. These developments should abet liquidity and help to give investors comfort that micronance is suitable for regular allocations of portfolio investment. In eect, investor demand for assets itself will become an important and self-fullling driver of progress in micronance. Moreover, as MFI owners and managers grow accustomed to an environment in which a deep pool of commercial funding is available for the well-run, expanding MFI, we can expect strategic transactionsmergers, acquisitions, buy-outs, roll-outs, listings, and so onto become integral elements in the life cycle of successful MFIs. This will result overall in stronger, more ecient, and more skilled institutions better serving clients needs. Of course, too rapid growth could also lead to speculation, overheating, and a crash, as we have seen many times before in nancial markets, from junk bonds to high tech to sub-prime. And certainly some MFIs will expand too quickly and lose control of their costs and their loan books, or cut rates too aggressively for competitive reasons, or push their clients into over-indebtedness. Micronance is no more immune to excess than any other business activity. But the inherent robustness of the micronance business model lays down a strong foundation for solid growth, and the sizable potential market ensures absorption capacity for substantial fresh nancing. Overall, the distinctive focus of micronance on banking the unbankablebringing nancial services to customers outside the formal nancial systemgives it a unique and attractive prole of risk and reward that can draw institutional investors seeking diversication and absolute return even those who are unmoved by the prospect of promoting social values.NOTES1. 2. Estimate by the author. Littleeld, E. (2007), Building nancial systems for the poor, presented at Cracking the Capital Markets Conference, New York, 19 March. The author adjusted the data to remove commercial investment in CDOs (which are treated in this chapter as transactions, not funds) and to add more recent funds aimed at the commercial market, such as the 160 million SNS Institutional Micronance Fund, managed by DWM, which had its rst closing in May 2007. CGAP (2007), Micronance investment vehicles, CGAP Brief, April, www.cgap.org/ portal/site/CGAP/menuitem.95cb370f4995240167808010591010a0/. For example, the CGAP brief cited in Note 3 states that portfolio investment by international nancial institutions doubled from 2004 to 2006, while micronance investment channels targeted to individuals report increasing assets, such as Kiva (www.kiva.org). See Walker, R. (2008), Extra helping, The New York Times Magazine, 27 January. DWM research.3. 4.5.The role of international capital markets6. 7. 8.459. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.In this chapter, capital markets means transactions or funds in which all or a major portion of the investment is raised from private sector institutional investors seeking fully risk-adjusted returns. DWM research. Data in this paragraph and the next three come from Optimizing capital supply in support of micronance industry growth, a presentation by McKinsey & Company to the Micronance Investor Roundtable in Washington, DC on 24 October 2006. The exception is the reference to 10 000 MFIs, an estimate widely quoted in the literature; see for example, Odell, Anne Moore (2008), Micronance: catch the swelling SRI wave, Sustainability Investment News, 11 January. Littleeld (2007), p. 3see note 2. According to the Dexia website, 27 January 2008, www.dexia.com/e/discover/ sustainable_funds 2.php. BlueOrchard uses a standard short-term rate as its benchmark, six-month LIBOR (London Inter-bank Oered Rate), the rate that prime banks charge each other for liquidity. Dexia Micro-Credit Fund Monthly Newsletter, BlueOrchard Finance, November 2007, www.blueorchard.org/jahia/Jahia/Site/blueorchard/Products/pid/42/dexiannews letter. A SICAV (socit dinvestissement capital variable) is an open-ended fund common in Western Europe especially Luxembourg, Switzerland, Italy, and France, comparable to a mutual fund in the USA. CDSF (2007), Capital markets-style risk assessment: testing static pool analysis on micronance, Center for the Development of Social Finance, March, www.cdso.org/downloads/MFIStudy-CDSF-Mar 07.pdf. CUSIP is an acronym for the Committee on Uniform Securities and Identication Procedures, a standards body. A CUSIP number uniquely identies a specic security to facilitate custody and trading of securities. MicroCapital, 27 January 2008, www.microcapital.org/?page_id=7. Press Release, ProCredit Holding AG and Deutsche Bank, 15 May 2006. Rahman, R. and S. Shah Mohammed (2007), BRAC micro credit securitization series I: lessons from the worlds rst Micro-credit backed security (MCBS), MF Analytics, Ltd, Boston, 20 March, www.micronancegateway.com/les/45785_le_11.pdf. DiLeo, P. and D. FitzHerbert (2007), The investment opportunity in micronance, Grassroots Capital Management LLC, June, www.grayghostnd.com/industry_ insights/viewpoints/the_investment_opportunity_in_micronance. ProFund Internacional, SA (2008) www.calmeadow.com/profund.htm. Rosenberg, R. (2007), CGAP reections on the Compartamos initial public oering: a case study in micronance interest rates and prots, Focus Note (42), June. USAID (2007), The Deutsche Bank global commercial micronance consortium and USAIDs DCA guarantee, United States Agency for International Development, January, www.microlinks.org/ev_en.php?ID=17450_201&ID2=DO_TOPIC. Abrams, J. and D. von Stauenberg (2007), Role reversal: are public development institutions crowding out private investment in micronance?, MicroRate, February, www.microrate.com/pdf/rolereversal.pdf. TCXthe currency exchange (2008), www.fmo.nl/smartsite.dws?id=88. Krauss, N. and I. Walter (2006), Can micronance reduce portfolio volatility?, Stern School of Business, New York University, November.3. Securitization and micro-credit backed securities (MCBS)Ray Rahman and Saif Shah MohammedINTRODUCTIONBRAC Micro Credit Securitization Series I, closed in August 2006, was the worlds rst securitization of micro-credit receivables and the rst of a new type of investment called micro-credit backed security (MCBS). This securitization was also the rst AAA-rated transaction within Bangladesh. A number of innovative transactions have taken place in the micro-credit industry in the last few years. In 2004, ICICI bank purchased 25 percent of SHARE Micron Ltds micro-loans in a US$4.9 million transaction. The purchased micronance receivables were valued at their net present value at an agreed-upon interest rate. A portion of the transaction was guaranteed by Grameen Foundation USA. Also in 2004, BlueOrchard issued its rst collateralized debt obligation (CDO). In these, and subsequent transactions, BlueOrchard pooled collections of loans made out to a number of micronance institutions around the world. Unlike the ICICI transaction, the BRAC securitization is scalable and allows for tranching. It is not a one-time sale of receivables at a discount. Unlike the BlueOrchard CDOs, the BRAC securitization involves the direct pooling of micro-credit receivables instead of the pooling of loans disbursed to micronance institutions. The investments are thus directly linked to the performance of the underlying portfolio of micro-credit receivables instead of the risk of the originating institution. The transaction is entirely in local currencythus removing any currency risk from the originating institution. As such, the BRAC securitization represents a step in the evolution of the linking of micronance to capital markets. In general terms, the transaction size is US$180 million in local Bangladesh currency. It consists of 12 equal tranches with the asset pool backing each tranche, mirroring the overall risk prole of BRACs microcredit portfolio. The tenor of the transaction is 6.5 years, with each tranche maturing in 12 months. It is a xed rate private placement sold pre46Securitization and micro-credit backed securities47dominantly to local Bangladesh investors. BRAC is the servicer, and Eastern Bank Limited (EBL) is the trustee. Having arranged and structured this transaction, we discuss the story behind it, and lessons from it for the future. The next section deals with the transaction rationale that convinced BRAC to agree to going forward with a securitization. The third section examines various political economy considerations that had to go into arranging the transaction. The fourth looks at the actual structure of the transaction. It highlights various data and logistical constraints that were reected in the nal structure. Working with BRACs micro-credit portfolio data allowed us to spot certain risks that were not anticipated at the beginning of the structuring process, and these risks were mitigated in the nal structure. Entering the eighth month of the transaction, we are able to draw some lessons about how well the structure has held, and what can be taken from it in future transactions. The fth section thus examines the performance of the bond and the underlying pool of securitized micro-credit receivables since the beginning of the transaction. We scrutinize the various credit enhancements that were put in place, and explore which of them could be reduced or removed in future securitizations. The nal section, as the conclusion, relates some nal lessons from the transaction and ideas for the future.TRANSACTION RATIONALEEstablished in 1972, BRAC is Bangladeshs largest NGO, with nearly 100 000 employees. BRAC takes a holistic approach to development, combining micro-credit activities with health care, education, and social advocacy. With its 1381 area oces, 35 000 primary schools and 35 health centers, BRAC has a presence in every district in Bangladesh. As of March 2007, BRACs micro-credit program had more than 5.3 million borrowers with outstanding loans. BRACs micro-credit activities are carried out through three programs targeted at dierent borrower groups: 1. 2. Dabiwith loans ranging from around US$50 to US$500 in size, targeting individuals with less than 1 acre of land; Unnatiwith loans ranging from around US$150 to US$850 in size, targeting individuals with more than 1 acre of land, or individuals who have graduated from the Dabi group; Progatiwith loans ranging from US$350 to US$5000 in size, targeting small enterprises.3.48MicronanceTable 3.1 BRAC micronance sources of funding 200205Source (%) Members savings Loan from PKSF (quasi-government) Grant from donors Retained earnings Loan from banks (including syndication) Others 2002 36.70 28.50 16.30 11.90 2003 45.70 23.85 12.97 14.05 2004 48.33 15.71 10.35 15.02 2005 46.30 8.14 8.53 16.34 CAGRa (three-year) 8.05 34.14 19.42 11.156.10 0.003.24 0.009.50 1.0920.69 0.0050.25Note: a. Compound annual growth rate. Source: BRAC.Loans generally have a one-year maturity. The interest rate is a at 15 percent, with collections taken on a weekly basis for Dabi and Unnati, and on a monthly basis for Progati. The BRAC portfolio has been growing rapidly in the last few years. Between January 2006 and January 2007, for example, the principal outstanding on loans grew by 35 percent from US$280 million to more than US$380 million, and the number of borrowers grew by more than 10 percent to more than 5 million borrowers. BRAC has been funding its micro-credit activities from a number of sources, including the savings of its group members/borrowers, donor grants, and funding from the government and some donor agencies through loans from Palli Karma Sahayak Foundation or PKSF, the apex micronance institution. In early 2004, BRAC borrowed the equivalent of nearly US$30 million from the local capital markets through its rst syndication. Table 3.1 gives the breakdown of BRACs micronance funding sources prior to the transaction. These nancing options had severe drawbacks. Donor funding was volatile, and could not be relied upon in long-term planning. BRAC also felt itself under increasing pressure from the government to reduce the interest rates charged to borrowers as a precondition for further loans through PKSF. However, BRAC considered its current interest rates necessary for the sustainability and viability of its micro-credit program and felt that they could not be reduced further. Financing through syndications was problematic in the long run as interest rates on future syndications would be driven by the volatile credit market and would not adequatelySecuritization and micro-credit backed securities49reect the historical performance of BRACs micro-credit program. Besides, syndications could not be carried out indenitely without falling foul of regulatory limits on borrowing. Securitization presented itself as an attractive alternative, or at least complement, to the existing sources of funding. Securitization would allow BRAC to have a more ecient balance sheet. It would improve BRACs asset/liability management, and reduce leverage. Properly structured, the transaction could obtain a higher rating than BRAC, and allow BRAC to raise lower-cost funds. A longer-term transaction would give BRAC the ability to plan out its explosive growth. And signicantly, BRAC would have a broader investor base. Through the transaction, BRAC would be able to remove its dependence upon PKSF funding, thereby alleviating some of the pressure to reduce interest rates.POLITICAL ECONOMY CONSIDERATIONSThe Approval Process Bangladesh Bank The central bank initially welcomed the concept of securitization as a way to deepen the capital markets. In December 2004, around the time the BRAC transaction was initially discussed, Bangladesh Bank arranged a well-attended workshop on securitizations as a source for funding for infrastructure and development projects. And yet, over the next two years Bangladesh Bank did little to actually implement many of the proposals that came out of the December 2004 workshop. For whatever reason, the bank simply refused follow-up on many of the changes one might deem necessary for having more ecient capital markets. For example, in the workshop (and in other public and private discussions with the bank), the need for a yield-curve, was noted by the local nancial community. As we write this report, the yield-curve is still non-existent in the Bangladeshi banking sector, and there are no ocial eorts in place to create such a benchmark. Our own discussions with Bangladesh Bank about securitizing microcredit receivables were well received in the months after December 2004. And based on these interactions with Bangladesh Bank, the structure for the transaction was created and submitted to the bank for approval around June/July 2005. The original signals from the bank indicated that the approval process would be a short one. After all, the structure addressed the concerns that it had raised, namely currency risk concerns and the involvement of local nancial institutions. The entire currency risk of the foreign investment was borne by FMO of the Netherlands, and two-thirds of the50Micronancetransaction would be subscribed by local investors. Yet, Bangladesh Bank withheld approval until the end of December 2005/January 2006. Bangladesh Bank raised three objections. First, we would have to further reduce the involvement of foreign investors by reducing the size of the tranche covered by a KfW/FMO guarantee. Second, the guarantees would have to be removed completely after the rst year. Third, Bangladesh Bank was worried about income from the bonds and fees being remitted outside of Bangladesh by FMO and KfW. This concern was directly related to the overheating of the local US dollar market in the second half of 2005, which saw the exchange rate move up in a matter of months from around taka 60/USD to more than taka 70/USD. Due to the increase in global oil prices, the government-owned Bangladesh Petroleum Corporation borrowed US dollars heavily from nationalized commercial banks. The structure was changed to meet the rst two concerns. For the third concern, we were able to show that the impact on the local dollar market from remittances would not be as strong as anticipated by Bangladesh Bank. The structure required FMO to buy a new tranche every six months, and this would actually help in the local currency crisis. Further, the regulators were simply an order of magnitude o in their understanding of the size of fees and other remittances. Securities and Exchange Commission The approval process at the Securities and Exchange Commission (SEC) was similarly fraught with delays. Since the bond issued in each tranche would have a one-year maturity under our structure, the existing rules did not require an SEC approval. The SEC conrmed this in our conversations with it while the transaction was being structured. However, the Bangladesh Bank approval letter required us to approach the SEC again on this question. In mid-January 2006, we approached the SEC for a nal conrmation that the approval process was not required under the rules. The SEC now decided that it would require us to get approval after all because of the six-year term of the transaction, even though the securities issued themselves would be of one-year maturity. Then in early February 2006, we were again surprised with a long request for documents and analyses. We submitted these documents and analyses almost immediately. In March 2006, we received an AAA rating for the transaction from the Moodys-aliated Credit Rating Agency of Bangladesh (CRAB). After this, the SEC did not logically have a reason to withhold approval. However, it was only in June 2006 that we received the approval.Securitization and micro-credit backed securities51What Explains the Delays in the Approval Process? We can only speculate about the reasons behind the delays in the approval process. Some of the concerns expressed by the regulatorssuch as pressures on the local dollar marketwere genuinely felt. And regulators may very well have been uncomfortable with the transaction due to their unfamiliarity with both micronance and with securitizations. After all, this was a rst-of-a-kind transaction with a number of moving parts (as described in the next section). Further, the bureaucratic regime is not one known for its transparency or nimbleness. Based on informal discussions, we can suggest that some of the delays may have been related to deeper concerns of the government. PKSF Established in 1990, PKSF is the apex micro-credit organization in Bangladesh, a privatepublic partnership between the micronance NGOs and the government. The organization receives funding from the government, the International Development Association (IDA)/World Bank, USAID, the Asian Development Bank (ADB), and the International Fund for Agricultural Development (IFAD). It lends these funds to partner micronance organizations at a below-market rate. PKSF has successfully utilized its funding clout to induce micronance organizations to strengthen their reporting and auditing processes, as well as implement computerized management information systems (MIS) at the head oce (and in some cases, area oce) level, thereby strengthening institutional capacity. The organization has played some role in coordinating the activities of micronance organizations to better target groups that may not have been receiving micronance services. It has also facilitated research on micronance activities. BRAC itself had received US$3040 million in loans from PKSF. But as discussed previously, BRAC felt itself under increased pressure to reduce the interest rates it charged as a precondition for being able to borrow again once the PKSF loans were paid down. One of the rationales for the securitization was thus to move away from dependence on PKSF funding. Further, PKSF may have felt its role as the apex micronance body in Bangladesh threatened by the prospect of a micronance organization being able to raise funds through securitization without PKSF involvement. The government 2005 was the UN International Year of Microcredit. In what is arguably the home of micro-credit, the governments reception of the year was schizophrenic. On the one hand, the government sponsored a number of52Micronanceworkshops and events to celebrate the year and highlight the achievements of micronance institutions. On the other, the headlines in Dhaka were at times dominated by statements from the Minister of Finance publicly doubting whether the activities of micronance institutions had any positive role to play in development. In the last quarter of 2005, there was a public disagreement between the Chairperson of BRAC and the Minister of Finance over the issue of interest rate caps. But the governments apathy to micronance activity in Bangladesh was not animated solely by policy concerns about high interest rates. Since 2001, the centre-right BNP had ruled Bangladesh in alliance with the rightwing Jamaat-i-Islami. Jamaat represented many of the most conservative elements of Bangladeshi society, and historically the relationship between these constituents of Jamaat and the activities of the development NGOs has been a tense one. Further, the NGOs have been vocal about the rights of minorities under the BNPJamaat coalition. The inuence of Jamaat on government policy since 2001 has been widely reported. A less well-publicized aspect of the relationship between the government and the micronance sector in Bangladesh has been concerns about the involvement of micronance institutions in political activities. With millions of borrowers and beneciaries of health care and education programs, micronance institutions have an unprecedented ability to mobilize voters. In past years, prominent organizations such as Proshika have been accused of openly partisan activities. The Proshika case is worth looking at closely to get a sense of government concerns. Like BRAC, Proshika was involved in education and social advocacy along with its micronance activities. Like BRAC, Proshika had also been able to scale up its micronance activities to reach millions of borrowers in nearly every part of Bangladesh. In the mid-1990s, Proshika played a prominent role in the opposition Awami League-led movement for elections under a neutral caretaker government. In the 1996 and 2001 elections, BNP accused Proshika of slanting its voter education material in favor of the Awami League. And in April 2004, Proshika was accused of trying to mobilize thousands of its borrowers to assemble in the capital to launch an opposition platform. In May 2004, the BNP/Jamaat government cracked down upon Proshika, arresting its chairperson and stopping the ow of donor funding to its programs. Whatever the merits of the Proshika case, government misgivings about the inuence of the large micronance institutions and their potential for quickly mobilizing thousands for political causes led in 2004 to the drafting of laws deepening government control of NGOs. It was in this climate that the BRAC securitization was proposed to the regulators. Concerns about the perception of Proshikas activities had led BRAC to form theSecuritization and micro-credit backed securities53Federation of NGOs of Bangladesh (FNB) with other NGOs, breaking away from the Proshika-led Association of Development Agencies in Bangladesh (ADAB), the apex development group. But the government likely felt that BRACs dependence upon the whims of government funding through PKSF was a vital lever for controlling its activities.STRUCTURE OVERVIEWThe Structure The parties BRAC is the originator in this transaction. BRAC also plays the role of servicer, depositing the collections from the securitized receivables and collateral to the special-purpose vehicles (SPVs) accounts on a monthly basis, updating the pool of securitized receivables and collateral, and reporting the performance of the pool to the investors. The trustee for the SPV is Eastern Bank Limiteda leading local bank. The investors in this transaction are FMO, Citibank, and two leading local banks, Pubali Bank and The City Bank. Citibanks investment shall (for the rst year) be guaranteed by FMO and counter-guaranteed by KfW of Germany. As in most securitizations, BRAC is also the residual beneciary of all cash ows after fees, principal payments, and interest payments are paid out each month by the SPV. MF Analytics shall provide continuing support to BRAC to maintain the securitized pool of receivables and collateral and assist in reporting performance to the investors. RSA Capital was the lead arranger of the transaction. FMO, KfW, and Citibank were co-lead arrangers. The transaction was rated by the Credit Rating Agency of Bangladesh, the local Moodys aliate. BRACs MIS system and MF Analytics pooling and reporting algorithms were audited by PriceWaterhouseCoopers. BRAC itself is audited by Ernst & Young Malaysia. Tranche structure This US$180 million transaction is divided into 12 equal tranches over six years. Every six months, the originator shall sell US$15 million worth of micro-credit receivables to the trust/SPV created for this transaction in return for cash. It shall also earmark another US$7.5 million worth of micro-credit receivables as additional collateral. The trust shall issue US$15 million worth of certicates or bonds of one-year maturity backed by the pool of securitized receivables and collateral to the investors. In return, the investors shall invest US$15 million in the trust on or before the date of issue for the tranche.54MicronanceEach tranche is divided into three sub-tranches. Sub-tranche A is the FMO investment, amounting to US$5 million. Sub-tranche B is the Citibank investment, also US$5 million. As noted already, Sub-tranche B is guaranteed by FMO for the rst two tranches, and counter-guaranteed by KfW. Sub-tranche C, the remaining US$5 million, is issued to the local investors. The certicates issued to the investors in each sub-tranche are pari passu (equitable). It should be noted that the cost of funds for BRAC from this transaction is between 150 to 200 basis points below what would have been available to it had it gone for a straight loan or syndication. The securitized asset pool The underlying asset pool for the securitization consists entirely of BRACs micro-credit receivables. At the outset of each tranche, the pool or the collateral underlying the tranche is selected to reect the distribution of loans in BRACs current portfolio (excluding all loans that have liens from PKSF and syndications) along three variables: 1. 2. 3. program (Dabi/Progati/Unnati); geography (identied by area oces); activity or purpose of the loan (identied by a schematic code).The securitized loans are selected within these categories randomly. As a result, the pool is not biased towards including loans of a particular size or age. The pool is over-collateralized by 50 percent, that is, at the beginning of each tranche, US$22.5 million (in equivalent Bangladesh taka) of receivables are pooled as per the criteria noted above. Furthermore, the asset pool is replenished with additional collateral from month to month if the forecasted cash ow from the pool is less than 140 percent of the SPV liability in the following month. Bond pay-down structure for each tranche The certicates or bonds issued for each tranche are of one-year maturity. The bonds are amortized monthly based on a predened pay-down schedule. Interest is also paid on the outstanding principal outstanding of the bonds on a monthly basis. The pay-down schedules of the certicates reects the actual pay-down of the underlying securitized receivables. The principal pay-down schedules for the rst two tranches is presented in Table 3.2. With new tranches issued every six months over a course of six years, BRAC is provided a committed, long-term source of funding. The tranches disburse funds at a rate that BRAC can absorb without trouble for distribution to its borrowers. The one-year maturity of the bonds reects the short-term nature of the underlying micro-credit receivables, allows theSecuritization and micro-credit backed securities55Table 3.2 Principal pay-down schedules for tranche 1 and tranche 2Montha Tranche 1 1 2 3 4 5 6 7 8 9 10 11 12 12.25 12.25 12.25 12.25 12.00 9.85 9.00 7.50 5.90 4.00 1.90 0.85 % Principal Paid Down Tranche 2 12.000 12.000 12.000 12.000 11.750 11.500 9.750 7.500 4.500 4.250 2.000 0.750Note: a. September 2007February 2008.securitized pool to track BRACs portfolio, and provides the investors an additional level of comfort for investing in this rst-of-a-kind transaction. Credit enhancements A number of credit enhancements were included in this transaction. The 50 percent over-collateralization of the securitized pool and the replenishment of the pool with additional receivables in the event of projected cash ows falling below 140 percent of the following months SPV liability, have already been discussed. A few additional credit enhancements were added for good measure: Substitution Loans identied as delinquent and loans with missing or corrupted data are removed on a monthly basis from the pool. (The delinquent loans are under our denition of delinquency. Loans identied as defaulted loans under BRACs denition are not removed from the securitized pool.) In their place, loans from the same program, geographical region (area oce), and type of activity as far as possible are purchased from BRAC as replacements by the SPV. These replacement loans are selected so that they would mature later than delinquent and missing data loans. Removing prepayment risks Instead of being used to pay the investors or released to the residual beneciary, prepayments are captured in a provisional account. Once loans with prepayments within the securitized pool mature, these prepayments are released at the rate at which these56Micronanceamounts would have been collected had no prepayment taken place. Prepayment risk is thus totally hedged out from the transaction. Debt service reserve account A debt service reserve account (DSRA) of US$2.5 million was provided by BRAC as an additional credit enhancement at the beginning of the rst two tranches. The DSRA amount is expected to be negotiated down after tranche 2. MF Analytics pool maintenance and reporting modules The structure required the creation of a software package by MF Analytics tailored to this transaction. The package included modules that create the pool of securitized receivables, and forecast the cash ow from the securitized pools and bond pay-down structure at the beginning of each tranche. It also included modules that on a monthly basis substitute delinquent and missing loans, trap and release prepayments to completely hedge out prepayment risk, and replenish the pool if necessary with additional collateral. Additional modules fulll all of the reporting needs of BRAC for this transaction. Issues The need for disaggregating data The investors desired the securitized pool at the time of pooling to reect the risk characteristics of BRACs entire micro-credit portfolio. However, once the pooling took place, the portfolio would not track the evolution of BRACs portfolio. Rather, because of substitution, prepayment trapping, and replenishment, the pool would have its own risk characteristics. The risk characteristics of the pool would thus diverge from those of BRACs portfolios over the course of the transaction. Further, as already noted, the substitution of loans on a monthly basis and the tracking and controlled release of prepayments above would require the processing of individual loans. Thus, it simply would not do to analyze BRACs historical portfolio data at an aggregated level for structuring. The structuring of the transaction required two levels of data analysis: 1. We had to identify the drivers of risk in the available data, and design the selection process of the securitized pool so the distribution of identied characteristics in BRACs portfolio would be reected in the selected pool. This involved identifying and analyzing the qualitative and quantitative data available on the loans and borrowers in BRACs portfolio. Once the pool was selected, we had to run simulations of the dynamic pool on the available historical data as well as on simulated data to reect various stress conditions to test and rene the structure.2.Securitization and micro-credit backed securities57Disaggregating the loan data required us to process and analyze enormous amounts of data. At the time of the structuring, BRAC had nearly 5 million borrowers, and the transactions of these loans were tracked monthly. Each securitized pool contained between 250 000 and 300 000 loans. Our analyses of BRACs information system and portfolio also revealed a number of issues that needed to be mitigated in the nal structure: Information and logistical issues Time lag BRAC updates the collection and disbursement data at its 1381 area oce computers daily. However, these data are transferred to the Dhaka head oce only once a month. The infrastructure simply does not currently exist for more frequent transfers of information to the head oce. At the head oce, it takes around a week to complete the process of updating and checking the database. (At the time of the structuring, this process took nearly 1012 days.) As a result, the longest gap between a transaction in the eld and the information reaching the head oce is nearly 42 days. A delinquency or prepayment will in many cases be reported to the head oce nearly 42 days after they took place. The structure would have to solve this lag. Changing collection dates BRACs loans to its borrowers are collected on a weekly or monthly schedule. However, the exact collection dates for a particular loan cannot be known until after the end of the month for a number of reasons. No collections take place on local holidays, and these holidays are often dependent upon the sighting of the moon. BRACs system will, during the month, allow the collection schedules to be updated to reect these local holidays. Similarly, at a country-wide level, national holidaysmany of them based on the lunar calendarrequire shifts in collection dates. Furthermore, changes in BRACs personnel itself can lead to changes in collection dates. At BRAC, each collection ocer is assigned two village groups a day to meet borrowers, examine their activities, collect payments, and make new disbursements. The transfer or promotion of a collection ocer can result in a change in the day of a week that a particular village group will be visited. Forecasts of collections from the pool of securitized receivables would have to reect such uncertainty about collection schedules. Missing, inconsistent, and unusable data The monthly data transferred to the head oce arrives in the form of CDs or zip drives carried by couriers from the 1381 area oces. These data are uploaded to the head oce servers before any reporting can take place. To say that the transfer process is not fail-safe is58Micronancean understatement. Each month, BRACs head oce nds missing information and corrupted data, and has to ask individual area oces to transfer the information again. But such checks do not spot all missing information. At each of the area oces, the information for thousands of collections and disbursements are entered each day by a BRAC-trained computer operator hired solely for this purpose. While the software used by BRAC does incorporate some checks on whether information was entered correctly, some mistakes inevitably creep in. As a result, BRACs data included some inconsistent information. Further, the software used for data entry at BRAC itself generated some inconsistencies. We were able to spot some of these inconsistencies in our due diligence process, and BRAC corrected them. Finally, at the eld oces, there is no way to x malfunctioning computers without sending the machines to Dhaka. Sometimes malfunctioning machines are not spotted early, and some of the information transferred to the head oce is corrupt and unusable. Because of res, theft, or natural disasters, there is always some underlying risk of spoilage of data. Notwithstanding these constraints, we found that errors and inconsistencies in BRACs dataset were relatively rare, less than 1 percent of the data every month. We also noticed that over time, BRACs monthly dataset got better, for a number of reasons. BRAC removed the software-generated inconsistencies that we spotted. Its internal checks on data quality improved. And the process of transferring data became more streamlined. However, the structure would have to take into account the risk of some missing, corrupt, or inconsistent data. Data issues in the risk analysis process Unavailability of historical data One of the limitations that we had to deal with in the structuring process was the incompleteness of the historical information available to us. BRAC area oces had been computerized at dierent times, and it was only near the end of 2005 that the computerization process was completed. Around the same time, BRAC was also in the midst of upgrading the data-entry and database software used at the area oces. While accessible computerized data for some area oces went back a few years, for most area oces accessible data went back only a few months. And because the upgrades to the area oce systems had not been completed, at the time of structuring the deal we were only able to analyze individual loans on a monthly basis, instead of being able to look at the actual dates of transactions. Thus, we would only be able to know if a particular loan failed to pay in a particular month, but not the exact date when it failed to pay.Securitization and micro-credit backed securities59Absence of data on borrower characteristics The BRAC data available to us at the time of the structuring process included information on where the loan was being used and what the loan was being used for. But important demographic characteristics were missing, such as the estimated age of the borrower, the length of time they had been borrowing from BRAC, information on defaults on previous loans, and even the number of installments missed in the current loan. These constraints severely limited our ability to create credit scores for individual borrowers. While the upgrades to BRACs system allow some of this information to be captured, BRACs database still does not include useful information that could be easily incorporated. BRACs health care and education programs, for example, are a rich source of information on borrower households. Yet these databases still do not speak to each other. Deciphering the meaning of defaults At the beginning of the transaction, there was some inconsistency in the way that BRAC and the investors understood what constituted a defaulting loan. BRAC denes a defaulting loan as one that failed to pay its total obligation of principal and interest by the end of the one-year period of the loan. Thus, a loan that missed all weekly payments for months but was still to reach the end of its one-year term is considered a current loan by BRAC. For BRAC, it is this denition that generates the astonishing, well-publicized nearly 100 percent repayment rates. Such a denition was simply not palatable for the investors. But a problem in creating a denition of default or delinquency more consistent with the conventional understanding of the concept was the fact that only monthly aggregates of the transactions of individual loans were available at the time of structuring the transaction. Taking into account data limitations, we dened delinquent loans as those loans that failed to pay their aggregate monthly installments in the immediately preceding month. The denition was stricter than BRACs own in that many BRAC current loans were identied as delinquent. But on the other hand, it was less strict than BRACs in that a loan that was beyond its one-year maturity period but was making its payments on time in the immediately preceding month would not be identied as delinquent. Table 3.3 shows the rate of delinquencies (dened as the principal outstanding of delinquent loans over the principal outstanding of BRACs total current portfolio) for the period January 2005 to April 2006. It also shows the cash ow impact of delinquencies in this period. It must be noted that Bangladesh did not suer from any natural disasters in this period. But even by our arguably stricter denition of delinquencies, the rate of delinquencies has never been greater than 8.25 percent.60MicronanceTable 3.3 Delinquency rate and cash impact of delinquencies, January 2005 to April 2006Month 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 Average Year 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2006 2006 2006 2006 Delinquency Rate (%) 8.14 7.48 8.25 7.81 6.62 6.14 6.42 5.93 5.47 6.51 4.77 4.24 4.83 4.45 4.51 4.89 6.03 Cash Impact (%) 7.69 7.33 7.49 7.35 6.49 5.53 6.09 5.98 5.88 6.62 5.72 4.57 5.18 4.83 4.77 4.88 6.02Source: BRAC data.Analysis of risk Risk variables identied In the available data, our analysis identied the location of the borrower and the type of activity of the loan as relevant risk variables. Borrowers in peri-urban and urban areas, for example, were less likely to be delinquent than borrowers in remote parts of the country. We also identied the age of the loan as a relevant risk variable. Loans that were six to eight months old were more likely to be delinquent than newer or older loans. We also found that loan size did have an impact upon delinquencies. The larger loans given in the Progati program to small enterprises almost never missed payments. Delinquencies were much more frequent for the smallest loans in the Dabi program. Prepayments Our analysis also revealed the existence of risks to the structure that had not been anticipated in the beginning of the structuring process. We noticed that a signicant number of loans prepay before their maturity date. Table 3.4 shows the cash impact of these prepayments between January 2005 and April 2006. It is worth noting that in some months, the positive cash impact of prepayments is comparable to the negative impact of delinquen-Securitization and micro-credit backed securities61Table 3.4 Prepayment rate and cash impact of prepayments, January 2005 to April 2006Month 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 Average Year 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2005 2006 2006 2006 2006 Prepayment Rate (%) 7.73 11.14 5.84 5.63 8.39 3.26 2.69 2.83 5.17 4.82 10.49 7.65 4.58 4.99 5.07 2.70 5.81 Cash Impact (%) 4.22 5.04 2.81 2.95 4.36 2.74 8.83 8.81 10.82 19.02 21.17 11.99 10.11 11.86 10.01 9.12 8.99Source: BRAC data.cies. Our structure would need to take into account the risk of prepayments not leaving enough cash ow for bond payments in future months.PERFORMANCEAt the time of the writing of this chapter, 11 (of 12) payment dates on Tranche 1 of the transaction had passed, and ve for Tranche 2. Over 99 percent of Tranche 1 has already been paid down, and 60 percent of Tranche 2. We are now in a position to look closely at the ecacy of the structure. Delinquency Rates Bangladesh has not witnessed any natural calamities in the last few months. However, there has been some political turmoil over the parliamentary elections, which were cancelled on 11 January 2007 with the declaration of a State of Emergency caretaker government. Table 3.5 looks at the delinquency rate of loans in the securitized pool. (Delinquency rate for a particular month is dened as the principal62MicronanceTable 3.5 Delinquency rate for Tranche 1 and Tranche 2 securitized pool, August 2006June 2007Month August September October November December January February March April May June AverageSource: BRAC Data.Tranche 1 (%) 2.60 2.80 8.20 6.10 4.00 5.80 6.90 14.00 18.00 15.00 16.00 9.04Tranche 2 (%)2.90 5.30 7.30 6.40 7.90 5.96outstanding of loans that missed a single payment in the previous month divided by the principal outstanding of the securitized pool.) Delinquencies, in the politically volatile period between November 2006 and January 2007 never rose above 6.10 percent. The relatively high delinquency rate in October 2006 (just after the ood season) is comparable to the rate from previous years. Delinquencies for Tranche 1 hovered above 15 percent in March 2007. This was because of the aging of the pool. A jump in the delinquency rate was expected, as older loans have a much higher probability of being delinquent. Bond Performance Tables 3.6a and 3.6b show the excess cash ow in the rst two tranches returned to BRAC as residual beneciary, and the future SPV liability covered by the amounts collectible from the securitized pool and cash trapped in the prepayment accounts. It is clear that the SPV is awash in liquidity. This was expected, and was one of the reasons that the credit rating report cited the AAA rating for the bond. For both tranches, at least 150 percent of the SPVs liability in future months is covered by the underlying securitized receivables and the cash trapped in the prepayment account without taking into account the US$2.5 million deposited in the debt service reserve account. But what would have happened had the credit enhancements not been put in place? Tables 3.7a, 3.7b, 3.8a and 3.8b examine the impact of removingSecuritization and micro-credit backed securities63Table 3.6a Tranche 1 excess cash ow and SPV liability covered by pool and trapped prepayments, September 2006July 2007Payment Month 9 10 11 12 1 2 3 4 5 6 7 Year 2006 2006 2006 2006 2007 2007 2007 2007 2007 2007 2007 % Excess Cash Flow 99 87 62 79 56 72 78 96 125 174 339 % Future SPV Liability Covered 157 157 166 172 184 199 222 286 413 762 1823Note: DSRA (debt service reserve account) amount not included in any calculation. Source: BRAC data.Table 3.6b Tranche 2 excess cash ow and SPV liability covered by pool and trapped prepayments, March 2007July 2007Payment Month 3 4 5 6 7 Year 2007 2007 2007 2007 2007 % Excess Cash Flow 70 84 87 84 76 % Future SPV Liability Covered 161 171 178 190 209Note: DSRA amount not included in any calculation. Source: BRAC data.the dierent credit enhancementsover-collateralization, the trapping and controlled release of prepayments, and the substitution and replacement of delinquent loans. Removing over-collateralization would have resulted in the SPV being unable to cover the bond obligations beyond the rst two months for both tranches. The securitized pool barely covers future SPV liabilities in the rstTable 3.7a Performance of dierent credit enhancements for Tranche 1: % excess cash ow of SPV liability returning to residual beneciary under dierent scenariosOver-collat., No Cr. Enhance. Over-collat., w/Prepay. Trap. Over-collat., w/Subst. No Over-collat., w/No Cr. Enhance.Payment MonthYearAll Credit Enhancements649 10 11 12 1 2 3 4 5 6 72006 2006 2006 2006 2007 2007 2007 2007 2007 2007 200799 87 62 79 56 72 78 96 125 174 339102 83 40 49 49 41 37 24 13 11 2687 64 24 30 16 18 12 2 18 23 1489 70 44 54 30 37 37 44 54 71 15435 21 7 1 1 6 9 17 25 26 16Note: Calculations do not include DSRA amount of US$2.5 million.Source: BRAC data.Table 3.7b Performance of dierent credit enhancements for Tranche 2: % excess cash ow of SPV liability returning to residual beneciary under dierent scenariosOver-collat., No Cr. Enhance.75 64 47 32 16 61 48 29 15 1 65 73 70 63 52Payment MonthYearAll Credit EnhancementsOver-collat., w/Prepay. Trap.Over-collat., w/Subst.No Over-collat., w/No Cr. Enhance.17 9 2 12 23653 4 5 6 72007 2007 2007 2007 200770 84 87 84 76Note: Calculations do not include DSRA amount of US$2.5 million.Source: BRAC data.Table 3.8a Performance of dierent credit enhancements for Tranche 1: % of future SPV liability for Tranche 1 covered by amount collectible from securitized pool and cash trapped in prepayment accountsOver-collat., No Cr. Enhance. Over-collat., w/Prepay. Trap. Over-collat., w/Subst. No Over-collat., w/No Cr. Enhance.Payment MonthYearAll Credit Enhancements669 10 11 12 1 2 3 4 5 6 72006 2006 2006 2006 2007 2007 2007 2007 2007 2007 2007157 157 166 172 184 199 222 286 413 762 1823152 147 148 148 148 150 156 152 185 266 566155 153 158 165 181 202 243 306 501 1092 3320156 154 161 165 173 182 199 251 353 633 1517101 98 99 99 98 100 104 101 123 177 378Note: Calculations do not include DSRA amount of US$2.5 million.Source: BRAC data.Table 3.8b Performance of dierent credit enhancements for Tranche 2: % of future SPV liability for Tranche 2 covered by amount collectible from securitized pool and cash trapped in prepayment accountsOver-collat., No Cr. Enhance. 155 137 134 126 127 157 142 144 142 152 159 167 172 182 198 Over-collat., w/Prepay. Trap. Over-collat., w/Subst. No Over-collat., w/No Cr. Enhance. 103 91 89 84 84Payment MonthYearAll Credit Enhancements673 4 5 6 72007 2007 2007 2007 2007161 171 178 190 209Note: Calculations do not include DSRA amount of US$2.5 million.Source: BRAC data.68Micronancefew months. The trapping and controlled release of prepayments would have allowed the securitized pool and trapped cash to cover future SPV liabilities more comfortably, particularly after the fth month of the transaction. Substitution and replacement of delinquent loans have a similar impact. However, the SPV still would not have been able to cover its liabilities between the second and sixth month. With over-collateralization but no other credit enhancements, Tranche 1 would have had some excess cash ow in all months, and the SPV would also have been able to cover at least around 150 percent of its future liabilities in any month. While the excess cash ow would have fallen to 15 percent in the last few months of Tranche 1, the structure could easily have been adjusted to accelerate payments in the earlier months where excess cash ow was higher. With over-collateralization and substitution (but no trapping of prepayments), there would have been a slight decrease in excess liquidity, which would still hover around 30 percent in the fth to seventh months of the transaction. Over-collateralization without substitution but with the trapping and controlled release of prepayments would have meant less SPV liquidity throughout the life of Tranche 1. In fact, the structure with the current amortization schedule would have been unable to pay its obligations in later months. However, the future liability covered by the pool and the trapped prepayments would increase signicantly in later months. The trapped prepayments could be used to extend the maturity of the bond to beyond 12 months. A viable structure could have been created with accelerated payments in the early months, lower months in the later months (where cash ows would have been negative) and the release of prepayments to pay bond obligations beyond 12 months. The Tranche 2 experience is consistent with that of Tranche 1.The transaction till date has not required the replenishment of the pool with new collateral, as excess cash ow has never dipped below 40 percent. We have thus not analyzed the impact of replenishment upon the robustness of the structure. Clearly, much of the robustness of the structure can be attributed to over-collateralization. The process of trapping prepayments and releasing them upon maturity adds to the comfort that the SPV shall be able to cover future obligations. However, even without the trapping of prepayments, at least 150 percent of future obligations are covered by the receivables from over-collateralized pool at any point in time for Tranche 1. Substitution has a larger impact on SPV liquidity than the trapping of prepayments, but is not crucial for a viable structure. These results suggest under normal circumstances, over-collateralization may be enough as a credit enhancement.Securitization and micro-credit backed securities69LOOKING FORWARDServicer Risk We have discussed many of the informational and logistical risks that were identied and addressed by the structure. However, a major risk noted by the investors was servicer risk. In the event of BRAC ceasing to exist, it would take a few months for another institution to step in and restart collections from BRACs borrowers. A few factors mitigate some of the servicer risk. PriceWaterhouseCoopers (PWC) ran a systems analysis of BRACs MIS. PWC noted that BRACs systems were quite robust. In particular, they noted the presence of data back-ups and system redundancies at both the head oce and area oce levels. They also independently checked the accuracy of the data. Further, the BRAC database includes the name and location of the individual borrower. Thus, in case the servicer needs to be replaced, the new servicer will be able to identify and locate the borrowers whose loan has been securitized. Yet, it is unavoidable that a new servicer will likely take some time to send its own collection ocers. Additionally, while all funds collected from the securitized pool can be identied in the current BRAC system, the 42-day time lag for information reaching the head oce means that there is always some commingling of funds in BRACs accounts. In the event of BRAC going bankrupt, there is some risk of not being able to retrieve all the collected funds for the immediately preceding month. However, the presence of a DSRA fund may mitigate some of these risks. Moving Beyond the BRAC Micro Credit Securitization Series I We have already discussed how information on borrower characteristics and households was limited. This constrained the risk analyses that could be done. Further, historical data were unavailable. In future transactions, more will be known about the risk prole of BRACs borrowers from its existing databases. But, as noted, BRAC is not tapping into rich sources of information that are readily available to it. BRACs health care and education programs collect detailed information on BRACs group members and their households. However, currently these databases cannot be linked to the microcredit database. Further, BRAC also collects information on the savings of the borrowers. These data were unavailable to us, and can yield valuable information about the risk proles of BRACs borrowers. As more is learnt about micro-credit borrowers, investors also have to be more exible in their approach to conditions they place on future securitizations. The 50 percent over-collateralization, for example, was dictated by70Micronancethe investors at the start of the transaction without much analysis. The 40 percent excess cash ow requirement and DSRA amount were also a function of some investors (and regulators) risk aversion and the need to guarantee an AAA rating in this rst-of-a-kind transaction. In future transactions, investors should allow analyses of the risk prole of the borrowers to guide optimal structures. Such analyses may also create the possibility of tiering risks to meet the appetites of dierent investors. The sharing of information about borrowers among dierent micronance institutions may also help these institutions better understand the risk proles of their own borrowers. It may also allow smaller institutions to pool their portfolios to achieve the necessary scale for accessing funds through securitizations. Sharing of informationparticularly in a country like Bangladesh with no national ID cards, let alone credit rating reports may help mitigate residual servicer risk. We have noted how BRACs own MIS in the area and head oces is quite robust, particularly given the dicult operational conditions. But the 42-day lag between actual collections and information reaching the head oce is problematic. BRAC has made tremendous strides in getting the information to the head oce quicker. But more frequent updates would reduce the risk of investors in future transactions, while also allowing BRAC to improve its own cash ow management. BRAC is considering the possibility of linking a few regional oces to the head oce through the Internet, and transferring information to these regional oces from the area oce weekly. It may also be possible to leverage mobile technology for live updates on collections and disbursements. During the course of the transaction, the very fact that BRAC had to expose and explain its processes and systems to investors, auditors, and a credit rating agency helped advance transparency and accountability at BRAC. Over the course of the transaction, issues spotted with BRACs systems or data were rectied as far as possible. BRAC has also made tremendous strides in the last year or so in strengthening its reporting systems. We predicted at the start of this transaction in 2004 that BRAC would mature as a result of this securitization. We were pleasantly surprised when this prediction was conrmed. A securitization such as this is really a learning process for the originator, investors, and regulators. It contains lessons for future transactions at many dierent levels, from political economy considerations to risks identied. The structure that was created for BRAC Micro Credit Securitization Series I incorporated many of these lessons. The structure has proved to be robust in the rst six months of Tranche 1. As the transaction moves forward into its new tranches, we hope to learn more lessons about what works in the current structure for replications of this transaction and the creation of new MCBS structures elsewhere.4. Cell phones for delivering micro-loansAnand ShrivastavI will give you a talisman. Whenever you are in doubt, or when the self becomes too much with you, apply the following test. Recall the face of the poorest and the weakest man whom you may have seen, and ask yourself, if the step you contemplate is going to be of any use to him. Will he gain anything by it? Will it restore him to a control over his own life and destiny? In other words, will it lead to Swaraj [freedom] for the hungry and spiritually starving millions? Then you will nd your doubts and your self melt away. (Mahatma Gandhi)1INTRODUCTION AND BACKGROUNDCutting-edge technology can provide a transformational role in delivering nancial, health, and educational services to the poor. This chapter focuses on India and the potential for providing nancial services to the poor through mobile phone technology. Such a delivery has the potential to cut signicantly the transactions costs of accessing small loans and this can lead to an improvement in the overall welfare of the poorer sections of society who are currently unable to access nancial services from the organized banking sector and credit markets. To provide the reader with a backdrop, it is rst useful to review some background data on India. Demography and Economy2 India is a large country with an area of 3.3 million square kilometers. The population of India has exceeded 1.1 billion, and is showing a year-onyear growth rate of about 2 percent. Of immediate interest to the subject matter addressed in this chapter is the fact that India has a fairly high savings rate (200405) of 29 percent, and a growing literacy rate. Moreover, there is a large pool of professionals, and the youth population is signicant as attested by the fact that more than 50 percent of the population is under 25 years old. An important fact that has been recognized by the policymakers is that there is signicant unemployment, especially in the age group 2024 in rural and urban India.3 The unemployment rate of educated7172Micronancewomen in rural and urban India is very high as well. Increased availability of credit and other nancial services is likely to be an important factor in alleviating unemployment, and in empowering women. Mobile Telecom4 Mobile telecom has already reached where other sectors are yet to reach. Telecom statistics and growth projections are spectacular: mobile subscribers have a cumulative annual growth rate of 86 percent and were expected to reach 143 million by November 2006. The Department of TelecomDOTputs forward a target gure of 500 million mobile connections and mobile access to every village over 1000 population by 2010. Despite this remarkable growth, the mobile penetration is only 10 percent and is one of the lowest in the world although, even with the 400 minutes/subscriber/month usage, which is close to the United States gure, Indian telecom is well placed in the world markets. Looking at the future, potential projections vary. However, even a conservative estimate sees 300 million subscribers by 2010. Considering that the mobile penetration is lower than many other comparable economies, the growth rate of mobile phones in India is likely to remain very high for the foreseeable future. Micronance5 The RBI (Reserve Bank of India) Internal Groups Report on Micro Finance has noted that the outreach of the Indian banking system has seen rapid growth in rural areas. Forty-eight percent of the total branches of the scheduled commercial banks (SCBs) and regional rural banks (RRBs) cater to the rural areas (32 303 branches translates to a population of about 23 000 people per branch). Of these, 31 percent (136.7 million) of deposit accounts and 43 percent (25.50 million) of borrower accounts are in the rural areas. This expansion of the organized nancial infrastructure has reduced the dependence of the rural population on the unorganized moneylending sector from 68.3 percent in 1971 to 36 percent in 1991. In spite of this growth, there continues to be wide gaps in the availability of banking services in the rural areas as the SCBs cover only 18.4 percent of the rural population through savings/deposit accounts and even a lower percentage of 17.2 percent of the rural households by way of loan accounts. Though the primary agriculture credit societies (PACS), with about 100 000 outlets, have a deep and wide presence in rural India, their impact in terms of extension of deposit and credit products has not only been minimal but concentrated in a few states only.Cell phones for delivering micro-loans73The decline in productivity of the rural branches of the commercial banks, fragility of the cooperative credit structure, and weakness of RRBs witnessed since the early 1990s has further accentuated the problem of inaccessibility of banking services for a large part of the rural population. Furthermore, as the banking sector has shown, a propensity towards the larger-size accounts has meant that the number of loan accounts of small borrowers with a credit limit range of less than Rs 25 000 decreased from 58.8 million in 1991 to 36.9 million in 2003. A vast majority of the rural population remains unintegrated with the organized banking sector. On analyzing the supply side, we observe some of the reasons for this, such as: (1) people are unbankable in the evaluation/perception of bankers, (2) the loan amount is too small to invite the attention of the bankers, (3) the person is bankable on a credit appraisal approach but distances are too far for servicing and supporting the accounts, and expanding branch network is not feasible and viable, (4) there are high transactions costs, particularly in dealing with a large number of small accounts, (5) there is a lack of collateral security, (6) an inability to evaluate and monitor cash ow cycles and repayment capacities due to information asymmetry, lack of data base, and absence of credit history of people with small means, (7) there are human resourcesrelated constraints both in terms of inadequacy of personnel and lack of proper orientation/expertise, (8) there is an adverse security situation prevailing in some parts of rural India, (9) a lack of banking habits and credit culture, (10) information-shadow geographical areas (geographical areas where information on identity and residential proof of persons is yet to be completed, which makes know your customer (KYC) compliance dicult for banks; hence the population in these areas remains unbanked), and (11) an inadequacy of extension services, which are crucial to improving the production eciency of the farmers, leading to better loan repayments. Similarly on the demand side, there are several reasons for the rural poor remaining excluded from the formal banking sector, such as: (1) high transactions costs at the client level due to expenses such as travel costs, wage losses, incidental expenses, (2) lack of awareness, (3) lack of social capital, (4) non-availability of ideal products, (5) very small volumes/size of transactions, which are not encouraged by formal banking institutions, (6) hassles related to understanding documentation and procedures in the formal system, (7) easy availability of timely and doorstep services from money lenders/informal sources and (8) prior experience of rejection by indierence of the formal banking system. Under the micronance program, loans are extended to the self-help groups (SHGs) who pool a part of their income into a common fund from74Micronancewhich they can borrow. The members of the group decide on the minimum amount of deposit, which ranges from Rs 20100 per month depending upon the size of the group. The group funds are deposited with a micronance institution (MFI) against which they usually lend and the deposits are usually placed with a bank by the MFI. The group funds are the way micro-savings are enforced, though it may seem like collateral. The loan ticket sizes are usually Rs 200015 000. Although loan repayment is a joint liability of the group, in reality individual liability is emphasized. Maintaining group reputation leads to the application of tremendous peer pressure.6 In India and other Asian countries the majority of SHGs typically consist of women because, in these countries, self-employment through micronance was perceived as a powerful tool for the emancipation of women. A World Bank report (2001see note 6) observes that gender equality is a necessary condition for economic development. It reports that societies that discriminate on the basis of gender are in greater poverty, have slower economic growth, weaker governance, and lower living standards. And the results are encouraging. Loans obtained from MFIs are utilized in agriculture and small businesses. Independent incomes and modest savings have made women self-condent and helped them to ght poverty and exploitation. This can be seen from a statement by a woman beneciary: Previously we had to cringe before our husbands to ask for one rupee. We do not have to wear tattered sarees anymore and, today, we have the condence to come and talk to you without seeking permission from our husbands (as told to the author of the UNPANs Field Survey [note 6]). Suvidha7 proposes to play a signicant role in micronance using the SWIFT mobile transaction platform and providing Beam (see The Product below) services, leveraging its distribution network and customer proles. Banks have the opportunity to partner with Suvidha to supplement their extension banking services too. Likewise, MFIs, RRBs, SCBs, NBFCs (non-banking nance companies), and banks can also partner Suvidha to extend their product deliveries and manage both inbound as well as outbound payments through the network of Beam Mobile Entrepreneurs. Payment System Ineciencies The GSM Association (GSMA) launched a pilot program in January 2006, aimed at tapping the ubiquity and ease-of-use of mobile technology to enable the worlds 200 million international migrant workers to send remittances easily and securely to their dependants, many of whom did not have bank accounts.8 By exploiting the extensive reach of the mobile networks,Cell phones for delivering micro-loans75Table 4.1 Payment realization for various payment modes in IndiaNature of Financial Transaction Cheques Location of Transaction Local (same bank) Outstation Local (same bank) Outstation (other bank) Electronic Credit card Debit card Electronic clearing system Postal money orderSource: Data collected by author.Time to Complete Transaction Three days 1030 days plus seven days by post. One to two days, plus seven days by post Three days, plus seven days by post Seven days (merchant payments) Seven days (merchant payments) Four days (limited coverage) 1045 daysDemand draftthe program complemented existing local remittances channels and made transferring money internationally signicantly more aordable. In India, the situation of payment realization for dierent modes is as shown in Table 4.1. It can be seen from Table 4.1 that the present payment systems in India are still quite inecient and can contribute to lower than optimal rates of economic growth.GLOBAL TRENDSAs per a report by IFC Washington-GSMA, the advantage of developing a market for micro-payments (also referred to as m-commerce), is that it continues to drive the economic system toward a cashless transaction environment.9 Elimination or minimization of physical cash has many advantages, including less opportunity for fraudulent or criminal activity, reduction of cash handling costs, and, for the user, less reliance on having the right amount of cash when needed. It also allows the value of money to be better utilized. Cash held outside the banking system is not available for short-term investment so that the time-value of the cash asset is lost.76MicronanceIn the more auent economies, there is already a good infrastructure for a cashless environment, with most people having bank accounts and an array of both debit and credit cards. Nevertheless there is an underlying need for cash for minor purchases and there is little incentive to eliminate cash entirely. These economies can manage quite well and there is no specic interest group that feels suciently under pressure to develop systems aimed at eliminating cash from the environment. Systems that have been developed in such markets are often expensive and hence not particularly attractive to the customer. In the developing economies however, there is a very large underclass that is totally reliant on cash for all their day-to-day expenses. Moreover, this underclass makes no use of the banking sector and so is invisible in terms of its cash value. At the same time, the need for cash forces the providers of goods and services in these markets to have adequate cashhandling facilities and this comes at some cost. In these cases, the commercial organizations have much more to gain by addressing the problem of cash transactions. Not only is the risk associated with cash holdings much greater, but the time-value of the cash being held outside the banking sector is entirely lost. Furthermore, the population in this category is lost, that is, unseen by the banking sector. For these reasons, there is likely to be more incentive in developing economies to move the population at large away from cash, than exists in developed economies. That being so, a solution that meets the needs of developing economies will also have extensive application in the developed economies. This arises because the solution must be accompanied by very low costs as, if it were otherwise, the solution would have no appeal in those developing economies. The resulting lowcost solutions can then be applied in the developed economies, resulting in further eciency gains. Further, according to the IFC report, the most successful micro-payment applications are to be found in the Philippines, with over 3.5 million m-commerce users on two mobile networks. The key success factors for that market included the ability to load prepaid airtime credits as well as the ability to transfer both cash and airtime credits between customers. Coupled with these were the low values set by the operator for such prepaid top-ups or credit transfers. Typical top-ups of 4757 (US) cents were allowed by the networks (equivalent to around four to ve minutes of calls) while transfers between customers of both cash and airtime credits were permitted as low as 4 cents. The target market surveyed by IFC is attuned to sachet purchasing or the practice of purchasing goods in very small quantities packed in sachets. This phenomenon is known to be common in other developing markets where the populace rely on cash for all trading and can aord to buyCell phones for delivering micro-loans77provisions for just a few days consumption. This market does not exhibit bulk purchase tendencies and m-commerce that involves a signicant cash deposit or payment will be unlikely to nd any signicant uptake from the target market. While the application of m-commerce to developing markets was not constrained to the Philippines, African (South Africa and Kenya) market developments seemed to reect the Filipino views, indicating that the target markets in these geographically diverse areas were very similar in their use of cash and their expectations. The range of features available in each market showed signicant uniformity as to be expected if the target markets were similar. With minor variations, the features of all systems included: provision for cash deposits and withdrawals; the ability for third parties to make deposits into a user account (employer, family member, or MFI); the ability to make retail purchases at selected outlets; over-the-air prepaid top-ups using the cash already in the account; the ability to transfer cash between users accounts; the ability to transfer airtime credits between users; provision for bill payments.These features could be used for micronance applications involving both loan repayments as well as loan advances, and this area in particular is being exploited in the Kenya trials and in Philippines services in conjunction with the Rural Bankers Association. Apart from the use of the services by MFIs all services studied by IFC operated on a debit account basis, that is, the account could only be operated in credit. As a result, bad debt is not an issue other than loss caused by fraudulent activity. No operator indicated any serious concerns in this area and provided the overall system security was ensured the possibility for fraud could be managed. In that regard, most of the systems studied involved a bank with normal banking systems in place as this arrangement results in the fraud issue being restricted to the banks area of involvement for which it will be well-equipped. The possibility of money laundering was considered by all service providers and it was noted that in all jurisdictions the banking regulatory authority had established appropriate policies governing the activities of the banks. These policies included monitoring transaction levels and frequency, looking for transaction patterns and stipulating both maximum account balances and daily transaction levels. While it is possible for a network operator to take almost full responsibility for the entire78MicronanceTable 4.2 Increase in SMS trac in Asia Pacic, 200410SMS Trac (billions) 2004 434.1 2005 540.1 2006 672.8 2007 802.4 2008 935.9 2009 1072.1 2010 1212.7Source: Portio Research, accessed at www.adnetasia.com/news/communications/0,39044192,39252956,00.micro-payment service, only one service in the Philippines was operating in that manner. Even then, the actual cash oat generated was held in one of the countrys regular banks. All other cases studied involved cooperative arrangements between banks and networks. In view of the regulatory issues surrounding the banking industry, this method of operation is more likely to appeal to intending service providers, given that the banks can bring additional advantages including the availability of debit cards through issuers such as MasterCard. While there were no quantiable gures available on system costs, various estimates placed a likely cost in the range of US$5 million to US$10 million with an expectation that an m-commerce system could be protable with as few as 25 000 users connected but that would depend on the overall investment and service operating costs. Various estimates placed the transaction level at around two per customer per day and average transaction values at between US$15 and US$30 per customer and airtime top-ups of around US$4 per time. These gures can only be regarded as indicative of the type of activity that may be encountered. According to a study by Portio Research, SMS (short message service) trac in the Asia Pacic (AP) region is expected to increase from 434 billion messages in 2004 to over 1.2 trillion by 2010.10 (See Table 4.2.) One of the key growth drivers is attributed to higher future mobile phone penetration in India and China. The report projected that the combined markets of India and China would account for over 800 billion SMS messages per annum by 2010. The study predicts that India, with a low mobile penetration rate has massive potential. SMS in India is expected to grow from 12.3 billion messages in 2004, to 180 billion in 2010. In summary, the situation provides an excellent and growing opportunity to Suvidha and its SWIFT technology platform, which uses SMS to perform both inbound as well as outbound transactions. Suvidhas network of Beam Mobile Entrepreneur women become self-employed by providing the front end services at the localities where they live, and thus gain better control over their lives.Cell phones for delivering micro-loans79Table 4.3 Domestic transaction companies operating in IndiaCompany Paymate JiGrahak MChek ITZ Cash Done Card Wallet 365 Fino A Little World Website www.paymate.co.in/web/Default.aspx www.jigrahak.com/site www.mchek.com/ www.itzcash.com/ https://www.donecard.com/index-1.aspx www.timesofmoney.com/tomsvc/jsp/home.jsp www.no.co.in/ www.alittleworld.com/Source: Gathered from market intelligence.THE INDUSTRYIndia currently has around eight domestic companies operating in the transaction space as shown in Table 4.3. Paymate is SMS-based but exclusively for use by customers having accounts with banks, for example, Citibank. JiGrahak requires download of software to a mobile handset and is thus limited to certain types of handsets. MChek works on GSM technology, as it is USSD (unstructured supplementary source data) technology-based solution and provides services to the subscribers of Airtel and some banks. ITZ Cash and Done Card are prepaid card-based solutions that work on the Internet. Wallet 365 is also an Internet-based service for banked customers. Fino is a closed-user proprietary technology service provider for MFIs. Last but not least, the technology focus of A Little World is on biometrics-based ID, RFID (radio frequency identication) smart cards (Java, PKI) and NFC (near eld communication) mobile phones as acceptance and enabling devices (with merchants, eld forces of MFIs, and at cashless ATMs). While ITZ Cash, Done Card, Paymate, and JiGrahak market themselves as m-commerce, their services can only be used if the subscriber has Internet accessmicronance is not their focus. Further, none are interoperable or neutral as they are exclusively tied either to a specic bank (Paymate, MChek, A little World) or telecom (MChek, JiGrahak) or can be used by special types of phones (A Little World, JiGrahak, MChek via GSM), thus excluding a signicant number of CDMA (code division multiple access) subscribers. Except for A Little World, none are focused towards micronance. Fino, on the other hand, acts only as an application service provider (ASP) for micronance institutions requiring core banking application and80Micronancepoint-of-sale devices. The parentage of Fino is ICICI Bank, hence possibilities of interoperability with other banks will be a challenge. Suvidha-Beam is focused on the micro-payments for the unbanked and enabling micronance oered by MFI/banks. Beam does not require the Internet, or software to download, change of SIM, or require special equipment. It is interoperable with subscribers of any telecom.THE PRODUCTBeam is an innovative and simple way of transacting money using mobile phones. It takes advantage and plugs the ineciencies in the payment systems of the economy. It enables subscribers to register and use a host of other services, anytime, anywhere, using short message service (SMS). The product has two parts. A robust, future-ready technology platform called SWIFT is at the backend; and a stored value prepaid card that consumers purchase for using services, called Beam. Backend Technology The backend SWIFT (subscriber wireless interaccount nancial transaction system) is a mobile commerce platform that took several years to develop. It leverages the cumulative knowledge and experience gained by Suvidha from product distribution and providing transaction management services to banks as well as telecom services. SWIFT is a sophisticated, robust, secure, and scalable application having disaster management and business continuity system too. It lets subscribers use the Beam services via SMS, IVRS (interactive voice response system) or the Internet. Services Beam as a service allows mobile phone subscribers to send money, give gifts, pay each other, make purchases from merchants besides a host of other servicestake credit, make deposits, obtain insurance, make investments, all using their mobile phone. Services can be accessed as soon as a mobile phone customer registers with Beam. This can be done by sending a simple SMS message to Beam and assigning him or herself a secure personal identication number (SPIN). The subscribers Beam account is established automatically by SWIFT and the subscriber receives an SMS to this eect within a few seconds. Beam prepaid cards can be purchased from any retailer, Beam Merchant, Beam Express (shops) franchisee or Beam Mobile Entrepreneurs (individ-Cell phones for delivering micro-loans81uals). Additionally the Beam prepaid cards can also be purchased from Suvidhas alternative channels comprising SCBs, cooperative banks, RRBs, NBFCs, MFIs, and India Post. Subscribers not having a bank account can purchase Beam prepaid cards to top up their account and perform a variety of transactions. Money can be gifted via Beam to another subscriber. A Beam Merchant can be paid by a subscriber using Beam. Similarly, refund of the residual amount in the subscribers Beam account can also be taken from any Beam Mobile Entrepreneur or Express franchisee. Besides micro-payment services, Beam Mobile Entrepreneurs can also extend micronance, micro-insurance, micro-investment as well as international money transfer services of Suvidha partners to the Beam subscribers. Additionally, the Beam Mobile Entrepreneurs can act as service delivery vehicles and extend micronance, micro-insurance, micro-investment, as well as international money transfer services in his or her locality. These will be to the customers located anywhere who may not have a mobile phone and/or may not be registered with Beam but are clients of banks, cooperative banks, RRBs, NBFCs, India Post, MFIs, SHGs, and Ladies Kitty Clubs (LKCs). Transaction Ecology Figure 4.1 shows the human ecology of the various types of transactions. Subscribers can be seen sending money, giving gifts, paying each other, making purchases from member Beam Merchants and also taking refunds from Beam Mobile Entrepreneurs using their mobile phones. Similarly, Beam Mobile Entrepreneurs can be seen providing refund services and also extending micronance, micro-insurance, micro-investment, and international money transfer services to Beam subscribers, as well as to the customers of banks, SCBs, RRBs, NBFCs, MFIs, SHGs, India Post and LKCs, who may have mobile phones and/or are not registered with Beam.THE FUTURE AND CHALLENGESSuvidha is starting o with micro-payment services. As it moves forward, it will use the exibility and scalability of SWIFT technology platform, leverage the distribution network and the proles of Beam subscribers to oer micronance products, micro-insurance (life and general insurance products like crop insurance and so on) of its partners. As the feeton-street, that is, Beam Mobile Entrepreneurs mature, it will oer the82MicronanceIntl money transfer InsuranceCREDIT - F2SBlankMutual fundMember Mobile entrepreneur franchisees merchant PAY - S2M REFUND-S2FMobile entrepreneur DEPOSIT - S2F franchisees INSURANCE - S2FINVESTMENT - S2FLadies-kitty Clubs Self-help groupsINTL MONEY XFR - F2SSubscriberGIFT - S2SSubscriberNote: OS2M = Subscriber to Merchant; S2F = Subscriber to Franchisee; S2S = Subscriber to Subscriber; F2S = Franchisee to Subscriber. Source: Anand Shrivastav.Figure 4.1 The ecology of Beam transactions micro-investment products of partners. Suvidha will also oer international money transfer services of partner money transfer organizations (MTOs) through the Beam Mobile Entrepreneurs and Beam Express franchisees. In addition to oering partner products, Suvidha will provide micro-payments transaction management services to customers of banks, SCBs, RRBs, NBFCs, micronance, micro-insurance, micro-investment, and MTO companies. It will move to other countries at an appropriate stage. Challenges1.2.3. 4.Regulation. The regulatory environment for payments, micronance, micro-insurance, micro-investment, and international money transfer is still evolving and there are no clear guidelines. Taxation. While service tax is well understood, VAT is administered by individual states who are not clear on what the treatment should be with regard to charging or not. Anti-money-laundering (AML). Customers may not yet have been issued the required documents with regard to anti-money-laundering. Combating the nancing of terrorism (CFT). Here again no clear dissemination of information has occurred at the enforcement level.Cell phones for delivering micro-loans83Micro-payments Microfinance Micro-insurance Micro-investment International money transfer Source: Anand Shrivastav.Figure 4.2 Services to Beam subscribers Nevertheless, Suvidha believes its services will not only improve the lives of its customers and the Beam Mobile Entrepreneurs, but will enable them have a greater control over their life and destiny, and in this manner make its humble contribution to the economic prosperity of India.NOTES1. 2. 3. 4. 5. Mahatma Gandhi (1958), Last Phase, vol. II, p. 65. From a note left behind in 1948. Census of India. Government of India-MOSPI (2004), Socio-economic dimension of unemployment in India; Government of India-MOSPI (2005) Employment and unemployment situation in Cities and Towns in India 200405 Part 1. Sources include Telecom Regulatory Authority of IndiaTRAI June 2006; Government of India Department of Telecommunication (2006), Telecom Vision 2010 and Merrill Lynch (2006), Global Mobile Matrix 2006. Sources include RBI (2005), Internal Group Report of the RBI on issues relating to rural credit and micro-nance 2005; RBI Banking Statistics 2003; National Sample Survey Organization (NSSO) (1999), All India debt and investment survey 1991; National Federation of State Cooperative Banks. Sources include UNPAN/Ghosh, R. (???) Field Survey; Sinha, F. and team (2003), Impact assessment of micronance in India, EDA Rural Systems Pvt Ltd, World Bank (2001), Engendering development through gender equality in rights-resources and voice; Sampark (2003), Mid-term impact assessment study of CASHE project in Orissa. Suvidha (Sue vee dhaa) is a Sanskrit word that means convenience. Suvidha is a mobile transaction service provider and is ISO 9001 certied. It was incorporated on 116.7.84MicronanceDecember 2002 and is headquartered in New Delhi. Its global transactions management system (GTMS) is used by corporate banking for cash, tax management, and cooperative payment management services via the Internet. Some of the banks using GTMS services are HSBC, Deutsche Bank, ICICI Bank, HDFC Bank, and UTI Bank. Suvidha is about to launch the Beam services in FY 200708, with SWIFT as its backend. GSMA (2007), Global money transfer uses pilot mobile to benet migrant workers and the unbanked, press release, www.gsmworld.com/news/press_2007/press 07_14.shtml. IFC Washington-GSMA-infoDev (2006), Micro-payment systems, infoDev report, www.infodev.org. ZDNet-Portio Research (2005), SMS trac to double in AP by 2010, September, www.zdnetasia.com.8. 9. 10.5. How should governments regulate micronance?Richard Rosenberg1INTRODUCTIONPowerful new micronance techniques are being developed that allow formal nancial services to be delivered to low-income clients who have long had no access to such services. But the micronance industry will not reach its full potential unless many of its service providers can eventually enter the arena of licensed, prudentially supervised nancial intermediaries. Regulations must eventually be crafted that allow this to happen. Dozens of developing and transition country governments are now at earlier or later stages of addressing this challenge. Many dierent actors are pushing for regulatory adjustments, from micronance institutions themselves (MFIs), to international development agencies, to government ocials who want to democratize nance or protect against perceived risks for the nancial system (or perhaps clamp down on annoying non-governmental organizationsNGOs). The interests and objectives of these actors diverge considerably. Thorny technical and practical issues are involved. We do not have decades of experience with regulated micronance to guide usmost of the countries with micronance regulations have only a few years of experience with implementing them. And country-specic circumstances loom large, so there can be no standard model for micronance regulation. Nevertheless, among people working on these topics there are surprisingly wide areas of consensus on some general principles that should bear on regulatory design for micronance. The author is condent that most of the material in this chapter is consistent with the views of most of the technical advisors who have multi-country experience and who do not represent the interests of any particular combatant in the fray. The discussion will begin with an important denitional distinction between prudential and non-prudential regulation. Non-prudential regulation will be treated in the third section, prudential regulation in the8586Micronancefourth, and the challenges of prudential supervision in the fth. The sixth section concludes.PREAMBLE: PRUDENTIAL AND NON-PRUDENTIAL REGULATIONGovernments regulate the behavior of all businesses. Such regulation may be aimed at protecting consumers, or employee safety, or the environment, but it usually does not try to protect the nancial health of the business that concern is generally left to the owners, at least where the owners are private. But in almost every country in the world, banks are treated dierently. Governments impose an elaborate regime of prudential regulation whose aim is to protect the solvency of banks. Various reasons are advanced for this. The main one is that nancial systems depend critically on condence, so that the failure of one bank can hurt many other banks and provoke a systemic crisis, with dire eects for the economy at large. Another reason is that banks are nanced predominantly by money of people other than the shareholders, which creates incentives for bank managers to take imprudent risks: when the gamble succeeds the bank and its shareholders capture the gain, but when the gamble fails, others especially depositorsmay bear a large part of the loss. Finally, governments believe that small, unsophisticated depositors need protection because they are in no position to appraise the riskiness of a bank on their own. Examples of prudential requirements include capital adequacy rules (how much of other peoples money a bank can use), restrictions on risky uncollateralized lending, limits on insider lending, or requiring maintenance of reserves for loans that are likely to go bad. Non-prudential regulation is an inelegant name for all the other banking rulesthe ones that dont involve the government in assessing and protecting the nancial health of banks. Such rules are sometimes referred to as conduct of business regulation. Examples include limits on interest rates charged to borrowers, other consumer protection like truth-in-lending laws, or anti-money-laundering rules that require screening and reporting of customers. Securities regulations are another non-prudential example: these rules usually require banks, just like other rms, to disclose all material information about their business to potential investors. The rules have been complied with when all the information about the banks business is disclosed. Investors are left to fend for themselves when it comes to weighing their risks. Implementing prudential regulation, where the government in some sense is vouching for the nancial soundness of each licensed bank takingHow should governments regulate micronance?87deposits, tends to be much more complex, dicult, expensive, and intrusive than implementing non-prudential regulation. Enforcing prudential regulation always requires a specialized banking authority, whereas many nonprudential regulations apply to non-deposit-taking rms as well, and might not necessarily require a specialized banking supervisor to enforce them. The reason for emphasizing this distinction is that in a number of countries, governments are applying burdensome prudential rules to nondeposit-taking MFIs whose failure would cause neither loss of depositors funds nor material disruption of the national nancial system. Prudential regulation has high costs not only for the supervisory authority but also for the supervised institution, which will eventually pass these costs along to its customers. Prudential regulation of credit-only MFIs uses a cannona very expensive cannonwhere a rie would be more than adequate in view of the risks involved. It is especially important to focus on the implications of regulation for the administrative costs of MFIs. In the centuries-old eort to improve nancial access for poor and low-income people, the critical factor is cost, more than the motivations of nancial service providers. Major, long-lasting improvements in access are usually associated with new ways to lower costs. Concern for the poor has played an important part in the micronance revolution of the last three decades. But concern for the poor has been around for a long time. The revolution became possible when Grameen Bank and other pioneers in Indonesia and Latin America discovered less costly ways to deliver and collect tiny uncollateralized loans, and mobilize and manage small savings. Some of the regulatory requirements discussed in this chapter have signicant cost implications for micronance providers. Decisions about such practices need to be handled carefully.NON-PRUDENTIAL ISSUESUsury Limits Lending a million (US) dollars in 10 000 loans of $100 each entails administrative costs that are hugely greater than the cost of lending out the same amount in one or two big loans. As a result, it is usually impossible to do micro-lending on a nancially sustainable basis without charging interest rates that are very substantially higher than what banks charge to larger borrowers. In 2005, the median annual interest rate collected by the hundreds of MFIs reporting to the MIX Market (www.themix.org) was about 31 percent. Rates above 50 percent are not uncommon, and a few MFIs charge more than 80 or even 100 percent. In most cases, these rates reect88Micronancenot high prots but high costs of micro-lending: the smaller the loan size, the higher the administrative costs are for lending a given amount. But this analysis of lending costs is ne print that is usually too small to show up on the screens of politicians or the general public. Charging poor borrowers 30 percent when fat cats pay only 10 or 15 percent shocks most consciences. Not all micronance interest rates can be explained by the costs of lending. In a recent well-publicized instance, the Mexican MFI Compartamos was charging interest of about 100 percent a year and producing annual prots that gave its shareholders more than a 50 percent return on equity. This cause clbre, despite being a highly exceptional case, has fanned the ames of a growing backlash against high micro-credit interest rates, a backlash that has already been underway in Latin America and the rest of the world for several years now.2 Limits on interest rates can hurt rather than help low-income borrowers if the interest cap is set too low for certain types of lending to be protable: providers will withdraw from the business and potential borrowers will lose access to services. In theory, an interest rate cap would avoid this result if it were set at just the right level. As a practical matter, however, nding the right level is hard, not least of all because loan products, clienteles, and costs vary. Moreover, it is politically dicult for governments to set interest caps at appropriate levels: a reasonable interest rate for tiny, high-cost micro-loans will inevitably seem exploitative to most people, because they do not understand the reasons for the high rates. Consumer Protection and Borrowers Rights When governments are concerned about micronance interest rates that sound abusive, they are sometimes advised to avoid interest rate caps and focus instead on other borrower protection issues such as truth-in-lending (disclosure of the full cost of loans in a format that makes it easy to compare rates oered by various lenders) or prohibition of certain unacceptable lending and collection practices. Most MFIs today do not quote their loan charges in the form of an annualized eective interest rate that includes all costs. In some cases there may be a legitimate concern that explicit quotation of rates this way would lead to a political backlash, resulting in interest rate caps that would make it impossible to continue serving their clients. In most cases, though, microlenders opposition to truth-in-lending policies and requirements probably stems mainly from normal, less noble motives. Loan-cost disclosure may not be a panacea, however: there are some indications that low-income clients have trouble understanding and using the information. This concern has led to scattered eorts to provide nancial education for consumers, butHow should governments regulate micronance?89most of these programs are still too young to be assessed, at least in developing and transition countries. Consumer protection regimes may also include restrictions on the way loans can be made and collected. Obvious examples would include prohibiting the use of violence or other heavy intimidation to collect loans, but other less dramatic practices may also be deemed abusive. In South Africa, for instance, so-called micro-credit consists mainly of rms making high-interest consumer loans to a clientele consisting mostly of salaried formal-sector employees.3 Taking possession of a borrowers ATM card or requiring delivery of a post-dated check for the loan amount were common practices, which were prohibited under a non-prudential regulatory regime created by the Micro Finance Regulatory Council (MFRC), a government agency lodged outside of the banking authority. MFRC rules were given teeth by stipulating that any loans issued in violation of those rules would be legally unenforceable. The Bolivian micronance sector suered huge losses when proigate Chilean-backed consumer lenders started marketing to unsalaried microentrepreneurs, passing out loans that bore no relation to the borrowers repayment capacity. Many borrowers got in over their heads, and since a large percentage of these were also borrowing from more responsible MFIs, those sound MFIs were badly hurt by the ensuing wave of defaults, not to mention borrowers who lost their credit rating. The Bolivian Superintendency of Banks responded by requiring all uncollaterized lending to include an assessment of repayment capacity. A the time of this Bolivian crisis, the governments credit reference bureau was not working well, so it was hard for lenders to know whether a potential borrower had loans outstanding from another source, or had a history of repayment problems. After the crisis, all of the actors found themselves considerably more enthusiastic about the credit bureau, and unlicensed lending-only MFIs were allowed to participate for the rst time. As a general matter, credit reference bureaus are a powerful tool for extending credit access to previously excluded groups, because the bureaus signicantly lower the costs of appraising borrower creditworthiness, and strengthen borrowers motivation to repay. Credit bureaus make it possible to lend to customers who would have been unprotable otherwise. Other consumer protection measures include privacy protection and accessible dispute-resolution systems. AML/CFT Regulation4 The Financial Action Task Force (FATF) is an international body that recommends standards for national legislation on anti-money-laundering and90Micronancecountering the nancing of terrorism (AML/CFT). The FATF standards do allow room for adjustment to t individual country circumstances, but developing and transition countries are often nervous of straying far from the standards, because winding up on the list of non-complying nations can have severe consequences. Among the FATF standards are know-yourclient rules (ascertaining and documenting the customers true identities and addresses), heightened surveillance of transactions, preserving transaction records, and reporting suspicious transactions to national authorities. Many banks complain loudly about the additional costs generated by these requirements when dealing with their normal customers. In the world of micronance, where transactions and balances are much tinier, full enforcement of regular AML/CFT standards would make it uneconomic to serve large groups of customers. The additional administrative costs are particularly problematic when transactions are so small, and compliance can be impractical for some kinds of clients. For instance, it is challenging to document identity and address for people who have no national identity card, are illiterate, and have never seen any document that species where they live. In its early years FATF was dominated by people coming from a lawenforcement perspective, who were not always instinctively sympathetic with concerns about how FATF rules might exclude low-income clients from services. More recently, this problem is getting more attention at international and local levels. Taking a risk-based approach to AML/CFT, it would not seem that transactions of, say, $30 or loan or savings accounts of $300 create substantial security risks. Governments should consider softer requirements, or waiving them altogether, for accounts and balances below dened limits. After this was done in South Africa, for instance, banks were able to oer basic, no-frills transaction accounts that in a few short years reached 1.6 million customers, most of whom would have remained unbanked if the AML/CFT rules had not been relaxed.PRUDENTIAL ISSUESWhether, When, and How to Open Prudential Licensing Regimes for Micronance Regulation as promotion In more than a few countries, the micronance sector consists mainly of weak non-governmental organizations that provide lending only, as well as credit unions or similar savings and loan cooperatives, few of which are large and stable. When a government is confronted with this situation, andHow should governments regulate micronance?91wants to catalyze a large expansion of quality nancial services for its lower-income population, one plausible-sounding response is to develop a new licensing window that allows institutions to become prudentially regulated, and take deposits, without facing minimum capital requirements as high as those required for a full banking license. (Deposit-taking is doubly important: not only does it provide a large funding source for expansion of lending, but it also gives the institutions low-income clients access to a savings service that is often even more valuable to them than credit.) In such countries, this build it and they will come approach is based on the hope that the special licensing window will attract new private sector entrants to the business, or encourage weak existing MFIs to tighten up their operations so as to meet the prudential standards for licensing. There is considerable controversy over this approach. It is premised on the belief that the binding constraint is absence of appropriate regulation, rather than scarcity of competent retail operators. International experience to date has been too limited to produce a general answer to the question; if anything, it suggests that the answer will vary from country to country. For instance, Tanzania spent a great deal of time, eort, and money on the development of a well-conceived licensing regime, but years afterward the results were disappointing. South Africa does not oer a micro-banking license, but the government took steps a decade ago to make it possible for micronance institutions to oer small loans at relatively high interest rates. Despite this change, South Africa still does not have many solid institutions oering uncollateralized loans to unsalaried micro-entrepreneurs. Pakistan has a huge unserved micronance market, and its hard to nd many countries with as good a licensing regime for micronance as the one Pakistan enacted in 2001. But until very recently, institutions licensed under this law contributed hardly at all to the growth of micronance in the country, and the overall nancial condition of licensed and unlicensed providers was worse than it had been when the law was passed. Within the last year, however, things are looking brighter: some of the newly licensed, privately owned micronance banks are expanding aggressively, and the overall nancial performance of the sector is improving. Regulation that follows, rather than leads, the market In most of the countries with eective prudential regimes in place today for micronance, the regulation came after, not before, the development of a critical mass of strong MFIs that were delivering loans on a sustainable basis. Bolivia has the longest and most solid record of successful micronance regulation, and this experience is often held up as an example where a new licensing window for micronance was a powerful contributor to the success of the industry. But the Bolivian licensing regime was put in place only after92Micronancethe country already had a number of strong NGO MFIs who had shown they could manage their lending business stably and protably. BancoSol, the leading MFI, did not use the micronance licensing windowit got a regular commercial banking license well before the specialized micronance law was passed. Most of the other MFIs who got licenses under the new law could probably have raised the money for a banking license if the easier and cheaper MFI license had not been available. In BancoSols rst year or two, there was a certain amount of supervision by winking, as the Superintendent waived application of some prudential rules that didnt t micronance very well. But the time the micronance law was passed and new prudential norms formalized, the Superintendency already had experience from supervising BancoSol. Its possible to argue that Bolivian micronance did not need a new specialized micronance window to reach its present vital state, and that a few adjustments to the countrys banking law and regulations would have created the necessary regulatory space. When countries design a new licensing window for micronance on the expectation that licenses will go mainly to existing NGO MFIs during the early years, the regulators sometimes dont pay enough attention to the condition of those MFIs and their loan assets. In Zambia, for instance, the foreign aid agencies of the United States and Sweden nanced development of a prudential regime in 1999 that would allow MFIs to take deposits. But at that time, sources say, the country had few if any MFIs whose cost recovery and loan collection would make them safe custodians of customers deposits. There may have been some expectation that donor-funded technical assistance would turn the MFIs into strong institutions, but it is hard to nd many examples of weakly managed MFIs that have been turned into vibrant, stable MFIs by outside technical assistance. This is not to suggest that such assistance is useless: MFIs that already have strong managers make good use of such support, but technical assistance can seldom turn a weak manager into a strong one. Political considerations prevented enactment of the Zambian law at the time. A set of micronance regulations was nally put into force in 2006, but a review of the MIX Market database as of the beginning of 2007 shows only a single sustainable Zambian MFI, and that one had only 12 500 clients.5 Adjusting Prudential Norms to Fit Micronance Products and Institutions Some regulations common in traditional banking need to be altered to accommodate micronance. Whether micronance is being developed through specialized stand-alone deposit-taking MFIs, or as a product line within retail banks or nance companies, the following norms will usually need re-examination:How should governments regulate micronance?93Minimum capital The kind of investors who are willing and able to nance MFIs may not be able to come up with the minimum capital required for a full banking license, especially as minimum capital requirements trend upward around the world. Setting a low minimum capital bar is often the central objective of those pushing for new licensing regimes for micronance. When banking authorities set minimum capital, bank safety and soundness may not be their primary concern. Rather, minimum capital is often used as a rationing device to manage the number of separate institutions that have to be supervised. The arguments for and against low minimum capital for MFIs will be treated in the next section, which deals with supervisory challenges. Capital adequacy Under the Basel Accords, the relationship between shareholders equity and bank risk assets is the foundation of prudential regulation. Equity is treated as a cushion that protects depositors and other creditors of the bank: the more of its assets are funded by shareholders money, the higher the losses the bank can sustain and still be able to repay its depositors. There has been controversy about whether solvency (capital adequacy) requirements should be tighter for specialized MFIs than for banks. If we want a level playing eld in the nancial sector, should micronance be penalized with tougher solvency requirements that lower shareholder protability? Several theoretical arguments point in the direction of higher equity-torisk-assets ratios for MFIs. In the rst place, deposit-taking micronance is a new business in most countries, which supervisorsand some MFI managers as welldo not have decades of experience with. Second, most MFI loan assets are not collateralized. Normally, MFI portfolio quality is very good, but if an MFI starts to have problems with loan delinquency, they can balloon out of control much faster than would be normal with collateralized loans. Third, administrative costs for MFIs are much higher than for commercial banks. When a signicant part of the MFIs loans are not being paid, the uncompensated administrative cost on those loans decapitalizes the MFI faster than would be the case with a normal bank. All of these considerations suggest tighter solvency requirements, at least in the early years. Balanced against those theoretical arguments is the clear empirical fact that licensed MFIs suer fewer loan losses than commercial banks do. There is also emerging evidence that licensed MFIs are more resilient than commercial banks in times of nancial or economic emergencies. In a recent banking crisis in Bolivia, all the commercial banks went insolvent94Micronanceand MFIs came through in good shape. During the nancial meltdown in Indonesia, repayment plummeted on the commercial loans of Bank Rakyat Indonesia (BRI), while there was hardly a blip in the repayment of its micro-loans. When times are uncertain, low-income people are especially anxious to maintain their access to a credit facility, which can be a life-saver if an unexpected shock hits. BRIs micro-borrowers understand that the only way to keep access to a future loan if and when they need it is to faithfully repay todays loan. Some micronance is delivered through credit unions and other nancial cooperatives. Application of capital adequacy norms to these institutions presents a particular issue with respect to the denition of capital. All credit union members have to invest a minimum amount of share capital into the institution. But unlike an equity investment in a bank, share capital can usually be withdrawn whenever a member decides to leave the credit union. From the vantage of institutional safety, such capital is not very satisfactory: it is impermanent, and is most likely to be withdrawn at precisely the point where it would be most neededwhen the credit union gets in trouble. Capital built up from retained earnings, sometimes called institutional capital, is not subject to this problem. One approach to this issue is to limit members rights to withdraw share capital if the credit unions capital adequacy falls to a dangerous level. A more conservative approach, now recommended by the World Council of Credit Unions, is to require credit unions to build up a certain level of institutional capital over a few years, after which time capital adequacy is based solely on those retained earnings. Unsecured lending limits and loan-loss provisions The experience in normal banking is that loans are more likely to default when they are not backed by collateral or guarantees. Thus, banking regulations often put tight limits on unsecured lendingfor instance, capping it at no more than 100 percent of the banks equity base. Such a limitation would make a portfolio of uncollateralized micro-credit impossible, at least in a specialized MFI. Some regulators have avoided the problem by treating group guarantees as collateral. But not all micro-lenders use a group methodology, and group guarantees are less eective than is commonly supposed. Many MFIs do not really enforce such guarantees, and loan losses in MFIs that use such guarantees are not markedly lower than loan losses in MFIs that do not. A more straightforward approach is to recognize the empirical evidence. Worldwide, with relatively few exceptions, uncollateralized loans in a countrys licensed MFIs suer less delinquency and default than collateralized loans in normal bank portfolios. The reasonable response to thisHow should governments regulate micronance?95evidence is to put no collateral requirements on micro-loans, but instead to concentrate on close supervision of the MFIs lending systems and repayment history. Banks in some places have been required to book a loan-loss provision expense to cover the full value of uncollateralized loans they make, even before they become delinquent. Again, this is unworkable for micro-credit. Even if the provision expense is later reversed when the loan is collected, the accumulated charges for loans that are showing no problems would produce a massive under-representation of the MFIs real net worth. And such a requirement has no empirical justication in the case of microcredit. Thus, general provisions (provisions booked when the loan is made, and so not based on the presence of any repayment delays) should be no more stringent for micro-credit than for normal portfolios. The picture changes, however, once a micro-loan has actually fallen late. When one narrows the focus down to the micro-borrowers who do run into repayment problems, experience shows that ultimate collection of their loans is less likely than collection of collateralized loans that fall late by the same amount of time. As a result, provisioning of already-delinquent loans needs to be more aggressive for micro-credit than for normal collateralized loans. Loan le requirements Given the nature of micronance borrowers, their informal businesses, and their loan sizes, it would be unreasonable to make micro-lenders generate the same loan documentation that is required for normal bank loans. This is particularly true in the case of nancial statements for the borrowers business, evidence that the business is formally registered, or collateral documentation. On the other hand, micro-loan les should always contain at least the loan contract, a record of the borrowers repayment history on prior or concurrent obligations, and a simple estimate of the borrowers income, expenses, and repayment capacity, at least where the MFIs methodology relies on loan ocers rather than fellow group members to determine repayment capacity. However, MFIs that make repeated short-term loans, for instance every three months, should not be required to do a fresh analysis of borrower cash ow before every single loan. Restrictions on co-signers as borrowers Regulations sometimes prohibit a bank from lending to someone who has co-signed or otherwise guaranteed a loan from that same bank. Such rules would need to be waived for MFIs that do group lending with crossguarantees among the group.96MicronanceInsider lending Loans made to owners, directors, or managers of a bank are not likely to receive the same objective scrutiny as loans to unrelated parties. In recognition of this fact, most banking authorities now restrict insider lending to some limited percentage of the banks assets or equity. This authors view is that insider lending should be completely prohibited in licensed MFIs, with the exception of small welfare loan programs for employees. When specialized MFIs are receiving favorable regulatory treatment for the sole reason that they are extending nancial access to low-income customers, it is hard to see any reason or need for insider lending. Frequency and content of reporting Banks may be required to report their nancial position frequently, even daily. In many settings, the undeveloped state of transportation and communication infrastructure may make this dicult or impossible for rural banks or branches. In addition, frequent or voluminous reporting to the banking supervisor can add substantially to the administrative costs of an intermediary, especially one that specializes in very small transactions. The chief nancial ocer of BancoSol once estimated that compliance with the banking supervisors reporting requirements cost the bank 5 percent of its assets the rst year, and 1 percent or more a year thereafter. On the other, eective supervision is impossible without adequate reporting. Specialized micronance banks or branches usually present a less complex set of risks than normal banking, so it should be possible to supervise them well based on reporting that is somewhat less burdensome and expensive than what conventional banks have to provide. Physical security and branching requirements Banks hours of business, location of branches, and security requirements are often strictly regulated in ways that could impede service to a micronance clientele. For instance, the convenience of clients who are running their micro-businesses all day might require operations outside normal business hours, or cost considerations might require that sta rotate among branches that are open only one or two days a week. Security requirements such as guards or expensive vaults, or other normal infrastructure rules, could make it too costly to open small-volume branches in poor areas. Branching and physical security requirements merit reexaminationbut not necessarily eliminationin the micronance context. Clients needs for nancial services have to be balanced against the security risks inherent in managing cash.How should governments regulate micronance?97Ownership requirements Some countries have ownership-diversication rules that prohibit any single party from controlling more than 20 percent (for instance) of a banks shares. Also, NGOs may not be eligible to own bank shares. Both of these rules serve legitimate prudential objectives, but they can cause serious diculty in the common case where the assets of a newly licensed MFI come almost exclusively from an NGO that has built up the business over a number of years. In recent years, commercial and quasi-commercial investors are showing greater interest in buying shares of newly licensed MFIs, but there are many transforming MFIs for whom attracting such investment is not a practical option, at least not at the time that they convert to licensed form. If the original NGO has to nd new owners who will purchase 80 or 100 percent of the business, transformation into licensed form could be delayed a long time. Some banking supervisors are allowing the original NGO to own most or all of the shares of a newly licensed MFI, with a requirement that the ownership structure has to be brought into line with normal banking rules over a reasonable period of years. Deposit insurance In order to protect smaller depositors and reduce the likelihood of runs on banks, many countries provide explicit insurance of bank deposits up to some size limit. Some other countries provide de facto reimbursement of bank depositors losses even in the absence of an explicit legal commitment to do so. There is considerable debate about whether public deposit insurance is eective in improving bank stability, whether it encourages inappropriate risk-taking on the part of bank managers, and whether such insurance would be better provided through private markets. In any event, if deposits in commercial banks are insured, the presumption probably ought to be that deposits in other institutions prudentially licensed by the nancial authorities should also be insured, absent strong reasons to the contrary. Branchless banking In a growing number of developing and transition countries, nancial services are being provided outside of conventional bank branches. The use of automated teller machines (ATMs) has been spreading for years. More recently, payment, transfer, and savings services are being oered through post oces or retail outlets like groceries, pharmacies, or gas stations. Such services may be used mainly by the middle class, but they hold promise for poor people as well, especially the rural poor. Using such retail agents, banks can reach places where building and stang a branch would be unprotable because of remoteness, low client density, or low client98Micronancetransaction sizes. In addition, mobile phone operators in countries like the Philippines, South Africa, and Kenya are exploiting their networks to provide fast and convenient payment and transfer services to their subscribers, who include increasing numbers of low-income people. Some branchless banking is bank-led: all of the clients transactions are with a licensed commercial bank, and the retail agents or mobile phone operators are acting as third-party agents to handle cash on the banks behalf. The bank remains responsible for any cash received. These arrangements raise some regulatory risk, including security of cash-handling and proper training of agents, but in general they do not add materially to the risks that are present in normal branch-based banking. More of a regulatory challenge is presented by non-bank-led models, where the clients cash is taken and held by a company like a mobile phone operator that is not licensed and prudentially supervised by the banking authorities. When such companies are holding signicant amounts of customers cash, should they be required to meet the same prudential standards as banks? South Africas answer to that question is a conservative one: any mobile phone operator that wants to provide e-money services is required to partner with (that is, operate under the license of) a commercial bank. This raises the mobile operators costs considerably, and these costs must eventually be passed along to the client. The Philippines is starting with a more liberal approach, allowing mobile operators to operate independent of a banking license. However, the amount of such transactions and the size of outstanding balances owed an individual customer are capped at low levels. So far this arrangement has not created signicant problems, but the central bank is moving to tighten restrictions further. It is not yet possible to identify best practices in dealing with this issue: so far, the European Community has not been able to agree on a common approach.SUPERVISORY ISSUESThe Burden of Supervising Small Intermediaries For bank supervisors in many developing countries (though certainly not all), the central fact of life is responsibility for supervising a commercial banking system with severe structural problems, often including some sizable banks teetering dangerously close to the edge of safety. The collapse of oneor a half-dozenof these banks could threaten the countrys nancial system with implosion. In trying to manage bank risk, supervisorsHow should governments regulate micronance?99may have to work in a political mineeld, because the owners of banks are seldom under-represented in the political process. The supervisors legal authority to enforce compliance or manage orderly clean-ups is often inadequate. They may not have enough control over the tenure, qualication, and pay of their sta. Monitoring healthy banks is challenging enough, but the real problems come when it is time to deal with institutions in trouble. When a sick bank nally crumbles, its president can start sleeping again (though perhaps in a dierent country), while it is the supervisor who has to stay awake at night worrying. If bank supervisors sometimes display resistance to adding MFIs mostly small, mostly new, mostly weak on protabilityto their basket of responsibilities, we should recognize that their reasons may be nobler than narrow-mindedness or lack of concern for the poor. The Philippines licenses hundreds of small intermediaries as rural banks. Originally, the minimum capital requirement for a rural bank was very low. These banks are not micronance institutions, but their operations do include credit and deposit services for lower-income clients. They have access to the national payments system and are supervised by the central bank. As of September 1997, 824 rural banks were serving half a million clients. These banks had only about 2 percent of the banking systems assets and deposits, but they made up 83 percent of the institutions the central bank had to supervise. Supervising the rural banks severely stretched the resources of the central banks supervision department, tying up as much as one-half of its total sta and budgetary resources at times. In the early 1990s one in every ve rural banks had to be shut down, and many others had to be merged or otherwise restructured. An unpublished 1996 analysis reported that about 200 inspectors were assigned to the rural banks, but even this level of resources was viewed as inadequate. Each on-site examination consumed up to three person-weeks or more. At one point the supervisory department found that this burden, combined with its budget limitations, was severely endangering its ability to function. One of the responses to the crisis was to raise the minimum capital substantially. But a knowledgeable observer has guesstimated that as of late 2007, perhaps half of the rural banks are still technically insolvent, though these tend to be the smaller ones. The larger rural banks have most of the assets and customers are said to be strong and expanding aggressively. The occurrence of a supervisory meltdown doesnt necessarily mean that licensing rural banks has been a failure in the Philippines. Hundreds of thousands of people are still getting services that would otherwise have been unavailable to them. But the experience, and similar ones in Indonesia and Ghana, teaches us to be realistic about the diculty of supervising100Micronancelarge numbers of small new nancial institutions. Some would argue that ineective supervision is worse than no supervision at all, because it misleads depositors and tarnishes the credibility of the banking authorities. Minimum Capital as a Rationing Device When regulators set minimum capital requirements for licensing MFIs, a major consideration should be limiting new licenses to a number that is consistent with available supervisory resources. Obviously, this has to be balanced against the objective of opening access to nancial markets, an objective that tends in the direction of keeping minimum capital as low as possible. There is a strong argument to be made that regulators should start with relatively high minimum capital for a new licensing window, and gradually relax the requirement after there has been more supervisory experience, and supervisors are better able to judge what they can take on. A country does not necessarily need large numbers of licensed MFIs in order to serve its market well. In most countries, a few large MFIs account for the vast majority of the outreach. In 2000, Bangladesh probably had over 1000 MFIs, but the largest ten served all but about 15 percent of the clients. Small Community-based Intermediaries Smaller institutions may not require as much supervision as big ones, but there are lower limits to how far supervision can be watered down. At some point, supervision lite is no longer eective, if eectiveness means that the supervisor can expect to ag most problems before they have gotten too serious to x. Some member-owned intermediaries take deposits but are so small, and sometimes so geographically remote, that they cannot be supervised on any cost-eective basis. This poses a practical problem for the regulator. Should these institutions be allowed to operate without prudential supervision, or should minimum capital or other requirements be enforced against them so that they have to cease taking deposits? Sometimes regulators are inclined to the latter course. They argue that institutions that cannot be supervised are not safe, and therefore should not be allowed to take small depositors savings. After all, are not small and poor customers just as entitled to safety as large and better-o customers? But this analysis is too simple if it does not consider the actual alternatives available to the depositor. Poor people can and do save. If formal deposit accounts are not available, they have to fall back on savings tools like currency under the mattress, livestock, building materials, or informal arrangements like rotating savings and credit clubs. All of these vehicles areHow should governments regulate micronance?101risky, and in many if not most cases, they are more risky than a formal account in a small unsupervised intermediary. Closing down the local savings and loan cooperative may in fact raise, not lower, the risk faced by local savers by forcing them back to less satisfactory forms of savings. Because of these considerations, most regulators facing the issue have chosen to exempt community-based intermediaries below a certain size from requirements for prudential regulation and supervision. The size limits are determined by number of members, amount of assets, or both. (Sometimes the exemption is available only to closed bond institutions whose services are available only to members of a pre-existing group such as employees of a company.) Once the size limits are exceeded, the institution must comply with prudential regulation and be supervised. If small intermediaries are allowed to take deposits without prudential supervision, a good argument can be made that their customers should be clearly advised that no government agency is monitoring the health of the institution, and thus that they need to form their own conclusions based mainly on their knowledge of the individuals running the institution. Supervisory Tools and their Limitations Some standard tools for examining banks loan portfolios are ineective for micro-credit. As noted earlier, loan-le documentation is a weak indicator of micro-credit risk. In a commercial bank, one can often capture most of the portfolio risk by examining a small number of large loans, but this is not true in a micro-credit portfolio consisting of thousands of tiny loans. Sending out conrmation letters to verify account balances is usually impractical for micro-credit, especially where client literacy is low. Instead, the examiner must rely more on analysis of the institutions lending systems and their historical performance. Analysis of these systems requires knowledge of micronance methods and operations, and drawing practical conclusions from such analysis calls for experienced interpretation and judgment. Supervisory sta are unlikely to monitor MFIs eectively unless they are trained and to some extent specialized. When an MFI gets in trouble and the supervisor issues a capital call, many MFI owners are not well-positioned to respond to it. NGOs who own shares may not have enough liquid capital available. Development agencies and development-oriented investors usually have plenty of money, but their internal procedures for disbursing it sometimes take so long that a timely response to a capital call is impractical. Thus, when a problem surfaces in a supervised MFI the supervisor may not be able to get it solved by a timely injection of new capital, as the Colombian banking supervisor found out when the MFI FinanSol ran into trouble.102MicronanceAnother common tool that supervisors use to deal with a bank in trouble is the stop-lending order, which prevents the bank from taking on further credit risk until its problems have been sorted out. A commercial banks loans are usually collateralized, and most of the banks customers do not necessarily expect an automatic follow-on loan when they pay o their existing loan. Therefore, a commercial bank may be able to stop new lending for a period without destroying its ability to collect its existing loans. The same is not true of most MFIs. Immediate follow-on loans are the norm for most micro-credit. If an MFI stops issuing repeat loans for very long, customers lose their primary incentive to repay, which is their condence that they will have timely access to future loans when they want them. When an MFI stops new lending, many of its existing borrowers will usually stop repaying. This makes the stop-lending order a weapon too powerful to use, at least if there is any hope of salvaging the MFIs portfolio. A typical MFIs close relationship with its clients may mean that loan assets have little value in the hands of a dierent management team. Therefore, a supervisors option of encouraging the transfer of loan assets to a stronger institution may not be as eective as in the case of collateralized commercial bank loans. The fact that some key supervisory tools do not work very well for micronance certainly does not mean that MFIs cannot be supervised. However, regulators should weigh this fact when they decide how many new licenses to issue, and how conservative to be in setting capital standards or required levels of past performance for transforming MFIs. Where to Locate Micronance Supervision Given the problem of budgeting scarce supervisory resources, alternatives to the conventional supervisory mechanisms used for commercial banks are frequently proposed for depository MFIs. Within the existing supervisory authority? The default option for MFI supervision would normally be the supervisory authority responsible for commercial banks. Using this agency to supervise micronance takes advantage of existing skills and lowers the incentive for regulatory arbitrage. If this option is chosen, the next question is whether to create a separate department of that agency. The answer will vary from country to country, but at a minimum, a specially trained supervisory sta is needed, given the diering risk characteristics and supervisory techniques in the case of MFIs and micronance portfolios. The location of micronance supervision becomes a more complicated question when both non-depository micro-lending institutions andHow should governments regulate micronance?103depository MFIs are to be addressed within a single, comprehensive regulatory scheme. The tasks involved in issuing permits to non-depository microlending institutions have relatively little to do with the prudential regulation and supervision of licensed depository institutions. In some contexts, lodging both of these disparate functions within the same regulatory body might be justied on pragmatic groundssuch as the absence of any other appropriate body, or the likelihood that the permit-issuing function would be more susceptible to political manipulation and abuse if carried out by another body. In other cases, non-depository MFIs are required to report to the banking supervisor in order to make it easier for them to move eventually into more services and more demanding prudential regulation. Often, however, the risks of consolidating prudential and non-prudential regulation of micronance within the banking authority will outweigh the benets. These risks include the possibility of confusion on the part of supervisors as to the appropriate treatment of non-depository institutions, and the possibility that the public will see the supervisory authority as vouching for the nancial health of the non-depository institutions, even though it is not (and should not be) monitoring the health of these institutions closely. A separate and independent agency? In some countries the banking authorities reluctance to take responsibility for micronance leads to plans or decisions to lodge micronance supervision in an independent agency. Building skills and experience in a dierent body can be time-consuming, and the new MFI supervisor may not be as politically independent as the banking supervisor. One approach to shortening the learning curve is to entrust prudential supervision of deposittaking MFIs to an apex agency that is already making wholesale loans to MFIs. This structure can present conicts of interest. If such an apex supervisor has large loans outstanding to a troubled MFI, will the agency be tempted to drag its feet when depositors interests are best served by shutting the MFI down? On the other hand, central banks frequently deal with similar conicts of interest. In developing countries, regulation and supervision of credit unions has usually been lodged outside the banking authority, often in the government department that is responsible for cooperatives of all sorts. The experience with this arrangement has usually been very disappointing. Delegated supervision? Sometimes the government nancial supervisor delegates part or all of the tasks of direct supervision to an outside body, while monitoring and controlling that bodys work. This seems to have worked, for a time at least, in some cases where the government nancial supervisor closely monitored104Micronancethe quality of the delegated supervisors oversight, although it is not clear that this model reduces total supervision costs. Where this model is being considered, it is important to have clear answers to three questions: (1) who will pay the substantial costs of the delegated supervision and the government supervisors oversight of it? (2) if the delegated supervisor proves unreliable and its delegated authority must be withdrawn, is there a realistic fallback option available to the government supervisor? and (3) when a supervised institution fails, which body will have the authority and resources to clean up the situation by intervention, liquidation, or merger? Because many MFIs are relatively small, there is some temptation to think that their supervision, or at least on-site inspection, can be safely delegated to external audit rms. Unfortunately, experience has been that external audits of MFIs, even by internationally aliated audit rms, very seldom include testing that is adequate to provide a reasonable assurance as to the soundness of the MFIs loan assets, which is by far the largest risk area for micro-lenders. If reliance is to be placed on auditors, the supervisor must require micronance-specic audit protocols that are more eective, and more expensive, than the ones now in general use, and must regularly test the auditors work. Self-regulation and supervision When regulators decide that it is not cost-eective for the banking authority to provide direct oversight of large numbers of MFIs, self-regulation is sometimes suggested as an alternative. Discussion of self-regulation tends to be confused because people use the term to mean dierent things. In this discussion, self-regulation refers to regulation (and/or supervision) by some body that is eectively controlled by the regulated entities. This is one point on which historical evidence seems clear. Self-regulation of non-bank nancial intermediaries in developing countries has been tried many times, and has virtually never been eective in protecting the soundness of the regulated organizations. One cannot assert that eective selfregulation in these settings is impossible in principle, but it can be asserted that such self-regulation is almost always an unwise gamble against very long odds, at least if it is expected that the regulation and supervision actually enforce nancial discipline and prudent risk management. Sometimes regulators have required certain small intermediaries to be self-regulated, not because they expect the regulation and supervision to be eective, but because this is politically more palatable than saying that these deposittakers will be unsupervised. This can be a sensible accommodation in some settings. While self-regulation probably will not keep nancial intermediaries healthy, it may have some benets in getting institutions to begin a reporting process, or in articulating basic standards of good practice.How should governments regulate micronance?105AFTERWORD: DOES PRUDENTIAL REGULATION WORK?There has long been a respectable minority of academic opinion that is skeptical about the conventional wisdom that government prudential regulation and supervision are eective in limiting bank failures and nancial system crises. Barth, Caprio and Levine (2006) conducted a cross-country analysis based on a 150-country database, and drew some jarring conclusions:In terms of what works best, our analyses raise a cautionary ag regarding the foundations of current international best practice recommendations. In particular, our results question the ecacy of Basel IIs rst two pillars on capital regulations and ocial supervision . . . Across the dierent statistical approaches, we nd that empowering direct ocial supervision of banks and strengthening capital standards do not boost bank development, improve bank eciency, reduce corruption in lending, or lower banking system fragility. Indeed, the evidence suggests that fortifying ocial supervisory oversight and disciplinary powers actually impedes the ecient operation of banks, increases corruption in lending, and therefore hurts the eectiveness of capital allocation without any corresponding improvement in bank stability . . . In contrast to these ndings on capital regulations and direct supervisory oversight of banks . . . supervisory and regulatory policies that facilitate private sector monitoring of banks improve bank operations, which endorses Basel IIs third pillar on market discipline. One mechanism for fostering private monitoring of banks is by requiring the disclosure of reliable, comprehensive, and timely information. Countries that enact and implement these pro-private monitoring regulations enjoy more ecient banks and suer from less corruption in lending. Furthermore, laws that strengthen the rights of private investors enhance the corporate governance of banks. In contract, policies (for example, deposit insurance) that weaken market monitoring of banks tend to have adverse ramications on the banking system [including reducing system stability rather than strengthening it] . . . We recognize, of course, that many countries do not have the legal and political institutions necessary to support eective market monitoring of banks. Consequently, many readers may conclude that a practical approach involves empowering ocial supervisors until countries develop the institutional foundations for market monitoring. The cross-country results thus far, however, do not support this conclusion. The results instead indicate that regulatory restrictions on bank activities, impediments to the entry of new banks, government ownership of banks, and reliance on powerful ocial supervisors to oversee banks have adverse eects on the operation of banks. Moreover, it is exactly in countries with weak political and legal institutions that empowering ocial supervisors is likely to be most detrimental. (Barth, Caprio and Levine, 2006, pp. 1016)106MicronanceOther econometricians disagree about interpretation of the data, and this author has neither the competence nor the courage to oer any judgment about the issue. In any event, hardly any of the worlds countries have been willing to take the advice of those who favor eliminating ocial supervision of banks. Whatever the merits of the issue, it would be politically impossible for most governments to back away from protecting depositors. Then why raise the question in a discussion about regulating micronance? The reason is that in crafting a regulatory regime for micronance, many decisions about detail and degree depend on ones underlying assumptions about the appropriateness and eectiveness of government prudential oversight. To the extent that one has doubts about such oversight, or is concerned that regulatory power put into the hands of a government supervisor may be used for personal rather than public benet, then one might be more cautious in deciding particular issues related to the scope of such power.NOTES1. Authors note: this chapter draws heavily on a set of consensus guidelines developed in consultation with a wide range of experienced regulators, policy advisors, and industry analysts (Christen, Lyman and Rosenberg, 2003) as well as the working experiences of the author and many colleagues. Citation of sources will be limited. The discussion is meant to apply to developing and transition economies only: the dynamics of micronance in rich countries, and the regulatory issues they present, can be substantially dierent. 2. It is important to note that if Compartamos priced its loans to produce no prot at all, it would still have to charge interest of about 77 percent, which would shock most consciences despite being completely driven by costs of lending. Compartamoss loans are extremely small in relation to the Mexican context, so its administrative costs alone are very high (Rosenberg, 2007). 3. In most countries, micro-credit is understood to refer mainly to small uncollateralized loans to people who have informal micro-businesses rather than jobs in the formal sector. 4. This section draws heavily on Isern, Porteus, Hernandez-Coss and Egwuagu (2005). 5. www.mixmarket.org/en/demand/demand.show.prole.asp?ett=1784&.REFERENCESBarth, J.R., G. Caprio and R. Levine (2006), Rethinking Bank Regulation: Till Angels Govern, New York: Cambridge University Press. Christen, R.P., T.R. Lyman and R. Rosenberg (2003), Guiding principles on regulation and supervision of micronance, CGAP, accessed at www.cgap. org/portal/binary/com.epicentric.contentmanagement.servlet.ContentDelivery Servlet/Documents/Guideline_RegSup.pdf. Isern, J., D. Porteus, R. Hernandez-Coss, and C. Egwuagu (2005), AML/CFT regulation: implications for nancial service providers that serve low-income people, CGAP Focus Note No. 29, accessed at. www.cgap.org/portal/binary/How should governments regulate micronance?107com.epicentric.contentmanagement.servlet.ContentDeliveryServlet/Publication s/html_pubs/FocusNote_29.html. Rosenberg, R. (2007), CGAP reections on the Compartamos initial public oering: a case study on micronance interest rates and prots, CGAP, Focus Note No. 42, accessed at www.cgap.org/portal/binary/com.epicentric.contentmanagement.servlet.ContentDeliveryServlet/Documents/FocusNote_42.pdf.6. Gender empowerment in micronanceBeatriz Armendriz1 and Nigel RoomeINTRODUCTIONEver since micronance was popularized in the mid-1970s in Bangladesh, one of its salient features has been the overwhelming representation of women. The trend has increased steadily, particularly during the 1980s. According to 2006 Microcredit Summit Campaign report, seven out of ten micronance clients are women.2 Millions of these women are married or live with a partner, and many have children. Relative to initial lending practices by the Grameen Bank in Bangladesh, the bias in favor of loans to women in micronance has been accompanied by an increasing trend to exclude men from micronance services, particularly at very low income levels. The practice of exclusion might however prove to be counterproductive, for it can generate frictions within households, as men feel increasingly threatened in their role as primary breadwinners within the household.3 In this chapter we argue that the promotion of women in micronance initiatives and the bias against men is taking place in the absence of solid empirical evidence on the eects of this strategy, and on the balance of power in households and on the health, education, and well-being of all household members, which we hold to be key aspects of development. We further argue that this issue deserves research given the possibility of unforeseen outcomes and adverse consequences that run counter to the goal of micronance initiatives to promote development. To clarify the central issues, on the one hand, higher household income in the hands of women might increase health and education for women and their household memberswe call this the women-empowerment eect. On the other hand, the exclusion of men from access to subsidized nance might create frictions, and rebound eects, that diminish the supportive role women play for their spouses and wider household members in the production of health and educationwe call this the women-disempowering eect. In the event that the latter eect dominates over the former, then108Gender empowerment in micronance109subsidized micronance for women might have no impact, or even worse, a negative impact on health and education. An even more challenging question is what social and institutional conditions most strongly inuence these empowerment and disempowerment eects and their outcomes. This chapter is structured as follows. First, it provides an overview of what we currently know about micronance, gender, health, and education in the context of Bangladesh, where most research has been conducted. Second, some anecdotal evidence from Bangladesh and Africa on the notion of micronance empowerment is presented and discussed. This raises questions about the structures on the enhanced capacity of women to assert their role as the main providers of health and education, mainly arising from the fact that the empowerment of women generates frictions with their partners, which in turn leads to a potential disempowerment eects. Third, anecdotal evidence from Chiapas in southern Mexico is outlined, which provided the basis for empirical research on new approaches to micronance now being undertaken in the region. Fourth, the chapter outlines this experimental intervention in southern Mexico, where the women borrowers in a micronance initiative invited their spouses to be part of women-only solidarity groups as borrowers, in order to see whether potential frictions could be eliminated as a way to enhance women empowerment and provide for better access to health and education at the household level. The main challenges of implementing this type of intervention, which were revealed through this empirical study so far are described. The nal section spells out some concluding remarks.CURRENT KNOWLEDGE OF MICROFINANCE, FINANCIAL RESOURCES, AND GENDER AS A BASIS FOR THE PRO-WOMEN BIAS4The most inuential empirical study on micronance and gender can be found in an article published by the Journal of Political Economy in 1998 by Mark Pitt and Shahidur Khandker. In their study, Pitt and Khandker develop a framework for estimating the impact of micronance using cross-section data from Bangladesh for 199192. The paper pins down the potential sources of bias, in identifying and estimating the impact of micronance initiatives alone on outcomes such as household expenditures on health and education. For example, Pitt and Khandker addressed the bias that might arise because the individuals who self-select into micronance programs may be the least poor and most entrepreneurial members of their community. This bias would lead to an overestimate of the overall potential of micronance110Micronanceon poverty reduction as and when micronance projects were expanded to include poorer and less entrepreneurial individuals. Pitt and Khandker faced the well-known endogeneity problem that entrepreneurial individuals may deliver for themselves better incomes, which then enable them to qualify for further loans, which in turn increase their incomes. Typical ways to resolve this problem of estimating is through the use of an independent variable, which correlates with entrepreneurial activity but not with outcomes. Pitt and Khandker used land-ownership as an instrumental variable: to qualify for a micronance loan, individuals (both men and women) had to be poor as proxied by their holdings of land not being more than half an acre. They used in this instrumental variable in studies that compared villages with micronance opportunities and control villages without. This approach meant that those who received loans in treatment villages did so because they were landless poor, with the same entrepreneurial abilities as those in the control villages where the landless poor did not access micronance loans because there were not that many micronance providers. Once village characteristics were controlled for, Pitt and Khandker extended their analysis to the role of gender. In particular, taking advantage of the fact that Bangladeshi micronance enterprises at the time were lending to men as well as women, Pitt and Khandker conducted a study to assess the relative impact of lending to men head-of-households compared with their women counterparts on development outcomes in terms of health and education. Their results regarding relative provision for health and education by both types of household heads were, at best, unclear. However, the main ndings of the study of expenditures by both types of households in relation to household expenditures (particularly on food and tools for expanding their respective businesses) are well known. These results have been highly inuential, particularly in shaping the trend in focusing donors aid through subsidized loans toward women.5 In particular, Pitt and Khandker showed that when a loan of 100 taka was extended to men it translated into 11 taka going into household expenditures (for food/nutrition/working tools), while the same amount lent to women household heads led to 18 taka being spent on household expenditures (for food/nutrition/working tools). While it would be too bold to claim that the ndings of Pitt and Khandker alone have inuenced the bias to women in recent micronance initiatives. It is our conjecture that in the absence of any countervailing empirical evidence, Pitt and Khandkers ndings contributed to the norms and operational practices of CGAP (Consultative Group to Assist the Poor) World Bank, as well as many other multilateral organizations engaged in providing subsidized micronance. Their priority has been to direct subsidized loans to women.Gender empowerment in micronance111The common practice in favor of subsidized micro-loans to women also ows from the research on the practice of delivering aid to women. For example, food stamps in the United Kingdom and Sri Lanka, and staple food and cash deliveries under the PROGRESA (now called OPORTUNIDADES) program in Mexico were directed to women rather than their spouses. This was done for fear that if such aid was given to men, they might sell the food stamps and mis-spend the resources, possibly wasting money on gambling, tobacco, and alcohol. There are a number of empirical studies on the practice of targeting aid to women. Emmanuel Skouas (2001) reports that the OPORTUNIDADES project aimed at women in rural Mexico led to sharp social improvements: poverty decreased by 10 percent, school enrollment increased by 4 percent, food expenditures increased by 11 percent, and adults health (as measured by the number of unproductive days due to illness) also improved considerably.6 Duncan Thomas (1990) reports that child health in Brazil (as measured by survival probabilities, height-for-age, and weight-for-height) along with household nutrient intakes, tended to rise if additional non-labor income was in the hands of women rather than men. He observed that income in the hands of a mother had, on average, 20 times the impact of the same income in the hands of a father with respect to childrens survival probabilities. In a subsequent study, also on Brazil, Thomas (1994) reports that increasing the bargaining power of women is associated with increases in the share of the household budget spent on health, education, and housing as well as improvements in child health. Patrice Engle (1993) similarly studies the relationship between a mothers and fathers income on child nutritional status (height-for-age, weight-for-age, and weight-for-height) for hundreds of households in Guatemala, and reports that childrens welfare improves as womens earning power increases relative to their husbands. Paul Schultz (1990) nds that in Thailand non-labor income in the hands of women tends to reduce fertility more than non-labor income possessed by men. He also nds that the impact of non-labor income has dierent eects on labor supply, depending on which household member controls that income.7 Anderson and Balands (2002)s article on Rotating Savings and Credit Associations (ROSCAs) reports on a survey of hundreds of women in Kenya. An overwhelming majority of the women responded that the principal objective for joining a ROSCA was to save money, and nearly all of the respondents were married. Anderson and Baland conclude that an important motive for women joining ROSCAs is thus the desire to keep money away from their husbands. Other studies, not necessarily conned to ROSCAs, suggest that savings motives (and the protection of assets) also apply to womens involvement in micronance institutions.112MicronanceChristopher Udrys (1996) research on agricultural practices in Burkina Faso provides evidence on the ways men and women invest in agriculture. Using panel data, controlled for soil quality and other variables, he nds that agricultural productivity is higher in plots cultivated by men than women. He also nds that compared with plots cultivated by women, the higher yields of plots cultivated by men are due to a greater intensity of productive inputs (including fertilizer and child labor). He thus concludes that productivity dierentials are attributed to the intensity of production between plots cultivated by men and women and not to inherent skill dierentials: an outcome he regards as inecient since there are sharply diminishing returns to the use of fertilizer. Not only are resources not fully shared, they are allocated in ways that diminish total household income. Udry suggests that by reallocating inputs to plots cultivated by women can thus enhance eciency. Another solution (that is, the micronance solution) is to provide women with credit sucient to purchase additional inputs. A second way that micronance can potentially address problems like this is by tackling the social norms that prevent women from having adequate access to inputs and marketing facilities in the rst place. This might be done through demonstration eects and from pressure created by the micro-lender to ensure higher returns to borrowers investments. From the point of view of practice, a eld loan ocer would see women as better customers for loans compared with men for at least four reasons. First, repayment rates on loans by women are higher, because women are more risk-averse and therefore more conservative in their investment strategy. Also, women are more vulnerable to peer pressure and the threat of public humiliation with regards to failure in the repayments on their loans, women have less opportunities than men to access alternative sources of credit, which in turn reduces the scope for moral hazard.8 Moreover, eld practitioners in micronance argue that women are less argumentative, which reduces the transaction costs of the loan, both for their peers and the bank. Women also lower the agency costs of bank ocers because as groups they are more punctual at repayment meetings, which avoids the bank ocer having to devote time looking for them at their homes/ businesses. Last but not least, women loan ocers cost less than men, and in many instances women are more ecient at granting and collecting repayments.9 Taken together, the ndings of empirical investigations, the perspectives of donors, and experience of practitioners, have led to an established wisdom in favor of lending to women. Moreover, the conventional wisdom has been that excluding men from micronance has no signicant or important detrimental outcomes. However, more recent views from the eldGender empowerment in micronance113expressed at the recent Micronance Forum in Beijing (2006) suggest otherwise:male exclusion can lead to negative consequences for women who join nancial services: they may meet resistance from men who see their exclusive participation as unfair and threatening; their loans may be hijacked . . . A family whose adult members all have access to nancial services is better o than one where half are ineligible. (Hugh Allen at the Micronance Forum, 2006)While the experiential knowledge of people like Hugh Allen should not be accepted without detailed investigation, his views have been a consideration for social scientists and anthropologists voicing similar concerns for some time now. Their observation, which run counter to conventional wisdom are reviewed in the following section.ANECDOTAL EVIDENCE FROM BANGLADESH AND AFRICAIn this section, we argue that there are potential dangers in excluding men from subsidized micronance as this may lead to frictions between household heads, leading to lower quality and quantity of health and education provision within the overall household. At this stage the evidence for this position is anecdotal, deriving from Bangladesh and Africa. It suggests that there is a need to take into account the potential danger of excluding the male head of household from micronance, as their exclusion can overburden women and lower health and education outcomes. Long before the 2006 Nobel Peace Prize was awarded to the creator of Grameen Bank, Muhammad Yunus, for his work in micronance, household surveys from Bangladesh, dating back to 1999, documented evidence that micronance was increasing frictions between husbands and wives, as husbands often felt threatened in their role as primarily income earners (Rahman, 1999). Moreover, well-known evidence, also from Bangladesh, suggests that micronance does not increase womens bargaining power entirely, because on average, women borrowers surrender nearly 40 percent of their control over the investment decisions they make. More alarmingly, over 90 percent of the returns these women realize from their investments are handled by their husbands (Goetz and Sen Gupta, 1996). In Africa, Linda Mayoux (1999), reports on a survey of 15 dierent micronance programs. She nds that the degree of womens empowerment is household- and region-specic, with womens empowerment dependent on inexible social norms and traditions. These ndings have to be weighed against the fact that impact on empowerment will also depend on how well114Micronanceparticular programs were designed. The issue of context factors and program design leads to the exceedingly preliminary observations from a eld experiment undertaken in southern Mexico.ANECDOTAL EVIDENCE FROM FIELD EXPERIMENTS IN SOUTHERN MEXICO10Grameen Trust Chiapas, AC (henceforth called GTC) is one of the rst replications of the Grameen model of micronance in Latin America. The project is located in the highlands of southern Mexico. It deploys funds from the Deutsche Gesellschaft fr Technische Zusammenarbeit (GTZ) via Grameen Trust Bangladesh. The replication in southern Mexico started by lending to women-only groups in 1997.11 In 2003, in sharp contrast to the original Grameen model, GTC took the risk of lending to men of some otherwise women-only groups. Since then the organization grew rapidly, and it now has over 12 000 borrowers in dierent groups, a large majority of those mixed groups of women and men. When branch managers in dierent geographical locations in the southern Mexican replication are asked why they have accepted men into womenonly groups, four explanations are oered. The rst relates to informational asymmetries between men and women. One loan ocer argues that even if loan disbursements and repayments are publicly known in women-only groups, men tend to overestimate the amount of money that women are handling, and they therefore contribute less to overall household expenditure, which often creates frictions within the household. This has dynamic eects. In many instances, women under these conditions are no longer using their loans for investment but for normal household expenditures on food, health, and education (particularly in the month of August when the academic year starts). They also often quarrel with their husbands who are no longer providing as much for these expenditures as they used to. Inviting some men to join the group allows them to have a more accurate estimate of womens real investments and their realized returns. With this information they are less likely to reduce their contributions to household expenditures. In those groups that became mixed, women borrowers invest more, and there were increased repayment rates by both men and women in such groups. A second explanation relies on the potential workload externalities of having women as the only recipients of loans within the household. In particular, another loan ocer argues that when women contract a loan from GTC, they become busier, and that the quality of the services that women traditionally provide to the household such as meals, and householdGender empowerment in micronance115chores, decreases in quantity or quality or both. This, the bank manager argues, irritates men and creates a tense atmosphere within the household. This family tension causes women to default more often or prevents them from making their repayments on time. When men are invited to join groups, they seem to internalize the negative workload externalities created by GTC micro-loans to women. In loan ocer Regis Ernesto Figueroas own words: invited men help more their spouses in their businesses and in household chores, which in turn, reduces tensions, and enables women to repay on time, as men become de-facto business partners of women. A third explanation relates to the absence of secure places for women to hide money while they save for two consecutive weeks or more in order to make their repayments to GTC. In particular, another loan ocer argues that women cannot open bank accounts in commercial banks as these banks do not accept their very small savings, as the transactions costs for the commercial banks are too high relative to the amounts deposited. Women borrowers of GTC therefore hide money away from their husbands in dierent places, generally in the house, because husbands steal the money and use it on alcohol and tobacco. When men are invited to join the group, this particular loan ocer argues, the situation changes because under the Grameen rules, he becomes responsible for the debt of the other male and women members of the group. Women become happier. They no longer complain about their husbands or men in general. The household heads work harmoniously together, the loan ocer explains. A loan ocer at the headquarters of GTC oers a fourth, and last, explanation of why men are invited into womens groups. She argues that the inclusion of men brings more women clients into the scheme, in particular more single women. She explains that the reason is because women generally face a trade-o between being nancially independent via a micro-loan from GTC, or, getting married. The argument goes that since GTC accepts men, women no longer face this trade-o, and they are therefore more likely to become clients. Moreover, the inclusion of men, according to this loan ocer, has increased marriage rates!ATTEMPTS TO MEASURE EMPOWERMENT AND DISEMPOWERMENT12The anecdotal evidence set out above suggests a substantive need to explore in greater depth the relationship between micronance structures and the issues of gender in development and empowerment around micronance. This calls for experiments designed to test the eects of the inclusion of male heads of households into women-only solidarity groups. Such116Micronanceexperiments are exceedingly demanding. Nevertheless they are important given the challenges to the conventional wisdom, that women are increasingly empowered by micronance that enables them to expand their businesses, earn a higher return so that their spouses would value them better, which translates into higher health and education for the household. In designing such studies it is recognized that cultural and institutional dierences may impact the results. Ideally, any test of the empoweringdisempowering hypothesis should therefore take place in Bangladesh, Africa, and Latin America to establish whether the results are culturally and institutionally robust. However, nding partner micronance institutions that would allow researchers to conduct scientic experiments of this kind, is dicult enough; to do so in three continents is yet more dicult. We report here progress to date with a pioneering study developed by Harvard and Yale researchers from the Innovations for Poverty Action (IPA) in their continuing study on the impact of gender issues of micronance and health and education outcomes in association with the Grameen Trust Chiapas, AC in southern Mexico. The elements of this study are reported below. Innovations for Poverty Action (IPA) researchers designed a survey and a follow-up random experiment, using a sample of approximately 2000 borrowers in women-only solidarity groups in 2006. In this experiment married-women-only solidarity groups were (randomly) selected into treatments and controls. Control solidarity groups were not subject to any kind of intervention while the treatments were. The treatments were divided into four dierent sub-groups. Intervention in the rst sub-group consisted of allowing women to voluntarily invite their partners/husbands to join the Grameen-style solidarity group in order to acquire a micro-loan. The study sought to take into account the possible network eects that might follow as invited male spouses joined solidarity groups, increasing the synergies through the span of group members. IPA researchers therefore allowed for a sub-group of women who could invite other women friends to join their group. Similarly, it was recognized that as partners/spouses were invited to join, so household income was increased. This factor was taken into account by extending larger loan sizes to a sub-group consisting of women-only clients. Last but not least, a treatment sub-group consisting of women who could invite their partners/ husbands via providing them high and low monetary incentives to better proxy womens marginal benet from being nancially independent. The IPA researchers are using a follow-up survey of the four subtreatment groups and the control group to assess and evaluate any behavioral changes at household level. Some of the questions that will be resolved are: did women borrowers decide to invite their male spouses, and if so, didGender empowerment in micronance117their willingness to do so increase as they were provided with incentives? In what way did the inclusion of male spouses alter outcomes in terms of health, education, and child labor, for example? It is recognized that there are many other possible dimensions of change to address, but it is not possible to develop a comprehensive questionnaire until the initial results from the 2008 follow-up survey of behavioral changes has been processed and analyzed. In the meantime, IPA researchers have been able to detect some interesting idiosyncrasies in the sample of borrowers, which include ve geographical areas, quite distant from one another. These preliminary ndings show that most of the decisions regarding investment and the expenditures from returns realized from micro-loans are taken by men heads of households, not by women, in the bank branch with the highest proportion of indigenous population. In addition, Grameen Trust Chiapas, as well as other group-lending institutions such as AlSol, another Grameen replication founded by Beatriz Armendriz, have been serving borrowers in these branches/regions for a long time. Health and educational expenditures in these two branches do not dier considerably. If micronance to women had not already empowered women in such traditional societies, and assuming the head of household relationship is basically frictionless as wives systematically defer decision-making to their husbands, it is dicult to imagine what an intervention of the sort we undertook in that region could actually bring about in terms of changes in behavioral patterns. However, when the women in these branches are given the power to invite their spouses into the group this provides an empowering tool in its own right, and, household income is expected to increase when women actually decide to include their spouses. On the other hand we might expect that if any anticipated increase in household income, once partners are included as micronance clients, is then controlled purely by men, there would not be the expected changes in outcomes, particularly with respect to health and education. Anticipating this, women might decide not to invite their spouses in the rst place. It is, however, much too early in the experiment to make any predictions of substantive outcomes from the project. A somewhat similar scenario seems to prevail in two more auent branches and regions. Interestingly, in at least one of the two branches, women borrowers have remained with the Grameen Trust Chiapas for a much longer period of time compared with the other four branches. At this time their higher income and expenditure, on average, might be due to this continued micronance activity, nevertheless, educational levels seem to be just as low as in the poorest branch, while it appears that health expenditures are somewhat higher. Whether women will opt for bringing their spouses in to the project, and whether this will translate into higher118Micronanceoutcomes in terms of health and education remains an open question at this stage in the program. An interesting situation exists in the other two branches and regions where the income of the borrowers is the highest. Women in both these regions are not just wealthier, but are also more educated and their health expenditures appear to be higher. Women seem to be more empowered in that they often declared themselves as being the main household head, and take most of the household decisions. Their spouses seem to be more supportive of their micro-businesses. In this situation these women might value their nancial independence, and this might be leading to a completely frictionless relationship, in which case we should probably not expect those already empowered women to actually invite their partners to join their project either. Again, it is too early to tell as we do not yet know the level of take-up of male membership, and if the invitation of spouses to join will lead to higher outcomes in terms of expenditures on health or education.CONCLUDING REMARKSAt present the baseline survey in Chiapas indicates that the degree of women empowerment is in line with Linda Mayouxs (1999) ndings in 15 dierent micronance programs in Africa. That is to say, expenditure is household- and region-specic and inexible as a consequence of social norms that seem exceedingly dicult to change. However, empowering women via an additional tool, namely by giving them the right to voluntarily invite their partners, might help to accelerate the process for change in those social norms. This might, however, prove more dicult in poorer regions where household heads seem mostly to be men. The question then is why should subsidized loans that make women responsible for repayment, but do not give them power over crucial decisions regarding their business and household expenditures in health and education seem to be endorsed by donors? Moreover, as the micronance industry becomes increasingly commercial, micro-credit becomes increasingly burdensome on women. Why should women take on the responsibility over higher repayments in the rst place? This view accords well with for-prot micronance enterprises in Latin America where men are increasingly self-selecting into programs oered by such enterprises. In the absence of subsidies, Grameen Trust Chiapas as well as other organizations in the region might be increasingly attracting men, not women. And the interest rates charged could be friendlier to women, if only because women are the main brokers of health and education within the household.Gender empowerment in micronance119As far as the more auent clients served by Grameen Trust Chiapas are concerned, it might be that the whole idea of excluding husbands can be counterproductive, because of informational asymmetries that appear to lead to mistrust, frictions within the couple, and worse, a decreased participation by men in household expenditures. This outcome is not what we understand as a preferred outcome for women. However, such disempowerment eects should be weighed against the value that women attach to their nancial independence. We see an important balance between greater nancial independence on one side, and more demands on time and loss of money for established levels of expenditure on the other. Given these scenarios women might be reluctant to invite their partners into their groups, but for dierent reasons. In the case of less auent households in very traditional societies, because having partners join the project would not change anything. And in the case of relatively more auent households because women attach too much value to their nancial independence. Final results on take-up as well as potential behavioral changes from this experiment on gender should further clarify the questions and shed light on implications of micronance structures around the exceedingly important issues of gender in the eld of micronance, development, and empowerment.NOTES1. I gratefully acknowledge the support of Alissa Fishman, Randall Blair, and Julio Luna from IPA in Mexico. Co-authors Dean Karlan and Sendhil Mullainathan have been incredibly patient in teaching me how to conduct eld work and without their guidance, my part in this chapter would never have been written. Valuable comments from seminar participants at Harvard, Columbia, Solvay Business School, and CERMi in Brussels are also greatly appreciated. Finally, the collaboration of Grameen Trust Chiapas management, and in particular, the support from Ruben Armendriz, Maricela Gamboa, branch managers, and loan ocers from Tuxtla, Ocosingo, Comitan, and Las Margaritas, have been exceedingly valuable. Nigel and I are solely responsible for the views and errors in this chapter. Daley-Harris, Sam (2003). Some evidence on this and follow-up debate is found in Mayoux (1999) and Rahman (2001), among others. This section borrows from Armendrizs joint work with Jonathan Morduch (2005). For a more general survey on gender issues in economic development, see Duo (2005). Pitt and Khandkers econometric estimations and results are, however, exceedingly controversial. For a more comprehensive critique, see Pit (1999) and Armendriz and Morduch (2005). Promoting women to powerful positions in villages and regions may, by the same token, bring social benets. In a recent paper on India, Raghabendra Chattopadhyay and Esther Duo (2003) show that by empowering women, and, in particular, by allowing2. 3. 4. 5. 6.120Micronancethem to be elected to local councils, spending on public goods most closely linked to womens concerns increased. Evidence from India also shows that there is a positive correlation between the relative size of a mothers assets (notably jewelry) and childrens school attendance and medical attention (Duraisamy and Malathy, 1991; Duraisamy, 1992). Hossain (1988): 81 percent of women had no repayment problems versus 74 percent of men. Khandker, Khalily, and Kahn (1995): 15.3 percent of male borrowers were struggling in 1991 versus 12.4 percent of female (missing some payments before the nal due date). This section draws from current eld work with Dean Karlan and Sendhil Mullainathan in southern Mexico. A year later, and under the auspices of Grameen Foundation USA, some of the Grameen Trust Chiapass managers founded AlSol, which currently serves approximately 3000 borrowers. For an explanation on random experiments, see Duo, Glennester and Kremer (2006).7. 8. 9. 10. 11. 12.REFERENCESAnderson, Siwan and Jean-Marie Baland (2002), The economics of ROSCAs and intrahousehold allocation, Quarterly Journal of Economics, 117 (3), 98395. Armendriz, Beatriz and Jonathan Morduch (2005), The Economics of Micronance, Cambridge, MA: MIT Press. Chattopadhyay, Raghabendra, and Esther Duo (2003), Women as policy makers: evidence from an India-wide randomized experiment, typescript, Cambridge, MA: MIT Economics Department. Daley-Harris, Sam (2003), State of Microcredit Campaign Report 2003, November, Washington, D.C.: Microcredit Summit, accessed at www.microcreditsummit. org/pubs/reports/socr/2003/socr 03_en.pdf. Duo, Esther (2005), Gender equality in development, Massachusetts Institute of Technology Poverty Action Lab working paper, Cambridge, MA. Duo, Esther, Rachel Glennester, and Michael Kremer (2006), Using randomization in development economics research: a toolkit, Massachusetts Institute of Technology Poverty Action Lab working paper, Cambridge, MA. Duraisamy, Paul (1992), Gender, intrafamily allocation of resources, and child schooling in South India, Yale University Economic Growth Center, working paper no. 667, New Haven, CT. Durisamy, Malathy (1998), Childrens schooling in rural Tamil Nadu: gender disparity and the role of access, parental and household factors, Journal of Educational Planning and Administration, XII (2), 13154. Engle, Patrice (1993), Inuences of mothers and fathers income on childrens nutritional status in Guatemala, Social Sciences and Medicine, 37 (11), 130312. Goetz, Anne Marie and Rina Sen Gupta (1996), Gender, power, and control in rural credit programs in Bangladesh, World Development, 24 (1), 4563. Hossain, Mahabub (1988), Credit for alleviation of rural poverty: institute research report 65, February, Washington, DC: International Food Policy Research. Khandker, Shahidur R., Baqui Khalily and Zahed Kahn (1995), Grameen Bank: performance and sustainability, World Bank Discussion Paper no. 306, Washington, DC. Mayoux, Linda (1999), Questioning virtuous spirals: micronance and womens empowerment in Africa, Journal of International Development, 11 (7), 95784.Gender empowerment in micronance121Pitt, Mark (1999), Reply to Jonathan Morduchs: Does micronance really help the poor? New evidence from agship programs in Bangladesh, typescript, Providence, RI: Department of Economics, Brown University. Pitt, Mark and Shahidur Khandker (1998), The impact of group-based credit programs on poor households in Bangladesh: does the gender of participants matter?, Journal of Political Economy, 106 (5), 95896. Rahman, Aminur (1999), Microcredit initiatives for equitable and sustainable development: who pays? World Development, 26 (12) December. Rahman, Aminur (2001), Women and Microcredit in Rural Bangladesh: An Anthropological Study of Grameen Bank Lending, Boulder, CO: Westview Press. Shultz, T. Paul (1990), Testing the neoclassical model of family labor supply and fertility, Journal of Human Resources, 25 (4), 599634. Skouas, Emmanuel (2001), Is PROGRESA working? Summary of the results by an evaluation by International Food Policy Research Institute (IFPRI), IFPRI Food Consumption and Nutrition Division discussion paper no. 118, Washington, DC. Thomas, Duncan (1990), Intrahousehold allocation: an inferential approach, Journal of Human Resources, 25 (4), 63564. Thomas, Duncan (1994), Like father like son, or, like mother like daughter: parental education and child health, Journal of Human Resources, 29 (4), 95088. Udry, Christopher (1996), Gender, agricultural production, and the theory of the household, Journal of Political Economy, 104 (5), 101046.IndexA Little World 79 administrative costs 87, 90 Africa 18, 20, 21, 11314, 118 see also individual countries agencies see aid agencies; development agencies; NGOs (nongovernmental organizations) agency costs 112 agricultural loan initiatives 4 agricultural pricing information 13 agricultural production 112 aid 111 aid agencies 5, 92 airtime credits 76, 77, 78 Allen, Hugh 113 AML (anti-money laundering) 1617, 778, 82, 8990 Anderson, Siwan 111 Asia 74 see also Asia Pacic; East Asia; South Asia; individual countries Asia Pacic 78 see also individual countries asset class 4041 assets, median total of lending institutions 20, 21 assortative matching 1112, 19 Baland, Jean-Marie 111 BancoSol 92 Bangladesh 4, 100, 10910, 113 see also Bangladesh Bank; BRAC; BRAC securitization; Eastern Bank Limited (EBL); Grameen Bank; PKSF (Palli Karma Sahayak Foundation); Proshika Bangladesh Bank 4950 bank failures 86, 934, 98, 105 banks and CDOs 31, 32 costs of access to 7 and equity 9, 37 India 72, 73, 74 and Internet 20 and mobile phone technology 1314, 20, 79, 80, 82, 83, 98 prudential regulation 86, 94, 95, 96, 978, 105 and securitization of micro-loans 35 supervision 98100, 102, 103, 105 womens empowerment 18, 19, 20 see also bank failures; commercial banks; overbanking; rural banks; rural branches of nationalized banks; village banks; individual banks barcode-reading point-of-sale (POS) terminals 1213 Barth, J.R. 105 Beam 74, 8083 BlueOrchard 27, 28, 32, 46 Bolivia 89, 912, 934 BOMSI (BlueOrchard Micronance Securities I) 2732 bonds 28, 30, 38, 50, 535, 628 borrower characteristics information 59, 6970, 73, 90, 95 borrowers 3, 78, 19, 26, 47, 48 see also borrower characteristics information; consumer protection; men; self-help groups (SHGs); solidarity groups; women; individual borrowers BRAC 479 BRAC securitization currency risk 46 future prospects 6970 performance 46, 618 political economy considerations 49 53 structure overview 35, 5360 transaction rationale 479123124 branching requirements 96 branchless banking 978 see also mobile phone technology Brazil 1213, 111 BRI (Bank Rakyat Indonesia) 94 broad borrowers 8 BTMs (biometric teller machines) 13 Burkina Faso 112Index m-commerce 78 micro-loans 878 prudential regulation 87 women as micronance customers 112 see also administrative costs; fees; transaction costs CRAB (Credit Rating Agency of Bangladesh) 50, 53 credit enhancements, BRAC securitization 556, 628 credit history 4, 8, 112 credit ratings 6, 11, 35, 38, 50, 62, 70, 89 see also CRAB (Credit Rating Agency of Bangladesh); MicroRate credit rationing 3 credit supply 4 credit unions 4, 7, 9, 18, 19, 20, 21, 94, 103 culture 2021 see also social norms currency 33 see also currency risk; local currency currency risk 33, 46, 4950 data 5660, 6970 debt/equity ratio 9 default insurance 3 default risk 11, 15, 19 defaults, loan see loan defaults delegated supervision 1034 delinquent loans and BRAC securitization 55, 56, 57, 5960, 612, 637, 68 collateralized versus uncollateralized 945 demography, India 712 deposit insurance 16, 97 deposits/savings BRAC micronance funding 48 in domestic capital markets 26 India 71, 72, 74 regulation 16, 867, 91, 92, 106 unsupervised small communitybased intermediaries 100101 and women 111, 115 development agencies 26, 27, 30, 35, 389, 40, 101 see also FMO; KfWcapital adequacy 934 capital calls 101 capital markets 5, 911 see also domestic capital markets; international capital markets capital structure 89 Caprio, G. 105 cash 14, 75, 76, 77, 111, 113, 114, 115 cash ow 6061, 62, 63, 65, 68, 70 CDOs (collateralized debt obligations) 10, 2734, 46 CFT (combating the nancing of terrorism) 17, 82, 8990 CGAP (Consultative Group to Assist the Poor) 67 child health 111 China 78 Citibank 53 Citigroup 10 co-signers as borrowers restrictions 95 collateralized loans 945 commercial banks 26, 38, 3940, 72, 74, 934 commercial investors 2931, 359 Compartamos 6, 10, 37, 88 competition, commercial and noncommercial investors 389 consumer protection 17, 889 contracts see loan contracts; puts cooperatives 18, 20 see also credit unions; self-help groups (SHGs); solidarity groups correspondent banking 1213 corruption 77, 105 costs access to credit and nancial services 69, 14 cash handling 75, 76 electronic matching of borrowers and lenders 11Index Dexia Microcredit Fund 27 domestic capital markets 10, 26, 40 Done Card 79 donor funding 48, 92 DSRAs (debt service reserve accounts) 56, 69, 70 DWM (Developing World Markets) 26, 27, 28, 2930, 31, 38, 41 East Asia 18, 20, 21 see also individual countries Eastern Bank Limited (EBL) 47, 53 eChoupals 13 economic downturns 31, 41 see also bank failures; sub-prime mortgage crisis economy, India 712 education 1089, 110, 111, 114, 116, 11718 see also nancial education; literacy levels; primary education electronic matching of borrowers and lenders 1112 Engle, Patrice 111 entrepreneurship 4, 19, 110, 116, 118 equity 9, 28, 357 Equity Bank (Kenya) 10, 37 equity capital 10 Europe 35, 37, 38 see also FMO; KfW exchange rates 50 exits 36, 37 FATF (Financial Action Task Force) 8990 fees 34, 50 nancial controls 42 see also regulation; supervision nancial education 889, 101 nancial return 34, 36, 412 nancial services 5, 79 Fino 7980 FMO 35, 40, 4950, 53 food expenditures 110, 111, 114 formal credit markets 23 see also banks formal markets 2 fraud 77 friction, marital 108, 113, 114, 115, 119 funding sources 48, 51 funds 279, 324, 38125G-cash 14 GCMF (Global Commercial Micronance Facility) 38 gender see men; women; women-only solidarity groups; womens disempowerment; womens empowerment geographic variation, womens empowerment 18, 1921, 11718 Gonzales, A. 78, 9, 18, 19 governments 5, 9, 48, 513 see also prudential regulation; supervision Grameen Bank 4, 20, 87, 108, 113 Grameen Trust Chiapas, AC (GTC) 11415, 116, 11719 group lending see self-help groups (SHGs); solidarity groups; women-only solidarity groups GSMA 748 Guatemala 111 health 2, 22, 1089, 110, 111, 114, 116, 11718 hedging 38, 3940, 44, 556 high-end borrowers 8 high-risk borrowers 3 HNWIs (high net worth individuals) 32, 38 household expenditure 110, 111, 114, 11718 household income 1, 108, 110, 111, 116, 11718, 119 housework 11415 ICICI 13, 46, 80 IFC Washington 10, 758 IFIs (international nancial institutions) 10 see also individual IFIs illiquidity 34, 41 income 1 see also household income India 4, 13, 1516, 715, 78 see also ICICI; mobile phone technology in India individual loans 8, 9, 18, 19126Index IPOs (initial public oerings) 10, 36, 37 ITC (India Tobacco Company) 13 ITZ Cash 79 JiGrahak 79 joint liability 3, 8, 74 Kenya 10, 77, 98, 111 KfW 10, 35, 50, 53 Khandker, Shahidur 10910 KIVA 12 land ownership 110 Latin America 4, 18, 19, 20, 21, 36, 87 see also Compartamos; individual countries legal vehicles see SPVs (specialpurpose vehicles) Lending Club 12 Levine, R. 105 licensing 912, 99100, 102 liquidations 28, 36 liquidity 34, 40, 44, 62 see also illiquidity literacy levels 12, 101 Little World, A 79 loan contracts 3, 8, 9, 11, 35 loan defaults 1516, 29, 42, 59, 89, 945 see also default insurance; default risk; delinquent loans loan les 42, 95, 101 loan loss provision 95 loan maturity 48 see also short-term loan contracts loan repayments and consumer protection 89 frequencies 3, 48 and inclusion of men in womens solidarity groups 114 and m-commerce 14, 76, 77 self-help groups (SHGs) in India 74 women 4, 112 loan size 1920, 26 see also broad borrowers; high-end borrowers; low-end borrowers; small borrowers local commercial banks 38, 3940, 53 see also individual banks local currency 33, 38, 3940, 46, 4950 logistics 578individual responsibility 42, 74 Indonesia 87, 94, 99100 informal credit markets 34 informal markets 3 information see agricultural pricing information; barcode-reading point-of-sale (POS) terminals; borrower characteristics information; data; eChoupals; information asymmetry; information availability; information disclosure; information sharing; Internet; loan les; MIS (management information systems); reporting; transparency information asymmetry 23, 114, 119 information availability 3 information disclosure 1415, 889 information sharing 70 insider lending 96 insiders, and equity 36 interest payment frequencies 3 interest rate caps 88 interest rates 4, 67, 11, 1416, 19, 48 9, 52, 878, 91 international capital markets amount of investment 25 asset class, on the path to 4041 CDOs versus funds 324 commercial investors, growing participation in CDOs 3032 commercial investors, introducing to CDOs 2930 and concerns about micronance 413 equity 357 from funds to CDO 279 importance 257 and local currency 33, 38, 3940 non-commercial investors 26, 27, 379 and securitization of micro-loans 345 international money transfer services 81, 82, 83 Internet 1112, 13, 20, 70, 79 intimidation 89 IPA (Innovations for Poverty Action) 11618Index long-term investment 279 low-end borrowers 7, 8 low-risk borrowers 3 m-commerce 758, 79 marriage rates 115 Mayoux, Linda 11314, 118 MChek 79 men 108, 110, 11113, 11418, 119 Mexico 111, 11415, 11619 MF Analytics 53, 56 MFBA (Micronance Bank of Azerbaijan) 38 MFIs (micronance institutions) borrowers, global number of 26 capital adequacy 934 costs 878 India 724 loan size 1920, 26 mission drift 423 ownership requirements 97 regulation see regulation size 100 stock exchange listing 10 supervision 99104 sustainability 92 and womens empowerment 18, 19, 2021 see also banks; NBFIs (non-bank nancial institutions); individual MFIs MFRC (Micro Finance Regulatory Council) 89 micro-credit backed securities 5 micro-insurance 81, 82, 83 micro-investment 10, 81, 82, 83 see also international capital markets micro-payments 82, 83 micronance, evolution 46, 9, 108 Micronance Securities XXEB 26, 31 MicroRate 6, 389 minimum capital requirements 93, 99, 100 MIS (management information systems) 51, 53, 568, 69, 70 mission drift 423 MIX data 6, 78, 9, 1821, 878, 92 mobile phone technology and banks 1314, 20, 79, 80, 82, 83, 98 and BRAC securitization 70127India see mobile phone technology in India and m-commerce 758 mobile phone technology in India future and challenges 813 industry 7980 and micronance 745, 78 products 8081 usage 72 money see AML (anti-money laundering); cash; cash ow; CFT (combating the nancing of terrorism); currency; fees; loan repayments; loan size; money laundering money laundering 77 see also AML (anti-money laundering) Morgan Stanley 32 NBFIs (non-bank nancial institutions) 5, 7, 9, 18, 19, 20, 21 see also credit unions; PACS (primary agriculture credit societies); ROSCAs (Rotating Savings and Credit Associations); self-help groups (SHGs); village banks NGOs (non-governmental organizations) and capital calls 101 and capital markets 1011 costs of access to nancial services 7, 9 in evolution of micronance 5, 9 ownership requirements 97 and womens empowerment 18, 19, 20, 21 see also individual NGOs non-commercial investors 26, 27, 30, 32, 379 non-prot organizations 26, 334 see also aid agencies; credit unions; development agencies; NGOs (non-governmental organizations); PACS (primary agriculture credit societies); ROSCAs (Rotating Savings and Credit Associations); self-help groups (SHGs); village banks128 non-prudential regulation 17, 86, 87 90, 103 Omidyar, Pierre 5 OPIC (Overseas Private Investors Corporation) 30 OPORTUNIDADES 111 origination risk 35 over-collateralization 638, 6970 overbanking 33, 412 ownership requirements 97Index prudential regulation adjusting regulations to t products and institutions 928 critique 1056 versus non-prudential regulation 867 regulation as promotion 9091 regulation that follows the market 912 puts 36 PWC (PriceWaterhouseCoopers) 53, 69 RBI (Reserve Bank of India) 72 regulation deposits/saving 16, 867, 91, 92, 106 formal markets 2 infrastructure problems 35 interest rates 1416 and m-commerce 778 and mobile phone technology in India 82 non-prudential regulation 17, 86, 8790, 103 prudential regulation (see prudential regulation) self-regulation 104 see also peer monitoring; supervision repayment capacity 89 reporting 53, 56, 70, 96, 103, 104 risk BRAC securitization 69, 70 CDOs (collateralized debt obligations) 2730, 31, 32, 33 and costs of access to nancial services 89 group lending and self-help groups (SHGs) 3 non-commercial investors as shield for commercial investors 378 and overbanking 33, 412 see also currency risk; default risk; high-risk borrowers; low-risk borrowers; origination risk; prepayment risks; risk analysis; risk aversion; risk/return ratios; security risks risk analysis 5861, 70 risk aversion 26, 70, 112 risk/return ratios 2930PACS (primary agriculture credit societies) 72 Pakistan 91 pawnbrokers 4 pay-down structure 545 payday lending 4 Paymate 79 payment systems 745 peer monitoring 14, 15, 18 peer pressure 15, 74, 112 performance, BRAC securitization 46, 618 Philippines 14, 768, 98, 99100 physical security 96 Pitt, Mark 10910 PKSF (Palli Karma Sahayak Foundation) 48, 51, 53, 54 political economy 4953, 70, 92, 99, 104, 105 pool maintenance, BRAC securitization 53, 56 poor persons 23, 5, 22, 100101, 110 see also poverty alleviation; poverty measurement Portio Research 78 poverty alleviation 78, 110, 111 poverty measurement 1 prepayment risks 556, 6061, 69 prepayments, and BRAC securitization 556, 57, 6061, 623, 647, 68, 69 primary education 1, 22 ProCredit Bank Bulgaria 35 ProCredit Holding AG 5, 35 ProFund Internacional SA 36 promotion of micronance 9091 Proshika 523 Prosper.com 11, 12Index ROSCAs (Rotating Savings and Credit Associations) 111 RRBs (regional rural banks) 72, 73, 74 rural areas 71, 72 see also RRBs (regional rural banks); rural banks; rural branches of nationalized banks; village banks; village Internet rural banks 99100 see also RRBs (regional rural banks); rural branches of nationalized banks; village banks rural branches of nationalized banks 5, 7, 9, 72, 73 sachet purchasing 767 savings see deposits/savings SCBs (scheduled commercial banks) 72, 74 Schultz, Paul 111 secondary markets 32, 43 Securities and Exchange Commission (SEC) 50 securitization of micro-loans 5, 10, 345, 46 see also BRAC securitization; CDOs (collateralized debt obligations) security risks 74, 90, 91, 100101 see also physical security self-employment 74 self-help groups (SHGs) 3, 14, 19, 734 see also solidarity groups; village banks; women-only solidarity groups self-regulation and supervision 104 servicers 35 see also BRAC securitization short-term investment 27, 50, 53, 545 short-term loan contracts 3, 11, 35 Skouas, Emmanuel 111 small borrowers 73, 87, 90, 91, 100101 see also low-end borrowers small community-based intermediaries 100101 SMS (short message service) 78, 79, 8081129social norms 112, 11314, 118 social pressure 15, 74, 112 socially responsible investors see noncommercial investors soft capital 5 solidarity groups 3, 8, 9, 14, 15, 18, 19 21 see also women-only solidarity groups solvency 934 South Africa 1314, 77, 89, 90, 91, 98 South Asia 18, 19, 20, 21 see also individual countries SPVs (special-purpose vehicles) 28, 53, 54, 55, 628 stand-alone loans see individual loans Stephens, B. 1011 stock exchange listing 10 stop-lending order 102 sub-prime mortgage crisis 32 substitution, and BRAC securitization 55, 56, 63, 647, 68 Superintendency of Banks (Bolivia) 89, 92 supervision 96, 98106 sustainability of MFIs 92 Suvidha 74, 80, 813 swaps 40 SWIFT 74, 80, 81 syndication 48, 49 Tanzania 91 taxation 7, 82 technological innovation 5, 1114, 1517 technology 2, 8, 22 see also Internet; mobile phone technology; technological innovation Thailand 111 Thomas, Duncan 111 TIAA-CREF 5 transaction costs 2, 7, 73, 78, 112 transparency 30, 33 trustees 28, 30, 47, 53 Udry, Christopher 112 uncollateralized loans 945 underclass 76 United Kingdom 11, 12, 111130 United States 4, 11, 12, 32 see also IFC Washington; MicroRate; OPIC (Overseas Private Investors Corporation) unsecured lending limits 94 USAID 14 usury limits 878 village banks 5, 8, 9, 1819 see also rural banks; rural branches of nationalized banks village Internet 13 violence 89 volatility 41, 489Index geographic variation 18, 1921, 11718 and health 108, 110, 111, 114, 116, 11718 and household expenditure 110, 111, 114, 11718, 119 and household income 108, 111, 116, 11718 and housework 11415 individual versus solidarity group lending 1819, 21 and lending institutions 18, 19, 2021, 108 and level of aggregation 2021 loan size 1920 and marital friction 108, 113, 114, 115, 119 measurement 11519 micronance in India 72 SHG (self-help group) lending in Asia 74 and social norms 112, 11314, 118 World Bank 1, 74, 110 Yanus, Muhammad 4, 113 Zambia 92 Zopa.com 11, 12Wallet 365 79 women 4, 111, 112, 115 women-only solidarity groups 11419 womens disempowerment 1089, 113, 11519 womens empowerment agricultural inputs and productivity 112 assets 20, 21 and education 108, 110, 111, 114, 116, 11718 entrepreneurship 4, 19, 110, 116, 118 in evolution of micronance 4, 108