The Theory and Practice of Microcredit

Wahiduddin Mahmud, S. R. Osmani

• Publisher: Routledge
• Year: 2016
• ISBN:
  9781315413150
• Edition: 1
• Topic: Business
• Subtopic: Business General
• Index: Business

The remarkable speed at which microcredit has expanded around the world in the last three decades has piqued the curiosity of practitioners and theorists alike. By developing innovative ways of making credit available to the poor, the idea of microcredit has challenged many traditional assumptions about both poverty reduction strategies and financial markets. While this has encouraged new theorising about how microcredit works, the practice of microcredit has itself evolved, often in unpredictable ways, outpacing the development of theory.

The Theory and Practice of Microcredit aims to remedy this imbalance, arguing that a proper understanding of the evolution of practice is essential both for developing theories that are relevant for the real world and for adopting policies that can better realize the full potential of microcredit. By drawing upon their first-hand knowledge of the nature of this evolution in Bangladesh, the birthplace of microcredit, the authors have pushed the frontiers of current knowledge through a rich blend of theoretical and empirical analysis. The book breaks new grounds on a wide range of topics including: the habit-forming nature of credit repayment; the institutional strength and community-based role of microfinance institutions; the relationships between microcredit and informal credit markets; the pattern of long-term participation in microcredit programmes and the variety of loan use; the scaling up of microenterprises beyond subsistence; the "missing middle" in the credit market; and the prospects of linking micro-entrepreneurship with economic development.

The book will be of interest to researchers, development practitioners and university students of Development Economics, Rural Development, or Rural Finance, as well as to public intellectuals.

1 Introduction

Money, says the proverb, makes money. When you have got a little, it is often easy to get more. The great difficulty is to get that little.

Adam Smith, The Wealth of Nations¹

Lawmakers and judges at all ages were tormented on the issue of recovery of unsecured loans that represent contracts freely arrived at and yet cannot be enforced without ‘oppression’.

John Hicks, The Theory of Economic History ²

Microcredit has emerged as a hugely popular tool all over the developing world for helping poor people to help themselves by engaging in self-employed, income-generating activities. Very few poverty intervention programmes have ever managed to attain a similar scale of operation. Although one may debate the extent to which microcredit can transform the livelihoods of the poor, the simple fact that it has managed to reach credit to millions of poor people hitherto bypassed by the traditional banking system is a remarkable achievement. By developing innovative ways of making the poor creditworthy, the ‘microcredit revolution’, as it has come to be called, has seriously challenged many traditional assumptions about poverty reduction strategies on the one hand and financial markets on the other. While this has encouraged new theorising about how microcredit works, the practice of microcredit has itself evolved, often in unpredictable ways, outpacing the development of theory. The idea of this book is based on the premise that a proper understanding of this evolution in practice of microcredit is essential for both developing theories that are relevant for the real world and adopting policies that can better realise the full potential of microcredit. Studying the growth and evolution of microcredit in Bangladesh is of particular interest since no other country can match its experience regarding the maturity and the extent of outreach of the microcredit programmes.

1.1 How microcredit works: the theories and their empirical tests

The remarkable speed at which the reach of microcredit has expanded around the world in the last three decades has piqued the curiosity of practitioners and theorists alike. Development practitioners want to know what that something special is about microcredit so that they, too, can practise it in their own environment, with necessary adaptations. Theorists want to know why it is that the special features of microcredit seem to work, at least in terms of embracing those who were previously excluded from the formal credit network, without compromising the financial viability of the lenders. This has led to an enormous outpouring of theoretical speculations, often drawing upon the latest theoretical advances in economics and finance. The ideas and tools developed in the economic theories of imperfect information, and the related theories of screening, incentives and mechanism design − which are themselves of fairly recent origin – as well as the tools of game theory have been enthusiastically applied by a new generation of economists to unearth the secrets of microcredit.³ At the same time, testing these theories with field-level data and experiments regarding how the microcredit model actually works has led to a new area of academic research, even though such empirical tests have not quite kept pace with new theorising.

The main features of the classic microcredit model pioneered by Grameen Bank in Bangladesh and replicated worldwide are well-known from the literature. It is a system of group-lending under which clients from poor families, predominantly women, form groups to receive collateral-free small loans that are repaid in regular weekly or frequent instalments, usually within a one-year loan cycle and for which the group bears joint responsibility. While, in theory, the model can be shown to solve a number of problems that historically plagued government efforts of providing credit to the poor, the relative importance of the various features of the model remains largely conjectural.

The problem is that there is no single theory of how microcredit works; what we have instead is a bewildering variety of theories, and the only common thread binding most of them is the idea that the way microcredit is delivered in practice helps overcome certain market imperfections – in particular, imperfections in information and in the enforcement of contract. But they differ greatly in their understanding of exactly which imperfections are being addressed and precisely how they are being overcome. Most of these theories have some a priori plausibility. Therefore, the only way to discriminate among them is to check empirically which theories seem to fit the reality better than others. Accordingly, researchers have increasingly turned their attention towards testing microcredit theories against empirical data. (The major theories of microcredit are discussed in Chapters 4 and 5, and empirical tests of theories are reviewed in Chapter 6.) The problem here faced by the researchers is that there may not be enough variations in the basic features of the existing lending modalities among the microfinance institutions (MFIs) in a country so as to provide data for econometric tests of the efficacy of particular lending mechanisms (Morduch, 1999, p.1586). Nevertheless, researchers are trying to get around the problem by using proxy information lying behind the actual working of the mechanisms and applying sophisticated econometric techniques, and also by drawing upon the tools of experimental behavioural economics. While this is currently a lively area of research already providing many valuable insights, this literature should not be confused with the more familiar and increasingly large body of studies aimed at empirically measuring the impact of microcredit on poverty.

Besides the econometric approach to analysing how microcredit works, another possible way is to look at the process of evolution of the system as the practitioners tried to overcome the limitations of the model and responded to the changing and varied needs of the borrowers. Bangladesh has now got a long enough history of experimentation with microcredit programmes for allowing such a Darwinian approach to analysing how the system has evolved through numerous innovations, mutations and adaptations. The classic Grameen model, which has been replicated worldwide, has evolved considerably in Bangladesh since its inception in the late 1970s. Other than Grameen Bank, the microcredit industry in Bangladesh is driven by numerous non-governmental institutions (NGOs) which were initially established for ‘social mobilisation’ of the rural communities along with service delivery, but later on they shifted their emphasis to the provision of microcredit to the poor – hence now commonly known as NGO-MFIs. While some of the defining features of the original Grameen model still remain widely prevalent, there now exists a wide range of variations in the lending practices, particularly among the relatively larger MFIs, with varying degrees of resemblance to the original model. More importantly, Grameen Bank itself overhauled its own lending system in 2001 with the introduction of what it calls Grameen II, allowing far more flexibility for its clients in accessing and repaying loans. An important feature of this evolutionary process is the gradual dilution of the rigorous discipline that the MFIs used to impose on borrowers through the rigid lending modalities, particularly through the system of joint liability of the group. But despite this dilution, the high loan recovery rates of MFIs do not seem to have suffered, while the credit needs of their clients are also now better served. How has this been possible?

Besides reviewing the limited amount of available literature specifically focusing on this topic, in this book we try to answer this question by drawing upon the insights of the microcredit practitioners who know from their experience how these adaptations, mutations and modifications of the original model have actually come about. (The place of microcredit in the overall rural credit market is discussed in Chapter 2 and the evolution of the microcredit model as practised in Bangladesh is analysed in Chapter 3.) In addition, we look at some recent survey findings focusing on the experience and opinion of the borrowers themselves about the evolving microcredit delivery system in Bangladesh. These findings come from two rounds of nationwide surveys of rural households: the Poverty Dynamics Survey carried out by the Institute of Microfinance (InM) in Dhaka in 2010 and 2013. A part of the later round of the survey, based on a sub-sample of households, was specifically designed to focus on the working of the microcredit system.

While group liability is the kingpin of how microcredit works in theory, there has been a gradual decline in the enforcement of group liability in all its usual forms: putting pressure on peers to obtain the weekly instalments one way or the other, or threatening to withhold loans to peers unless default was prevented, or actually cutting off loans to peers in case of actual default. Most MFIs in Bangladesh now follow a flexible interpretation of joint liability, namely, to expect the group members to help assess the defaulting member’s actual capacity to repay and to apply pressure to recover as much loan as is possible before letting her go. Another deviation from theory is that the system of group lending, to the extent that it works in Bangladesh, is based on mutual help and trust among group members rather than the threat of applying social sanctions on the defaulting group member by her peers. In this latter respect, the microcredit movement in Bangladesh seems to have benefited from a kinship-based rural social structure in dense settlements that has helped in forming the microcredit groups from the existing networks of relatives, friends and neighbours.

Many authors have pointed out that the role of group lending contracts in ensuring high repayment rates is overplayed in the theoretical models; instead, they point to other important features that explain the successful performance of the microcredit programmes, such as the proactive role played by the loan officers of MFIs, the ‘dynamic incentive’ created by threat of cutting off future loans, the open public meetings, and the predominance of female clients who are particularly susceptible to public shaming (Jain and Moore, 2003; Armendáriz and Morduch, 2010, pp.137–62). In the case of Bangladesh, there have been two other reinforcing factors, namely, the socially embedded role of NGO-MFIs, who deliver a variety of non-credit services and have been able to assume a ‘social guardian’ role within the communities they serve; and the long experience of the MFI clients helping them to realise that it is worthwhile to invest in a long-term relationship with MFIs.

But the most unique feature of Bangladesh’s experience is the evidence that loan repayment can be habit-forming and that high repayment rates maintained over a long period can lead to a social norm and culture of repayment, thus creating a social stigma attached to loan default (Mahmud, 2003; Osmani, 2015b). In the theoretical models, and also in the actual practice of repayment enforcement in some contexts, the determining factors are said to be peer pressure or sanctions along with a permissible degree of coercion. The situation is quite reversed when the repayment can be explained by the cost of non-repayment from the point of view of the borrower; the cost would include, besides the foregone benefits of future loans, the psychological cost of the social stigma and the sense of guilt from breaking a trust-based relationship. This has several implications, both for theory and practice. For the practitioners, it implies that there will be less need for coercion for repayment enforcement and for monitoring of the loan use to prevent imprudent loan use. It also makes it easier to relax the repayment modalities, like weekly instalment payments, which are major constraints in the choice of loan-financed projects and for poor households to participate in the microcredit programmes. From the point of view of theory, the implications are even deeper and more nuanced.

There is an obvious problem of moral hazard arising from the risk that the borrowers will use the loans ‘imprudently’ for non-productive purposes under social or economic pressures, leading to problems of loan recovery. However, most of the theoretical constructs are based on another, more subtle and nuanced kind of ex ante moral hazard, which arises from the so-called ‘limited liability’ of the borrower while using collateral-free loans for a production project, and which can lead to a systematic divergence between the objectives of the borrower and the lender regarding the use of the loan. In theory, the ‘limited liability’ assumption is taken to imply that the borrower has to repay the loan only if the project is successful, but not if the project fails. This can lead to the borrower’s deliberate ex ante strategy of undertaking too-risky projects, since she is cushioned against the underside of her risky behaviour, which can be passed off to the lender. The result can be an inefficient choice of projects or too little effort expended on the project and a higher risk of repayment default. However, these sources of perverse incentives for preferring too-risky projects or putting in too little effort will disappear as soon as the ‘limited liability’ assumption is replaced by that of ‘full liability’ – that is, if the borrower could somehow be made to repay the loan under every contingency regardless of the success or failure of projects (or, as if the borrower is investing out of her own funds).

The applicability of these theoretical results, however, will depend on how the borrower actually perceives her liability for the loan, which in turn will depend on her assessment of the repayment enforcement mechanisms employed by the MFIs and(or) her own mental calculation regarding the cost of non-repayment. It is true that a collateral-free loan cannot be recovered if the borrower household does not possess the wherewithal or the necessary cash flows to repay the loan; but the question is whether a borrower interprets this as her own limited liability for the loan while making decisions about the use of the loan. It is easy to see how the nuanced form of moral hazard disappears if the cost of non-repayment is high enough to ensure repayment.⁴ Even if a coercion-based repayment system works as well as the one based on inducement and moral obligation in addressing the problem of moral hazard, the latter is obviously far superior in promoting the social mission of MFIs and in strengthening the resilience of the system.

The issues regarding loan recovery by ‘moral rather than material sanctions’ as well as those involving limited or unlimited liability of loans are reminiscent of a debate around the lending modalities of rural co-operative societies in the colonial Bengal (Islam, 1978, p.175). According to the so-called Maclagan Committee, which was appointed to examine these issues, the principle of unlimited liability for the loans given to poor farmers could be meaningful only if ‘honesty and moral obligations’ could be a substitute for material assets. The failure of the co-operatives, according to the Committee, was because of the failure in materialising what it called a ‘romantic’ approach of unlimited liability. The success of microcredit in Bangladesh lies in no small way in finding ways of converting what was once considered a romantic approach into a realistic one. This success is also remarkable given the age-old problem of how to recover, without coercion, loans given to the poor, as mentioned in the quote from John Hicks with which we began this chapter and the book.

Overall, the microcredit system in Bangladesh has come a long way from its reliance on group liability and rigid repayment modalities to a much more flexible system that can better meet the needs of the clients, such as by making repayment schedules match with income flows from projects, or helping clients to manage livelihood vulnerabilities, or by providing larger-sized loans to progressive borrowers. The extent of these developments is often underestimated by academics – for example, when they think that the improvements to the original Grameen model ‘will not happen on its own’ and will need ‘tweaking, tinkering and testing’ until the lending system can serve various needs of financial services of the poor without, say, ‘forcing one to pay for the other’ (Karlan and Appel, 2011, p.113).

1.2 Measuring the poverty impact of microcredit

The part of the microcredit literature that has stirred the most controversies, debates and emotions has to do with the assessment of the poverty impact of microcredit. The early studies on the impact of microcredit almost invariably found that microcredit made a positive contribution not only in reducing poverty, but also in a host of other economic and social dimensions. These studies soon came to be questioned, however, on the grounds of econometric methodology. Doubts were raised about whether the methodologies employed in them were able to correctly identify the causal effect of credit on economic outcomes because of what is known in the econometric literature as the ‘identification problem’. It was argued, in particular, that various kinds of ‘selection bias’ vitiate the findings of the studies and lend an ‘upward bias’ to the estimates of the impact of microcredit. The bias may arise from programme placement when MFIs deliberately go for areas known as less ‘risky’ or when individuals with better entrepreneurial ability may ‘self-select’ themselves into the MFI’s programme.

A number of subsequent studies have tried to address this problem by using appropriate econometric techniques, and in general they confirm the existence of the beneficial impact found by the early studies – although the magnitudes of benefit may differ. These studies have been much more conscious of the likely biases in impact estimation, and the search for an appropriate identification strategy has been at the heart of their research methodology. However, questions were raised about the findings of these second-generation studies as well, resulting in a prolonged and sometimes obscure debate on econometric methodology. Indeed, one sometimes gets the feeling that the methodological concern with identification has been all-consuming, often superseding the concern with substance. An extreme version of the critique has recently emerged, however, which holds that the kind of observational data on which these studies are based are fundamentally incapable of allowing a satisfactory solution of the identification problem. According to this view, what is needed are experimental data generated by randomised controlled trials (RCTs). Such experimental data have indeed been generated over the last decade to assess the impact of microfinance in a number of developing countries, and they generally fail to find any significant developmental impact. Although Bangladesh is not included in these countries, the negative findings of these studies – pertaining to a wide variety of contexts – have cast serious doubt on the validity of the findings of beneficial impact of microcredit.

Randomised experiments with microcredit face a dilemma. If the objective is to test hypotheses about microcredit’s presumed ability to transform the lives of borrowers, one cannot rely on experiments carried out over a short time span; but then experiments carried out over longer time spans are difficult to implement due to practical reasons. The short-run impact is not only likely to be small, but it also will be heterogeneous depending on the use of the loan. Setting up a controlled experiment for measuring the long-run impact is difficult because of keeping the comparator group of non-participant households from joining a microcredit programme or other competing programmes of poverty alleviation. It is not easy to see how this dilemma can be resolved.⁵ Under the circumstances, the best the researchers can do is to work with observational data, recognise that there are inherent problems with such data arising, in particular, from the existence of selection bias of the kind discussed earlier, and to do their best to tease out the causal connections by using appropriate econometric methodologies. It is true that identification of causal connections in this manner would never be as precise as would be the case with an ‘ideal’ randomised experiment, but we may have to settle for the second best since an ‘ideal’ microcredit experiment for capturing the poverty impact of microcredit in its...

Table of contents

• Cover
• Title
• Copyright
• Dedication
• Contents
• List of figures
• List of tables
• Preface
• 1 Introduction
• 2 Microcredit in Bangladesh: how the credit markets work
• 3 Microcredit in Bangladesh: how the microcredit model works
• 4 Theories of microcredit: group lending and moral hazard
• 5 Theories of microcredit: adverse selection and repayment enforcement
• 6 When theory meets reality: testing the theories of microcredit
• 7 Economic impact of microcredit: the experience of Bangladesh
• 8 The patterns of loan use
• 9 The economics of microenterprise
• 10 Micro-entrepreneurship and economic development
• References
• Index