Handbook of Microfinance, Financial Inclusion and Development 9781789903874, 1789903874

Citation preview

HANDBOOK OF MICROFINANCE, FINANCIAL INCLUSION AND DEVELOPMENT

ELGAR HANDBOOKS IN DEVELOPMENT The Elgar Handbooks in Development series is a collection of works edited by leading international scholars within the field. The series provides an overview of recent research in all aspects of Development Studies, thereby forming an exhaustive guide to the field. These Handbooks aim to be prestigious, high-quality works of lasting significance, discussing research areas including the politics of international development, the global economic impacts of development, and the challenges faced by those driving development, on both a national and international scale. Each Handbook will consist of original contributions by leading authors that aim both to expand current debates within the field, and to indicate how research in Development Studies may progress in the future. This series will form an essential reference point for all students of Development Studies. Titles in the series include: Research Handbook on Democracy and Development Edited by Gordon Crawford and Abdul-Gafaru Abdulai Handbook of Communication and Development Edited by Srinivas Raj Melkote and Arvind Singhal Handbook of Development Policy Edited by Habib Zafarullah and Ahmed Shafiqul Huque Handbook on the Politics of International Development Edited by Melisa Deciancio, Pablo Nemiña and Diana Tussie Handbook on the Governance of Sustainable Development Edited by Duncan Russel and Nick Kirsop-Taylor Handbook on Governance and Development Edited by Wil Hout and Jane Hutchison Handbook of Microfinance, Financial Inclusion and Development Edited by Valentina Hartarska and Robert Cull

Handbook of Microfinance, Financial Inclusion and Development

Edited by Valentina Hartarska Department of Agricultural Economics and Rural Sociology, and Department of Finance, Auburn University, USA

Robert Cull Development Research Group, The World Bank, USA

ELGAR HANDBOOKS IN DEVELOPMENT

Cheltenham, UK · Northampton, MA, USA

© International Bank for Reconstruction and Development/The World Bank 2023 The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. 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 USA A catalogue record for this book is available from the British Library Library of Congress Control Number: 2022950616 This book is available electronically in the Economics subject collection http://dx​.doi​.org​/10​.4337​/9781789903874

ISBN 978 1 78990 386 7 (cased) ISBN 978 1 78990 387 4 (eBook)

EEP BoX

Contents

vii

Editors and contributors PART I 1

INTRODUCTION Overview of microfinance, financial inclusion, and development Robert Cull and Valentina Hartarska

PART II

2

CONCEPTUAL FRAMEWORKS FOR MICROFINANCE, FINANCIAL INCLUSION, AND DEVELOPMENT

2

Rethinking poverty, household finance, and microfinance Jonathan Morduch

3

Assessment of microfinance institutions and their impact: evidence from a scientometric study Begoña Gutiérrez-Nieto and Carlos Serrano-Cinca

4

Financial inclusion and gender Isabelle Guérin

5

Toward a theory of fair interest rates on microcredit: balancing the needs of clients and institutions Marek Hudon and Joakim Sandberg

21

41 66

6

Resilience in emergencies, savings, and credit Saniya Ansar, Jake Hess, and Leora Klapper

7

When is financial education successful? Taking stock of the new wave of field evidence Bilal Zia

83 99

119

PART III DELIVERING FINANCIAL SERVICES TO CLIENTS 8

Group lending in theory and practice Christian Ahlin and Godwin Debrah

135

9

Alternative delivery channels and impacts: agent banking Sinja Buri, Robert Cull, and Xavier Giné

150

10

Digital financial inclusion and development Greta Bull and Leora Klapper

164

11

Building inclusive value chains for smallholders: the role of finance Alan de Brauw and Johan Swinnen

181

v

vi  Handbook of microfinance, financial inclusion and development

12

Index insurance for developing countries: a primer Mario J. Miranda and Denis Nadolnyak

194

PART IV VIEW FROM PRACTITIONERS AND FUNDERS 13

14

Measuring the evolution of client vulnerability: innovation at the BBVA Microfinance Foundation Claudio Gonzalez-Vega, Laura Mo, and Giovanni di Placido

212

An investor’s perspective on measuring and managing social performance and impact Gregor Dorfleitner, Dina Pons, and Noémie Renier

248

PART V

EVIDENCE FROM REGIONS AND COUNTRIES

15

Financial inclusion in high-income countries: gender gap or poverty trap? Anastasia Cozarenco and Ariane Szafarz

272

16

Financial literacy and the use of financial services by US households James R. Barth, Valentina Hartarska, Jitka Hilliard, and Nguyen Nguyen

297

17

Financial inclusion, microfinance, and financial education in Latin America Alejandro Javier Micco Aguayo and Patricio Andrés Valenzuela Aros

316

18

Gender and financial inclusion in Latin America and the Caribbean Victor Motta

339

19

Inclusive finance and agricultural development in Africa Calum G. Turvey and Apurba Shee

357

20

Evaluating digital financial inclusion: a Kenyan perspective on morality and finance Susan Johnson and Silvia Storchi

21

Inclusive finance and inclusive rural transformation in China Calum G. Turvey

22

Does microfinance cause banking sector development and economic growth? An application to Mongolia Batkhuyag Myagmar, Robert Lensink, and Wim Heijman

23

Index

Financial inclusion and poverty: evidence from Armenia Aleksandr Grigoryan, Knar Khachatryan, Knarik Ayvazyan, and Pundarik Mukhopadhaya

383 402

425 449

467

Editors and contributors

EDITORS Valentina Hartarska is an Alumni Professor in the Department of Agricultural Economics and Rural Sociology and the Department of Finance at Auburn University, USA. Robert Cull is the Research Manager for Finance and Private Sector Development in the Development Research Group of the World Bank and the Co-Director of the World Development Report 2021: Data for Better Lives.

CONTRIBUTORS Christian Ahlin is a Professor in the Department of Economics at Michigan State University, USA. Saniya Ansar is an Economist in the Development Research Group at the World Bank. Knarik Ayvazyan works at the Central Bank of Armenia and is a Scholar at the Fletcher School of Law and Diplomacy, USA. James R. Barth is a Professor and Eminent Scholar of Finance in the Department of Finance at Auburn University, USA. Greta Bull is the Director of Women’s Economic Empowerment at the Bill & Melinda Gates Foundation and formerly the CEO of CGAP (the Consultative Group to Assist the Poor). Sinja Buri is a Project Manager at the Munich Climate Insurance Initiative, which is hosted at the United Nations University Institute for Environment and Human Security, Germany. Anastasia Cozarenco is an Associate Professor in Economics at the Montpellier Business School, France, and the Centre for European Research in Microfinance (CERMi), Belgium. Alan de Brauw is a Senior Research Fellow at the International Food Policy Research Institute, USA. Giovanni di Placido is the Director of Analyses and Studies at the BBVA Microfinance Foundation. Gregor Dorfleitner is a Professor in the Department of Finance at the University of Regensburg, Germany, an associated researcher of the Centre for European Research

vii

viii  Handbook of microfinance, financial inclusion and development

in MIcrifinance (CERMi), Belgium, and an Honorary Fellow of the Hanken Centre for Accounting, Finance and Governance, Hanken School of Economics, Finland. Xavier Giné is a Lead Economist in the Finance and Private Sector Development Team of the Development Research Group at the World Bank. Godwin Debrah is a Lecturer at the University of Ghana. Claudio Gonzalez-Vega is a Professor Emeritus in the Department of Agricultural, Environmental, and Development Economics at the Ohio State University, USA, and a Trustee of the BBVA Microfinance Foundation. Aleksandr Grigoryan is an Associate Professor at the American University of Armenia, Armenia, and a CERGE-EI Foundation Teaching Fellow in Prague, Czech Republic. Isabelle Guérin is a Senior Research Fellow at the French National Research Institute of Sustainable Development (IRD) at CESSMA/Université de Paris Cité, affiliated with the French Institute of Pondicherry, France. Begoña Gutiérrez-Nieto is a Professor of Accounting and Finance at the University of Zaragoza, Spain. Wim Heijman is a Professor in the Department of Economics at the Czech University of Life Sciences, Prague, Czech Republic, and at Wageningen University, the Netherlands. Jake Hess is an independent consultant. Jitka Hilliard is the J.K. Lowder Family Professor in the Department of Finance at Auburn University, USA. Marek Hudon is a Professor at the Solvay Brussels School of Economics and Management (SBS-EM), Université Libre de Bruxelles (ULB), Belgium, the Co-Director of the Centre for European Research in Microfinance (CERMi), Belgium, and the Director of the Centre d’Etudes Economiques et Sociales de l’Environnement (CEESE), Belgium. Susan Johnson was formerly Associate Professor and Director of the Centre for Development Studies at the University of Bath, UK. Knar Khachatryan is an Associate Professor at the American University of Armenia, and an Associate Researcher at the Centre for European Research in Microfinance (CERMi), Belgium. Leora Klapper is a Lead Economist in the Development Research Group at the World Bank.

Editors and contributors 

ix

Robert Lensink is a Professor of Finance and Vice Dean for Research in the Faculty of Economics and Business at Groningen University, the Netherlands, and a Professor at Wageningen University, the Netherlands. Alejandro Javier Micco Aguayo is a Professor in the Department of Economics at the University of Chile. Mario J. Miranda is a Professor in the Department of Agricultural, Environmental and Development Economics at the Ohio State University, USA. Laura Mo is a Data Scientist in Client Solutions and Digital Innovation at the BBVA Microfinance Foundation. Victor Motta is an Associate Professor at SKEMA Busines School, Brazil, and Université Côte D’Azur, Nice, France. Jonathan Morduch is a Professor of Public Policy and Economics and the Executive Director of the Financial Access Initiative at New York University, USA. Pundarik Mukhopadhaya is a Professor in the Economics Department of the Macquarie Business School, Macquarie University, Australia. Batkhuyag Myagmar is a Professor at the Mongolian University of Life Sciences, Mongolia, and the Wageningen University and Research, the Netherlands. Denis Nadolnyak is a Professor in the Department of Agricultural Economics and Rural Sociology at Auburn University, USA. Nguyen Nguyen is an Assistant Professor in the Department of Finance at Minnesota State University Mankato, USA. Dina Pons is a Managing Partner at Incofin Investment Management. Noémie Renier is a Partner at Incofin Investment Management. Joakim Sandberg is a Professor of Practical Philosophy at the University of Gothenburg, Sweden, the Director of the Financial Ethics Research Group in Gothenburg, and the Vice Director of the Sustainable Finance Lab. Carlos Serrano-Cinca is a Professor of Accounting and Finance at the University of Zaragoza, Spain. Apurba Shee is an Associate Professor of Applied Economics at the University of Greenwich, UK.

x  Handbook of microfinance, financial inclusion and development

Silvia Storchi works in the Inclusive Rural Transformation and Gender Equality Division (ESP) of the Food and Agriculture Organization of the United Nations (FAO). Johan Swinnen is the Director General of the International Food Policy Research Institute, USA. Ariane Szafarz is a Professor of Finance at the Solvay Brussels School of Economics and Management (SBS-EM) of the Université Libre de Bruxelles (ULB), Belgium, Co-Director of the Centre for European Research in Microfinance (CERMi), Belgium, and Co-Director of the SBS-EM Doctoral Programme in Management Sciences, Belgium. Calum G. Turvey is the W.I. Myers Professor of Agricultural Finance at Cornell University, USA. Patricio Andrés Valenzuela Aros is an Associate Professor at the Universidad de los Andes, Chile. Bilal Zia is a Senior Economist in the Finance and Private Sector Development unit of the Development Economics Research Group at the World Bank.

PART I INTRODUCTION

1. Overview of microfinance, financial inclusion, and development Robert Cull and Valentina Hartarska

Microfinance institutions continue to provide essential financial services on a large scale that the poor cannot access otherwise. Microfinance emerged as a potential tool to fight poverty, rising to fame with the awarding of the Nobel Peace Prize to Muhammad Yunus and Grameen Bank, though a later series of influential randomized trials showed only marginal income gains for some clients and, most recently, widespread moratoria on repayment of microloans during COVID-19 proved necessary. This volume marshals evidence to help readers understand who microfinance reaches, how it helps them, and why clients come back. It also identifies the limitations of microfinance and why it has fallen short of grandiose expectations for reducing poverty. A part of the microfinance story is about how well-being and financial services are conceptualized and measured and how high-frequency data better reveal the financial challenges faced by households that hover near official poverty lines. Moreover, data that capture the perspectives of alternative actors—clients, providers, and donors—provide a fuller picture of how microfinance is practiced and offer a lens on microfinance outcomes with respect to gender, equity (especially in pricing), meeting urgent needs, and resilience to shocks. A theme throughout many of the chapters is that the benefits from financial inclusion increase if key foundations are in place. For low-income users of financial services, these include basic financial education and literacy, which also require both numeracy and literacy themselves. With respect to digital financial services, digital connectivity and capabilities become crucial, as well as systems for personal identification and protection of personal data. For providers, reaping the most benefit from services that promote greater financial inclusion requires continuous adaptation to the evolving needs and capabilities of potential clients. The ways financial services are delivered also play an important role in reducing the physical, social, and economic distance between users and providers. Group liability lending, for example, unlocks social connections between users who lack collateral and formal credit histories to facilitate credit from providers who otherwise would not lend to poor borrowers. Agent banking can reduce not only the physical distance to points of service, but also the social distance between low-income clients and the representatives of the providers, both of which can help facilitate greater usage of financial services. Digital financial services can eliminate travel to a point of service altogether, and thus further reduce the costs of providing services to this market segment. A final recurring theme emphasized in the country and regional studies in the volume is that context plays a major role in determining both the needs of low-income users and the feasibility of potential solutions from providers. For example, value-chain finance that brings crops to market at lower costs to farmers, and insurance and savings products that can protect smallholders when yields are low, are crucial in the agricultural context throughout the less developed world, but especially in Africa where the geographical distance to providers and frequent droughts pose acute challenges. Rural smallholders also face financial challenges in China, but there the political context has determined both the institutional structures that 2

Overview of microfinance, financial inclusion, and development  3

serve them and limited the feasibility of private sector-led solutions to this point. Digital financial services show promise in circumventing some contextual challenges, but, as noted, lower-income countries will also need to overcome substantial challenges related to connectivity and affordability, the digital and financial capabilities of low-income users, consumer protection, and the creation of systems and processes that safeguard and protect personal data. The handbook begins with a section that offers new conceptual frameworks to think about microfinance, financial inclusion, and economic development. Chapter 2 challenges us to rethink poverty itself, microfinance and household finance, and the impact that finance can have on the lives of the poor. Chapter 3 uses a scientometric approach to summarize how researchers have evaluated the impact of microfinance institutions. Chapter 4 presents a more nuanced view of the interaction between gender roles and microfinance in line with Chapter 2, and Chapter 5 discusses the concept of fair interest rates. Chapter 6 builds on the idea from Chapter 2 that volatility matters and shows how savings and credit help build resilience in emergencies. Chapter 7 closes this section, illustrating the types of financial educational programs that can help people improve their financial lives and identifying the circumstances under which they work best. Chapter 2, entitled “Rethinking Poverty, Household Finance, and Microfinance,” is by Jonathan Morduch—a leading microfinance thinker. It sets the tone for the need to examine anew the role of microfinance, and financial inclusion more broadly, as tools to promote economic development. Building on insights from his work that analyzed high-frequency financial data, Morduch proposes a new way of thinking about poverty as an overall insufficiency in the lives of the poor rather than the specific insufficiency captured by annual measures of income or consumption. These traditional poverty measures differ from how poverty is experienced, with important consequences for academic research and policies designed to alleviate poverty. Morduch’s insights are that life in poverty is best described through the interaction of insufficiency, instability, and illiquidity, rather than insufficiency alone. Therefore, a reduction in instability and/or illiquidity can help the poor even if it does not change the insufficiency measured by income or consumption. Instability, in turn, would matter less if households could borrow, save, and use insurance. Morduch proposes that household finances are better understood from the perspective of the needs of the poor to manage their cash flows in the short-term to fend off shocks rather than from the perspective of long-term goals for building assets because the poor constantly build up and deplete assets. The poor have fundamental needs to engage in consumption “smoothing” (or “distribution”—taking lump sums and shifting these to other times when they are needed) or “spiking” (“aggregation” or cobbling together bits of money to form larger sums). The combination of borrowing and saving helps to meet needs for a variety of lumpy spending needs such as health, education, and other lifecycle events. Such a view is different from the traditional focus on only one type of lumpy spending—entrepreneurial investment. Borrowing can be viewed as a commitment savings device, where structured relationships between the household and the lender help the household to use lump sums within a structure but with some flexibility. Traditional credit products do not pay attention to volatility and do not have such characteristics. Households, however, are vulnerable to cycles of persistent debt, and their need for flexibility is often met by informal rather than by formal lenders. That is why microfinance seems well positioned to address some of these problems, but the prevailing focus on business finance has prevented us from learning more about how microfinance is helping the poor. Morduch proposes that we recognize the new aspects of poverty and household finance— the low overall earning power, pervasive instability, and illiquidity. From that perspective,

4  Handbook of microfinance, financial inclusion and development

microfinance can be re-examined and some of the elements of microfinance contracts can be recognized as well suited to help the poor. This helps answer important questions such as why microfinance is so popular even if it seemingly has little impact on annual incomes or asset growth. Moreover, contracts could be improved by making them more suitable to meet the consumer finance needs of poor households. This approach is in contrast to the narrative and even ideology that microfinance exists to meet entrepreneurial investment needs and, in that sense, helps explain why the promise that microfinance can lift the poor out of poverty is unfulfilled. In practice, the original group lending mechanism that relied on informed peer pressure to ensure repayment is declining in popularity. Group members’ inability to repay is often driven by bad luck rather than “strategic default” and group members are typically too poor themselves to provide effective insurance mechanisms to their peers. Most microfinance contracts today look more like consumer loans and rely on division into, and repayment of, small sums of money and other dynamic incentives. The small sums operate like a credit card/consumer loan and, even if not used for business purposes, help with “smoothing” and “spiking,” which is important in the lives of the poor. Empirical evidence of modest impacts of microcredit on income or assets may derive from non-representative samples or may reflect the fact that households divert loans away from business investment. Microfinance continues to be popular even without demonstrating significant income/consumption impacts in experimental settings, and borrowers repay and continue borrowing precisely because microfinance meets other needs in combatting low, unstable incomes and illiquidity, as described above. In particular, microfinance loans have features similar to consumer installment loans and both have features of structured savings products that the poor need. The challenge that microfinance contracts need to address then is striking the right balance between structure and flexibility to help the poor manage their cash flows. The chapter concludes by cautioning that when microfinance is evaluated in terms of gender empowerment or burdens, it is important to recognize that it is used as a cash management tool by women facing very different circumstances in countries across the globe. Morduch’s insights raise important questions for microfinance research, some of which are also raised in Chapter 3 by Begoña Gutiérrez-Nieto and Carlos Serrano-Cinca, who present scientometric evidence on the assessment of microfinance institutions and their impacts. First, the authors identify the dual objective of the institutions as the main challenge of studies assessing microfinance institutions and their impact. These challenges are presented in their historical context as microfinance was gaining popularity and the enthusiasm for microfinance developed. Consequently, studies are linked to the Nobel Peace Prize received by Muhammad Yunus in 2006 (pre and post), as well as the Nobel Prize in Economics received by two MIT researchers, Abhijit Banarjee and Esther Duflo, in 2019. The authors describe the challenges of defining and measuring social performance, carefully presenting the supply-side literature that offers an institutional perspective and relies on the analysis of accounting information and the demand-side literature, which is focused on the quality of outreach. They use a scientometric approach to evaluate both strands of the literature and to offer insights into what kind of work has gained the most popularity in the scientific community. The authors first review measures of financial and social performance and summarize the work focused on social performance. They also describe the challenges related to measuring mission drift and its link to governance to conclude that, in most studies, mission drift is not observed when it is treated as a subset of the main inquiry in a study. The authors review the original outreach and social efficiency literature and the outcomes and impacts that it studies and conclude that it was qualitative in nature and helped create exaggerated expectations. They also observe

Overview of microfinance, financial inclusion, and development  5

that the subsequent quasi-experimental strand of the literature was more rigorous and careful in its conclusions while rigorous studies based on randomized controlled trials (RCTs) found limited impacts that may be realized only under some circumstances. The last part of the chapter is a scientometric analysis that relies on information from both Web of Science and Scopus to derive a sample of quality articles on microfinance and to create maps that illustrate the progress of the literature between 1995 and 2019, and separately 2020 as it was the first year after the Nobel Prize in Economics was awarded to researchers that used RCTs to evaluate the impact of microfinance programs. The authors present maps that illustrate several topics (general microfinance, sustainability, efficiency, mission drift, governance, outreach, and RCTs). The first map shows how studies cluster on health impacts and general impacts. The second and third show the country of origin of authors, as well as universities and research centers that focus on microfinance, while the fourth identifies the main journals in which microfinance articles are published, again with those in medical fields separate from the economics field. Next, the authors present efficiency knowledge maps and mission drift knowledge maps. The authors generate knowledge maps using the publications before 2007 (when the Nobel Peace Prize was awarded to Yunus) which show that the only topic related to assessment was impact and illustrate the difference between welfarist and institutionalist approaches. Articles from 2008 until 2018 were mapped and show that performance, sustainability, outreach, efficiency, and governance were added as topics, and a clear distinction between microfinance and health literatures emerged. The last knowledge map shows that since 2019, financial inclusion emerges as a new topic within the discipline. The authors offer concluding remarks indicating that despite the proliferation of microfinance research and ever more sophisticated methods to study its impacts, these impacts are inherently hard to study and should continue to be an important area of research in which a variety of subfield RCTs and quasi-experiments, as well as other methods, will continue to be used. In Chapter 4, Isabelle Guérin tackles the question of whether financial inclusion is good or bad for women. Guérin highlights the advantages of financial inclusion and illustrates some of the challenges that prevent finance from positively impacting the lives of women, consistent with the ideas presented in Morduch’s lead chapter. She starts by framing the controversy of whether finance helps or hurts women and thus women should be targeted in the context of differences in regional needs and as part of a dialog between two sides. One side believes that finance serves to manage liquidity constraints, which is essential in women’s lives especially since informal financial tools have proven insufficient in this regard. The second side argues that only structural changes can truly help women and thus improvements in finance will be insufficient. She also discusses the new hopes for the growth of digital finance and links that to its increasing importance in the economy. Guérin asks next if financial inclusion creates debt or helps women to manage scarcity. In line with Morduch’s ideas, Guérin offers evidence that financial services are mainly used for day-to-day management and less for investment. She reveals that managing household budgets and debt, in particular, has been a traditional role for women, one that has grown harder as household debt has increased to the extent that it has exacerbated inequalities. In this sense, financial inclusion offers women more choices to manage budgets, but it also further solidifies the gendered division in some societies that money management is the responsibility of women, even if that role is seen as shameful. One key aspect of this chapter is the emphasis on the diverse contexts, constraints, and needs that women have and the need for financial services to be well attuned to this diversity. Guérin offers evidence of how gender norms make women’s entrepreneurial potential

6  Handbook of microfinance, financial inclusion and development

impossible to realize in some contexts and more possible in other contexts. Moreover, she shows that targeting women to receive financial services does not necessarily prevent discrimination. Discrimination can occur in other, more subtle dimensions, such as loan size or women can be forced to participate in complementary training, or to receive other services that increase their costs in terms of time commitment. Guérin also cautions that the argument that targeting women creates a win-win situation by improving both efficiency and equity and thus avoiding mission drift may be weaker than expected, since a large share of the loans that women take is on behalf of their husbands. This chapter presents a careful argument for rethinking women’s empowerment and microfinance. It argues that the change in gender relations is often a zero-sum game. Men may view women’s access to financial services and entrepreneurial activities as detrimental to family obligations. The evidence seems to support the idea that financial inclusion can empower mostly women who already have some sort of preexisting power in business, politics, or social relations and may actually lead to the disempowerment of women who do not. This would then suggest that women who are seeking out financial services are doing so to negotiate interdependencies within their communities rather than to think and act autonomously to pursue economic opportunities for themselves, as might be expected by the donor community. Even when women use new products like mobile money or access their savings accounts, it is done to maintain the network of relationships or modify it. Thus, mobile money may not promote autonomy but help to reshape interdependencies. Overall, the conclusion is that financial inclusion can be helpful to decrease gender discrimination and inequality, but likely only marginally and only if services are affordable, non-discriminatory, and genuinely tailored to women’s needs and constraints, which are very diverse in different contexts and social groups. In Chapter 5 Marek Hudon and Joakim Sandberg review the existing theories about the fairness of interest rates in general and develop a conceptual framework for thinking about the fairness of prices for microcredit. They start by outlining the main debate on the fairness of interest rates in microfinance. One side of the debate emphasizes the welfare of the customers (welfarists) while the other emphasizes the need to build strong institutions (institutionalists). The authors point out that interest rates are difficult to study and compare because the interest payment itself is not the only cost for borrowers. They first describe the procedural approach to interest rates, which stipulates that if the interest rate is the result of a free negotiation process where neither the lender nor the debtor is coerced or deceived, then interest rates are fair. This would happen with transparency and free and voluntary exchange. However, they argue that loan contracts and their interest rates rarely represent the genuine will of the borrowers because of their impoverished situation and lack of acceptable credit alternatives. According to the second competitive market approach to interest rate fairness, high interest rates are a result of market inefficiency due to market imperfections. Interest rates would be lower and fairer in more competitive markets but this may be difficult to achieve in reality. Moreover, in practice increased competition can have negative consequences because it cannot guarantee sufficiently low interest rates to the poorest, and providers’ focus may shift toward the less poor (upper poor) who can afford higher rates. The credit-as-a-right approach to the fairness of interest rates is based on the welfarist ideas of Yunus, who argued that credit is a human right and should be affordable. However, there are a number of obstacles to establishing credit as a right. For example, other, arguably more pressing needs such as housing and clothing could be better candidates to be considered a human right. In addition, in Yunus’s own view, artificially lowering interest rates is not a viable alternative because MFIs must cover their costs. And, in practice, imposing ceilings to create fair

Overview of microfinance, financial inclusion, and development  7

interest rates has proven ineffective. The consequentialist approach argues for lowering interest rates because this may ameliorate the (financial) situation of the poor. It also recognizes that rates must be utility-maximizing, and that utility includes long-run access to financial services, which requires sustainable institutions. This approach is similar to the competitive market approach in arguing that MFIs should not be able to transfer unnecessary costs onto their clients. However, the authors identify problems with this approach related to the inability to ensure that utility is maximized by setting interest rates as low as possible for example. Ultimately, the authors propose an approach to fair interest rates that is a combination of the rates-as-rights and the consequentialist approaches. Consistent with the first approach, they argue that in a just society, rates should be set “low enough.” The authors then propose to connect the ethical responsibility principle of the agents involved in microfinance to the latter approach. They therefore argue for the use of strategies that push interest rates toward the lowest possible level. Their proposal is similar to achieving a pooling equilibrium in insurance contracts. Specifically, the authors argue that microfinance agents should charge higher interest rates to one class of clients that can afford to pay to cross-subsidize the lower rates charged to another (e.g. from urban to rural or from the “upper” poor to the poorest). The second strategy would involve improvements in institutional efficiency that naturally lead to lower costs of doing business and thus lower rates. Finally, the authors support a variety of help from socially responsible investors and donors who would continue to support MFIs through guarantees, improvements in management information systems, and, most importantly, smart subsidies, with an awareness that subsidies should not go only to the largest MFIs to avoid potential “crowding out” of smaller institutions. In Chapter 6, Saniya Ansar, Jake Hess, and Leora Klapper provide an overview of the role of savings and credit in helping the poor be more resilient in emergencies. They start by clarifying how emergencies, including COVID-19, challenge people’s finances in both developing and developed countries. Their review suggests that the income level of the countries is not significantly correlated with people’s ability to come up with emergency funds, but that within-country inequalities explain most of the variation in financial resilience. The authors enumerate the sources of emergency funds and observe that the main source in developed countries is employment, while in developing countries it is friends and family. With COVID19 affecting the ability of people to work and generate income in all countries, access to emergency funds has been a significant challenge. The authors describe also how financial resilience is affected by the way people save—in formal financial institutions versus informally, with the latter more prevalent in lower-income countries. COVID-19, moreover, has had differential impacts on savings accumulation in developed and developing countries with a savings accumulation effect in the former and savings depletion in the latter group. For vulnerable populations such as women and the elderly, the advantages of access to formal savings mechanisms are further highlighted. Access to credit is important for financial resilience, and the authors argue that access to formal credit is especially important in times of crisis. Regulatory interventions aimed at protecting borrowers during the COVID-19 crisis have been especially beneficial. This stands in contrast to microfinance studies that find more muted impacts of access to credit on incomes and women’s empowerment. Modest impacts of microfinance might also stem from how poverty and insufficiency are measured in examining the role of credit in some studies, in line with messages from the lead chapter in this volume that caution researchers to carefully consider how these concepts are operationalized and tracked.

8  Handbook of microfinance, financial inclusion and development

Digital financial services, the authors further argue, help people to prepare for financial challenges. Services such as default and automatic deductions from accounts and “nudges” such as text message reminders have been shown to be useful in helping people to save more and to strengthen their resilience to shocks. Women, in particular, were able to have more control over their financial resources when they used digital financial technologies. Moving from cash to account-based transactions and digital payments also improves the efficiency of government interventions by decreasing the administrative costs of transfers while increasing the speed with which support reaches the target populations. This, too, has been particularly important for COVID-19 interventions in several countries. Since many digital services are complicated, the authors next describe some of the related risks and consumer protection options that help mitigate these risks. Vulnerable consumers (e.g. the elderly or the very poor) may be misled by some service providers or may lack the capacity to understand well the digital world and the terms of their financial contracts. General educational programs are rarely effective, but targeted and specific education prior to service provision can be more effective. Technological solutions are also available but government involvement to improve transparency is essential. Finally, the authors investigate empirically which individual characteristics and government policies are effective in improving the use of formal savings, and thus financial resiliency. Using data from the 2017 World Bank Financial Inclusion and Consumer Protection Survey in a two-stage Heckman-type selection model, they show that, perhaps unsurprisingly, individuals who are in the upper-income quintiles, older, more educated, employed, married, and resident in urban areas are more likely to save. The results also show that consumerfriendly government interventions and policies, such as capping the maintenance fees for savings accounts, special tax incentives for saving schemes, lowering know-your-customer requirements, and stronger consumer protection monitoring measures are associated with a higher probability of saving and thus are policies worth encouraging. In Chapter 7, Bilal Zia discusses the role of financial education, which has gathered a lot of enthusiasm among policymakers, especially in the wake of the financial crisis in 2008. Education helps if it addresses the reasons why people make suboptimal financial decisions in their lives. Since the structure of the financial system is complicated to navigate and financial products are difficult to evaluate, the idea of financial education seems attractive and has been embraced by politicians in both developed and developing countries. However, the early academic research did not offer support for the effectiveness of financial education. It evaluated classroom-based programs that relied on standardized curricula with some local contextualization. The results in general showed that such programs not only struggled to attract adults but also failed to change their financial behaviors, likely because other factors like income were the binding constraints. On the supply side, financial education programs were expensive, difficult to design, and it was difficult to find qualified staff to deliver them. However, while the average treatment effects of financial education were muted, some subgroups like the illiterate or those with lower levels of financial literacy benefited. Appreciation of heterogeneity in customers increasingly led researchers to study the impacts on high-risk and vulnerable populations for which education may be more effective or appropriate. The later wave of research paid closer attention to measurement, targeting, and timing and explored complementarities with potential add-on services and features. Financial literacy, which is ultimately a measure of the success of financial education, is easier to define and monitor in advanced countries. Zia argues that in lower-income countries the impact should be measured more in terms of people’s access to financial services and take-up rates. He

Overview of microfinance, financial inclusion, and development  9

proposes a new approach to financial literacy that assesses numeracy skills, basic financial awareness, and attitudes toward financial decision-making. Research to date using these concepts finds weaker evidence that numeracy improves financial decisions but stronger evidence that the other two dimensions do. Zia argues that targeting youth could be most effective, not only because education improves the youth’s own knowledge, attitudes, and decisions but also because it has the potential to spill over to older relatives. Indeed, positive outcomes have been found in several studies for youth in advanced countries, but research on interventions in lower-income countries does not confirm those positive effects. Zia summarizes experimental evidence indicating that appropriate timing of training and triggering teachable moments affect migrant workers’ financial attitudes and actions. The evidence also shows that complementary services like gaming and experiential learning and add-ons such as reminders and goal-setting tools work well, particularly in low-literacy contexts. The delivery channels (e.g. via TV edutainment) and the potential for scale-up of programs combined with local control and customization are also described as effective. Finally, the evidence presented shows that rural women and high-risk groups in developing countries improve their financial knowledge and behavior with financial education. The author concludes that the long-term impacts of financial education are actually still unknown and are therefore a promising research area. Other exciting research avenues in this area could involve the use of digital platforms to reach new population segments and machine learning approaches to customize educational content. The next section of the volume explores how financial services are delivered to poor clients. The modalities are as varied as clients’ needs and a recurrent theme of these chapters is that products and delivery methods are adapted to the local context to better meet those needs. But gaps between what is needed and what is available and affordable persist, and the process of imperfect adaptation continues. Chapters 8–10 explore three specific approaches for delivering financial services: joint liability (or “group”) lending (Chapter 8), agent networks (Chapter 9), and digital technologies (Chapter 10). Chapters 11–12 examine financial products used by agricultural producers, specifically credit products that enable smallholders to participate in agricultural value chains (Chapter 11) and, more generally, credit and insurance (Chapter 12). In Chapter 8, Christian Ahlin and Godwin Debrah describe the theoretical underpinnings for group lending, explaining how joint liability for group members’ loan repayments, the denial of future credit for all group members in cases of default, and the ability to harness local information and social pressure mechanisms enable this type of lending to flourish. Another common feature of joint liability lending is group meetings, which can help reduce the operational costs of lenders (by providing information about multiple borrowers and projects at the same time) and may foster social capital among group members, leading to better loan repayment and more efficient risk sharing. At the same time, group meetings impose additional costs on borrowers, who must travel to meetings and sit through the settling of others’ loan accounts. The authors conclude that the local context determines whether these costs to borrowers outweigh the benefits in terms of increased credit access. This framework also provides a lens for interpreting some of the empirical findings on group lending. For example, studies using large samples of MFIs across countries have found that non-profit MFIs use group lending more extensively than for-profit MFIs and that MFIs that rely on group lending tend to make smaller loans, operate less profitably, and lend more to women than MFIs that rely on traditional individual liability loans. Based on these patterns,

10  Handbook of microfinance, financial inclusion and development

the authors speculate that group lending is used for borrowers who are difficult to reach, and in environments where social capital is more plentiful than collateral. Field experiments have provided some indications that group lending engenders high repayment rates and can improve the lives of female borrowers (in terms of food consumption and business ownership), but there are few studies that compare group lending against other alternatives. Moreover, the details of the experimental tweaks to lending style may be the critical drivers of patterns, and experiments require a willing institutional partner—that willingness may itself signal that the partner is not representative of other institutions. In general, concerns about external validity make it difficult to generalize about the efficacy and usefulness of group lending from field experiments. Finally, local context and trade-offs may also be at the root of the supposed shift away from group liability lending. Lacking a census of microfinance institutions, it is difficult to document this trend precisely, but even studies that focus on within-MFI variation have found declining reliance on joint liability lending. Appealing again to theoretical findings, Ahlin and Debrah note that non-profit lenders can offer group lending as long as it breaks even. But when social capital is low, and thus the risk of default on joint liability loans is relatively high, individual lending can be feasible when group lending is not. Competition, which reduces the interest margins on group loans, therefore provides one potential reason for non-profit lenders to shift toward individual lending. The changing composition of lenders is another potential driver. As the microfinance industry becomes more commercialized, and lenders increasingly seek to extract profits from micro-borrowers (rather than breaking even), group lending may be replaced by the more-profitable individual lending. Finally, the shift towards individual lending may also be driven by MFIs gaining experience—as group lenders accumulate information on borrowers’ credit histories, some of the asymmetric information that group lending is meant to overcome erodes. Also, because the social capital created through group meetings is likely to be durable, lenders may shift from group meetings when the diminishing returns to creating additional social capital no longer exceed the costs of meetings. While competition, commercialization, and experience provide reasons for the decline in group lending, the authors conclude that its ability to harness social assets to resolve informational asymmetry and make lending feasible implies that it will continue to offer substantial benefits in many settings. Local knowledge and social connectedness can unlock microcredit in certain contexts, but the impediments and resulting innovations differ for other financial services. In Chapter 9, Sinja Buri, Robert Cull, and Xavier Giné describe how agents—small retail businesses whose owners are trained to collect deposits and process withdrawals and payments for a formal financial institution (most often a bank or a microfinance institution)—reduce the costs and increase the security of financial transactions for market segments that are typically underserved. Because of their proximity to poorer clients, agents reduce the costs of financial transactions associated with travel time and waiting at branches, while offering greater security relative to informal financial services. And because of their social proximity to those clients, agents can help familiarize formal financial services for people with limited knowledge and experience with them. Consistent with this notion, mounting evidence shows that microfinance clients conduct more financial transactions when agents are available. More specifically, emerging evidence indicates that clients become more financially active with respect to depositing, withdrawing, and transferring funds. While the convenience of agents could also lead to more client savings, the available evidence suggests that commitment

Overview of microfinance, financial inclusion, and development  11

devices may also be required, and there is little evidence that agents expand credit to clients, though they can facilitate loan repayment. There is also limited evidence that clients may have greater trust in agents than other bank staff, especially if agents are of the same gender, though privacy concerns may influence them to continue conducting larger, more sensitive transactions at branches. Emerging evidence also suggests that consumers are price-sensitive and that the pricing of agent transactions needs to be an important consideration when planning for customer activity. The chapter closes by presenting preliminary evidence on the business model for microbanking through agents, emphasizing the trade-offs between rapid network expansion and the quality of agents’ services. Challenges can be especially acute in rural areas that can sustain few providers, leading to concentrated market structures or even local monopolies, which have been linked with complaints about service quality and concerns about high fees. Even when tariff structures are regulated, there is evidence that some agents charge extra or hidden fees, and these higher prices are more likely to be charged to women, less educated clients, and those living in remote areas. While there is a role for regulation to protect clients with limited or no experience with financial services, the chapter also points out that fees, deposit mobilization on a large scale, and cross-selling of products can be crucial to agent profitability. The tensions between consumer protection and agent profits can therefore be difficult to navigate, despite the documented cost reductions that agents offer. In Chapter 10, Greta Bull and Leora Klapper explore the potential for digital financial services (DFS) to extend and deepen financial inclusion. By eliminating travel to financial institutions, DFS can reduce costs even further than agents can. For example, mobile money accounts provided by telecommunications companies enable people to use basic mobile phones to send money over long distances quickly and safely to friends and relatives facing a financial emergency. Those friends and relatives still often use a mobile agent network to “cash out” the funds sent to their accounts, but the ease of sending and receiving those funds has been shown to improve recipients’ resilience to financial shocks. In general, cost reductions derive from a business model that is not predicated on face-to-face contact with financial services providers, one that is less reliant on deposit mobilization, fees, and cross-selling for its viability. Many of the benefits of DFS relate to making payments for retail purchases, bills, and fees (e.g. to schools or governments) and to receiving payments, such as wages from businesses, remittances from family members, or safety net benefits from governments. And Bull and Klapper summarize evidence showing that the use of digital payments increased substantially during the COVID-19 pandemic as more apparel factories paid wages into accounts, agribusiness chains paid farmers directly to mobile money accounts, and local grocers and other merchants allowed customers to pay from their mobile money accounts. DFS also offers potential benefits in terms of savings, though the evidence to date indicates that savings accumulate when funds are automatically deposited into accounts. Making a digital service/account the default option for receiving funds and other commitment devices (as mentioned above) may be crucial for DFS to enable savings accumulation, and the subsidies and lack of fees on deposits that have featured in experimental settings may not be feasible in the real world. In short, despite the technological advances of DFS, savings remain a tough nut to crack. The payments and savings benefits of DFS may also reduce the relatively high rates of financial exclusion among women, though for it to do so, legal strictures and social norms may need to be changed, as well as the subordinate economic role in households and lack of access to technology such as mobile telephony that confronts women in many countries.

12  Handbook of microfinance, financial inclusion and development

At the same time, and precisely because DFS does not rely on social connections in the ways that group liability lending and banking through microfinance agents do, it raises a host of thorny issues regarding consumers’ data protection and privacy. These challenges can be especially problematic with respect to credit and for clients who lack experience with financial services, including many women. Recent evidence from East Africa has shown high rates of default and late payments on digital loans, in part due to a lack of transparency about these credit products, and that a substantial share of borrowers (20 percent in a study in Kenya) have reduced food purchases to make a payment on these loans. And while data from digital payments can help establish creditworthiness and enable merchants to tailor offerings to individual preferences, there are also indications that clients, including lowincome ones, are willing to spend more to obtain financial services that come with privacy protections. Evidence from analog financial products based on mystery shopping exercises and lab experiments has shown that inexperienced borrowers receive less information on credit products than more experienced ones, and that printed materials on product specifications are often incomplete, misleading, and in violation of guidelines, even though simplified statements of key facts about credit and savings products have been shown to induce better choices by consumers. Evidence in this area is sparse (though emerging), but it stands to reason that consumer and data protection concerns related to DFS will be especially challenging. This presents a conundrum going forward—DFS may be the most effective means of expanding financial inclusion among inexperienced clients, including women, but those potential clients may also be most ripe to be taken advantage of by financial services providers and others who would exploit their personal data. In Chapter 11, Alan de Brauw and Johan Swinnen zoom in on the agricultural sector and the financial challenges faced by smallholders. They detail how risks they face limit their production choices and their willingness to make investments, including climate and weather, pests, price risk, crop damage, and perishability. Relative to grains, the prices of fruits, vegetables, and animal source foods have been rising in the long term, and thus participating in value chains for those products offers a path to greater profits, but these farmers lack the financial tools necessary to overcome the associated risks such as affordable credit and borrowing and saving products that are aligned with agricultural production schedules and needs. Like other forms of microcredit, making small loans to smallholders is costly per dollar lent, and, since it is spread across rural areas, agricultural production and the effects of weather on it are difficult for lenders to monitor, leading to moral hazard concerns. Moreover, using collateral to secure credit is often unworkable because property rights over land are ambiguous or incomplete. De Brauw and Swinnen describe how agricultural value chain finance (ACVF) is overcoming these challenges in some contexts. ACVF is a specific type of agricultural finance arrangement that involves at least three parties, two of which are value chain participants, and one of which is a financial institution. One of the two remaining parties is the end borrower. The second remaining party can either be directly or indirectly involved in providing finance to the end borrower (examples include grower cooperatives, marketing associations, insurers, or loan guarantors). Pre-conditions for successful ACVF arrangements include a stable macroeconomic environment and the absence of distortionary policies such as price and interest rate controls. Competition both within value chains and within the segment of the financial sector related to agriculture is also crucial. Adaptation and experimentation in the delivery of credit

Overview of microfinance, financial inclusion, and development  13

(e.g. through mobile money platforms) and new forms of collateral (e.g. warehouse receipts or repayment histories) can also help facilitate ACVF. Finally, capacity building among farmers, through for example extension services bundled with financial products can enable them to grow crops that are more commercially viable, while on the financial sector side, loan officers may need to learn more about agriculture, including production timing and risks. In Chapter 12, Mario J. Miranda and Denis Nadolnyak examine the difficulties of insuring smallholders when agricultural yields are low, focusing particularly on index insurance, which provides payouts based on the value of a specified “index,” a publicly observable random variable that is highly correlated with the losses of the insured, but which cannot be influenced by the actions of the insured or the insurer. Index insurance does not, therefore, indemnify the insured’s pure loss, but instead makes payouts when a triggering event such as a drought or other natural disaster occurs. The prototypical example is insurance that provides payouts when rainfall measured at a nearby meteorological station falls below a specified level. In principle, index insurance appears to be a viable way to safeguard against production risks that plague smallholders, and indeed, the authors describe the myriad pilot programs and studies initiated by researchers, agricultural ministries, and NGOs over the past 30 years to test and demonstrate its viability in less developed countries. But they also note that, despite those efforts, there is not a single index insurance program anywhere in the world that is currently generating significant sales volume without heavy reliance on government subsidies, and to explain why, they draw on a series of intuitive theoretical models. The fundamental impediment to widespread adoption of commercial index insurance is “basis risk” which refers to its limitations in providing indemnities that closely match the losses of the insured because the basis for the payouts (e.g. measurement of rainfall at a specific station) may not be closely correlated with the insured’s losses, which are based on reduced, lower-quality agricultural yields. This can be because of the distance between a farmer’s plot and the measurement station; simplified payout schedules that fail to capture the complex relationship between yields and rainfall; and/or idiosyncratic production risks that are uncorrelated with the index such as pestilence, disease, windstorms, fires, vandalism, and destruction caused by animals. If basis risk could be eliminated or greatly reduced, static models of smallholder utility indicate that index insurance could be beneficial to them and commercially viable for providers, though even here the correlation between losses and payouts is necessarily imperfect because payouts are capped at a maximum and depend critically on the “trigger” level of rainfall at which payouts start. The authors then illustrate that dynamic utility models that allow smallholders to self-insure through savings reveal index insurance to be substantially less attractive than static utility models indicate—poor farmers cannot afford index insurance while rich ones do not need it because they have better self-insurance alternatives. These problems are only compounded as basis risk and premium “loads” to cover insurers’ administrative costs and other expenses increase. The authors also discuss recent literature on behavioral influences that further dampen demand for index insurance stemming from farmers’ risk attitudes, lack of trust, and limited understanding of insurance. This combination of simple theoretical modeling and empirical evidence provides a compelling explanation for why index insurance has failed to take off absent heavy subsidies. What can be done about it? Using technology to reduce basis risk offers some promise. Index insurance contracts based on vegetation and rainfall-proxy indices computed from

14  Handbook of microfinance, financial inclusion and development

satellite observation allow finer geographic resolution, though disparities between remotely sensed data and conditions on the ground introduce another form of basis risk. Basing payouts on average per-hectare yields and revenues also offers promise because those measures can account for losses due to more factors (pestilence, disease, floods, or low prices), though the costs and timeliness of yield- and revenue-based estimates have thus far proven a major challenge. The authors’ main recommendation is, therefore, to bundle index insurance with credit. They argue that index insurance has the potential to decrease loan defaults dramatically, enabling lenders to manage their portfolio risk more effectively. With greater financial incentives for lenders to offer more loans at lower interest rates, which farmers understand easily and demand, this market could grow with limited reliance on subsidies. The third section of the volume assesses the impact of microfinance from the perspectives of practitioners and investors. In Chapter 13, Claudio Gonzalez-Vega, Laura Mo, and Giovanni di Placido provide an in-depth description of how BBVA Microfinance Foundation uses longitudinal data to measure and track the impact of its financial services on clients in Latin America. In 2007, Spain’s Banco Bilbao Vizcaya Argentaria (BBVA) established the BBVA Microfinance Foundation, a legal entity independent from the BBVA Group with its own governance and management to promote inclusive finance through investments in and consolidation of a group of leading microfinance institutions in Latin America. A critical element of that consolidation process was a unique approach to measuring client-level outcomes which uses a centralized longitudinal dataset of the universe of clients from those institutions. By tracing out the arc of long-term relationships with clients, these data enable the Foundation to track their poverty and vulnerability status and link it to the financial services they have received from its institutions. These rich data have also made it possible to describe the full distribution of various client outcomes by annual cohorts and for different client segments. Analysis of cohorts and market segments has enabled the Foundation to assess whether these institutions have drifted from their original mission to service poorer clients (they haven’t) and to describe typical pathways out of poverty. In addition, those analyses indicate that poorer clients and, especially, women have tended to experience larger percentage gains in income and surplus after receiving microcredit than richer ones. A point of emphasis throughout the chapter is the potential heterogeneity of impacts from microfinance due to characteristics of clients, who differ in terms of preferences, ambitions, resource endowments, demography, age, and economic opportunities; the microfinance lending technologies themselves; and environmental characteristics and shocks that lenders and clients have no control over. The chapter closes with a brief discussion of the differential effects of the COVID-19 pandemic across institutions and market segments— female microentrepreneurs, for example, could fare worse because they tend to be concentrated in informal trade and services sectors, which have been especially impacted, while women have also assumed larger responsibilities to care for other household members. The rapid pace of digitalization of microfinance during the crisis is also noted, and the authors raise concerns about its effects on clients with less experience with and understanding of financial services. In Chapter 14, Gregor Dorfleitner, Dina Pons, and Noémie Renier delve into measuring and managing social performance and impact by MFIs, but from the perspective of investors, focusing specifically on microfinance investment vehicles (MIVs). The chapter opens

Overview of microfinance, financial inclusion, and development  15

by tracing the historical evolution of MFI performance measurement, from a focus on doing good (or at least no harm) while being financially sustainable in the 1990s, to a recognition that benign intentions were insufficient and measurement of social performance was crucial to assessing whether MFIs were achieving their mission in the 2000s, and to incentivizing MFIs to manage simultaneously for financial and social performance while encouraging more systematic approaches to the measurement of impact on clients after 2010. In promoting measurement of social performance and impact among investee institutions, many MIVs have made strides to avoid investments that could result in social (or environmental) harm, though few use thorough analysis of positive social impact to guide their investment decisions, and social audits of MFI portfolios by investors are rare. While there remain important challenges in building the business case for social performance management, various bodies have promoted guidelines for the governance of socially responsible financial services providers (FSPs). At the industry level, investors are playing an important role in emphasizing the importance of local currency funding for MFIs to protect against sudden currency depreciations that have crippled the industry at times, promoting responsible lending practices in good and bad economic conditions, and encouraging greater pricing transparency for MFI financial services, though these all remain works in progress. Aligning the expectations of asset owners (investors) and managers of FSPs with respect to social returns is central to these efforts in terms of incorporating financial and social returns in a single measure of return on investment, responsible exits from institutions through sales to investors that share social performance goals, and detailed assessment of the social performance of investors (to date, the focus has been only on FSPs). Finally, the chapter highlights the ways that MIVs have assisted FSPs during the COVID-19 crisis. For example, the Social Investors Working Group (SIWG) of the Social Performance Taskforce, a non-profit organization with more than 3,000 members globally, has provided guidance on avoiding debt collection processes and loan restructuring that could harm clients, responsible staff movement and the protection of field staff, and transparency for clients as FSPs have shifted toward digital channels to provide services. The volume closes with chapters that offer regional and country case studies to illustrate some aspects of the link between microfinance, financial inclusion, and economic development. The section starts with two chapters focused on financial inclusion in developed countries. The first is a study that compares gender and poverty gaps in financial inclusion between the US and the Euro Area while the second chapter evaluates the link between financial literacy and the use of financial services in the US. The chapters could serve as a benchmark for the progress in developed countries. The following two chapters focus on Latin America, with the first exploring financial inclusion, microfinance, and financial education, while the second focuses more narrowly on the gender gap in financial inclusion. The following two chapters are an empirical study of the role of inclusive finance in Africa and a narrower study on Kenya where the role of digital financial inclusion is evaluated from the perspective of the link between morality and finance. The last three chapters are country studies on inclusive finance and the rural transformation in China, the role of microfinance for financial sector development in Mongolia, and the link between financial inclusion and poverty in Armenia. In Chapter 15, Anastasia Cozarenco and Ariane Szafarz study financial inclusion gaps in developed countries and ask how they relate to gender and to poverty. The authors analyze several rounds of the Global Findex survey together with OECD data and show that financial inclusion in the Euro Area differs from that in the USA. The gaps in access to financial

16  Handbook of microfinance, financial inclusion and development

services between men and women, and between rich and poor individuals in these two developed regions are demand-side driven, and distinct from the gaps in the developing countries, where financial inclusion gaps are described as supply-side driven. Financial inclusion is measured through rates of account ownership, formal borrowing, formal savings, and financial resilience, which is the ability to find savings equivalent to one-twentieth of GNI per person in response to an emergency. The analysis shows that the two regions are similar in terms of access to bank accounts and emergency savings but in the US access to credit and formal savings is better than in the Euro area. The authors compare rates of use of financial services by groups and map the underlying income differences by gender and by poverty percentile to evaluate whether they explain the observed differences in financial inclusion. The authors find larger gender gaps in the Euro area, except with respect to financial resilience, while they find that financial exclusion of the poor is worse in the US. Surprisingly, the study finds that the Euro area and North America were less able than other regions to narrow the gender and poverty gaps for saving and to increase borrowing opportunities offered to the poor between 2012 and 2017. In Chapter 16, James R. Barth, Valentina Hartarska, Jitka Hilliard, and Nguyen Nguyen evaluate how financial literacy is related to the use of financial services by low-income households in the US. The chapter uses household survey data from the Financial Institutions Regulatory Agency (FINRA) Foundation’s National Financial Capability Study for the years 2009, 2012, 2015, and 2018. Since in the US use of banking services is cheaper than the use of alternative financial services (AFS) such as payday lenders or pawnshops, among other providers, the main hypothesis is that lower levels of financial literacy would be associated with the use of more AFS. Consistent with the approach by Zia in Chapter 7, financial literacy here is measured by the number of correct answers to five financial literacy questions that capture the complicated nature of an advanced financial system, namely questions about compound interest rates, bond prices, mortgage loans, inflation, and risk. The results show that greater financial literacy (as reflected in one additional correctly answered finance question) increases the probability that a household is banked (by about 3.8 percent) and decreases the probability of the household being underbanked (i.e. using banks but also relying on some AFS) or unbanked (these households may or may not use AFS). Moreover, financial literacy decreases the probability that the household uses AFS as well as the number of AFS that it uses. Overall, these results indicate that in high-income countries financial literacy is related to using less costly services, suggesting that it is related to improvements in the financial situation, and thus the well-being, of the households. In Chapter 17, Alejandro Javier Micco Aguayo and Patricio Andrés Valenzuela Aros take stock of financial inclusion in Latin America, focusing specifically on innovations to reach previously unbanked population segments, such as branch and account expansion programs, banking agent networks, mobile banking and mobile money, and other products provided by fintech companies. They also emphasize that the success of efforts to deepen financial inclusion depends crucially on financial education to improve the knowledge and capabilities of previously un- or underserved clients, both so that consumers are not taken advantage of and to facilitate genuine competition in this market segment. They therefore pull together available information on financial education in the region to provide a current snapshot and point out wide disparities in this dimension across and within countries.

Overview of microfinance, financial inclusion, and development  17

In Chapter 18, Victor Motta looks more closely at the state of financial inclusion for women in Latin America and the Caribbean. Regression results indicate that well-educated women with relatively high incomes are among the most likely users of financial services, even more so than men with similar characteristics. However, there are relatively few highly educated women with high incomes, and thus financial inclusion indicators for most Latin American women remain low and lag behind those of men. Like the previous chapter, this chapter emphasizes the importance of building financial knowledge and capabilities, especially among women, to address these financial inclusion gaps. In Chapter 19, Calum G. Turvey and Apurba Shee offer an assessment of financial inclusion and agricultural development in Africa, emphasizing the wide variation across countries and regions in access to and usage of deposits, credit, insurance, and mobile telephony for digital financial services. Despite this heterogeneity, and acknowledging that efforts to promote financial inclusion across the continent have met with mixed success, the authors emphasize that a significant and growing number of micro-studies find a relationship between agricultural credit and agricultural productivity and that insurance outcomes look promising except, perhaps, for the very poor. One interesting aspect of credit market failures in Africa is risk rationing, which occurs when insurance markets are absent, and lenders facing asymmetric information shift so much contractual risk to the borrower that the borrower voluntarily withdraws from the credit market. This can arise even when the borrower has the collateral wealth in the form of productive assets needed to qualify for a loan contract but does not want to pledge that collateral because its forfeiture would severely undermine future productive activities. Estimates are presented from surveys in multiple countries indicating that risk rationing may be more prevalent in Africa than in other developing regions. Turvey and Shee argue that inclusive financial policies are more likely to encounter success if they move beyond microfinance as it has traditionally been practiced to better target tailored insurance and credit products to under-served market segments. Greater flexibility in the design and terms of financial products, particularly credit, appears to be important to bridge farmers’ needs and lenders’ willingness to supply. Survey evidence indicates that, in many cases, lenders have very limited understanding of farmers’ financial needs, while farmers have little experience with or understanding of financial services. In addition, regulatory policy and oversight will need to adapt to foster the adoption of these products, and alternative delivery channels such as mobile technologies may be key to success. Many of the themes and recommendations in this chapter echo those in Chapter 11 by de Brauw and Swinnen on overcoming the financial challenges faced by smallholders. In Chapter 20, Susan Johnson and Silvia Storchi offer a new way to evaluate the potential for the success of digital finance in Kenya. Instead of using traditional financial inclusion metrics of access, use, and impact on poverty, the authors offer an economic anthropologist’s perspective to enhance our understanding of the impacts of finance. Building on Sen’s capability framework, the authors define what it means to live well in Kenya by analyzing survey data. They find that relationships and morals are an integral part of how material—including financial—resources are understood and used. They further find that a good life is found to be not an individual affair but a relational and collective experience. Financial technologies such as M-Pesa (for mobile money transfer and access to credit based on analysis of data from phone calls), Kenya’s Equity Bank lending practices, or the app M-Changa (a crowdfunding platform) have been built to incorporate local practices and were successful. In contrast, an alternative such as the app MaTontine (in Senegal) undermines local practices because

18  Handbook of microfinance, financial inclusion and development

it digitalizes information from the transactions of ROSCAs to offer larger individual loans and improve repayment. The authors conclude that “financial sector development can seek to uphold and develop indigenous logics or pursue mechanisms that fragment and undermine them.” In Chapter 21, Calum G. Turvey describes the historical evolution of financial services to rural residents in China. Lacking the institutional arrangements that underpin financial services provision by private firms such as property rights, the Chinese rural financial system has evolved through government initiatives and experimentation. From its third incarnation in 1963, the Agricultural Bank of China (ABC) played a pivotal role in delivering agricultural credit to farmers as well as making policy loans on behalf of the People’s Bank of China (PBC). As China increasingly pursued market-based economic reforms beginning in the late 1970s, the financing of Township and Village Enterprises (TVEs) became a priority, and ABC’s role adapted accordingly. It was directed to oversee the large network of rural credit cooperatives (RCCs), which had been created to support agricultural production cooperatives starting in the 1950s. However, the issuance of land use rights (LUR) under the household responsibility system meant that land could not be used as collateral for loans. As a substitute, banks and RCCs instituted a system of group guarantees based on joint liability when lending to TVEs and farmers. Over time, RCCs were granted greater freedom to operate commercially and were formally separated from ABC in 1996. In 2002, new policies allowed RCCs to set their own interest rates within upper and lower bounds to adapt to demand and supply conditions, and some RCCs were converted to joint-liability stock rural credit banks. And, since the 2006 New Countryside Campaign, new financial institutions were licensed including the formation of village banks, postal savings institutions, non-deposit joint-stock lending institutions, microfinance institutions, and loan guarantee companies. In 2015, China opened an office for financial inclusion and promulgated new laws to enable the transfer of land use rights. Against this backdrop of government-led development of rural finance and the dominance of ABC and then RCCs, typical Grameen-style MFIs did not emerge on a large scale. Turvey notes that, in fact, MFIs in the Chinese context usually refer to lending-only companies that make loans to TVEs and SMEs, and not to poor farmers or individuals. In addition, he offers evidence of the pervasiveness of informal lending in China based largely on kinship networks, which are rooted in trust, reciprocity, and altruism. MFIs simply could not compete with the subsidized interest rates offered by formal institutions supported by the government, nor displace the huge social networks of friends and relatives that willingly lend to each other at zero interest rates. These formal and informal financial arrangements will likely evolve as China moves toward a more commercially oriented agricultural economy supported by the legal transfer of land use rights. In Chapter 22, Batkhuyag Myagmar, Robert Lensink, and Wim Heijman study the relationship between financial development and economic growth in Mongolia for the period 2006–2016. Mongolia has a highly concentrated bank-dependent financial sector with about a dozen banks as well as a vibrant microfinance sector. This justifies a focus on the relationship between the banking sector, the microfinance sector, and economic well-being. The authors use a principal component analysis to create measures of microfinance and banking sector development, while economic well-being is measured by GDP growth. They perform a cointegration analysis, followed by additional analysis to identify the causality of impacts. Their results show that the microfinance sector has helped the banking sector to develop in both the

Overview of microfinance, financial inclusion, and development  19

short and long term, possibly because microfinance institutions are clients of the banks. In contrast, the banking sector has adversely affected microfinance development in the short run and had no impact over the long term. The authors conclude that microfinance has become an integral part of the financial sector in Mongolia even if it has had only a short-term measurable effect on banking sector development. The lack of a long-term relationship between economic growth and either banking or microfinance in Mongolia is consistent with the ideas in this handbook. Namely, financial services in general, and especially those for the poor, can be most effective in providing money management for liquidity purposes or alleviation of financial constraints. Thus, this chapter’s results are in line with the idea that microfinance in particular brings essential financial services to households throughout the year and that these benefits are hard to capture with the typical annual or aggregate-level measures of income and poverty. In the last Chapter 23, Aleksandr Grigoryan, Knar Khachatryan, Knarik Ayvazyan, and Pundarik Mukhopadhaya study the link between financial inclusion and poverty in Armenia. The authors acknowledge the multi-dimensional aspects of poverty and rely on measures of income, education, health, living standards, and employment to create an index of multidimensional poverty. Using that index, they find that reducing health expenditures is the main pathway to reducing poverty, while income and education are less important channels. Next, they define a measure of financial inclusion using data on access to finance, savings behavior, and financial literacy from annual surveys collected by the Central Bank for the period 2016– 2018. An association between multidimensional poverty and financial inclusion is then established showing that better access to saving facilities, the ability to save from monthly income, and better financial literacy are all associated with a lower chance of deprivation, especially among the poorest. At the same time, indebtedness through excess (more than three) credit lines hurts vulnerable households and increases their risk of deprivation. This, too, resonates with the theme throughout this volume that financial literacy and capabilities are crucial for poor households and help them use wisely financial services.

PART II CONCEPTUAL FRAMEWORKS FOR MICROFINANCE, FINANCIAL INCLUSION, AND DEVELOPMENT

2. Rethinking poverty, household finance, and microfinance* Jonathan Morduch

INTRODUCTION Sometimes new data shakes up thinking and forces us to revisit understandings that had once seemed settled. This chapter describes how understandings of poverty and finance shift when viewed with week-by-week and month-by-month household-level economic data.1 High-frequency data collected regularly during the year shows how the experience of poverty extends beyond the insufficiency of earning power captured by annual snapshots. Yearby-year measures of poverty narrow metrics of progress to changes that raise average annual earnings. Month-by-month views of poverty broaden that view, showing that deprivation can worsen and improve through the year, with people close to poverty lines moving in and out of poverty, sometimes facing “chronic instability” within the year. The high-frequency data show that the material condition of poverty is only partly captured by overall insufficiency. Instead, life in poverty is better captured by the interaction of insufficiency × instability × illiquidity. The three elements—insufficiency, instability, and illiquidity—are entwined. High-frequency data show that reducing instability and/or illiquidity can reduce exposure to poverty even when average earning power (overall insufficiency) is unchanged. We see this by reconceiving poverty in shorter time units than the year and taking a high-frequency view of data. In turn, the high-frequency view reshapes understandings of household finance. Perhaps most important, the high-frequency view shows how improving household finance can reduce high-frequency poverty by reducing illiquidity. The high-frequency view shows the power of intra-year consumption smoothing—a building block of intertemporal household economics. When distinguishing consumption from spending, it also shows that consumption smoothing often requires the spiking of spending. The high-frequency view—and the instability that it reveals—also helps to illuminate the challenging context for behavioral interventions. Behavioral economics demonstrates the value of structured contracts (like “commitment saving”) in the face of challenges like “kin taxes” and lack of self-control. High-frequency data, in contrast, highlight the value of flexibility in order to respond to unexpected shocks. These two qualities—structure and flexibility—are inherently opposed, and the need for flexibility helps to explain the low take-up rates of structured behavioral saving products (John 2020; Karlan and Morduch 2009). These insights explain some of the success and limits of microfinance. When microfinance emerged in the 1980s, advocates (most prominently Muhammad Yunus) did more than present a new tool in the fight against poverty. In making the case for microfinance, advocates also presented a theory of why poor people stayed poor. The idea behind microfinance rested on the assertion that poverty could be traced back to a lack of access to loans to finance business investments. Microfinance was positioned as a way to increase entrepreneurial income and 21

22  Handbook of microfinance, financial inclusion and development

thus reduce overall insufficiency. Poverty was distilled as low earning power, and household finance was narrowed to concerns with entrepreneurial finance (Morduch 1999). Illiquidity (apart from challenges in getting business loans) was not central to the narrative. Seeing high-frequency instability makes it easier to see why microfinance borrowing and saving are, in practice, often used to address the ups and downs of household spending needs rather than business needs. For households, microfinance holds the promise of addressing high-frequency poverty and facilitating high-frequency smoothing and spiking. Highfrequency instability explains why ex post moral hazard (“strategic default”) is a particular problem for lenders (rather than the textbook ex ante moral hazard depiction) and, in turn, why joint liability is difficult to sustain. The installment structure of typical microfinance loan contracts (i.e., high-frequency repayments) is similar to the structure of consumer lending products and contractual saving products, helping to explain how microfinance loans work naturally for purposes other than business investment, even when that departs from lenders’ nominal intentions. At the same time, microfinance loan contracts provide useful structure, but they are often too inflexible. Taking everything together, the high-frequency view—and the broad use of microfinance to meet spending needs—helps to show why microfinance loans remain popular as financial tools despite modest measured impacts on average household income. The view also points to ways to improve contracts to better meet households’ needs.

RETHINKING POVERTY The most familiar notion of poverty is the annual poverty rate measured and tabulated by policy experts, academics, and governments (e.g., World Bank 2020). The annual poverty rates quantify gaps between annualized measures of household needs (what is the minimum that a household needs for a year?) and annualized measures of household resources (are yearly income and spending high enough?).2 Academics debate how high to set poverty lines and how to form aggregate poverty measures (Atkinson 2019; Ravallion 2016), but what is seldom debated is the decision to measure poverty in year-long accounting units and to define “poverty” on the basis of yearly averages, although alternatives are possible (Atkinson 2019, chapter 3). For most households, poverty is not just experienced as overall, steady insufficiency across the year. Instead, material challenges are tied to instability during the year, with urgent needs that rise and fall, and with moments when incomes may dip well below yearly averages. This creates a gap between poverty as measured and poverty as experienced by households. A household that lives on less than the World Bank’s $1.90 a day per person poverty line in a given year may have months when average daily earnings are much greater than $1.90 per person and months with much less. The ups and downs can stretch across months and seasons. Once we see that, we can understand why even poor households save (Deaton 1991), and why they often borrow to support spending. When calculating annual incomes and expenditures, the ups and downs of households’ resources during the year disappear in the process of aggregation. Volatility and instability are averaged out, so the problem of poverty when seen through the conventional yearly lens is mostly a problem of low overall earnings. Similarly, the ups and downs of needs through the year are smoothed out to yield a notion of average need as reflected in the single poverty line for the year. Emergencies and seasonally changing requirements blur into the aggregates.

Rethinking poverty, household finance, and microfinance  23

Recognizing the divergence between poverty as measured and poverty as lived is essential for understanding how, in practice, people make economic choices and how they use household finance and microfinance.3 The centrality of month-to-month instability is the most important observation from the financial diaries completed in Bangladesh, India, and South Africa, and published as Portfolios of the Poor: How the World’s Poor Live on $2 a Day (Collins et al. 2009). The financial diaries show households wrestling with low incomes, instability, and a lack of reliable financial tools to manage important economic transactions. Together, these three elements (what the authors call “the triple whammy”) shape the experience of poverty in terms of material deprivation.4 Similar findings are reported by Morduch and Schneider (2017) in financial diaries from four sites in the United States. When poverty is redefined from an annualized quantity to a monthly measure, households are seen moving in and out of poverty during the year (Morduch and Schneider 2017, chapter 7; Morduch and Siwicki 2017). The US financial diaries showed that households whose yearly income placed them below local poverty lines still spent, on average, about three months with income above poverty lines. Non-poor households whose yearly income was between the poverty line and 150 percent of the poverty line spent an average of five months below the poverty line. Even households with average income greater than twice the poverty line spent, on average, 1.6 months with income below the poverty line. Households were less poor when resources were measured by spending rather than income, but the shift in metric did not eliminate the dips into poverty (Morduch and Schneider 2017, table 7.1).5 Re-conceptualizing “poverty” in terms of shorter accounting units leads to measures of poverty—and understandings of poverty—that more closely capture the experiences of households living with scarcity. The argument is not that we should entirely shift poverty measurement from year-long units to shorter units. Annual poverty rates importantly highlight broad changes in earning and spending power. The argument is that much can be better understood by prying loose the definition of “poverty” from its tight attachment to annual sums, putting greater weight on households’ experiences of poverty during the year. The idea that poverty can be seasonal is well-documented and has had a prominent place in development economics (e.g., Longhurst et al. 1986; Devereux et al. 2012; Khandker 2012). Still, discussions of seasonality tend to be walled off from broader conversations about poverty, largely confined to a list of concerns for farmers and agricultural laborers whose work follows the agricultural cycle (e.g., Morduch 1994). A contribution of Collins et al. (2009) was to show that instability occurs for many reasons beyond farming and agriculture. A taxi driver in Dhaka, for example, saw his earnings rise and fall by the week depending on when he could drive, whether it rained during his shifts, and the luck of being in the right place at the right time to pick up fares. A snack shop owner in a South African township similarly saw her revenues swing sharply with local economic conditions and the availability of supplies. Second, rural households who were not engaged in farm work also experienced seasonal income variation since their incomes were ultimately tied to the fluctuating fortunes of the farmers around them. Third, volatility is not just an issue of income variation; needs vary as well. The instability would not matter, or would matter much less, if households could borrow, save, and insure without difficulty. Then, they could smooth the financial peaks and fill in the financial valleys, yielding a flatter plain for the year. Yet it is the poorest individuals who, typically, are least able to get hold of the amounts of money they need at the needed times. Put

24  Handbook of microfinance, financial inclusion and development

a different way: the standard framework for analyzing poverty makes the most sense when financial constraints do not bind for individuals, but, in practice, financial constraints are especially likely to bind for people with limited resources. Measuring yearly poverty on the basis of consumption, rather than income, shifts the picture somewhat (because it accounts for consumption-smoothing across years), but it does not fundamentally reshape the picture. People with limited resources face the peaks and valleys during the year, leading to strategies and struggles that remain largely invisible (or puzzling) to poverty experts whose views only come through the annual aggregates.6 In contrast, highfrequency household data on household consumption shows that greater liquidity does more than help households to “manage” poverty; it can effectively reduce measured poverty itself— when poverty is measured in sub-year units—and can lessen deprivation. This is one place where household finance and microfinance can be critical tools. Households know this well, and it shapes their actions.

RETHINKING HOUSEHOLD FINANCE The conventional focus of household finance for small-scale entrepreneurs is finance to invest in business, buy equipment and materials, and the like. In other words, it is “entrepreneurial finance” for the self-employed. This kind of finance is at the center of the microfinance narrative. For others, “productive” loans for education, or to migrate to a place where wages are higher, may be the focus. In each case, finance is seen as a tool to increase long-term earning possibilities. For economists, these are clear steps to improve efficiency and productivity. A second, related focus is on ways to meet long-term goals, usually aligned with life-cycle stages and asset building (Sherraden 1991). Examples from the United States include saving for retirement, borrowing to purchase a house, and building an “emergency fund.” These forms of finance are legible in annual aggregates and household balance sheets, even when the ups and downs of short-term spending are not.7 The focus on household balance sheets can take attention away from the concern with short-term cash flows (Collins et al. 2009). Just as yearly poverty rates do not capture the ups and downs of poverty within the year, focusing on balance sheets obscures the within-year ups and downs of resources and needs that can be seen by tracking cash flows. The importance of managing cash flows is at the heart of Angus Deaton’s work on saving models (Deaton 1991). Deaton simulates optimal consumption trajectories in environments with substantial high-frequency income risk and limited borrowing possibilities. He labels this as “high frequency” saving in the sense that shocks come often and responses must come often too—in contrast to the low frequency need to adjust choices in preparation for large, distant life-cycle events like retirement. In Deaton’s simulations, saving and dis-saving are actively managed. Resources are accumulated and drawn down, and then accumulated again, in an ongoing cycle that responds to incoming shocks (Morduch and Schneider 2017; Elmi et al. 2020). In this context, households hold few assets on average because they are constantly building up and depleting them. The aim, in fact, is not to build long-term assets; short-term shocks come too often and needs are too urgent. Deaton’s simulations align with a key finding from Collins et al. (2009) that asset balances for low-income households are often low at any given time, but nonetheless households often engage in a great deal of financial activity, borrowing and saving frequently. Low stocks accompany large flows. This is also consistent with

Rethinking poverty, household finance, and microfinance  25

Morduch and Schneider (2017, figure 4.1) who find that American families may have low saving balances over the long term but that saving accounts are used actively to manage shortterm needs. Averaging across their respondents, 72 percent of the money in saving accounts was earmarked for needs arising within six months; 83 percent was earmarked for spending within a year. Morduch and Schneider argue that saving choices are often not so much about whether to save or not, as in the textbook depiction; instead, some of the most difficult saving choices are about what to save for. The choices often involve saving for high-frequency needs arising “soon” (in six months, say) versus saving for low-frequency needs that will occur “later” (in five, ten, or more years). Continually making these choices is not easy, and can be exhausting. Mullainathan and Shafir (2013) draw in part on the framework from Collins et al. (2009) to argue that instability, together with limited resources to accommodate the ups and downs, impose a cognitive tax as households are forced to repeatedly calibrate financial choices against opportunity costs as they figure out how to cope (Shah et al. 2012). The repetition of that task, again and again in varying forms, distracts attention from other choices that can improve long-term well-being (Mullainathan and Shafir 2013). Smoothing and Spiking, Aggregation and Distribution Deaton’s model draws on theories of consumption smoothing, stripped down to illuminate the key insights on high-frequency versus low-frequency saving. The model is simplified by assuming that needs are identical in each period. The challenge for households is then to try to spend as evenly as possible across time despite instability and illiquidity (Morduch 1995; Jappelli and Pistaferri 2017).8 Yet, as described above, in reality, needs vary too, requiring different amounts of resources at different moments. A consequence is that the need to “smooth” is often accompanied by the need to “spike” (Morduch and Schneider 2017, chapter 3). Even if income is relatively steady, households can face substantial instability as they address unexpected (and sometimes urgent) spending needs throughout the year. Moreover, the spending needs may be inherently lumpy (i.e., indivisible). Addressing a health problem may require a fixed sum, for example, and half the sum does little good. Households, in short, often need “usefully large sums” (Rutherford with Arora 2009; Collins et  al. 2009). Another way of thinking of the basic financial problem is that there are two common challenges involving moving money through time: “distribution” (or “smoothing”—taking lumps of money received at a certain moment, dividing them, and moving all or part to other times when they are needed) and “aggregation” (or “spiking”—taking bits of money and amassing them to form larger sums).9 Although it may seem contradictory, or at least complicated, households often need to distribute and aggregate—to smooth and spike—simultaneously. Spiking is sometimes necessary for smoothing. Smoothing is usually defined in terms of consumption (Jappelli and Pistaferri 2017). Spiking, in contrast, is often about spending. But the two are directly related in that smoothing consumption may sometimes require spikes in spending. For example, buying a house—typically an enormous spending spike—subsequently provides years’ worth of steady consumption flows. The same, at a smaller scale, is true for buying a television or phone. At an even smaller scale, buying a large sack of rice once a month can translate to smooth daily or weekly consumption as the bag is slowly depleted. It is tempting to equate “consumption smoothing” with “spending smoothing” but they are

26  Handbook of microfinance, financial inclusion and development

distinct notions, and one does not imply the other. Because spiking and smoothing seem at odds, spiking is often ignored in discussions of consumption smoothing, but they are of a piece, especially when moving to a high-frequency view. While household finance tends to be organized around saving, borrowing, and insuring, it is often more helpful to think in terms of the underlying goals of distribution and aggregation. Most importantly, distribution and aggregation are often achieved by combining saving and borrowing—or are achieved by using saving and borrowing in ways that depart from textbook depictions. Afzal et al. (2021) show that the desire to form meaningfully large sums can be so great that households willingly pay for saving services that facilitate the accumulation of lumps, just as they would pay for loans that achieve the same goal (albeit with different timing). From a different angle, Morduch (2010) draws on Collins et al. (2009) to show cases in which people willingly borrow at high interest rates in order to protect already-accumulated savings (which would otherwise need to be rebuilt after being drawn down). Bauer et al. (2012) provide evidence from India suggesting that the structure of microfinance loans provides a valuable way to form large sums in the absence of similarly structured ways to save. In other words, people might rather save if appropriately structured products were available, and borrowing is a next-best option in the absence of appropriate saving products. Rutherford (with Arora 2009) makes a similar point by describing borrowing of this sort as “saving down” in contrast to the conventional “saving up.”10 Putting a focus on the value of aggregation helps to explain seemingly puzzling preferences. For example, Casaburi and Machiavello (2019) find that in Kenya dairy farmers prefer being paid less frequently rather than more often. The less frequent payments can be seen as combining a saving service together with compensation. Rather than the farmers having to save up on their own to meet larger needs (and having to overcome the challenges of doing so), the farmers are presented with lump sums that have already been accumulated for them. This would have limited appeal if the farmers could save easily on their own, but—as with the depiction of microfinance in Bauer et al. (2012)—the favored mechanism can be seen as a next-best saving equivalent. See also Brune et al. (2021), who find related results in an experiment in rural Malawi. Perhaps even more surprising, Herskowitz (2021) explains sports betting in Kampala, Uganda, in a similar way. Beginning with the need for indivisible, lumpy sums, he shows how gambling can be a next-best way for bettors to get hold of usefully large sums, albeit at a high cost. He designs an experiment to establish that demand for betting falls when sports bettors are offered a better way to save. In short, households need to smooth and spike. They need lumps, and they work hard to create them, sometimes in costly ways. Financing business investment is one example of lumpy spending, but it is only one of many lumpy needs. Borrowing as a Commitment Service Behavioral economists have shown that people think about financial choices in ways that depart from textbook depictions—and have shown that “commitment saving devices” can increase saving rates by providing a helpful structure that constrains time inconsistency and overcomes present bias for individuals with hyperbolic preferences (e.g., Laibson 1997; Karlan and Morduch 2009). The examples above provide a related way to think about borrowing. As Bauer et al. (2012) and Morduch (2010) show, borrowing is not just a transaction in which money is lent to

Rethinking poverty, household finance, and microfinance  27

borrowers at a given time to be repaid at a later date with interest. Instead, borrowing is also a structured relationship. An installment plan is often decided on. A plan is made and agreed on by the borrower and lender. The lender becomes a partner in creating accountability and enforcing early commitments. If plans go awry, the lender becomes an enforcer and negotiating partner. In all instances, the borrower is not left alone to repay. The rules might not be optimal, and lenders might be coercive and predatory. Borrowers may find themselves living under a growing burden of debt and may regret their decisions to have borrowed. But, even as we recognize the dark sides, we can recognize that the qualities of most lending relationships are designed to ensure timely completion of the arrangement—and these qualities may, in themselves, be worth paying for. Behavioral economics shows that these kinds of structures are not found in traditional saving products. Saving and borrowing thus have contrasting qualities that go beyond timing, and it becomes clearer why a patient person may choose to borrow simply for the advantages provided by the structure, even when saving might otherwise seem logical. Structure and Flexibility Behavioral economics in the spirit of Laibson (1997) shows how structure is useful for savers who face temptations and competing pressures (Dupas and Robinson 2013a). On the other hand, the financial diaries show that volatility calls for the opposite—flexibility (Collins et  al. 2009). As Morduch and Schneider (2017) argue, one of the most difficult—but most important—financial tasks is to figure out how to create structure that keeps plans in line while maintaining flexibility to address unexpected changes. Structure and flexibility seem at odds, and they can be hard to fit together. Most commercial financial products have too little useful structure, while others have too little flexibility. Thus, for all of the insight of behavioral economics, the relative inattention to volatility is a sharp limit when explaining the choices of poorer households who lack reliable means to smooth ups and downs. New evidence shows that structure can help up to a point: barriers that are too high, and commitments that are too strong, can backfire or dissuade self-aware customers in the first place (John 2020). Finding the right mix of structure and flexibility can also be difficult when the user themself is unsure how much structure is enough or too much, and they might pay a high cost when they get it wrong. Households thus seek an often-elusive but important balance: how to simultaneously enact rules while making sure that the rules can also be broken when needed. Dupas and Robinson (2013a, 2013b) show how the competing demands for structure and flexibility, both implicit and explicit, complicate choices. Morduch and Schneider (2017) show that people often found a balance not by finding a financial product with appropriate attributes but by shaping the context for the use of conventional products. One man in Brooklyn, New York, asked his mother to hold his savings, which she kept in her own (conventional) savings account. His mother provided the structure and flexibility in her role as gatekeeper. Similarly, a woman in rural Mississippi found that her local bank was too accessible and convenient. She found herself too tempted to withdraw money that she would regret later. Realizing this, she had her savings be deposited automatically in a (conventional) credit union that was an hour away from home and had inconvenient hours. The barrier of time and distance created a way for her to discipline withdrawals while not ruling them out if really needed. Both examples show how—given the absence of better options—people used conventional products in unconventional ways to balance structure and flexibility.

28  Handbook of microfinance, financial inclusion and development

Debt and Other Troubles It is tempting to read financial diaries as stories of remarkable ingenuity in the face of complicated challenges. It is true that households put great effort into managing instability (Longhurst et al. 1986; Collins et al. 2009). But the challenges are steep and the options are often limited. Households often fail, in whole or in part. Little formal insurance is available, making loans and savings important ways to cope with risk. Households are vulnerable to cycles of persistent debt (Vishwanath et al. 2020), particularly high-cost unsecured debt such as from moneylenders (Reserve Bank of India 2017).11 Unless forced by regulation, banks are reluctant to lend, especially at low interest rates, to customers seeking only small loans and who cannot offer much in the way of collateral (Johnston and Morduch 2008; Cull et al. 2018). The turn to the informal sector and to noninstitutional debt is also due to perceived inadequacies of the formal banking institutions that are available. Potential customers cite lack of trust, fear of paperwork and hassles, and lack of appropriate products (Reserve Bank of India 2017). As noted above, households seek flexibility to deal with uncertainties. But, for customers without collateral and limited credit histories, such flexibility is more likely to be offered by informal lenders than by banks. When hit by a pressing emergency, a high-priced moneylender may be a compelling option, especially when the resources of family and friends have been exhausted. In a way, microfinance was designed to address this problem. It was conceived as a largescale financial solution that could displace moneylenders and do better than conventional banks, reaching poor households who lack collateral and want to borrow in small sums. By focusing on business finance, however, the language of microfinance made it harder to address households’ broad needs for debt—and their vulnerability to debt burdens. As discussed below, despite the business rhetoric, customers use microfinance broadly, and greater transparency about actual needs and uses is a first step toward addressing debt burdens.

RETHINKING MICROFINANCE The discussion above lays a foundation for rethinking microfinance.12 In principle, microfinance can be an answer to many of the problems described above—low overall earning power, high-frequency instability, and general illiquidity. To make it so, borrowers use microfinance in ways other than those advertised by microfinance leaders, opening ways for rethinking microfinance to better align with customers’ needs. That begins with recognizing how microfinance rhetoric departs from existing practices. Microfinance is usually described as entrepreneurial finance: capital to aid small-scale business investment. In practice, it is frequently used in broader ways, in keeping with the preceding discussion. Above, poverty as measured by the year was distinguished from poverty as experienced by the week, month, or season. Taking a high-frequency view on poverty moves households’ instability during the year to the center of discussion. This reality helps to explain why, in practice, microfinance borrowers often combat their poverty by using microfinance for general household needs rather than business investment. In doing so, they are addressing high-frequency poverty rather than the annualized measures calculated by statistical agencies. There is another layer to the story. As described below, design elements in standard microfinance contracts are especially well-suited for loans for non-business purposes. The installment structure of standard microfinance loans makes the loans function more like standard

Rethinking poverty, household finance, and microfinance  29

consumer loans than business loans (Armendàriz and Morduch 2000). Similarly, microfinance loan products share important features with structured saving products (Bauer et al. 2012). The “diversion” of microfinance loans away from business finance is thus less surprising. The broader uses of microfinance happen not despite the microfinance contract structure but because of it. The primary question is not about what microfinance should be used for, it is about what microfinance is—in reality—used for. That question then leads to a set of questions about goals and strategies which can help frame normative queries: 1. If the measured impacts of microfinance on business profit and income are modest or are only large for a minority of borrowers, why do so many customers continue to borrow? 2. If the contractual forms of microfinance loans look more like consumer loans than business loans, can microfinance loans be restructured to work better for those borrowers who are primarily motivated by business investment? Can microfinance loans be restructured to work better for consumers? 3. If customers seek a balance between structure and flexibility, are there ways in which microfinance designs can be improved? 4. If microfinance loans are best thought of as general-use loans, and if women are targeted as customers, how does microfinance contribute to “empowerment”? Or can microfinance—to the contrary—add to burdens for women especially? 5. Ultimately, should using microfinance loans for consumption be celebrated, or should attempts be made to limit such “diversion” (if even possible)? What would embracing microfinance for general uses look like for the sector? These questions help take us beyond the conventional narratives of microfinance. The Conventional View of Microfinance The potential to reduce annual poverty rates was central to the narrative that Muhammad Yunus created to promote microfinance. Yunus argued that microfinance could reduce annual poverty rates dramatically, even in Bangladesh—a country that gained independence only in 1971 and, in its early decades, had to immediately confront high rates of population growth and low average standards of living. In the process, microfinance became framed largely as entrepreneurial finance: loans (mainly) for business investment, with saving and insurance products as helpful add-ons. Poor borrowers were seen primarily as capital-limited, small-scale entrepreneurs, and microfinance was seen as the fuel that could power borrowers’ emancipation from poverty by easing financial constraints and thereby increasing earning power. Microfinance reinforced the argument that poverty should not be seen as a result of the personal failings of individuals, nor of toxic environments, “backward” cultures, or lack of knowledge and training. To Muhammad Yunus (1999), poor people were not seen as longtoiling laborers or farmers, but were instead recast as entrepreneurs. They were frustrated entrepreneurs, for sure, lacking capital but possessing energy and ambition. Poor people, in this view, sought opportunities foremost: chances to make the most of their talents by building and growing small businesses. The aim of microfinance was to provide the missing financing and to do so at costs well below the interest rates charged by moneylenders.

30  Handbook of microfinance, financial inclusion and development

At a conceptual level—at the level of ideology—microfinance was a clear success in its first decades.13 By depicting poor people as entrepreneurs, and as customers for financial services rather than beneficiaries of government handouts, microfinance saw dignity and agency in poor people. Microfinance also elevated women and, especially in South Asia, made poor women the central focus. Microfinance succeeded in part because it capitalized on emerging understandings in economics. For economists, the diagnoses of market failure hinged on examples of businesses thwarted in their pursuit of loans by asymmetric information, limited liability, and the attendant challenges of moral hazard and adverse selection (Stiglitz and Weiss 1981; Townsend 1979; Armendáriz and Morduch 2010, chapter 2; Banerjee 2013). Theories of moral hazard and adverse selection became the basis of theories of microfinance, and the focus was on how to lend to risky businesses (see, e.g., chapters 2, 4, and 5 of Armendáriz and Morduch 2010). The theoretical structures highlighted a fundamental challenge for lenders: how to know if repayment difficulties are caused by risks outside of a borrower’s control versus difficulties due to insufficient care taken by the borrower. For the most part, the lenders’ challenge was depicted as ex ante moral hazard (e.g., Stiglitz 1990). The lender’s difficulty was in their inability to easily observe borrowers’ efforts (e.g., Holmström 1979). Group lending with joint liability was posed as a solution (Ghatak and Guinnane 1999). If a borrower’s neighbors have good local information by virtue of their proximity (and if they are interested in loans for themselves), the neighbors can monitor their neighbors and enforce contracts. The solution that microfinance innovators suggested was to make neighbors jointly liable for each others’ loans (sometimes implicitly, sometimes explicitly). In theory, loan officers do not need to acquire the information held by the groups; the loan officers just need to enforce the optimal incentive-compatible contract. With the contract in force, borrowers will work hard, investments will succeed and loans will be repaid. High microfinance loan repayment rates appear to confirm the power of the mechanism. Ex Post Moral Hazard Crucially missing from this picture is serious consideration of how to address the broader risks described in previous sections, those arguably outside the borrowers’ control. These are not necessarily business risks. Risks that undermine the ability to repay loans include risks to health, shelter, and food security. They include risks to income earned from sources other than the particular investment to which microfinance is attached. The risks may involve obligations to extended family members and the community, linked in mutual support. So, even if ex ante moral hazard (will borrowers work hard enough to succeed at the given business?) is managed, ex post moral hazard still arises (will borrowers succeed at the business yet still not repay?). Ex post moral hazard is often depicted as “strategic default,” but the ethics and contexts can be complicated. Labeling the choice to default as “strategic” implies greater agency than may actually be the case, and reducing default rates to zero may be cruel, undermining the welfare gains from lending when it means taking strong actions to discipline defaulters who are already seriously down on their luck, perhaps through little fault of their own (Gertler et al. 2021). Most evidence on the causes of repayment difficulty focuses on the efficacy of microfinance contracts (e.g., Ahlin and Townsend 2007), but the question here is about the nature of the underlying risks and triggers of non-payment. Simtowe, Zeller, and Phiri (2006) report on

Rethinking poverty, household finance, and microfinance  31

survey data from Malawi, for example, which shows that the single most prevalent cause of default is ex post moral hazard in microfinance groups (the unwillingness to repay, not the inability to repay). A quarter of their sample had the money to repay, but chose not to. Another 16 percent could not repay due to mismanagement of various kinds (ex ante moral hazard). Thus, 41 percent of default could be attributed to either mismanagement or misuse of funds. Of the balance, 18 percent was due to natural disasters (i.e., slightly more than was attributed to ex ante moral hazard) and 24 percent was due to “low profit” (but not mismanagement), about the same as was attributed to ex post moral hazard.14 The lines between misuse, mismanagement, and poor luck are not always clear, and the efficient level of risk-sharing implied in joint-liability contracts can be hard to reach. Borrowers’ groups are thus implicated, through joint liability, in shouldering a wide range of risks faced by households. The risks can be a heavy load to bear for group members, and it’s unsurprising that microfinance borrowers complain about the burden imposed by joint liability, especially as loan sizes (and thus obligations to others) grow. As Ahlin (2020) notes, drawing on evidence from Thailand, the joint-liability contract may itself push toward group formation which is not particularly well diversified against risk. When customers get into trouble, loan officers then face a dilemma. Recognition of the broader risks facing borrowers—risks that arise at least in part from life in risky environments—leads loan officers to step in to deliver fairer outcomes than the strict application of group lending under joint liability would produce. Rather than insisting that group members pitch in to repay for all overdue loans of others, and rather than cutting off all group members from future borrowing, loan officers may try to negotiate and adjudicate. One solution is to replace the borrower in default, re-constituting the group. With steps like this, the excesses of group lending are reduced. But so too is the bite of group lending. It is a difficult contract to maintain because of the inherent contradictions: given their poverty, borrowers are ill-equipped to provide broad insurance mechanisms for each other, yet the contract pushes for it and in some cases punishes borrowers who, to the best of their ability, appear to be doing everything right.15 It is thus not surprising that joint liability has ceased to be seen as the key to microfinance (Attanasio et al. 2015).16 Grameen Bank itself dropped joint liability at the start of this century (Dowla and Barua 2006). A broader study by De Quidt et al. (2018) documents the decline of joint liability in the global MIX Market dataset and ties the fall to increased commercialization. The use of groups and group meetings may persist, but the use of joint liability—the best-known microfinance innovation—has greatly faded. Microfinance Loans Look Like Consumer Loans How then does microfinance work? Microfinance has always worked through multiple, overlapping mechanisms, even though group lending with joint liability took most of the attention. Two other mechanisms are dynamic incentives (borrowers are eligible for the next loan only if they have successfully repaid prior loans) and the division of repayments into small, frequent installments that start soon after the loan is disbursed (Armendáriz and Morduch 2000; Bauchet and Morduch 2019). The latter is an odd feature. A conventional business loan has a single “balloon” or “bullet” repayment to allow borrowers time to invest and reap profits before repaying. In some cases, there might be more than one installment, but not the small weekly/bi-weekly/monthly installments that are usual in microfinance contracts.

32  Handbook of microfinance, financial inclusion and development

The frequency of microfinance installments, though, has the advantage of making it possible for borrowers to repay loans with flows of household income that are independent of revenues from the targeted business. The small sums needed for the installments can be put together with income from various sources. In fact, it is necessary to do so, at least at first, in the period before businesses produce revenues (Field et al. 2013). More striking, it is not strictly necessary to even have a business in order to want a microfinance loan and to be able to repay it on time. For better or worse, this structure with small, regular payments makes microfinance operate much more like a credit card (or an installment-based consumer loan) than a business loan, and customers understand the possibilities that this opens (Morduch 2018). As with credit cards, microfinance can help with both smoothing and spiking. The possibility of using microfinance for consumption purposes (to pay for healthcare, to keep food on the table, to pay for travel or durable goods) means that microfinance is in practice often not used entirely for business. This is not quite “misuse” if the borrower can afford to repay the loans through reliable strategies, although the choice may deviate from the assumption that loans will be used for business investment. Surveys of borrowers reveal that a large share of microfinance loans are used for broad purposes, both to smooth and spike. Stuart Rutherford concludes from his research on a sample of Grameen Bank borrowers that only a minority borrow mainly for business: On the one hand, it is clear that an early hope of microfinance lending—that virtually every loan would be invested in a microenterprise—has not come about. On the other hand, businesses and asset-investment uses are responsible for more than half the value of loans disbursed, though concentrated among the minority of borrowers well placed to use them in this way. (Collins et al. 2009, p. 167)

Similar results—about half of microfinance loans going to non-business purposes—are documented, for example, in Indonesia by Johnston and Morduch (2008) and in Mongolia by Attanasio et al. (2015).17 Islam and Maitra (2012) consider health shocks in Bangladesh. Using a large panel data set from rural Bangladesh, they show that, given the lack of access to health insurance, households often sell livestock to address health shocks. But once microfinance is introduced, households are more likely to cope with microfinance loans instead, a less costly strategy. Gertler et al. (2009) similarly find that households with access to microfinance in Indonesia are better able to protect their consumption after health shocks. Calis et al. (2017) turn to coping with natural disasters, investigating a large cyclone that struck India in 2013. They find that microfinance helped borrowers to mitigate the shock, again by providing needed liquidity.18 Why Microfinance Loans Are Popular When Measured Impacts Are Modest Early randomized trials yielded a mixed verdict for microfinance impacts: they found a clear impact on business activity but little impact on overall household income or consumption (Banerjee et al. 2015). The modest findings of the initial RCTs remain part of the picture and align with expectations from earlier non-experimental studies (Morduch 1999). But the results raise a critical question: if impact on average income and consumption is hard to find, why do borrowers continue to borrow? The early RCT results force us to think harder about consumers’ financial choices. Are borrowers as ignorant or irrational as the results imply?

Rethinking poverty, household finance, and microfinance  33

One response to this demand puzzle is that the results are from margins that may be unrepresentative (Morduch 2020). Subsequent work finds a positive impact on incomes in other settings. Breza and Kinnan (2020), for example, analyze the impact of halting microfinance in South India and show increases in income that occur through effects on wages. Cai et al. (2020) show a large, positive impact on incomes in rural China in a particularly poor group of villages, driven partly by increased migration. These studies suggest a complementary answer to the demand puzzle. The low levels of impact found in the early RCT studies could partly arise because borrowers divert money away from business investment in order to spend on health needs and consumer goods and pay down more expensive debt, etc. The outcomes from those uses are unclear, but they align with the discussion of high-frequency poverty and finance above and suggest very different terms by which to evaluate microfinance.19 Annual poverty rates and annual household income are thus not necessarily the outcomes that are most likely to be affected by microfinance. Attanasio et al. (2015) find, for example, that food consumption increases even if income does not. Since microfinance loans can be used in multiple ways, evaluating their impact on business investment or yearly income provides a test of Yunus’s narrative but misses other potential impacts. Microfinance Loans Look Like Structured Saving Products One response to the observation that microfinance loans are used to fund consumption is that customers should be encouraged to save instead. Without getting into whether that should be so, it is worth noting that the structure of microfinance loans already resembles that of contractual saving products, where savers are expected to deposit a given amount on a regular schedule until a certain goal is met (e.g., Dupas and Robinson 2013b). Put another way: microfinance loans share features with consumer installment loans, and both share features with structured saving products. This is the sense above that a fundamental element of loans (including microfinance loans) is that they come bundled with “behavioral” commitment services (Bauer et al. 2012). As noted in the discussion of household finance, saving and borrowing can both be ways to translate flows of money into “usefully large sums.” With borrowing, the lump is delivered sooner than through saving and is more costly. But the function of reliably transforming flows into lumps may be more important than the timing and costs (Bauer et al. 2012; Afzal et al. 2021). Microfinance allows people to aggregate resources when functionally equivalent saving products do not exist. Kast and Pomeranz (2018) show this in reverse: when improved saving possibilities were introduced in Chile, short-term borrowing declined by 5–20 percent. An attraction of microfinance borrowing for some is as a way to build up lump sums when it is not possible to save in a structured format. Too Much Structure, Too Little Flexibility The microfinance installment structure transforms a “business” loan into something that looks more like a consumer loan or a structured saving product. In a way, that is the hidden genius of microfinance. But that does not mean that the contract is optimal, for either business or consumer purposes. Field et al. (2013), for example, work with a lender in Kolkata to experimentally test the impact of giving borrowers more time to invest. Providing borrowers with a “grace period” before they start to repay increases business investment and average profitability and encourages risk-taking by customers.

34  Handbook of microfinance, financial inclusion and development

Similarly, Battaglia et al. (2018) provide an experiment with year-long loans from BRAC in Bangladesh, in which customers were allowed to pay two installments late without penalty. (The usual structure involved monthly installments.) The resulting flexibility led borrowers to invest more, increasing assets by 51 percent relative to the control group. Their revenues increased on average by 87 percent and profits by 25 percent. Borrowers also saw more risk, with sales volatility rising 80 percent relative to the control group. Still, given the flexibility, loan defaults dropped and customer retention increased. Banerjee (2013) notes that while structure may be helpful, weekly repayments may impose too much structure. For some borrowers, monthly installments, as at BRAC, may provide a better trade-off between structure and flexibility. Gendered Empowerment or Burden? Of all the claims for microfinance, the most compelling may be the promise that access to small loans can empower women. By targeting women with business finance, the hope is to raise their earning power—and, with that, to increase women’s bargaining power, status, and autonomy. The vision is tied to business finance, and, as described above, this is only one element of microfinance. As shown by Bernhardt et al. (2019), investments in the businesses of women are often diverted to those of men in the same household. Similarly, Riley (2020) shows that women are more successful in business when they can hide their resources from family members. Finance may be helpful to households, but it operates within cultural and social constraints that are difficult to erode. The focus on financial management above points to a different set of issues for female borrowers. As some note, microfinance targeted at women can create burdens of debt that women are forced to shoulder (Karim 2011). Guérin (2014) finds from her research in Tamil Nadu, India, that the main demand for microfinance in her site is for non-business purposes (e.g., food security, health, religious and social obligations, and repaying other debt). For the women that she and her collaborators study, access to microfinance can provide liquidity, giving women more options as they juggle debt. Yet the backdrop to such juggling is a heavy burden of debt borne by women, money owed both to microfinance lenders and to informal lenders (Guérin et al. 2020). These burdens land most heavily on poorer women with fewer social advantages. By targeting women, microfinance lenders assist in shifting the task of “making ends meet” to women rather than men, even as they (helpfully) provide women with better financial tools to do so. Only by recognizing the reality of microfinance as a cash management tool can these unequal burdens be seen, described, and assessed.

CONCLUDING THOUGHTS People manage their economic lives at different frequencies: week by week, month by month, and year by year. Governments and researchers, however, usually collect data that only give snapshots of annual aggregates like yearly income or expenditure. In principle, poverty can also be measured by the month, for example, or by the season. These alternative choices help to highlight households’ short-term but often chronic instability—and the limited liquid assets and financial tools that households have available to respond. I argue that household finance and microfinance are best understood by also using a highfrequency lens to follow households as they experience ups and downs throughout the year. The

Rethinking poverty, household finance, and microfinance  35

same is true for poverty, as households with limited resources move in and out of poverty. The intersecting concerns—overall insufficiency, instability, and illiquidity—are bound together. Typical poverty analyses ignore short-term instability and thus pay little attention to short-term illiquidity. But once we see the challenge of instability, we can immediately see broader needs for access to reliable finance. This, in turn, pushes for a broader perspective on household finance and microfinance as tools to manage instability and illiquidity in order to facilitate general spending needs. In this conception, business investment may or may not be an important goal. Yunus wove a coherent narrative of microfinance and its emancipatory possibility (Yunus 1999). His result was grand and sweeping, and, with that, both powerful and problematic. Theory and practice, research and experiment, reveal inherent contradictions in this conventional microfinance narrative. Within those contradictions lie an answer, or the seeds of an answer, as to why microfinance has failed to manifest its transformative promise yet why it has thrived, continuing to draw customers and investors. To better understand the realities of microfinance—its possibilities and its contradictions— it is necessary to return to the notions of poverty and household finance upon which the initial claims and ambitions were based. This essay is an attempt to do that, selectively drawing together threads rather than providing a review of related literatures. Policy experts may be tempted to minimize concern with the short-term ups and downs of poverty and the short-term maneuvers of finance. They may be tempted to focus only on policies and programs that promise large, long-term transformations. For households living with scarcity, however, the short term is the path to the long term. Without the ability to get through the short term, long-term goals too frequently lie out of reach, unmet. Being able to ignore short-term challenges is a privilege, usually accessible only to better-off households.

NOTES *

This chapter pulls a thread through older and newer work with co-authors. Tim Ogden provided very useful comments, and I appreciate the chance to test-drive these ideas at the 5th Dvara Research Conference in June 2021. I am grateful for financial support from the Mastercard Impact Fund in collaboration with the Mastercard Center for Inclusive Growth provided to the NYU Financial Access Initiative. The views here are mine only and are not necessarily those of the funders. 1. The ideas here and in much of the chapter grew from Morduch and Schneider (2017) and Collins et al. (2009). I also draw on ongoing research on high-frequency poverty with Joshua Merfeld using the most recent ICRISAT household data from India. An older literature on seasonal poverty and episodic poverty (e.g., Longhurst et al. 1986) aligns with the discussion here. 2. The ideas here and in much of the chapter draw on chapter 7, “Sometimes Poor,” of Morduch and Schneider (2017) and on Collins et al. (2009) and Morduch (2012). I also draw on ongoing empirical research with Joshua Merfeld with the ICRISAT data from India. 3. The divergence can be hard to see in typical data sets, and as Longhurst et al. (1986, p. 86) note: “it is crucially important that awareness is increased: that urban-based, season-proofed professionals become more aware of what rural people know only too well about how adverse seasonality affects them and how they try to handle it. Far greater knowledge and appreciation is required of the pattern of income earning and food acquiring activities of vulnerable people, especially when urban-based professionals are least likely to travel at the times of year when things are worst for rural people.” 4. Collins et al. (2009) focus on material deprivation, and I follow that path here. The full experience of poverty is much broader, including deprivations that may be political and social, that may entail challenges to physical and mental health, and limits to other capabilities in the sense of Amartya Sen (1999). 5. The sample in Morduch and Schneider (2017) is limited to four sites and is not representative. Parolin et al. (2020) make a similar argument based on broader data in the United States.

36  Handbook of microfinance, financial inclusion and development

6. The urgency and frequency of the within-year challenges have another consequence for poor households. Having to cope with the downturns exposes households with limited means to exploitation, forces reliance on the beneficence of friends and family, and increases the risk of persistent debt (Vishwanath et al. 2020). It may also create gendered burdens through the unequal responsibilities for coping (Guérin et al. 2020). These elements of poverty are also hidden from view as data are swept into annual aggregates. Paying attention to poverty within the year brings into focus these broader implications of material deprivation. 7. The attention to this kind of asset-building and changes to annual household balance-sheets is the basis for the framework for poverty reduction advanced by Sherraden (1991), and it has analogues with the rhetoric of microfinance. Grameen Bank has long offered higher education loans and housing loans alongside loans designated for business investment. 8. The optimal consumption path will depend on interest rates, discount rates, and various preference parameters. Moreover, the model is most naturally applied to non-durables. With consumer durables, the timing of spending and consumption can diverge substantially. 9. Addressing risk is a related challenge, and, for brevity, it is left to the side here. Saving and borrowing, though, are common ways to cope with risk given the unavailability of formal insurance. 10. These studies align with insights from behavioral economics, especially the limits to conventional ways of thinking about saving and borrowing through the lens of steady discount rates and interest rates. 11. See comparative data from India, China, Thailand, the United Kingdom, the United States, Australia, and Germany in Badrinza et al. (2019). 12. This section draws on many sources, including chapter 6 of Collins et  al. (2009), “Rethinking Microfinance: The Grameen II Diaries,” which draws on Stuart Rutherford’s diaries of Grameen Bank borrowers. I am grateful for long conversations with Rutherford over the years, comparing and testing ideas. 13. By a “clear success,” I mean that the rhetorical formulation was convincing to many and aligned with prevailing political and social imperatives. See, e.g., Geismer (2020). Not everyone was convinced, and microfinance has had its share of critics, especially from the left (e.g., Bateman and Chang 2012). 14. Default is endogenous, and it is not possible to describe underlying risks that created difficulty but which did not result in default. 15. In theory and in lab experiments, joint liability can even exacerbate risk through allowing freeriding via the implicit insurance mechanism (Giné et al. 2010; Fischer 2013). 16. See Mahmud (2020) for a counter-example documenting how switching from individual to joint liability was helpful for a lender in Pakistan. 17. Karlan et al. (2016) show in a sample from Manila that borrowers added to business investment in a magnitude comparable to loan sizes. Since money is fungible, the money directly received from the microfinance lender was not necessarily spent on business, yet it freed other money that could be spent on business, and, ultimately, investment took place. This is consistent with the demand for general-use loans, for which business is one use. 18. Some of the reduction in harm could come from increased overall income, or the diversification of income flows, rather than from consumption-smoothing. Note too that mobile money has been shown to provide liquidity in a similar way, by facilitating payments within social and family networks (e.g., Jack and Suri 2014; Lee et al. 2021). 19. As noted earlier, by enabling consumption smoothing, the loans may be reducing poverty as experienced by households (and as measured by sub-year poverty measures, even if not by annualized rates).

REFERENCES Afzal, Uzma, Giovanna d’Adda, Marcel Fafchamps, Simon Quinn, and Farah Said. 2021. “Two Sides of the Same Rupee? Comparing Demand for Microcredit and Microsaving in a Framed Field Experiment in Rural Pakistan.” Economic Journal 128: 2161–2190. Ahlin, Christian. 2020. “Group Lending, Matching Patterns, and the Mystery of Microcredit: Evidence from Thailand.” Quantitative Economics 11: 713–759.

Rethinking poverty, household finance, and microfinance  37

Ahlin, Christian and Robert Townsend. 2007. “Using Repayment Data to Test across Models of Joint Liability Lending.” The Economic Journal 117(517): F11–F51. Armendáriz, Beatriz and Jonathan Morduch. 2000. “Microfinance Beyond Group Lending.” Economics of Transition 8: 401–420. Armendáriz, Beatriz and Jonathan Morduch. 2010. The Economics of Microfinance, Second edition. Cambridge, MA: MIT Press. Atkinson, Anthony B. 2019. Measuring Poverty Around the World. Princeton, NJ: Princeton University Press. Attanasio, Orazio, Britta Augsburg, Ralph De Haas, Emla Fitzsimons, and Heike Harmgart. 2015. “The Impacts of Microfinance: Evidence from Joint-Liability Lending in Mongolia.” American Economic Journal: Applied Economics 7(1): 90–122. Badrinza, Cristian, Vimal Balasubramaniam, and Tarun Ramadorai. 2019. “The Household Finance Landscape in Emerging Economies.” Annual Review of Financial Economics 11: 109–129. Banerjee, Abhijit. 2013. “Microcredit under the Microscope: What Have We Learned in the Past Two Decades, and What Do We Need to Know?” Annual Review of Economics 5(1): 487–519. Banerjee, Abhijit V., Esther Duflo, Rachel Glennerster, and Cynthia Kinnan. 2015. “The Miracle of Microfinance? Evidence from a Randomized Evaluation.” American Economic Journal: Applied Economics 7(1): 22–53. Banerjee, Abhijit, Dean Karlan, and Jonathan Zinman. 2015. “Six Randomized Evaluations of Microcredit: Introduction and Further Steps.” American Economic Journal: Applied Economics 7(1): 1–21. Bateman, Milford and Ha-Joon Chang. 2012. “Microfinance and the Illusion of Development: From Hubris to Nemesis in Thirty Years.” World Economic Review 1: 13–36. Battaglia, M., S. Gulesci, and A. Madestam. 2018. “Repayment Flexibility and Risk Taking: Experimental Evidence from Credit Contracts.” CEPR Discussion Paper 13329. Bauchet, Jonathan and Jonathan Morduch. 2019. “Paying in Pieces: A Natural Experiment on Demand for Life Insurance under Different Payment Schemes.” Journal of Development Economics 139: 69–77. Bauer, Michal, Julie Chytilová, and Jonathan Morduch. 2012. “Behavioral Foundations of Microcredit: Experimental and Survey Evidence from Rural India.” American Economic Review 102(2): 1118–1139. Bernhardt, Arielle, Erica Field, Rohini Pande, and Natalia Rigol. 2019. “Household Matters: Revisiting the Returns to Capital among Female Microentrepreneurs.” American Economic Review: Insights 1(2): 141–160. Breza, Emily and Cynthia Kinnan. 2020. “Measuring the Equilibrium Impacts of Credit: Evidence from the Indian Microfinance Crisis.” National Bureau of Economic Research Working Paper No. 24329. Brune, Lasse, Eric Chyn, and Jason Kerwin. 2021. “Pay Me Later: Savings Constraints and the Demand for Deferred Payments.” American Economic Review 111(7): 2179–2212. Cai, Shu, Albert Park, and Sangui Wang. 2020. “Microfinance Can Raise Incomes: Evidence from a Randomized Control Trial in China.” Manuscript. Calis, T., S. Gangopadhyay, N. Ghosh, R. Lensink, and A. Meesters. 2017. “Does Microfinance Make Households More Resilient to Shocks? Evidence from the Cyclone Phailin in India.” Journal of International Development 29(7): 1011–1015. Casaburi, Lorenzo and Rocco Macchiavello. 2019. “Demand and Supply of Infrequent Payments as a Commitment Device: Evidence from Kenya.” American Economic Review 109(2): 523–555. Collins, Daryl, Jonathan Morduch, Stuart Rutherford, and Orlanda Ruthven. 2009. Portfolios of the Poor: How the World’s Poor Live on $2 a Day. Princeton, NJ: Princeton University Press. Cull, Robert, Asli Demirgüç-Kunt, and Jonathan Morduch. 2018. “The Microfinance Business Model: Modest Profit and Enduring Subsidy.” World Bank Economic Review 32(2): 221–244. De Quidt, Jonathan, Thiemo Fetzer, and Maitreesh Ghatak. 2018. “Commercialization and the Decline of Joint Liability Microcredit.” Journal of Development Economics 134: 209–225. Deaton, Angus. 1991. “Saving and Liquidity Constraints.” Econometrica 59: 1221–1248. Devereux, Stephen, Sabates-Wheeler, Rachel, and Richard Longhurst (eds). 2012. Seasonality, Rural Livelihoods and Development. London and New York: Earthscan and Routledge.

38  Handbook of microfinance, financial inclusion and development

Dowla, Asif and Dipal Barua. 2006. The Poor Always Pay Back: The Grameen II Story. West Hartford, CT: Kumarian Press. Dupas, Pascaline and Jonathan Robinson. 2013a. “Savings Constraints and Microenterprise Development: Evidence from a Field Experiment in Kenya.” American Economic Journal: Applied Economics 5(1): 163–192. Dupas, Pascaline and Jonathan Robinson. 2013b. “Why Don’t the Poor Save More? Evidence from Health Savings Experiments.” American Economic Review 103(4): 1138–1171. Elmi, Sheida Isabel and the Aspen Financial Security Program. 2020. “The Cycle of Savings: What We Gain When We Understand Savings as a Dynamic Process.” Aspen Financial Security Program. Field, Erica, Rohini Pande, John Papp, and Natalia Rigol. 2013. “Does the Classic Microfinance Model Discourage Entrepreneurship among the Poor? Experimental Evidence from India.” American Economic Review 103(6): 2196–2226. Fischer, Greg. 2013. “Contract Structure, Risk Sharing and Investment Choice.” Econometrica 81(3): 883–939. Geismer, Lily. 2020. “Agents of Change: Microenterprise, Welfare Reform, the Clintons, and Liberal Forms of Neoliberalism.” Journal of American History 107(1): 107–131. Gertler, Paul, David Levine, and Enrico Moretti. 2009. “Do Microfinance Programs Help Families Insure Consumption against Illness?” Health Economics 183: 257–273. Gertler, Paul, Brett Green, and Catherine Wolfram. 2021. “Digital Collateral.” NBER Working Paper No. 28724, May. Ghatak, Maitreesh and Timothy W. Guinnane. 1999. “The Economics of Lending with Joint Liability: Theory and Practice.” Journal of Development Economics 60: 195–228. Giné, Xavier, Pamela Jakiela, Dean Karlan, and Jonathan Morduch. 2010. “Microfinance Games.” American Economic Journal: Applied Economics 2(3): 60–95. Guérin, Isabelle. 2014. “Juggling with Debt, Social Ties, and Values: The Everyday Use of Microcredit in Rural South India.” Current Anthropology 55(S9): S40–S50. Guérin, Isabelle, Christophe Jalil Nordman, and Elena Reboul. 2020. “The Gender of Debt and Credit: Insights from Rural Tamil Nadu.” IZA Discussion paper series DP No. 13891. Herskowitz, Sylvan. 2021. “Gambling, Saving, and Lumpy Liquidity Needs.” American Economic Journal: Applied Economics 13(1): 72–104. Holmström, Bengt. 1979. “Moral Hazard and Observability.” Bell Journal of Economics 10(1): 74–91. Islam, Asadul and Pushkar Maitra. 2012. “Health Shocks and Consumption Smoothing in Rural Households: Does Microcredit Have a Role to Play?” Journal of Development Economics 97: 232–243. Jack, William and Tavneet Suri. 2014. “Risk Sharing and Transactions Costs: Evidence from Kenya’s Mobile Money Revolution.” American Economic Review 104(1): 183–223. Jappelli, Tullio and Luigi Pistaferri. 2017. The Economics of Consumption. Oxford: Oxford University Press. John, Anett. 2020. “When Commitment Fails—Evidence from a Field Experiment.” Management Science 66(2): 503–529. Johnston, Don Jr. and Jonathan Morduch. 2008. “The Unbanked: Evidence from Indonesia.” World Bank Economic Review 22(3): 517–537. Karim, Lamia. 2011. Microfinance and Its Discontents: Women in Debt in Bangladesh. Minneapolis, MN: University of Minnesota Press. Karlan, Dean and Jonathan Morduch. 2009. “Access to Finance” (with Dean Karlan). In Dani Rodrik and Mark Rosenzweig (eds), Handbook of Development Economics, Volume 5. Amsterdam: Elsevier: 4704–4784. Karlan, Dean, Adam Osman, and Jonathan Zinman. 2016. “Follow the Money Not the Cash: Comparing Methods for Identifying Consumption and Investment Responses to a Liquidity Shock.” Journal of Development Economics 121: 11–23. Kast, Felipe and Dina Pomeranz. 2018. “Saving Accounts to Borrow Less: Evidence from Chile.” Working paper, June. Khandker, Shahidur. 2012. “Seasonality of Income and Poverty in Bangladesh.” Journal of Development Economics 97: 244–256.

Rethinking poverty, household finance, and microfinance  39

Laibson, David. 1997. “Golden Eggs and Hyperbolic Discounting.” Quarterly Journal of Economics 112(2): 443–478. Lee, Jean N., Jonathan Morduch, Saravana Ravindran, Abu Shonchoy, and Hassan Zaman. 2021. “Poverty and Migration in the Digital Age: Experimental Evidence on Mobile Banking in Bangladesh.” American Economic Journal: Applied Economics 13(1): 38–71. Longhurst, Richard, Robert Chambers, and Jeremy Swift. 1986. “Seasonality and Poverty: Implications for Policy and Research.” IDS Bulletin 17(3): 67–70. Mahmud, Mahreen. 2020. “Repaying Microcredit Loans: A Natural Experiment on Liability Structure.” Journal of Development Studies 56(6): 1161–1176. Morduch, Jonathan. 1994. “Poverty and Vulnerability.” American Economic Review (AEA Papers and Proceedings) 84(2): 221–225. Morduch, Jonathan. 1995. “Income Smoothing and Consumption Smoothing.” Journal of Economic Perspectives 9(3): 103–114. Morduch, Jonathan. 1999. “The Microfinance Promise.” Journal of Economic Literature 37(4): 1569–1614. Morduch, Jonathan. 2010. “Borrowing to Save.” Journal of Globalization and Development 102(2). Morduch, Jonathan. 2012. “Notre façon de voir la pauvreté” [‘How We See Poverty’].” FACTS, Special Issue 4 (Lutte contre la pauvreté) January: 14–19. Morduch, Jonathan. 2018. “Microfinance as a Credit Card?” Limn 9 (Little Development Devices / Humanitarian Goods). Stephen J. Collier, Jamie Cross, Peter Redfield and Alice Street (eds). Morduch, Jonathan. 2020. “Why RCTs Failed to Answer the Biggest Questions about Microcredit Impact.” World Development 127: 104818. Morduch, Jonathan and Rachel Schneider. 2017. The Financial Diaries: How American Families Cope in a World of Uncertainty. Princeton, NJ: Princeton University Press. Morduch, Jonathan and Julie Siwicki. 2017. “In and Out of Poverty: Poverty Spells and Income Volatility in the US Financial Diaries.” Social Service Review 91(3): 390–421. Mullainathan, Sendhil and Eldar Shafir. 2013. Scarcity: Why Having Too Little Means so Much. New York: Times Books. Parolin, Zachary, Megan Curran, Jordan Matsudaira, Jane Waldfogel, and Christopher Wimer. 2020. “Monthly Poverty Rates in the United States during the COVID-19 Pandemic.” Columbia University Poverty and Social Policy Working Paper, October 15, 2020. Ravallion, Martin. 2016. The Economics of Poverty: History, Measurement, and Policy. Oxford: Oxford University Press. Reserve Bank of India. 2017. Report of the Household Finance Committee: Indian Household Finance. July 2017. Riley, Emma. 2020. “Resisting Social Pressure in the Household Using Mobile Money: Experimental Evidence on Microenterprise Investment in Uganda.” Manuscript. Rutherford, Stuart with Sukhwinder Arora. 2009. The Poor and Their Money, 2nd edition. Bourton on Dunsmore, Warwickshire: Practical Action Publishing. Sen, Amartya. 1999. Development as Freedom. New York: Knopf. Shah, Anuj, Sendhil Mullainathan, and Eldar Shafir. 2012. “Some Consequences of Having Too Little.” Science 338(2 November): 682–685. Sherraden, Michael. 1991. Assets and the Poor: A New American Welfare Policy. Armonk, NY: M.E. Sharpe. Simtowe, Franklin, Manfred Zeller, and Alexander Phiri. 2006. “Determinants of Moral Hazard in Microfinance: Empirical Evidence from Joint Liability Lending Programs in Malawi.” African Review of Money Finance and Banking: 5–38. Stiglitz, Joseph. 1990. “Peer Monitoring and Credit Markets.” World Bank Economic Review 4(3): 351–366. Stiglitz, Joseph and Andrew Weiss. 1981. “Credit Rationing in Markets with Imperfect Information.” American Economic Review 71: 393–410. Townsend, Robert. 1979. “Optimal Contracts and Competitive Markets with Costly State Verification.” Journal of Economic Theory 21: 265–293.

40  Handbook of microfinance, financial inclusion and development

Vishwanath C., Monami Dasgupta, and Misha Sharma. 2020. “Household Finance in India: Approaches and Challenges.” Dvara Research, India. World Bank. 2020. Poverty and Shared Prosperity 2020: Reversals of Fortune. Washington, DC: World Bank. Yunus, Muhammad. 1999. Banker to the Poor: Micro-Lending and the Battle against World Poverty. New York: PublicAffairs.

3. Assessment of microfinance institutions and their impact: evidence from a scientometric study Begoña Gutiérrez-Nieto and Carlos Serrano-Cinca

INTRODUCTION The provision of microcredits—small loans for people excluded from financial markets—has been part of development strategies in recent years. One of the recurring goals of microfinance institutions (MFIs) is to help eradicate poverty, which is the first of the Sustainable Development Goals adopted by the United Nations Member States in 2015. At least, so far, MFIs have been improving financial inclusion by serving people previously unable to access financial services (Brown et al., 2015). The assessment of MFIs is complex because they try to be sustainable and efficient, but given their social purpose, they have to avoid drifting from their mission and charging unfair interest rates. Among development initiatives, microfinance is probably the most concerned about its impact (Balkenhol, 2018). While Yunus won the Nobel Peace Prize in 2006 for the development of microcredit, the Nobel Prize in Economics went to Kremer, Banerjee, and Duflo in 2019 for their experimental approach to alleviating global poverty, with work on the impact of microfinance among their research contribution (Banerjee et al., 2015a). Little convincing evidence exists that microcredit actually has had a significant positive impact on development (Committee for the Prize in Economic Sciences in Memory of Alfred Nobel, 2019). All stakeholders involved in microfinance—practitioners, investors, donors, regulators, or policymakers, as well as academia—are concerned about the impact of MFIs. This chapter describes the main methodologies and standards for measuring financial and social performance in microfinance. There are many microfinance literature reviews (Cull & Morduch, 2017; Duvendack et al., 2011; Duvendack et al., 2014; Duvendack & Mader, 2020; Hermes & Hudon, 2018; Van Rooyen et al., 2012). We adopt a scientometric approach, based on the keyword co-occurrence analysis and citation network analysis of major academic studies. An assessment of the performance of MFIs must take into account both their social and financial achievements. This is the so-called double bottom line, proposed by Yaron (1994), which deals with both financial and social aspects. There is still a controversy over how to best measure both the financial and social performance of MFIs (Hermes & Hudon, 2018). The evaluation of the financial performance of MFIs adopts measures and standards of conventional finance used by other financial institutions. These include analysis of accounting information, computing financial ratios, and efficiency measures which then are used to generate financial reports’ ratings (Gutiérrez-Nieto & Serrano-Cinca, 2007). On the one hand, MFIs are expected to be self-sustainable and avoid dependence on donations (Hudon & Traca, 2011). On the other hand, MFIs are expected to show that their financial sustainability does not come at the expense of deviating from their outreach mission by charging high-interest rates that penalize the poorest and, thus, worsen their social performance. Although the study of the 41

42  Handbook of microfinance, financial inclusion and development

relationship between financial and social performance is a fruitful line of research (Hermes & Hudon, 2018), there is no consensus in the literature on how the profit orientation of MFIs affects their social and financial outcomes (Blanco-Oliver & Irimia-Diéguez, 2019). One of the reasons for the lack of consensus on the relationship between financial performance and social performance is that the social performance of MFIs can be assessed in many ways. One approach is to evaluate outreach from the supply side, that is, by analyzing MFIs’ accounting information (Schreiner, 2002). Outreach can be measured in terms of depth, value to users, cost to users, breadth, length, and scope (Navajas et  al., 2000). The breadth of outreach is measured using indicators such as the number of active borrowers and the depth of outreach is measured using indicators such as the average loan size. The latter indicator is often used as a measure of mission drift (Copestake, 2007). Indicators such as the percentage of women served by the institution or the percentage of rural clients are also used. Besides, it is expected that MFIs duly serve other stakeholders, remunerate their employees fairly, pay taxes, and adopt appropriate governance practices (Hartarska, 2005; Labie, 2001; Mersland & Strøm, 2009). Another way of measuring MFIs’ social performance focuses on the quality of outreach and consists of impact evaluations that do not analyze the institution itself but its clients, i.e., analyses are carried out from the demand side (Copestake, 2007). Impact studies try to demonstrate whether microfinance improves the well-being of the poor (Hermes & Hudon, 2018), that is, whether microfinance intervention has brought about a specific and positive change in the lives of their clients (Duvendack, 2019). Balkenhol (2018) defined impact as changes in the situation of customers that can be causally attributed to access to micro-financial services. Unfortunately, measuring the impact of microfinance is not easy. Balkenhol (2018) stated that there is controversy in the selection of variables, measurement tools, dimensions, and levels of impact, as well as evaluation methodologies. There have been different reviews of the literature on microfinance assessment. Brau and Woller (2004) provided a comprehensive review of more than 350 papers, addressing the issues of MFI sustainability, management practices, clientele targeting, regulation, and impact assessment, among others. Hermes and Hudon (2018) systematically evaluated the potential of microfinance to reduce poverty from the supply side, by examining the performance of MFIs in reaching out to the poor by providing the services the poor need. The impact of microcredit was initially based on anecdotal evidence (Gaile & Foster, 1996; Goldberg, 2005; Sebstad & Chen, 1996; Odell, 2010), rather than careful analysis of the impact on poverty. Armendáriz and Morduch (2005) summarized the results from rigorous quantitative evidence on the nature, magnitude, and balance of microfinance impact. Duvendack et al. (2011) performed a meta-analysis of 58 studies, focusing on the technical challenges of conducting precise microfinance impact evaluations. Van Rooyen et al. (2012) performed a systematic review of the evidence of the financial and non-financial impacts of microfinance on poor people in sub-Saharan Africa (Stewart et  al., 2010). Duvendack et  al. (2014) conducted a new metaanalysis study, this time on the impact of microcredit on household decisions by women. Cull and Morduch (2017) performed an integrative literature review, highlighting the diversity in evidence on impacts and the role of subsidies. Duvendack and Mader (2020) performed a systematic review of reviews; they analyzed 32 meta-studies concluding that impacts are more likely to be positive than negative, but the effects vary. This chapter aims to analyze the research carried out by the scientific community on the assessment of microfinance with a scientometric approach. Previous literature provides

Assessment of microfinance institutions and their impact  43

general reviews of microcredit under systematic review approaches, which attempted to collect all empirical evidence to answer a specific research question (García-Pérez et al., 2017; Hermes & Hudon, 2018; Pinz & Helmig, 2015; Rasel & Win, 2020). Meta-analyses, which use statistical procedures to aggregate and combine the results of independent studies, were also performed (Fall et  al., 2018; Reichert, 2018). Finally, the scientometric approach has also been employed, but these studies had different goals (Gutiérrez-Nieto & Serrano-Cinca, 2019; Roy & Goswami, 2013). Roy and Goswami (2013) conducted a scientometric analysis of 71 research papers focusing on the overall performance of microfinance institutions. Their review was carried out along with different parameters: financial performance, social performance, outreach, sustainability, efficiency, productivity, institutional characteristics, and governance. Gutiérrez-Nieto and Serrano-Cinca (2019) studied the time evolution of two research traditions: papers focusing on clients (welfarists) and papers focusing on microfinance entities themselves (institutionalists). This chapter contributes to the literature by studying the assessment of MFIs and especially the impact of microcredit, analyzing bibliographic data following a scientometric approach, with a sample of 3,588 papers. This methodological approach enables obtaining microcredit impact knowledge maps and trends of microfinance impact research, which is the main contribution of the study. The remainder of the chapter is organized as follows: the second section is on microfinance financial performance. The third section presents some MFIs social assessment methodologies. The fourth section provides the scientometric analysis. The fifth section presents a discussion of the results along with the main conclusions. The chapter closes with the reference list.

FINANCIAL PERFORMANCE Although with multiple goals, MFIs are financial entities, so their financial performance indicators are standardized and universal. The analysis of their financial performance is based on accounting information used to compute financial ratios. A consortium of 28 public and private development agencies agreed on a set of guidelines on definitions of financial terms, ratios, and adjustments for microfinance to achieve a standardized method of calculating financial indicators in the microfinance sector (CGAP, 2003). These consensus indicators remain valid today and are grouped into four categories: portfolio quality, assets and liability management, profitability and sustainability, and efficiency and productivity. Rating agencies rate the financial performance of MFIs by adapting the financial methodologies offered by agencies such as Standard & Poor’s or Moody’s or the international rating system used by bank supervisory authorities (Gutiérrez-Nieto & Serrano-Cinca, 2007). The microfinance business consists of lending money and recovering it, so its portfolio quality is a key aspect. Despite the a priori reluctance to lend to the poor, who do not have collateral, Yunus applied in practice the proverb “the poor always pay back”, supported by credit methodologies such as solidarity groups or peer monitoring. These and other novel credit methodologies were developed to keep non-performing loans at levels similar to those of banks (Schreiner, 2000). However, Karim (2008) questioned the overly aggressive loan recovery programs implemented by some MFIs, whose loan officers often put excessive pressure on clients by shaming them or even destroying their properties. Two indicators commonly used in microfinance to measure portfolio quality are portfolio at risk, defined as the proportion of

44  Handbook of microfinance, financial inclusion and development

the loan portfolio in arrears for longer than 30 days, and the proportion of the loan portfolio that is written off and accounted as a loss for the MFI (Zamore et al., 2019). Asset and liability management is crucial in financial institutions, which have asset-liability committees (ALCO) to manage the financial structure of the institution’s balance sheet. In this aspect, the strategic management of liquidity risks and the interest rate spread are analyzed (Kusy & Ziemba, 1986). MFIs have some peculiarities when managing funds, as many of them do not collect deposits, but funds come from investors and donors. Both the term and the price of assets and liabilities need managing, using indicators such as (1) the asset coverage ratio that relates the entity’s assets to debt and (2) the net interest margin that measures the difference between the interest paid on deposits and interest earned on loans. The management of net interest margin is a delicate issue for MFIs because setting high-interest rates has an impact on poor clients, who pay more money for their loans than non-poor clients, which leads to a form of poverty penalty (Gutiérrez-Nieto et al., 2017). This is why good asset and liability management can lead to bad social management practices, meaning a low social performance position. In addition to the traditional profitability ratios, MFIs calculate some specific indicators, such as self-sustainability, measured with a ratio that relates revenues to costs, with adjustments made to account for implicit subsidies. Cull et al. (2018) analyzed 1,335 microfinance institutions to measure how many of them could actually continue operations without external donor funding and found that just over two-thirds of microfinance borrowers are served by MFIs not earning profits, suggesting that there are still substantial subsidies running through the sector. An important aspect of the evaluation of MFIs is their efficiency, which relates the inputs of MFIs (personnel expenses, operating expenses) to net income (Gutiérrez-Nieto et al., 2007). The efficiency ratio answers the question of how much the MFI needs to spend to obtain $100. Efficiency studies often handle several inputs and outputs, so multi-output approaches such as parametric frontiers and data enveloping analysis (DEA) are used. More recent studies reveal that the efficiency of MFIs has increased over time; however, the level of efficiency of the industry as a whole remains low and should be improved (Fall et al., 2018). The greatest source of inefficiency of MFIs is their high operating expenses (both personnel and administrative), which represent 62 percent of charges to borrowers, while financial expenses represent 23 percent, profits 10 percent, and losses from defaults 5 percent (González, 2007). MFIs focus on the least profitable clients in the financial system, those in the long tail of Pareto’s distribution of loans (Serrano-Cinca & Gutiérrez-Nieto, 2014). Operating in such a business environment, profits can only be made by raising the price of the product, i.e., by increasing the lending rate or by using technologies that dramatically reduce management costs, such as electronic banking or mobile applications.

SOCIAL PERFORMANCE Social performance is the effective translation of an institution’s social goals into practice in line with accepted social values (SEEP, 2006). These social values include sustainably serving increasing numbers of poor and excluded people, improving the quality and appropriateness of financial services, improving the economic and social conditions of clients, and ensuring social responsibility to clients, employees, and the community they serve (Hashemi, 2007). The measurement of the social performance of MFIs is more complex than the financial

Assessment of microfinance institutions and their impact  45

performance, in terms of the concepts, the diversity of methodologies applied, and the variables to be analyzed. One of the most widely accepted procedures for assessing social performance has four steps (World Bank, 2007). The first is to identify the social aim of the institution. The second is the evaluation of the institution’s internal systems and activities. The third analyzes outputs, assessing whether the institution serves the poor and whether the products meet their needs. The fourth involves the analysis of outcomes, i.e., whether clients have experienced improvements in their social and financial situation, culminating with an impact analysis, which attempts to establish causality between participation in the program and improvement in the conditions of clients. Several social performance assessment tools have been developed such as CERISE, SPA, Social Action, PPI, and FINCA (World Bank, 2007). Besides, there are rating agencies that perform social ratings such as M-CRIL, Microfinance, and Planet Rating. Given the diversity of methodologies, the Social Performance Task Force advocated the creation of a common reporting format for social reporting in microfinance that included organizational and client indicators, which was materialized in the Universal Standards for Social Performance Management (Wardle, 2017). Standardizing social indicators and evaluation procedures is a very necessary task, although in explicit situations it will be necessary to use specific evaluation frameworks and indicators (Sierra et al., 2019). Mission Drift and Governance The first step in social performance evaluation is to identify the social aim of the institution and verify the degree of compliance with it. Mission drift in microfinance arises when an MFI finds it profitable to reach out to unbanked wealthier individuals while at the same time crowding out poor clients (Armendáriz & Szafarz, 2011). Serrano-Cinca and Gutiérrez-Nieto (2014) argued that some MFIs have a tendency towards mission drift, by merely applying Pareto’s 80/20 Principle, which states that the least profitable customers are placed in the long tail of the wealth distribution function. They conducted an empirical study finding a pattern of a mission-centered MFI: a small NGO, with high labor productivity, receiving donations, and obtaining a high yield. Beisland et al. (2019) explored additional internal reasons for MFIs’ mission drift and suggest that changing credit officer behavior over time might explain why MFIs drift from social motivations toward financial motivations. Several indicators quantify whether the MFI meets its mission statement. If the aim of the MFI is poverty reduction, the number of poor served should be analyzed, and a common indicator of success is the average loan size because it is understood that large loans are not extended to the poorest. If the mission of the MFI is women’s empowerment, the percentage of women clients should be analyzed. If the mission is to improve rural financial inclusion, the percentage of rural clients or the percentage of agricultural loans are indicators measuring its performance. Several authors have studied the time evolution of mission drift in the MFI sector, finding a significant increase in the average loan size. This result, however, can also be a symptom of mission expansion, that is, the MFI also lends to additional clients without abandoning the poor (D’Espallier et al., 2017). Most studies indicate that mission drift is not a trend in the whole microfinance sector, although there are institutions that clearly deviated from their mission (Beisland et al., 2019). An assessment of the social performance of MFIs must include aspects related to their governance, which is closely related to the protection of the mission. Corporate governance involves a set of relationships between a company’s management, its board, its shareholders,

46  Handbook of microfinance, financial inclusion and development

and other stakeholders (OECD, 2004). To this end, several indicators have been developed that deal with board composition, employee compensation, donor relations, tax payments, relations with customers and suppliers, the actions that the MFI does to improve the environment, the contributions it makes to the community, and the transparency with which it carries out all its actions. The relationships between governance and performance, both financial and social, have also been studied. Hartarska (2005) found a significant positive relationship between board independence and performance, embracing the idea that independence of the microfinance board should be promoted. Outreach and Social Efficiency Outreach is defined as the degree to which an MFI provides financial services to a large number of people, especially the poorest. According to the last edition (10th) of the Microfinance Barometer (Convergences, 2019), an annual publication disclosing the main trends of the sector, on a global basis from MIX Market data, microfinance outreach has risen from 98 million borrowers in 2009 to 139.9 million borrowers in 2018, which is a 42.75% growth in ten years. The Barometer states that the growth of the sector, measured by its credit portfolio, is $124.1 billion in 2018 with an average annual growth rate of 11.5% in the previous five years. It seems clear that MFIs have succeeded in the financial inclusion of many people (Brown et al., 2015). It is expected that an MFI not only serves as many clients as possible but also does it efficiently, using few resources. Hence, social efficiency is another step in the study of outreach. Social efficiency considers the same inputs used to calculate financial efficiency, such as personnel expenses and other operating expenses, as well as separate social outputs such as the number of poor served, female clients, or the group on which the mission of the institution is focused (Gutiérrez-Nieto et al., 2009). Products Offered and Prices An analysis of social performance should include the review of financial products and services offered by the MFI, which should be designed to meet the needs of the poor. In addition to microcredit, microsavings, and microinsurance, non-financial services such as social services, business development services, and business training should also be considered. Lensink et al. (2018) found that the provision of social services was associated with improved loan quality and greater depth of outreach. Few doubt the role that technology has in promoting financial inclusion, and access through mobile electronic banking should be valued as an important aspect that affects MFIs service quality. Additionally, the loan methodology—individual lending, group lending, and village banking—is also important because certain methodologies transfer the credit risk to the client. Interest rates can be very high and lead to a poverty penalty, so a social performance assessment should ensure that the institution does everything possible to maintain fair interest rates (Gutiérrez-Nieto et al., 2017). For this purpose, the yield on the gross loan portfolio of the MFI is usually evaluated, but it would be more appropriate to evaluate the interest rates actually paid by the clients in that institution to establish comparisons with what they would pay for the same product in other institutions (Waterfield, 2015). Evaluation of Outcomes and Impact The analysis of MFIs outcomes aims to evaluate whether customers have experienced improvements in their social and financial situation. An impact evaluation is the last step of

Assessment of microfinance institutions and their impact  47

the social performance analysis aimed to establish a causal relationship between participation in the program and improvement in customers’ conditions. Establishing impact means demonstrating that the program causes the observed changes (Rossi et al., 1999), i.e., that changes are more likely to occur with participation in the program rather than without participation. Early microcredit studies avoided calculations of poverty impact, often treating the fact that small loans are being made as proof that the poor are being reached, and the fact that loans are being repaid as proof that incomes have increased (Mosley & Hulme, 1998). Some researchers even took the financial health of the entity as a proxy indicator of impact, arguing that the popularity of the services offered among people was sufficient to show that they were beneficial (Johnson & Rogaly, 1997). In fact, until the end of the 20th century, most microfinance impact assessment studies were qualitative commentaries and only relied on anecdotal evidence (Duvendack, 2019). For example, the description of a woman, who thanks to a microloan, was able to buy a sewing machine was considered as evidence that she was helped out of poverty. However, even if the sample was extended to include several cases, and even if a microcredit program improves the client’s income, this does not mean that the microfinance program is the cause of this better income since correlation does not mean causation. It is essential to properly design the methodology to ensure quality (Duvendack et al., 2011; Graham-Rowe et al., 2011). Studies in which pre-intervention measures were not recorded are unreliable. Take, for example, an MFI that grants microcredits in a village. Three years later, the MFI claims that the standard of living of its current clients is good enough but the MFI did not even measure the poverty level of its clients before joining the program. In the following stage, we can find studies that recorded pre-intervention measures but did not report data from a control or comparison group, making it difficult to assess the effectiveness of the interventions. This would be an example of a typical study in which an MFI affirms that its clients were poor, and they have been lifted out of poverty three years later, but we do not know if microcredit is the cause in the absence of a control group. Maybe the villagers who did not ask for microcredit reached a better level than microfinance clients did, perhaps microfinance clients would have done even better without the microcredit. The next stage in terms of quality is the cohort-analytic method, that is, an observational study where groups, which have or have not been exposed to intervention are compared, but where intervention exposure is not controlled by the researchers (Graham-Rowe et  al., 2011). However, these groups were not randomized or matched, which is a notable shortcoming. Quasi-experimental designs use matched but not randomized control groups. The selection of the groups is not randomized; the researcher divides the sample by a certain criterion. Then, self-selection by program participants may lead to an overestimation of the beneficial impact of microfinance. For example, it could be the case that applicants for microcredit are better skilled, and have more entrepreneurial spirit than those who choose not to apply for microcredit. Balkenhol (2018) noticed that participants joined microcredit programs for unobserved reasons and that those reasons might explain why their achievements were superior to non-participants. Randomized controlled trials (RCT) are experimental designs where the investigators randomly allocated individuals to an intervention or control group, overcoming the problem of quasi-experimental designs. Participants in both groups are identical in all but one aspect: their participation (or not) in the microfinance program (Coleman, 1999). Few papers used this methodology in the field of microfinance, although their number has grown in recent years

48  Handbook of microfinance, financial inclusion and development

(Banerjee et al., 2013; Banerjee et al., 2015a, 2015b, 2015c; Banerjee et al., 2019; De Mel et al., 2009; Karlan & Valdivia, 2011; Karlan & Zinman, 2011). Karlan and Valdivia (2011) analyzed the effect of business training on a micro entrepreneurs group, finding little or no evidence of changes in business revenue, income, employment, or subjective well-being; however, they observed business knowledge improvements and increased client retention rates for the MFI. Some impact studies have been useful to contrast the assumption that microcredit was not very useful for poverty alleviation (Banerjee et al., 2015c). Banerjee et al. (2015a) found positive financial impacts (investment and profits) but no significant changes in non-financial aspects (health, education, women empowerment). Banerjee et al. (2015b) studied the impact of a multifaceted program providing a productive asset grant, training, and support, life skills coaching, health information, temporary cash consumption support, and access to savings accounts, finding that it caused lasting progress for the very poor. However, Banerjee et al. (2019) found that the effects of access to formal credit through microfinance are highly heterogeneous, finding positive effects on household businesses and consumption for those individuals with an existing business acumen before their use of microfinance. Guidelines for conducting impact studies are provided by several initiatives—the International Initiative for Impact Evaluation, the Qualitative Impact Protocol, and the Regulatory Impact Assessment. RCT studies are the most appropriate from a methodological point of view, but they are not exempt from some problems. Let us compare them with medical studies, which profusely use RCT. First, the measurements must be unbiased. Whether or not a person develops cancer is an easily measurable variable. However, poverty has multiple dimensions: financial, economic, social, and environmental, and for each of them, there is a host of indicators (Balkenhol, 2018). We can know precisely the number of cigarettes smoked each day, but it is difficult to pinpoint accurately when we talk about the impact of microfinance because there are very different microfinance products. The type of loans or services offered matters; there are also different loan methods, and different kinds of microfinance institutions, i.e., the list of variables is daunting (Balkenhol, 2018). Odell (2010) stated that given the variety of microfinance tools and local markets it is impossible to answer, in a general way, the question of whether microfinance works or not. Similarly, Morduch (2020) recognized that RCTs are interesting and informative in their own terms and in their own idiosyncratic contexts but fail to answer the biggest questions about microcredit impact. Following on with the comparison to medical studies, if the RCT study is well done its ability to generalize results can be high. For example, if the study determines the relationship between smoking and lung cancer, the results could be extrapolated to almost all of humanity because people are more or less alike, although there may be immune collectives. In fact, when a drug is approved, its use spreads around the world. However, demonstrating that a microcredit experience was indisputably a success or a failure provides a valid result only in that context, being unsuitable to extrapolate the results. In other words, a result that is true in one place, at one time, and under one set of circumstances, will typically not be true in another place, another time, or under different circumstances (Deaton, 2020). Karlan and Zinman (2011) encouraged the expansion of the number of studies and carried them out themselves in different contexts, to the point where there was sufficient evidence of the impact (or not) of microcredit. Let us point to RCT studies that opened the way, but to demonstrate the direct causal relationship between tobacco and lung cancer, more than a single statistical study is necessary. Increased understanding of the molecular biology of tobacco-related cancers identified 60 carcinogens that are responsible for multiple genetic changes in lung tissue and the

Assessment of microfinance institutions and their impact  49

development of lung cancer (Hecht, 2002). In the case of microcredit, even if the statistical evidence becomes irrefutable, it will be necessary to develop a theory that explains the causes, which will hardly be equivalent to the experimental context of genetic science. Finally, these experimental studies can have an ethical problem: how to justify that an entity with a social mission randomly grants credits for methodological reasons. Hudon et al. (2019) pointed out that randomized studies focus on the methodological aspects and that ethical issues are left aside, suggesting that ethical committees employ a rule known as equipoise, used in medicine, which means that the patient consents to provide ex-ante equally attractive options to the treated and control groups.

SCIENTOMETRIC ANALYSIS OF MICROFINANCE RESEARCH ON THE ASSESSMENT OF INSTITUTIONAL PERFORMANCE AND IMPACT ON CLIENTS Web of Science (WoS) and Scopus are two world-leading abstract and citation databases. Scopus is increasingly used in academic papers (only a little less than the competitor WoS) and is challenging the dominating role of WoS (Zhu & Liu, 2020). In this section, we analyze all the papers published between 1995 and 2019 in both databases, performing a keywords co-occurrence analysis and a citation network analysis (Radhakrishnan et al., 2017). This type of analysis allows for identifying the main topics of interest for microfinance researchers and for studying the evolution of interest in these topics. The citation networks analysis enables obtaining cognitive maps (Small, 1973) that allow classifying the works, authors, and journals in clusters, and, by analyzing the clusters, drawing relevant conclusions about the microfinance state of the art. The study was carried out chronologically, which additionally allows the analysis of microfinance research trends, analyzing emerging topics by evolving the keywords used in academic articles. First, we opted for the WoS database. We searched by theme (which includes title, abstracts, and keywords) with the keywords “microfinance” and its synonyms. The following search was used: TS = (microfinance) OR TS = (“micro finance”) OR TS = (micro-finance) OR TS = (microcredit*) OR TS = (“micro credit*”) OR TS = (“micro-credit*”) OR TS = (microbank*) OR TS = (“micro bank*”) OR TS = (“micro-bank*”) OR TS = (microinsurance*) OR TS = (“micro insurance*”) OR TS = (“micro-insurance*”) OR TS = (microsaving*) OR TS = (“micro saving*”) OR TS = (“micro-saving*”)

This search returned 3,588 results. Afterward, we performed the same search on the Scopus database, which required a slight change in syntax: TITLE-ABS-KEY(“microfinance”) OR TITLE-ABS-KEY(“micro finance”) OR TITLE-ABSKEY(“micro-finance”) OR TITLE-ABS-KEY(“microcredit*”) OR TITLE-ABS-KEY(“micro credit*”) OR TITLE-ABS-KEY(“micro-credit*”) OR TITLE-ABS-KEY(“microbank*”) OR TITLE-ABS-KEY(“micro bank*”) OR TITLE-ABS-KEY(“micro-bank*”) OR TITLEABS-KEY(“microinsurance*”) OR TITLE-ABS-KEY(“micro insurance*”) OR TITLE-ABSKEY(“micro-insurance*”) OR TITLE-ABS-KEY(“microsaving*”) OR TITLE-ABS-KEY(“micro saving*”) OR TITLE-ABS-KEY(“micro-saving*”)

This search returned 4,371 results. The coverage of WoS and Scopus is different, but scientific studies that opt for one or the other often come to similar conclusions (Harzing & Alakangas, 2016). This is because the most cited articles are usually published in journals with the greatest

50  Handbook of microfinance, financial inclusion and development

impact and these journals are usually included in both databases. However, there may be exceptions. One of the exceptions is Schreiner (2002), a highly cited paper on the concept of microfinance outreach that was published in the Journal of International Development. This journal is part of Scopus, but not of WoS. It may even be the case that journals are not indexed in either database, such as the defunct Journal of Microfinance; we can find here a highly cited paper on credit scoring in microfinance from the same author: Schreiner (2000). One way to solve this problem is to select the papers indexed in any of the databases but examine the articles that cite those papers. This backward procedure ensures that none of the relevant articles are forgotten. Finally, it is also worth using Scholar Google, which contains by far the largest database of academic publications, although it does not have the quality filters of WoS and Scopus. From now on, we will use WoS. We removed conference proceedings and books from the search, focusing only on Science and Social Sciences Citation Indexes. The AND operator was subsequently added to find articles dealing with specific assessment topics, one at a time. The chosen topics were experimental designs using RCT, for which the keyword “randomized” was used. Other topics analyzed were: “sustainability”, “efficiency”, “outreach”, “governance”, and “mission drift”. Variants were identified for each keyword. Table 3.1 shows the number of articles published for each of the topics analyzed. There may be many publications on a topic but with a low impact, measured as citations per publication. The table shows the number of citations annually received for each topic. It is interesting to study the average number of citations received for each topic. A ratio was calculated by dividing the number of citations received each year by the cumulative sum of the articles on each of the topics. The table shows that in each of the most recent years more than 400 studies were written on microfinance, of which 20 apply RCT, less than the number of studies on governance or efficiency, but higher than topics such as mission drift. RCT is the topic whose articles receive the highest number of citations per year, about 5.45 in 2019, compared to the 1.95 that microfinance articles receive on average. In addition to a great deal of interest in the topic, two factors explain it. The first is that some of the articles are published in medical journals, which have higher impact factors than economic journals. For example, The Lancet has an impact factor of 59.1 in 2018 compared to 3.9 for World Development. Also, the impact of microcredit is a topic of interest not only to microfinance researchers but also to other scientists, so their studies receive citations from outside the microfinance field (Gutiérrez-Nieto & Serrano-Cinca, 2019). Table 3.2 shows the 15 most cited microfinance articles on impact studies using RCT. The number of citations received by them is displayed, ranked by the number of citations received in 2019. All these papers share the same school of thought, but to Banerjee-Duflo-Karlan we add co-authors such as Giné, Angelucci, Attanasio, Ashraf, and Crépon (Angelucci et al., 2015; Ashraf et al., 2010; Attanasio et al., 2015; Banerjee et al., 2015a, 2015b, 2015c; Crépon et al., 2015; Giné & Karlan, 2014; Giné & Yang, 2009; Karlan & Valdivia, 2011; Karlan & Zinman, 2009; Karlan & Zinman, 2011). Among them, the three most cited papers (Banerjee et al., 2015a, 2015b, 2015c) stand out as very recent because they were published in 2015. Winning the Nobel Prize may lead to an increase in citations because of the Matthew effect, which means that established researchers are extremely successful in obtaining citations (Merton, 1968). The name of the effect comes from the Bible’s Matthew 25:29, “For whoever has will be given more, and they will have an abundance”; in other words, the rich get richer and the poor get poorer. However, there is no Matthew effect because Banarjee and Duflo won in October

51

Randomized

Outreach

Governance

Mission drift

Efficiency

—

—

—

Citations

Ratio

—

Ratio

Papers

—

Citations

1

2

Ratio

Papers

—

2

Papers

—

—

Citations

Ratio

Citations

—

Papers

3

1

Citations

Ratio

0.15

—

Ratio

Papers

2

Citations

3

0.27

Ratio

Sustainability Papers

21

17

Papers

—

—

—

0.5

2

—

1

2

—

—

—

-—

0.6

3

2

0.63

12

6

0.63

62

35

—

—

—

1.25

5

—

0.5

1

—

—

—

—

0.33

2

1

0.5

11

3

0.58

71

23

—

—

—

0.4

4

6

0.25

1

2

—

—

—

0.14

1

1

0.88

29

11

0.74

129

53

—

—

—

0.18

2

1

0.29

2

3

—

—

—

0.25

2

1

0.53

21

7

0.55

119

40

—

—

—

1.42

17

1

1.88

15

1

—

—

—

1.44

13

1

1.21

51

2

0.93

236

38

—

—

1

1.2

18

3

3.38

27

—

—

—

—

0.67

6

—

1.47

66

3

1.11

333

48

6

12

1

1.11

21

4

3.1

31

2

0.5

1

2

1.08

13

3

1.14

65

12

1.08

403

73

4.1

41

8

1.47

28

—

1.93

27

4

0.25

1

2

1.2

18

3

0.9

63

13

1.11

549

120

5.77

75

3

1.59

46

10

2

42

7

0.33

2

2

1.38

29

6

1.43

123

16

1.32

812

121

170

2011 199

2012 212

2013 234

2014 371

2015 382

2016 436

2017

427

2018

5.59

123

9

2.06

74

7

2.36

66

7

1.5

12

2

1.52

41

6

1.58

177

26

1.61

4.39

136

9

2.04

94

10

1.81

65

8

1.44

13

1

1.74

66

11

1.51

205

24

1.47

4.56

164

5

3.13

169

8

2.26

104

10

1.89

17

—

2.24

103

8

2.1

329

21

1.65

6.76

250

1

2.92

193

12

1.95

109

10

5.7

57

1

2.98

182

15

2.33

403

16

1.79

6.35

305

11

2.49

214

20

2.71

179

10

2.15

28

3

2.76

243

27

2.45

503

32

1.87

2.53

5.62

388

5.65

435

8

3.26 21

372

28

2.45

235

14

2.45

54

7

2.56

405

33

2.31

741

52

1.76

388

33

2.82

231

16

4.4

66

2

2.7

338

37

2.39

643

64

1.96

6.17

605

21

2.83

537

43

2.27

304

38

2.33

84

14

2.73

565

49

2.57

1,014

74

2.02

84

1.95

7,066

414

2019

2.32

2.57

6.03

699

5.45

736

19

3.46 18

743

47

1.98

380

27

2.7

154

15

2.61

792

48

837

52

2.65

437

31

3.19

134

6

3.3

844

49

2.71

1,271 1,284

74

2.08

1,266 1,405 1,906 2,447 2,995 3,870 4,153 5,622 6,701

171

Microfinance

Citations

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Keyword

Table 3.1  The number of articles published for each of the topics analyzed, as well as the number of citations and a ratio showing the average number of citations received annually for each topic

52

—

—

—

—

—

—

—

Attanasio et al. (2015)

Karlan & Zinman (2011)

Karlan & Valdivia (2011)

Ashraf et al. (2010)

Pronyk et al. (2008)

Giné & Karlan (2014)

—

Giné & Yang (2009)

Crépon et al. (2015)

—

Kim et al. (2007)

—

12

Pronyk et al. (2006)

—

—

Banerjee et al. (2015b)

Karlan & Zinman (2009)

—

Banerjee et al. (2015c)

Angelucci et al. (2015)

—

Banerjee et al. (2015a)

2007

—

3

—

—

—

—

—

—

—

—

7

28

—

—

—

2008

—

6

—

—

—

—

—

—

—

—

11

39

—

—

—

2009

—

16

—

—

—

—

—

5

—

7

13

40

—

—

—

2010

—

22

1

3

2

—

—

8

—

11

15

37

—

—

—

2011

Table 3.2  The 15 most cited articles on RCT in microfinance

—

10

4

6

12

—

—

12

—

8

14

34

—

—

—

2012

—

12

7

16

12

—

—

22

—

17

21

47

—

—

—

2013

1

16

7

33

13

—

—

18

—

14

27

49

—

—

—

2014

8

12

6

30

13

10

11

22

7

12

29

44

5

6

13

2015

8

13

15

20

17

8

8

13

8

21

27

42

23

15

25

2016

14

12

9

19

15

13

15

19

17

21

32

42

29

35

35

2017

9

16

21

36

13

15

16

20

14

16

30

48

38

36

44

2018

15

16

16

16

19

19

20

20

25

25

25

35

45

46

55

2019

55

154

86

179

116

65

70

159

71

152

251

497

140

138

172

Total

Assessment of microfinance institutions and their impact  53

2019 so the award cannot be the driving force behind the citations. Finally, the articles that investigated the health effects of microfinance through RCT are relevant contributions (Kim et al., 2007; Pronyk et al., 2006, 2008). We then generated knowledge maps using the Vosviewer software, which aggregates references and illustrates the progress of a research field over time (Van Eck & Waltman, 2017). We took the 135 microfinance studies conducted using RCT, that is, which used the keyword “randomized” and its synonyms. Figure 3.1 shows the knowledge map. These maps are similar to road maps in which two nearby cities appear close to each other. Anyone can simply use a ruler to measure the distances between several cities to obtain a distance chart. Performing the opposite process, that is, creating a map from a distance chart is complicated but fortunately, there are several mathematical algorithms capable of doing so. For example, a multidimensional scaling algorithm converts information about the pairwise distances among a set of n objects into a configuration of n points mapped into a Euclidean space, usually a plane (Torgerson, 1952). The measure of distance (similarity) is a unit of length in the case of road maps. In the scientometrics area, the measure of distance can be obtained from a co-citation analysis to obtain the knowledge maps of a discipline. A citation link is a link between two items where one item cites the other. By contrast, a co-citation link is a link between two items that are both cited by the same document. If two documents are co-cited by a third article, they can be similar, so that the more co-cites they share, the greater the affinity between them, and the closer they appear on the knowledge map. In the figures generated by the software, the volume of the sphere is proportional to the number of citations, which were normalized to correct for the fact that older documents have had more time to receive citations than more recent documents. Figure 3.1 visualizes two clusters, which are clearly not bibliographically coupled. The authors of the cluster on the right carry out studies that analyze the impact of microcredit on health (Kim et  al., 2007; Pitt & Khandker, 1998; Pronyk et  al., 2006; Sherman et  al., 2010; Ssewamala et  al., 2010; Weiser et  al., 2015). The cluster on the left contains studies on the impact of microcredit. In that part, Karlan and Valdivia (2011) and Banerjee et  al. (2015a) stand out, compared to the other articles (Angelucci et al., 2015; Attanasio et al., 2015;

Figure 3.1 Knowledge maps obtained from co-citation analysis of microfinance publications using RCT. Unit of analysis: authors

54  Handbook of microfinance, financial inclusion and development

Banerjee et al., 2015b; Crépon et al., 2015; De Mel et al., 2011; Field & Pande, 2008; Giné & Yang, 2009; Tarozzi et al., 2015). Figure 3.2 illustrates the origins of the authors of the citations. The absolute predominance of the USA is noteworthy, and a very small cluster with some other countries, mostly European. Figure 3.3 illustrates the authors’ universities and research centers. Three clusters can be seen; the one on the left is the hard core with authors Banerjee and Duflo (MIT), Karlan (formerly at Yale, and now at Northwestern), and Zinman (Dartmouth College). The presence of universities outside the US orbit is very limited. Figure 3.4 presents the results of the citation analysis applied to journals. Again, two clusters appear that clearly distinguish the publications on the health area (at the right) from those on economics (at the left). In the latter case, the American Economic Journal: Applied Economics, which published a special issue in January 2015, American Economic Review, World Development, and Journal of Development Economics stand out. Figure 3.5 shows the microfinance governance knowledge map, with a selection of the ten most cited papers (Barry & Tacneng, 2014; Beisland et al., 2014; Galema et al., 2012; Hartarska, 2005; Hartarska & Mersland, 2012; Mersland, 2009; Mersland & Strøm, 2009; Périlleux et al., 2012; Servin et al., 2012; Strøm et al., 2014). In this case, the bibliography is grouped around several seminal papers. The cluster at the bottom right contains papers related to ownership and leadership; the papers in the cluster at the bottom left analyze external governance mechanisms, while the papers in the cluster at the top are focused on outreach and clients. Figure 3.6 shows the microfinance efficiency knowledge map created from the ten most cited papers (Caudill et al., 2009; D’Espallier et al., 2017; Gutiérrez-Nieto et al., 2007; Hartarska et al., 2013; Hermes et al., 2011; Hermes & Hudon, 2018; Hudon & Traca, 2011; Louis et al., 2013; Servin et  al., 2012). The cluster at the bottom left represents social efficiency papers, the

Figure 3.2 Knowledge maps obtained from a citation analysis of microfinance publications using RCT. Unit of analysis: countries

Assessment of microfinance institutions and their impact  55

Figure 3.3 Knowledge maps obtained from a citation analysis of microfinance publications using RCT. Unit of analysis: organizations

Figure 3.4 Knowledge maps obtained from a citation analysis of microfinance publications using RCT. Unit of analysis: journals cluster at the top includes papers focused on inputs and costs and the papers in the cluster at the bottom right deal with the funding sources of MFIs, with a special emphasis on subsidies. Figure 3.7 shows the microfinance mission drift knowledge map with the ten most cited papers on this topic (Ault, 2016; Chahine & Tannir, 2010; Copestake, 2007; D’Espallier et al., 2017; Kar & Swain, 2014; Mersland & Strøm, 2010; Mia & Lee, 2017; Serrano-Cinca & Gutiérrez-Nieto, 2014; Serrano-Cinca et al., 2016; Vanroose & D’Espallier, 2013). The papers in the cluster at the right contain mission drift models, whereas the papers in the cluster at the left are focused on the role of transformation and commercialization of MFIs. Knowledge maps can also be obtained from the similarities among keywords in order to visualize groups of a network of co-occurring terms. The number of co-occurrences of two keywords is the number of publications where both keywords appear together in the title, abstract, or keyword list (Van Eck & Waltman, 2017). If two keywords are included in many publications, their co-occurrence is high and, hence, they will appear close together on the knowledge map. The size of the circle in the map is proportional to the number of

56  Handbook of microfinance, financial inclusion and development

Figure 3.5 Knowledge maps obtained from a citation analysis of microfinance publications on governance. Unit of analysis: documents

Figure 3.6 Knowledge maps obtained from a citation analysis of microfinance publications on efficiency. Unit of analysis: documents publications that have the corresponding term in their title or abstract or keywords. Three maps were generated. Figure 3.8 shows the map created from all articles on microfinance published until 2007, the year when the boom in microfinance research began after Yunus won the Nobel Prize. The keywords reflect the topics of interest to academics. Two main groups are appreciated. The first cluster, at the upper right corner, focuses on the study of microfinance institutions, the lending methodologies they use, and the impact of these

Assessment of microfinance institutions and their impact  57

Figure 3.7 Knowledge maps obtained from a citation analysis of microfinance publications on mission drift. Unit of analysis: documents

Figure 3.8 Visualizing a network of co-occurring terms from microfinance publications. Unit of analysis: documents. Years: 1997–2007 programs on poverty alleviation. The second cluster, at the bottom left corner, focuses on research on microcredit and its clients, especially rural credit programs and women empowerment. In this cluster, the most used keyword is “Bangladesh”. Therefore, the division between welfarism and institutionalism is evident between clusters. It is noteworthy that “impact” was the only word related to assessment until 2007. Concepts such as sustainability, outreach efficiency, or governance were incipient and did not appear as keywords until later.

58  Handbook of microfinance, financial inclusion and development

Figure 3.9 Visualizing a network of co-occurring terms from microfinance publications. Unit of analysis: documents. Years: 2008–2018 Figure 3.9 shows the map created using the articles published from 2008 to 2018. Each of the two previous clusters was split into two groups. The most striking issue is that in the block that studies microfinance institutions, the interest in assessing MFIs can be clearly seen, with

Figure 3.10 Visualizing a network of co-occurring terms from microfinance publications. Unit of analysis: documents. Years: 2019–2020

Assessment of microfinance institutions and their impact  59

keywords such as performance, sustainability, outreach, efficiency, or governance. Also, in the welfarist cluster, an important research topic emerges in this period: microfinance and health. Figure 3.10 shows the map created using only the articles published in 2019 and 2020. There is continued interest in the assessment of microfinance institutions as well as studies evaluating the impact of microcredit on poverty alleviation. This third map serves to detect the most recent trends in the sector. “Financial inclusion” (the new mantra in the industry) is emerging as a widely used keyword.

DISCUSSION AND CONCLUSIONS The dual nature of MFIs, social and financial, is a source of conflicts that are difficult to solve. MFIs aim to “kill several birds with one stone”: achieve financial self-sustainability and increase the income and employment of beneficiaries and reduce poverty (Hulme & Mosley, 1996). Depending on how financial performance and social performance are defined, a positive, negative, or no relationship between them can be found. It is sometimes difficult to find entities with outstanding behavior in several aspects because it is logically impossible to maximize in more than one dimension at the same time unless the dimensions are monotone transformations of one another (Jensen, 2001). For example, MFIs cannot simultaneously maximize the financial margin (to get the maximum benefit) and minimize the financial margin (to avoid the poverty penalty). There is a great deal of ongoing debate in the literature about the trade-off between the two main dimensions of microfinance performance, and which one should prevail (Reichert, 2018). Assessment of microfinance social performance should monitor whether the institution has improved in self-sustainability by deviating from its mission, charging very high-interest rates, or both. It should also investigate loan recovery practices, and whether efficiency gains are achieved through improvements in technology and reduction of redundant costs or by having employees underpaid and overworked. A complete evaluation of microfinance performance would require a cost-benefit or a costeffectiveness analysis to compare the social value with social cost in general equilibrium (Navajas et al., 2000). A key aspect of the performance of MFIs is their impact on clients. It is a process of determining whether microcredit helps the poor to be less poor, or only adds debt as one more problem to their existing problems. Duvendack et al. (2011) performed a meta-analysis on the impact of microcredit, reviewing 58 studies, and found that almost all impact evaluations of microfinance suffered from weak methodologies and inadequate data, concluding that it remains unclear under what circumstances, and for whom, microfinance has been and could be of real, rather than imagined, benefit to poor people. Roodman and Morduch (2014) claimed that we have little solid evidence that microfinance improves the lives of its clients in a measurable way. Many of the studies on the impact of microcredit suffered from methodological challenges until RCT was used (Banerjee et  al., 2015a, 2015c, 2019; Karlan & Valdivia, 2011; Karlan & Zinman, 2011). These studies found that microfinance programs did not have the development effects that many had claimed when these programs were introduced on a large scale (Committee for the Prize in Economic Sciences in Memory of Alfred Nobel, 2019); although some programs showed some positive effects (Banerjee et al., 2019). Yet even RCTs are not exempt from dangers and drawbacks underscoring the challenges to study the impact of microfinance (Deaton, 2020; Morduch, 2020).

60  Handbook of microfinance, financial inclusion and development

In short, in spite of the proliferation of research over the past 30 years, the literature is far from offering a unified conclusion about the impact of microfinance (Duvendack, 2019). The challenge is to find the characteristics of MFIs and the external circumstances under which microcredit can have a positive impact on clients and to identify the conditions under which microcredit works best (Gutiérrez-Nieto & Serrano-Cinca, 2019). Nevertheless, a final comment should be made about the number of studies and their length. The Surgeon General’s report by the US Department of Health and Human Services (2014) reviewed hundreds of articles conducted with RCT before presenting its findings on the health consequences of smoking. By contrast, only about 20 studies a year are carried out on the impact of microcredit in all its forms, most of them by the same group of researchers. One of the studies that determined the relationship between smoking and lung cancer lasted 50 years, from 1951 to 2001 (Doll et al., 2004). Without the material possibility of going to such extremes, the fact is that studies on the impact of microcredit have a short duration while it would be advisable to capture the possible long-term effects. With the recognition by the Nobel Committee of the RCT use in economics, it is expected that the number of studies that use RCT will grow. The hegemony of impact studies through RCT in the USA is striking, compared to the rest of the world, even though the net official development assistance by European donors doubles that provided by the USA. We encourage other researchers from other research centers to develop more impact measurements using RCT.

ACKNOWLEDGMENTS This work was supported by the Spanish Ministry of Science, Innovation, and Universities and European Regional Development Fund (ERDF) A way of making Europe, RTI2018093483-B-I00, and PID2019-107822RB-I00, by the Spanish Ministry of Economy and Competitiveness and European Regional Development Fund (ERDF) A way of making Europe, ECO2016-74920-C2-1-R, and by the Government of Aragon [Ref. S38_17R].

REFERENCES Angelucci, M., Karlan, D., & Zinman, J. (2015). Microcredit impacts: Evidence from a randomized microcredit program placement experiment by Compartamos Banco. American Economic Journal: Applied Economics, 7(1), 151–182. Armendáriz, B., & Morduch, J. (2005). The economics of microfinance (1st ed.). Cambridge, MA: MIT Press. Armendáriz, B., & Szafarz, A. (2011). On mission drift in microfinance institutions. In Armendáriz, B., & Labie, M. eds, The handbook of microfinance. Hackensack, NJ: World Scientific Publishing. Ashraf, N., Karlan, D., & Yin, W. (2010). Female empowerment: Impact of a commitment savings product in the Philippines. World Development, 38(3), 333–344. Attanasio, O., Augsburg, B., De Haas, R., Fitzsimons, E., & Harmgart, H. (2015). The impacts of microfinance: Evidence from joint-liability lending in Mongolia. American Economic Journal: Applied Economics, 7(1), 90–122. Ault, J. K. (2016). An institutional perspective on the social outcome of entrepreneurship: Commercial microfinance and inclusive markets. Journal of International Business Studies, 47(8), 951–967. Balkenhol, B. (2018). Microfinance: Research, Debates, Policy. Abingdon, UK: Routledge. Banerjee, A., Breza, E., Duflo, E., & Kinnan, C. (2019). Can microfinance unlock a poverty trap for some entrepreneurs? National Bureau of Economic Research. NBER working paper No. w26346.

Assessment of microfinance institutions and their impact  61

Banerjee, A., Chandrasekhar, A. G., Duflo, E., & Jackson, M. O. (2013). The diffusion of microfinance. Science, 341(6144), 1236498. Banerjee, A., Duflo, E., Glennerster, R., & Kinnan, C. (2015a). The miracle of microfinance? Evidence from a randomized evaluation. American Economic Journal: Applied Economics, 7(1), 22–53. Banerjee, A., Duflo, E., Goldberg, N., Karlan, D., Osei, R., Parienté, W., … Udry, C. (2015b). A multifaceted program causes lasting progress for the very poor: Evidence from six countries. Science, 348(6236), 1260799. Banerjee, A., Karlan, D., & Zinman, J. (2015c). Six randomized evaluations of microcredit: Introduction and further steps. American Economic Journal: Applied Economics, 7(1), 1–21. Barry, T. A., & Tacneng, R. (2014). The impact of governance and institutional quality on MFI outreach and financial performance in Sub-Saharan Africa. World Development, 58, 1–20. Beisland, L. A., D’Espallier, B., & Mersland, R. (2019). The commercialization of the microfinance industry: Is there a ‘personal mission drift’ among credit officers? Journal of Business Ethics, 158(1), 119–134. Beisland, L. A., Mersland, R., & Randøy, T. (2014). The association between microfinance rating scores and corporate governance: A global survey. International Review of Financial Analysis, 35, 268–280. Blanco-Oliver, A., & Irimia-Diéguez, A. (2019). Impact of outreach on financial performance of microfinance institutions: A moderated mediation model of productivity, loan portfolio quality, and profit status. Review of Managerial Science, doi:10.1007/s11846-019-00353-4. Brau, J. C., & Woller, G. M. (2004). Microfinance: A comprehensive review of the existing literature and an outline for future research. The Journal of Entrepreneurial Finance, 9(1), 1–27. Brown, M., Guin, B., & Kirschenmann, K. (2015). Microfinance banks and financial inclusion. Review of Finance, 20(3), 907–946. Caudill, S. B., Gropper, D. M., & Hartarska, V. (2009). Which microfinance institutions are becoming more cost effective with time? Evidence from a mixture model. Journal of Money, Credit and Banking, 41(4), 651–672. CGAP (2003). Microfinance consensus guidelines. Definitions of selected financial terms, ratios and adjustments for microfinance, 3rd edition. Consultative Group to Assist the Poorest: Washington, DC, USA. Chahine, S., & Tannir, L. (2010). On the social and financial effects of the transformation of microfinance NGOs. Voluntas: International Journal of Voluntary and Nonprofit Organizations, 21(3), 440–461. Coleman, B. E. (1999). The impact of group lending in Northeast Thailand. Journal of Development Economics, 60(1), 105–141. Committee for the Prize in Economic Sciences in Memory of Alfred Nobel (2019). Scientific background on the Sveriges Riksbank prize in economic sciences in memory of Alfred Nobel 2019, Understanding Development and Poverty Alleviation, October 14, 2019. Convergences (2019). Microfinance Barometer 2019. http://www​.convergences​.org​/wp​-content​/uploads​/ 2019​/09​/ Microfinance​-Barometer​-2019​_web​-1​.pdf [retrieved 22.11.2019]. Copestake, J. (2007). Mainstreaming microfinance: Social performance management or mission drift? World Development, 35(10), 1721–1738. Crépon, B., Devoto, F., Duflo, E., & Parienté, W. (2015). Estimating the impact of microcredit on those who take it up: Evidence from a randomized experiment in Morocco. American Economic Journal: Applied Economics, 7(1), 123–150. Cull, R., Demirgüç-Kunt, A., & Morduch, J. (2018). The microfinance business model: Enduring subsidy and modest profit. The World Bank Economic Review, 32(2), 221–244. Cull, R., & Morduch, J. (2017). Microfinance and economic development. Handbook of finance and development. Cheltenham: Edward Elgar Publishing. D’Espallier, B., Goedecke, J., Hudon, M., & Mersland, R. (2017). From NGOs to banks: Does institutional transformation alter the business model of microfinance institutions? World Development, 89, 19–33. De Mel, S., McKenzie, D. J., & Woodruff, C. (2009). Measuring microenterprise profits: Must we ask how the sausage is made? Journal of Development Economics, 88(1), 19–31. De Mel, S., McKenzie, D. J., & Woodruff, C. (2011). Getting credit to high return microentrepreneurs: The results of an information intervention. The World Bank Economic Review, 25(3), 456–485.

62  Handbook of microfinance, financial inclusion and development

Deaton, A. (2020). Randomization in the tropics revisited: A theme and eleven variations. In Bédécarrats, F., Guérin, I., & Roubaud, F. eds, Randomized controlled trials in the field of development: A critical perspective. Oxford: Oxford University Press. Doll, R., Peto, R., Boreham, J., & Sutherland, I. (2004). Mortality in relation to smoking: 50 years’ observations on male British doctors. British Medical Journal, 328(7455), 1519–1528. Duvendack, M. (2019). What do we really know about microfinance impact? In Hudon, M., Labie, M., & Szafarz, A. eds, A Research Agenda for Financial Inclusion and Microfinance. Cheltenham, UK: Edward Elgar Publishing. Duvendack, M., & Mader, P. (2020). Impact of financial inclusion in low- and middle-income countries: A systematic review of reviews. Journal of Economic Surveys, 34(3), 594–629, doi:10.1111/joes.12367. Duvendack, M., Palmer-Jones, R., Copestake, J. G., Hooper, L., Loke, Y., & Rao, N. (2011). What is the evidence of the impact of microfinance on the well-being of poor people? London: EPPI-Centre, Social Science Research Unit, Institute of Education, University of London. Duvendack, M. Palmer-Jones R., & Jos Vaessen (2014). Meta-analysis of the impact of microcredit on women’s control over household decisions: Methodological issues and substantive findings. Journal of Development Effectiveness, 6(2), 73–96. Fall, F., Akim, A. M., & Wassongma, H. (2018). DEA and SFA research on the efficiency of microfinance institutions: A meta-analysis. World Development, 107, 176–188. Field, E., & Pande, R. (2008). Repayment frequency and default in microfinance: Evidence from India. Journal of the European Economic Association, 6(2–3), 501–509. Gaile, G. L., & Foster, J. (1996). Review of methodological approaches to the study of the impact of microenterprise credit programmes. Report submitted to USAID assessing the impact of microenterprise Services (AIMS), June. Galema, R., Lensink, R., & Mersland, R. (2012). Do powerful CEOs determine microfinance performance? Journal of Management Studies, 49(4), 718–742. García-Pérez, I., Muñoz-Torres, M. J., & Fernández-Izquierdo, M. Á. (2017). Microfinance literature: A sustainability level perspective survey. Journal of Cleaner Production, 142(4), 3382–3395. Giné, X., & Karlan, D. S. (2014). Group versus individual liability: Short and long term evidence from Philippine microcredit lending groups. Journal of Development Economics, 107, 65–83. Giné, X., & Yang, D. (2009). Insurance, credit, and technology adoption: Field experimental evidence from Malawi. Journal of Development Economics, 89(1), 1–11. Goldberg, N. (2005). Measuring the impact of microfinance: Taking stock of what we know. Grameen Foundation USA Publication Series, December. González, A. (2007). Efficiency drivers of microfinance institutions (MFIs): The case of operating costs. Microbanking Bulletin, 15, 37–42. Graham-Rowe, E., Skippon, S., Gardner, B., & Abraham, C. (2011). Can we reduce car use and, if so, how? A review of available evidence. Transportation Research Part A: Policy and Practice, 45(5), 401–418. Gutiérrez-Nieto, B., & Serrano-Cinca, C. (2007). Factors explaining the rating of microfinance institutions. Nonprofit & Voluntary Sector Quarterly, 36(3), 439–464. Gutiérrez-Nieto, B., & Serrano-Cinca, C. (2019). 20 years of research in microfinance: An information management approach. International Journal of Information Management, 47C, 183–197. Gutiérrez-Nieto, B., Serrano-Cinca, C., Cuéllar-Fernández, B., & Fuertes-Callén, Y. (2017). The poverty penalty and microcredit. Social Indicators Research, 133(2), 455–475. Gutiérrez-Nieto, B., Serrano-Cinca, C., & Mar-Molinero, C. (2007). Microfinance institutions and efficiency. Omega, 35(2), 131–142. Gutiérrez-Nieto, B., Serrano-Cinca, C., & Mar-Molinero, C. (2009): Social efficiency in microfinance institutions. Journal of the Operational Research Society, 60(1), 104–119. Hartarska, V. (2005). Governance and performance of microfinance institutions in Central and Eastern Europe and the newly independent states. World Development, 33(10), 1627–1643. Hartarska, V., & Mersland, R. (2012). Which governance mechanisms promote efficiency in reaching poor clients? Evidence from rated microfinance institutions. European Financial Management, 18(2), 218–239. Hartarska, V., Shen, X., & Mersland, R. (2013). Scale economies and input price elasticities in microfinance institutions. Journal of Banking & Finance, 37(1), 118–131.

Assessment of microfinance institutions and their impact  63

Harzing, A. W., & Alakangas, S. (2016). Google Scholar, Scopus and the Web of Science: a longitudinal and cross-disciplinary comparison. Scientometrics, 106(2), 787–804. Hashemi, S. M. (2007). Beyond good intentions: Measuring the social performance of microfinance institutions. Consultative Group to Assist the Poor (CGAP), Focus Note, 41. Washington, DC. Hecht, S. S. (2002). Cigarette smoking and lung cancer: chemical mechanisms and approaches to prevention. The Lancet Oncology, 3(8), 461–469. Hermes, N., & Hudon, M. (2018). Determinants of the performance of microfinance institutions: A systematic review. Journal of Economic Surveys, 32(5), 1483–1513. Hermes, N., Lensink, R., & Meesters, A. (2011). Outreach and efficiency of microfinance institutions. World Development, 39(6), 938–948. Hudon, M., Labie, M., & Szafarz, A. (2019). The microfinance alphabet. In O’Connor, M., & Silva Afonso J. eds, Emerging Challenges and Innovations in Microfinance and Financial Inclusion. London: Palgrave Macmillan. Hudon, M., & Traca, D. (2011). On the efficiency effects of subsidies in microfinance: An empirical inquiry. World Development, 39(6), 966–973. Hulme, D., & Mosley, P. (1996). Finance against poverty. London: Routledge. Jensen, M. C. (2001). Value maximization, stakeholder theory, and the corporate objective function. Journal of Applied Corporate Finance, 14(3), 8–21. Johnson, S., & Rogaly, B. (1997). Microfinance and poverty reduction. Oxford: Oxfam (UK and Ireland). Kar, A. K., & Swain, R. B. (2014). Interest rates and financial performance of microfinance institutions: Recent global evidence. The European Journal of Development Research, 26(1), 87–106. Karim, L. (2008). Demystifying micro-credit: The Grameen Bank, NGOs, and neoliberalism in Bangladesh. Cultural Dynamics, 20(1), 5–29. Karlan, D., & Valdivia, M. (2011). Teaching entrepreneurship: Impact of business training on microfinance clients and institutions. Review of Economics and Statistics, 93(2), 510–527. Karlan, D., & Zinman, J. (2009). Expanding credit access: Using randomized supply decisions to estimate the impacts. The Review of Financial Studies, 23(1), 433–464. Karlan, D., & Zinman, J. (2011). Microcredit in theory and practice: Using randomized credit scoring for impact evaluation. Science, 332(6035), 1278–1284. Kim, J. C., Watts, C. H., Hargreaves, J. R., Ndhlovu, L. X., Phetla, G., Morison, L. A., … Pronyk, P. (2007). Understanding the impact of a microfinance-based intervention on women’s empowerment and the reduction of intimate partner violence in South Africa. American Journal of Public Health, 97(10), 1794–1802. Kusy, M. I., & Ziemba, W. T. (1986). A bank asset and liability management model. Operations Research, 34(3), 356–376. Labie, M. (2001). Corporate governance in microfinance organizations: Along and winding road. Management Decision, 39(4), 296–302. Lensink, R., Mersland, R., Vu, N. T. H., & Zamore, S. (2018). Do microfinance institutions benefit from integrating financial and nonfinancial services? Applied Economics, 50(21), 2386–2401. Louis, P., Seret, A., & Baesens, B. (2013). Financial efficiency and social impact of microfinance institutions using self-organizing maps. World Development, 46, 197–210. Mersland, R. (2009). The cost of ownership in microfinance organizations. World Development, 37(2), 469–478. Mersland, R., & Strøm, R. Ø. (2009). Performance and governance in microfinance institutions. Journal of Banking & Finance, 33(4), 662–669. Mersland, R., & Strøm, R. Ø. (2010). Microfinance mission drift? World Development, 38(1), 28–36. Merton, R. K. (1968). The Matthew effect in science: The reward and communication systems of science are considered. Science, 159(3810), 56–63. Mia, M. A., & Lee, H. A. (2017). Mission drift and ethical crisis in microfinance institutions: What matters? Journal of Cleaner Production, 164, 102–114. Morduch, J. (2020). Why RCTs failed to answer the biggest questions about microcredit impact, World Development, 127, 104818, doi:10.1016/j.worlddev.2019.104818. Mosley, P., & Hulme, D. (1998). Microenterprise finance: Is there a conflict between growth and poverty alleviation? World Development, 26(5), 783–790.

64  Handbook of microfinance, financial inclusion and development

Navajas, S., Schreiner, M., Meyer, R. L., Gonzalez-Vega, C., & Rodriguez-Meza, J. (2000). Microcredit and the Poorest of the Poor: Theory and Evidence from Bolivia. World Development, 28(2), 333–346. Odell, K. (2010). Measuring the impact of microfinance: Taking another look. Washington, DC: Grameen Foundation. OECD (2004). OECD principles of corporate governance, organisation for economic co-operation and development (OECD). Paris. Périlleux, A., Hudon, M., & Bloy, E. (2012). Surplus distribution in microfinance: Differences among cooperative, nonprofit, and shareholder forms of ownership. Nonprofit and Voluntary Sector Quarterly, 41(3), 386–404. Pinz, A., & Helmig, B. (2015). Success factors of microfinance institutions: State of the art and research agenda. Voluntas: International Journal of Voluntary and Nonprofit Organizations, 26(2), 488–509. Pitt, M. M., & Khandker, S. R. (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), 958–996. Pronyk, P. M., Hargreaves, J. R., Kim, J. C., Morison, L. A., Phetla, G., Watts, C., … Porter, J. D. (2006). Effect of a structural intervention for the prevention of intimate-partner violence and HIV in rural South Africa: a cluster randomised trial. The Lancet, 368(9551), 1973–1983. Pronyk, P. M., Kim, J. C., Abramsky, T., Phetla, G., Hargreaves, J. R., Morison, L. A., … Porter, J. D. (2008). A combined microfinance and training intervention can reduce HIV risk behaviour in young female participants. Aids, 22(13), 1659–1665. Radhakrishnan, S., Erbis, S., Isaacs, J. A., & Kamarthi, S. (2017). Novel keyword co-occurrence network-based methods to foster systematic reviews of scientific literature. PloS one, 12(3), e0172778. Rasel, M. A. & Win, S. (2020). Microfinance governance: A systematic review and future research directions. Journal of Economic Studies, 47(7), 1811–1847, doi:10.1108/JES-03-2019-0109. Reichert, P. (2018). A meta-analysis examining the nature of trade-offs in microfinance. Oxford Development Studies, 46(3), 430–452. Roodman, D., & Morduch, J. (2014). The impact of microcredit on the poor in Bangladesh: Revisiting the evidence. Journal of Development Studies, 50(4), 583–604. Rossi, P. H., Lipsey, M. W., & Henry, G. T. (1999). Evaluation: A systematic approach. Thousand Oaks, CA: Sage Publications. Roy, A., & Goswami, C. (2013). A scientometric analysis of literature on performance assessment of microfinance institutions (1995-2010). International Journal of Commerce and Management, 23(2), 148–174. Schreiner, M. (2000). Credit scoring for microfinance: Can it work? Journal of Microfinance/ESR Review, 2(2), 6. Schreiner, M. (2002). Aspects of outreach: A framework for discussion of the social benefits of microfinance. Journal of International Development, 14(5), 591–603. Sebstad, J., & Chen, G. (1996). Overview of studies on the impact of microenterprise credit. Report submitted to USAID assessing the impact of microenterprise services (AIMS), June. SEEP (2006). Social performance map. The SEEP network social performance working group, small enterprise education and promotion network (SEEP). Washington, DC. Serrano-Cinca, C., & Gutiérrez-Nieto, B. (2014). Microfinance, the long tail and mission drift. International Business Review, 23(1), 181–194. Serrano-Cinca, C., Gutiérrez-Nieto, B., & Reyes, N. M. (2016). A social and environmental approach to microfinance credit scoring. Journal of Cleaner Production, 112, 3504–3513. Servin, R., Lensink, R., & Van den Berg, M. (2012). Ownership and technical efficiency of microfinance institutions: Empirical evidence from Latin America. Journal of Banking & Finance, 36(7), 2136–2144. Sherman, S. G., Srikrishnan, A. K., Rivett, K. A., Liu, S. H., Solomon, S., & Celentano, D. D. (2010). Acceptability of a microenterprise intervention among female sex workers in Chennai, India. AIDS and Behavior, 14(3), 649–657. Sierra, J., Muriel-Patino, V., & Rodríguez-López, F. (2019). The quest for the evaluation of microfinance social performance. Iberoamerican Journal of Development Studies, 9(2). doi:10.26754/ojs_ried/ ijds.351. Small, H. (1973). Co‑citation in the scientific literature: A new measure of the relationship between two documents. Journal of the Association for Information Science and Technology, 24(4), 265–269.

Assessment of microfinance institutions and their impact  65

Ssewamala, F. M., Ismayilova, L., McKay, M., Sperber, E., Bannon Jr, W., & Alicea, S. (2010). Gender and the effects of an economic empowerment program on attitudes toward sexual risk-taking among AIDS-orphaned adolescent youth in Uganda. Journal of Adolescent Health, 46(4), 372–378. Stewart, R., van Rooyen, C., Dickson, K., Majoro, M., & De Wet, T. (2010). What is the impact of microfinance on poor people? A systematic review of evidence from sub-Saharan Africa. Technical Report, EPPI-Centre, Social Science Research Unit, University of London. Strøm, R. Ø., D’Espallier, B., & Mersland, R. (2014). Female leadership, performance, and governance in microfinance institutions. Journal of Banking & Finance, 42, 60–75. Tarozzi, A., Desai, J., & Johnson, K. (2015). The impacts of microcredit: Evidence from Ethiopia. American Economic Journal: Applied Economics, 7(1), 54–89. Torgerson, W. S. (1952). Multidimensional scaling: I. Theory and method. Psychometrika, 17(4), 401–419. US Department of Health and Human Services (2014). The health consequences of smoking—50 years of progress: A report of the Surgeon General. Atlanta, GA. Van Eck, N. J., & Waltman, L. (2017). Citation-based clustering of publications using CitNetExplorer and VOSviewer. Scientometrics, 111(2), 1053–1070. Van Rooyen, C., Stewart, R., & De Wet, T. (2012). The impact of microfinance in sub-Saharan Africa: A systematic review of the evidence. World Development, 40(11), 2249–2262. Vanroose, A., & D’Espallier, B. (2013). Do microfinance institutions accomplish their mission? Evidence from the relationship between traditional financial sector development and microfinance institutions’ outreach and performance. Applied Economics, 45(15), 1965–1982. Wardle, L. (2017). “The Universal Standards for Social Performance Management. Implementation Guide.” Social Performance Task Force. Online: https://sptf​.info​/images​/A2​_USSPM​_Impl​_Guide​ _English​_20171004​.pdf [retrieved 13.12.2019]. Waterfield, C. (2015). Advocating transparent pricing in microfinance: A review of MFTransparency’s work and a proposed future path for the industry. MicroFinance Transparency. Weiser, S. D., Bukusi, E. A., Steinfeld, R. L., Frongillo, E. A., Weke, E., Dworkin, S. L., … Cohen, C. R. (2015). Shamba maisha: Randomized controlled trial of an agricultural and finance intervention to improve HIV health outcomes in Kenya. AIDS, 29(14), 1889. World Bank (2007). Beyond good intentions: measuring the social performance of microfinance institutions (English). CGAP focus note; no. 41. Washington, DC: World Bank. http://documents​. worldbank​.org​/curated ​/en ​/220581468142504387​/ Beyond​-good​-intentions​-measuring​-the​-social​performance​-of​-microfinance​-institutions. Yaron, J. (1994). What makes rural finance institutions successful? The World Bank Research Observer, 9(1), 49–70. Zamore, S., Beisland, L. A., & Mersland, R. (2019). Geographic diversification and credit risk in microfinance. Journal of Banking & Finance, 109, 105665. Zhu, J., & Liu, W. (2020). A tale of two databases: The use of Web of Science and Scopus in academic papers. Scientometrics, 123, 321–335.

4. Financial inclusion and gender Isabelle Guérin

Out of all the controversies over financial inclusion, the issue of gender has probably been the most intensely debated. The very history of financial inclusion, which is intrinsically linked to the history of microcredit, is inseparable from the dimension of gender. “We are here because of women,” to quote Sam Daley-Harris, founder of the first Microcredit Summit, at the opening of the 2006 Summit in Halifax, Canada. When in 1997 the Microcredit Summit Campaign began to attract investors and disseminate “good practices,” women and the poor were a priority target. Very soon, however, discordant voices were heard, arguing that microcredit exploits women more than it emancipates them and reporting situations of violence, humiliation, and misappropriation of women’s microcredits by men (Goetz & Gupta 1996; Rahman 1999). The purpose of this chapter is to revisit this controversy. Several meta-analyses have converged to produce mixed, and at best modest, results (Garikipati et al. 2017). However, the promises of financial inclusion continue to be (unfairly) promoted by some, and digital finance has even been giving a fresh boost to this optimistic discourse. The criticism has also persisted, notably from activist movements, researchers, and more recently, multilateral international organizations under the umbrella of the United Nations (UN 2020; UNCTAD 2019). Dialogue between these opponents and critics is not easy, as it is based on diverse and conflicting visions of development and empowerment: macro versus micro, public goods versus private goods, structure versus behavior, power versus agency, short-term versus long-term, the individual versus the collective, and so on. Our attempt here will be to support dialogue between these diverging positions. Criticisms must be taken seriously. Financial services alone cannot do much to address the structural problems of gender inequality. Moreover, excessive optimism is not only simplistic but dangerous, especially in view of the massive and growing over-indebtedness of the populations of the Global South, with which its women, as household budget managers, largely have to cope. Equally, in today’s financialized world, whether we like it or not, everyone needs financial services for at least two fundamental reasons. First, when properly set up, financial services help people to manage their liquidity constraints and payments and give them a safe place to put their savings or transfer them when they have excess liquidity, allowing them to access additional liquidity through credit when they need it. Second, informal sources of financing, however widespread, almost never meet all the financial needs of a given population. Yet it is clear that the current financial services supply suffers from diverse weaknesses, some of which may do more harm than good to customers, especially women. Improving the quality of services must remain a major priority. The very idea of financial inclusion therefore deserves neither to be overlooked nor forgotten. It is a highly diverse financial tool with highly varied effects. The question is not whether financial inclusion is “good” or “bad,” “works” or “does not work,” but is a matter of exploring the conditions that make it useful for its users, or conversely which factors make financial inclusion do more harm than good. In addition to this, “women” as a category is so diverse that there is little to be gained by looking to list universal statements, to reiterate a conclusion 66

Financial inclusion and gender  67

reached almost two decades ago (Johnson 2005; Kabeer 2005) and which bears repeating (Garikipati et al. 2017). In this chapter, I combine my own work with secondary data. My own work is mostly based on teamwork, drawing on various surveys and ethnographies conducted in South India since the early 2000s, but also on shorter field visits in Senegal, Morocco, and the Dominican Republic. What the studies cited in this chapter have in common is their focus on processes rather than “difference-making” analyses (Shaffer 2015). Quantitative studies that are decontextualized from their environment and limited to measuring changes in indicators are illequipped to capture the complexity and the direction of causality chains. This is particularly the case for randomized studies, which will therefore not be mentioned here.1

THE CONTROVERSY In the history of institutional credit to the poor (which is much older than microcredit), the female clientele has most often been overlooked. Targeting women is a key specificity of contemporary financial inclusion policies, albeit with regional variations. According to the latest available microcredit data, the proportion of female microfinance clients is 25.8 percent in the Middle East and North Africa, 53.8 percent in Sub-Saharan Africa, 61.6 percent in the Caribbean, and 81 percent in Asia Pacific (Cull & Morduch 2017: 9). Bank account ownership has increased significantly over the past decade, with 515 million adults worldwide opening an account at a financial institution or through a mobile money provider between 2014 and 2017 (Demirgüç-Kunt et al. 2018: 2). However the gender gap persists: 72 percent of men have an account, as opposed to 65 percent of women, and this gap has remained unchanged since 2011. In countries with a high penetration of mobile money, mobile money is less discriminating and may help close the gender gap in the future (Demirgüç-Kunt et al. 2018: 25–6). Digital payments have made considerable progress: roughly half of the adult population is now affected (as sender or as receiver) with a gender gap of five percentage points that has remained unchanged since 2014 (Demirgüç-Kunt et al. 2018: 57). As discussed below, these numbers say nothing about the actual use of these services, let alone their effects on users. But they do give an idea of the rapidly changing financial landscape and the persistent gender gaps. While the rhetoric of associating microcredit with women’s empowerment has probably grown less naïve in tone, it still persists. Following on from the broken promises of microcredit, notably since the various over-indebtedness crises, digital finance has been raising new hopes. This holds true for the pioneers of financial inclusion like Muhammad Yunus, but also more recent players, observers, and thinkers from the financial inclusion field.2 When the World Bank’s latest Findex report was published in April 2018, the press release stated that “Globally, 69 percent of adults—3.8 billion people—now have an account at a bank or mobile money provider, a crucial step in escaping poverty” (our emphasis).3 The press release quotes several personalities from the development world who affirm the very positive effects of financial services in terms of poverty alleviation, and for women in particular.4 None of these claims have been solidly demonstrated. Moreover, the promotion of financial inclusion must take into account a growing problem in the Global South: household overindebtedness. Of course, many promoters of financial inclusion consider that access to a bank account can provide access to cheap credit. But this remains highly controversial, and a key question is: does financial inclusion alleviate or contribute to over-indebtedness (Guérin, Labie, et al. 2015)?

68  Handbook of microfinance, financial inclusion and development

According to UNCTAD analyses, the global stock of debt, compared to GDP, has returned to its pre-global financial crisis level (260 percent in 2017, as opposed to 240 percent at the onset of the global financial crisis and 140 percent in 1980). This growth has largely been driven by the rise in private debt, which has increased more than 12-fold since 1980 and accounted for two-thirds of global debt in 2017. Developing countries, where private debt has long been confined to relatively low levels, are following the same trend: as a share of global GDP, private debt rose from 79 percent in 2008 to 139 percent in 2017 and is largely driven by non-financial corporations. “The unprecedented explosion of private debt should clearly raise the loudest alarm bells” (UNCTAD 2019: 74–5). A central question, of course, is what this private debt is being used for. Yet a comparison with productive investment shows that in many developing countries, between 2008 and 2015, private debt grew significantly faster than investment (UNCTAD 2019: 83). This confirms a more general trend, specific to contemporary financial systems, which are more focused on financing household social reproduction needs such as food, health or education (Lapavitsas 2011). Note too that the available data often underestimate actual household debt, as they do not (or poorly) take informal debt into account. Yet in many contexts, informal debt remains high. The rise in private household debt is the outcome of several factors (Servet & Saiag 2013; UN 2020). Low and irregular real incomes are one such factor, given the persistence of informal and vulnerable employment (ILO 2018). Access to basic services such as water, electricity, housing, and healthcare is becoming increasingly commodified. Demand for private education is also increasing, including among poor households seeking a better future for their children. The depletion of certain natural resources and of hitherto free common goods (drying up of groundwater, the disappearance of forest areas or restricted access to them, etc.) is causing a substantial loss of earnings for some segments of the population. The adoption of urban lifestyles has led to a considerable increase in the supply of new consumer goods, both related to identity and functional in nature. This is the case with mobile phones, but also with household appliances, computers, internet access, and motor vehicles (motorcycles or cars). Social transfer policies have not compensated for this gap. In some contexts, such as in Brazil or Argentina, social transfers are used as collateral and viewed by the population as a dangerous means of getting into debt (Lavinas 2017; Saiag 2020). Is financial inclusion part of the solution or part of the problem? Some critics have blamed microcredit for over-indebtedness (Bateman 2010). This criticism may be valid in some contexts, but not everywhere. Microcredit is only a small part of private household debt, which includes various forms of consumption credit and credit cards provided by financial companies and supermarkets, informal credit, and sometimes mortgages. Furthermore, the key issue leading to over-indebtedness is the discrepancy between income and expenses, a gap that microcredit may simply fill. Depending on the terms and conditions and the context of microcredit, it can reduce risk by offering better terms than other sources of borrowing. However, microcredit can also exacerbate over-indebtedness through aggressive marketing policies that induce people to take on more debt and accept repayment rigidities that are illsuited to irregular incomes. A good borrower repayment history does not imply an absence of over-indebtedness. In many places, borrowers make various sacrifices (Schicks 2014) such as selling assets, migrating or sending one of their siblings off to migrate to nearby regions or abroad (Bylander 2015; Morvant-Roux 2013), enduring various forms of social costs (Guérin, Morvant-Roux, et al. 2013), or even engaging in transactional sex (Guérin & Kumar 2020).

Financial inclusion and gender  69

All these coping strategies mask latent crises and allow saturated markets to hold on (Guérin, Labie, et al. 2015). As the UNCTAD and UN reports point out, massive indebtedness is unsustainable and a major driver of increasing inequalities. Debates on financial inclusion need to take into account the reality of growing household debt. Moreover, as the following section discusses, women are often the first to be affected.

FINANCIAL INCLUSION: CREATING WEALTH OR MANAGING SCARCITY? Historically, the ability to save, borrow, and circulate money has played an important role in women’s daily lives. In pre-industrial Europe, whether widows or brides, aristocrats or commoners, housewives, artisans, or small traders, the ability to access various forms of informal borrowing and lending was not a guarantee of emancipation, but a tool for coping in a world where economic opportunities were restricted and the vagaries of daily life were legion (Fontaine 2008; Lemire et al. 2001). This lesson from history is still valid today: financial services are mainly used for day-to-day management and much less for investment. For the reasons mentioned above, in the present world, the need for financial services appears more relevant than ever. In some cases, financial services may be used to create, sustain or improve a small business. However, it is now a well-known fact that even very small entrepreneurship experiences multiple challenges and is restricted to a few. The most common financial needs are intended to make ends meet and to maintain social relations (which also helps to ensure the material protection of people, as we shall see below). This includes food, healthcare, housing improvement, debt repayment, education, ceremonies, and sometimes durable consumer goods.5 Even when they are not “financially included,” women have no choice but to deploy multiple strategies of saving, borrowing, lending, gift-giving, and sharing. Although this is true for both men and women (Collins et al. 2009), women often bear the responsibility for making ends meet on a daily basis (Guérin 2011; Johnson 2004; Kusimba 2018). Regardless of gendered financial inclusion policies, household budget management in poor and workingclass families is primarily a woman’s responsibility. This seems to be a historical constant (Bruce & Dwyer 1988; Guérin 2011; Johnson 2004). With the sharp rise in private household debt, managing household budgets is becoming more and more akin to managing debt within households. Quantitative evidence is scarce since most debt data don’t disaggregate gender patterns within households. Qualitative evidence, however, indicates that survival debt is primarily female, for two reasons that seem to cut across space and time. The first one relates to a gendered division of roles: within poor households, managing debt is one of the domestic tasks usually assigned to women. The second reason reflects the gendered codes of honor and shame. Debt management is left to or imposed upon women as survival debt is a strong threat to masculinity and the breadwinner male model. Publicly admitting to being unable to make ends meet, personally facing debt collectors, lenders, or bailiffs, and handling rumors and mockery all seem to fall on women, shielding male honor, as has been observed in various parts of the world.6 Faced with this permanent liquidity constraint, any new opportunity is welcome: being eligible for new financial services is as such taken as an additional opportunity to better smooth over income and expenses over time (Collins et al. 2009). As such, microfinance should first

70  Handbook of microfinance, financial inclusion and development

and foremost be considered as “liquidity services” (Cull & Morduch 2017: 32). This potential role in smoothing over income and consumption has been a key finding from financial diaries (Collins et al. 2009). By offering women access to a wider range of services, financial inclusion has had a mixed impact. On the one hand, it potentially gives women some room for maneuver. On the other hand, it may reinforce the gendered division of responsibilities and further reinforce their role as household budget managers. “It’s marked on our foreheads,” Tamil women say when asked about their heavy debt burdens. Thus microcredit, entirely targeted at women in this region, exacerbates their responsibilities (Guérin & Kumar 2020). In urban West Bengal, “getting a loan [is] now constructed as another job for a married woman to do” (Kar 2018: 137). “It is easy for women to ask,” Kenyan women say of their multiple financial responsibilities and their higher propensity to go into debt. Yet digital finance further reinforces gendered norms that making ends meet is a female responsibility (Kusimba 2018: 258).

DIVERSE CONTEXTS, DIVERSE CONSTRAINTS, DIVERSE NEEDS As suggested in a recent collection of articles, financial inclusion “can hardly be expected to have one single, consistent impact story over the long assortment of product variations and geographical differences” (Garikipati et al. 2017: 642). As feminist research has shown time and again, gender norms vary widely across regions and social groups. This is particularly true with regard to women’s access to money, finance, and markets. Until recently, and probably still today in specific places, women’s access to markets can be severely restricted. In some Hindu and Islamic communities, until a few decades ago, it was said that a business run by a woman was an “abomination”; women refrained not only from selling but also from buying (Boserup 1970; Papanek & Schwede 1988). In Upper Egypt in the 1980s, it was said that women should not touch or talk about money, “even covered with gold” (Hoodfar 1988: 130). In some parts of Morocco in the same period, it was “an affront to her dignity and to God” for a woman to sell or buy (Maher 1981: 124). By contrast, in other parts of the world (in many countries in Africa, especially sub-Saharan Africa, South Asia, and Latin America), women have long been involved in trading and buying. This great diversity of norms and practices obviously shapes the opportunities and constraints that women face and their use of financial services. In highly patriarchal contexts, women’s entrepreneurship is often considered something extremely challenging. The poorest are often freer to move around, but may lack the necessary contacts to locate customers and suppliers and to navigate administration (Guérin, D’Espallier, et  al. 2015; Kalpana 2016). Access to regular paid employment makes a huge difference. In the textile industrial region of Tiruppur, Tamil Nadu, wage employment is widespread among not only men but also some women. This regular source of income allows women to take on debt for household needs without taking additional risks (Carswell et  al. 2020). The lack of control over the means of production is also a major barrier (Garikipati 2008). In contexts where women have long been engaged in markets, it is easier for them to use financial services to consolidate (rather than start) an income-generating activity. This has been observed, for example, in Paraguay (Schuster 2015), Kenya (Johnson 2004), Malawi (Johnson 2005), Cameroon (Mayoux 2001), and Senegal (Guérin 2006). However, market congestion and saturation, confinement within low capital-intensive and low-paying sectors, and the burden of domestic obligations often prevent women from expanding their activities, including with new financial services. Some

Financial inclusion and gender  71

women entrepreneurs have nothing to envy men, as evidenced by the success of the “mamabenz” in Togo, the Ghanaian or Yoruba cloth wholesalers, the Wolof gold traders linked to the pilgrimage networks in Mecca, or the dealers in contraband diamonds from Sierra Leone and Zaire. On average, however, women have smaller, less profitable, and more precarious businesses (WIEGO & ILO 2014).

DIVERSE MODELS The diversity of models of financial inclusion equally matters in shaping outcomes. The quality of financial services is a first and central issue. By this, I mean how services meet local needs at an acceptable cost. Although the issue may seem trivial, it is far from settled. The rigidity of services and their inability to adapt to fluctuating and unpredictable revenues remains a major challenge. Added to this are time and mobility constraints, which are often greater for women. Many microfinance providers continue to require time-consuming procedures, demanding that women travel, attend many meetings, and perform time-consuming, stressful relational and emotional work (Kar 2018; Schuster 2015). Access to individual loans, which are on the rise, can also be time-consuming since it often demands finding guarantors (Angulo Salazar 2013). In rural areas and in agriculture, as is well-known, credit terms and conditions remain poorly matched. With male migration, women are often in charge of family farming. Here too, needs can be diverse (inputs, equipment, livestock, cash flow to finance the lean season, etc.), and in turn relate to the diversity of agriculture models (cash or food crops, agriculture in dry or rainfed areas, intensive or extensive, family-based or professional, independent or contractual through integration into agro-business sectors or producers’ cooperatives, etc.). The targeting of women (although varying by region, as seen above) does not prevent discrimination. The fact that women represent a large share of the clientele does not necessarily mean that they are as well-served as men (Garikipati et al. 2017: 644). A study evaluating a Brazilian MFI shows that “all things equal,” women who apply for larger loans are discriminated against, and the authors conclude that there is a “glass-ceiling on loan size” (Agier & Szafarz 2013). This glass ceiling has also been observed in the French context (Cozarenco & Szafarz 2018) but not in Uganda (Corsi & De Angelis 2017). Discrimination against women is thus not systematic, but should be a cause for concern. Due to the pervasiveness of stereotypes, women are too often considered as a single category, contenting themselves with small amounts, and few providers treat women entrepreneurs properly. Associated complementary services can also be important, but this all depends on their quality. Like any form of education, training (whether in entrepreneurship, financial education, leadership, and so forth) can certainly be useful, but only under certain conditions. All too often, women attend them out of obligation or fear of not receiving their microcredit, while considering them as a waste of time (and therefore an additional cost), because the content is at odds with their own constraints. This is the case with entrepreneurship training courses that teach how to make a business model, keep accounts, and be innovative, without taking into account the multiple—and structural—barriers that women face. This is also the case with financial education training. Programs teach the virtues of saving while ignoring the fact that women already save in multiple ways. They recommend against multiple loans, but ignore the social role of some loans, or simply that women have no choice (Guérin 2012). With regards to over-indebtedness, some training modules focus on “honesty, thrift and the virtue

72  Handbook of microfinance, financial inclusion and development

of debt repayment,” and serve to secure repayment rather than to improve women’s well-being (Bylander & Res 2020). When the objective of practitioners or their funders is indeed to promote women’s empowerment, it seems that the most promising complementary services are those that enable women to improve their position in both the productive and reproductive spheres (Holvoet 2005; Kabeer 2016). This includes for instance the pooling of means of production or sales channels, intensive professional training, negotiating contracts with providers and buyers, and affordable and sustainable childcare and healthcare.

MISSION DRIFTS: WHO TAKES THE RISKS? It has been argued that targeting women meets equity and efficiency objectives. This “winwin” claim has been made since the beginning of microcredit promotion campaigns and remains relevant today (Armendariz & Morduch 2005). With the growing commercialization of the sector, the efficiency argument has undoubtedly taken over. Many organizations are content to target women, and few make real efforts to adapt their services to the specificities of female clientele, whether in terms of types of services, hours of operation, or staff training (Mayoux 2010). Many financial providers are convinced that targeting women is good in itself, both for women and their families. Simply checking indicators, such as the proportion of women clients, loan volumes, and repayment rates, does not vouchsafe positive effects for women. A strong demand for credit by women is sometimes misleading and can mask pressure from husbands or in-laws. For example, men may lack access to credit, or are unwilling to attend meetings and discipline themselves to access it. When loan collection practices drift toward coercion, women are often on the front line (D’Espallier et  al. 2011). Contrary to the official history of the Grameen Bank, targeting women was not an initial choice: it emerged as a necessity in the face of men’s poor repayment records (Todd 1996: 159–60). MIX Market data from the late 2000s has shown that women repay better, on average (D’Espallier et al. 2011). Women’s low physical mobility and sensitivity to social pressure explain why they repay better, as observed in a very wide range of contexts.7 While they claim to fight against gender inequalities, some financial providers rely precisely on patriarchal norms to optimize repayments. This drift is rarely a patriarchal conspiracy, but the outcome of a gradual shift, of which MFIs’ leaders, maybe out of touch with ground realities, are little or unaware. The main problem has been excessive targets for portfolio volumes and repayment rates. The rhetoric of “win-win” and “double bottom-line” on which the growth and success of microfinance have been built is in fact very problematic. It is now well-known that the commercial/social winwin vision has not been borne out (Cull & Morduch 2017: 12–15). Constrained by excessive objectives in terms of portfolio and repayment quality, loan officers may tend to turn to a more captive and disciplined clientele, and therefore a female clientele.

THE DIVERSE AND CONTRADICTORY MEANINGS OF “EMPOWERMENT” In very patriarchal contexts, the stated goal of empowerment and self-affirmation in regards to men and the local community may not meet expected goals. Many, although not all, women operate much more within the logic of avoidance, resistance, and compromise toward men,

Financial inclusion and gender  73

while also being involved in ambiguous intra-feminine relations that combine solidarity, rivalry, and competition. The point here is not to argue in favor of the status quo or the idea that women’s perceptions and aspirations should be the sole criterion for emancipation. The poor, the oppressed, the dominated, and women in particular, often tend to adapt to their condition and may have limited aspirations, even reproducing their own domination. But taking account of local perceptions and norms is crucial to capturing resistance to change, the varied ways through which women use the services provided to them, and the sometimes perverse effects of external interventions that are out of touch with local realities. Empowerment as a process is something contingent, difficult to predict, with sometimes conflicting results, and which may take unexpected forms. As far as financial inclusion is concerned, financial services alone cannot truly empower women. As Naila Kaber rightly observed in her groundbreaking study of Bangladesh in the 1990s, access to financial services can at best provide greater space and bargaining power, within a structural framework that remains unchanged (Kabeer 2001). One of the notable effects observed by Naila Kabeer (2001) relates to dignity. In an increasingly monetarized context, women greatly appreciate having access to their own sources of cash without begging for help from others. They also appreciate being better able to honor the rules of hospitality and to financially contribute to family ceremonies. For those who have managed to start their own business, it is often at the cost of hard work, but it is much more rewarding, they say, than being cooped up or working as a maid. Women also attach great importance to the dignity of their own husbands—allowing them to rent land rather than work for others is considered a success in itself. Naila Kabeer also points out ambiguous results: in a context where control over women and their bodies is greater among the upper classes, those who are economically successful tend to conform more to patriarchal norms (for instance by confining themselves to domestic space) as a mark of their social status. Further ethnographic examples highlight the diversity of processes and their ambivalent and at times conflicting qualities. In rural Tamil Nadu, the few women who have succeeded at setting up businesses have had a very turbulent experience, and many have failed due to lack of profitability and opportunities, but also because of overly conflictual relationships within their families, with two recurring areas of discord: their reduced availability for household chores and their transgression of femininity norms such as discretion, dependence, and modesty. The few rural women who have managed to set up a business are sometimes referred to as “prostitutes,” referring to the frequent relationships they are forced to establish with business partners and the administration (Guérin, Kumar, et al. 2013). An ethnography conducted in Andhra Pradesh has observed similar contradictions (Still 2015). Also in Tamil Nadu, a study based on mixed methods comparing the different sources of borrowing available to women, and their impact in terms of decision-making power, has highlighted other forms of ambivalent effects (Garikipati et al. 2017). The authors show that “instant” loans (informal loans, loans from relatives, pawnbrokers, street lenders) enhance women’s bargaining power in various types of household financial decisions. However, “planned loans,” including microcredit, which involve larger amounts but have more rigid terms and conditions, have no impact. This surprising result is best understood when exploring the nature of instant loans and the social hierarchy of debt. Instant loans are easy to access, but involve coercive repayment methods and are considered socially degrading:

74  Handbook of microfinance, financial inclusion and development

By accessing instant loans and by bearing the brunt of the social humiliation associated with such loans, women perform a convenient service for their households. The greater role that these women have in household financial decisions can then be explained as a concession by their families in return. (Garikipati et al. 2017: 719)

In Malawi, female microfinance clients pride themselves on being able to better cover some family expenses, but they are also stressed and anxious—their businesses have low profitability and weekly repayments are a real headache. They also face male suspicion because of their limited availability, although men frequently disengage from their own obligations (Johnson 2005). The diversity of contexts, methods, and modes of appropriation of microcredit by women certainly explains the contrasting effects in terms of domestic violence: some studies conclude that microcredit contributes to the reduction of domestic violence (Schuler et al. 1996), but others find the opposite (Rahman 1999). The issue of sexual violence has been little studied but deserves attention. Here too we can assume contrasting effects, even within the same context. For example, in Tamil Nadu, access to microcredit allows some women to avoid transactional sex—i.e. sexual exchange for material gain—with informal moneylenders. However, for those who struggle to repay their microcredit, transactional sex is sometimes the only option (Guérin & Kumar 2020). As far as digital finance is concerned, a series of studies conducted with the support of the Institute for Money, Technology and Financial Inclusion has concluded that it is impossible to predict a priori its impact on gender (Rea & Nelms 2017: 17). Sometimes women are excluded: men see mobile phones as a threat since they may prevent them from controlling women’s privacy. When women access mobile money, this can be very useful to enable them to receive transfers and make payments, to avoid them having to travel, sometimes saving considerable time. But it can also reinforce their obligations as mothers and wives. Some even prefer not to use a mobile phone to avoid being overly solicited by their entourage (Rea & Nelms 2017: 14).

WHO CAN BE EMPOWERED? Power can be perceived either as a relationship of domination or as an ability to think and act (agency). The concept of empowerment, as it is used in the development and financial inclusion world, stresses the issue of agency while all too often forgetting the power component. However, in certain spheres of action, such as gender relations, change is often a zero-sum game. What one gains is lost to others. Some studies have observed a monopolization of the benefits of microcredit by a few women. And in many cases, it is those who already have economic, social, and political resources who best manage to appropriate the effects of financial inclusion services. This includes for instance wives of civil servants, local notables or businessmen, women who are themselves already businesswomen, professional lenders, brokers of all kinds who are part of dominant political, community, or religious networks, first wives in polygamous contexts, etc.8 This raises the question of the causality between financial inclusion and empowerment: having some sort of pre-existing power seems to be a prerequisite to be able to use financial services usefully. It is also worth noting that from women’s perspectives, optimizing the use of financial services can take many forms. It can mean improving an existing business

Financial inclusion and gender  75

(rather than creating a new one) or becoming a local leader. It can also be about investing in ceremonial expenses (Buggenhagen 2011), trying to achieve middle-class respectability, e.g. by investing in the employment of men (Kabeer 2001) or the education of children (Kar 2018: 135), or becoming a professional informal lender (Karim 2011; Perry 2002) or a gatekeeper in accessing various resources (Guérin & Kumar 2017). The empowerment of some women may also result in the disempowerment of other women. Gains of any kind involve some form of control over other women within the household and neighborhood.9 The role of male spouses can also be decisive. To exclude men, as many financial organizations (particularly in Asia) practice, may not be a solution. While such exclusion is not the only cause, it has certainly helped reinforce certain perverse effects that have been observed in various contexts: the misappropriation of loans by men, who also need financing and have little understanding as to why they are excluded (Goetz & Gupta 1996; Montgomery 1996), the reproduction of stereotypes of “men as unreliable, drunk, and likely to waste money” (Kar 2018: 135), and the misappropriation of businesses when they become profitable (Garikipati 2008), which results in increased responsibilities, overwork, stress, and fatigue.10

NEGOTIATING INTERDEPENDENCIES OVER AUTONOMY AND SELF-RELIANCE Although many financial inclusion supporters and researchers argue in favor of autonomy and self-reliance as an indicator of success, the evidence contradicts this expectation. First, the modalities of financial inclusion might go against the objective of autonomy. This contradiction is particularly true for microcredit, since clients’ creditworthiness precisely depends on their ability to mobilize social resources, either through joint liability groups or personal guarantors (Schuster 2015). Moreover, both men and women users of microfinance services are not necessarily looking to be “autonomous.” In many contexts, as anthropology has long shown, being connected and dependent on others is both a mode of action and a deliberate strategy, for two main reasons: to construct and maintain one’s identity, and to safeguard a network of material protection (Ferguson 2015). As a result, people’s agency translates into the ability to choose certain forms of dependency and interdependence. It is through this frame of analysis that we should seek to understand how financial services are either accepted, rejected, or used for reasons other than those intended by their promoters (Guérin et al. 2019; Johnson 2016; Kusimba 2020). First, we shall take the case of savings. Over recent decades, several initiatives have sought to design innovative services to tackle income irregularity, chronic uncertainty, illiteracy, and spatial and social isolation. Since Collins et al. (2009), providing “convenient, flexible and reliable” products has been made a key objective. This has been facilitated by growing innovation in mobile banking and financial technologies. Various governments’ shifts towards making social benefit payments directly into bank accounts have been credited with helping incentivize saving.11 Saving is often less successful than microcredit or digital payments (Karlan et al. 2014). Although the number of bank account owners is rising, many accounts are dormant (25 percent in developing countries; Demirgüç-Kunt et al. 2018: 8), and “more account ownership does not necessarily translate into more formal saving” (Demirgüç-Kunt et  al. 2018: 71). Such resistance to saving is due to the fact that in many contexts, savings in a bank account

76  Handbook of microfinance, financial inclusion and development

are experienced as something that cuts savers off from their social relations. Formal saving, whether through banks or microfinance providers, is only one of many forms of storing, accumulating, and circulating value. Circulating value may carry more meaning than storing it, both individually and collectively, especially over the long term (Guyer 1997). In rural Tamil Nadu for instance, bank account ownership has risen to almost 100 percent at household levels, but the median amounts saved are extremely low. Gold, lending to others, and ceremonial gifts (to be reciprocated) remain the main forms of saving. Of course, these are specific savings, but they remain very liquid, and certainly more liquid than bank savings. A neighbor who has been lent money is supposed to reciprocate quickly. In the same way, gold operates as a quasi-money since there are pawnbrokers everywhere. Ceremonial spending, on the other hand, is closer to long-term investment saving. These forms of saving play a crucial role in household reproduction. It works both economically, by helping to accumulate lump sums and smooth expenses and income over time, and socially, by helping maintain or even strengthen family and clan status and reputation. India has the highest share of dormant accounts (48 percent) and is certainly a particular case (Demirgüç-Kunt et al. 2018: 65). But ethnographies from various parts of the world have shown our study on India to be far from unique.12 By contrast, mobile banking has spread at great speed and scale, at least in the regions where it is authorized through legislation. Its success has been down to the fact that users take advantage of the service not to save money, but to maintain their network of relationships, or possibly to modify it by more carefully selecting to whom they give and how much (Morvant-Roux & Peixoto-Charles 2020; Rea & Nelms 2017). In the case of Kenya, contrary to what some digital finance promoters predicted, so far digital finance has rarely been used as a springboard for banking inclusion: it is mainly designed as a means of “moving money within interpersonal networks” (Johnson 2016: 92). Most users withdraw funds completely after receiving a transfer. Mobile banking promises improved peer-to-peer banking relationships, an end to the “inefficiencies” of intermediaries, and the promotion of cashless transactions. But here too, real usage goes against theory and most transactions have not tended to be peer-to-peer. Various empirical studies have shown that cash remains “king” (Stuart & Cohen 2011: 17) and that mobile money users have remained strongly embedded in social networks: a complex web of intermediaries is necessary to build trust and access the service, whether for cash in or cash out (Morvant-Roux & Peixoto-Charles 2020; Rea & Nelms 2017). Ultimately, while formal financial services do not promote women’s “autonomy” or selfreliance, they may help to reshape interdependencies (Guérin et  al. 2019; Schuster 2015). This is also why financial inclusion rarely replaces pre-existing financial practices. “Juggling” practices prevail. These involve resorting to a wide range of borrowing sources, borrowing from one place to pay back to another, lending out in one place while borrowing elsewhere, saving from one place while borrowing from another, etc. In the face of scarcity, such juggling is a necessity. But it also adheres to complex strategies aiming to multiply and diversify social relationships, strengthening or weakening the burden of dependency ties. Depending on how much leeway women have to juggle, which is based not only on their economic wealth, but also on their social skills and their degree of cooperation with their spouses and relatives, the practice of juggling can be more or less empowering. However, the outcome is difficult to measure, let alone predict. Even for the most successful businesses women, it is precisely the ability to enmesh themselves in neighborhood and kinship networks that is a source of success rather than

Financial inclusion and gender  77

an independent and autonomous business spirit (Schuster 2015: 53). Ethnographic evidence shows that some women manage to diversify their social relations, including extending and strengthening kinship relationships, while others are now “addicted” to credit (Kar 2018: 199). Yet this “addiction,” given its cost, can be an important source of extraction of their income (Reboul et al. 2019).

CONCLUSION Financial inclusion policies are not part of a vacuum. Besides their diversity of models, which is a first differentiating factor, the way in which they are implemented is affected by a wide range of social, cultural, economic, and political institutions. These include women’s place in the productive and reproductive spheres, the existence of protection systems, the extent of discriminatory norms, and prior debt levels, which are growing but at varying magnitudes. Financial tools cannot promote women’s “autonomy” and “self-reliance,” but may enhance their ability to manage their income and expenditure over time, to better choose forms of interdependence, and to better negotiate within this set of institutions. Regarding microcredit, it may also increase their indebtedness due to aggressive commercial policies, and rigid repayments that are incompatible with low and irregular incomes, which in turn reinforce oppressive bonds of dependency. The perverse effects of female-only targeting, which is widespread in Asia, are mainly founded on the logic of securing repayments. Historically, credit has never been considered a tool for women’s empowerment, but simply a means for making economic relations more fluid. There is no reason why it should be otherwise today. Financial inclusion supporters may need to tone down claims that financial inclusion can solve the problem of gender discrimination and inequality. In a competitive context where different players have to promote the effectiveness of the projects or tools they develop and finance, it is easy to understand the rationale behind their appealing slogans. Such simplistic discourses are part of a broader trend in development. The withdrawal of the state as planner and developer has led to “thinking small” (Cohen and Easterly 2010). This has gone along with the privatization of interventions and players, and the marketization of the goods and services delivered. The promises of financial inclusion have served as an additional incentive for policymakers to turn away from structural measures, such as investing in social infrastructure, setting up effective social protection systems, broadening their tax base, securing women’s property rights, etc. Yet it is precisely the absence or decline of such structural measures which is at the root of the explosion of private household debt. Financial inclusion, far from alleviating the problem of over-indebtedness, can sometimes make it worse. It is also the absence or decline of such structural measures which is at the root of gender inequalities. Many financial inclusion actors certainly have good intentions. However, when they offer excessive promises, they unintentionally feed into this wider dynamic of “thinking small.” In many contexts, women are already juggling multiple sources of borrowing, saving, and sharing. They may need better quality services than they already have. However, care must be taken to ensure that female targeting is not primarily a strategy for ensuring a captive and disciplined clientele. Financial inclusion can only benefit women if services are affordable, non-discriminatory, and genuinely tailored to their needs and constraints, which are very diverse in different contexts and social groups.

78  Handbook of microfinance, financial inclusion and development

NOTES 1.

2.

3. 4.

5. 6. 7. 8. 9. 10. 11. 12.

This is not to claim that RCTs are without use. But they serve very specific purposes for particular development policies, and are ill-equipped to explain or even adequately measure financial inclusion impacts. For specific criticisms of RCTs and financial inclusion, see (Bédécarrats, Guérin, Morvant-Roux, et al. 2019; Bédécarrats, Guérin, & Roubaud 2019; Kabeer 2019; Wydick 2016). Muhammad Yunus, for instance, stated in a 2017 speech: “There are roughly 160 million people all over the world in microcredit, mostly women. And they have proven one very important thing: that we are all entrepreneurs. Illiterate rural women in the villages, in the mountains, take tiny little loans—$30, $40—and they turn themselves into successful entrepreneurs.” www​.theguardian​. com ​/sustainable ​-business​/2017​/mar​/29​/we ​-are ​-all​- entrepreneurs​-muhammad​-yunus​- on​- chan ging​-the​-world​-one​-microloan​-at​-a​-time [last accessed March 28, 2017]. www​.worldbank​.org ​/en ​/news​/press​-release​/2018​/04​/19​/financial​-inclusion​-on​-the​-rise​-but​-gaps​remain​-global​-findex​-database​-shows. For instance, World Bank President Jim Yong Kim said: “In the past few years, we have seen great strides around the world in connecting people to formal financial services [. . .] Financial inclusion allows people to save for family needs, borrow to support a business, or build a cushion against an emergency. Having access to financial services is a critical step towards reducing both poverty and inequality, and new data on mobile phone ownership and internet access show unprecedented opportunities to use technology to achieve universal financial inclusion.” A comment by Melinda Gates, Co-Chair of the Bill & Melinda Gates Foundation, is even stronger and explicitly mentions women: “We already know a lot about how to make sure women have equal access to financial services that can change their lives [. . .] When the government deposits social welfare payments or other subsidies directly into women’s digital bank accounts, the impact is amazing. Women gain decision-making power in their homes, and with more financial tools at their disposal they invest in their families’ prosperity and help drive broad economic growth.” www​.worldbank​.org​/en​/news​/ press​-release​/2018​/04​/19​/financial​-inclusion​-on​-the​-rise​-but​-gaps​-remain​-global​-findex​-database​shows. Regarding microcredit, see, for instance, Collins et al. (2009), Guérin, D’Espallier, et al. (2015), Johnston Jr. and Morduch (2008), and Morvant-Roux et al. (2014). Data from Kenya indicate that this is also true of digital credit borrowing (Mustafa et al. 2017). See Garikipati et  al. (2017) and Karim (2011) for the Global South; Callegari et  al. (2019), Goldberg (2005), and Thorne (2010) for the Global North; and Reboul (forthcoming) for an historical overview. See Angulo Salazar (2013), Hummel (2013), Joseph (2013), Kar (2018), Karim (2011), Rahman (1999), Rankin (2002), Schuster (2015), and Taylor (2011). See, for instance, Garikipati (2008), Guérin, Kumar, et al. (2013), Kalpana (2016), Karim (2011), Mayoux (2001), Pattenden (2010), Rankin (2002), Rao (2008), and Wright (2004). See, for instance, Guérin (2006), Guérin, Kumar, et al. (2013) and Mayoux (2001). Observations made by Ackerly (1995) are confirmed by various ethnographies quoted in the rest of the chapter. See, for instance, in Mexico (Fouillet & Morvant-Roux 2018) or India (Goedecke et al. 2018). See Maurer (2015: 102) and Peebles (2010) for general overviews. See Guérin (2006) for Senegal and Johnson (2016) and Shipton (2010) for Kenya.

REFERENCES Ackerly, B. A. (1995). ‘Testing the tools of development: Credit programmes, loan involvement, and women’s empowerment’, IDS Bulletin, 26/3: 56–68. Agier, I., & Szafarz, A. (2013). ‘Microfinance and gender: Is there a glass ceiling on loan size?’, World Development, 42: 165–81. Angulo Salazar, L. (2013). ‘The social costs of microfinance and over-indebtedness for women’, in Guérin, I., Morvant-Roux, S., & Villarreal, M. (eds), Microfinance, debt and over-indebtedness. Juggling with money, pp. 232–52. London: Routledge.

Financial inclusion and gender  79

Armendariz, B., & Morduch, J. (2005). The economics of microfinance. Cambridge, MA: MIT Press. Bateman, M. (2010). Why doesn’t microfinance work?: The destructive rise of local neoliberalism. London: Zed Books. Bédécarrats, F., Guérin, I., Morvant-Roux, S., & Roubaud, F. (2019). ‘Estimating microcredit impact with low take-up, contamination and inconsistent data. A review of Crépon, Devoto, Duflo and Pariente (American Economic Journal: Applied Economics, 2015)’, International Journal for Re-Views in Empirical Economics, 3, 1–22. Bédécarrats, F., Guérin, I., & Roubaud, F. (2019). ‘All that Glitters is not Gold. The political economy of randomized evaluations in development’, Development and Change, 50/3: 735–62. Boserup, E. (1970). Woman’s role in economic development. London: George Allen & Unwin. Bruce, J., & Dwyer, D. (1988). A home divided: Women and income in the third world. Stanford, CA: Stanford University Press. Buggenhagen, B. (2011). ‘Are births just “women’s business”? Gift exchange, value, and global volatility in Muslim Senegal’, American Ethnologist, 38/4: 714–32. Bylander, M. (2015). ‘Credit as coping: Rethinking microcredit in the Cambodian context’, Oxford Development Studies, 43/4: 533–53. Bylander, M. (2020). ‘“If You Fall, Stand Up Again”: The moral nature of financial literacy in the global south’, Development and Change, 52(1): 26–53. Callegari, J., Liedgren, P., & Kullberg, C. (2019). ‘Gendered debt–a scoping study review of research on debt acquisition and management in single and couple households’, European Journal of Social Work, 23(5), 1–13. Carswell, G., Neve, G. D., & Ponnarasu, S. (2020). ‘Good debts, bad debts: Microcredit and managing debt in rural south India’, Journal of Agrarian Change, 21(1), 122–42. Cohen, J., & Easterly, W. (2010). What works in development?: Thinking big and thinking small. Washington, DC: Brookings Institution Press. Collins, D., Morduch, J., Rutherford, S., & Ruthven, O. (2009). Portfolios of the poor: How the world’s poor live on $2 a day. Princeton, NJ: Princeton University Press. Corsi, M., & De Angelis, M. (2017). ‘Gender discrimination in microfinance? Some evidence from Uganda’, The Journal of Development Studies, 53/5: 723–40. Cozarenco, A., & Szafarz, A. (2018). ‘Gender biases in bank lending: Lessons from microcredit in France’, Journal of Business Ethics, 147/3: 631–50. Cull, R., & Morduch, J. (2017). Microfinance and economic development. Policy Research Working Paper 8252. Washington, DC: World Bank. Demirgüç-Kunt, A., Klapper, L., Singer, D., Ansar, S., & Hess, J. (2018). The global findex database 2017: Measuring financial inclusion and the fintech revolution. Washington, DC: The World Bank. D’Espallier, B., Guérin, I., & Mersland, R. (2011). ‘Women and repayment in microfinance: A global analysis’, World Development, 39/5: 758–72. Ferguson, J. (2015). Give a man a fish: Reflections on the new politics of distribution. Durham, NC: Duke University Press. Fontaine, L. (2008). L’économie morale: pauvreté, crédit et confiance dans l’Europe préindustrielle. Paris: Editions Gallimard. Fouillet, C., & Morvant-Roux, S. (2018). ‘L’inclusion financière au service de l’État en Inde et au Mexique ?’, International Development Policy | Revue internationale de politique de développement [online], 10/1. Garikipati, S. (2008). ‘The impact of lending to women on household vulnerability and women’s empowerment: Evidence from India’, World Development, 36/12: 2620–42. Garikipati, S., Johnson, S., Guérin, I., & Szafarz, A. (2017). ‘Microfinance and gender: Issues, challenges and the road ahead’, The Journal of Development Studies, 53/5: 641–8. Goedecke, J., Guérin, I., D’espallier, B., & Venkatasubramanian, G. (2018). ‘Why do financial inclusion policies fail in mobilizing savings from the poor? Lessons from rural South India’, Development Policy Review, 36: O201–19. Goetz, A. M., & Gupta, R. S. (1996). ‘Who takes the credit? Gender, power, and control over loan use in rural credit programs in Bangladesh’, World Development, 24/1: 45–63. Goldberg, L. (2005). ‘Dealing in debt: The high-stakes world of debt collection after FDCPA’, Southern California Law Review, 79: 711.

80  Handbook of microfinance, financial inclusion and development

Guérin, I. (2006). ‘Women and money: Lessons from Senegal’, Development and Change, 37/3: 549–70. Guérin, I. (2011). ‘Do women need specific microfinance services?’, in Armandariz, B. & Labie, M. (eds), Handbook of Microfinance, pp. 563–89. London and Singapore: World Scientific Publishing. Guérin, I. (2012). ‘Households’ over-indebtedness and the fallacy of financial education: Insights from economic anthropology’, Microfinance in Crisis Working Papers Serie, 1. Paris I Sorbonne University/IRD Paris. Guérin, I., D’Espallier, B., & Venkatasubramanian, G. (2015). ‘The social regulation of markets: Why microcredit fails to promote jobs in rural South India’, Development and Change, 46/6: 1277–301. Guérin, I., & Kumar, S. (2017). ‘Market, freedom and the illusions of microcredit. Patronage, caste, class and patriarchy in Rural South India’, The Journal of Development Studies, 53/5: 741–54. Guérin, I., & Kumar, S. (2020). ‘Unpayable debts. Debt, gender and sex in financialized India’, The American Ethnologist, 47(3): 219–33. Guérin, I., Kumar, S., & Agier, I. (2013). ‘Women’s empowerment: Power to act or power over other women? Lessons from Indian microfinance’, Oxford Development Studies, 41(suppl 1): S76–94. Guérin, I., Labie, M., & Servet, J.-M. (eds). (2015). The crises of microcredit. London: Zed Books. Guérin, I., Morvant-Roux, S., & Servet, J.-M. (2019). ‘Breaking away from normative approaches to financial practices’, in Hudon, M., Labie, M., & Szafarz, A. (eds), A research agenda for financial inclusion and microfinance, pp. 54–68. Cheltenham: Edward Elgar Publishing. Guérin, I., Morvant-Roux, S., & Villarreal, M. (eds). (2013). Microfinance, debt and over-indebtedness: Juggling with money. London and New York: Routledge. Guyer, J. I. (1997). ‘Endowments and assets: The anthropology of wealth and the economics of intrahousehold allocation’, in Haddad, J., Hoddinott, J., & Alderman, H. (eds), Intrahousehold resource allocation in developing countries, pp. 112–29. Baltimore, MD: The Johns Hopkins University Press. Holvoet, N. (2005). ‘The impact of microfinance on decision-making agency: Evidence from South India’, Development and Change, 36/1: 75–102. Hoodfar, H. (1988). ‘Household budgeting and financial management in a lower-income Cairo neighborhood’, in Bruce, J. & Dwyer, D. (eds), A home divided: Women and income in the third world, pp. 120–42. Stanford, CA: Stanford University Press. Hummel, A. (2013). ‘The commercialization of microcredits and local consumerism: Examples of over-indebtedness from indigenous Mexico’, in Guérin, I., Morvant-Roux, S., & Villarreal, M. (eds), Microfinance, debt and over-indebtedness. Juggling with money, pp. 253–71. London: Routledge. ILO. (2018). Paid employment vs vulnerable employment (ILOSTAT. Spot Light on Work Statistics No. 3). Geneva: ILO. Johnson, S. (2004). ‘Gender norms in financial markets: Evidence from Kenya’, World Development, 32/8: 1355–74. Johnson, S. (2005). ‘Gender relations, empowerment and microcredit: Moving on from a lost decade’, The European Journal of Development Research, 17/2: 224–48. Johnson, S. (2016). ‘Competing visions of financial inclusion in Kenya: The rift revealed by mobile money transfer’, Canadian Journal of Development Studies / Revue canadienne d’études du développement, 37/1: 83–100. Johnston Jr, D., & Morduch, J. (2008). ‘The unbanked: Evidence from Indonesia’, The World Bank Economic Review, 22/3: 517–37. Joseph, N. (2013). ‘Mortgaging used saree-skirts, spear-heading resistance: Narratives from the microfinance repayment standoff in Ramanagaram, India, 2008–2010’, in Guérin, I., Morvant-Roux, S., & Villarreal, M. (eds), Microfinance, Debt and Over-indebtedness. Juggling with money, pp. 272–94. London: Routledge. Kabeer, N. (2001). ‘Conflicts over credit: Re-evaluating the empowerment potential of loans to women in rural Bangladesh’, World Development, 29/1: 63–84. Kabeer, N. (2005). ‘Is microfinance a “magic bullet” for women’s empowerment? Analysis of findings from South Asia’, Economic and Political Weekly, 4709–18. Kabeer, N. (2016). ‘Gender equality, economic growth, and women’s agency: The “endless variety” and “monotonous similarity” of patriarchal constraints’, Feminist Economics, 22/1: 295–321.

Financial inclusion and gender  81

Kabeer, N. (2019). ‘Randomized control trials and qualitative evaluations of a multifaceted programme for women in extreme poverty: Empirical findings and methodological reflections’, Journal of Human Development and Capabilities, 20/2: 197–217. Kalpana, K. (2016). Women, microfinance and the state in neo-liberal India. New Delhi: Routledge India. Kar, S. (2018). Financializing poverty: Labor and risk in Indian microfinance. Stanford, CA: Stanford University Press. Karim, L. (2011). Microfinance and its discontents. Women in debt in Bangladesh. Minneapolis, MN: University of Minnesota Press. Karlan, D., Ratan, A., & Zinman, J. (2014). ‘Savings by and for the poor: A research review and Agenda’, Review of Income and Wealth, 60: 36–78. Kusimba, S. (2018). ‘“It is easy for women to ask!”: Gender and digital finance in Kenya’, Economic Anthropology, 5/2: 247–60. Kusimba, S. (2020). Reimagining money. Stanford, CA: Stanford University Press. Lapavitsas, C. (2011). ‘Theorizing financialization’, Work, Employment and Society, 25/4: 611–26. Lavinas, L. (2017). The takeover of social policy by financialization. The Brazilian paradox. New York: Palgrave Macmillan. Lemire, B., Campbell, G., & Pearson, R. (eds). (2001). Women and credit: Researching the past, refiguring the future, Vol. 24. Oxford and New York: Berg Publishers. Maher, V. (1981). ‘Work, consumption and authority within the household: A Moroccan case’, in Young, K., Wolkowitz, C., & McCullagh, R. (eds), Of marriage and the market. Women’s subordination internationally and its lessons, pp. 117–35. London: Routledge and Kegan Paul. Maurer, B. (2015). How would you like to pay? How technology is changing the future of money. Durham, NC: Duke University Press. Mayoux, L. (2001). ‘Tackling the down side: Social capital, women’s empowerment and micro-finance in Cameroon’, Development and Change, 32/3: 435–64. Mayoux, L. (2010). ‘Reaching and empowering women: Towards a gender justice protocol for a diversified, inclusive, and sustainable financial sector’, Perspectives on Global Development and Technology, 9/3–4: 581–600. Montgomery, R. (1996). ‘Disciplining or protecting the poor? Avoiding the social costs of peer pressure in micro-credit schemes’, Journal of International Development, 8/2: 289–305. Morvant-Roux, S. (2013). ‘International migration and over-indebtedness in rural Mexico’, in Guérin, I., Morvant-Roux, S., & Villarreal, M. (eds), Microfinance, debt and over-indebtedness: Juggling with money, pp. 170–92. London and New York: Routledge. Morvant-Roux, S., Guérin, I., Roesch, M., & Moisseron, J.-Y. (2014). ‘Adding value to randomization with qualitative analysis: The case of microcredit in rural Morocco’, World Development, 56: 302–12. Morvant-Roux, S., & Peixoto-Charles, A. (2020). ‘Here and there? Mobile money and the politics of transnational living patterns in West Africa’, Oxford Development Studies, 48/2: 181–94. Mustafa, Z., Wachira, M., Bersudskaya, V., Nanjero, W., & Wright, G. A. N. (2017). Where credit is due. Customer experience of digital credit in Kenya. MicroSave. Papanek, H., & Schwede, L. (1988). ‘Woman are good with money: Earning and managing in an Indonesian city’, in Bruce, J. & Dwyer, D. (eds), A home divided: Women and Income in the Third World, pp. 80–98. Stanford, CA: Stanford University Press. Pattenden, J. (2010). ‘A neoliberalisation of civil society? Self-help groups and the labouring class poor in rural South India’, The Journal of Peasant Studies, 37/3: 485–512. Peebles, G. (2010). ‘The anthropology of credit and debt’, Annual Review of Anthropology, 39: 225–40. Perry, D. (2002). ‘Microcredit and women moneylenders: The shifting terrain of credit in rural Senegal’, Human Organization, 61/1: 30–40. Rahman, A. (1999). ‘Micro-credit initiatives for equitable and sustainable development: Who pays?’, World Development, 27/1: 67–82. Rankin, K. N. (2002). ‘Social capital, microfinance, and the politics of development’, Feminist Economics, 8/1: 1–24. Rao, S. (2008). ‘Reforms with a female face: Gender, liberalization, and economic policy in Andhra Pradesh, India’, World Development, 36/7: 1213–32.

82  Handbook of microfinance, financial inclusion and development

Rea, S. C., & Nelms, T. C. (2017). ‘Mobile money: The first decade’, Institute for Money, Technology and Financial Inclusion Working Paper, 1. Reboul, E. (Forthcoming). ‘Microcredit, a revolution? A brief historical perspective on women’s debt, between (in)visibility and (il)legitimacy’, in Chiapello, E., Engels, A., & Gresse, E. (eds), Financialisations of development: Global games and local experiments. London: Routledge. Reboul, E., Guérin, I., Raj, A., & Venkatasubramanian, G. (2019). ‘Managing economic volatility. A gender perspective’, Working Papers CEB 19-015, ULB. Saiag, H. (2020). ‘Financialization from the margins Notes on the incorporation of Argentina’s subproletariat into consumer credit ( 2009–2015)’, Focaal—Journal of Global and Historical Anthropology, 87: 16–32. Schicks, J. (2014). ‘Over-indebtedness in Microfinance–an empirical analysis of related factors on the borrower level’, World Development, 54: 301–24. Schuler, S. R., Hashemi, S. M., Riley, A. P., & Akhter, S. (1996). ‘Credit programs, patriarchy and men’s violence against women in rural Bangladesh’, Social Science & Medicine, 43/12: 1729–42. Schuster, C. (2015). Social collateral women and microfinance in Paraguay’s smuggling economy. Stanford, CA: Stanford University Press. Servet, J.-M., & Saiag, H. (2013). ‘Household over-indebtedness in northern and southern countries: A macro-perspective’, in Guérin, I., Morvant-Roux, S., & Villarreal, M. (eds), Microfinance, debt and over-indebtedness, pp. 44–65. London and New York: Routledge. Shaffer, P. (2015). ‘Two concepts of causation: Implications for poverty’, Development and Change, 46/1: 148–66. Shipton, P. M. (2010). Credit between cultures: Farmers, financiers, and misunderstanding in Africa. New Haven, CT: Yale University Press. Still, C. (2015). The imperatives of honour: Dalit women and patriarchy in south India. New Delhi: Social Science Press. Stuart, G., & Cohen, M. (2011). Cash-in, cash-out: The role of M-PESA in the lives of low-income people (Financial Services Assessment). Washington, DC: Microfinance Opportunities. Taylor, M. (2011). ‘“Freedom from poverty is not for free”: Rural development and the microfinance crisis in Andhra Pradesh, India’, Journal of Agrarian Change, 11/4: 484–504. Thorne, D. (2010). ‘Extreme financial strain: Emergent chores, gender inequality and emotional distress’, Journal of Family and Economic Issues, 31/2: 185–97. Todd, H. (1996). Women at the center: Grameen Bank borrowers after one decade. Dhaka: University Press. United Nations. (2020). Private debt and human rights (Report of the independent expert on the effects of foreign debt and other related international financial obligations of States on the full enjoyment of human rights, particularly economic, social and cultural rights). Geneva: United Nations. UNCTAD. (2019). Trade and development report 2019. Financing a global green new deal. Geneva: United Nations. WIEGO, & ILO. (2014). Women and men in the informal economy. A statistical picture, 2nd éd. Geneva: International Labour Office (ILO). Wright, K. (2004). ‘The darker side to microfinance’, in Fernando, J. (ed.), Microfinance. Perils and prospects, pp. 154–71. London and New York: Routledge. Wydick, B. (2016). ‘Microfinance on the margin: Why recent impact studies may understate average treatment effects’, Journal of Development Effectiveness, 8/2: 257–65.

5. Toward a theory of fair interest rates on microcredit: balancing the needs of clients and institutions Marek Hudon and Joakim Sandberg

INTRODUCTION Debates about fairness in pricing are seemingly on the rise on many fronts in contemporary society. For example, some people are upset about how the current patent regime makes various life-saving drugs too costly for poor people. Similarly, the so-called fair trade movement lobbies strongly for fair remunerations for low-income workers (Huybrechts et al., 2017). However, the concern for fair prices has a long history (Johnson, 1938; Elegido, 2009). It probably originates with Aristotle’s (1946 [350 BC]) famous comments on how the “true” price of various goods and services sometimes differs from the market price. Aristotle’s main object of concern was interest, i.e. the price of loans, which he famously denounced as the unnatural fruit of a barren parent. This chapter concerns microcredit which is the practice of extending loans to poor or lowincome clients; quite often these are tiny loans (of less than $100) to the very poorest of the global poor (Cull et al., 2018; Tchakoute Tchigoua, 2016; Morduch and Cull, 2017). This practice was in a way developed precisely to offer fairer prices to small-scale entrepreneurs in developing countries. Prior to microcredit, these borrowers had to rely on loans from informal lenders (or “money sharks”) who often charge rates of 1,000 percent in interest (Armendariz and Morduch, 2010). Moreover, microfinance institutions (MFIs) have proven capable of reaching clients that previously lacked access to financial services altogether, such as women in rural areas (Hermes and Hudon, 2018). However, a perhaps unexpected fact is that also MFIs charge very high interest rates. They charge rates that are much higher than what commercial banks in developed countries charge. The high rates are mainly due to the high transaction and operating costs involved in the upkeep of very small loans. The average nominal rate on microloans is around 30 percent globally (D’Espallier et al., 2017a, 2017b), and the absolute majority of MFIs charge between 20–60 percent in interest per year. However, in some extreme cases, MFIs charge much more. The most infamous case is Compartamos, one of the fastest-growing commercial MFIs in Latin America, which made its clients pay around 100 percent per year plus VAT (Rosenberg, 2007). As the rates charged by MFIs have become more widely known in the last decade, various outside agents have raised ethical concerns about the practices of MFIs. Branches of large MFIs have been closed by authorities in Ecuador, India, and Nicaragua (Counts, 2008). Ugandan and Cameroonian branches have also been closed, and the central bank of Bangladesh has decided to regulate the industry by imposing a strict ceiling (or cap) on interest levels. Not all of this attention is due to the interest rates. Some concerns come from the fragile status of borrowers. MFIs have been criticized for their loan collection practices or their lack of concern for clients 83

84  Handbook of microfinance, financial inclusion and development

that have become overindebted. In any case, ethical issues – and particularly debates on interest rates – are now widely accepted to constitute a major threat to the credibility of the entire microfinance sector (Hudon and Sandberg, 2013). Given the emergence of this critical issue, the present chapter looks to analyze the ethical dimensions of lending to the poor. We build on previous work, including Hudon (2007), Hudon and Ashta (2013), and Sandberg (2012), in order to develop a conceptual framework for thinking more clearly about the fairness of prices in microcredit. The structure of the chapter is as follows. The second section outlines the main tenets of the debate as it has taken shape in the microfinance industry itself. In the third through sixth sections, thereafter, we attempt to deepen the debate by organizing thoughts into four more theoretical approaches to fairness in interest rates. The third section considers what we call the procedural approach, the fourth section the competitive market approach, the fifth section the credit-as-a-right approach, and the sixth section the consequentialist approach. Finally, the seventh section presents our own perspective which is a combination of the two last approaches. We believe that such a combination is needed in order to adequately balance the needs of clients with the needs of the institutions. A clarification is in order before we begin: interest rates are notoriously difficult to compare, since their level may vary “naturally” with factors such as social and economic environments, customs, taxes, currencies, public policies, and even cultural and historical aspects (Conard, 1959; Homer and Sylla, 2005). This problem is even bigger in microcredit since the interest payment is seldom the only cost for borrowers. They may also be obliged to keep deposits (also called compulsory savings) during the reimbursement period, or even months prior to taking the loan, or they may have to pay various other fees on top of the interest. In addition, the largest part of the cost of lending is sometimes not related to the loan itself but to transaction costs associated with the lending methodology (Collins et al., 2009). For these reasons, it is perhaps wise to talk about effective interest rates, or more specifically the annual percentage rate (APR). The APR includes a wider range of variables such as the concrete interest payments, both upfront and consequent, the loan (or service) fee, and the contribution to group funds (Ledgerwood and White, 2006). For the remainder of the chapter, then, when we talk about interest rates, we mean APRs.

THE DEBATE ON INTEREST RATES IN THE MICROFINANCE INDUSTRY Prior to the 1970s, the interest rates charged on loans to poor entrepreneurs by development projects, particularly in rural settings, were very low indeed. It is difficult to know the rationale behind this but one may speculate in that it did not seem socially desirable to charge higher prices to the most deprived, or perhaps the idea was that the poorest would not be able to repay higher rates. In any case, the natural inclination of these original development projects seems to have been to set prices with an eye toward the welfare levels of the borrowers. During the 1970s and 1980s, fierce debates took place on the issue of these low interest rate policies. These debates developed hand in hand with the emergence of the kind of commercial microfinance institutions that we are concerned with here. Representative of this debate is Adams et al.’s (1984) seminal book Undermining Rural Development with Cheap Credit and several other works from the so-called “Ohio School” – a group of influential economists at Ohio State University (see Hulme and Mosley, 1996). Ohio School scholars argued that

Toward a theory of fair interest rates on microcredit  85

almost nothing had come out of the many billions of dollars spent on credit programs targeting the poor since World War II, and the reason for this was that subsidized loans are dysfunctional. Access to cheap credit gives no incentive to save, and as a result, distorts the way that lenders allocate funds (Adams et al., 1984, p. 75). Furthermore, they argued that “low interest rates on loans to rural people end, paradoxically, by restricting their access to financial services” – mainly since few commercial institutions are willing to enter this market without subsidies (Von Pischke, 1983, p. 176). Finally, they argued that many had underestimated or neglected the fact that low rates would create an underdeveloped financial infrastructure. The Ohio School scholars accepted that the commercialization of development loans most likely would lead to higher interest rates for poor borrowers. However, their motto was that “access is more important than price” – that is, having reliable access to loans is more valuable for the borrowers than having access to cheap, large, or long-term loans (Adams and Von Pischke, 1980). Moreover, a common argument was that it is essential to guarantee the continued existence of the relevant institutions, since if the institutions were to fail to be profitable and go bankrupt, the clients would end up with nothing (or fall into the hands of the moneylenders who charge even higher rates). It is in this sense that low and subsidized interest rates were thought to be counter-productive. Rather than focusing on the borrowers’ “need” for cheap credit, then, they argued that MFIs should offer “cost-covering interest rates” which would enable MFIs to continue to operate (Adams and Von Pischke, 1980). We believe that the ensuing debate on interest rates within the microfinance industry, from the 1990s to the present time, can be understood as a continuous search for balance between the two elements above: i.e. the welfare of the customers on the one side and the imperative of building strong institutions on the other. Following Woller et al. (1999), it has become common to characterize the debate as one between the “welfarist” and the “institutionist” camps. On the one hand, welfarist proponents – such as Nobel Peace Prize Laureate Muhammad Yunus – stress the idea that the overarching goal of microfinance must be the clients’ interest in escaping poverty. Interest rates should therefore be set from the perspective of helping the impoverished, rather than supporting the economic stability of the lending institutions. From this perspective, the suggestion is that interest rates should be low enough to give borrowers enough margin to develop their microbusinesses. While proponents of this view have been critical of high interest rates ever since the 1970s, similar criticisms have really intensified during the last decade (Hudon et al., 2018; Tchakoute Tchigoua, 2016). On the other hand, institutionist proponents – such as the Consultative Group to Assist the Poor (CGAP), a microfinance donor organization housed at the World Bank – put emphasis first and foremost on the MFIs themselves and their financial sustainability. In line with the Ohio School’s thinking, access is considered to be more important than price and it is indeed thought that poor households can afford very high rates (CGAP, 1996). Moreover, the ethical justification of microcredit is partly thought to stem from a comparison with the prices charged by informal lenders and pawnbrokers because, as noted above, microcredit is normally cheaper than what the borrowers used to pay to finance their activity on the informal market (McKenzie and Woodruff, 2008; de Mel et al., 2008). Now, which side of the debate above is correct? Are today’s high rates on microloans unfair or not? Unfortunately, both sides tend to be vague on the philosophical underpinnings of their arguments or views. More specifically, the previous literature on microcredit contains no systematic discussion of various conceptions of fairness – that is, of more theoretical principles that aim to capture the essence of what people take to be fair or just. This is where we think

86  Handbook of microfinance, financial inclusion and development

that philosophy can help to bring further clarity and new insights. The aim of philosophical ethics is exactly to try to elevate our everyday feelings of what is right or just to a higher degree of theoretical precision – formulated in terms of abstract principles or perspectives – which can help to clarify on which points we agree or disagree about various ethical issues. In what follows, we present and discuss four more theoretical approaches to fairness that we think are relevant to the contemporary debate about fairness in microcredit. Our aim with this exercise is both to consolidate the existing debate in the industry, as well as to develop it further with the help of a philosophical method of inquiry. We will use the approaches to identify relevant points of tension in the previous debate and evaluate the arguments for different views with greater precision. The ultimate goal is to get closer to a “theory” of the fairness of interest rates that strikes a reasonable balance between different interests and viewpoints.

THE PROCEDURAL APPROACH TO FAIRNESS According to a first perspective, which is fairly common in the business ethics literature, the fairness of a given market transaction depends primarily on the degree of voluntariness of the transacting parties and their consent to the terms of the transaction. Gielissen et al. (2008, p. 372) explain that a procedural analysis of prices focuses on “whether the seller has ‘played fair’ by adhering to the rules of process when setting the price”. Applied to the microcredit context this means that any interest rate is fair as long as it is the result of a free negotiation process where neither the lender nor the debtor is coerced or deceived. We may call this the procedural approach to fairness in interest rates. This approach could be said to be the default, or “no theory needed”, option in the circumstances since it represents the idea that only MFIs and their clients themselves can decide whether a given interest rate is fair or not. Proponents of this view may legitimately ask: who are we as commentators to moralize over interest rate levels from afar? Should we not let this be an issue strictly between the MFIs and their clients? We start with the procedural approach here for two reasons: firstly because we believe that it underpins a central argument from proponents of the institutionist view, namely that high repayment rates indicate borrowers’ acquiescence to the current interest levels. It is an interesting fact about today’s microfinance industry that the average repayment rate on microloans is very high. According to Lensink et al. (2018), the average “portfolio at risk” (30 days) on microloans was as low as 6 percent. The argument from some commentators is then that the fact that clients always pay back their loans (including the interest) – and indeed typically come back for more – can be taken as an (at least preliminary) justification of the interest rates set by the relevant MFIs. For instance, this is a central argument in CGAP’s official advice on interest rates to MFIs, which contains a section on “The Theory and Practice of ‘Exorbitant’ Interest Rates”. Part of this section reads: For the past ten years, the author of this paper has been asking in conferences, courses, and (more recently) Internet newsgroups whether anyone present has ever heard of a microfinance program that ran into trouble by driving away clients with interest rates that were too high. No one has yet pointed to a single example. This remarkable piece of data does not indicate that there is no limit to the interest rates that the microcredit market can bear, but it does suggest that the limit is probably considerably higher than what even the more aggressive MFIs are presently charging. (Rosenberg, 2002, p. 10)

Toward a theory of fair interest rates on microcredit  87

We will return to the issue of just how much clients can pay in the next section. The second reason for starting with the procedural approach is that it also seems to underpin several initiatives stemming from the welfarist side, which concern transparency in interest rate setting. Several initiatives in the microfinance industry – such as the MIX Market database, MFTransparency, and the Smart Campaign – all include a call for increased transparency in pricing of microfinancial products and services. The most straightforward reason given for this is that MFIs target very poor clients who frequently are illiterate and therefore need further help to understand the contracts into which they enter. Of course, transparency is also important for other stakeholders such as donors and investors who evaluate the social and financial performance of MFIs, and regulators or competitors wanting fair competition (Augustine, 2012). Most of the initiatives in question focus on educating MFIs on how to calculate effective interest rates or the APR – as we have said, when comparing interest rates it is vital to not only include the basic interest rate charged on the loans but also the fees, mandatory savings, or additional burdens that clients cannot easily identify. The call for increased transparency implicitly accepts the procedural approach to fairness in interest rates since it focuses exactly on making sure that clients know what they are getting themselves into, and that they therefore accept the relevant contracts freely and voluntarily. As long as MFIs price their loans with full transparency, then, and clients keep coming back for loans, interest rates are fair. We think it is clear that the procedural approach here has some promising aspects, at least if it is supplemented with a sufficiently rich view of what it means for clients to consent to the contract terms. Many commentators have stressed the value, either inherent or instrumental, of more active participation by clients in the loan process (Guérin et al., 2015). If a central goal of development in general and microfinance in particular is the empowerment of the poor, one may hold participation in the loan process (or indeed in loan contract negotiations) as a kind of financial empowerment. Furthermore, such participation may have positive side effects both for the relevant MFIs and for society in general. To the extent that the procedural approach could be viewed as an ideal of increased borrower participation in the loan process, then, it is certainly on to something important. Moreover, given that loan contracts indeed represent the will of both MFIs and their clients, one would seem to have at least prima facie moral reasons to respect them. This is so because autonomy is inherently valuable and any kind of paternalism (acting in the interest of clients against their will) therefore requires further justification. This is an important moral principle that the procedural approach correctly highlights. In the end, however, we believe that the procedural approach has more problems than strengths, and calls for increased transparency do not take us far enough toward the goal of fairer interest rates on microloans. The most disturbing problem in this context is that it simply seems improbable that typical loan contracts in today’s industry represent borrowers’ genuine will. It should be noted that coercion can take several forms. The procedural approach correctly highlights that contracts would be unfair if MFIs used direct force or the threat of force, or if borrowers were deceived into contracts with opaque pricing procedures. Such types of coercion may not be typical in the industry and could decrease further with increased transparency in pricing. However, there are also more indirect and elusive forms of coercion – such as structural coercion – which may be more pervasive and problematic. One may safely assume, first of all, that borrowers are forced by their impoverished situation as such to do something, well anything, that can bring food to their table and pay for their

88  Handbook of microfinance, financial inclusion and development

housing. Secondly, one may assume that choice of activity is restricted due to a lack of acceptable alternatives – in many cases, the poor’s only alternatives may be to take out microloans or go to the local moneylenders (since they have no money and there is no social security). Now if they decide to go for the microloan, the unfortunate reality is that they often will have only one or two MFIs to choose from, and these will probably give the same deals anyway. This is so because the microfinance industry is far from competitive in most places and there is no external pressure on MFIs to adapt. Finally, if and when they approach their chosen MFI, the borrower has no collateral to bargain with and typically little knowledge of financial matters. For these reasons, borrowers will realistically have very little power in “negotiations” about loan contracts. We take the considerations above to suggest that the procedural approach, in a morally problematic way, sides too much with the needs of the institution rather than the needs of the clients. A fundamental idea of this chapter is that, if we are to settle the perennial debate between welfarists and institutionists, we must find a better balance between these two needs. One simply cannot take borrowers’ acceptance of the loan terms at face value as proponents of the institutionist side sometimes do, since the important question is not whether the poor pay back but why they do so.

THE COMPETITIVE MARKET APPROACH TO FAIRNESS Can the procedural approach somehow be revised to avoid the problems above? Judging from the debate above, it seems that the natural reaction in the present context – at least to most economists – is to characterize the problem as one primarily concerning market imperfections, that is, deviations from the theory of efficient markets. As noted above, in non-competitive markets the client may well decide to retake a loan even if the price is exorbitant, and they may also lack the bargaining power to influence the price or approach another lender. According to a second approach to fairness, which we have also taken from the business ethics literature, we should not be misled by such imperfections but instead identify fairness with what would have happened in a more competitive or efficient market. This is sometimes called the Paretian or “market failures” approach to business ethics (Heath, 2014; Norman, 2011), from which we derive the competitive market approach to fairness. Put simply, this approach holds that the fair interest rate is the one that would be agreed upon in a (sufficiently) competitive or efficient market. Such a market is typically characterized by the so-called “perfect market assumptions” of completely rational economic actors, no transaction costs (such as taxes or transportation costs in microfinance), and equal access to information by all market participants (no asymmetric information). This approach is also popular among proponents of the institutionist camp, although it often is used to point out problems with the route that the microfinance industry currently is taking. For instance, reflecting on their own part in the development which culminated with the Compartamos case in 2007, CGAP writes: [S]ince our founding in 1995, CGAP has been vocal about the need for interest rates that are high enough to cover costs, but we have been less emphatic about the loss to clients when interest rates are driven by inefficiency or exorbitant profits. We never made concrete predictions about how quickly competition would fix these problems, but we were probably too optimistic on this score. The Compartamos IPO gives all of us an opportunity to take another look at these questions. (Rosenberg, 2007, p. 15)

Toward a theory of fair interest rates on microcredit  89

The suggestion that increased competition is a key to solving the problem of exorbitant interest rates gains further support from Fernando (2006). He suggests that donors must focus on measures that will decrease interest rates in a sustainable manner, encourage the entry of various kinds of institutions, and stimulate more competitive markets. We take the competitive market view to be an improvement over the first and strictly procedural approach. In this second view, it is possible to ethically criticize at least the most aggressive MFIs such as Compartamos. What these companies are doing wrong is knowingly exploiting the fact that the market is less than perfect – that is, that there is too little competition in many places, and that borrowers are in dire need of funds. Furthermore, it may be noted that the competitive market approach can give practical advice on how one best curbs the problem with exorbitant interest rates. This is advice that should be fairly straightforward to anyone that is familiar with economics, namely to promote an enabling environment, encourage the entry of further agents, and control inflation. Recent evidence also confirms that competition decreases microcredit interest rates (Al-Azzam and Parmeter, 2019), even though the issue may be more complicated in segmented rural markets. However, we think that the competitive market approach also has important flaws. The first flaw is that it cannot guarantee that interest rates would be different in a more competitive market. Working out the most efficient rates is a hugely complex issue but, interestingly, some quick considerations indicate that they need not be low. Firstly, on the supply side, the administrative costs associated with extending a large number of small loans are immense (more on this below). Secondly, on the demand side, there are reasons to believe that at least some poor people can accept very high rates. According to what economists call the “law of diminishing marginal returns on capital”, relative returns on investments are higher the less capital you have to start with. What this means is that, to use an example, the tycoon who invests yet another £1,000 in his already proven company is likely to get much less out of this extra money than the temp who uses £1,000 to start his first micro-business. But since interest rates are exactly relative, this also means that the temp will be able to handle a much higher interest rate than the tycoon. Of course, the “law” above is only a generalization and may not be correct in all settings. There are a few empirical studies that identify settings relevant for MFIs (e.g. in agriculture) in which it seems incorrect (Armendáriz and Morduch, 2010). In any case, the two considerations above are fairly common arguments from the likes of CGAP in defense of today’s high interest rates on microloans. However, we suggest that they are problematic to the extent that great differences in the prices that the poor and the non-poor are offered, unless they are justified by sufficiently weighty moral reasons, feel inherently unfair to most people. We will return to this egalitarian intuition below. A second problem with the competitive market approach is the risk of negative side effects from increased competition in the microfinance industry. According to some empirical studies, a higher concentration of MFIs in a given region is associated with more aggressive marketing, which in some cases leads to borrowers taking out loans from multiple sources and thereby becoming overindebted. Furthermore, competition may lead to a focus on the “upper poor” (the relatively wealthier) instead of the “ultra-poor” (the relatively poorer) since it is cheaper to handle a smaller number of larger loans than a larger number of smaller ones (Tchakoute-Tchuigou and Soumaré, 2019). One may thus say that competition makes MFIs behave more like standard commercial banks than like development-oriented organizations.

90  Handbook of microfinance, financial inclusion and development

These considerations can be taken to exemplify a more general problem for the competitive market approach, namely that it tries to “smuggle in” concern for clients without really referring to their situation directly. Proponents of the present approach typically argue that subsidized and low interest rates would distort the market and hinder the entry of competitors. Instead of subsidizing loans and thereby supporting the customers directly, the emphasis is on building strong institutions and competitive markets that are thought to drive interest rates down indirectly. But one may wonder what happens if this is not the case. Our point is simply that lower rates at best are unintended by-products of institutionist thinking, where the overarching concern is the best interests of the institutions. In conclusion, then, also the competitive market approach fails to reach the kind of balance between the needs of clients and institutions that we are looking for. This approach still sides too much with the “institutionist” camp.

THE CREDIT-AS-A-RIGHT APPROACH TO FAIRNESS The considerations above suggest that we need to look in a completely different direction in order to find a more plausible theory of fairness in interest rates on microcredit. At this point, we turn to a perspective defended by Muhammad Yunus and others from the “welfarist” camp. Yunus (1998, 2007) talks about microcredit in general, and also interest rates in particular, in a language that is quite different from that of the economists described above. Yunus famously argues that access to credit should be considered a human right – that is, that it should be included in the United Nations’ Universal Declaration of Human Rights. The reason for this is that everyone should have a right to unleash their entrepreneurial capacities and strive for personal development through self-employment. According to Yunus, it is seldom the lack of entrepreneurial spirit that prevents poor people from lifting themselves out of poverty; more often it is the lack of the money to get things started. Yunus is not talking about just any kind of credit here, but affordable credit (Hudon, 2009). He has a rather straightforward classification system for what is affordable. Truly povertyfocused MFIs, he says, charge interest rates that fit into one of two zones: the Green Zone, which equals the cost of funds at the market plus up to 10 percent, and the Yellow Zone, which equals the cost of funds at the market rate plus 10 to 15 percent.

Beyond that, MFIs “that charge an interest rate higher than the Yellow Zone operate in the Red Zone, which is moneylenders’ territory” (2007, p. 68). If we put the two ideas together, then, Yunus is arguing that poor people have a human right to credit with a rate of interest that is less than 15 percent over and above the cost of funds (that is, the interest rate that the MFI itself gets on loans from governments or commercial banks). While the competitive market approach above was problematic since it could not guarantee sufficiently low interest rates for the poor, we see that offering low interest rates for the poor is exactly what Yunus’s approach sets out to do. According to what we may now call the creditas-a-right approach, poor people simply have a right to affordable credit – that is, they have a right to (have access to) credit with a price that is considerably lower than what many MFIs charge today. Corresponding to this right there are certain ethical obligations on the part of both lenders and governments. The basic idea is that MFIs act unethically if they fail to give

Toward a theory of fair interest rates on microcredit  91

clients a sufficiently low interest rate (because this violates clients’ rights) – so they are subject to a social responsibility to give affordable rates. Governments also have obligations that go in the same direction, such as an obligation to protect the poor from violations of their right to credit and (possibly) an obligation to create an environment where everyone has access to credit. Some recent developments on the political side may be understood along the lines of the credit-as-a-right approach. Helms and Reille (2004) list about 40 developing and transitional countries that had introduced various kinds of interest rate ceilings prior to 2004. Similar regulatory measures have become even more popular during the last decade due to the global financial crisis and an increasing fear of overindebtedness. For example, Bangladesh has decided to cap interest rates and a sub-committee of the Reserve Bank of India has suggested imposing a 24 percent cap on MFIs’ interest rates plus a 10 percent cap on their mark-up over and above the cost of funds. Other countries have simply asked their MFIs to “voluntarily” decrease their interest rates, threatening to revoke the MFIs’ licenses or even close down their branches if they fail to comply. The credit-as-a-right approach obviously represents a more egalitarian philosophy than the competitive market approach does, one which focuses on the legitimate claims of people at the lower end of society’s ladder. This is a clear advantage in the present context since, as we noted above, most readers will agree that the disparity in interest rates for the poor and the non-poor feels inherently unjust. Yunus formulates this egalitarian intuition nicely when he writes: “Isn’t it outrageous that low-income people who are struggling to make ends meet are the ones who have to pay the most for basic financial services – when they can get access to those services at all?” (2007, p. 50). Of course, the fact that many of us share this egalitarian intuition does not make it true in and of itself. However, we contend that the special context of microcredit makes it natural to attach much weight to egalitarian concerns. After all, the idea of microcredit in the first place is exactly to support borrowers that are poor or disadvantaged. The credit-as-a-right approach seems particularly attractive as a theory of how interest rates would be set in an ideally just society. However, it should be noted that the approach is not without problems. For instance, a number of commentators have argued against Yunus’s idea of credit as a human right (Gershman and Morduch, 2015; Sorell, 2015). A central argument in this literature is that access to credit at best can have instrumental value for borrowers – that is, what the poor really want is housing and clothing and such things and not access to credit per se. If this is the case, it seems strange to insist on a human right to credit per se. The relevant human rights must then be the rights to housing and clothing, and access to affordable credit should at best be considered a political instrument, which one may or may not use for securing those more fundamental rights. A different kind of criticism concerns the practicalities of Yunus’s idea. It is often argued that interest rate ceilings are a very rough political instrument and that they have a wide range of negative side effects. According to Fernando (2006), caps on microcredit interest rates will make it harder for MFIs to raise capital, both from their own operations as well as from external sources, which in turn leads to less willingness to expand and more focus on shortterm loans to the “upper poor”. Some commentators similarly argue that providing subsidized loans to the poor is an ineffective way of tackling the problem of exorbitant interest rates. According to Helms and Reille (2004), governments and the international community do better by focusing on improving the infrastructure (telecommunications, roads, education) of the countries where microfinance borrowers live. This will have more positive effects in terms of

92  Handbook of microfinance, financial inclusion and development

alleviating poverty, and at the same time lead to lower interest rates on microloans indirectly, since it will reduce certain key costs for MFIs. We believe that both arguments above should be taken seriously, but they need not be devastating for proponents of the credit-as-a-right approach. If talking about human rights is problematic one could instead talk a bit more loosely about legitimate justice-based claims. The poor have legitimate claims to affordable credit to the extent that it may help them out of poverty. Furthermore, there are several ways in which either the government or other parties could meet those claims, and proponents of the present approach need not insist on any more specific way – that is, they need not insist on interest rate caps or indeed any policy that targets interest rates directly. The point could simply be that some of these agents have moral responsibilities to do the things which most effectively protect or promote the legitimate claims of the poor. We will return to develop this point in the penultimate section. In the end, however, there is another problem with the credit-as-a-right approach which is more devastating. If the interest rate to which the poor have legitimate justice-based claims is very low, there is an apparent risk that not all relevant agents can afford to provide credit in sufficient quantities. That is, the institutions as such may not afford to give affordable rates. As we noted previously, a common argument from the institutionist side is that MFIs need to charge “cost-covering” rates in order to be financially self-sustainable – at least as long as there is no steady source of subsidies or grants to rely on (which is seldom the case). One may of course argue that governments and the international community have a responsibility to step in and secure the poor’s right to affordable credit if and when MFIs are unable to do so themselves for this reason. But it is wise to be pragmatic on some level and accept that both governments and MFIs may have burdensome costs and other obligations which prevent them from meeting the needs of the poor. In sum, then, it would seem like the credit-as-a-right approach actually goes too far in its support for the needs of the clients over the needs of the institutions. In this sense, also this approach fails to find the necessary balance. So are there any further possibilities?

THE CONSEQUENTIALIST APPROACH TO FAIRNESS The final approach that we will discuss here draws on the ethical theory of consequentialism (Driver, 2012; Sinnott-Armstrong, 2019). According to a consequentialist approach to microcredit, its ultimate or background justification must be its good effects or specifically the amelioration of the (financial) situation of the poor. All of the involved agents’ goals should therefore be to try to create as much of this good effect as possible. Decreasing interest rates may be one way of doing this since it rather straightforwardly lets the poor keep more of their money – which in turn may increase their chances of lifting themselves out of poverty. On the other hand, one must also be aware of how the relevant institutions work: as noted above, MFIs can seldom find stable sources of subsidies and need to cover their costs. Thus, giving away too much now may make it more difficult to give away anything later. On a consequentialist approach, the fair interest rate is here the one that weighs these concerns against each other and maximizes the overall utility – that is, the balance which makes the outcome the best. Few commentators in the microfinance literature defend this approach explicitly, or they at least seldom mention consequentialism as such. However, we believe that it is close to some central ideas that are going around. Formulations that go in this direction can actually be found on both sides of the debate on commercialization in microfinance. For instance,

Toward a theory of fair interest rates on microcredit  93

Elizabeth Rhyne, the former Vice-President of Accion (one of the largest microfinance NGOs in the US), defines fair pricing as that which “allows the institution to operate as a (on)going concern, but at the same time is as low cost to the customer as possible” (Accion, 2004). Furthermore, the Accion “consumer pledge” states that “interest rates will not provide excessive profits, but will be sufficient to ensure that the business can survive and grow to reach more people” (Accion, 2004). What seems to be sought in these quotes is exactly a balance between the needs of the poor and the need for MFIs to continue to exist. The statements above are obviously vague, and indeed the theory of consequentialism as such is infamously plastic. In order to avoid confusion, we wish to stress that our version of the approach cannot be used to defend just any kind of “balance” between the needs of clients and institutions. More specifically, the consequentialist approach can only justify setting interest rates so to cover what we may call necessary costs – that is, costs that are absolutely vital for the survival of the institution. It is not clear whether Accion’s appeal to “ongoing concern” refers just to such costs, or more broadly to all current costs which include the (unnecessary) costs associated with delivering attractive profits to commercial investors. In any case, the consequentialist approach seems to side with the competitive market approach here in forbidding MFIs to transfer such unnecessary costs onto the clients. What most straightforwardly sets the consequentialist approach apart compared to the previous approaches is that it so explicitly seeks to find a balance between the needs of the clients and the needs of the institutions. Compared to the credit-as-a-right approach, it certainly seems to be more realistic if we aim at the sustainable provision of financial services for the poor – as we argued above, both governments and MFIs may have burdensome costs and other obligations which sometimes prevent them from meeting the claims of the poor. MFIs are thus doing the right thing and should not be morally castigated for setting the interest rates wherever they set them as long as they give as much as possible back to their clients after having covered their necessary costs (Sandberg, 2012). The same could also be said for governments – they should not be criticized as long as they, to the best of their (current) ability, ameliorate the financial situation of the poor through either subsidies of interest rates or something else. However, the consequentialist approach also has problems that are troubling in the present context. First, it should be noted that we cannot be absolutely certain that utility is maximized if only MFIs avoid all unnecessary costs. Since consequences by definition are events in the future, and it is impossible to know with any certainty what the consequences of a certain policy or action will be, it would seem that it also is impossible to give concrete guidelines on how to find the utility-maximizing balance between the needs of clients and the needs of the institutions. This is a serious flaw of the consequentialist approach which many philosophers have noted (see Sinnott-Armstrong, 2019). We take this problem seriously and therefore admit that we cannot completely rule out various unexpected consequences and scenarios. A second and related problem is that we cannot guarantee that the utility-maximizing interest rate will be much lower than the levels noted and which may seem unfair to many observers. Interest rates may remain high even in this sort of idealized scenario, since the costs involved in providing a great number of microloans to poor people are very high. According to statistics from the MIX database, the operating expenses and cost of funds of MFIs drive interest rates up. Average profits even among sustainable MFIs represent about 14 percent of the interest yield, which means that even if an MFI were to eliminate all profits it would only be able to reduce its interest rate by roughly one-seventh (Rosenberg et al., 2009).

94  Handbook of microfinance, financial inclusion and development

A third problem is related to the fact that it may feel unfair that even some of the necessary costs (such as transaction costs) are transferred to and have to be borne by the poor clients. We believe that the competitive market approach, for example, is on to something when it suggests that costs arising from poorly functioning markets should not have to be borne by clients. Similarly, one may think that it is not the clients’ fault that they live in remote rural areas where communications are expensive and therefore also loan administration is costly. However, from the perspective of the consequentialist approach, such factors are unavoidable and will therefore determine the utility-maximizing interest rate. The problems above are important in the present context. Where do they then put us?

OUR OWN PROPOSAL: A COMBINATION OF RIGHTS AND CONSEQUENCES Our aim in this chapter has been to find theoretical input on how to set prices on microcredit in a more balanced way, i.e. to get closer to a theory of fairness in interest rates. The time has now come to present and elaborate on what we take to be the most plausible view in the present context. We have shown that there are serious flaws in both the procedural and competitive market approaches. We see more value and promise in the credit-as-a-right and consequentialist approaches. Our proposed “solution” is to try to combine these latter two perspectives. We acknowledge that combining views makes the philosophical foundation more complex, but we think that there is a place for balance in both theory and practice. The main problems afflicting the consequentialist approach above have to do with people’s egalitarian intuitions that focus on the plight of the poor. For this reason, we suggest that the approach is most plausible in combination with the credit-as-a-right approach. More specifically our suggestion is that the latter is the most plausible principle for how interest rates would be set in an ideally just society, or roughly what the overarching politico-philosophical goal of the industry should be. However, the consequentialist approach is a more plausible principle of the ethical responsibilities of (or criteria for moral praise and blame concerning) the agents involved in the contemporary microfinance industry. Let us now explain what this amounts to in practice. During the last decade, the microfinance sector has put more emphasis on building profitable institutions than on decreasing the final cost to the borrowers. However, recent events in Bangladesh, Nicaragua, and India where many branches of MFIs have been closed – as well as the launching of major campaigns like the Smart Campaign – have shown that the balance will have to evolve if microfinance wants to remain acceptable and sustainable. There are many things that contemporary MFIs themselves both can and should do in order to change this. Most importantly, they ought to cut down on unnecessary costs in order to avoid passing these on to their poor clients. Armendàriz and Morduch (2010) suggest that operational inefficiencies were a big part of why Compartamos charged such high interest rates: most strikingly, they had very high staff turnover. It is perhaps unrealistic to expect that MFIs can become as efficient and profitable as commercial banks, but this does not mean that there is nothing they can do to reduce their rates. A strategy often suggested by donors and practitioners is cross-subsidies. The goal of cross-subsidies is to reach out to wealthier clients (by extending relatively larger loans) and to combine such lending with loans to a larger number of poor clients whose average loan size is relatively small (Armendàriz and Szafarz, 2011). MFIs would be able to use the margin

Toward a theory of fair interest rates on microcredit  95

generated by their services to the relatively better-off clients to also serve the “ultra-poor”. In this context, MFIs could indeed charge lower interest rates on the range of their products targeting the ultra-poor. Alternatively, they could provide a standardized product in both urban and rural areas even if operating expenses would be higher in the rural ones. Cross-subsidies make sense in terms of social justice. Furthermore, they would make the relevant institutions less dependent on access to government grants or other kinds of subsidized funds for serving the most disadvantaged. However, cross-subsidization often comes with a practical dilemma for MFIs. On the one hand, the pricing, geographical targeting, and service delivery methodologies of MFIs often make their services less attractive to middleincome and wealthy individuals. As a result, the typical form of cross-subsidization would be between different groups that are already poor. On the other hand, targeting the non-poor to increase income invites the risk of “mission drift” which is a big debate topic in the microfinance literature (Mersland and Strøm, 2010; Armendariz and Szafarz, 2011; Serrano-Cinca and Gutiérrez-Nieto, 2014). Once they start serving a less poor clientele, MFIs may well be tempted to focus on this “easier” segment of the population. Moreover, the combination of a development logic related to poverty alleviation and a more commercially oriented logic of the less-poor is very difficult to manage in MFIs (Battilana and Dorado, 2010). In conclusion, we believe that cross-subsidies are part of the solution but must be managed with much precaution to avoid mission drift and internal conflicts. Besides cross-subsidization, operating expenses and inefficiency could also be decreased through technical or other innovations (Hartarska and Mersland, 2012). Mobile banking is a typical example of how technological innovation could decrease both management and infrastructure costs in very remote areas (Venet, 2019). Furthermore, the forming of partnerships with traditional banks or other NGOs could allow MFIs to benefit from the expertise of other kinds of institutions. The measures above should be able to reduce interest rates on microcredit to at least some extent and, as noted, we believe it is MFIs’ obligation to do this (along the lines of the consequentialist approach). However, there is most likely a limit to how much they will be able to reduce costs. Beyond this limit, we have argued that the responsibility more fruitfully should be located with governments and the international community (along the lines of the creditas-a-right approach). There are certainly many things which these agents can do to alleviate the situation of poor microcredit borrowers. A traditional strategy is to give subsidized credit lines. Many donors and socially responsible investors have already started to provide cheaper funds to MFIs. While we have noted the warnings raised about a risk of market distortion from such subsidies (Cull et al., 2018), they are still relatively easy and inexpensive ways to decrease total costs. Yet others have warned against a “crowding-out” effect. They argue that many of these subsidized funds are given to large and already profitable institutions that could well be financed by commercial players, while many smaller institutions would need these funds. Moreover, it is difficult to tell whether the institutions benefiting from these funds really end up decreasing their interest rates rather than increasing their profitability. Another example is guarantee funds, financed by donors, that could help MFIs to lower their cost of funds. These funds help creditworthy institutions to get access to alternative sources of funding while they lack real track records or the financial collateral requested by commercial lenders. Thanks to the guarantees, part of the default risk would be supported by the donor rather than the MFI. Finally, donors could make sure that MFIs benefit from modern management systems or digital techniques that help to decrease costs. As suggested above,

96  Handbook of microfinance, financial inclusion and development

we believe that donors have many instruments or policies to implement that could decrease interest rates paid by the clients of MFIs. While the institutions themselves obviously are at the front line and set the interest rates directly, donors could indirectly facilitate the decrease of rates.

CONCLUSION One of the most salient ethical debates concerning microcredit pertains to the unexpectedly high rates of interest charged on microloans. Microcredit is supposed to be to the advantage of borrowers in some of the poorest regions of the world, but at the same time, commercial institutions need to cover their comparably high costs. In this chapter, we have sought to get closer to a theory of fairness in interest rates when lending to the poor. We have presented four major theoretical perspectives on the issue, with inspiration taken both from the contemporary debates in the microfinance industry as well as more general philosophical debates on fairness. Our conclusion is that the procedural and competitive market approaches pinpoint some key concerns related to fairness but that they also have severe flaws. Our own proposal is ultimately to favor a combination of consequentialism and the credit-as-a-right approach. As long as MFIs are doing all they can to ameliorate their clients’ situation they are ethically in the clear. However, there is also a need to continue the political discussion about what interest rate levels the poor have legitimate justice-based claims to. Our goal here has mainly been to point out a direction for future thinking, and we do not claim to once and for all have settled the issue of fairness in interest rates on microloans. A lot more research is needed on all parts of the debate on interest rates in contemporary microfinance. More specifically, we hope that we have shown the great need for further theoretical research on the politico-philosophical perspective on microcredit, as well as further practical research on ways in which MFIs can cut unnecessary costs.

REFERENCES Accion (2004), Accion Consumer Pledge, Accion, Cambridge. Adams, D., Graham, D. and Von Pischke, J. (1984), Undermining Rural Development with Cheap Credit, Westview Press, Boulder. Adams, D. and Von Pischke, J. (1980), Fungibility and the design and evaluation of agricultural credit projects. American Journal of Agricultural Economics, 62(4): 719–726. Al-Azzam, M. and Parmeter, C. (2019), Competition and microcredit interest rates: international evidence. Empirical Economics. https://doi​.org​/10​.1007​/s00181​- 019​- 01766​- 6. Aristotle 1946 [350 BC]. The Politics of Aristotle, trans. E. Baker, Clarendon Press, Oxford. Armendariz, A. and Szafarz, A. (2011), On mission drift in microfinance institutions, in B. Armendariz and M. Labie (eds), The Handbook of Microfinance, Scientific Work, London, 341–366. Armendariz, B. and Morduch, J. (2010), The Economics of Microfinance, second edition, MIT Press, Cambridge. Augustine, D. (2012), Good practice in corporate governance: transparency, trust, and performance in the microfinance industry. Business & Society, 51: 659–676. Battilana, J. and Dorado, S. (2010). Building sustainable hybrid organizations: the case of commercial microfinance organizations. Academy of Management Journal, 53(6): 1419–1440. Collins, D., Morduch, J., Rutherford, S. and Ruthven, O. (2009). Portfolios of the Poor: How the World’s Poor Live on $2 a Day, Princeton University Press, Princeton. Conard, J. (1959), An Introduction to the Theory of Interest, University of California Press, Los Angeles. Consultative Group to Assist the Poorest [CGAP] (1996), Microcredit interest rates. Occasional Paper No. 1.

Toward a theory of fair interest rates on microcredit  97

Counts, A. (2008), Reimagining microfinance. Stanford Social Innovation Review, 8(1) : 45–53. Cull, R., Demirgüç-Kunt, A. and Morduch, J. (2018), The microfinance business model: enduring subsidy and modest profit. The World Bank Economic Review, 32(2): 221–244. de Mel, S., McKenzie, D. and Woodruff, C. (2008), Returns to capital in microenterprises: evidence from a field experiment. The Quarterly Journal of Economics, MIT Press, 123(4): 1329–1372. D’Espallier, B., Hudon, M. and Szafarz, A. (2017a), Aid volatility and social performance in microfinance. Nonprofit and Voluntary Sector Quarterly, 46(1): 116–140. D’Espallier, B., Goedecke, J., Hudon, M. and Mersland, R. (2017b), From NGOs to banks: does institutional transformation alter the business model of microfinance institutions? World Development, 89: 19–33. Driver, J. (2012), Consequentialism, Routledge, London and New York. Elegido, J. (2009), The just price: three insights from the Salamanca school. Journal of Business Ethics, 90(1): 29–46. Fernando, N. A. (2006). Understanding and Dealing with High Interest Rates on Microcredit, Asian Development Bank, Manila. Gershman, J. and Morduch, J. (2015), Credit is not a right, in Tom Sorell and Luis Cabrera (eds), Microfinance, Rights and Global Justice, Cambridge University Press, Cambridge, 14–26. Gielissen, R., Dutilh, C. and Graafland, J. (2008), Perceptions of price fairness: an empirical research. Business & Society, 47(3): 370–389. Guérin, I., Labie, M. and Servet, J. M. (2015), The Crises of Microcredit, University of Chicago Press, Chicago. Hartarska, V. and R. Mersland (2012), Which governance mechanisms promote efficiency in reaching poor clients? Evidence from rated microfinance institutions. European Financial Management, 18(2): 218–239. Heath, J. (2014), Morality, Competition, and the Firm: The Market Failures Approach to Business Ethics, Oxford University Press, Oxford. Helms, B. and Reille, X. (2004). Interest rate ceilings and microfinance: the story so far. CGAP Occasional Paper 9. Washington, DC, CGAP. Hermes, N. and Hudon, M. (2018), Determinants of the performance of microfinance institutions: a systematic review. Journal of Economic Surveys, 32(5): 1483–1513. Homer, S. and Sylla, R. (2005), History of Interest Rates, fourth edition, John Wiley & Sons, Hoboken, NJ. Hudon, M. (2007), Fair interest rates when lending to the poor. Ethique et économique/ Ethics and Economics, 5(1): 1–8. Hudon, M. (2009). Should access to credit be a right? Journal of Business Ethics, 84: 17–28. Hudon, M. and Ashta, A. (2013), Fairness and microcredit interest rates: from Rawlsian principles of justice to the distribution of the bargaining range. Business Ethics: A European Review, 22(3): 277–291. Hudon, M. and Sandberg, J. (2013), The ethical crisis in microfinance: issues, findings, and implications. Business Ethics Quarterly, 23(4): 561–589. Hudon, M., Labie, M. and Reichert, P. (2018), What is a fair level of profit for social enterprise? Insights from microfinance. Journal of Business Ethics, 162(3): 1–18. Hulme, D. and Mosley, P. (1996), Finance Against Poverty, Vol 1. Routledge, London. Huybrechts, B., Nicholls, A. and Edinger, K. (2017), Sacred alliance or pact with the devil? How and why social enterprises collaborate with mainstream businesses in the fair trade sector. Entrepreneurship & Regional Development, 29(7–8): 586–608. Johnson, E. (1938), Just price in an unjust world. International Journal of Ethics, 48(2): 165–181. Ledgerwood, J. and White, V. (2006), Transforming Microfinance Institutions: Providing Full Financial Services to the Poor, The World Bank, Washington, DC. Lensink, R., Mersland, R., Vu, N. T. H. and Zamore, S. (2018), Do microfinance institutions benefit from integrating financial and nonfinancial services? Applied Economics, 50(21): 2386–2401. McKenzie, D. and Woodruff, C. (2008), Experimental evidence on returns to capital and access to finance in Mexico? World Bank Economic Review, 22(3): 457–482. Mersland, R. and Strøm, T. (2010), Microfinance mission drift? World Development, 38(1): 28–36. Morduch, J. and Cull, R. (2017), Microfinance and economic development, in T. de Beck and R. Levine (eds), Handbook of Finance and Development, Edward Elgar, Cheltenham, 550–572.

98  Handbook of microfinance, financial inclusion and development

Norman, W. (2011), Business ethics as self-regulation. Journal of Business Ethics, 102(1): 43–57. Rosenberg, R. (2002), Microcredit interest rates. CGAP Occasional Paper 1. CGAP, Washington, DC. Rosenberg, R. (2007), CGAP Reflections on the Compartamos Initial Public Offering: A Case Study on Microfinance Interest Rates and Profits, CGAP, Washington, DC. Rosenberg, R., Gonzalez, A. and Narain, S. (2009), The new moneylenders: are the poor being exploited by high microcredit interest rates? CGAP Occasional Paper No. 15. Sandberg, J. (2012). Mega-interest on microcredit: are lenders exploiting the poor? Journal of Applied Philosophy, 29(3): 169–185. Serrano-Cinca, C. and Gutiérrez-Nieto, B. (2014), Microfinance, the long tail and mission drift. International Business Review, 23(1): 181–194. Sinnott-Armstrong, W. (2019), Consequentialism, in Edward N. Zalta (ed.), Stanford Encyclopedia of Philosophy (Summer 2019 Edition), accessible online at https://plato​.stanford​.edu​/archives​/sum2019​/ entries​/consequentialism/. Sorell, T. (2015), Is there a human right to microfinance? in Tom Sorell and Luis Cabrera (eds), Microfinance, Rights and Global Justice, Cambridge University Press, Cambridge, 27–46. Tchakoute Tchuigoua, H. (2016), Buffer capital in microfinance institutions. Journal of Business Research, 69(9): 3523–3537. Tchakoute-Tchuigoua, H. and Soumaré, I. (2019), The effect of loan approval decentralization on microfinance institutions’ outreach and loan portfolio quality. Journal of Business Research, 94: 1–17. Venet, B. (2019), Fintech and financial inclusion, in M. Hudon, M. Labie and A. Szafarz (eds), A Research Agenda for Financial Inclusion and Microfinance. Edward Elgar, Cheltenham, UK, 162–172. Von Pischke, J. D. (1983), The pitfalls of specialized farm credit institutions in low-income countries, in J. D. Von Pischke, D. Adams and G. Donald (eds), Rural Financial Markets in Developing Countries, World Bank Group, Washington, DC, 175–182. Woller, G. M., Dunford, C. and Woodworth, W. (1999) Where to microfinance? International Journal of Economic Development, 1(1): 29–64. Yunus, M. (1998), Banker to the Poor. The University Press Limited, Dhaka. Yunus, M. (2007). Creating a World without Poverty: Social Business and the Future of Capitalism, Public Affairs, New York.

6. Resilience in emergencies, savings, and credit Saniya Ansar, Jake Hess, and Leora Klapper*

ECONOMIC SHOCKS, COVID-19, AND RESILIENCE People face many financial risks, such as sudden illness, job losses, and natural disasters. The ability to manage financial risk is particularly important for people living in the poorest households. Their income often barely allows them to get by and any income shock may have severe consequences. A study found that, in the event of a loss of income, more than two-thirds of adults in Kenya, Vietnam, Greece, Chile, Colombia, and Bangladesh would not be able to cover basic needs for three months using just their savings or sales of assets (Gubbins 2020). Another study found that when the poor are affected by natural disasters the share of their wealth lost is two to three times that of the non-poor, largely because of the nature and vulnerability of their assets and livelihood (Hallegatte et al. 2017). Such income shocks force poor households to make choices that have detrimental long-term effects, such as withdrawing a child from school or cutting health care spending. COVID-19 has affected both high-income and developing economies and deepened the financial vulnerability of millions of people. A lack of employment-linked social insurance or social assistance programs like cash transfers has put poor households, the elderly, and informal workers at particular risk (Alfers and Moussié 2020). Households that would typically rely on informal jobs or gig-work, money from their social network, or money from informal savings may struggle to produce emergency funds during the crisis. Research shows that women are disproportionately affected during a crisis. They are more likely to lack access to identification, financial accounts, and mobile phones, which are needed to receive digital payments (Gelb and Mukherjee 2020a). In many fragile and conflict-affected economies, displacement and internal instability have burdened the already fragile health care systems. Having an effective resilience mechanism during such a time is crucial for governments and people. Are People Financially Resilient to Unexpected Expenses? Financial resilience is dependent on a variety of factors. Arguably the most important is the ability to meet ongoing financial obligations and consumption needs, absorb and recover from shocks, and fulfill longer-term goals (Rhyne 2020). For many people, even access to health care remains limited due to the inability to come up with funds in an emergency. People in developing economies are particularly vulnerable as they are far less likely than those in highincome economies to be able to come up with these funds (Figure 6.1). In developing economies on average, men are significantly more likely than women to say they would be able to come up with emergency funds; there is no such gender gap in high-income economies. In developing economies as well as high-income economies, financial resilience is higher among wealthier adults than poorer adults. 99

100

Figure 6.1 People in high-income economies are more likely to be able to raise emergency funds. Adults able to raise emergency funds (%), 2017

Source:  Global Findex database.

Note:  The 2017 Global Findex survey asked respondents whether or not it would be possible to come up with an amount equal to 1/20 of gross national income (GNI) per capita in local currency within the next month

Resilience in emergencies, savings, and credit  101

Globally, the ability to come up with emergency funds is not significantly correlated with an economy’s income level, implying that many economies with a high GDP could be just as vulnerable to a crisis as low-income economies. But a variety of structural and cultural factors beyond GDP might help explain cross-economy differences in financial resilience. A recent study found that inequalities within countries—rather than average conditions across countries—explain the dominant share of the global variation in financial resilience. These inequalities include local market conditions, policies or norms that shape the distribution of income and opportunities for adults with different socioeconomic backgrounds, and the financial behaviors and tools that adults use when managing their financial lives (Gubbins 2020). For example, the poor were significantly less likely to come up with emergency funds than the rich in both high-income and developing economies. Cultural differences might influence the type of emergency people imagine when they answer the surveys or whether people are inclined to say that it is possible to come up with emergency funds. What Are the Sources of Emergency Funds? Emergency funds can come from different sources. Often, people in high-income economies resort to using their savings to come up with emergency funds. In developing economies, money from working and money from family and friends are the main sources, but this usually comes with strings attached (Demirgüç-Kunt et  al. 2022). People who are already employed and in the workforce are also more likely to rely on making additional money from their jobs. Some people may also borrow from a bank, an employer, or a private lender in high-income economies. Relying on Savings in an Emergency Savings methods matter for financial resilience. People can save in different ways. One way to save is at a bank or another type of formal financial institution. An alternative to formal saving is a rotating savings and credit association (ROSCA). These associations generally operate by pooling weekly deposits and disbursing the entire amount to a different member each week. Other savings options may include saving in cash at home (“under the mattress”) or saving in the form of livestock, jewelry, or real estate. It may also include using investment products offered by equity and other traded markets or purchasing government securities. Readily accessible savings in any form can help people handle emergencies. But saving in less formal or liquid ways—such as through a savings club or in the form of livestock, jewelry, or real estate—may mean that the savings will not be readily accessible in an emergency. The savings club might have spent the money, and selling livestock, jewelry, or real estate quickly or without a loss might not be possible. And while saving cash at home may keep it readily accessible, saving money at a bank or another type of financial institution offers potential advantages. One is safety from theft. Another is that formal saving can curb impulse spending and therefore encourage better cash management, ensuring that money is available in an emergency. Money kept in a savings account might also earn interest income. A quarter of adults who reported that they can come up with emergency funds in developing economies said they would rely on savings as a source. But among those who cited savings as their main source of emergency funds, about half reported only saving in nonformal, less-safe ways. By contrast, in high-income economies, formal savings were used by three-quarters of those who said they would rely on savings to raise emergency funds.

102  Handbook of microfinance, financial inclusion and development

Financial Resilience and Employment Access to emergency funds is strongly influenced by employment. In high-income economies as well as low-income economies, adults active in the labor force were more likely to report being able to come up with emergency funds. For these adults, money from working was by far the most common source of emergency funds. Money from working can be defined in various ways. It could include working more hours at an existing job, seeking additional sources of employment, or getting a salary advance from an employer. But with heightened job instability in the era of COVID-19, this source of funding has become the most vulnerable. Relying on money from working could be interpreted as working more—by putting in more hours or seeking additional work. Other times, it could be interpreted as money from the regular salary a person receives for their labor. In this case, these funds could also be considered savings since any share of salary that is not spent is technically savings.

SAVINGS Having access to reliable savings can help people in emergencies. People save for a variety of reasons. On average, the most popular reason to save is for old age, followed by saving to start, operate, or expand a business. People might also save for education or large purchases, such as a home purchase. How People Save Saving at a formal institution is a popular choice in high-income economies (Figure 6.2). Adults in developing economies are more likely to rely on a common alternative, saving by using a savings club or a person outside the family. Such semi-formal savings tools are particularly widespread in sub-Saharan Africa. Savings patterns vary by demographics. In high-income economies, men and women are equally likely to use formal savings. But in developing economies, formal saving is more widespread among men than women. Poorer adults are less likely to use formal savings in high-income economies as well as developing economies. How Saving Impacts Financial Resilience Formal saving tools offer many advantages over the alternatives. They can provide safety against theft and curb impulse spending allowing for a better cash flow during an emergency (Dupas and Robinson 2013). Saving through an account can help achieve a range of development goals (for an overview see Karlan, Ratan, and Zinman 2014). In Nepal, female household heads provided with savings accounts were better able to cope with income shocks, reallocated their expenditures (more spending on education and food; less on health and dowries), and reported that their overall financial situation improved, though the study could not document statistically significant increases in savings compared to the control group (Prina 2015). Formal savings might have especially big benefits for women (Dupas and Robinson 2013). They can help improve a woman’s ability to save, invest in household durables relevant to herself, and empower her in household decision-making due to the greater control and restricted access savings accounts offer (Ashraf, Karlan, and Yin 2010; Karlan, Ratan, and Zinman 2014). All of this can provide women with the economic security needed when their household

103

Figure 6.2 Formal saving around the world. Adults saving at a financial institution in the past year (%), 2017.

Source:  Global Findex database.

104  Handbook of microfinance, financial inclusion and development

is faced with an emergency. Women in Chile who received free savings accounts reduced their reliance on debt and improved their ability to make ends meet during an economic emergency (Kast and Pomeranz 2014). Savings tools can lessen women’s reliance on dangerous coping methods. Women in Kenya who received mobile savings accounts were less likely to sell sex to raise money during a financial emergency (Jones and Gong 2019). Savings can also significantly reduce women’s vulnerability to health shocks (Dupas and Robinson 2013). A field experiment in Kenya tested four different technologies, including safe boxes and lock boxes kept at the individual’s home as well as accounts, both group- and individual-based, at a local savings and credit group. The effects were particularly large for savings accounts. Individuals with individual health savings accounts were 12 percentage points less likely to be unable to afford medical care, compared to 31 percent of individuals in the control group. COVID-19 influenced saving patterns in both developing and high-income economies. For example, in the euro area, household saving rates reached unprecedented levels in the first half of 2020 (Dossche and Zlatanos 2020). Lockdown measures discourage households from overspending. In parallel, the high unpredictability of future income and unemployment led to forced savings. By contrast, savings in developing economies have decreased. McKinsey’s Financial Insights Pulse Survey which was conducted in select economies finds that by June 2020, more than 70 percent of respondents in Indonesia and South Africa reported a decline in savings. In Mexico, India, the Russian Federation, and Turkey, this share exceeded 60 percent. With fewer employment opportunities present during lockdown, these changes mirrored reductions in household income. Saving for Old Age Many economies face a retirement crisis due to their aging populations. A lack of structured pensions and regular savings leaves many people potentially vulnerable to the financial challenges of old age. In some cases, savings or pensions may be insufficient as many adults are financially unprepared for the change in post-retirement payments compared with preretirement incomes. For example, adults in Germany report lower satisfaction with their current income when they retire. In parallel, an aging population implies that relatively fewer young people are generating economic revenues to finance social programs for the rest of the population. This makes older adults particularly vulnerable, and even more so during a health crisis, like the COVID-19 pandemic. In the United Kingdom, studies suggest that economic well-being decreases in the years leading up to and following retirement (Demirgüç-Kunt, Klapper, and Panos 2016). Despite this, relatively few adults save for old age. Older adults are more likely to save for old age, especially in their late 40s. But older adults in developing economies are less likely to use formal savings as they age. Instead, they rely on less reliable savings methods or might have no formal savings to begin with. While formal savings for old age may be a popular choice in high-income economies, the share declines among people entering their 60s as people begin to draw on their savings to meet their financial needs. A recent survey in the United States found that half of low- and moderate-income adults aged 50 or older had inadequate short-term savings, while nearly 40 percent struggled with excessive debt (Dunn, Andres, and Wilson 2019). Globally, the number of people aged 60 years or over has more than doubled since 1980, and the share of older adults is projected to double again by 2050. Even with these growing

Resilience in emergencies, savings, and credit  105

numbers, public safety nets, including pensions, fall short. For example, on average in the economies of partner countries to the Organization for Economic Cooperation and Development (OECD), pensions cover only 63 percent of a person’s pre-retirement income (OECD 2017). For these adults, public safety nets, including pensions, are vital for their financial wellbeing. Such programs help older manage their financial risk during an emergency. A study in the United States found that disability benefits for older adults ensured that people were 30 percent less likely to face bankruptcy, 30 percent less likely to deal with foreclosure, and 20 percent less likely to sell their homes to make ends meet during a financial emergency (Deshpande, Gross, and Su 2021).

CREDIT Financial resilience also encompasses access to credit from formal financial institutions that allows people to invest in educational and business opportunities, as well as the use of formal insurance products (see Box 6.1) that help people better manage their financial risk. A common reason for borrowing money is to make major purchases, such as land or a home. Others might borrow for health or medical purposes, to save, operate, or expand a business, or to pay for education. Different Ways People Borrow Sources of credit vary considerably around the world. In high-income economies, where financial institutions are well regulated, people are more likely to borrow formally—through a regulated financial institution such as a bank or credit union. Many people also rely on short-term credit such as through the use of a credit card. In fact, in high-income economies borrowing through the use of a credit card dominates formal borrowing. In developing economies, people are more likely to borrow from family and friends. In an emergency, this money could be either a gift or a loan. But in most cases, money received from family and friends in an emergency is likely based on reciprocity: if someone receives money from family or friends in an emergency they are likely expected to provide financial assistance if family or friends were to find themselves in a similar financial emergency. This provides an informal insurance mechanism, especially for the poor. This mechanism is well documented in the literature (Townsend 1994; Coate and Ravallion 1993; Fafchamps and Lund 2003). As such, money from family or friends in an emergency is not free money but likely comes with strings attached. How Credit Impacts Financial Resilience Access to credit can help people meet unexpected costs. For example, a one-time cash or credit subsidy to cover the cost of migration for work during the lean agricultural season succeeded in increasing seasonal, domestic migration among rural households, leading to improvements in household consumption and food security in Bangladesh (Bryan, Chowdhury, and Mobarak 2014). Alternative credit arrangements in Kenya which offered dairy farmers asset-collateralized loans, instead of requiring a sizeable initial deposit or guarantor, helped provide dairy farmers reliable and convenient access to water and improve their productivity (Jack et al. 2016). However, the evidence from the microfinance literature on the development impact credit access is mixed. Microloans can help skilled entrepreneurs with preexisting businesses get higher business assets and profits, and increase spending on durable non-business goods

106  Handbook of microfinance, financial inclusion and development

(Banerjee et al. 2019). Recent studies find that microloans can improve business outcomes for some entrepreneurs, while impacts on welfare or women’s empowerment are less strong (JPAL 2018). In India, a microcredit program that targeted women did not lead to new entrepreneurial activity and benefited only a handful of firms. But the study did find that microcredit had a small but positive impact on some broader social welfare indicators such as increases in female decision-making, happiness, and trust in each other and decreases in depression and need for aid (Banerjee et al. 2015). A study documents that expanding access to individual consumer loans in South Africa at high interest rates (200 percent APR) led to a clear increase in income (Karlan and Zinman 2010). Randomly assigned loans to marginally rejected loan applicants resulted in increases for constructed indices capturing consumption, economic self-sufficiency/maintaining employment, and optimism and perception of socioeconomic status although measures of stress and depression increased as well. Credit comes with potential downsides as well. A study of poverty transitions in three dozen villages in Andhra Pradesh, India, found that people often turn to expensive informal money lenders when faced with medical bills and that such high-interest debt is one of the factors most commonly associated with people becoming poor (Krishna 2006). COVID-19 has also influenced the ability of people to repay their loans. To discourage people from defaulting, regulators in many countries permitted, encouraged, or even mandated financial institutions to offer moratoria to borrowers. Almost 90 percent of the microfinance institutions participating in CGAP’s Pulse Survey reported using moratoria and other forms of rescheduling more than usual. Of those reporting the extent of their use, nearly a quarter had restructured more than 30 percent of their portfolios, mainly through moratoria (Rhyne and Dias 2020). But the nonregulated institutions to which lower-income people often turn have received little or no support. Early evidence from Pakistan shows that many microfinance loan officers continued to pressure borrowers for repayments, even after moratoria were declared (Malik et al. 2020). On the supply side, COVID-19 has reduced the availability of loans at a time when people may need them the most. For example in Kenya financial service providers rejected loans due to negative credit scores (Business Daily Africa 2020). By contrast, in Uganda, the government provided additional funding to SMEs during the early months of COVID-19 in an effort to increase household income (IMF 2020).

BOX 6.1: THE IMPORTANCE AND LIMITATIONS OF AGRICULTURAL INSURANCE Insurance products can help people manage financial risks stemming from large, unexpected expenses. Formal insurance products can pool risk over a large population, which offers households broader coverage than they would have if they relied on their own savings, credit, or community (Demirgüç-Kunt, Klapper, and Singer 2017). The ability to manage financial risks is especially crucial for farmers, who live at the whims of weather patterns, disease shocks, and market fluctuations. The 2017 Global Findex surveyed select adults in 15 economies in Sub-Saharan Africa about their use of agricultural insurance. The findings suggest that few adults who rely on agriculture

Resilience in emergencies, savings, and credit  107

for income are covered by insurance, and payouts for agricultural losses are rare. In the 15 surveyed economies in Sub-Saharan Africa, about one in three adults reported that their household relied on agriculture by growing crops or by raising livestock as the main source of their household’s income. Among adults living in agricultural households, the share with agricultural insurance is in the single digits on average. The low rates of insurance usage are not due to low risk. In fact, most adults living in agricultural households reported that their household experienced a bad harvest or significant livestock loss in the past five years. And most of the households facing agricultural loss bear the entire financial risk of such a loss, receiving no compensation through either an insurance payout or government assistance (Klapper, Ansar, and Hess 2021). Agricultural insurance coverage is limited and studies suggest that it suffers from low demand and takeup without significant subsidies. But research suggests that it can help embolden farmers to invest more in agricultural equipment and plant riskier but more lucrative crops. A study of index-based rainfall insurance in India found that insurance increased the cultivation of riskier rice varieties (Mobarak and Rosenzweig 2012). Similarly, randomized controlled trials on weather-based index insurance in India (Cole, Gine, and Vickery 2013) and Ghana (Karlan, Ratan, and Zinman 2014) showed that it encouraged farmers to adopt high-return, high-risk crops. In India, farmers who received free rainfall insurance against rainfall risk significantly increased production of cash crops which have higher expected returns but are more sensitive to rainfall. In Ghana, farmers who received free insurance invested more in cultivation and also shifted their mix of crops to riskier, more rain-sensitive crops. Insured farmers had higher total revenue and liquid post-harvest assets. In terms of welfare outcomes, the households of insured farmers were eight percentage points less likely to report missed meals but the study found no significant impact on select expenditures in other categories.

USING DIGITAL TECHNOLOGY TO EXPAND ACCESS TO FINANCE Digital financial services can help people to prepare better for financial challenges. The security of regularly scheduled transfers into interest-bearing accounts can ensure long-term savings possibilities. Defaults and automatic deductions can increase participation in savings plans to help people to save more money and reach their savings goals (Benartzi and Thaler 2007). For example, in India, researchers gave one group of people weekly payments in cash, and another group of people identical payments directly into savings accounts. The group that got the payments into accounts had significantly higher savings than the cash group (Somville and Vandewalle 2018). Similarly, in Afghanistan, workers who automatically deposited part of their salary into a mobile savings account had higher savings and financial security than workers who received a mobile savings account but did not enroll in automatic deposits (Blumenstock, Callen, and Ghani 2018). In the United States, setting automatic enrolment in 401(k) retirement savings plans as the default option led to a 50 percent increase in participation (Madrian and Shea 2001).

108  Handbook of microfinance, financial inclusion and development

Additional tools such as electronic savings reminders (“nudges”) can also help. In the Philippines, sending text messages to remind individuals of their savings goals increased savings amounts, making people more likely to reach their goals (Karlan et al. 2016). For individuals receiving digital payments, they can also be easier and less costly to navigate than cash. Governments typically implement their safety net programs through in-kind payments, food vouchers, or cash. Digital financial services can play an important role to help governments make such payments during an emergency. In Niger, government transfers through mobile phones cut administrative costs for the government by 20 percent compared to manual cash distribution (Aker et  al. 2016). In Mexico, the government’s shift to digital payments trimmed its spending on wages, pensions, and social welfare by 3.3 percent annually, or nearly US$1.3 billion (Babatz 2013). Shifting cash payments into accounts can also increase transparency and ensure that people receive wages even during emergencies where physical conditions do not allow easy movement. For example, the government of Sierra Leone was able to quickly pay thousands of Ebola workers, often working in rural areas, by opening accounts for them and making payments digitally (Bangura 2016). During COVID-19, the need for social distancing and lockdowns requires governments to explore digital frontiers. The use of digital payments is at the forefront of this change and can help with broader policy measures such as payment programs, pensions, and other safety nets. During a crisis, digitized payments can ensure efficiency by increasing the speed of payments and reducing the cost of disbursing and receiving them. Countries with stronger digital infrastructure, including ID and payment systems and social registers, have generally been able to implement and disburse emergency assistance programs more rapidly than those without these assets (Palacios 2020). India channeled payments to 200 million women holders of Jan Dhan bank accounts registered under the Aadhaar ID system, while Bangladesh announced a new program to provide wage support for workers in the garments sector who had lost their jobs. The private sector also helped increase payment efficiency. For example, Kenya’s Safaricom, the largest mobile money operator, waived fees for most transactions for three months to deal with the COVID-19 crisis (Gelb and Mukerjee 2020b). Credit products can also be better designed to help households mitigate risk. For example, text messages can also remind people of their financial obligations and increase loan repayment (Karlan, Morten, and Zinman 2015). Insurance products can be combined with monetary incentives and credit offerings to ensure the adoption of insurance against risks. Economic transactions significantly decrease during a crisis. Regulators can help facilitate transactions such as merchant payments or online purchases through digital platforms. In addition, shifting payments, especially regular bill payments, from cash into accounts can also help people build a payments data history which can then be leveraged for better credit access. Digital payments can also allow women to be more financially resilient by offering greater control over financial resources. Giving women more control over their money may also have larger societal and development benefits. A large body of research suggests that income in the hands of women, compared to men, is associated with larger improvements in children’s health and higher spending on health, housing, and nutritious food (Duflo 2012). In India, women who opened free bank accounts to receive wage payments from their jobs in a rural employment guarantee program increased their labor force participation in the public and private sectors, and in the long run, they helped liberalize gender norms. There was no such effect for a control group that received training as well as a bank account that was not

Resilience in emergencies, savings, and credit  109

linked to the employment program or their private-sector labor market (Field et al. 2019). When women in South Africa received a bank card linked to a government social benefits program, they increased their household bargaining power, and eventually, their labor force participation. The effects were driven by women who were previously unbanked, suggesting that their inclusion in the financial system is what drove the results (Biljon, Fintel, and Pasha 2018).

RISK AND CONSUMER PROTECTION Digital financial services can help people become more financially resilient. But this does not come without risk. Early research suggests that the rise of small loans delivered through mobile phones in East Africa has resulted in high rates of default and late payment, partly due to irresponsible lending and a lack of transparency (Izaguirre, Kaffenberger, and Mazer 2018). A mystery shopper audit of 1,000 microfinance firms in Uganda found that information on cost was inconsistent, inexperienced borrowers received less information than experienced borrowers, and printed materials with product specifications were often missing or in violation of guidelines (Atuhumuza et al. 2019). A study in Mexico argued that loan officers voluntarily provide little information to low-income clients and that clients are never offered the cheapest product that fits their needs (Gine, Cuellar, and Mazer 2017). Often the recipients of digitized government transfers are unaware of the process and find the financial products difficult to use. Many need help from the agent or a counterpart, thereby compromising the confidentiality of their account. Transfer recipients also struggle to get help through customer service when they have a question or a problem with their payments. Others have reported being targeted for fraud (Zimmerman and Baur 2016). Providing protections in these situations is especially important for women, the elderly, and low-income people, who are most likely to be financially inexperienced. Many poor women lack the skills they need to effectively use financial services. Traditional approaches to financial education—such as classroom-based training—have generally proved ineffective (see IPA 2017 for a discussion). Training might be more fruitful if provided during teachable moments when people have a specific reason for learning financial skills. This can be an effective strategy to connect real-life decision-making together with financial capability interventions and helps to ensure that information is more likely to be retained and used. For example, researchers provided financial training to migrants and their families. They found that the training had positive impacts on financial knowledge and behavior such as savings (Doi, McKenzie, and Zia 2014). In China, a financial capability program focusing on compound interest resulted in an increase of 40 percent in retirement savings by providing a short financial lesson to individuals immediately before being asked about the level of contribution they would like to make toward their pension plan (Song 2019). Among the elderly, low digital capability was highlighted as a leading cause of financial exclusion among older adults (de Soto 2000). A medical journal, The Lancet, found that the second-most prevalent form of exploitation of older adults was financial abuse, which impacts about one in seven older adults living in institutional settings, such as assisted living facilities (Yon et al. 2017). Digital tools could create new ways to financially exploit older adults. Traditional face-to-face banking might have security benefits that digital banking lacks. For example, bank tellers or financial advisers might be able to detect financial abuse during inperson visits with older clients. Detection could be more difficult if, say, fraudsters steal an

110  Handbook of microfinance, financial inclusion and development

older person’s personal information and use it to make transactions online or over the phone (Deane 2018). Such problems could increase as digital banking expands. Digital technology can also be used to identify and prevent the financial abuse of older adults through tools such as two-method verification, voice recognition, and facial recognition (Deane 2018). A range of new digital applications aims to help older adults and their families with financial management in older age. A study found that most dementia patients in the United States received financial management assistance, mostly from relatives, and that such help dramatically reduced dementia’s negative financial effects (Belbase, Sanzenbacher, and Walters 2018). A range of digital tools exists to help older people to navigate their financial challenges. For instance, some financial services apps are designed to help people to monitor their older parents’ financial activities, set up reminders and automatic bill payments, identify relevant government benefits, and alert families about potential scams. Others offer older adults specialized services in detecting and guarding against fraud and exploitation by monitoring credit reports and financial accounts for signs of suspicious or illegal activity. Research on the effectiveness of such apps is scarce; more studies are needed to understand if and how they can help to address the financial abuse of older adults. A strong consumer protection policy enabled through effective government regulation can help mitigate these risks. Targeted financial literacy and capability training can have a positive impact in areas such as increasing savings and promoting financial skills like record keeping (Miller et al. 2014). However, research has found that a wide approach that focuses on general financial education programs is less effective than enforcing effective disclosure and pricing transparency regimes associated with financial products (Gine, Cuellar, and Mazer 2017). Financial institutions can also provide quick, convenient, and effective redress mechanisms to respond to consumer concerns which can play a critical role in ensuring the sustainability of both new and old customers. Regulation does not have to stymie financial inclusion. Regulators can facilitate account access by introducing policies such as tiered documentation requirements, requiring banks to offer basic or low-fee accounts, and embracing opportunities to use new technologies to expand access to formal financial services.

ANALYZING DATA FOR FORMAL SAVINGS, FORMAL CREDIT, AND RESILIENCE Individual-level characteristics such as gender, age, education, and level of income can play an important role in determining access to savings. Additionally, there are a host of regulatory policies such as know-your-customer (KYC) requirements, low-fee accounts, and consumer protection frameworks that influence the use of accounts. Many national regulatory and supervisory agencies worldwide have taken steps toward financial inclusion by easing entry barriers to non-traditional financial service providers, increasing consumer protection standards, and improving financial literacy (World Bank 2013). In countries where regulatory quality is within the top quartile, individuals are 12.4 percent more likely to have an account at a financial institution compared to bottom-quartile countries (Chen and Divanbeigi 2019). To explore further, we update the analysis first completed by Allen et al. (2016), which used data from the 2011 Global Findex to study the individual and country characteristics associated with the use of formal accounts and which policies are effective (Klapper, Ansar, and Hess 2021). We assume that we only observe whether an individual uses a bank account to save if he or she owns an account. Estimating the use of accounts to save involves running a

Resilience in emergencies, savings, and credit  111

sample selection model. Because using an account to save is a binary variable, we use a twostage probit selection model where equation (1) defines the probit selection specification and equation (2) captures individuals’ decision to use their account to save. Among the individual-level characteristics, we include several socioeconomic variables from the Gallup World Poll such as gender, income quintiles, education level, employment status, marital status, urban classification, and log of household size. Not surprisingly, the data show that the likelihood of using an account to save is higher for adults who are in the upperincome quintiles, older, more educated, employed, married, and resident in urban areas. It is important to note that the cross-sectional nature of the data allows us to interpret these results only as significant correlations, not causal relationships. The disparity in the use of accounts across different economies suggests that the variance in country-level indicators is not determined only by individual characteristics. Hence, in our estimations, we consider a host of other country-level characteristics and policies as potential determinants of access to formal savings. Aside from controlling for individual-level variables, our estimations also consider a set of country-level characteristics and policies that might influence formal savings. We also look at policies that are targeted to promote inclusion— such as offering basic or low-fee accounts, granting exemptions from onerous documentation requirements, or making provisions for consumer protection. We use data from the 2017 World Bank Financial Inclusion and Consumer Protection Survey, which tracks supervisory activities and enforcement powers of the consumer protections agencies in a country. For our analysis, we use policy indicators such as the regulation on maximum maintenance fees for savings, tax incentives on saving schemes, KYC requirements, consumer protection monitoring index, and enforcement index (Table 6.1). Our analysis finds that the likelihood of formal savings is higher in countries where there are regulations on capping the maintenance fees for savings or current accounts in commercial banks. The likelihood of formal savings is also higher in countries that introduce special tax incentives for saving schemes. There is no relationship between regulations around KYC and formal savings, conditional on having an account. Strong and effective consumer protection (monitoring and enforcement index) is correlated with a higher probability to save, reinforcing the fact that consumer protection could be an important building block to build trust in financial institutions. Overall, these results underscore the importance of consumer-friendly regulations that increase access to basic financial services and ensure the safety of funds and reduce the risk of fraudulent activities.

CONCLUSION People around the world face a variety of financial risks from shocks such as health emergencies, loss of income, and weather-related events. The ability to manage financial risk is especially important for people living in the poorest households. Their income often barely allows them to get by and any income shock may have large and lasting effects on their livelihoods. The COVID-19 crisis has made millions of people more financially vulnerable. Access to and use of financial services can indeed build household resilience to financial vulnerability. Financial services play an especially important role in helping poor households adapt to and recover from shocks in a manner that protects their livelihoods and prevents them from resorting to coping strategies such as reducing spending on health or education. Accounts can help people save, possibly by helping people better resist impulse spending or demands on their

112  Handbook of microfinance, financial inclusion and development

Table 6.1  Relationship between formal savings and country characteristics Two-stage Heckman-probit model (outcome equations) (1) Variables

(2)

(3)

(4)

Formal savings (share of adults with an account)

Regulation on max maintenance fees for 0.133* savings/current accounts –0.079 Tax incentive savings schemes

0.175** (0.082)

KYC requirements index

–0.032 (0.087)

Consumer protection (enforcement and monitoring index)

0.391** (0.182)

Note:  Complete regression results are shown in Appendix 6.1. In the first stage, we regress account ownership (1/0) on each regulatory indicator separately. The second stage regresses formal savings (1/0). Data controls for individual characteristics and GDP per capita. Analysis includes data from 93 to 100 economies. Standard errors in parentheses (*** p < 0.01, ** p < 0.05, * p < 0.1). Standard errors are clustered at the country level. The share of adults who save formally but do not have an account is marginal at 0.65 percent. Since using an account to save is a binary variable to be estimated with a probit model, we cannot use Heckman’s (1979) two-step estimation procedure. The inverse Mills ratio, or Heckman’s lambda, only enters in the second step of this procedure in the case of a linear model; see Greene (2012, p. 880). Therefore, we jointly estimate the probit selection procedure and the probit model by maximum likelihood.

income from family and friends, can allow people to send or receive money in an emergency, and can help people build a relationship with financial institutions that they can leverage for better access to credit in case of an emergency. A sound legal and regulatory framework is crucial in providing the right conditions for financial resilience. An absence of such policies can increase financial burdens on consumers. For example, India simplified its account-opening process, allowing millions of previously unbanked adults to open a bank account. But a survey found that two-thirds of adults who applied for an account under the new system were asked to post a minimum balance even though the program had no such requirement, while nearly a fifth reported having to pay bribes (Demirgüç-Kunt et al. 2017). The success of financial services, especially digital payments, is largely dependent on the enforcement of appropriate regulations. Governments can play an important role in enforcing a sound regulatory framework that not only creates opportunities to make access to finance efficient and cost-effective but also ensures consumer protection and recourse.

NOTE *

We are grateful to Sakshi Hallan for her excellent assistance. This chapter’s findings, interpretations, and conclusions are entirely those of the authors and do not necessarily represent the views or policies of the World Bank or International Monetary Fund, their executive directors, or the countries they represent.

Resilience in emergencies, savings, and credit  113

REFERENCES Aker, Jenny C., Rachid Boumnijel, Amanda McClelland, and Niall Tierney. 2016. “Payment Mechanisms and Anti-Poverty Programs: Evidence from a Mobile Money Cash Transfer Experiment in Niger.” Economic Development and Cultural Change 65 (1): 1–37. Alfers, Laura and Rachel Moussié. 2020. “The ILO World Social Protection Report 2017–19: An Assessment.” Development and Change 51: 683–697. Allen, Franklin, Asli Demirgüç-Kunt, Leora Klapper, and Maria Soledad Martinez Peria. 2016. “The Foundations of Financial Inclusion: Understanding Ownership and Use of Formal Accounts.” Journal of Financial Intermediation 27: 1–30. Ashraf, Nava, Dean Karlan, and Wesley Yin. 2010. “Female Empowerment: Impact of a Commitment Savings Product in the Philippines.” World Development 38 (3): 333–344. Atuhumuza, Elly, X. Gine, Rafe Mazer, and Joeri Smits. 2019. “Mystery Shopping Assessment of Credit Cost Disclosures in Uganda.” Study Summary, Innovations for Poverty Action, New Haven, CT. Babatz, Guillermo. 2013. “Sustained Effort, Saving Billions: Lessons from the Mexican Government’s Shift to Electronic Payments.” Study Series: Mexico Study. Better Than Cash Alliance, New York. Banerjee, Abhijit, Emily Breza, Esther Duflo, and Cynthia Kinnan. 2019. “Can Microfinance Unlock a Poverty Trap for Some Entrepreneurs?” Working Paper 26346, National Bureau of Economic Research, Cambridge, MA. Banerjee, Abhijit, Esther Duflo, R. Glennerster, and C. Kinnan. 2015. “The Miracle of Microfinance? Evidence from a Randomized Evaluation.” American Economic Journal: Applied Economics 7 (1): 22–53. Bangura, Joe Abass. 2016. “Saving Money, Saving Lives: A Case Study on the Benefits of Digitizing Payments to Ebola Response Workers in Sierra Leone.” Better Than Cash Alliance, New York, NY. Belbase, Anek, Geoffrey T. Sanzenbacher, and Abigail Walters. 2018. “Do People with Dementia Get Help Managing their Money?” Working Paper, Center for Retirement Research, Boston College, Chestnut Hill, MA. Benartzi, Shlomo and Richard Thaler. 2007. “Heuristics and Biases in Retirement Savings Behavior.” Journal of Economic Perspectives 21 (3): 81–104. Biljon, C.V., D.V. Fintel, and A. Pasha. 2018. “Bargaining to Work: The Effect of Female Autonomy on Female Labour Supply.” Stellenbosch Economic Working Papers. Blumenstock, Joshua, Michael Callen, and Tarek Ghani. 2018. “Why do Defaults Affect Behavior? Experimental Evidence from Afghanistan.” American Economic Review 108 (10): 2868–2901. Bryan, Gharad, Shyamal Chowdhury, and Ahmed Mushfiq Mobarak. 2014. “Underinvestment in a Profitable Technology: The Case of Seasonal Migration in Bangladesh.” Econometrica 82 (5): 1671–1748. Business Daily Africa. 2020. “Banks Face Fines for Locking Out Blacklisted Borrowers.” Accessed January 28, 2021. Chen, Ron and Raian Divanbeigi. 2019. “Can Regulation Promote Financial Inclusion?” Policy Research Working Paper 8711, the World Bank, Washington, DC. Coate, Stephen and Martin Ravallion. 1993. “Reciprocity without Commitment: Characterization and Performance of Informal Insurance Arrangements.” Journal of Development Economics 40: 1–24. Cole, Shawn A., Xavier Gine, and James Vickery. 2013. “How Does Risk Management Influence Production Decisions? Evidence from a Field Experiment.” Policy Research Working Paper 6546, the World Bank, Washington, DC. Deane, Stephen. 2018. “Elder Financial Exploitation: Why it is a Concern, What Regulators are Doing about it, and Looking Ahead.” U.S. Security and Exchanges Commission, Office of the Investor Advocate, Washington, DC. Demirgüç-Kunt, Asli, Leora Klapper, and Dorothe Singer. 2017. “Financial Inclusion and Inclusive Growth: A Review of Recent Empirical Evidence.” Policy Research Working Paper 8040, the World Bank, Washington, DC. Demirgüç-Kunt, Asli, Leora Klapper, and Georgios A. Panos. 2016. “Saving for Old Age.” Policy Research Working Paper 7693, the World Bank, Washington, DC.

114  Handbook of microfinance, financial inclusion and development

Demirgüç-Kunt, Asli, Leora Klapper, Saniya Ansar, and Aditya Jagati. 2017. “Making It Easier to Apply for a Bank Account: A Study of the Indian Market.” Policy Research Working Paper 8205, the World Bank, Washington, DC. Demirgüç-Kunt, Asli, Leora Klapper, Dorothe Singer, and Saniya Ansar. 2022. Global Findex Database 2021: Financial Inclusion, Digital Payments, and Resilience in the Age of COVID-19. Washington, DC: World Bank. Deshpande, Manais, Tal Gross, and Yalun Su. 2021. “Disability and Distress: The Effect of Disability Programs on Financial Outcomes.” American Economic Journal: Applied Economics 13 (2): 151–178. de Soto, Hernando. 2000. The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else. New York: Random House. Doi, Yoko, David McKenzie, and Bilal Zia. 2014 “Who you Train Matters: Identifying Combined Effects of Financial Education on Migrant Households.” Journal of Development Economics 108: 39–55. Dossche, Maarten and Stylianos Zlatanos. 2020. “COVID-19 and the Increase in Household Savings: Precautionary or Forced?” Economic Bulletin Boxes, European Central Bank, vol. 6. Duflo, Esther. 2012. “Women Empowerment and Economic Development.” Journal of Economic Literature 50 (4): 1051–1079. Dunn, Andrew, Karen Andres, and Eric Wilson. 2019, “Redesigning the Financial Roadmap for the LMI 50+ Segment: New Challenges and Opportunities,” Financial Health Network, Washington, DC. Dupas, Pascaline and Jonathan Robinson. 2013. “Why Don’t the Poor Save More? Evidence from Health Savings Experiments.” American Economic Review 103: 1138–1171. Euart John, Nuno Ferreira, Jonathan Gordon, Ajay Gupta, Atakan Hilal, and Olivia White. 2020. Financial life during the COVID-19 Pandemic – An Update. New York: McKinsey and Company. Fafchamps, Marcel and Susan Lund. 2003. “Risk-Sharing Networks in Rural Philippines.” Journal of Development Economics 71: 261–287. Field, Erica, Rohini Pande, Natalia Rigol, Simone Schaner and Charity Troyer Moore. 2019. “On Her Own Account: How Strengthening Women’s Financial Control Impacts Labor Supply and Gender Norms.” Working Paper 26294, National Bureau of Economic Research, Cambridge, MA. Gelb, Alan and Anit Mukherjee. 2020a. “Digital Technology in Social Assistance Transfers for COVID-19 Relief: Lessons from Selected Cases.” CGD Policy Paper, 181, Center for Global Development, Washington, DC. Gelb, Alan and Anit Mukherjee. 2020b. “COVID-19: How Countries Can Use Digital Payments for Better, Quicker Cash Transfers.” CGD Policy Paper, 181, Center for Global Development, Washington, DC. Gine, Xavier, Martínez Cuellar, Cristina Mazer, and Rafael Keenan. 2017. “Information Disclosure and Demand Elasticity of Financial Products: Evidence from a Multi-Country Study.” Policy Research Working Paper 8210, the World Bank, Washington, DC. Gubbins, Paul. 2020. “The Prevalence and Drivers of Financial Resilience among Adults: Evidence from the Global Findex.” FSD Kenya, Nairobi. Hallegatte, Stephane, Adrien Vogt-Schilb, Mook Bangalore, and Julie Rozenberg. 2017. “Unbreakable: Building the Resilience of the Poor in the Face of Natural Disasters.” Climate Change and Development, the World Bank, Washington, DC. IMF. 2020. “Policy Responses to COVID-19. Policy Tracker: Uganda.” International Monetary Fund, Washington, DC. IPA. 2017. “Beyond the Classroom: Evidence on New Directions in Financial Education.” Financial Inclusion Program Brief, Innovations for Poverty Action, New Haven, CT. Izaguirre, Juan Carlos, Michelle Kaffenberger, and Rafe Mazer. 2018. “It’s Time to Slow Digital Credit’s Growth in East Africa.” Consultative Group to Assist the Poor (blog). Accessed October 8, 2020. Jack, William, Michael Kremer, Joost de Laat, and Tavneet Suri. 2016. “Borrowing Requirements, Credit Access, and Adverse Selection: Evidence from Kenya.” Working Paper 22686, National Bureau of Economic Research, Cambridge, MA. Jones, Kelly and Erick Gong. 2019. “Precautionary Savings and Shock-Coping Behaviors: The Effects of Promoting Mobile Bank Savings on Transactional Sex in Kenya.” Working Paper 6, American University, Washington, DC.

Resilience in emergencies, savings, and credit  115

JPAL. 2018. “Microcredit: Impacts and Limitations.” J-PAL Policy Insights, Abdul Latif Jameel Poverty Action Lab, Cambridge, MA. Karlan, Dean, Aishwarya Lakshmi Ratan, and Jonathan Zinman. 2014. “Savings by and for the Poor: A Research Review and Agenda.” The Review of Income and Wealth 60 (1): 36–78. Karlan, Dean, Margaret McConnell, Sendhil Mullainathan, and Jonathan Zinman. 2016. “Getting to the top of Mind: How Reminders Increase Saving.” Management Science 62 (12): 3393–3672. Karlan, Dean, Melanie Morten, and Jonathan Zinman. 2015. “A Personal Touch in Text Messaging Can Improve Microloan Repayment.” Behavioral Science and Policy 1 (2): 25–31. Karlan, Dean, Robert Osei, Isaac Osei-Akoto, and Christopher Udry. 2014. “Agricultural Decisions after Relaxing Credit and Risk Constraints.” The Quarterly Journal of Economics 129 (2): 597–652. Karlan, Dean and Jonathan Zinman. 2010. “Expanding Credit Access: Using Randomized Supply Decisions to Estimate the Impacts.” Review of Financial Studies 23 (1): 433–64. Kast, Felipe, and Dina Pomeranz. 2014. “Saving More to Borrow Less: Experimental Evidence from Chile.” Working Paper 14-001, Harvard Business School, Cambridge, MA. Klapper, Leora, Saniya Ansar, and Jake Hess. 2021. “Financial Inclusion and Financial Regulation.” Policy note. World Bank Global Findex database. Krishna, Anirudh. 2006. “Pathways out of and into Poverty in 36 Villages of Andhra Pradesh, India.” World Development 34 (2): 271–88. Madrian, Brigitte C. and Dennis F. Shea. 2001. “The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior.” The Quarterly Journal of Economics 116 (4): 1149–1187. Malik, Kashif, Muhammad Meki, Jonathan Morduch, Timothy Ogden, Simon R. Quinn, and Farah Said. 2020. “COVID-19 and the Future of Microfinance: Evidence and Insights from Pakistan.” New York University Wagner Research Paper. Miller, Margaret, Julia Reichelstein, Christian Salas, and Bilal Zia. 2014. “Can You Help Someone Become Financially Capable? A Meta-Analysis of the Literature.” Policy Research Working Paper 6745, the World Bank, Washington, DC. Mobarak, Ahmed Mushfiq and Mark Rosenzweig. 2012. “Selling Formal Insurance to the Informally Insured.” Working Paper, Yale University, New Haven, CT. OECD. 2017. “Pensions at a Glance 2017: OECD and G20 Indicators.” Organisation for Economic Co-operation and Development Publishing, Paris. Palacios, R. 2020. “Social Protection and Jobs Responses to COVID-19: A Real-Time Review of Country Measures.” The World Bank, Washington, DC. Prina, Silvia. 2015. “Banking the Poor via Savings Accounts: Evidence from a Field Experiment.” Journal of Development Economics 115(C): 16–31. Rhyne, Elizabeth. 2020. “Measuring Financial Health: What Policymakers Need to Know.” Insight2impact. Rhyne, Elizabeth and Denise Dias. 2020. “Moratoria during COVID-19: How Are They Working Out?” Microfinance and COVID-19: Insights from CGAP’s Global Pulse Survey (blog), Consultative Group to Assist the Poor, Washington, DC. Somville, Vincent and Lore Vandewalle. 2018. “Saving by Default: Evidence from a Field Experiment in Rural India.” American Economic Journal: Applied Economics 10 (3): 39–66. Song, Changcheng. 2019. “Financial Illiteracy and Pension Contributions: A Field Experiment on Compound Interest in China.” The Review of Financial Studies 33 (2): 916–949. Townsend, Robert M. 1994. “Risk and Insurance in Village India.” Econometrica 62 (3): 539–591. World Bank. 2013. “Global Financial Development Report 2013: Rethinking the Role of the State in Finance.” The World Bank, Washington, DC. Yon, Yongjie, Christopher R. Mikton, Zachary D. Gassoumis, and Kathleen H. Wilber. 2017. “Elder Abuse Prevalence in Community Settings: A Systematic Review and Meta-Analysis.” The Lancet Global Health 5 (2), doi:10.1016/S2214-109X(17)30006-2. Zimmerman, Jamie M. and Silvia Baur. 2016. “Understanding How Consumer Risks in Digital Social Payments Can Erode Their Financial Inclusion Potential.” CGAP Brief, Consultative Group to Assist the Poor, Washington, DC.

116

Income: Middle 20% (0/1)

Income: Second 20% (0/1)

Income: Poorest 20% (0/1)

Female (0/1)

Consumer protection (enforcement and monitoring index)

KYC exception (0/1)

–0.417*** (0.025)

(0.030)

(0.026)

(0.038)

–0.317***

–0.618***

(0.038)

(0.049)

–0.425***

–0.776***

(0.017)

(0.022)

–0.520***

–0.027

–0.057**

—

—

— —

—

—

—

—

—

—

(0.078) —

—

–0.136

Tax incentive savings schemes (0/1)

0.133*

–0.075

Max maintenance fees for savings/ current accounts (0/1)

Formal savings (of adults with an account)

Account ownership (of adults)

(2)

Variables

(1)

(0.030)

–0.319***

(0.038)

–0.428***

(0.049)

–0.523***

(0.023)

–0.057**

—

—

—

—

(0.124)

0.12

—

—

Account ownership (of adults)

(3)

(0.025)

–0.419***

(0.026)

–0.620***

(0.038)

–0.779***

(0.017)

–0.031*

—

—

—

—

(0.082)

0.175**

—

—

Formal savings (of adults with an account)

(4)

(0.031)

–0.315***

(0.038)

–0.419***

(0.050)

–0.512***

(0.023)

–0.057**

—

—

(0.125)

–0.037

—

—

—

—

Account ownership (of adults)

(5)

(0.025)

–0.412***

(0.027)

–0.613***

(0.039)

–0.768***

(0.017)

–0.028

—

—

(0.087)

–0.032

—

—

—

—

Formal savings (of adults with an account)

(6)

(0.031)

–0.310***

(0.038)

–0.422***

(0.049)

–0.514***

(0.024)

–0.057**

(0.283)

0.603**

—

—

—

—

—

—

Account ownership (of adults)

(7)

(0.026)

–0.419***

(0.027)

–0.623***

(0.039)

–0.778***

(0.018)

–0.032*

(0.182)

0.391**

—

—

—

—

—

—

Formal savings (of adults with an account)

(8)

APPENDIX 6.1: RELATIONSHIP BETWEEN HAVING A BANK ACCOUNT AND FCP REGULATORY VARIABLES

117

Out of workforce (0/1)

Unemployed (0/1)

Employed for employer (0/1)

Divorced (0/1)

Married (0/1)

Log of HH size

Education: Primary level or less (0/1)

Rural (0/1)

Age squared

Age

Income: Fourth 20% (0/1)

–0.356*** (0.041)

(0.064)

(0.040)

(0.050)

–0.270***

–0.427***

(0.032)

–0.138***

(0.046)

(0.037) 0.099***

(0.039)

0.398***

–0.082**

(0.041)

0.045

0.115***

(0.053)

(0.034)

0.132**

0.080**

(0.039)

(0.044)

(0.058)

–0.039

–0.321***

(0.034)

(0.039)

–0.503***

0.087**

0.125***

(0.000)

(0.000)

(0.003) –0.000

–0.000***

(0.004)

(0.024) 0.006*

(0.024)

0.023***

–0.284***

–0.198***

(0.064)

–0.269***

(0.051)

–0.138***

(0.047)

0.399***

(0.038)

0.043

(0.054)

0.131**

(0.039)

–0.034

(0.057)

–0.501***

(0.039)

0.123***

(0.000)

–0.000***

(0.004)

0.023***

(0.024)

–0.199***

(0.041)

–0.353***

(0.040)

–0.423***

(0.032)

0.102***

(0.036)

–0.079**

(0.042)

0.112***

(0.034)

0.082**

(0.044)

–0.324***

(0.033)

0.088**

(0.000)

–0.000

(0.003)

0.006*

(0.024)

–0.284***

(0.063)

–0.265***

(0.051)

–0.140***

(0.046)

0.404***

(0.039)

0.039

(0.054)

0.130**

(0.039)

–0.042

(0.057)

–0.503***

(0.041)

0.124***

(0.000)

–0.000***

(0.004)

0.022***

(0.024)

–0.193***

(0.041)

–0.355***

(0.041)

–0.425***

(0.033)

0.109***

(0.038)

–0.089**

(0.041)

0.113***

(0.034)

0.074**

(0.044)

–0.329***

(0.036)

0.091**

(0.000)

–0.000

(0.003)

0.006*

(0.024)

–0.279***

(0.062)

–0.285***

(0.055)

–0.157***

(0.046)

0.396***

(0.039)

0.038

(0.055)

0.131**

(0.041)

–0.040

(0.057)

–0.481***

(0.041)

0.099**

(0.000)

–0.000***

(0.004)

0.024***

(0.025)

–0.192***

(Continued)

(0.040)

–0.367***

(0.040)

–0.443***

(0.031)

0.097***

(0.038)

–0.083**

(0.042)

0.116***

(0.035)

0.079**

(0.044)

–0.322***

(0.035)

0.082**

(0.000)

–0.000

(0.003)

0.006*

(0.025)

–0.285***

118

70,629

(3)

109,585

100

(0.445)

–3.583***

(0.250)

1.306***

(0.048)

0.430***

Account ownership (of adults)

(4)

70,269

100

(0.333)

–4.057***

—

—

(0.033)

0.394***

Formal savings (of adults with an account)

(5)

106,611

97

(0.447)

–3.524***

(0.275)

1.372***

(0.050)

0.431***

Account ownership (of adults)

(6)

67,592

97

(0.335)

–4.072***

—

—

(0.035)

0.402***

Formal savings (of adults with an account)

(7)

104,571

95

(0.428)

–3.771***

(0.381)

1.439***

(0.050)

0.413***

Account ownership (of adults)

67,833

95

(0.343)

-4.252***

—

—

(0.036)

0.393***

Formal savings (of adults with an account)

(8)

*** p

q . (12.3) a

Thus, if the insurance is actuarially fair or carries a positive load, q ³ 0, a necessary, but not sufficient, condition for the farmer to purchase index insurance is that the payout be positively correlated with losses. Furthermore, the greater the load, the greater the correlation must be for the farmer to purchase insurance. Equation (12.2) reveals that if the insurance carries a positive load, q > 0, the optimal amount of index insurance q* purchased by the farmer, if positive, increases with the covariance Cov(l, p ) between losses and payout, increases with the farmer’s risk aversion α and decreases with the premium load θ . However, if the insurance is subsidized and carries a negative load, q < 0, we obtain the perverse results that the farmer might purchase index insurance even though his losses are negatively correlated with the payouts and the amount of insurance purchased decreases with the farmer’s risk aversion α.

CONTRACT DESIGN The previous section revealed that the potential value of index insurance to farmers depends on the correlation between the farmer’s losses and the contract payouts, which in turn depends on the choice of index z and the payout schedule p( z ). In practice, identifying an index and structuring a payout schedule so as to maximize the loss–payout correlation presents many challenges. In particular, the number of indices available for use in index insurance contract designs is limited and the payout schedule must be kept relatively simple so that the contract terms are transparent to the farmer, insurer, regulator, and policymaker. Tradeoffs between contract complexity and basis risk are inevitable (International Fund for Agricultural Development, 2015; Mahul, 1999). Choosing the contract’s payout schedule to maximize the correlation between losses and payouts is not a straightforward exercise, given that the relation between crop yields and rainfall can be complicated and naturally varies across farms. A crop’s watering needs vary across phenological phases of development, both in the amount of water needed and the regularity with which it is needed (Dalhaus, Musshoff, and Finger, 2018). For example, during the tasseling phase, maize requires sustained regular rainfall more than large amounts of rainfall. However, during the flowering stage, maize crop development depends less on the regularity of rainfall than on the total rainfall that saturates the ground. Moreover, the planting dates for crops can vary from year to year, depending on weather conditions around the traditional start of the growing season. Not surprisingly, there is evidence that farmers prefer contracts that provide multiple and more frequent payouts, for example, contracts that insure monthly deficits. However, although more flexible contracts can provide better risk protection, farmers usually trust them less than simple ones (Hill, Hoddinott, and Kumar, 2013). Efforts to reduce basis risk by developing contract designs based on watering needs have led to more complex designs. For example, the maize drought insurance contract offered by the Ghanaian Agricultural Insurance Pool allows for dynamic start dates and provides three distinct payouts, one for each of the three main phenological phases, with one of the payouts a function of the number of consecutive days without rainfall and one of the payouts a function of total rainfall during the entire phase (Stutley, 2010).

Index insurance for developing countries  199

Yet another challenge to designing an effective index insurance contract is that the correlation between losses and payouts varies across farmers. For example, for a drought insurance contract with payouts based on rainfall measured at a specified meteorological station, the correlation between losses and payouts typically declines with the distance between a farmer’s plot and the station. Given that official weather stations in rural areas of developing countries are sparsely distributed, index insurance contracts whose payouts are tied to rainfall at weather stations may be useful only to very few farmers who live near the station. This problem is referred to by many as “spatial adverse selection”. Spatial adverse selection has been addressed by designing index insurance contracts that employ vegetation and rainfall-proxy indices computed from remotely sensed (i.e., satellite) observations, which allow a greater variety of contracts to be offered at finer geographical resolutions. However, remotely sensed data introduces additional basis risk because remote observations do not generally correlate perfectly with conditions on the ground. Yet other indices that in principle embody less basis risk have been employed, including area-yield and area-revenue indices, which base payouts on average per-hectare yields and revenues in a circumscribed area. Such contracts would cover a wider range of sources of systemic losses, such as losses from pestilence, disease, floods, or low prices. However, compilation of area yields and area revenues takes time and thus delays payments past the time when they are most badly needed. Finally, although index insurance avoids many of the costs associated with indemnity insurance (e.g., farm-specific rating and loss adjustment), it is nonetheless subject to transaction costs that must be incorporated into premiums or otherwise covered by subsidies. These costs include administrative costs of marketing and settling insurance contracts (Miranda and Farrin, 2012).

INDEX INSURANCE VERSUS SELF INSURANCE Much of the theoretical and applied research on the demand for index insurance is based on static expected utility models such as the one presented in the fourth section. A shortcoming of this approach is that a static model is inherently incapable of capturing the most natural substitute for index insurance: self-insurance through the use of savings. Failing to account for risk management alternatives to index insurance leads to an overestimate of the value of index insurance to the farmer. To illustrate this, we now develop and analyze a dynamic model that allows the farmer to save. Consider an infinitely lived farmer who employs a subsistence rain-fed agricultural production technology. Each period, the farmer receives an income that depends on whether a drought has occurred and an exogenous idiosyncratic random shock e that is independent of drought conditions. Specifically, the farmer receives income m1e if drought occurs and income m0e otherwise, where m0 > m1 > 0 and e is a positive serially i.i.d. random variable with mean 1. Each period, drought occurs with probability q, independently of past occurrences. Available to the farmer is a standardized drought insurance contract that pays 1 in the event of drought but 0 otherwise. The farmer may purchase as many contracts or fractional part thereof as he wishes up to his expected income shortfall from a drought, x º m0 - m1. Each contract commands a premium p = (1 + q )q where θ is the premium load. The farmer begins each period endowed with predetermined wealth w, which he then allocates among consumption, savings, and purchases of drought insurance. The farmer chooses how much to save s ≥ 0 and how many drought insurance contracts to purchase x ≥ 0 so as to

200  Handbook of microfinance, financial inclusion and development

maximize the sum of current and discounted expected future utility of consumption over an infinite horizon. By Bellman’s Principle of Optimality, the farmer’s decision problem is characterized by the Bellman equation

V ( w) = max {u ( w - s - p x) + d s ³ 0, x ³ x ³ 0

åq E V (m e + s + xi)} (12.4) i

e

i

i =0,1

where V ( w) which denotes the maximum attainable sum of current and discounted expected future utility of consumption given the farmer’s current wealth w. Here, δ is the farmer’s subjective discount rate, q1 º q , q0 º 1 - q , and u is the farmer’s utility of consumption function, which is assumed to be strictly increasing and strictly concave. Under mild regularity conditions, the farmer’s Bellman equation possesses a well-defined solution. However, neither the value function V(w) nor the attendant optimal insurance coverage policy x(w) are expressible in closed form. These functions, however, can be approximated to an arbitrary precision numerically using collocation methods (Miranda and Fackler, 2002). In order to render the model numerically soluble, we assume that: without loss of generality, the farmer’s expected income equals 1; the coefficient of variation of income is 30 percent; the probability of drought is 20 percent; the idiosyncratic shock is lognormally distributed and accounts for 50 percent of the variance in log income; savings deposits earn 5 percent per period; the farmer discounts each period by a factor of 0.9; the farmer exhibits constant relative risk aversion 2; and the index insurance contract is actuarially fair and carries no load or embody a subsidy. Figure 12.1 illustrates the farmer’s optimal index insurance coverage as a function of his wealth, with and without the ability to self-insure through saving. As seen in the figure, a farmer with a very low wealth will not purchase index insurance because the cost of forgone consumption from purchasing the insurance in the current period exceeds its expected future benefits. This is true whether or not he has the ability to self-insure. However, at levels of wealth above w1, it becomes optimal for the farmer to purchase index insurance and, initially, in increasing quantities as his wealth increases. If the farmer is unable to save, as assumed in

Figure 12.1 Optimal index insurance coverage as a function of wealth with and without savings

Index insurance for developing countries  201

all static models, the farmer buys greater quantities of index insurance as his wealth increases, eventually reaching the maximum allowable. However, if the farmer can self-insure by saving, a very different picture emerges. Once the farmer’s wealth reaches and begins to exceed w2, the farmer increasingly favors self-insuring through increased savings and reduces the amount of index insurance he buys. Once the farmer’s wealth reaches and exceeds w3, he no longer demands index insurance. Thus, the ability to self-insure through saving generally reduces the demand for index insurance, something that is not revealed in static models of index insurance demand. In short, poor farmers cannot afford index insurance, and, given self-insurance alternatives, rich farmers do not need it. Figure 12.2(a) illustrates the farmer’s optimal index insurance coverage as a function of his wealth, given the ability to self-insure, for different levels of basis risk. Basis risk is

Figure 12.2 Optimal index insurance coverage as a function of wealth for different levels of systemic risk and different premium loads, when self-insurance is possible

202  Handbook of microfinance, financial inclusion and development

operationally defined as the proportion η of the variance of income attributable to the idiosyncratic shock, or equivalently, one less the correlation between the farmer’s income and the index insurance payout. This proportion is varied, with η = 0.2 representing low basis risk, η = 0.4 representing moderate basis risk, and η = 0.6 representing high basis risk. As seen in the figure, demand for index insurance at all levels of wealth falls as basis risk increases. However, the impact of basis risk when self-insurance is possible is relatively modest, and the basic results hold. Figure 12.2(b) illustrates the farmer’s optimal index insurance coverage as a function of his wealth, given the ability to self-insure, for different premium loads θ , with θ = −0.7 representing massively subsidized insurance, θ = 0 representing actuarially fair insurance, and θ = 0.2 representing insurance a more commercially realistic premium load. As seen in the figure, demand for index insurance at all levels of wealth rises as the premium load falls. With sufficient subsidies, all but the very poor will purchase maximum coverage. These results suggest that subsidies benefit wealthy farmers more than poor ones. The results of this section illustrate how most models of index insurance used in applied work, which generally rely on static expected utility theory, can overstate the benefits of index insurance to individual farmers and fail to capture its limitations. The results provide some possible explanations as to why virtually all index insurance pilot programs aimed at smallholder farmers in developing countries have proven unsustainable unless heavily subsidized. The results also bring into question whether the observed uptake of index insurance products, where it occurs, is attributable primarily, if not exclusively, to the subsidies they embody, rather than their risk reduction benefits.

INDEX-INSURED CREDIT Given the challenges of designing index insurance contracts to manage farmers’ farm-level risk directly, economists have explored other ways to use index insurance to benefit farmers, most notably using it to support their access to affordable agricultural credit (Miranda and Gonzalez-Vega, 2011; Farrin and Miranda, 2015; Carter, Elabed, and Serfilippi, 2015). The basic idea is to “bundle” index insurance with loans, effectively making the purchase of insurance mandatory for anyone acquiring a loan. Making the purchase of index insurance mandatory for securing a loan at first may appear not to benefit the farmer, given that in most cases the farmer would not otherwise be willing to pay for it, even if modestly subsidized. However, bundling index insurance with loans has the potential to reduce widespread loan repayment defaults in the event of a drought, protecting the lender against the threat of bankruptcy. This potentially would allow lenders to offer a greater volume of loans to poor farmers at lower interest rates, thus yielding indirect but nonetheless tangible benefits to them. To explore the impact of bundling index insurance with credit on the supply of credit let us consider again the farmer introduced in the preceding section. Suppose now that a lender offers the farmer an in-kind loan in the form of “hi-tech” seed that raises the farmer’s income by a multiplicative factor γ > 1 in all states of nature in the following period. Bundled with the loan is an index insurance contract which, in the event of a drought, provides a payout equal to the debt obligation L. For simplicity, assume the farmer cannot save and maximizes the expected lifetime utility of income. The farmer, however, may or may not repay his loan the following period. If he defaults, however, he is permanently banned from future credit and additionally suffers a nonpecuniary

Index insurance for developing countries  203

utility penalty f ³ 0 due to shame, moral regret, and/or decline in commercial and social standing. We refer to f simply as the farmer’s creditworthiness, noting that, other things being equal, the greater the farmer’s creditworthiness, the less inclined he is to default on his loan. Let V ( w) denote the maximum attainable present value of current and expected future utility of consumption for an indebted farmer who begins the period with liquid wealth w. Then, by Bellman’s Principle of Optimality (Bellman, 1957), for all levels of wealth w, the farmer’s value function V must satisfy the Bellman functional fixed-point equation

V ( w) = max{u ( w) + A - f , u ( w - L) + C} (12.5)

where C = d EV ( y ) is the present value of expected future utility for a farmer with a loan, A is the present value of expected future utility for a farmer who has been permanently banned from access to credit, and

ìïgm e y = í 0 îïgm1e + L

with probability 1 - q (12.6) with probability q

is the following period’s random income for a farmer with an insured loan. Bellman equation (12.5) reveals that a farmer with an outstanding loan obligation possessing liquid wealth w chooses between: (1) defaulting on his loan obligation and accepting a value equal to the current utility derived from consuming his liquid wealth u ( w), plus the future utility expected if permanently banned from credit A, less the non-pecuniary utility penalty incurred from defaulting f ; or, (2) repaying his loan and taking out a new one, accepting a value equal to the current utility of consumption u ( w - L) after repaying the loan, plus the future utility expected by a credit-eligible farmer with a loan C. The Bellman value function lacks known closed-form expression. However, it is well defined and the present value of expected future utility with a loan C is a strictly decreasing function of both the farmer’s creditworthiness f and the interest rate charged on the loan (Miranda and Chen, 2022). That the value of a loan to a farmer declines with the interest rate is not surprising. That its value declines with the farmer’s creditworthiness, however, is somewhat less obvious and begs an explanation. An unsecured loan embodies a value deriving from the farmer’s option to default if he realizes an income shortfall such that the utility forgone from repaying exceeds the benefits of retaining access to credit in the future. The option value depends on the farmer’s creditworthiness. The more creditworthy the farmer, the greater his sense of obligation to repay the loan. As such, the more creditworthy the farmer, the lower the default option value and, thus, the lower the value the farmer places on the loan. A farmer will accept a loan if, and only if, his expected future utility from accepting loan C exceeds his expected utility from declining loan A. It can further be shown that only farmers with sufficiently low creditworthiness will take out a loan. Specifically, for any interest rate r ³ 0, there exists well-defined critical creditworthiness f (r ) ³ 0 that is strictly decreasing in r, such that a farmer of creditworthiness f will accept a loan if, and only if, f < f (r ) . It can further be shown that, if the lender is a risk-neutral expected profit maximizer and observes the farmer’s creditworthiness, then, for any interest rate r, there exists a unique critical creditworthiness f (r ) ³ 0 such that the lender will offer a loan carrying interest rate r to a farmer of creditworthiness f if, and only if, f ³ f (r ) . The market for credit thus exhibits both adverse

204  Handbook of microfinance, financial inclusion and development

selection and credit rationing, with farmers of high creditworthiness electing not to take out a loan and farmers of low creditworthiness unable to obtain one. Figure 12.3(a) shows, for uninsured loans and varying rates of interest r, the minimum creditworthiness for a lender to offer a loan f (r ) and maximum creditworthiness for a farmer to accept a loan f (r ) with the levels in between that satisfy both borrower and lender (shaded region). As seen in the figure, the lender will not offer loans at interest rates below the cost of funds rf but will offer loans to a greater range of farmers at modestly greater interest rates. However, as the interest rate rises, the range of farmers to whom the lender is willing to offer loans eventually begins to shrink due to increasing likelihoods of default. The range of farmers willing to accept a loan, on the other hand, declines steadily with the interest rate. For interest rates beyond r0 , the lender is unwilling to offer a loan to any farmer willing to accept one, and the market for credit disappears.

Figure 12.3 Minimum creditworthiness to accept, maximum creditworthiness to offer, and range of market participation versus interest rate, uninsured and insured credit

Index insurance for developing countries  205

Figure 12.3(b) shows the range of farmer creditworthiness at which farmers will demand a loan and a lender is willing to offer them one, once bundling with index insurance is introduced. As seen in the figure, bundling index insurance with the loan will increase both the number of farmers demanding a loan and the number of farmers to which the lender is willing to offer a loan at every interest rate. The maximum interest rate that will sustain the market for credit r1 also increases. Bundling index insurance with loans creates opportunities to substantially increase the number of farmers that will have access to credit, while increasing both lender profits and the value derived by farmers from borrowing, potentially giving rise to credit markets where they would otherwise not exist.

BEHAVIORAL AND INSTITUTIONAL FACTORS AFFECTING DEMAND FOR INDEX INSURANCE Analysis of index insurance demand based on the neoclassical paradigm employed in the preceding sections has been supplemented by analyses based on behavioral theories in recent years. Most of these reveal yet additional problems that might explain the low demand for index insurance and its failure to achieve sustainability and scalability without substantial subsidies. These problems are intertwined to a certain extent because both institutional imperfections and cultural peculiarities trigger behavioral responses. Behavioral critiques of insurance fall into three broad classes: behavioral aspects of risk attitudes, lack of trust, and limited understanding of insurance. Empirical research on behavioral aspects of risk attitudes and demand for insurance seeks to find empirical confirmations, and perhaps reconciliation, of the two main behavioral theories used in economics—the standard expected utility theory and the prospect theory. While expected utility theory predicts risk-averse farmers will benefit from purchasing insurance, empirical research shows that the demand is much lower than the theory would predict (Patt, Suarez, and Hess, 2010). Much recent empirical research on risk attitudes and insurance demand uses a prospect theory framework that, in general, suggests that people like both to insure and to gamble (Barberis, 2013; Camerer, 1991). Prospect theory also suggests that individuals are risk averse to the worst outcomes and risk seeking otherwise, which limits demand for insurance covering small risks and with limited coverage or a small probability of non-payout, although the empirical evidence is mixed (Gunning and Zeitlin, 2019; Ito and Kono, 2010; Clarke and Kalani, 2012). Barberis (2013) reviews research on the largest insurance markets: property and causality. Empirical findings suggest that homeowners have unreasonably high risk-aversion levels according to the expected utility framework (Barseghyan et  al., 2013). Prospect theory, on the other hand, explains this by allowing homeowners to overweigh tail events—outcomes in which the deductible is paid. Additionally, the theory’s concept of reference point explains the bias toward a certain premium paid annually compared to a low probability deductible associated with greater loss aversion (Holt and Laury, 2002; Holden, 2014). Similar results are found in research on people at retirement age indicating that they spend less on annuities in mortality insurance. Overall, behavioral research on risk attitudes finds more empirical support for the prospect theory than for expected utility theory, although neither seems to fully explain risk attitudes and behavior. One simple explanation for the discrepancy between theoretical predictions and empirical observations could be cognitive dissonance, which in this context is related to adopting subjectively comfortable risk attitudes whereby the insured underestimate or ignore the likelihood of rare but catastrophic events and thus undervalue insurance. This can be

206  Handbook of microfinance, financial inclusion and development

reinforced by prolonged periods of no payouts, which is likely with index insurance, given that it is generally designed to cover one-in-ten- or one-in-twenty-year events in order to keep premiums low. Empirical studies also point out several other behavioral factors. One is ambiguity aversion—knowing what to expect when not buying insurance but not knowing what to expect from the insurer (Bryan, 2019; Elabed and Carter, 2015). Another is hyperbolic discounting— time-inconsistent preferences reflected in empirical findings suggesting that people tend to discount more distant future payoffs at an increasing rate (Ito and Kono, 2010). In addition, there could be a discontinuity of preferences for certainty associated with an individual’s preference for certain versus uncertain outcomes, i.e. preferring insurance with a certain payout to the same insurance with a 1 percent probability of non-payout but being indifferent between contracts with 1 percent and 2 percent probabilities of non-payout (Andreoni and Sprenger, 2012). Experimental evidence from Sub-Saharan Africa shows that about a third of the farmers are willing to pay more for insurance with uncertain costs (like premium rebates in bad years) suggesting that demand can be increased by redesigning contracts accordingly (Carter, Elabed, and Serfilippi, 2015; Dercon et al., 2014). In many developing countries, farmers may not trust the insurer to fulfill their end of the bargain. The lack of trust has been attributed to the high cost of enforcement, limited property rights, and poorly functioning court systems. The demand for index insurance also depends on the level of trust in the product itself because of the commitment to regular premium payments in exchange for an uncertain future payout. Some argue that educational programs can help to improve understanding of insurance products, although the existence of basis risk still undermines trust. Patt, Suarez, and Hess (2010) point out that, apart from economic factors that influence participation (premium, payout, etc.), trust in the organization that offers and administers it as well as trust in the process and in one’s own ability to make a good decision is also important. Evidence from India, Africa, and South America suggests that repeated interactions with potential customers as well as field games help to build this trust. Trust in the institution and its management often depends on the level of acquaintance with the administrators and the frequency of payouts (Cole, Stein, and Tobacman, 2014). Farmers usually trust index insurance more if the index measurement is certified by the government. The decision to purchase insurance also depends on interpersonal trust among insured individuals (Berg, Dickhaut, and McCabe, 1995). In addition, instead of perceiving insurance as redistribution across states of nature, many poor farmers view it with the principle of reciprocity in mind, which leads to disappointment and even demands for refunding the premiums paid if the payout does not happen (Basaza, Criel, and Van der Stuyft, 2008). Another trust-related issue arises from the discrepancy between rainfall recorded at weather stations and rainfall measured at the insured’s farm (Mobarak and Rosenzweig, 2013; Patt, Suarez, and Hess, 2010). Farmers and insurers may both understand that the contract covers losses due to drought but may have different opinions about what constitutes a drought. The insurance contract explicitly defines what is meant by drought in terms of prescribed precipitation shortfalls at a prescribed location that may be some distance from the insured location. If a farmer sees his and his neighbors’ crops wither due to lack of rainfall, he would conclude that a drought has occurred and that he is due a payout, even though the rainfall trigger has not been met at the weather station. Such experiences can significantly reduce insurance uptake in the long run.

Index insurance for developing countries  207

Yet another factor that undermines the demand for index insurance is the lack of full understanding of how insurance works. A number of behavioral studies based on field experiments suggest that, while educational efforts increase the demand for index insurance, the expected impacts are rather small (Giné, Townsend, and Vickery, 2008; Patt, Suarez, and Hess, 2010; Giné and Yang, 2009; McPeak, Chantarat, and Mude, 2005). While training and financial literacy have been shown to increase the demand for index insurance, the experiences shared among peers also influence demand, although the impacts are generally modest (Barr and Genicot, 2008; Karlan et al., 2014; Cole, Stein, and Tobacman, 2014). McPeak, Chantarat, and Mude (2005) describe an experiment game with index-based livestock insurance in Kenya mimicking local conditions and designed to help to explain insurance and other financial products to rural producers. They find mixed evidence and report that the process is labor intensive and complicated by competing claims to livestock with complex property rights as insurance interacts with social ideas of stock ownership. Giné, Karlan, and Ngatia (2013) cluster farmers by geography and randomly pick and assign low or high intensity of (1) financial literacy materials and (2) discount vouchers for index-based drought insurance. They find that having been given financial literacy materials increases the likelihood of buying index insurance. Overall, empirical research on behavioral aspects of demand for index and other types of insurance finds that the actual demand is lower than predicted by behavioral theories and offers explanations in terms of particular behavioral and institutional factors not captured by the theory. It seems that further behavioral research may help not only to improve understanding of the factors driving the demand for index insurance but also to design insurance contracts that better suit the needs of the rural producers.

CONCLUSION Index insurance has failed to achieve the promises made by many of its proponents. Over the past 30 years, innumerable index insurance pilot programs and studies have been implemented by researchers, government agencies, and non-governmental agencies in a wide variety of settings throughout the world and using a variety of indices and index insurance payout designs. However, despite all of these efforts, a commercially sustainable and scalable index insurance scheme for poor farmers remains elusive. No index insurance program currently generating any significant volume of sales anywhere in the world subsists without massive government subsidies. The primary obstacle is the inability to reduce basis risk sufficiently to make these contracts of any value to farmers. But other factors, such as self-insurance alternatives, also help account for smallholder lack of willingness to pay anything close to actuarially fair rates for index insurance. This presents a series of unanswered questions that need to be addressed by those who promote index insurance for smallholder farm-level risk management. First, if farmers will purchase index insurance only if heavily subsidized, do they do so because it reduces their risk, the stated rationale for index insurance, or do they do so simply to capture the subsidies? And if so, would they be better off simply receiving a cash grant? Second, if index insurance is to be subsidized, does it pass the public cost-benefit test? More to the point, does the use of subsidized index insurance by farmers generate positive externalities in the form of benefits to rural communities or contributions to national food security that justify the subsidies? Third, whether or not index insurance passes the public cost-benefit test, might not the subsidies be

208  Handbook of microfinance, financial inclusion and development

more cost-effectively put to other uses, such as the promotion of small-scale irrigation, health care, and education services (Binswanger-Mkhize, 2012)? While we are highly pessimistic that index insurance can succeed as a farm-level risk management tool, it may yet prove useful in expanding credit to small farmers if it is used to manage the risk embodied in lender loan portfolios. Bundling index insurance with loans has the potential to decrease loan defaults dramatically and provide strong incentives for lenders to offer a greater volume of loans at lower interest rates, and with only modest subsidies.

REFERENCES Akerlof, George A. 1970. “The Market for Lemons: Quality Uncertainty and the Market Mechanism.” Quarterly Journal of Economics 84(3): 488–500. Andreoni, James and Charles Sprenger. 2012. “Risk Preferences Are Not Time Preferences.” American Economic Review 102(7): 3357–3376. Barberis, Nicholas C. 2013. “Thirty Years of Prospect Theory in Economics: A Review and Assessment.” Journal of Economic Perspectives 27(1): 173–196. Barnett, Barry J. and Olivier Mahul. 2007. “Weather Index Insurance for Agriculture and Rural Areas in Lower Income Countries.” American Journal of Agricultural Economics 89(5): 1241–1247. Barr, Abigail and Garance Genicot. 2008. “Risk Sharing, Commitment, and Information: An Experimental Analysis.” Journal of the European Economic Association 6(6): 1151–1185. Barseghyan, Levon, Francesca Molinari, Ted O’Donoghue, and Joshua C. Teitelbaum. 2013. “The Nature of Risk Preferences: Evidence from Insurance Choices.” American Economic Review 103(6): 2499–2529. Basaza, Robert, Bart Criel, and Patrick Van der Stuyft. 2008. “Community Health Insurance in Uganda: Why Does Enrolment Remain Low? A View from Beneath.” Health Policy 87(2): 172–184. Bellman, Richard E. 1957. Dynamic Programming. Princeton, NJ: Princeton University Press. Berg, Joyce, John Dickhaut, and Kevin McCabe. 1995. “Trust, Reciprocity, and Social History.” Games and Economic Behavior 10(1): 122–142. Binswanger-Mkhize, Hans P. 2012. “Is There Too Much Hype About Index-Based Agricultural Insurance?” Journal of Development Studies 48(2): 187–200. Bryan, Gharad. 2019. “Ambiguity Aversion Decreases the Impact of Partial Insurance: Evidence from African Farmers.” Journal of the European Economic Association 17(5): 1428–1469. Camerer, Colin F. 1991. “Bounded Rationality in Individual Decision Making.” Experimental Economics 1(2): 163–193. Carter, Michael, Ghada Elabed, and Elena Serfilippi. 2015. “Behavioral Economic Insights on Index Insurance Design.” Agricultural Finance Review 75(1): 8–18. Chambers, Robert G. 1989. “Insurability and Moral Hazard in Agricultural Insurance Markets.” American Journal of Agricultural Economics 71(3): 604–616. Chantarat, Sommarat, Andrew G. Mude, Christopher B. Barrett, and Michael R. Carter. 2013. “Designing Index-Based Livestock Insurance for Managing Asset Risk in Northern Kenya.” The Journal of Risk and Insurance 80(1): 205–237. Clarke, Daniel and Gautam Kalani. 2012. “Microinsurance Decisions: Evidence from Ethiopia.” Technical report, International Labour Office. Clement, Kristina Yuzva, Wouter Botzen, Roy Brouwer, and Jeroen Aerts. 2018. “A Global Review of the Impact of Basis Risk on the Functioning of and Demand for Index Insurance.” Journal of Risk Analysis and Crisis Response 28: 845–853. Cole, Shawn, Daniel Stein, and Jeremy Tobacman. 2014. “Dynamics of Demand for Index Insurance: Evidence from a Long-Run Field Experiment.” American Economic Review 104(5): 284–290. Dalhaus, Tobias, Oliver Musshoff, and Robert Finger. 2018. “Phenology Information Contributes to Reduce Temporal Basis Risk in Agricultural Weather Index Insurance.” Scientific Reports 8 (46). Dercon, Stefan, Ruth Vargas Hill, Daniel Clarke, Ingo Outes-Leon, and Alemayehu Seyoum Taffesse. 2014. “Offering Rainfall Insurance to Informal Insurance Groups: Evidence from a Field Experiment in Ethiopia.” Journal of Development Economics 106: 132–143.

Index insurance for developing countries  209

Elabed, Ghada and Michael R. Carter. 2015. “Compound-Risk Aversion, Ambiguity and the Willingness to Pay for Microinsurance.” Journal of Economic Behavior & Organization 118: 150–166. Farrin, Katie and Mario J. Miranda. 2015. “A Heterogeneous Agent Model of Credit-Linked Index Insurance and Farm Technology Adoption.” Journal of Development Economics 116(5): 199–211. Giné, Xavier, Dean Karlan, and Muthoni Ngatia. 2013. “Social Networks, Financial Literacy and Index Insurance.” Discussion paper, The World Bank, Washington, DC. Giné, Xavier, Robert M. Townsend, and James Vickery. 2008. “Patterns of Rainfall Insurance Participation in Rural India.” World Bank Economic Review 22(3): 539–566. Giné, Xavier and Dean Yang. 2009. “Insurance, Credit, and Technology Adoption: Field Experimental Evidence from Malawi.” Journal of Development Economics 89(1): 1–11. Gunning, Jan Willem and Andrew Zeitlin. 2019. “The Demand for Insurance Under Limited Trust: Evidence from a Field Experiment in Kenya.” CSAE Working Paper 2019-06, Centre for the Study of African Economies, University of Oxford, Oxford, United Kingdom. Hill, Ruth Vargas, John Hoddinott, and Neha Kumar. 2013. “Adoption of Weather Index Insurance: Learning from Willingness to Pay among a Panel of Households in Rural Ethiopia.” Agricultural Economics 44(4–5): 385–398. Holden, Stein Terje. 2014. “Risky Choices of Poor People: Comparing Risk Preference Elicitation Approaches in Field Experiments.” Working Paper No. 10/2014, Norwegian University of Life Sciences, Centre for Land Tenure Studies, Aas, Norway. Holt, Charles A. and Susan S. Laury. 2002. “Risk Aversion and Incentive Effects.” American Economic Review 92(5): 1644–1655. International Fund for Agricultural Development. 2015. “Weather Index-Based Insurance in Agricultural Development: A Technical Guide.” Technical report, International Fund for Agricultural Development, Rome, Italy. International Water Management Institute. 2010. “Managing Water for Rainfed Agriculture.” Technical report, International Water Management Institute, Colombo, Sri Lanka. Ito, Seiro and Hisaki Kono. 2010. “Why is the Take-up of Microinsurance So Low? Evidence from a Health Insurance Scheme in India.” The Developing Economies 48(1): 74–101. Jensen, Nathaniel and Christopher Barrett. 2017. “Agricultural Index Insurance for Development.” Applied Economic Perspectives and Policy 39(2): 199–219. Karlan, Dean, Robert Osei, Isaac Osei-Akoto, and Christopher Udry. 2014. “Agricultural Decisions After Relaxing Credit and Risk Constraints.” Quarterly Journal of Economics 129(2): 597–652. Khalil, Abedalrazq F., Hyun-Han Kwon, Upmanu Lall, Mario J. Miranda, and Jerry R. Skees. 2007. “El Niño-Southern Oscillation-Based Index Insurance for Floods: Statistical Risk Analyses and Application to Peru.” Water Resources Research 43(10): W10416. Mahul, Olivier. 1999. “The Design of an Optimal Area Yield Crop Insurance Contract.” Geneva Papers on Risk and Insurance - Theory 24(2): 159–171. McPeak, John G., Sommarat Chantarat, and Andrew Mude. 2005. “Explaining Index Based Livestock Insurance to Pastoralists.” Agricultural Finance Review 70(3): 333–352. Mechler, Reinhard and Joanne Linnerooth-Bayer. 2006. “Disaster Insurance for the Poor: A Review of Microinsurance for Natural Disaster Risks in Developing Countries.” Technical report, ProVention Consortium/IIASA, Geneva, Switzerland. Miranda, Mario J. 1991. “Area-Yield Crop Insurance Reconsidered.” American Journal of Agricultural Economics 73(2): 233–242. Miranda, Mario J. and Jian Chen. 2022. “A Dynamic Model of Index-Insured Agricultural Microcredit with Strategic Default.” Technical report, The Ohio State University, Department of Agricultural, Environmental, and Development Economics. Miranda, Mario J. and Paul L. Fackler. 2002. Applied Computational Economics and Finance. Cambridge, MA: MIT Press. Miranda, Mario J. and Katie Farrin. 2012. “Index Insurance for Developing Countries.” Applied Economic Perspectives and Policy 34(3): 391–427. Miranda, Mario J. and Claudio Gonzalez-Vega. 2011. “Systemic Risk, Index Insurance, and Optimal Management of Agricultural Loan Portfolios in Developing Countries.” American Journal of Agricultural Economics 93(2): 399–406.

210  Handbook of microfinance, financial inclusion and development

Miranda, Mario J. and Dmitry V. Vedenov. 2001. “Innovations in Agricultural and Natural Disaster Insurance.” American Journal of Agricultural Economics 83(3): 650–655. Mobarak, Ahmed Musfiq and Mark Rosenzweig. 2013. “Informal Risk Sharing, Index Insurance and Risk-Taking in Developing Countries.” American Economic Review: Papers and Proceedings 103(3): 375–380. Norton, Michael T., Calum Turvey, and Daniel Osgood. 2012. “Quantifying Spatial Basis Risk for Weather Index Insurance.” The Journal of Risk Finance 14(1): 20–34. Patt, Anthony, Pablo Suarez, and Ulrich Hess. 2010. “How Do Small-Holder Farmers Understand Insurance, and How Much Do They Want It? Evidence from Africa.” Global Environmental Change 20(1): 153–161. Shavell, Steven. 1979. “On Moral Hazard and Insurance.” Quarterly Journal of Economics 93(4): 541–562. Stutley, Charles. 2010. “Innovative Insurance Products for the Adaptation to Climate Change Project Ghana (IIPACC) Crop Insurance Feasibility Study 2010.” Technical report, Deutsche Gesellschaft für Technische Zusammenarbeit GmbH.

PART IV VIEW FROM PRACTITIONERS AND FUNDERS

13. Measuring the evolution of client vulnerability: innovation at the BBVA Microfinance Foundation Claudio Gonzalez-Vega, Laura Mo, and Giovanni di Placido*

As an initiative within its social responsibility strategy, by 2007 the BBVA bank made the unrestricted transfer, in perpetuity, of a founding endowment for the pursuit of a well-defined mission and created the BBVA Microfinance Foundation (BBVAMF) as an autonomous entity. The endowment has been invested in the acquisition, merging, and consolidation of several Latin American microfinance institutions, leading to the upgrading and management of a group of currently six: Fondo Esperanza and Emprende Microfinanzas in Chile, Bancamia in Colombia, Banco Adopem in the Dominican Republic, Microserfin in Panama, and Financiera Confianza in Peru.1 From their NGO origin, the largest three were transformed and operate as prudentially regulated financial intermediaries, authorized to offer a full range of services and engaged in constrained profit maximization, with the Foundation as a patient controlling shareholder, focused on its mission (Gonzalez-Vega, 2020). The BBVAMF recognizes that knowledge about the clients is critical in placing them at the center of its operations. It learns about them in ways that are reasonably cost-effective, by immersing systematic data generation within the process of interaction with clients, while taking advantage of economies of scale—from the centralized construction of a dataset—and economies of scope—from comparative analytical efforts, facilitated by the joint management of six institutions. Diversity within the Group increases the range of variation of relevant variables and offers a variety of contexts within which data are generated and interpreted. Measurement of indicators about the circumstances of clients is a fundamental concern, while following their evolution is a key component of efforts to develop long-term relationships and to accompany clients with a sequence of value offers that respond to changing circumstances and demands. The Foundation must also respond to concerns about staying on target with its mission. Thus, it has generated longitudinal panel data for the universe of clients in its institutions. These data make it possible to track, over time, the poverty and vulnerability status of each client as well as describe the complete distribution of outcomes, for different cohorts, the totality, or diverse segments of the universe of clients. Claiming attribution for these outcomes is not the purpose. Rather, the Group attempts to understand the sources of client demands for various services, how these demands evolve over time, and what can be learned from measurable relevant outcomes that accompany the evolution of the client relationship. As reported here, some of these outcomes are quite impressive.

GOALS FOR MIDE As a central dimension of its knowledge management strategy, in the provision of financial services to excluded and vulnerable populations, the BBVAMF has implemented a unique and 212

Measuring the evolution of client vulnerability  213

innovative approach for the measurement of outcomes at the client level. This measurement tool (MIDE) pursues three goals: (i) The corroboration—for internal reassurance—and ability to show—as evidence of achievements, in external dealings—that the mission entrusted to the BBVAMF is being pursued and desired outcomes are being achieved. In essence, it proves that there is no mission drift (Gonzalez-Vega et al., 1997; Armendariz and Szafarz, 2011; Quiros et al., 2019). While this verification is of upmost interest to the founder, in implementing its social responsibility agenda, impact investment, and exercise of smart philanthropy, the Foundation hopes that its endeavors generate broad social value. (ii) The creation of useful knowledge for the development and implementation of financial technologies that reflect an understanding of and respond to the individual preferences, resource endowments, and productive opportunities of each client, within the realities of their household firms and environments where they operate. (iii) The demonstration—to financial inclusion constituencies—that it is feasible to achieve looked-for outcomes, believed to be out of reach for institutional providers, and the unveiling of the approaches that generate these outcomes. This transparency reflects the desire to make the lessons learned available—as public goods—to regulators, development agencies, and the microfinance industry at large.

PRODUCTIVE AND RESPONSIBLE FINANCE The founder entrusted the BBVAMF with the management of a group of specialized microfinance institutions, engaged in the supply of “productive and responsible” financial services for vulnerable household-firms that undertake self-employment endeavors. Productive finance means that the key piece in building long-term, personalized client-institution relationships is the identification of a productive opportunity—a client’s sufficiently profitable business— capable of generating high marginal rates of return when a credit constraint is removed. Such opportunity must be capable of spawning sustainable net income flows that demonstrate a high probability of the client being able to repay principal and interest on loans and keep a surplus, to increase—over time—the household-firm’s wealth and thereby its capabilities, freedom, and welfare. As relationships evolve and the extent and dynamic nature of the opportunities are validated, additional financial and non-financial services (payments, deposits, insurance, training) are added to the institution’s value offer—after verification of legitimate demands—for further exploitation of these opportunities. Screening Is about the Selection of Promising Clients For the Group, screening is the first task, namely, acquiring the know-how to select, among applicants within the target population, those revealed as “promising”—that is, those with the ability to repay loans and gradually escape poverty as well as being potentially capable of sustaining standard-of-living improvements over time. Success is grounded on the quality of client selection, given institutional preferences about outreach, risk-taking, patience, and long-term profits. The BBVAMF institutions are concerned with two types of errors (analogous to type I and type II errors in statistical inference). If they fail to correctly identify opportunities and reject promising applicants, they fail in achieving the breadth of outreach implicit in their mission.

214  Handbook of microfinance, financial inclusion and development

Further, if insufficient knowledge leads them to reject poorer applicants, their depth of outreach declines, contributing to the persistence of financial exclusion. If, instead, they misidentify opportunities, which do not exist to the extent assumed, losses from default follow and the clients are impoverished—through collateral forfeiture, loss of reputation and other intangible assets, or unfavorable credit ratings that block future opportunities to borrow. In either case, the prospect of building a lasting relationship is wasted. From this perspective, microfinance can either assist in the alleviation of poverty or it can increase poverty (Adams, 1998; Gonzalez-Vega, 1998). The actual outcome depends on the quality of credit decisions, based on relevant information, handled by skilled loan officers. Responsible finance means that, through careful screening, the Group avoids either of two mistakes: default or the exclusion of the creditworthy poor. Compatible Incentives for Both Borrowers and Loan Officers A second task is to create—through signaling and contract design—a structure of compatible incentives that strengthens client relationships. These incentives reflect the present value of potential future transactions, given by the increasing quality of the value offer to clients over time. Both tasks—screening and contract design—are information-intensive and benefit from the sunk costs associated with the accumulation of a stock of useful knowledge (Deaton, 2013; Hausmann, 2016). Moreover, the two tasks are complements. Successful screening is based on good information, while the right incentives encourage the revelation of information that reinforces the screening process. Influence on the life of clients depends on the appropriateness of innovations incorporated in lending technologies and their ability to overcome specific information and incentive imperfections, in matching contract terms to heterogeneous client characteristics (GonzalezVega, 2013). Remarkably, the critical role of specific credit technology features is ignored in many experiments about microfinance impacts on clients. This omission might explain the lack of significance found in some exercises (Person and Hernandez, 2019). Successful implementation of any financial technology also depends on the quality of the human capital involved, particularly loan officers’ skills, the institution’s induction and training practices, and the incentives—emerging from the organizational design—that govern staff behavior, discretionary choices, and private knowledge generation and sharing (Chaves and Gonzalez-Vega, 1996; Schmidt and Tschach, 2001; Holtmann, 2002; Hartarska, 2005). Wide differences across microfinance institutions in these regards condition their impact. A key strength of the BBVAMF is the careful screening of potential loan officers, to select those with the right attitudes and skills (Naranjo Landerer, 2017). Knowledge-Intensive Microfinance for the Creation of Relationships Microfinance is knowledge intensive. It requires the collection, handling, interpretation, accumulation, and use in decision-making of the knowledge that makes cost-effective financial transactions with vulnerable clients possible. Not only is microfinance more knowledgeintensive than traditional banking (given weak collateral), but it relies on different kinds of information, more opaque, more costly to collect, and more dispersed, with the relevant data on individual features varying over wider ranges. Traditional banking requires audited financial statements, tax returns, feasibility studies, and investment plans, while parametric scores are used for consumption credit. The

Measuring the evolution of client vulnerability  215

relationships are impersonal and hard information is required for every transaction (Berger and Udell, 2002). Incentives to repay emerge from collateral, typically a mortgage on real estate or a lien on some tangible asset (Fleisig and De la Peña, 2003). Contracts are enforced through judicial foreclosure. When the clients live and work far away (speak different languages or do not trust formal institutions), are informal and keep no accounting records, are too poor to own assets eligible as collateral, and take out loans too small to be worth judicial enforcement, a traditional banking approach is not appropriate technology (GonzalezVega, 2003). In contrast, microfinance loan officers visit applicants in their businesses, farms, or homes, speak their language, and are familiar with social and cultural norms. Information is gathered in situ, through screening processes that recognize the indivisible unity of the household-firm and are implemented by experienced loan officers, with the ability to assess both tangible and intangible characteristics of potential borrowers, such as honesty, reputation, creativity, and hard work. The credit decision relies on soft information and the accumulation, over time, of personalized knowledge. While, in traditional banking, the borrower’s tangible assets are pledged as collateral, in microfinance mostly intangible assets are at risk. Reputations are at stake, particularly in group credit technologies, with the loss of access to other transactions and the erosion of social capital when the borrower’s reputation is stained by being delinquent. The most important innovation of microfinance has been the widespread use of the “client relationship” as an incentive to repay (Gonzalez-Vega, 2019). Relationships are built from individualized knowledge and direct contact with the client, reducing risks of default, making it possible to grant ever larger loan sizes, at longer terms to maturity, and creating opportunities for the sale of other financial services, through a gradual progression of the relationship. Digital innovations and individualized attention have allowed the Group to sustain personalized contact with the clients during the COVID-19 pandemic. The value of relationships depends on how products match client circumstances and demands—given risk preferences and time discount rates—including terms to maturity compatible with the investments undertaken and amortization schemes consistent with cash flows. It also depends on timeliness in the disbursement of funds, the convenience and safety of deposit facilities, as well as the variety of the menu of services offered, and client benefits from economies of scope in dealing with a single provider of financial services, A key dimension of client relationships is for the institutions to be able to “grow with their clients” and offer them a wider range of products, as the maturity of their businesses may require. Recognizing Pervasive Heterogeneity Heterogeneity is substantial, ubiquitous, and critical. Financial services—characterized by vectors of diverse contract terms and conditions—are not homogeneous goods (Baltensperger, 1978). Financial technologies are not identical production functions (Gonzalez-Vega, 1976). Organizational design and stocks of human capital are not homogeneous inputs in the production of financial services. Distinguishing these disparities is key, however, in explaining differences in impacts on the life of clients. Diverse financial products (e.g. deposit facilities versus credit) may have different signs in their impact on household-firm behavior and outcomes (Guizar et al., 2015). Different lending technologies (e.g. group versus individual loans) show differential impacts through different average loan sizes and costs of lending and borrowing. Beyond technology types, the ways in

216  Handbook of microfinance, financial inclusion and development

which information is collected and used in decisions affect loan sizes and other credit terms and conditions and, thereby, potential impacts. The skills and professional profile of loan officers shape outcomes in rural versus urban environments. In microfinance institutions with a broad menu of services and wide geographic scope, like the BBVAMF, various bundles of products and delivery channels influence the lives of clients in complex ways and through multiple pathways. These disparities are key dimensions of the broader context within which the efficient supply of financial services counts for the vulnerable. Context matters (Pritchett and Sandefur, 2015; El-Zoghbi, 2019; Hernandez, 2020). Generalizations about the impact of (any or all) microfinance that ignore these distinctions and other dimensions of context frequently lack relevance (Vivalt, 2015). Achievement of the BBVAMF goals requires a nuanced understanding of its own specific institutions, service terms and conditions, technologies, client features, and context. The MIDE data and analysis address some of the issues resulting from the evaluation of social performance. By describing the transactions and relationships involved, MIDE is an effort to build—from the ground—a versatile dataset useful in policy, management, and innovation.

BUILDING THE DATASET As client relationships evolve, the BBVAMF institutions compile “relationship histories,” a sort of augmented credit histories that combine the institution’s private information about the clients with public information from credit bureaus and link it to relevant information about context (location, climate, infrastructure, market trends, prices, and value chains). This stock of data reduces information asymmetries. Over time, it lowers the costs and risks of additional transactions for both parties in the contract. A source of sunk costs for the institution, it increases the value of relationships and creates intangible wealth for the clients, which may be deployed in numerous other market transactions (Di Placido, 2020). A huge longitudinal dataset is built from the accumulation of these histories. Both Individual and Universal Knowledge MIDE’s multiple purposes conditioned several choices. Creating the dataset required the collection of information for the whole universe of clients. At any point in time, it includes data on all current clients in five countries. The dataset reveals both universal and individual knowledge about the clientele. MIDE generates time series data both for various aggregates and about each client, key in screening and building long-term relationships grounded on individual features. The more individualized the information, the more supplying a personalized bundle of services approaches a differentiating rather than a pooling offer. This reduces adverse selection (Villafani-Ibarnegaray and Gonzalez-Vega, 2007). Looking at the Shape and Shifts of the Whole Distribution The BBVAMF focuses both on learning about each individual client and gaining a deep understanding of the shape and behavior—over time—of the whole distribution of clients. For example, making the whole distribution explicit enabled the BBVAMF to identify how and how rapidly lifestyle changes occur for clients in different deciles of the sales, assets, or business surplus distributions, as various financial services do not influence the life of clients—at different levels of relevant variables—in the same way.

Measuring the evolution of client vulnerability  217

Comparisons of distributions built for specific groups, differentiated by gender, location, sector of economic activity, levels of education, age, and other key characteristics promote insights into how segments of the clientele use different types of financial services and the differential outcomes observed. The nature and magnitude of client-level outcomes—correlated to specific financial technologies and management approaches—and lessons learned about pathways that lead to those outcomes encourage debate about reasons for such differences and how additional innovation may assist in addressing observed lags. This approach inspired, for example, BBVAMF to develop explicit gender emphases and distinctions. In general, it improves understanding of increasing demand, as client businesses grow, or about what happens to deposit balances over economic cycles or during emergencies—which may lead to the verification of increased precautionary savings during adverse systemic shocks.

A PARTICULAR WAY OF LEARNING Recognizing the importance of knowledge management for the success of microfinance, the BBVAMF has consolidated several layers of knowledge, generated by a chain of diverse actors (clients, loan officers, national entities, and Madrid headquarters) to address the complexity of its challenge: to foster the sustainable development of excluded and vulnerable populations, through the supply of financial services (Gonzalez-Vega, 2020). Is the Mission Being Accomplished? Following its productive finance mandate, the Group grants loan amounts commensurate with verifiable opportunities by implementing screening technologies that—for an applicant that belongs to the target group—over time lead to a stream of loans and other services that impose the least transaction costs and risks to both parties and allow the fuller exploitation of the borrower’s opportunity. Success is revealed by a sustained relationship with borrowers who meet three characteristics: initial vulnerability, repayment ability, and gradual poverty alleviation. MIDE’s task is to discover how often and to what extent this happens. BBVAMF expects to accomplish this for increasing numbers of vulnerable household-firms (growing breadth of outreach). It hopes to incorporate household firms excluded from institutional financial services (depth of outreach). The declining costs associated with engagement in long-term relationships allow voluntarily constrained profits—still required to guarantee institutional sustainability and credibility of the promise of improved future services that support the relationships. Given substantial information imperfections, in the presence of heterogeneity, screening might be too expensive when lending to first-time entrepreneurs, who have not yet revealed the actual nature of their opportunity and skills to take advantage of it (Hartarska and GonzalezVega, 2006). Thus, the Group generally requires verification of prior business efforts, as their existence better forecasts the potential of significant impact since, for proven entrepreneurs, engaging in the client relationship with the institution puts their businesses on a different trajectory (Banerjee et al., 2019). MIDE offers information on clients in default and the timing of exit—after how many cycles they disappear from the dataset. When exit responds to removal by the institution— given unsatisfactory borrower performance—it is still not clear if this has resulted from inadequate screening or from idiosyncratic shocks suffered by the client. Historic client information assists in assessing the source of the problem. Exit may even reveal client success when, given

218  Handbook of microfinance, financial inclusion and development

market competition, the borrower gains access to more convenient sources of funds—possibly using the intangible assets created by the original relationship—or self-finances its efforts from saved surpluses. Exit questionnaires and survey methods complement the MIDE data while the performance of former clients may be followed in credit bureaus. While Not Claiming Attribution Attribution is a difficult and expensive task (Deaton, 2020). While interested in the progress of clients in emerging out of poverty and vulnerability and devoting substantial efforts to verify this outcome, the BBVAMF does not claim attribution. It does not allege that average outcomes at the client level are the unbiased and significant results of its supply of financial services. It does not claim that the Group institutions, by themselves, pull clients out of poverty. Why, how, and when this happens depends, among other things, on the client’s initial conditions and endowment of resources (how poor and credit-constrained they were), the nature and extent of productive opportunities available to them, context, and the incidence of adverse shocks. The BBVAMF hopes that a considerable proportion of its clients manage to escape poverty by taking advantage of their own productive opportunities—identified as accurately as possible by the lending technology—and by deploying their own skills and efforts, as the circumstances of their dynamic environments allow. These opportunities are (usually) independent of the relationship, and the assumption is that they would not have been as extensively exploited in the absence of the financial services supplied. MIDE measures how often poverty alleviation happens and at what stage of a client relationship it can be expected to occur. The Foundation understands that exit from poverty is a process that takes time and may suffer reverses, due to choices and exogenous events. The working assumption is that the Group enables this outcome by providing valuable financial services, powerful but not the only or necessarily the most important tool for success in these endeavors. Client outcomes are contingent on the whole vector of initial conditions, while the role of loans critically depends on the extent to which clients are initially credit-constrained. BBVAMF then contributes financial services that suitably accompany clients in their growth—their journey out of poverty. There is tremendous heterogeneity among clients—in terms of preferences, ambitions, resource endowments, demography and life cycle, opportunities, and what they are lacking (Sen, 1999). A microfinance lending technology is at its best precisely when it recognizes heterogeneity and thereby makes differentiating—rather than pooling—offers of financial services (Gonzalez-Vega, 2011). Since poverty is a dynamic condition, fueled by random events outside the control of clients, the fate of the poor constantly changes—as evidenced by the substantial volatility of incomes around poverty lines shown by the MIDE dataset. Extreme heterogeneity makes attribution of microfinance impacts challenging (Banerjee et al., 2019). Further, the randomization needed to address selection bias may not make everything equal and factors that influence both dependent and independent variables may lead to spurious associations (Deaton and Cartwright, 2018). The modest role of MIDE is to derive stylized facts that facilitate mission achievement. Tracking Progress While Acknowledging Complexity Poverty and vulnerability are multidimensional processes, the outcome of a multitude of determinants interacting in dynamic evolution along multiple pathways. Escaping it is a complex

Measuring the evolution of client vulnerability  219

process and complexity makes attribution almost unsurmountable (Morvant-Roux et al., 2014; Bamberger, 2016). Because there are numerous pathways out of poverty, different financial services influence exit in different and multiple ways (Collins et al., 2009; Collins et al., 2019; Readdaway et al., 2020). As a tracking exercise, MIDE is interested in how these pathways work, in order to design products that accompany clients in their progress toward achievement of individual household goals. Exercises that segment observed outcomes recognize that the relevant pathways and goals that influence outcomes depend both on context and the class of clients considered (Vaessen et al., 2014; Field et al., 2016; Kabeer, 2019). BBVAMF is interested in tracking how outcomes evolve over time (the dynamics of progress out of poverty), how they differ in various contexts (within and across countries), and how they differ for various types of clients. MIDE attempts to understand clients and assess how key dimensions of their lives evolve during the span of the relationship. The longer the relationship, the richer the lessons learned and the higher the probability that clients sustain their income gains. Both parties—institution and client—invest in the relationship and expect the returns associated with learning about each other to increase with its continuity. Mostly, MIDE is a characterization tool, where critical stylized facts can be derived to guide sustainable business decisions. A Centralized Dataset on Individual Clients Centralization of the dataset (Madrid headquarters) creates economies of scope and some diseconomies. Standardized and consistent variable definitions, metrics, and collection methods are used at all locations. Complex issues about appropriate inflation rates, to transform nominal into real magnitudes, exchange rates for currency conversions, macroeconomic shocks, poverty lines and the like are centralized for uniform treatment. The initially high costs in creating a uniform dataset become sunk costs, while increasing benefits are reaped from comparisons across countries—given different institutional contexts, local shocks, and clientele composition. This creates a huge laboratory where hypotheses are explored. While the unified dataset is available in Madrid, all institutions benefit from the Group’s stock of knowledge, complemented with locally available data. Time series information about each client makes these data somewhat independent of specific loan officers. While BBVAMF insists on the importance of personal relationships with clients, loan officer turnover is inevitably high, particularly in countries with intense market contestability, like Peru. The Group further understands that beyond the wide-ranging information collected about each client for MIDE, the loan officer’s intuition and private information play key roles in credit decisions. Gaining access to standardized information, in comparative mode, offers new managers and loan officers a valuable perspective on the evolution of the relationship during earlier periods, and facilitates its sustainability. Most of these data are private information, rather than the public information available in credit bureaus (De Janvry et al., 2010). This private information is what gives institutional competitive advantages. Pouring private information into MIDE, combined with careful recruitment and human resources (talent) management—to hire loan officers with a propensity to remain with the organization and with skills to recognize soft attributes of clients— shields these advantages (Naranjo Landerer, 2017). For the household-firm as a unit of analysis, by recording information at the time of loan disbursements, MIDE ascertains the time pattern of monthly sales, asset holdings, business surplus (earnings minus costs), per capita household surplus (as a proxy for per capita income

220  Handbook of microfinance, financial inclusion and development

to be compared to the country’s poverty lines), degrees of volatility of surpluses, index of housing quality, changes in education and the like. These indicators are consolidated for the Group and computed by country, for clientele segments (sex, rural or urban location, age, levels of education, sectors of economic activity, and various features of the environment). Given space limitations, the remainder of this chapter offers a few examples of ways the information may be reported and interpreted.

REMARKABLE OUTCOMES The institutions in the Group offer a broad range of financial services as well as advice, training, market access, and support to vulnerable household-firms such that, when they use financial services to take advantage of profitable opportunities, they improve their living conditions in a sustainable fashion. The accomplishments of this complex undertaking have been, so far, impressive. Some are described next. Breadth of Outreach By 2019, the Group had accumulated a gross loan portfolio of US$1,252 million, and disbursed US$1,607 million in new loans (Table 13.1),2 The six BBVAMF institutions range broadly in size. The outstanding portfolios of Financiera Confianza in Peru and Bancamia in Colombia were 45 times and 35 times, respectively, the portfolio of Emprende Microfinanzas in Chile. For the flow of annual disbursements, Financiera Confianza was almost 66 times and Bancamia 38 times larger than Emprende. Differences reflected a diverse history, prudential regulation status, and market size, maturity, and saturation of the microfinance sector in these countries. The three largest institutions secured charters to mobilize deposits from the public. Offering deposit facilities is one of the reasons why supervised microfinance institutions show an overall better performance than non-regulated NGOs, at least in Latin America (Quiros et  al., 2019). Outstanding deposits in these institutions amounted to US$676 million. About half of the loan portfolios of Colombia (49 percent), Peru (57 percent), and the Dominican Republic (58 percent) were funded with deposits. The ability to mobilize deposits not only leverages their equity, but also creates organizational structures and corporate cultures that promote sustainability and portfolio growth (Vogel, 1984; Gomez and Gonzalez-Vega, 2007). At the same time, it offers clients a valuable service that is in high demand (Vasquez, 1986; Guerrero, 1988; Mauri, 1993; Robinson, 2001). While Fondo Esperanza in Chile specializes in village banking, all other entities focus on individual loan technologies. Financiera Confianza (Palabra de Mujer) and Banco Adopem added village banking members to their mostly individual clientele. The mix of financial technologies and market features adds value to MIDE as quasi-laboratory. Given the systemic nature of the COVID-19 pandemic shock, the performance of joint liability groups has suffered more than proportionately, while individual credit clients have shown more resilience, as was identified for Bolivia by González-Vega and Villafani-Ibarnegaray (2011). The Group’s breadth of outreach is admirable. With a staff of 8,304 collaborators and a network of 596 branches and service points, complemented by 533 exclusive corresponding agents and 42,969 external agents and third-party points of attention, it reached large numbers of (mostly) previously excluded and vulnerable household-firms. By 2019, over 2.2 million net clients were reached (Table 13.1). Among them, over 1.9 million owned deposit accounts,

221

1,334 685 675,678

Average amount disbursed

Median amount disbursed

Outstanding deposits (000s USD)

2,234,413 1,108,278 951,776 100 1,963,178 100 1,282,637 764,585 271,224

Net number of clients*

Number of credit clients

Credit clients in Group (%)

Number of deposit clients

Deposit clients in Group (%)

Number of clients with only deposits

Number of clients with only loans

26

247,356

5

15,893

56

35

85,666

332,222

25

1,092,385 493,912

36

343,693

579,578

206

2,419

382,892

761

2,061

372,822

768,555

557,307

Notes:  — Not applicable or not available. * This net number avoids duplication when a client is both a borrower and a depositor.

Clients with only loans / total clients (%) 28

580

Number of branches

217

8,315

3,445

212,981

779

Number of employees

Operational data

1,204,220 296,713

Number of disbursements 1,289

1,606,865 382,483

Total amount disbursed (000s USD)

434,244

9

18,506

185,830

19

376,881

22

209,557

395,387

74

1,466

79,805

493

798

186,345

148,688

137,823

— —

— —

100

100

6,635

—

—

128,090

1

6,635

6,635

14

92

13

128,090

128,090

56

619

—

—

100

16,434

—

—

—

2

16,445

16,445

13

274

—

1,241

—

687

12,590

19,535

27,508

1,552

5,128

10,168

12,523

1,983

839

330,622

277,435

82,269

Bancamía Financiera Confianza Banco Adopem Fondo Esperanza Emprende Microserfin

1,251,673

Group

Gross outstanding portfolio (000s)

Financial data

Table 13.1  BBVAMF: financial and operational data and numbers of clients, as of December 31, 2019 (US dollars and percentages)

222  Handbook of microfinance, financial inclusion and development

while almost one million had loans outstanding. Bancamia contributed the highest proportion of net clients to the Group (49.6 percent), followed by Financiera Confianza (25.9 percent) and Banco Adopem (Dominican Republic). Most credit clients (72 percent) accessed both deposit and loan services, thereby strengthening their relationships with the institution. Of the total, 58 percent were clients with only deposits, 12 percent were clients with only loans, and 30 percent had both loans and deposits. Additionally, 28 percent had purchased insurance. While the higher number of depositors than borrowers is consistent with international experience (Gonzalez-Vega and Poyo, 1985), many deposits are small and show few transactions (Rozas and Erice, 2014). Among credit clients, 60 percent are women and over one-third live in rural areas (35 percent) and have completed, at best, primary education (37 percent). At least 85,000 borrowed in the formal market for the first time in 2019. Depth of Outreach: Poverty and Vulnerability The depth of outreach is impressive, given the size of loan outstanding of US$1,315 and of loan disbursement of US$1,334, on average (Table 13.1). Mean disbursements ranged between low US$798 (Adopem) and US$839 (Fondo Esperanza), and high US$1,983 (Emprende). A small loan size is an imperfect proxy for client poverty. Instead, the BBVAMF approximates per capita household incomes by carefully measuring surpluses generated by the client’s enterprise. Although this is also a partial measure, it is a reasonable proxy for households where most incomes are earned from self-employment in their own informal enterprises. Net surpluses from these enterprises are the main source of loan repayment and the welfare impacts of microfinance. The poverty and vulnerability status of credit clients is approximated from detailed information obtained during the loan evaluation process. Clients are ranked according to their position below the extreme poverty and poverty lines, as defined in each country. The BBVAMF has constructed a vulnerability line, defined as three times the country’s poverty line.3 These lines were roughly equivalent to US$1.0 to US$2.4 per day for extreme poverty, US$1.7 to US$5.8 per day for poverty, and US$5.2 to US$14.2 per day for vulnerability, depending on the country and urban or rural area. By 2019, of the total number of outstanding credit clients in the Group, about 9 percent were extremely poor, while of the new clients added in that year, 15 percent were extremely poor. This higher proportion among new than among all clients reflects the Group’s sustained commitment to reaching this population segment, when adding new clients, and the gradual escape from extreme poverty of clients added to the portfolio in earlier years. In turn, 27 percent of borrowers were poor (because 9 percent were extremely poor, 18 percent were poor but not extremely poor). Among new clients, 38 percent were poor (22 percent were poor but not extremely poor). Finally, 79 percent were below the vulnerability line (52 percent were vulnerable but not poor) while, among new clients, 84 percent were vulnerable (46 percent were vulnerable but not poor). There Is No Mission Drift There is no mission drift within the Group. Despite rigorous screening criteria, for over 12 years the Foundation’s focus on its mission has kept its institutions serving the same population segments it has targeted. The extremely low variability—over time—of these shares

Measuring the evolution of client vulnerability  223

is exceptional. The share of vulnerable clients has been between 80 percent and 90 percent, while the share of poor clients has been just above or below 40 percent of the number of new borrowers added to the portfolio each year. The gap between the shares for new and all borrowers shows the impact of a gradual exit from poverty. The commitment to focus on the target group has been true for all institutions, ranging from 75 percent (Financiera Confianza, operating in the extremely competitive, saturated microfinance market in Peru) to 95 percent (Fondo Esperanza). Considering all cohorts of borrowers incorporated during 2011–2019, of those who were poor at the time of the first loan—and had continued in the relationship—63 percent were no longer poor at the end of the fourth loan cycle.4 Alternatively, of those who were not poor at the time of the first loan, at the end of the fourth cycle, 13 percent were poor. These shifts accounted for a net reduction of 44 percent of poverty incidence after four loan cycles, compared to the initial incidence (Figure 13.1). A proportion of poor clients (36 percent) escapes poverty after the first loan (an outcome learned when evaluated for a second loan), but because 12 percent of those not poor fall into poverty, there is a 15 percent net reduction in its incidence with the first loan. Both the share of those having escaped poverty and the net reduction in the incidence of poverty increase with each loan cycle (although at a declining rate), as long as client relationships endure. In microfinance, each new loan cycle enjoys a longer term to maturity, so additions to the share of clients escaping poverty take longer to show in the data as the number of cycles

Note:  Computed from the numbers of clients for all cohorts of borrowers from Banco Adopem, Bancamia, and Financiera Confianza, with a first loan in any year, 2011–2019.

Figure 13.1 BBVAMF: proportions of clients not poor at the time of their first loan who fell into poverty, proportions of clients initially poor who escaped poverty, and net poverty reduction, by loan cycle, 2011–2019 (percentages)

224  Handbook of microfinance, financial inclusion and development

increases. Overcoming poverty takes time, while it seems to be facilitated by long-term client relationships. MIDE records the evolution of monthly per capita surpluses from the household enterprise, for every cohort of clients. This indicator is closely related to the household’s poverty status. To appreciate the depth of poverty, MIDE defined an index of 1 when a per capita household enterprise surplus equals the poverty line in each country. Figure 13.2 shows the evolution of the mean and median per capita surpluses relative to the poverty line. For annual cohorts from 2011 to 2019, when they get their first loan, for average borrowers in extreme poverty the index is 0.35. They earn per capita surpluses of just over one-third of the poverty line. It takes them four cycles of borrowing to climb over this line. For the median

Note:  Per capita household surplus compared to the country´s poverty line, as of the corresponding year (indexed at 1), for clients of Banco Adopem, Bancamia, and Financiera Confianza, and for cohorts of extremely poor, poor, and vulnerable clients at the time of receiving their first loan, 2011–2019.

Figure 13.2 BBVAMF: mean and median per capita household surplus relative to the poverty line, for initially extremely poor, poor, and vulnerable clients, by loan cycle, for 2011–2019 cohorts

Measuring the evolution of client vulnerability  225

extremely poor borrowers, this takes five cycles of borrowing. By the fifth cycle, the index mean reaches 1.36 (almost four times its initial value), while the median reaches 1.05 (almost three times its initial value). The median extremely poor household firm barely manages to progress above poverty and remains vulnerable. In turn, when they get their first loan, for borrowers in poverty but not in extreme poverty, the average index is 0.75. It takes them two cycles of borrowing to move above the poverty line. For the median borrower, this takes three cycles. The mean of the index reaches 1.67 by the fifth cycle (2.2 times its initial value), while it reaches 1.31 for the median borrower (1.7 times its initial value). This may not be sufficient to prevent relapses into poverty.5 Finally, when they get their first loan for vulnerable but not poor borrowers, the average index is 1.72 and the median is 1.60. With both indicators below two times the poverty line, these clients are vulnerable enough to be within the Foundation’s target group. The index reaches a mean of 2.42 by the fifth cycle (1.4 times its initial value), and a median of 1.90 (1.2 times its initial value). As highlighted, for each group, the average grows faster than the median index, but both indicators grow fastest for the extremely poor, and faster for poor than for vulnerable clients. Thus, the index means (as well as the index medians) of the three groups gradually converge over time. This has distributional implications. While the BBVAMF does not claim attribution for these results, they suggest the presence of substantial and complex poverty-alleviation processes among the clients. To some extent, these gains reflect the success of the screening practices in identifying household firms, within the target population, with a potential to escape poverty. Once these clients’ potential to improve is revealed, there is a strong presumption that access to finance assists them in reaching this potential. Gains from finance are then associated with accurate screening and delivering appropriate services. Among microfinance institutions around the world, there is a broad range of these attributes. This should be considered when evaluating the impact of finance. Growing Enterprise Size as the Relationship Matures (Sales) Tracing the evolution of key indicators—which offers insights into processes that contribute to poverty alleviation—is made possible by the dataset. Data consolidated for borrowers in Banco Adopem, Bancamia, and Financiera Confianza show how the monthly sales of client enterprises evolve, as their relationship matures (Figure 13.3). For all cohorts with their first loan any year from 2011 to 2019, the monthly sales at the time of screening—before the first cycle of borrowing—in real terms (2019 prices) averaged US$1,067, while the median sales amounted to US$766. The difference reflects the right-skew asymmetric distribution of sales, with a high proportion of clients enjoying sales well below the cohort mean (Figure 13.4). Increasing with each new loan cycle, at a compound rate of growth (CRG) of 10.9 percent per cycle, mean monthly sales reached US$1,612 by the beginning of the fifth cycle while, at a CRG of 10.8 percent, the median reached US$1,157. After four loans, mean enterprise sales reached 1.47 times while median sales reached 1.51 times their initial level.6 Increasing sales are predictors of potential growth of their business, as relationships with the institution mature. While—to some observers—these amounts may seem modest, they are sufficient to support the poverty alleviation results already highlighted. Moreover, since the difference between the current and initial values of both mean and median strictly increases, as the number of loans increases, the potential improvements are greater, and the longer the client remains in the relationship.

226  Handbook of microfinance, financial inclusion and development

Note:  The computations exclude clients not vulnerable before the first loan. Amounts in domestic currencies, in real terms, and converted into US dollars.

Figure 13.3 Mean and median business assets, sales, and surplus, by loan cycle, consolidated for client cohorts with first loans at Banco Adopem, Bancamia, and Financiera Confianza, 2011–2019 (US dollars) Clients of Financiera Confianza showed initial average monthly sales (US$1,184) higher than in the other two institutions, while median monthly sales (US$652) were lower. Differences reflected a greater skewness of the distribution and heterogeneity of clients in Peru (due, in part, to a broader set of sectors of economic activity reached, including a higher share of agriculture) than in Bancamia (US$1,030 and US$732, respectively) and Banco Adopem (US$1,001 and US$870), with the latter showing the least skewness and greatest homogeneity of its clientele. Mean and median sales grew fastest in Financiera Confianza (CRGs of 12.2 percent and 13.9 percent, respectively), in part reflecting more rapid economic growth in Peru. Sales of clients of Bancamia enjoyed CRGs of 10.6 percent (mean) and 10.1 percent (median), while clients of Banco Adopem enjoyed CRGs of sales of 9.1 percent and 9.2 percent respectively. Median sales in Peru grew to 1.7 times their initial value, compared to 1.5 times in Colombia and 1.4 times in the Dominican Republic. Despite these differences, which mostly reflect contextual disparities and, maybe, differences in screening attitudes and experience in the three countries, what is surprising is how similar they are. This reflects the success of the BBVAMF in instilling in its largest institutions (93 percent of its clientele) well-defined expectations of outreach to its target clientele and in avoiding mission drift in all of them. Rapid Wealth Accumulation and Declining Inequality (Assets) Figure 13.3 shows the evolution of the value of enterprise assets as relationships mature. For cohorts with their first loan from 2011 to 2019, the mean value of initial assets was US$ 4,713

Measuring the evolution of client vulnerability  227

Note:  Computation excluded clients above the vulnerability line before the first loan. For comparisons, the equivalent normal distributions are shown by the smooth lines.

Figure 13.4 Distributions of the value of assets, monthly sales, and monthly business surplus, prior to the first loan, consolidated for clients of Banco Adopem, Bancamia, and Financiera Confianza, for the 2011–2019 cohorts of clients (percentages)

228  Handbook of microfinance, financial inclusion and development

and the median value was US$2,177. Mean assets value was 2.2 times the median, while mean sales were 1.4 times the median. There is more skewness and heterogeneity with respect to assets than to sales (Figure 13.4). Sale-to-assets relationships suggest high rates of return. Annualized, mean sales (US$12,802) represented 2.7 times the mean value of assets, implying a mean gross rate of return, in terms of sales, of 272 percent per year. In turn, annualized median sales (9,197) represented 4.2 times the median value of assets, implying a median gross rate of return of 422 percent per year, for these households with smaller asset holdings than the mean. The difference hints at diminishing marginal rates of return. The high rates are commensurate to those found by McKenzie and Woodruff (2006) and De Mel et al. (2008) for informal microenterprises in Mexico and Sri Lanka and by Udry and Anagol (2006) for poor rural enterprises in Africa. Strictly increasing over time, at a CRG of 15.1 percent per cycle, by the time of negotiation of their fifth loan, the average value of the borrower’s assets had grown to US$8,265. Namely, after four loans, on average the borrower’s enterprise assets had grown to 1.8 times their initial value, reflecting a sustained process of wealth accumulation. Similarly, strictly increasing at a CRG of 21.4 percent, the median value of the borrower’s assets had grown to US$4,721. After four loans, the median assets had grown to 2.2 times their initial level, faster than for the average borrower. Differences in asset values represent a key initial condition, influencing both the client’s poverty status as well as the extent of wealth accumulation. Clients of Financiera Confianza showed an initial mean value of assets (US$6,874) and initial median value (US$2,177) higher than for the other institutions, reflecting a greater skewness of the distribution and heterogeneity of clients (with a mean of 3.2 times the median), compared to Bancamia (with initial mean value of assets of US$4,451, namely 2.2 times the median value of assets of US$2,069) and Banco Adopem (with initial mean value of assets of US$3,690, namely 2.3 times median assets of US$1,588). Mean and median values of assets grew the least rapidly for clients of Banco Adopem (mean CRG of 11.4 percent and median CRG of 19.0 percent). These rates were higher at Financiera Confianza (16.0 percent and 21.4 percent) and at Bancamia (14.9 percent and 20.9 percent). Thus, mean asset values increased, compared to their initial level, to 1.8 times in Peru, 1.7 times in Colombia, and 1.5 times in the Dominican Republic. In turn, median asset values increased to 2.2 times in Peru and in Colombia and 2.0 times in the Dominican Republic. The more rapid rate of asset accumulation for the median than the mean client reflected a reduction in wealth inequality, as in each country the mean and median values of enterprise asset holdings gradually converged. This is an important distributional consequence, consistent with the Foundation’s mission. High Returns, but Growing at a Diminishing Rate (Enterprise Surpluses) Figure 13.3 shows the evolution of monthly enterprise surpluses (approximated by earnings minus observed costs). In view of less variability in household size, compared to variations in asset holdings and rates of return, differences in enterprise surpluses mostly determine differences in per capita surpluses—the proxy for per capita income. For all cohorts that received their initial loan between 2011 and 2019, the mean monthly enterprise surplus amounted to US$357, in real terms. Annualized, this surplus was equivalent to 90.9 percent of the mean value of assets, suggesting high net rates of return to these

Measuring the evolution of client vulnerability  229

enterprises. Strictly increasing over time, at a CRG of 10.2 percent per cycle, by the time of negotiation of their fifth loan, the firm’s mean monthly surplus had grown to US$527. That is, on average, after four loans, the enterprise surpluses had reached 1.48 times their initial preloan level, reflecting a sustained process of net income growth. These surpluses become available for loan repayment, additional investment in assets, firm growth, and increased purchasing power to augment household welfare. The gap between the initial business surplus and its value at any point in time strictly increases as the relationship matures, but at a declining rate. By negotiation of the fifth loan, the average annual surplus was equivalent to 76.6 percent of the mean enterprise assets, suggesting again—when compared to 90.9 percent just before the first loan—diminishing marginal returns to the growth of the firm (Figure 13.4). By the time of the first loan, the median of the monthly enterprise surplus amounted to US$320. Annualized, this was equivalent to 176.6 percent of the median asset values. This is higher than the implicit return of 90.9 percent corresponding to mean assets—which are more valuable than median assets—again hinting at diminishing returns. This difference may reflect that the median borrower would have been more credit-constrained, at the time of the first loan, than the mean borrower. Strictly increasing over time, at a CRG of 8.5 percent, by negotiation of the fifth loan, the median value of the monthly surplus had grown to US$445. After four loans, the median borrower’s enterprise surpluses had reached 1.47 times their initial pre-loan level, almost in parallel with the growth of mean surpluses. The gap between the initial median surplus and its value at any point in time strictly increases as the relationship matures, but at a decreasing rate. By the negotiation of the fifth loan, the median annual surplus was equivalent to 112.8 percent of the median borrower’s enterprise assets, suggesting again diminishing returns to the growth of the firm. Because the intervals between loans increase as the relationship progresses, over time returns decrease less rapidly than is implied by these approximations (rates per cycle). After several borrowing cycles, additional increases in loan size might not be any longer marginally profitable. This creates challenges for the clients, in terms of how to expand the household’s productive opportunities, and the institutions, in terms of what kinds of products and service terms appropriately match the new circumstances of the clients. Frequently, however, quantity loan rationing blocks valuable opportunities (Gonzalez-Vega, 1977; Keeton, 1979; Stiglitz and Weiss, 1981). Across countries, median surpluses differ less than the other magnitudes. The initial pre-loan median monthly surplus ranges from US$287 for Bancamia to US$333 for Banco Adopem and to US$355 for Financiera Confianza. After four loans, these median surpluses are US$477 (and thus 1.43 times their initial value) for Banco Adopem, US$392 (and thus 1.37 times their initial value) for Bancamia, and US$460 (and thus 1.30 times the initial value) for Financiera Confianza. The corresponding CRGs are 9.4 percent per cycle for the Dominican Republic, 8.1 percent for Colombia, and 6.7 percent for Peru. Substantial and Unequal Distribution Dispersion While there is significant dispersion for each distribution, indicating a presence of ample heterogeneity, the degree of dispersion differs substantially for the distributions of assets, sales, and business surplus. While the median of each distribution is markedly lower than the mean, this is more pronounced for the distribution of assets, less for the distribution of sales, and least for the distribution of business surplus (Figure 13.4).

230  Handbook of microfinance, financial inclusion and development

The coefficients of variation (standard deviation divided by the mean) for the distribution of assets (152) are close to three times those for the distribution of surpluses (56), while the coefficients for the distribution of sales (98) are almost two times those for the distribution of surpluses. This is also reflected by the high skewness parameter of sales in Financiera Confianza (4.15) and the low skewness in Banco Adopem (1.93). Context matters. The distinctions highlight clientele differences (more homogeneous in Adopem) and differences in lending technology (more extensive use of village banking in Adopem). Similar country differences explain the high positive kurtosis of the aggregate distributions of sales (47.9) and assets (32.7), but low kurtosis for the distribution of surpluses (3.6). Some values of assets and sales are high, compared to the mean and median, but this happens less for surpluses. Any improvement in the distribution of assets across clients induced by the Group would be socially valuable. There is no space for details of the interesting story about the clientele revealed by these data. The stylized facts signal a vulnerable and excluded population, with great disparity in asset holdings before incorporation in institutional credit portfolios. This inequality is reflected in a Gini coefficient of 0.68 for the asset distribution (with minor discrepancies across countries). Such wealth disparities reflect different stages in life cycles, diverse asset accumulation strategies (land, physical capital, human capital), different histories of adverse shocks that depleted earlier wealth accumulation, age of indivisible capital goods, diverse factor intensities of various sectors of economic activity, and the extent to which clients had been credit-constrained before being reached. The somewhat lesser inequalities from the distribution of sales (Gini coefficient of 0.56) further reflect demographics, local market size, and huge transaction costs dispersion. These costs represent constraints to productive opportunities and barriers to enterprise growth. In the case of sales, country differences are substantial, which is not the case for assets. These differences may reflect the breadth of economic activities of the clients, with the larger and more diverse geographic and economic space of Peru in part explaining the higher Gini coefficient (0.62) compared to the Dominican Republic (0.38). Explaining the lesser inequality of business surpluses (and observed similarity of Gini of 0.31) will require further research. When the comparatively less dispersed business surpluses of these households are combined with a highly unequal initial distribution of assets, a highly unequal distribution of gross and net rates of return on assets emerges. This broad dispersion in marginal rates of return is characteristic of fragmented economies, and it mostly responds to the level and dispersion of risk and transaction costs (McKinnon, 1973). Informal financial markets possess a limited ability to reduce this dispersion (Collins et al., 2009). The depth of financial intermediation precisely reduces the dispersion and reallocates the country’s resource endowment in a more productive fashion (Shaw, 1973; GonzalezVega, 1984; Robinson, 2001). The resulting increased productivity of resources encourages the BBVAMF to expect positive outcomes from expanding the financial inclusion of the target clientele. The heterogeneity of the target population suggests that there are multiple pathways out of poverty as well as instruments for sustaining the escape from poverty. Given the fungibility of money (Von Pischke and Adams, 1980), formal finance provides additional generalized purchasing power and inter-temporal and risk-management services that allow diverse households to escape poverty in multiple different ways, when they pursue different livelihood strategies and care about different goals in life (Gonzalez-Vega, 2013; Gonzalez-Vega et al., 2021). Moreover, as the relationships evolve, the reduction in the dispersion of the relevant

Measuring the evolution of client vulnerability  231

distributions is an indirect indicator of the pro-poor gains in household outcomes from obtaining access to a broad range of financial services. For the period under analysis (2011–2019), the magnitudes of all key indicators strictly increased with each loan cycle. This may not necessarily always be the case, particularly in the presence of sufficiently strong adverse systemic shocks. These magnitudes kept strictly increasing, however, despite several major weather-related events (such as El Niño and severe floods), revealing the robustness of the Group’s clientele and appropriate responses adopted by the institutions in assisting clients in need. This also reflected sufficient diversification, while index insurance was used to protect portfolios from catastrophic events. Improvements in Income Distribution So far, the inquiry has focused on mean and median outcomes related to the influence of credit services on per capita surpluses and the role that the evolution of asset holdings, sales volumes, and business surpluses play in shaping them. The results show that, for the 2011–2019 period, strictly increasing mean and median outcomes are associated with each new loan, as the client relationship evolves. An important question relates to the varied extent to which these outcomes benefit clients differently situated along the initial distribution of per capita surpluses. A first component of the answer is that the poorer the client, the gain is proportionately greater. Access to BBVAMF loans not only improves surpluses per capita but—by improving proportionately more the surpluses of the poorer—it reduces inequality in the income distribution across the clientele. The convergence, as new loan cycles are added, of the mean and median per capita surpluses for extremely poor, poor, and vulnerable clients reflects this outcome (Figure 13.2). In effect, compared to the poverty line, by the time of the first loan, mean per capita surplus is 5.0 times higher for vulnerable than for extremely poor clients. By the time of the second loan, it is 2.6 times, and by the fifth loan, it is 1.8 times higher. Similarly, by the time of the first loan, median per capita surplus is 4.5 times higher for vulnerable than for extremely poor clients. By the time of the second loan, it is 3.0 times, and by the fifth loan, it is 1.8 times higher. The means and medians for the three groups converge. This outcome results from higher rates of growth of per capita surpluses (reflecting growth of the household’s business surplus) for the extremely poor than for the poor and the vulnerable. Over the five-loan cycle, the CRG of the mean per capita surplus of the extremely poor is 40.9 percent per cycle, compared to 22.0 percent for the poor, and 9.0 percent for the vulnerable. Similarly, the CRG of the median per capita surplus of the extremely poor is 31.1 percent per cycle, compared to 14.5 percent for the poor, and 4.4 percent for the vulnerable. While, after four loans, the mean per capita surplus of the extremely poor increased 3.9 times, for the poor it increased 2.2 times, and for the vulnerable 1.4 times. Similarly, the median per capita surpluses of the extremely poor, poor, and vulnerable increased by 3.0, 1.7, and 1.2 times, respectively. In consequence, inequality of the income distribution within the clientele declined. Once the screening process identifies each client’s productive opportunities, even if the opportunities of the extremely poor were less attractive in generating gross income (in absolute terms) than those for the rest of applicants, by having been excluded from institutional financial markets and, thereby, by being sufficiently more strictly credit-constrained, the marginal rate of return from the command over resources acquired with the loan could be higher for the extremely poor. Thus, at the margin, the rates of return and the rates of growth of their

232  Handbook of microfinance, financial inclusion and development

surpluses could be higher than for their more fortunate peers. By declining more rapidly, their rates of return gradually converge with those of the less constrained groups. In the process, nevertheless, their gains from finance are greater (Gonzalez-Vega, 1976). This virtuous distributional outcome requires a willingness to include (given by the mission) and a lending technology capable of incorporating previously excluded—extremely poor and poor—clients. The screening process must be able to correctly assess—at the margin— the returns to their enterprises and then grant each applicant loan amounts that can be repaid from those returns. Combined with very low delinquency rates, these outcomes suggest that, to some important extent, the BBVAMF has accomplished this feat. While rates of return and rates of growth partially converge, over time they will not be equalized if there are positive transaction costs. These costs are particularly high for both borrowers and lenders in the case of the extremely poor. Further, the more rapidly the marginal returns diminish as the firm size grows (most likely for the extremely poor), the more rapidly their rates of return and of growth converge toward the levels for not-so-poor and vulnerable clients (Gonzalez-Vega, 1984). Further, over time the coefficient of variation of the distribution of assets declines with each new loan cycle. This implies that not only income but also the distribution of wealth becomes less unequal, signaling an important distributional impact. The Whole Distribution of Per Capita Surpluses Shifts Outward A second component of the answer is that per capita surplus gains are observed at all levels of the distribution, cycle after cycle. Figure 13.5 shows the cumulative distributions of per capita

Note:  The poverty index equals one; the vulnerability index equals three. Clients not vulnerable before the first loan are excluded. The table shows surplus values relative to the poverty line for percentiles 25, 50, and 75.

Figure 13.5 Cumulative distributions of the per capita business surplus relative to the country’s poverty line, by loan cycle, consolidated for clients of Banco Adopem, Bancamia, and Financiera Confianza, receiving their first loan in 2015

Measuring the evolution of client vulnerability  233

surpluses, for each loan cycle, consolidated for the clients of Banco Adopem, Bancamia, and Financiera Confianza. Comparing the distribution for the first cycle with the distributions for each successive cycle, with each new cycle the entire distribution moves rightward. This is the equivalent of first-order stochastic dominance. The distribution of client surpluses for each new cycle completely dominates the previous distribution. Namely, for any percentile of the distribution (on the vertical axis), successive cycles are associated with higher per capita surpluses, relative to the poverty line (horizontal axis). With every new loan cycle, additional proportions of vulnerable clients move beyond vulnerability (three times the poverty line, shown by the vertical line at 3). Similarly, the proportion of clients below the poverty line—the vertical line where per capita surplus equals 1—declines with each loan cycle, with these clients thereby leaving the ranks of the poor and filling the ranks of the vulnerable. There is (at 1) entry into and (at 3) exit from vulnerability. Movements at the time of the second and third cycles are the largest, indicating that large early gains are obtained by the cohort once the first “exploratory” loan cycle is left behind and before marginal returns decline further. For a quantitative understanding of this process, the table shows the relative per capita surplus for clients at percentile 25, percentile 50 (median), and percentile 75 of the distribution. Per capita surpluses are read horizontally for each percentile. A client at percentile 25 reaches the poverty line by the fourth loan. At this point, three-quarters of the clientele are above the poverty line. The client at the percentile 25 of Bancamia falls slightly short of this goal by the fifth loan (0.99), but those of Banco Adopem and Financiera Confianza cross the poverty line (1.02 and 1.05, respectively) by the fifth and fourth loan. The client at the percentile 75 crosses two times the value of the poverty line by the second loan. This also happens in Bancamia, and by the third loan, it happens in Banco Adopem. The client of Financiera Confianza at the 75 percentile is at two-times the poverty line since the first loan. Identity Matters Two cautionary comments are useful. First, at the individual household level, identity matters (Beneke de Sanfeliu and Gonzalez-Vega, 2000). The client at the median (or any other percentile) of the distribution is not the same household when the distribution shifts rightward. Some clients reap gains, in various magnitudes, and improve their position in the distribution; others suffer losses and see their distributional rank decline. In the new distribution, a different household fills the position of the client at percentile 25. The dominance of each cumulative distribution over the previous one does not mean that all clients improve in their vulnerability status. Rather, it means that there is no systematic bias in the shift outward, as if some client classes (percentiles) improved and others deteriorated. Instead, in the figure, each new cumulative distribution does not intersect the earlier one. The Challenge of Attrition Second, the analysis includes only clients who return for an additional loan, at which time information (related to what happened during the previous period) is collected. There is, however, attrition, and the fate of clients when they are no longer in the relationship is not known from the MIDE dataset. Analysis of attrition shows the largest number of exits taking place between the first and second loans (Mo, 2020). The proportion of exits is highest for Financiera Confianza (21 percent after the first year), driven mostly by clients of the Palabra de Mujer program (directed at joint liability groups of very poor women). After the first year,

234  Handbook of microfinance, financial inclusion and development

in Banco Adopem exits reach 10 percent, and in Bancamia 9 percent. After the second year, all institutions show similar few exits. Clients in urban locations, the young, women, those working in retail trade and services, and the poor are significantly more likely to exit earlier (except for Banco Adopem). Highly educated clients also leave early, possibly because self-employment is transitory for them. Because they do not return for the next loan, what happened during the period of the last loan and what the reasons are for deserting are not known by MIDE, except from exit questionnaires. The higher exit rate after the first loan may be a consequence of information imperfections. Screening those until-then-excluded poor and vulnerable applicants, frequently without credit histories, is costly. Investing too much in screening is unprofitable (unsustainable) for the institution. After some degree of screening, the institution may prefer to learn more about the client through the observation of actual performance in the relationship than through additional pre-loan costly screening. Further, given the BBVAMF mission, it is preferable to offer easier entry into potential relationships, even if some of these attempts fail, than err when actually creditworthy but not well-known applicants are excluded. Similarly, recently included clients possess only a vague understanding of their obligations and own ability to service institutional loans. After a first loan, some discover that the relationship does not work for them. Early exit is then a predictable outcome of a (mutual) investment in acquiring information about the potential of the relationship (Gonzalez-Vega et al., 2006). The sooner the parties find out that the relationship is not attractive to either one of them, the less costly this investment in learning is. Information management by both parties in the contract, prior to and during the evolution of the relationship, is a key component of microfinance as a learning undertaking. Both parties invest in this exercise—mostly learning by doing— and benefit from the sunk costs of the accumulation of relevant past information (GonzalezVega, 2019). The value of this information thus influences the decision to stay or leave the relationship. In later cycles, exit is mostly due to adverse shocks that the borrower cannot overcome, despite rescheduling offers by the institution. It may also respond to changes in borrower circumstances (enterprise growth), which make other providers of financial services more attractive.

APPLIED EXERCISES The MIDE dataset allows interesting, applied exercises. For example, we explore the marginal productivity of an increasing loan amount for each borrower, in terms of the resulting additional business surplus. Two ways of measuring this productivity effect are as an elasticity (what percentage increase in surplus is induced by a 1 percent increase in the loan amount) or as a per dollar incremental outcome (if the loan amount is increased by one dollar, how much does the surplus increase?). A clarification about the timeline is useful. At any time (t), the applicant requests a (new) loan and the screening process so triggered creates a MIDE record of the business surplus (St). The previous loan (Lt – 1) would have facilitated this outcome. If still found creditworthy, the loan is granted in amount Lt. At time (t + 1), the surplus facilitated during the period of loan Lt is measured again (St + 1), and a new loan is granted (Lt + 1), and so on. When Lt – 1 = 0, that is, when the applicant has not received any institutional loan yet, the surplus S1—corresponding to self-financing—would have been generated by employing the

Measuring the evolution of client vulnerability  235

applicant’s own resources (generated by savings efforts or obtained from informal gifts or loans). This is the surplus measured at the time of the first application and used to rank the new client, according to vulnerability status. The change in surplus that follows (St + 1 – St) would be the marginal contribution of borrowing (for the first time) from the institution, recognizing that S1 would have resulted from the applicant’s savings and access to other—informal—sources of credit (friends, relatives, moneylenders, input suppliers) or even competing institutional lenders. Whatever impact the loan may have is the marginal impact of one additional source of finance (Dahal and Fiala, 2018; Ogden, 2020; Morduch, 2020). If the vector of prior informal sources of funds (unknown to MIDE) is ignored and a prior institutional loan does not exist, S1 is equaled to zero in the denominator, and the elasticity would be undefined. To properly measure this first elasticity, the client’s own endowment of resources, to be enlarged by the first loan, is needed (but unknown to MIDE). For this reason, the first elasticity, assumed to be very high (as suggested by the degree to which these applicants are credit-constrained under exclusion) and representative of the impact of inclusion in institutional markets, is not reported in Figure 13.6. Gender Focus: The Marginal Productivity of Increasing Loan Size Figure 13.6 shows values for the mid-point arc elasticity of business surplus with respect to loan amounts, namely ℮ = [percentage change in business surplus during cycle (t), given a percentage change in loan amount from that disbursed in cycle (t – 1)]. The figure displays values for women and for men, and for four groups of clients, according to the distribution of pre-loan per capita surpluses: Q1, those with pre-loan surpluses below or equal to that corresponding to the first quartile (percentile 25); Q2, those with surpluses below or equal to the median and above the first quartile; Q3, those with surpluses below or equal to the third quartile and above the median; and Q4, those with surpluses above the third quartile. To abbreviate, the four groups are designated as the first, second, third, and fourth quartiles. Each group’s median is centered at the percentiles 12.5, 37.5, 62.5, and 87.5, respectively. The computation follows (13.1).

Dls i : j = median éë D ( surplus j + 1, j) /D ( disbursement amount i, j) ùû (13.1)

The data come from the 2015 cohort of Banco Adopem clients, as an example. Exercises for other institutions generate similar results. Median per capita surpluses with respect to the poverty line differ by gender, for each of the quartile groups. For women, these values are 0.46, 0.83, 1.27, and 2.04 times the poverty line. For men, these values are 0.56, 0.95, 1.42, and 2.18 times the poverty line. Differences by gender show that, in each of the quartile groups, men exhibit per capita surpluses higher than women—namely, men are less poor or less vulnerable than women, within each group. For men, the median per capita surplus was 1.22 times that of women, in the first quartile, but it was only 1.07 times for those in the fourth quartile. Thus, even when contrasted by gender, there is convergence in per capita surpluses. The initial disadvantage of women is gradually reduced as per capita surpluses increase. This is another way of thinking about improvements in income distribution (equity). Are these differences significant? The corresponding t-test shows that, in 2015, women starting as clients of Banco Adopem enjoyed average per capita surpluses significantly lower than those enjoyed by male clients, with a level

236  Handbook of microfinance, financial inclusion and development

Note:  Computed excluding clients not vulnerable before the first loan. Q1, pre-loan surplus ≤ first quartile; Q2, first quartile < surplus ≤ median; Q3, median < surplus ≤ third quartile, and Q4, third quartile < surplus.

Figure 13.6 Median business surplus elasticities with respect to loan amounts and median incremental monthly surplus in response to loan amount changes, by gender, per capita surplus, and loan cycle (Banco Adopem 2015 cohort) of significance of 0.001. A Kolmogorov-Smirnoff test allows rejection of the null hypothesis that the per capita surplus distributions for men and women come from the same reference distribution (p = 0.001). A plot of the cumulative distributions of per capita surplus indicates that the distribution for men is entirely to the right of the distribution for women, reflecting stochastic dominance

Measuring the evolution of client vulnerability  237

of the first order. For any percentile of the distribution, the per capita surplus of men is higher than for women at the same level of the distribution. The distribution for men shows a higher variance of relative per capita surpluses (0.432) than the distribution for women (0.408). However, for women, the coefficient of variation (55.7) is higher than for men (51.1). Thus, not only are women poorer at the time when they are incorporated in the BBVAMF portfolios, but they come from a more dispersed distribution. This suggests the desirability not only to think of women as a different clientele segment but also to appreciate that the target population includes women in different initial circumstances and who will, as a result, demand different products and benefit from different channels for their delivery. First, almost all elasticities are, for the corresponding quartile and loan cycle, higher for women than for men. This suggests that women are initially and throughout more credit-constrained than men (which may, in part, explain their deeper poverty). A household firm is more credit-constrained if, given its productive opportunity, it has so few resources of its own that it can barely take advantage of the opportunity and, as a result, its marginal rate of return is higher than elsewhere. Extremely high transaction costs as well as information and incentive imperfections explain the presence of credit constraints in poorly developed markets (Jaffee and Modigliani, 1969; Gonzalez-Vega, 1976; Jaffee and Russell, 1976). This result confirms predictions about gender gaps (Malapit, 2012; Fiala, 2018). This key outcome (higher elasticities of per capita surpluses with respect to loan size for women than for men) justifies a gender focus—beyond a core set of values on women’s empowerment—for an organization interested in poverty alleviation because a 1 percent increase in a client’s initial loan amount leads to a higher percentage increase in surplus (that is, more poverty alleviation) if the disbursement is for a woman rather than for a man. This is the case even when ignoring the implications of different consumption preferences, which assume that women’s choices contribute more to household well-being (Davies and Zhang, 1997; Haddad et al., 1997; Duflo and Udry, 2004). Given diminishing marginal returns, women’s surplus increments may decline rapidly and, at some point, additional funds loaned to a man would trigger a greater household surplus increase. In the absence of transaction costs and other frictions, an institution solely interested in poverty alleviation would attempt to optimize its portfolio composition by equating the marginal product of loan size for borrowers of both genders. One consequence of the higher elasticity observed would be that the institution ends up having more women than men in its portfolio. Indeed, women represent 60 percent of the borrowers in the BBVAMF Group. The institution may have other goals, in addition to poverty alleviation, that drive this result (Fernandez-Lord, 2019). In general, a wise and prudent institution would balance the gender composition of its portfolio, to reflect potential impacts on poverty alleviation, considerations from gender goals in its mission, and opportunities to diversify risks by lending to both women and men. Differential barriers and frictions characterize the inclusion of women and men in financial markets and a central question concerns the differential barriers and why, within the excluded population, women are more credit-constrained than men (Field et al., 2010; Cornwall, 2016). An exploration of these barriers would suggest the need to adopt a gender focus in the design of products and channels. Second, for both women and men and for all quartiles, the elasticities decline with each new loan cycle. This reflects diminishing marginal productivity: each 1 percent increase in loan amount induces smaller and smaller percentage increases in surpluses. This suggests that

238  Handbook of microfinance, financial inclusion and development

client businesses are subject to diminishing marginal returns, so additional purchasing power for the enterprise (from the loan) does not increase surpluses as much with each new loan. For these household-firms, diminishing marginal returns arise from the limitation of local market size, agency problems in the control of resources beyond the household, and the costs of reaching formality, among other reasons. As the productivity of growing loan amounts in augmenting business surpluses declines, the value of the client relationship may decline, and the institution would have to strengthen its value offer in some other ways. For example, both client and institution would benefit if the borrower took a qualitative jump into a new, more productive opportunity, through some investment or technological change. This is a riskier proposition for both, requiring new screening abilities for the institution, and access to new types of financial services (deposits for precautionary savings, long-term investment loans, and insurance) for the client (GonzalezVega, 2017). Particularly, for poor, vulnerable household enterprises, deposit facilities may play a stronger role than credit in encouraging innovation and investment in lumpy, indivisible capital goods (Gonzalez-Vega et al., 2021). Third, with each loan cycle, the elasticities decline more rapidly for men than for women. If women initially are more credit-constrained, under self-financing they might not have yet encountered a region of too rapidly diminishing returns, and they can increase the size of their businesses without suffering this challenge, while less credit-constrained men, if they had initially been able to exploit their opportunities further, may be operating already in a region of more rapidly diminishing returns. Figure 13.6 reports incremental monetary changes, as ∆ = (St + 1 − St) / (Lt − Lt – 1), indicating the increase in the dollar value of surpluses given an increase in the dollar amount of loans. For example, a US$100 increase in amount between the first and second loan boosts median business surpluses by US$44 per month (US$525 per year) for first-quartile women and by US$33 per month (US$390 per year) for first-quartile men. In almost all cases, the incremental change is greater for women than for men, adding a metric to the extent to which women are more credit-constrained than men and can benefit more from financial inclusion. This implies gross marginal rates of return of 525 percent (women) and 390 percent (men) per year, consistent with the literature and more than sufficient to pay market interest rates (De Mel et al., 2009). An extra US$44 a month represents US$1.46 a day, a substantial increase for the extremely poor. As with elasticities, the incremental change in surplus due to extra loan amounts declines with each new cycle. For example, a US$100 increase in amount between the second and third loans boosts median monthly business surpluses by US$34 (US$411 per year) for first-quartile women and US$20 (US$237 per year) for first-quartile men, less than in the case of the earlier loan. Finally, with each loan cycle, the incremental change in surplus declines more rapidly for men than for women. While, on average, women in each quartile enter into credit relationships with per capita monthly surpluses lower than men, additional loan amounts are more productive for them, and the gap gradually closes. In these circumstances, additional loan amounts for women (rather than men) contribute more to both poverty alleviation and gender parity. COVID-19 Impacts and Response Large segments of the clientele were adversely affected by the COVID-19 pandemic and sharp reduction in productive activity, due to health measures—that locked down sectors of the economy—and to a generalized decline in demand and in the availability of inputs,

Measuring the evolution of client vulnerability  239

particularly imported. Financial inclusion gains were in danger of reversing (Alonso Gispert et al., 2020). During adverse systemic shocks, when assets are destroyed, sales collapse, and surpluses may be insufficient for sustaining household welfare, other financial services (liquid precautionary reserves held securely in a deposit, remittances and government assistance efficiently transferred, insurance used to cope with losses) help the target population not to fall into poverty traps. The broad range of services offered by the Group and efforts to sustain client relationships became particularly valuable, as was the institutional sustainability guaranteeing that relationships would be continued when credit became available again for the recovery. During 2020, most portfolios were forcibly frozen, by moratoria decreed by the prudential authorities (forgiving interest payments and postponing borrower obligations). Some loans have been reprogrammed since 2021. MIDE reveals the impact of the shock and helps in answering questions about its influence on key indicators. Because the dataset responds to these questions monthly and relates answers to the time-series history of the clientele, it is a valuable tool. The data paint a broad picture of sharply declining credit portfolios followed by faster-thanexpected recovery, on one hand, and increasing deposit mobilization, on the other. The number of credit clients reached a peak of 959,124 by February 2020. It then declined to 828,928 by October and, a year later (October 2021), it had partially recovered to 847,208 (15 percent increase). The most notable immediate impact was (deliberate) stagnation in the incorporation of new clients as the pandemic progressed, until it slowly resumed in recent months. Almost the entire BBVAMF efforts focused on existing clients and their needs. Special assistance to clients included massive loan rescheduling, emergency loans, development of new channels, increased authority for remote transactions, and cooperation with public entities for government-to-people transfers and access to guarantee funds. The goal was to assure uninterrupted service at all levels and offer clients alternative service channels and sustained interaction, involving social media. This also included the adaptation of financial education activities. In contrast, the number of depositors steadily increased to 2,213,766 by October 2020 and 2,545,018 by October 2021, aided by the rapid advance of digital services. An important driver of the growth in the number of depositors was the participation of the Group in disbursing government emergency transfers, particularly in Colombia. Given credit and deposit trends, the number of net clients increased to 2,419,016 by October 2020 and 2,744,616 by October 2021 (a 13 percent increase over these 12 months). Bancamia accounted for 54 percent, Financiera Confianza for 26 percent, and Banco Adopem for 15 percent of the clientele. The gross credit portfolio increased by 7 percent, the flow of disbursements increased by 18 percent, the average loan amount declined by 2 percent, and the rate of arrears of more than 30 days increased to 5.4 percent of the portfolio. Relevant questions answered by MIDE are, how do per capita surpluses behave during the crisis? What percentages of the non-poor clients fall back into poverty? What are the paths of post-pandemic sales, asset accumulation, and business surpluses? How do they vary by gender, sector of economic activity, urban or rural location, and even across countries in response to different policy interventions? (Garcia Van Gool, 2021). During the crisis, BBVAMF kept reaching its target clientele. The data show a reduction in the rate of exit from poverty. After four loan cycles, 62 percent rather than 68 percent of the initially poor had escaped poverty, while 17 percent rather than 15 percent of the not poor had fallen into it. Net poverty

240  Handbook of microfinance, financial inclusion and development

alleviation was 36 percent rather than 44 percent. Between March and September 2020, entry into poverty exceeded exits, and a negative net reduction of poverty ensued. The volatility of incomes increased, combined with slower rates of growth of sales and surpluses. Despite these less favorable outcomes, during 2020, over 350,000 microentrepreneurs gained access to formal financial services for the first time, while 73,793 migrants (from Venezuela) and refugees became BBVAMF clients. Women are heavily concentrated in the—mostly informal—trade and services sectors, which were more heavily impacted by lockdown measures enacted to address the pandemic. Increased household and caring obligations have reduced the time available for women to tender to their businesses. In contrast, the reduction of activity was less severe in rural areas, thereby offering some portfolio diversification opportunities. With the recent recovery of lending, farmers—closely associated with food value chains—have been among the most active borrowers, while clients linked to transportation and construction have been delayed in their recovery. Thus, the different dynamics of sectors of economic activity matter much in the evolution of impacts of the pandemic. The pace of digitalization accelerated. While this reduces operational costs and increases efficiency, the gains are not equally distributed, and the consequences might be regressive. Experts warned about gender gaps in digitalization and the differential access between rich and poor, urban and rural (Hilbert, 2011; Antonio and Tuffley, 2014; Mariscal et al., 2019). BBVAMF is making exceptional efforts—including greater emphasis on financial education and the introduction of satellite connectivity in remote areas—to close these gaps. A major challenge is uncertainty about the intensity, length, and speed of the recovery and the shape of the “new normal” (Gonzalez-Vega, 2022). The associated structural changes might have rendered credit scoring based on the old realities useless. This has required constant adjustment and added flexibility in risk management. In forecasting client demand and behavior, probabilities of impact on surpluses are being used for segmentation, considering geographic location (urban and rural and population density), sectors of economic activity, asset levels, and other debt. What challenges emerge for a microfinance business model based on personalized frequent contact as the basis of client relationships? The BBVAMF attempts to sustain personalized contact with clients, leveraging mobile phone and digital interactions followed by—less frequent—face-to-face encounters. Revised regulatory frameworks in response to the pandemic created barriers to microfinance portfolio growth, while the introduction of repressive regulation, such as interest rate ceilings, is making it harder to reach the most difficult clientele. BBVAMF operations in five different countries offer opportunities to assess and compare the impact of differential government actions. Applied exercises may explore the uneven consequences of COVID-19 on diverse BBVAMF clientele. In some ways, the pandemic offers the equivalent of a natural experiment in learning about these effects.

CONCLUSIONS We have examined the features of MIDE as an institutional innovation in knowledge management that has created a versatile data set and a culture of connecting data and practice. The BBVAMF focuses on learning about clients at the individual household-firms level—to build long-term relationships—and about the universe of clients, by identifying the shape and evolution of whole distributions. As a client tracking exercise, MIDE makes it possible to

Measuring the evolution of client vulnerability  241

identify a series of stylized facts that help in understanding the characteristics of client segments and how BBVAMF accompanies them in their business journey. This has led to several conclusions. 1. There is no mission drift within the BBVAMF Group. Its institutions reach the populations targeted in its mission. The share of vulnerable clients (between 80 percent and 90 percent) and of poor clients (around 40 percent) has stayed stable. 2. Despite differences across institutions, due to context, strikingly similar outcomes reflect the success of BBVAMF in clearly defining and instilling in its institutions’ ambitious outreach expectations, within a framework of sustainability and constrained profit-optimization. 3. As client relationships grow longer, high proportions of borrowers exit their initial poverty status. For the 2011–2019 cohorts, on average 63 percent of those initially poor were no longer poor at the end of the fourth loan cycle, while 13 percent of those originally not poor had fallen into poverty. This represented a 44 percent net reduction in the incidence of poverty. 4. Initial vulnerability matters. Clients with monthly per capita surpluses below the extreme poverty line (0.35 times this line) took an average of four and a median of five loan cycles to escape poverty. Clients below the poverty line (0.75 times this line) took an average of two and a median of three cycles to escape. This reflects the success of selection practices in revealing the potential of clients for improving their lives. 5. Average and median magnitudes of all key indicators (sales, assets, business surplus) strictly increase with each loan cycle. Client gains are greater, the longer the length of the relationship. 6. Average and median magnitudes increase at a declining rate, reflecting diminishing marginal returns as the household firm grows. This creates challenges for clients, in expanding their productive opportunities, and for institutions, in appropriately matching client demands with an evolving offer of services. 7. Marginal rates of return of the clients’ businesses are high, as clients are initially creditconstrained, but these rates decline over time. The extent to which highly heterogeneous clients are initially credit-constrained determines their comparative gains from access to BBVAMF services. 8. There is significant distribution dispersion, reflecting ample heterogeneity within the clientele. The distribution of assets is most unequal, followed by the distribution of sales, while the distribution of business surpluses is least unequal. When the less dispersed business surpluses are combined with a highly unequal initial distribution of assets, a highly unequal distribution of gross rates of return on assets emerges. This broad dispersion creates opportunities for financial intermediation to improve the productivity of available resources. 9. The poorer the client is, the gains from the relationship are proportionately greater. Thus, a large share of BBVAMF clients not only improve their per capita surpluses and leave the ranks of the poor, but inequality in their income distribution declines. 10. There is convergence, with each new loan cycle, of the mean and median per capita surpluses of extremely poor, poor, and vulnerable clients. By the first loan, the mean per capita surplus is 5.0 times and the median per capita surplus is 4.5 times higher for vulnerable compared to extremely poor clients. By the fifth loan, the mean and median per capita surplus are both 1.8 times higher for vulnerable than extremely poor clients.

242  Handbook of microfinance, financial inclusion and development

11. Convergence reflects higher marginal rates of return and higher rates of growth of per capita surpluses of the extremely poor (on average 40.9 percent per cycle), compared to poor but not extremely poor (22.0 percent) and to vulnerable but not poor clients (9.0 percent). The corresponding median rates of growth of per capita surpluses are 31.1 percent, 14.5 percent, and 4.4 percent per cycle. Even if the productive opportunities of the extremely poor are in absolute terms less attractive, by being more strictly credit-constrained, their marginal rates of return are initially higher. Diminishing marginal returns explain the convergence. 12. The virtuous distributional outcome requires lending technologies capable of incorporating previously excluded clients—with no prior credit histories—and correctly identifying the marginal returns of their enterprises, to grant them loan amounts that they can repay and leave a surplus that allows the accumulation of wealth and improves their lives. 13. The entire distribution of client surpluses for each new loan cycle completely dominates the previous distribution. For any percentile of the distribution, each successive cycle is associated with a higher per capita surplus. The identity of the household firm matters, though, as the same client may be ranked in a different percentile of the distribution, as some clients gain, and others lose in their ranking. 14. Each cycle there is some attrition in the cohort of clients. Most exits from the relationship happen after the first loan, reflecting information imperfections, investments in learning by both institution and borrower, and a broad offer to avoid rejection of potentially creditworthy applicants, given the BBVAMF outreach mission. 15. Gender differences in income are statistically significant. Initially, men are less vulnerable than women, possibly in reflection of larger asset holdings and less binding credit constraints. 16. Elasticities of surplus with respect to loan amounts (indicating marginal productivity of additional loan amounts in increasing surpluses) are higher for women than for men because women are more credit-constrained than men. This justifies a gender focus by an institution interested in poverty alleviation. A 1 percent increase in loan amount is associated with a higher percentage increase in surplus for women than for men. As surplus increments decline rapidly, portfolios include both women and men. 17. The COVID-19 pandemic induced reductions and later recovery of loan portfolios, in contrast with steadily increasing deposit mobilization. The rates of exit from poverty as well as rates of growth in sales and surpluses declined, risk management had to become more flexible, and digitalization increased, assisting in sustaining contact with clients.

NOTES *

The authors are grateful for the support of Javier Flores Moreno, Director General, and Stephanie Garcia Van Gool, Director of Impact Measurement and Strategic Development, at the BBVA Microfinance Foundation. Views are those of the authors and are not necessarily shared by the BBVA Microfinance Foundation and the BBVA bank. 1. We will indistinctly refer to the BBVAMF, the Foundation, or the Group. 2. Data are reported as of December 2019, unless otherwise noted. All data are in US dollars. Much of these data and additional information can be retrieved from the Foundation’s website (http:// mfbbva​.org​/en/) and the Foundation’s Social Performance Report issued every year. 3. For 2019, for the five countries, extreme poverty lines ranged from US$32 to US$59 monthly per capita income in rural areas and from US$39 to US$71 in urban areas. Poverty lines ranged from

Measuring the evolution of client vulnerability  243

US$54 to US$107 in rural areas and from US$90 to US$144 in urban areas. Vulnerability lines ranged from US$161 to US$321 and from US$269 to US$433, respectively. 4. Graphs about the Group include only clients of Banco Adopem, Bancamia, and Financiera Confianza (93 percent of the total). Clients who did not demand new loans or were excluded due to default are not included in the data for dates beyond their exit. 5. Fondo Esperanza, not considered here because, due to differences in its village banking technology, it reaches even poorer clients. 6. Amounts in real terms are at 2019 prices. They are adjusted from the nominal amounts in local currencies, using the GDP deflators for each year (base 2019 = 100) and converting into US dollars using the exchange rate as of December 31, 2019. GDP deflators may not reflect inflation rates relevant for these populations, while to the extent to which exchange rates do not reflect purchasing power parities, country comparisons are illustrative. As pure fractions, rates of growth suffer less from these limitations, but here they are compound over loan cycles that get longer as the relationship ages. For this reason, the CRGs used here are approximations.

REFERENCES Adams, Dale W (1998). “Altruistic or Production Finance? A Donors’ Dilemma.” Mwangi S. Kimenyi, Robert C. Wieland, and J.D. Von Pischke (eds). Strategic Issues in Microfinance. Aldershot: Ashgate, 87–98. Alonso Gispert, Tatiana, Erik Feyen, Tatsiana Kliatskova, Davide Salvatore Mare, and Matthias Poser (2020). COVID-19 Pandemic: A Database of Policy Responses Related to the Financial Sector. Washington, DC: World Bank Finance, Competitiveness, & Innovation Global Practice. Antonio, Amy, and David Tuffley (2014). “The Gender Digital Divide in Developing Countries.” Future Internet, 6 (4): 673–687. Armendariz, Beatriz, and Ariane Szafarz (2011). “On Mission Drift in Microfinance Institutions.” Beatriz Armendariz and Marc Labie (eds). The Handbook of Microfinance. London: World Scientific Publishing Ltd., 341–366. Baltensperger, Ernst (1978). “Credit Rationing: Issues and Questions.” Journal of Money, Credit and Banking, 10 (2):170–183. Bamberger, Michael (2016). “The Importance of a Mixed Methods Approach for Evaluating Complexity.” Michael Bamberger, Jos Vaessen, and Estelle Raimondo (eds). Dealing with Complexity in Development Evaluation. London: SAGE. Banerjee, Abhijit V., Emily Breza, Esther Duflo, and Cynthia Kinnan (2019). “Can Microfinance Unlock a Poverty Trap for Some Entrepreneurs?” NBER Working Paper 26346. BBVA Microfinance Foundation (2019). Social Performance Report 2019. http://www​.fun​daci​onmi​crof​ inan​zasbbva​.org​/informes​/2019​/en ​/ FMBBVA​.pdf Beneke de Sanfeliu, Margarita, and Claudio Gonzalez-Vega (2000). “Dynamics of Rural Household Incomes in El Salvador: 1995–1997 Panel Results.” Annual Meetings of the Latin American and Caribbean Economics Association, Rio de Janeiro, Brazil, October 12–14. Berger, Allen N., and Gregory F. Udell (2002). “Small Business Credit Availability and Relationship Lending: The Importance of Bank Organizational Structure.” Economic Journal, 112: F32–F53. Chaves, Rodrigo A., and Claudio Gonzalez-Vega (1996). “The Design of Successful Rural Financial Intermediaries: Evidence from Indonesia.” World Development, 24 (1): 65–78. Collins, Darryl, Jonathan Morduch, Stuart Rutherford, and Orlanda Ruthven (2009). Portfolios of the Poor: How the World’s Poor Live on $ 2 a Day. Princeton, NJ: Princeton University Press. Collins, Daryl, Liz Larson, and Abby Butkus (2019). Pathways to a Better Life: The Intricate Role of Digital Finance in Reaching the Sustainable Development Goals. Mapping the Journey from Financial Service Usage to Improving Low-Income People’s Lives. Boston, MA: UNCDF and BFA Global. Cornwall, Andrea (2016). “Women’s Empowerment: What Works?” Journal of International Development, 28 (3): 342–359. Dahal, Mahesh, and Nathan Fiala (2018). “What Do We Know about the Impact of Microfinance. The Problems of Power and Precision.” Ruhr Economic Papers, 756.

244  Handbook of microfinance, financial inclusion and development

Davies, James B., and Junsen Zhang (1997). “The Effects of Gender Control on Fertility and Children’s Consumption.” Journal of Population Economics, 10 (1): 67–85. De Janvry, Alain, Craig McIntosh, and Elisabeth Sadoulet (2010). “The Supply- and Demand-side Impacts of Credit Market Information.” Journal of Development Economics, 93 (2): 173–188. De Mel, Suresh, David McKenzie, and Christopher Woodruff (2008). “Returns to Capital in Microenterprises: Evidence from a Field Experiment.” Quarterly Journal of Economics, 123 (4): 1329–1372. De Mel, Suresh, David McKenzie, and Christopher Woodruff (2009). “Are Women More Creditconstrained? Experimental Evidence on Gender and Microenterprise Returns.” American Economic Journal: Applied Economics, 1 (3): 1–32. Deaton, Angus (2013). The Great Escape: Health, Wealth and the Origins of Inequality. Princeton, NJ: Princeton University Press. Deaton, Angus (2020). “Randomization in the Tropics Revisited: A Theme and Eleven Variations.” Florent Bédécarrats, Isabelle Guérin and François Roubaud (eds). Randomized Controlled Trials in the Field of Development: A Critical Perspective. Oxford: Oxford University Press. Deaton, Angus, and Nancy Cartwright (2018). “Understanding and Misunderstanding Randomized Control Trials.” Social Science and Medicine, 210: 2–21. Di Placido, Giovanni (2020). Measuring Clients’ Intangible Financial Attributes: Cases in Colombia, Peru and the Dominican Republic. Madrid: BBVAMF Working Paper. Duflo, Ester, and Christopher Udry (2004). “Intrahousehold Resource Allocation in Cote D’Ivoire: Social Norms, Separate Accounts, and Consumption Choices.” NBER Working Paper Series No. 10498. El-Zoghbi, Mayada (2019). “Toward a New Impact Narrative for Financial Inclusion.” CGAP Research and Analysis, October. Fernandez-Lord, Laura (2019). “El Momento Es Ahora.” Progreso No. 18, March. Fiala, Nathan (2018). “Returns to Microcredit, Cash Grants and Training for Male and Female Microentrepreneurs in Uganda.” World Development, 105 (C): 189–200. Field, Erica, Seema Jayachandran, and Rohini Pande (2010). “Do Traditional Institutions Constrain Female Entrepreneurship? A Field Experiment on Business Training in India.” American Economic Review: Papers & Proceedings, 100: 125–129. Field, Erica, Rohini Pande, Natalia Rigol, Simone Schaner, and Charity Troyer Moore (2016). “On Her Account: Can Strengthening Women’s Financial Control Boost Female Labor Supply?” Working Paper. Cambridge, MA: MIT Press. Fleisig, Heywood, and Nuria De la Peña (2003). “Legal and Regulatory Requirements for Effective Rural Financial Markets.” International Conference on Paving the Way Forward for Rural Finance. Washington, DC: WOCCU and USAID. Garcia Van Gool, Stephanie (2021). “Nuestra Evolución en Gestión de Impacto.” BBVAFM Report, December. Gomez-Soto, Franz, and Claudio Gonzalez-Vega (2007). “Determinantes del Riesgo de Liquidez y Volatilidad Diferenciada de los Depósitos en el Sistema Financiero Boliviano. Desempeño de las Entidades de Microfinanzas ante Múltiples Shocks Sistémicos.” Latin American Journal of Economic Development, 8: 53–86. Gonzalez-Vega, Claudio (1976). “On the Iron Law of Interest Rate Restrictions: Agricultural Credit Policies in Costa Rica and in Other Less Developed Countries.” Ph.D. Dissertation. Department of Economics, Stanford, CA: Stanford University. Gonzalez-Vega, Claudio (1977). “Interest Rate Restrictions and Income Distribution.” American Journal of Agricultural Economics, 59 (5): 973–976. Gonzalez-Vega, Claudio (1984). “Credit Rationing Behavior of Agricultural Lenders: The Iron Law of Interest Rate Restrictions.” Dale W. Adams, Douglas H. Graham and J.D. Von Pischke (eds). Undermining Rural Development with Cheap Credit. Boulder, CO: Westview Press, 120–132. Gonzalez-Vega, Claudio (1998). “Do Financial Institutions Have a Role in Assisting the Poor?” Mwangi S. Kimenyi, Robert C. Wieland, and J.D. Von Pischke (eds). Strategic Issues in Microfinance. Aldershot: Ashgate, 11–26. Gonzalez-Vega, Claudio (2003). “Deepening Rural Financial Markets: Macroeconomic, Policy, and Political Dimensions.” International Conference on Paving the Way Forward for Rural Finance. Washington, DC: WOCCU and USAID.

Measuring the evolution of client vulnerability  245

Gonzalez-Vega, Claudio (2011). “Microfinance, Heterogeneity, and the Balance between Financial and Non-financial Services.” Plenary Session on Using Microfinance to Ensure Sustainable Livelihoods and Food Security while Mitigating Climate Change, The Global Microcredit Summit. Valladolid, Spain, November 14–17. Gonzalez-Vega, Claudio (2013). Innovation in the Design of Formal Financial Services for Greater Inclusion. Turin: The Mastercard Foundation Symposium on Financial Inclusion, July. Gonzalez-Vega, Claudio (2017). “The Barriers to Agricultural Financing and the Emergence of New Models of Financial Intermediation for the Sector.” Emilio Hernandez (ed.). Innovative Risk Management Strategies in Rural and Agriculture Finance. The Asian Experience. Rome: FAO. Gonzalez-Vega, Claudio (2019). “The Evolving Role of the Client Relationship.” Rodolfo Quiros, Claudio Gonzalez-Vega, and Pedro Fardella (eds). Do Clients Still Matter? Contrasts and Evolution. San Jose: Calmeadow, 18–21. Gonzalez-Vega, Claudio (2020). “The Future of Microfinance as Knowledge Management: The Experience of the BBVA Microfinance Foundation.” Ira W. Lieberman, Paul DiLeo, Todd A. Watkins, and Anna Kanze (eds). The Future of Microfinance. Washington, DC: Brookings Institution Press, 103–126. Gonzalez-Vega, Claudio (2022). “Microfinanzas en la Postpandemia. Mirando al Futuro, Recuperando el Pasado.” Progreso, forthcoming. Gonzalez-Vega, Claudio, and Jeffrey Poyo (1985). “Rural Savings Mobilization in the Dominican Republic: Challenges, Accomplishments and Lessons.” Economics and Sociology Occasional Paper No. 1226. Columbus, OH: The Ohio State University. Gonzalez-Vega, Claudio, and Marcelo Villafani-Ibarnegaray (2011). “Microfinance in Bolivia: Foundation of the Growth, Outreach and Stability of the Financial System.” Beatriz Armendariz and Marc Labie (eds). The Handbook of Microfinance. London: World Scientific Publishing Ltd. Gonzalez-Vega, Claudio, Mark Schreiner, Richard L. Meyer, Jorge Rodriguez, and Sergio Navajas (1997). “The Challenge of Growth for Microfinance Organizations: The Case of Banco Solidario in Bolivia.” Hartmut Schneider (ed.). Microfinance for the Poor? Paris: OECD. Gonzalez-Vega, Claudio, Geoffrey Chalmers, Rodolfo Quiros, and Jorge Rodriguez-Meza (2006). “Hortifruti in Central America: A Case Study about the Influence of Supermarkets on the Development and Evolution of Creditworthiness among Small and Medium Agricultural Producers.” Micro Report No. 57. Washington, DC: USAID. Gonzalez-Vega, Claudio, Isai Guizar, and Mario J. Miranda (2021). “¿Crédito o Depósitos? Influencias en la Adopción y Retención de Tecnologías Avanzadas.” Eduardo Lizano Fait, Gloriana Ivankovich and Josue Martinez (eds). Ensayos en Honor de Miguel Ángel Rodríguez Echeverría. San Jose, Costa Rica: Academia de Centroamerica. Guerrero, Jose A. (1988). Determinants of Successful Rural Deposit Mobilization: Banco Agricola in the Dominican Republic. Master’s Thesis, Department of Agricultural, Environmental and Development Economics. Columbus, OH: The Ohio State University. Guizar, Isai, Claudio Gonzalez-Vega, and Mario Miranda (2015). “Uneven Influence of Credit and Savings Deposits on the Dynamics of Technology Decisions and Poverty Traps.” Fourth European Research Conference on Microfinance. Geneva: University of Geneva, June 1–3. Haddad, Lawrence James, John Hoddinott, and Harold Alderman, eds (1997). Intrahousehold Resource Allocation in Developing Countries. Models, Methods and Policies. Baltimore, MD: Johns Hopkins University Press. Hartarska, Valentina (2005). “Governance and Performance of Microfinance Institutions in Central and Eastern Europe and the Newly Independent States.” World Development, 33 (10): 1627–1643. Hartarska, Valentina, and Claudio Gonzalez-Vega (2006). “What Affects New and Established Firms’ Expansion? Evidence from Small Firms in Russia.” Small Business Economics, 27 (2): 195–206. Hausmann, Ricardo (2016). “Economic Development and the Accumulation of Know-how.” Welsh Economic Review, 24: 13–16. Hernandez, Emilio (2020). “Financial Inclusion for What?” CGAP Blog on Impact and Evidence in Financial Inclusion, February. Hilbert, Martin (2011). “Digital Gender Divide or Technologically Empowered Women in Developing Countries? A Typical Case of Lies, Damned Lies, and Statistics.” Women’s Studies International Forum, 34 (6): 479–489.

246  Handbook of microfinance, financial inclusion and development

Holtmann, Martin (2002). Developing Staff Incentive Schemes. Nairobi: MicroSave. Jaffee, Dwight M., and Franco Modigliani (1969). “A Theory and Test of Credit Rationing.” American Economic Review, 59 (5): 850–872. Jaffee, Dwight M., and Thomas Russell (1976). “Imperfect Information, Uncertainty, and Credit Rationing.” The Quarterly Journal of Economics, 90 (4): 651–666. Kabeer, Naila (2019). “Randomized Control Trials and Qualitative Evaluations of a Multifaceted Programme for Women in Extreme Poverty: Empirical Findings and Methodological Reflections.” Journal of Human Development and Capabilities, 20 (2): 197–217. Keeton, William (1979). Equilibrium Credit Rationing. New York: Garland Press. Malapit, Hazel Jean L. (2012). “Are Women More Likely to be Credit Constrained? Evidence from LowIncome Urban Households in the Philippines.” Feminist Economics, 18 (3): 81–108. Mariscal, Judith, Gloria Mayne, Urvashi Aneja, and Alina Sorgner (2019). “Bridging the Gender Digital Gap.” Economics, 13 (9): 1–12. Mauri, Arnaldo (1993). “A Policy to Mobilize Rural Savings in Developing Countries.” J.D. Von Pischke, Dale W. Adams, and Gordon Donald (eds). Rural Financial Markets in Developing Countries. Baltimore, MD: Johns Hopkins University Press. McKenzie David J., and Christopher Woodruff (2006). “Do Entry Costs Provide Empirical Basis for Poverty Traps? Evidence from Mexican Microenterprises.” Economic Development and Cultural Change, 55: 3–42. McKinnon, Ronald (1973). Money and Capital in Economic Development. Washington, DC: The Brookings Institution. Mo, Laura (2020). “Análisis de Fuga.” BBVAMF Working Paper. Morduch, Jonathan (2020). “Why RCTs Failed to Answer the Biggest Questions about Microcredit Impact.” World Development, 127 (3): 104818. Morvant-Roux, Solène, Isabelle Guérin, Marc Roesch, and Jean-Yves Moisseron (2014). “Adding Value to Randomization with Qualitative Analysis: The Case of Microcredit in Rural Morocco.” World Development, 56: 302–312. Naranjo Landerer, Martin (2017). “Quintin.” Progreso, 11 (2), June. http://www​.fun​daci​onmi​crof​i nan​ zasbbva​.org​/revistaprogreso​/quintin/. Ogden, Timothy (2020). “Understanding the Impact of Microcredit.” Ira W. Lieberman, Paul Di Leo, Todd A. Watkins, and Anna Kanze (eds). The Future of Microfinance. Washington, DC: Brookings Press, 103–126. Person, Jennie, and Emilio Hernandez (2019). “Looking Beyond the Average Impact of Financial Inclusion.” CGAP Blog on Impact and Evidence in Financial Inclusion, June 18. Pritchett, Lant, and Justin Sandefur (2015). “Learning from Experiments when Context Matters.” American Economic Review, 105 (5): 471–475. Quiros, Rodolfo, Claudio Gonzalez-Vega, and Pedro Fardella (2019). Do Clients still Matter? Contrasts and Evolution. San Jose: Calmeadow. Readdaway, Alex, Daryl Collins, Liz Larson, and Abby Butkus (2020). Impact Pathways. Mapping the Benefits of Financial Inclusion. Suva: Pacific Financial Inclusion Programme. Robinson, Marguerite S. (2001). The Microfinance Revolution: Sustainable Finance for the Poor. Washington, DC and New York: The World Bank. Rozas, Daniel, and Gabriela Erice (2014). Microfinance and Savings Outreach: What Are We Measuring? Luxembourg: European Microfinance Platform. Schmidt, Reinhard H., and Ingo Tschach (2001). “Microfinance as a Nexus of Incentives.” Working Paper 87, Frankfurt am Main: Department of Finance, Goethe University. Sen, Amartya (1999). Development as Freedom. New York: Random House. Shaw, Edward S. (1973). Financial Deepening in Economic Development. Oxford: Oxford University Press. Stiglitz, Joseph E., and Andrew Weiss (1981). “Credit Rationing in Markets with Imperfect Information.” The American Economic Review, 71 (3): 393–410. Udry, Christopher, and Santosh Anagol (2006). “The Return to Capital in Ghana.” The American Economic Review, 96 (2): 388–393.

Measuring the evolution of client vulnerability  247

Vaessen, Jos, Ana Rivas, Maren Duvendack, Richard Palmer Jones, Frans L. Leeuw, Ger van Gils, Ruslan Lukach, Nathalie Holvoet, Johan Bastiaensen, Jorge Garcia Hombrados, and Hugh Waddington (2014). “The Effects of Microcredit on Women’s Control over Household Spending in Developing Countries: A Systematic Review and Meta-Analysis.” Campbell Systematic Reviews, 10 (1): 1–205. Vasquez, Archibaldo (1986). “Determinants of Household Deposit Behavior in the Dominican Republic.” Master’s Thesis, Department of Agricultural, Environmental and Development Economics. Columbus, OH: The Ohio State University. Villafani-Ibarnegaray, Marcelo, and Claudio Gonzalez-Vega (2007). “Tasas de Interés y Desempeño Diferenciado de Cartera de las Entidades de Microfinanzas ante Múltiples Shocks Sistémicos. ¿Se Cumple el Teorema de Stiglitz-Weiss en las Microfinanzas Bolivianas?” Latin American Journal of Economic Development, 8 (2): 11–52. Vivalt, Eva (2015). “Heterogeneous Treatment Effects in Impact Evaluation.” American Economic Review, Papers and Proceedings, 105 (5): 467–470. Vogel, Robert C. (1984). “Savings Mobilization: The Forgotten Half of Rural Finance.” Dale W. Adams, Douglas Graham and J.D. Von Pischke (eds). Undermining Rural Development with Cheap Credit. Boulder, CO: Westview Press, 248–265. Von Pischke, J.D., and Dale W Adams (1980). “Fungibility and the Design and Evaluation of Agricultural Credit Projects.” American Journal of Agricultural Economics, 62 (4): 719–726.

14. An investor’s perspective on measuring and managing social performance and impact Gregor Dorfleitner, Dina Pons, and Noémie Renier

INTRODUCTION The question of how one can ensure quality financial services that truly benefit the low-income population and go beyond the mere financialization of the people excluded from the financial sector has been and still is at the heart of the preoccupations of impact investors specialized in financial inclusion. The active pursuit of the double (or triple) bottom line, i.e. aiming at financial and social (and environmental) purposes, has characterized and differentiated financial inclusion from the mainstream banking sector since the early days (Lapenu, Brusky, and Sallé, 2017). Dorfleitner, Röhe, and Renier (2017) show that a high share of female borrowers and a low average loan size both increase the chances of financial service providers (FSPs) to receive funding from financial inclusion or microfinance investment vehicles (MIVs), while Arnold et al. (2019) find a similar effect for FSPs charging responsible interest rates. These findings are supported by a practitioner survey, confirming that pursuing impact is central to the mission of a majority of impact investors (Mudaliar et al., 2019). According to Symbiotics Group (2018), increasing access to financial services is one of the primary social goals of MIVs, together with improving the livelihoods of the clients and the creation of work opportunities. However, the definitions and measures of impact remain rather imprecise. Therefore, it appears worthwhile to analyze the reality of financial inclusion investing from the perspective of investors. For this chapter, and building on an industry-wide accepted definition (SPTF, 2017), social performance management (SPM) means “the systems that organizations use to achieve their stated social goals and put customers at the center of strategy and operations”. If an FSP’s social goal is to serve poor rural households, a relevant SPM system would be the use of a poverty-targeting tool to assess clients’ socio-economic profiles. Similarly, developing a rural network of branches to minimize clients’ travel time is another type of SPM system. Impact, on the other hand, when understood in the strict academic sense of the term, is defined as the change caused by an intervention (Duvendack, Palmer-Jones, and Copestake, 2011). To rigorously monitor whether an institution’s poor rural credit product contributes to poverty alleviation, requires tracking clients’ and non-clients’ poverty profiles over time to assess whether one positively evolves faster than the other. It is important to understand that the definition of impact used in the financial inclusion community is different. When talking about signs of the positive social or environmental effect of their work, the term outcome is used. Outcome refers to the change for the end clients that is plausibly associated with an FSP’s services (Spaggiari, 2016). For instance, over a multi-year period, an FSP annually tracks a representative sample of clients’ poverty profiles and witnesses an overall improvement. Because the FSP does not know the evolution of the poverty status of a non-client during the same period, it can only conclude that the outcome of its poor rural credit products on its own clients is at least positive. 248

An investor’s perspective on social performance and impact  249

Through 150 years of discussion on accounting practices, assessing financial performance was made consistent and objective. Through 20 years of collection of field practices and exchange between FSPs, rating agencies, social auditors, and investors, standards for social performance finally have emerged and changed the conversation on social mission intent into an objective one. Continuous efforts to clarify concepts, improve best practices, broaden the use of standardized methodologies and tools, and build knowledge on social performance and impact management are critical to guarantee and enhance the quality of responsible investment practices in financial inclusion. In parallel, the whole impact investment industry, of which financial inclusion is only a subset,1 has been growing stronger thanks to the formulation of the Sustainable Development Goals (SDGs) adopted by the United Nations in 2015, and the adoption of the EU Sustainable Finance Disclosure Regulation (SFDR), which was introduced in 2019 and came into force in March 2021. Mudaliar et al. (2019) define impact investment as “investments made with the intention to generate positive, measurable social and environmental impact alongside a financial return”. While this momentum is positive and critical for achieving the SDGs by 2030, the authors warn about the risks of “impact washing”, i.e. risk of investors conveying a false public image of positive impact. This risk may arise from the lack of a commonly defined measurement of impact (and social returns), and could eventually lead to a misperception of impact investments and divert funding from the achievements of the SDGs. Often perceived as the “grandmother of the impact investment”, the financial inclusion industry has learned several lessons over its journey. After more than 20 years of effort, the financial inclusion sector has emerged as a laboratory and testimony of what can be achieved through the dialogue between investors in competitive markets. The purpose of this chapter is to share and reflect on the best practices that have emerged. Gathering insights from a dozen industry experts (the “Experts”), the authors highlight key achievements, as well as remaining collective insufficiencies in the areas of promotion, measurement, execution, and impact of SPM. To do so, we have structured our analysis in four parts: the next section looks back at the chronology and process that led to the emergence of globally accepted social performance standards, including the progressive shift from the passive “do not harm” approach to “doing good”. In the following sections, the authors distinguish two levels of impact. The third section analyses the impact of MIVs at the level of the investees. The authors show how MIVs integrate impact through their investment process to effectively promote SPM at the level of their investees. In the fourth section, the authors gather examples of contributions of MIVs to build responsible financial inclusion eco-systems. These are typically non-direct activities, harder to monitor, including stirring dialogue with central banks to promote code of lending practices, creation of credit bureaus, or promotion of pricing transparency. Finally, the fifth section opens the debate on what could be further improved by the industry. This section inquires about what differentiates an impact investor from a commercial investor and how the interests of the asset owners and asset managers could be better aligned.

A RETROSPECTIVE OF CONCEPTS With more than three decades of existence, the financial inclusion industry offers a unique example of a process that has led to an objective definition of social performance and more recently also to a definition of impact.

250  Handbook of microfinance, financial inclusion and development

The 1990s: With the Intention to Do Good, Be Financially Sustainable! Providing financial services to the people excluded from the financial sector and positively impacting the final beneficiaries have been the primary objectives of financial inclusion from the onset (Rhyme and Otero, 1992). Yet, in the 1990s, the focus was mainly on financial and institutional sustainability. Financial analysis tools were developed and adapted, while the social performance of the FSPs was often considered as “given”. Financial sustainability, i.e. the ability of an FSP to cover all costs, was considered the necessary condition to reach significant numbers of poor people (Robinson, 2001; CGAP, 2004). Besides, proving that the poor could constructively use and be sustainably provided with a comprehensive range of financial services—which was unthinkable 20 years ago—is why the industry focused its efforts on proving that microfinance could be financially sustainable to achieve its very mission (DiLeo, Watkins, and Kanze, 2017). Building financially sustainable institutions was not the ultimate objective. It was deemed the only way to reach a significant scale and impact far beyond what donor agencies could fund. The 2000s: Good Intentions Are Not Sufficient Starting from the early 2000s, a dozen socially driven FSPs started voicing concerns that focusing mainly on financial performance could overshadow—even undermine—their social mission. In 2002, the French organization CERISE started looking at the questions of social performance definition and measurement. To achieve the social promise of financial inclusion, practitioners gathered and created the Social Performance Taskforce (SPTF) in 2005, a nonprofit membership organization gathering more than 3,000 global stakeholders, which has facilitated the “bottom-up” development of social performance standards. In 2006, CERISE released its third version of the Social Performance Indicators (SPI-3), which provided for the first time clear and measurable indicators for tracking and measuring social performance in an audit framework. This framework inspired financial inclusion rating agencies, which developed the first social rating tool whose aim was to assess whether an FSP is achieving (or is likely to achieve) development objectives (M-Cril, 2005). The SPI-3 framework also inspired investors, such as Incofin Investment Management (Incofin IM), an impact asset manager, which in 2007 developed its first proprietary social audit tool, the ECHOS©. In line with the emerging industry practices at that time, ECHOS© looked at the environment, customer service, human resources, outreach, and social mission. The year 2009 saw the creation of the SMART Campaign, an industry-led, self-regulatory effort aiming at promoting ambitious client protection practices by financial inclusion institutions. Structured around seven principles, the Client Protection Principles (CPPs), the SMART Campaign’s message was to say that out of broad SPM practices, absolute priority shall be put on preventing FSPs from harming or negatively impacting the lives of vulnerable borrowers (the “do not harm” approach). The CPPs defined guidelines on responsible financial inclusion including appropriate product design and delivery, prevention of over-indebtedness, transparency, responsible pricing, fair and respectful treatment of clients, privacy of client data, and mechanisms of complaints. These CPPs were rapidly endorsed by the financial inclusion community.2 Yet, a clear consensus on how to measure implementation did not yet exist. The 2010s: The Emergence of Social Performance Management The sector continued its remarkable expansion, spurred by the channeling of public and private MIV capital into the market. The total estimated MIV assets tripled between 2007 and 2009,

An investor’s perspective on social performance and impact  251

recording growth rates of 86 percent in 2007, 34 percent in 2008, and 25 percent in 2009 (Symbiotics Group, 2011). This period of high growth came to an end in 2010, as the sector faced major turmoil triggered by irresponsible lending practices in a state of India, Andra Pradesh. Some FSPs had failed to live up to their social responsibilities and instead pursued aggressive growth strategies. Investors’ processes and methodologies had somewhat failed to identify and prevent these predatory practices. The Indian crisis seriously tarnished the reputation of MIVs. MIV assets growth dropped by 10 percent in 2010 and redemptions in the financial inclusion investment sector reached some industry records. In the aftermath of the financial crisis and financial inclusion crisis in India, the sector entered the second phase of profound introspection, learning lessons and doubling its efforts to develop more socially responsible, ethical practices and safeguards to avoid harmful effects. Because claiming to have a social mission does not necessarily mean achieving it, the sector had long agreed on the necessity to develop effective social performance measurement tools (Lapenu, Brusky, and Sallé, 2017). Instead of “proving impact”, practitioners focused on “improving impact”, i.e. improving the processes and systems in place to ensure more ethical, sustainable, and responsible financial inclusion. The concept of SPM was developed further and adopted more widely. The social performance then referred to the FSP’s effectiveness in achieving its stated social goals and creating value for clients. If a provider had strong SPM practices, it was more likely to achieve a strong social impact. This period also led to the creation of the Universal Standards of Social Performance Management (USSPM), which were released in 2012 by the SPTF. Resulting from the collaboration amongst a broad range of stakeholders, the USSPM was the first comprehensive manual of best SPM practices. The USSPM were organized around six key dimensions (SPTF, 2019): (1) Define and monitor social goals. (2) Ensure board, management, and employee commitment to social goals. (3) Design products, services, and delivery channels that meet clients’ needs and preferences. (4) Treat clients responsibly. (5) Treat employees responsibly. (6) Balance financial and social performance. While in 2013, the SMART Campaign launched a certification (CPP Certification) tool aiming to externally validate compliance with minimum do not harm standards (CPPs), the USSPM raised the bar of what was expected at the FSP level and provided the investors with clear standards of SPM and benchmarks to assess their partner FSPs. While the Client Protection Principles had focused on “do not harm”, the USSPM were deemed to be more comprehensive moving toward a positive approach to social performance management, beyond client protection (the “doing good” approach). In an attempt to operationalize further the USSPM, CERISE released the fourth version of the Social Performance Indicators (SPI4) in September 2014, translating standards into a measurement tool. The MIVs acknowledged still that the wide adoption of a standard due diligence SPM tool would not only reduce the reporting burden of FSPs but also enable greater accountability, allowing better quality and benchmarking across the industry. This was so far not possible with numerous individual tools. According to Cecile Lapenu from CERISE, the increasing standardization of function of Social Performance and Impact Manager in both

252  Handbook of microfinance, financial inclusion and development

the financial and investment sector as well as the uptake of social due diligence tool such as SPI4 demonstrate not only that social performance can be measured, but moreover is being embedded into operations. Nonetheless, with over 200 indicators, the full SPI4 was deemed too heavy by the investor’s arena to be applied in the context of investment due diligence, as pointed out by some of the Experts. Still wanting to use the SPI4, the SPTF Social Investors Working Group (SIWG), an informal group of investors within the SPTF, decided to work with CERISE to select 80 indicators that were deemed the most relevant among the full SPI4. A dozen MIVs participated and co-funded the process of developing an investorfriendly version of the SPI4: the SPI4 ALINUS (a common social data collection tool for aligning investors’ due diligence with the Universal Standards of Social Performance Management).3 According to Cecile Lapenu, SPI4 ALINUS helps investors track what matters and guides them to make informed decisions using standard social metrics to select and support their investees. As of December 2019, about 30 investment managers and development financial institutions were actively using, training their staff, or strategically planning to use SPI4 ALINUS, representing about 30 percent of the funds investing in microfinance (CERISE, 2020). Today, stakeholders collect data aimed at holding FSPs accountable to their social mission. In practice, this means encouraging them to integrate social intentions into their strategy and management systems, and to monitor them with key performance indicators. From 2015 Onward: Outcome Measurement, the Missing Piece of the “Impact” Puzzle Despite the achievements of the last decade in building a common language around social performance and impact, a big gap has remained, the absence of evidence of positive change at the level of the end clients. This is how the call for the need to measure impact emerged. Anecdotical stories do not suffice anymore to convince asset owners about the FSPs’ capacity to contribute to their impact thesis. Yet, proving the positive impact of financial inclusion is not a new topic. Since the mid-1990s millions of dollars and hours have been spent on hundreds of studies dedicated to impact assessment, with almost 140 major studies dedicated to impact assessment in financial inclusion between 1995 and 2011 (Lapenu, Brusky, and Sallé, 2017). Proving impact, i.e. the causality between an investment and the changes in the life of the final beneficiary, remains challenging, expensive, and subject to methodological flaws leading to mixed results and ethical challenges (Duvendack, Palmer-Jones, and Copestake, 2011). Experimental methods, such as the so-called randomized control trials (RCTs), which borrow their methodology from medical science, have been applied to assess the impact of a development field. Such RCTs are based on numerous hypotheses, which ultimately affect their reliability (Bédécarrats, Guérin, and Roubaud, 2015). Besides, a rigorous impact survey requires a control group, i.e. in the case of microcredit, a group of clients to whom micro-borrowing shall never be offered regardless of their preferences and needs, raising questions of feasibility and ethics. Most studies require also long-term data collection over several years, posing issues of clients’ drop-out and progressive reduction of the studies’ sample. In an industry in constant need of growth to reach economies of scale and higher client centricity, deciding to allocate hundreds of thousands of dollars to an impact survey rather than to a training module for field staff, product development research, or client satisfaction surveys, is a decision seldom made

An investor’s perspective on social performance and impact  253

by investors. This is one of the reasons why impact measurement has long become an elusive goal for the financial inclusion industry. More recently, the debate was reactivated, in particular as non-financial inclusion subsectors were emerging as rival impact investment fields. The Experts highlight that, since its creation, there exists limited consensus on what “impact” is expected within the financial inclusion industry: is it poverty alleviation or only financial inclusion? Is it supporting woman empowerment or local entrepreneurship? Meanwhile, other impact investment sectors have emerged with a more advanced set of outcome indicators. Certain asset managers were involved in clean energy, education, agriculture, or health, rewarding asset managers based on the social outcomes achieved and the public spending saved (Spaggiari, 2016). It is within this context that the financial inclusion industry sought further additional insights into the well-being of its end clients. Several initiatives to measure and track outcomes were launched. The most credible stakeholder to coordinate such new development was again the SPTF. Under its umbrella, several outcome-measurement working groups emerged. Through their work, financial inclusion reshaped its impact narrative (El-Zoghbi and Holle, 2019). From a broad and hard-to-prove “poverty alleviation” mandate, financial inclusion stakeholders tested and agreed that there were four key social goals that financial inclusion contributed to (Sinha and Greenberg, 2017). Through multi-stakeholder working group discussion and field indicator testing, the industry is now building consensus that financial inclusion’s outcome shall be assessed by looking at positive or negative changes in business and entrepreneurship, economic poverty, assets and housing, resilience and vulnerability, and health (SPTF, 2019).

PROMOTING SPM AND IMPACT TOWARD INVESTEES To improve their social performance and impact thesis, the USSPM were enforced by MIVs, through their investment and due diligence processes or financing contracts, sending messages to the FSPs on “how to behave to obtain financing”. By 2017, 99 percent of MIVs were endorsers of the CPPs of the SMART Campaign (Symbiotics Group, 2016). In turn, when assessing the social performance of an organization, MIVs assess the likelihood of this organization achieving its social goals and ultimately contributing to the MIV’s social strategy. This framework assumes that strong SPM increases the likelihood of achieving a positive impact on financial beneficiaries. Thanks to the implementation of the USSPM and the increasing use of assessment tools linked to them (SPI4, CPP Certification, social ratings), investors in financial inclusion can make investment decisions with the insurance that social performance management systems within investees meet certain industry standards. By doing this, they can be held accountable to their asset owners for the alignment of their investment process with their investment and impact strategy. Measuring SPM and Using It to Make Investment Decisions Experts agree that managing and creating impact starts with a clear definition of an impact strategy, which can then be integrated into the investment process. A clear impact strategy refers to the problem and to the target population being addressed and is articulated in clearly defined social performance objectives that are measurable and quantifiable. The authors emphasize that impact investment managers who want to be equally serious about both dimensions of the double bottom line need to devote equally strong processes, policies,

254  Handbook of microfinance, financial inclusion and development

and procedures to measure, monitor, and quantify the achievement of their impact thesis along financial performance. Two types of approaches exist for the selection and screening of socially responsible investments: a negative or a positive approach (O’Rourke, 2003; Renneboog, Ter Horst, and Zhang, 2008). An MIV can opt for either one or a combined approach. In the negative approach, environmental, social, and governance or “ESG”-related exclusion criteria are applied to ensure that the proposed investment is in line with the MIV’s investment strategy (De Corte et al., 2013). These filters aim at safeguarding the reputational risk of the investors and preventing any harmful consequences to the society or environment (in line with the “do not harm” approach). Fund managers can refer to the prevailing exclusion lists published by their asset owners or development financial institutions, which indicates typically the types of projects that they do not want to finance (e.g. trade or production of weapons, alcoholic beverages, tobacco, gambling, etc.) or the types of labor practices that they do not want to be associated with (child labor, missing compliance with minimum standards). Other fund managers go further, in order to intentionally select the most socially relevant investment projects and retain the ones that are most likely to contribute to their impact strategy. Such an approach, therefore, requires to be able to assess and benchmark the social performance of the selected investments. According to the Symbiotics MIV Survey 2018, two-thirds of the MIVs conduct internal social assessments on their financial inclusion investees using tools developed in-house. In the authors’ view, the positive approach distinguishes impact investors from traditional socially responsible investors. This positive approach is expected to gain further momentum in the global impact sector as investors require more accountability and demonstration of the social returns of their investments (De Corte et al., 2013). Industry standards such as the USSPM provide investors with a relevant and universal set of criteria to understand the social performance of the FSPs in which they invest. Besides, this facilitates the dialogue between the MIVs and the FSPs on social expectations, through the use of one common set of social performance indicators. Limited Uptake of Standard Social Audit Tools Used by Investors Although the USSPM and CPPs have been largely promoted and recognized by the investor community, there is some room for improvement when it comes to the true adherence to international standards by MIVs within their investment process. Symbiotics estimates that the majority of MIVs collect outreach indicators, using in-house developed tools to assess the social performance of their investees, but only less than half of them also used tools developed by the industry such as SPI4 ALINUS described in the previous section. Besides, only a third of the investees in the direct financial inclusion portfolio of the MIVs had carried out a CPP assessment process in 2017. Although increasing as a result of continuous sector efforts to encourage responsible financial inclusion, this figure remains on the low side and reflects the need for further enforcement of the USSPM and CPPs. Experts point out that some MIVs are concerned that the use of standardized tools will limit their capacity to promote their impact measurement expertise and differentiate themselves in the eyes of the asset owners. Some also consider such tools too cumbersome and time consuming, or not adapted to their size or strategy. An investment manager spends typically between one and three days of due diligence in the field, and tools such as the SPI4 ALINUS do require significant time to collect all the necessary information in a consistent and reliable manner. Besides the qualitative character of several parameters increases the risk of discrepancies

An investor’s perspective on social performance and impact  255

between assessment and perception of social performance, which may also vary according to the interest of the person conducting the assessment (i.e. investment versus risk manager). Ongoing training is part of the solution, together with the development of independent internal skills and control mechanisms within MIV’s organization. The Experts do recognize also that there exists room for improvement, maybe in the form of modular solutions, with skinnier tools that would improve the uptake of standard SPM reporting, and in fine, the relevance and reliability of SPM data. Building Further the “Social Performance” Business Case While the industry has achieved significant progress in measuring the social performance of investments, how and to which extent social performance influences the final decision remains critical. Some investments may require the decision-making body to compromise between financial and social performance. Therefore, the question of how one can ensure that the impact strategy prevails and is not diluted by financial objectives remains a key challenge. The relationship between the social and financial performance of FSPs has been studied by numerous academics. Nonetheless, such a relationship remains complex and ambiguous. Academicians such as Frank (2008), Hermes, Lensink, and Meesters (2011), and Hartarska, Shen, and Mersland (2013) found supporting evidence of the existence of such a trade-off, i.e. between efficiency/transformation and outreach. Other researchers emphasize a positive relationship between financial and social performance (Quayes, 2012, 2015; Abdullah and Quayes, 2016) and how both are complementary and indispensable to building a solid financial inclusion sector (Rhyne and Otero, 2006). Yet most empirical studies derive their conclusions out of a set of few social indicators, using the (growth in) number of clients as proxies for the breadth of outreach and the percentage of women or average loan size as proxies for assessing the depth of outreach (D’Espallier and Goedecke, 2019). Only limited research has been able to capture the multidimensional character of social performance. Using more than 91 social and financial indicators, Incofin IM found a positive association between the social performance score and the financial performance score (Dewez and Perez 2010). Hoepner et al. (2012) found broadly empirical evidence of a parabolic, non-linear relationship between the social and financial performance of FSPs. Their findings provide evidence that while investments in strong SPM systems can be costly at the onset (i.e. correlated with a lower return on equity), efforts in fine pay off once a “minimum critical mass” is achieved to ensure the trust of the clients and investors, resulting in a U-shape. The U-shape shows that putting clients at the center of any business strategy is ultimately financially rewarding. These findings help investors to understand the synergies and the trade-offs between social and financial performance. Further research should leverage the uptake of standardized social performance measurement tools to look at how synergies and trade-offs evolve over the lifecycle of FSPs, hereby further supporting the investment decision-making process of the MIVs. Clear investment procedures and criteria, requiring a minimum social performance score and FSP’s adherence to certain industry standards also support the MIV investment process. Another possible differentiator for social investors may also be their willingness to accept higher risks or lower financial risk premiums, and conditions that other investors would not accept (Morduch, 2019). Preferential terms can take the form of lower interest rates but also technical assistance, flexible tenors, unsecured loan, or more lenient financial covenants. While SPI4 scores globally have improved from 61 percent (with 138 quality audits) in December 2016 to 65 percent as of October 2019 (for 358 quality audits) (Bauwin, 2019),

256  Handbook of microfinance, financial inclusion and development

the average interest rate charges by MIVs to these FSPs do not take into account their SPM improvements and continues to be based predominantly on country and credit risk consideration. Hence FSPs receive only limited incentives to improve their SPM to avail of funding at better terms and conditions, which can be then passed on and reflected in better conditions for the end borrowers. Promoting Best Governance Practices, Building Socially Responsible FSPs Locally The authors highlight also that several MIVs provide technical assistance to their investee, which complements their investment with grant money dedicated to reinforcing the institutional capabilities of the FSP, be it, for example, through improving their risk management systems, creating audit procedures, or supporting their transformation. Additionally, in the case of equity investment, MIVs can also play an important role in pushing SPM and impact actions to be taken by their FSPs investees during the time of their investment (Drake and Rhyne 2002; Helms, 2006). Specialized equity funds can participate in the governance and institutional robustness of the FSP in which they invest. In particular, poor governance was often considered one of the major weaknesses in the financial inclusion industry (CGAP, Citigroup, and CSFI, 2008). Good governance is all the more needed in high-growth markets where rapidly increasing outreach and capturing market shares can easily lead to social mission drift, lack of resources to invest in SPM processes, and dilution of core founders’ groups in a broader management team through new recruitment. In those instances, strong governance is recognized as a key safeguard (Di Benedetta, Lieberman, and Ard, 2015). Governance mechanisms may indeed play a key role in ensuring a clear strategic vision and transparent organizations that are efficient and accepted by all the stakeholders involved (Lapenu and Pierret, 2006). This is all the more crucial in the transformation of NGOs into regulated financial entities, which requires finding new owners and a new board while keeping in mind both social and financial objectives (White and Campion, 2002). In that respect, balanced board composition is expected to help in balancing the double objective of the FSP (Dorado and Molz, 2005). In particular, the Experts emphasize the critical support of MIVs guiding FSPs in their transformation from credit-only institutions towards saving mobilization institutions, which requires often additional capital but also the strengthening of their risk management procedures and systems, and cautious management of liquidity to protect savers’ rights. The SPTF has developed several guidelines for boards and board members to manage social performance and support FSPs’ commitment to social impact (Wardle, 2015). Strong governance is key to ensuring the sustainable development of the FSP. In that respect, when asked about the type of governance-related clauses included in their shareholder agreements, most equity MIVs (67 percent) included the CPPs in their shareholder agreements and half had a clause regarding a social and environmental management system creation. One-third of the funds either had agreements ensuring no mission drift by new shareholders or clauses about the creation of an SPM committee at the board level or nominated a responsible person for environmental and social risk management (Symbiotics Group, 2018).

INVESTORS’ INFLUENCE AT THE INDUSTRY LEVEL Looking at the impact of investments of KfW in the financial sector, Wisniwski and Maurer (2004) find positive demonstration effects on the financial institutions and the financial sector,

An investor’s perspective on social performance and impact  257

and provision of a learning experience for regulators and central banks. Yet, most research has been focusing on assessing the impact of financial inclusion at the level of the end borrower (micro-level), and only limited studies have looked at the wider impact of financial inclusion at the level of the sector. This is the attempt of this section. Bridging the Funding Gap through the Promotion of Local Currency Funding MIVs aim to finance FSPs serving economically vulnerable populations in emerging markets. These FSPs often structurally face funding needs and the role of MIVs is therefore critical in filling in such funding gaps. MIVs have played an important role in channeling large amounts of public and private capital into the financial inclusion sector (Kloppenburg, 2006). In 2008, MIVs accounted for approximately half of the total cross-border foreign capital investments in financial inclusion (CGAP, 2011). One of the key impediments to the structuring of international investments across countries is the matter of foreign exchange (FX) risk. International investments in developing countries have been typically denominated in “hard” currencies, such as the US dollar (USD) because FX risks associated with funding projects in local currencies in emerging economies are typically considered very high by the investors. Due to that, the FX risk is left at the FSPs in developing countries, which, because of underdeveloped financial markets, have little to no possibility of mitigating currency risk (Priberny and Dorfleitner, 2013). This exposes FSPs, and sometimes even the final borrowers, to the risk of suffering FX-related losses and financial instability. Therefore, mitigating FX risks is critical for ensuring sustainable financial-sector development, inclusive economic growth, and poverty alleviation in developing economies (TCX, 2018). It is within this context that the Currency Exchange Fund (TCX), and its sister organization MFX Solutions, were founded in 2007 and 2008, respectively. Their sponsors were a group of development finance institutions (DFIs), MIVs, and donors.4 Both entities provide FX riskmanagement tools in the form of hedging instruments, or products, for local currencies in more than 70 developing countries. The demand for local currency financing has increased heavily over the past years. A good indication of this growth is TCX’s derivative portfolio, which grew from USD 2.2 billion in 2017 to USD 4.54 billion in 2019. The objective for TCX is to act as a catalyst, and by taking the role of the first mover, crowding in other providers to establish a market. According to the Experts, managing FX risks is a key pillar of socially responsible investment. In certain countries with high currency volatility, the absence of hedging options would simply mean that MIVs would not lend in these countries because currency risk is deemed too high. By offering hedging products such intermediaries contribute to an incremental investment flow and therefore influence the investors’ confidence in certain countries. Promoting Responsible Lending Practices The lack of harmonized or adequate regulatory and supervisory frameworks for financial inclusion typically increases the risk of free riders and irresponsible behavior from FSPs (CGAP, 2012). In certain markets, MIVs have greatly contributed to the development and promotion of responsible practices within the financial inclusion sector. Cambodia represents one significant example of investors influencing positively local markets (Incofin IM, 2018a, 2019). During the 1990s, financial inclusion contributed significantly to the socio-economic reconstruction of Cambodia in the aftermath of the civil war.

258  Handbook of microfinance, financial inclusion and development

A supportive regulatory framework and a low level of financial inclusion made it very attractive to investors, especially in the context of a dollarized economy. The size of the financial inclusion industry grew rapidly, largely enabled by the inflow of foreign capital. Yet, persistently high portfolio growth and increasing average loan sizes raised concerns about overheating market. Concerned by the risks of over-indebtedness, several initiatives started in 2012 to map the excessive debt concerns, coordinated and financed by various social investors. These projects facilitated, amongst others, the creation of the Credit Bureau of Cambodia (CBC). Yet, in 2017, data from the CBC revealed that 50 percent of loans disbursed were for refinancing. To safeguard a healthy financial inclusion sector, tame market growth, and protect clients, Incofin IM together with other international institutions ADA, BIO, FMO, and PROPARCO, and thanks to the active participation of the Cambodia Financial Inclusion Association (CMA), the CBC pioneered the Responsible Lending Guidelines, a pro-active, self-regulatory initiative aiming to encourage FSPs to publicly endorse a set of responsible lending rules. In addition to being endorsed by the entire FSPs community through the leadership of the CMA, more than 20 MIVs endorsed the document, and the SMART Campaign decided to include compliance with the Lending Guidelines as a requirement in its certification methodology. Such a decision led to several institutions temporarily losing their CPP certification, prompting several lenders to decide not to lend to these institutions until they won their certification back. Today, the Credit Bureau of Cambodia reports monthly on the FSP’s compliance with the Lending Guidelines. These reports allowed MIVs to be more selective while remaining active in Cambodia in the pursuit of their impact strategy. The risk of over-indebtedness remains in Cambodia, yet sufficient industry-level information exists on market saturation thanks to the guidelines initiative, allowing MIVs to make informed investment decisions. The case of Cambodia is not a remote example of how social investors have teamed up to contribute to the responsible financial inclusion sector. In 2012 investors led a similar initiative in Kyrgyzstan, supporting research on the risk of over-indebtedness, promoting systematic use of the credit bureau, and formulating a responsible lending charter to limit multiple borrowing and refinancing loans. Numerous credit bureaus in emerging countries have been promoted by DFIs acting at the local level. Complementing their efforts in expanding access to finance to the poor and underserved, DFIs have played an active role in advancing credit bureau development, through feasibility studies, legal and regulatory advice, research, public awareness projects, and long-term coaching and support. Promoting Responsible Behavior during Good and Bad Times The implementation of responsible covenants and social undertakings can also be seen as another achievement of the sector in continuously encouraging responsible financial inclusion. Indeed, created in 2014 the lenders’ guidelines for setting reasonable covenants (the SPTF’s Lenders’ Guidelines5) is a common set of covenants and social undertakings developed by a group of public and private investors under the SIWG. By providing a framework to harmonize, wherever possible, the definition of covenants and undertakings, they intended to ease reporting constraints for FSPs and enforce further implementation of SPM. For instance, recognizing the importance to protect local savers’ rights, a financial covenant requires a minimum liquidity level for saving mobilization operating in lowly regulated environments. A social undertaking also mentions that for any FSP with an ROA higher than 7.5 percent, further justification shall be provided by the FSP to assess the risks of irresponsible pricing. While agreeing to discuss and align proved the “constructive spirit” of the MIVs community,

An investor’s perspective on social performance and impact  259

actual implementation remains limited nonetheless. Out of the 59 MIVs responding to the Symbiotics Survey in 2017, 25 reported being aligned with the SPTF’s Lenders’ Guidelines. Out of the 34 remaining MIVs, 24 still reported to include social undertakings despite not being fully aligned with the guidelines. According to the Experts, one reason for this misalignment is the difficulty to capture the requirements and specificities of the fund’s constraints or risk appetites. Finally, MIVs’social responsibility is also evidenced in the event of distress of FSPs. In such cases, they must safeguard their binding fiduciary responsibilities while minimizing adverse impact on the sector and protecting the clients. The pursuit of social impact does not mean that impact investors should take a subordinated position to “commercial” investors. Yet the authors highlight how the dual objectives of both social and financial performance increase the complexity of workout management. Again, the sector demonstrated its ability to join efforts to promote responsible financial inclusion by proactively working together on FSP’s debt restructurings and developing concrete practices and tools for MIVs, including the voluntary debt workout principles and inter-creditor agreement templates. The workout principles provide the best ethical practices on transparency, disclosures of interest, good faith, time being of the essence, favoring long-term and going-concern solutions and fair burden-sharing (IAMFI, 2011). Limitation: Improving Pricing Transparency The price at which clients borrow their loans has been historically and remains a sensitive topic in financial inclusion. In the nineties, raising interest rates was often deemed the only way for sustainable financial inclusion (Otero and Rhyne, 1994). Yet rapidly, the higher level of interest rates charged by some FSPs led to serious ethical concerns and sometimes even the accusation of usury. Microfinance interest rates usually range between 15 and 40 percent per year, while a study done on 17 microfinance markets concluded that the effective interest rate was in the range of 64 to 82 percent (Beisland, Mersland, and Randøy, 2014), depending on the macro- and micro-environment. Higher interest rates are primarily linked to the heavy transaction costs associated with lending small amounts to poor people and a risk premium (Dorfleitner et al., 2013). Higher interest rates have also been justified by the typically higher return on investment of the smaller business compared to larger ones, implying that microborrowers could afford to pay higher interest rates charged by the FSPs (Harper, 1998). Yet, because profit and returns to equity investors remain intrinsically correlated to the interest rates paid by end borrowers, large profits are often perceived as derived at the expense of the protection of poor clients, which are not always properly and transparently informed in the financial transaction. The links between returns of clients, FSPs, and investors make pricing recurrent in investors’ reputation management and regulators’ concerns. As early as 2009, fairness and transparency pricing were included within the CPPs, while from 2012, the USSPM promoted the need for balanced financial returns, i.e. respecting the interests of the end clients. The USSPM called also for measuring the performance of investments beyond shareholders’ return and integrating other stakeholder groups, in this case, end clients. Yet, little data on comparable prices is available to make informed decisions, with the risk of underestimating opaque and irresponsible pricing, or, on the opposite, over-reacting to the risk with mixed results through for example the use of interest rate caps (Spaggiari, 2019). The pricing transparency advocate, Chuck Waterfield (2015), dedicated a huge amount of energy to creating a publicly available platform to collect data on pricing and increase

260  Handbook of microfinance, financial inclusion and development

transparency on the interest rates charged by FSPs, the Microfinance Transparency (MFT). After years of advocacy work, MFT failed to collect a large volume of data in a sustainable way. To some extent, numerous FSPs were concerned about making the first step in sharing their data and possibly being publicly attacked while other competitors would have kept quiet. As these FSPs were not required by any law to publish their pricing, the lack of incentives for FSPs to share their data negatively the uptake of MFT. The sector also somewhat failed in achieving sufficient progress in improving pricing transparency and divulging the true price FSPs charged on loans to the poor. In this context, several double bottom line investors have integrated pricing in their due diligence analysis such as Triple Jump, which developed a sophisticated tool called the Pricing Traffic Light. With the need to bring objectivity in its assessment of responsible pricing and determining the appropriate balance between social and financial returns, Triple Jump developed the Interest Traffic Light as early as 2011. At the first level of analysis, the Traffic Light focuses on three factors that, if above a certain threshold, trigger a second level of analysis, which focuses on an additional set of six factors to determine if the levels of interest and profits are justified. The Interest Traffic Light synthesizes these and other factors into an overall score, which determines if the financial institution is eligible for financing. More recently a rating agency MFR launched an online private platform, ATLAS, aiming at providing access to a large and comparable database to benchmark FSPs pricing offerings.6 As the Experts pointed out, transparency is a necessary condition of responsible pricing, and can only be achieved through detailed benchmarks to inform deliberate and balanced decisions along the investment value.

PUSHING THE IMPACT MANDATE ONE STEP HIGHER: DEFINING SOCIAL RETURNS AND ALIGNING ASSET OWNER AND MANAGERS’ EXPECTATIONS The following section opens the debate on what truly differentiates an impact investor from a commercial, non-impact driven investor. It summarizes the ongoing debates on measuring social return and provides insights on the hot topics related to responsible social and financial returns on realized equity investments and so-called “responsible exits”. Encompassing Financial and Social Returns into One ROI According to the GIIN definition of impact investing, impact investing means that the success of an asset manager’s investment strategy would eventually be evaluated by the asset owner according to financial return metrics but also according to social and environmental impact metrics, or “social returns”. Most MIVs target and measure both financial and social returns (Symbiotic Group, 2018). However, the authors point out the limited progress that has been achieved and the lack of consensus on measuring and reporting social returns. The concept of social return or SROI is, however, not new and has been studied for two decades already (Emerson 2003; Hall, Millo, and Barman, 2015). The SPTF defines the social return on investments (SROI) as a financial concept that incorporates principles from return on investment and cost-benefit analysis to derive an estimate of net social benefit (SPTF, 2019). Typically, the net social benefit would be expressed as either the dollar value of social benefits minus social costs or the ratio of social benefits to social costs (Nicholls et al., 2009). In particular, while exploring existing SROI frameworks, Lingane and Olsen (2004) emphasize the challenges in

An investor’s perspective on social performance and impact  261

quantifying the amount of social value created. In financial inclusion literature, the concept of social return has been often broadly associated with several measures of output, such as the average loan size or number of women borrowers, taken as a proxy of social performance. Unfortunately, these proxies fail to capture the multiple dimensions of social performance (D’Espallier and Goedecke, 2019). The interviewed Experts point out that while funds’ documentation typically clearly lays out the expected financial return, in most cases only broad social performance goals or basic output data are mentioned. The lack of definition and limited formalization of social return expectations create a disconnect between the fiduciary duties of an asset manager towards its asset owners and its incentive to reach its impact mandate wherever a trade-off between financial and social return emerges. The possibility of a contradiction between financial and social return achievement strategies is not yet legally recognized and could be an interesting area for improvement for the industry. Certain asset owners, while recognizing that outcome will not be possible to be measured, have nonetheless decided to integrate certain SPM and impact-related metrics in the fund documentation in which they invest to incentivize MIVs to deliver both on their financial and social targets. In that context, the example of the debt fund for financial inclusion in Congo (Fonds pour l’inclusion financière en DRC or FPM) offers an interesting example of the formalization of alignment of interests between the asset owners of the fund and the advisor of the fund. The shareholders of FPM SA have clearly stated their twofold financial and social objectives which are formalized in the fund documentation. To ensure these objectives are followed in the execution of the investment strategy, non-financial criteria are considered in the remuneration of the advisor. For instance, governance and social indicators (such as the number of cooperatives and locally owned FSPs, the number of provinces where these FSPs operate, and the number of female clients served by FSPs) account for 35 percent of the total calculation of the advisor’s remuneration, while financial performance indicators stand at 50 percent. Such a remuneration structure ensures a full alignment of all stakeholders on how to measure the success of the deployment of the fund according to its double bottom-line objectives. Responsible Exits Financial inclusion struggled for years to prove the financial sustainability of such a risky business model. Efforts toward refinement of institutional and financial performance have borne fruit, and the financial inclusion sector is now seen as one of the most profitable impact asset classes. In particular, financial inclusion equity returns have outperformed by far those of emerging market banks, sparking debate on responsible return. The IPO of the Mexican FSP Compartamos in 2007 was in this regard quite heavily criticized (Lewis 2008). That controversy raised already several concerns about responsible exit and undermined the impact of the massive private-sector infusion of investments to alleviate poverty (Rhyne and Guimon, 2007). Yet the concept of responsible exit was formalized later in 2014 when the CGAP led the publication of a paper on “The Art of the Responsible Exit in Financial inclusion Equity Sales”.7 In this paper, CGAP highlighted some lessons drawn from past realized exits and highlighted that the lack of flexibility of close-ended funds in exit horizons made them in essence not the most suitable investment vehicle. Five years later, with many more exits realized, the debate around responsible exit matured one step further and both practitioners and researchers start defining a potential “framework” for responsible exits. Based on expert interviews, Mendelson and Rozas (2018) draw

262  Handbook of microfinance, financial inclusion and development

a framework to help investors navigate the complex terrain of responsible exits. Their model relies on three steps: (1) excluding buyers with bad reputations and with which regulator’s approval would be difficult, (2) reviewing the financial offer, and (3) assessing whether the proposed buyer’s strategic objective aligns with the social mission and the best interests of the FSP. Mendelson and Rozas (2018) conclude that a buyer selection practice that gives primacy to the financial offer and considers social mission and strategic value to the investee, only to reject egregiously unsuitable buyers and fail to keep in mind that the best interests of the FSP and its clients are the primary reason for investing in the financial inclusion sector. To allow a more balanced consideration between financial and social impact in an exit strategy, the investor’s expectations need to be clearly incorporated in the fund documentation. More specifically, similarly to the way asset managers include certain exit clauses in investees’ shareholder agreements to ensure a timely exit in line with the fund’s life expectancy, certain legal provisions referring to the preservation of the investee’s social mission or social performance governance structure could be considered. While the enforceability of such clauses can be questioned, including and defending them during exit negotiation has the merit to test the intentions of the buyer. In 2018, on the occasion of its exit from one of its financial inclusion investees in Cambodia, Incofin IM, acting as the fund manager of the Rural Impulse Fund II, developed its own exit framework called the Fitness and Compatibility Review tool. After the successful use of this framework, Incofin IM embedded this approach as part of its overall impact methodology. Any exit proposed to an investment committee needs to be structured according to the Fitness and Compatibility Review for their approval.

BOX 14.1: INCOFIN IM’S FITNESS AND COMPATIBILITY REVIEW TOOL TO PROMOTE RESPONSIBLE EXITS: A CASE STUDY OF CAMBODIA Founded in 2003, AMK is one of the largest and most recognized rural FSPs in Cambodia serving around 700,000 clients. AMK’s strong customer centricity and deep focus on rural clients were key decision factors for Incofin IM, on behalf of Rural Impulse Fund II (RIF II). Demonstrating the extent of their social mission, AMK had set up a social performance committee at the level of the board, and was and still is certified by the SMART Campaign, holding also an “α” rating on its social performance by M-Cril. In 2012, Incofin IM invested a 25 percent stake in the institution and exited six years later, selling its stake to the Shanghai Commercial and Saving Bank (SCSB). Given the importance of social performance in AMK’s business strategy, when shareholders, including Incofin IM, started planning their exit, they decided to implement a formal screening process using the Fitness and Compatibility Review to assess the suitability of any potential buyers based on a list of financial and non-financial criteria. A committee was established to conduct reverse due diligence on interested parties, including interviewing the potential future members of the FSP’s board. All the information gathered was then used to screen the selected buyers against a list of 17 key criteria covering their reputation, market knowledge, financial performance, social performance, strategy and vision, corporate culture, and additionality. Amongst these criteria, investment return

An investor’s perspective on social performance and impact  263

expectations, investment horizon, and commitment to the FSP’s social mission were considered critical to preserving the impact mission of the investee. By assigning a grade against each criterion for each buyer, the selling investors could engage in meaningful conversations, such as understanding their true commitment to AMK-specific target poor clientele. As a result of this process, SCSB was selected. SCSB is a retail bank serving the MSMEs of its home markets in Taiwan for more than 100 years. Its strategy, its banking expertise in less emerging economies of South East Asia, its investment horizon, and its commitment to support local economic development in Cambodia were all factored into the assessment tool. The Fitness and Compatibility Review tool offered a framework for selecting the most suitable next shareholder of AMK. One other result of the discussion around the framework was that the selected buyer, SCSB, accepted to include a specific clause in the shareholder agreement, requiring that any change to AMK’s social mission or target client would require the positive vote of the original promoter shareholder, Agora. A similar clause was embedded regarding Cambodian management protection (MicroCapital, 2018; Incofin, 2018b).

Need for Further Industry Introspection: Assessing the Social Performance of Investors While social rating and CPP certification tools exist to assess the FSPs, similar rating tools have yet to emerge to be able to rate and compare MIVs and rebalance the impact accountability not only at the level of the investees but throughout the whole investment chain (asset managers and asset owners). In 2010, responding to the need for transparent and systematic information at the MIV level, four MIVs, including Blue Orchard’s Dexia Microcredit Fund (DMCF), Incofin CVSO, Rural Impulse Fund (RIF), and Oikocredit, underwent the first pilot rating exercise with M-CRIL (Sinha and Sinha, 2010). In 2011, the UN-supported Principles for Responsible Investment (PRI) initiative launched the Principles for Investors in Inclusive Finance (PIIF), a specific set of practices catered to financial inclusion investors only. While investors were only expected to endorse and self-report, the PIIF constituted a first attempt to apply and check the implementation of SPM standards by investors. In 2019 though, PRI decided to discontinue the PIIF-specific initiative and integrate it back within the UN PRI framework (UN PRI, 2016). The Luxembourg Labelling Agency, LuxFlag, launched its microfinance label in 2006 to prove to investors that the MIV invests in the microfinance sector. Yet the label covered only limited features, mostly focusing on the concentration of assets in microfinance. Later, intending to increase the expected requirements of its label, LuxFlag introduced social performance standards in its methodology, including the adoption of social performance objectives and indicators by the MIVs in their due diligence process, regular monitoring of the portfolio’s social performance, and reporting to investors. Such a label, although in theory open worldwide to regulated MIVs, represented only 32 MIVs in 2019. In 2018, CERISE has also developed an assessment tool dedicated to impact-driven investors, IDIA (Impact-Driven Investor Assessment). IDIA aims at helping investors, fund managers, and donors to assess the implementation of their impact thesis, thanks to the appraisal of their social strategy, governance, practices, and products (CERISE-IDIA, 2018). The tool was

264  Handbook of microfinance, financial inclusion and development

created in 2018 (based on a previous audit tool for MIV used by six microfinance funds) and four investors/funds have gone through the exercise for now. In April 2019, IFC reactivated the debate around the need for higher scrutiny of investors’ SPM and launched the IFC Operating Principles for Impact investors, which offer a framework for understanding and managing environmental and social risks for impact-driven fund managers. To encourage transparency and exchange of good practices, all endorsers are asked to disclose their implementation of the principles on an annual basis. The disclosures are then independently verified by either internal or external partners. The first disclosure exercise will be conducted in 2020. The recent EU regulation on disclosures relating to sustainable investments and sustainability risks, published in 2019, goes in the same direction, i.e. ensuring a greater level of transparency and accountability of impact fund managers on their impact promise. This regulation introduces additional requirements for the formalization and integration of ESG risks in both investment decision-making processes and disclosures. The Experts highlighted that, so far, if several initiatives and tools have emerged over the last decade, no unique framework nor tool stood out as setting the reference, hereby inviting asset owners and regulators to accelerate dialogue on the topic to agree on common standards and be able to increase the visibility of investors’ differentiated impact and SPM strategy and execution.

CONCLUSIONS The attempt of financial inclusion investors to define and measure social performance and impact is a real and undeniable industry achievement. It demonstrates the industry’s unique capacity to create standards and, linked to these, measurement tools, where nothing had existed before. The authors have endeavored to shed some light on the true efforts of MIVs to promote and push the execution of SPM with FSPs first. The interviewed Experts point out that those fund managers who can neither clearly show responsible social performance practices nor demonstrate the social relevance of their activities ultimately risk a discrediting of their reputation. Hence, if MIVs and/or investment managers want to be equally serious about both dimensions of the double bottom line, they need to devote equally strong processes, policies, and procedures to monitor and quantify the achievement of their social mission throughout the whole investment process. In doing so, numerous tools recognized by the industry are available to fund managers. In Figure 14.1, the authors summarize some of these ways and tools available to MIVs to formalize their impact thesis. The Experts also highlight the impact that MIVs have caused within the industry at large. Through engagement with national central banks and microfinance associations, MIVs have opened conversations and supported initiatives related to pricing transparency, responsible lending, local currency funding benefits, and the use of credit bureaus. In a lot of cases, these conversations became the blueprint for regulators and industry stakeholders when setting up guidelines and regulations. In our view, be it through applying social due diligence tools, promoting responsible lending practices, playing an active role in FSPs governance, or promoting outcome data tracking on end clients, MIVs have contributed to building and promoting one universal language and to a common understanding of social performance and impact. Nonetheless, we also see areas

An investor’s perspective on social performance and impact  265

Figure 14.1 Integrating social performance and impact management throughout the investment process for further collective reflection and improvement. To start with, the promotion of outcome tracking is only nascent and needs to be mainstreamed. It is only through the analysis of what is happening at the level of end clients that the financial inclusion industry will be able to reflect and adjust on which social and business strategy works and which one does not. This will also help nurture the debate around social return and how to formally capture social goals in the remuneration of MIVs to ensure a stronger alignment between asset owners and asset managers on which impact investment strategy is expected. More specifically, the authors provide insights on how investors can ensure responsible social and financial returns on realized equity investments, by aiming for “responsible exits”. The way an investor sells the holdings in an investee should be part of the impact investment process continuum and the way to go about it, embedded in the fund documentation. The authors finally point out the current absence of a social audit tool for investors recognized industry-wide, which limits the true accountability of investors while hampering their capacity to assess and differentiate the impact or social returns promised by each asset manager. As a broader impact investment community is structuring itself using the language of the United Nations Sustainable Development Goals and complying with the EU SFDR Regulations, the financial inclusion industry has a lot of lessons learned, standards, and tools to offer as public goods. With 20 years of success and failures, financial inclusion investors have seen what claiming without objectively measuring can lead to. To prevent risks of impact washing, creating standards and measurement tools beyond logos is the next challenge ahead. Only such a rigorous approach will allow us to truly differentiate an impact investor from a commercial, non-impact driven investor. The journey of financial inclusion investors can humbly serve in that respect.

MIVS’ RESPONSE TO THE COVID-19 CRISIS COVID-19 was the black swan of 2020; none of us could have anticipated that such a pandemic would have hit the world at such a pace, transforming our societies and economies in various ways and intensities. All sectors and markets have been impacted by a complex and unprecedented combination of domestic and external shocks. Despite containment measures being continuously adjusted across the globe, and notwithstanding the uncertainty lingering

266  Handbook of microfinance, financial inclusion and development

around the timing and pace of recovery, the microfinance sector has demonstrated again its ability to cooperate in these turbulent times, pooling efforts to battle the impact of the pandemic on multiple fronts and adapt to the new operating environments.8 At the onset of the crisis, several MIVs gathered to ensure coordination of funding support to stabilize the funding base of FSPs in response to the COVID crisis. To that end, Incofin Investment Management, together with a group of MIVs representing more than USD 10bn of total assets, contributed to the elaboration of a Memorandum of Understanding providing a framework for managing debt refinancing and limiting financial risk—particularly liquidity risk—that FSPs faced due to COVID. Standard monitoring tools were developed to reduce the reporting burden on investees. These initiatives allowed for swift positive results in the field and prevented several institutions from entering more heavy distress. While liquidity needs have varied greatly across geographies and sectors, preliminary data show better sector performance than anticipated, with no evidence of a global liquidity crunch throughout 2020. Although the crisis required significant attention to solving immediate liquidity issues, ensuring equally rigorous social due diligence standards was critical. Within the SPTF, the SIWG launched regular virtual meetings to coordinate efforts and share resources. The SPTF launched also a “Crisis Checklist”, supporting a social response to the crisis based on the Universal Standards for SPM. The checklist highlighted which social performance practices were more likely to be put at risk due to COVID and provided guidance to investors on how to adjust their social due diligence to maintain the same level of scrutiny during the pandemic. First, the traditional “do not harm approach” called for extra prudence to identify risks of bad collection practices, irresponsible restructuring, and caution in the calculation of clients’ debt repayment capacity given their likely income loss. Second, human resources were also key areas of focus to ensure that any staff movement was done responsibly and that field staff but also agents were properly protected. Thirdly, digitalization and the related increased client protection risks were also mentioned, asking the sector to abide by best transparency practices when shifting to digital channels. Eventually, attention to governance aspects including assessing board conversations on social topics such as increasing credit risks or decisions to not distribute dividends were also highlighted as key areas to be mindful of to ensure the highest level of social performance even during crisis time. The authors emphasize that customer centricity and strong social safeguards were key to allowing FSPs to navigate through the crisis and build the resilience of the most vulnerable micro-borrowers. Various groups at the level of the SPTF provided forums for discussion and guidance to address these concerns in a structured manner.

ACKNOWLEDGMENTS The authors would like to thank all the interviewed experts. In particular, we express our great appreciation and special gratitude for the valuable and constructive discussions with: ● ● ● ● ●

Lucia Spaggiari, Business Development Director, MFR Daniel Rozas, Senior Microfinance Consultant, e-MFP Jurgen Hammer, Managing Director, SPTF Cécile Lapenu, Managing Director, CERISE Christophe Bochatay, Social Performance and Impact Manager, Triple Jump

An investor’s perspective on social performance and impact  267 ● ● ● ●

Loïc De Cannière, Founder and Managing Partner, Incofin IM Amelia Greenberg, Deputy Director, SPTF Anne Contreras, Of Counsel, Arendt and Medernach Amadeus Bringmann, Analyst, TCX

Their input took our insights to a new level.

NOTES 1.

Microfinance was estimated to account for 13 percent of total impact assets under management in 2018 (Mudaliar et al., 2019). 2. As per January 2020, more than 5,000 organizations had endorsed the CPPs, including 1,841 FSPs and 192 investors and donors (SMART Campaign, 2019). 3. SPI4 ALINUS is available on http://cerise​-spm​.org. 4. Current investors in TCX are 22 multilateral and bilateral DFIs and MIVs and the Dutch and German governments. Please refer to the TCX website. 5. The complete SPTF Lender’s Guidelines for Setting Covenants in Support of Responsible Financial Inclusion can be found here: https://sptf​.info​/images​/investor​%20wg​_2014​%20lendersguidelines​ _reasonablecovenants​_final​_2014​.pdf. 6. ATLAS centralizes pricing, social, and financial data of FSPs in more than 70 countries and calculates APR data using the microfinance transparency method, allowing reliable and comparable results. ATLAS is accessible at www​.atlasdata​.com. 7. See Rozas (2014). 8. The results of these group discussions are publicly available on the SPTF’s website. A forum was also set up under the website www​.covid​-finclusion​.org as a collaborative project between the European Microfinance Platform (e-MFP), the Center for Financial Inclusion at Accion (CFI), and the Social Performance Task Force (SPTF). Its objective was to provide a platform to promote discussion about the challenges and solutions for the financial inclusion sector in response to the COVID-19 pandemic as well as sharing key initiatives.

REFERENCES Abdullah, S., Quayes, S. (2016): Do women borrowers augment financial performance of MFIs? Applied Economics 48, 5593–5604. Arnold, L., Booker, B., Dorfleitner, G., Röhe, M. (2019): Refinancing MFIs with market power: theory and evidence, Small Business Economics 56(4), 1485–1505. Bauwin, M. (2019): Social performance management in microfinance: practices, results and challenges, ADA, Cerise, https://cerise​-spm​.org​/app​/uploads​/sites​/3​/2019​/04​/ Etude​-SPI4​-EN​-final​.pdf. Bédécarrats, F., Guérin, I., Roubaud, F. (2015): The gold standard for randomized evaluations: from discussion of method to political economy, Working Paper UMR DIAL. Beisland, L., Mersland, R., Randøy, T. (2014): Transparency and disclosure in the global microfinance industry: implications for practice and policy makers, in: Forssbaeck, J., Oxelheim, L. (Eds), The Oxford handbook of economic and institutional transparency, Oxford University Press, New York. CERISE-IDIA (2018): IDIA BRIEF, https://cerise​-spm​.org​/app​/uploads​/sites​/3​/2018​/12​/ Newsletter​_ IDIA​_ EN​.pdf. CGAP (2004): Key principles of microfinance, https://www​.cgap​.org​/sites​/default​/files​/CGAP​Consensus​-Guidelines​-Key​-Principles​-of​-Microfinance​-Jan​-2004​.pdf. CGAP (2011): Cross-border funding of microfinance, CGAP Focus Note No. 70, https://www​.cgap​.org​/ sites​/default ​/files​/researches​/documents​/CGAP​-Focus​-Note​- Cross​-border​-Funding​-of​-Microfinance​ -Apr​-2011​.pdf. CGAP (2012): A guide to regulation and supervision of microfinance, Consensus Guidelines, https://www​.cgap​.org ​/sites​/default ​/files​/researches​/documents​/Consensus​- Guideline​-A​- Guide​-to ​Regulation​-and​-Supervision​-of​-Microfinance​-Oct​-2012​_0​.pdf.

268  Handbook of microfinance, financial inclusion and development

CGAP, CitiGroup, CSFI (2008): Microfinance banana skins 2008: risk in a booming industry, Centre for the Study of Financial Innovation. De Corte, J. M., Labie, M., Urghege, L., Vansinck, J. (2013): Microfinance investment vehicles and social performance, in: Basu, R. (Ed.), Microfinance: a practitioner’s handbook, Globe Business Publishing, London, 165–191. D’Espallier, B., Goedecke, J. (2019): Social performance measurement in microfinance, in: Hudon, M., Labie, M., Szafarz, A. (Eds), A research agenda for financial inclusion and microfinance, Edward Elgar, Cheltenham, 62–74. Dewez, D., Perez, L. (2010): INCOFIN’s experience in assessing MFIs social performance, European Dialogue n°3, European Microfinance Platform, 73–83, https://www​.findevgateway​.org​/sites​/default​/files​/ mfg​-en​-paper​-making​-microfinance​-investment​-responsible​-state​-of​-the​-practice​-in​-europe​-nov​-2010​.pdf. Di Benedetta, P., Lieberman, I., Ard, L. (2015): Corporate governance in microfinance institutions, International Bank for Reconstruction and Development/The World Bank, http://documents​ .worldbank ​ . org ​ /curated ​ /en ​ / 567251468189242027​ / pdf ​ / 96022 ​ - May​ -WP​ - P130793 ​ - PUBLIC​ -Box391458B​-microfinance​-APRIL​-15​-2015​.pdf. DiLeo, P., Watkins, T. A., Kanze, A. (2017): Microfinance is dead; long live microfinance, Syminvest Blog, https://www​.syminvest​.com​/news​/p​-dileo​-t​-a​-watkins​-and​-a​-kanze​-microfinance​-is​-dead​-long​ -live​-microfinance​/2017​/4​/20​/6563. Dorado, S., Molz, R. (2005): Co-evolution of boards of directors in microfinance organizations: the case of Bancosol and Los Andes, Journal of Developmental Entrepreneurship 10, 99–121. Dorfleitner, G., Leidl, M., Priberny, C., von Mosch, J. (2013): What determines microcredit interest rates? Applied Financial Economics 23, 1579–1597. Dorfleitner, G., Röhe, M., Renier, N. (2017): The access of microfinance institutions to debt capital: an empirical investigation of microfinance investment vehicles, The Quarterly Review of Economics and Finance 65, 1–15. Drake, D., Rhyne, E. (2002): The commercialization of microfinance: balancing business and development, Kumarian Press, West Hartford, CT. Duvendack, M., Palmer-Jones, R., Copestake, L. (2011): What is the evidence of the impact of microfinance on the well-being of poor people? EPPI-Centre, Social Science Research Unit, Institute of Education, University of London. El-Zoghbi, M., Holle, N. (2019): Is poverty reduction the right outcome for financial services? CGAP blog series impact and evidence in financial inclusion: taking stock, https://www​.cgap​.org​/ blog​/ poverty​-reduction​-right​-outcome​-financial​-services. Emerson, J. (2003): The blended value proposition: integrating social and financial returns, California Review Management 45, 35–51. Frank, C. (2008): Stemming the tide of mission drift: microfinance transformations and the double bottom line, WWB Focus Note, Women’s World Banking, https://www​.findevgateway​.org​/sites​/ default ​/files​/mfg​- en​-paper​-stemming​-the​-tide​- of​-mission​- drift​-microfinance​-transformations​-and​ -the​-double​-bottom​-line​-apr​-2008​.pdf. Hall, M., Millo, Y., Barman, E. (2015): Who and what really counts? Stakeholder prioritization and accounting for social value, Journal of Management Studies 52, 907–934. Harper, M. (1998), Profits for the poor: cases in micro-finance, Intermediate Technology Publications Ltd., London. Hartarska, V., Shen, X., Mersland, R. (2013): Scale economies and input price elasticities in microfinance institutions, Journal of Banking and Finance 37, 118–131. Helms, B. (2006): Access for all: building inclusive financial systems, International Bank for Reconstruction and Development/The World Bank, CGAP. Hermes, N., Lensink, R., Meesters, A. (2011): Outreach and efficiency of microfinance institutions, World Development 39, 938–948. Hoepner, A., Liu, H., Moauro, A., Perez-Rocha, B. Spaggiari, L. (2012): Financial results of microfinance institutions: social performance matters. Microfinanza Rating Srl. IAMFI Microfinance Lenders Working Group (2011): Charting the course: best practices and tools for voluntary debt restructuring in microfinance, International Association of Microfinance Investors, https://www​.findevgateway​.org ​/ library​/charting​- course ​-best​-practices​-and​-tools​-voluntary​- debt​ -restructuring​-microfinance.

An investor’s perspective on social performance and impact  269

Incofin (2018a): Incofin spearheads pioneering self-regulatory initiative to prevent over-indebtedness in Cambodia, http://www​.incofin​.com​/incofin​-spearheads​-pioneering​-self​-regulatory​-initiative​-to​ -prevent​-over​-indebtedness​-in​-cambodia/. Incofin (2018b): Impact newsletter, Incofin investment management, http://www​.incofin​.com​/wp​ -content​/uploads​/20181009​_ Newsletter​_Impact​.pdf. Incofin (2019): Microfinance significantly contributes to Cambodia’s socio-economic development – since 2012, Incofin has been encouraging responsible lending in Cambodia, http://www​.incofin​.com​ /microfinance​-significantly​- contributes​-to ​- cambodias​-socio ​- economic​- development​-since ​-2012​ -incofin​-has​-been​-encouraging​-responsible​-lending​-in​-cambodia/. Kloppenburg, N. (2006): Microfinance investment funds: Where wealth creation meets poverty reduction, in I. Matthaus-Maier & J.D. Von Pischke (Eds), Microfinance investment funds: leveraging private capital for economic growth and poverty reduction, Springer, Berlin, pp. 11–46. Lapenu, C., Brusky, B., Sallé, J. (2017): Universal standards for social performance management: an inspiring framework for impact investing, CERISE. Lapenu, C., Pierret, D. (2006): Handbook for the analysis of the governance of microfinance institutions, IFAD, CERISE-IRAM. Lewis, J. C. (2008): Microloan sharks, Stanford Social Innovation Review, 54–59, https://www​ .findevgateway​.org​/sites​/default ​/files​/mfg​-en​-paper​-microloan​-sharks​-2008​.pdf. Lingane, A., Olsen, S. (2004): Guidelines for social return on investment, California Management Review 46, 116–135. M-CRIL (2005): Pioneering a new rating service: social rating of MFIs, https://www​.findevgateway​ .org​/sites​/default ​/files​/mfg​-en​-paper​-social​-rating​-overview​-of​-mfis​-2005​.pdf. Mendelson, S., Rozas, D. (2018): Caveat venditor: towards a conceptual framework for buyer selection in responsible microfinance exits, COLOPHON, http://www​.e​-mfp​.eu​/sites​/default ​/files​/resources​ /2018​/04​/ Buyer​%20Selection​%20in​%20Responsible​%20Microfinance​%20Exits​.pdf. MicroCapital (2018): Special Report: a model from Cambodia for preventing overheating – not just multiple lending, https://www​.microcapital​.org​/special​-report​-a​-model​-from​-cambodia​-for​ -preventing​-overheating​-not​-just​-multiple​-lending​-to​-be​-presented​-at​-european​-microfinance​-week​ -november​-14​-16​-2018/. Morduch, J., Ogden, T. (2019): The challenges of social investment through the lens of microfinance, in: Hudon, M., Labie, M., Szafarz, A. (Eds), A research agenda for financial inclusion and microfinance, Edward Elgar, Cheltenham, 12–26. Mudaliar, A., Bass, R., Dithrich, H., Nova, N. (2019): Annual impact investor survey 2019, GIIN, New York. Nicholls, J., Lawlor, E., Neitzert, E., Goodspeed, T. (2009), A guide to social return on investment, Office of the Third Sector, Cabinet Office, https://neweconomics​.org​/uploads​/files​/aff3779953c5b88d53​ _cpm6v3v71​.pdf. O’Rourke, A. (2003): The message and methods of ethical investment, Journal of Cleaner Production 11, 683–693. Otero, M., Rhyne, E. (1994): The new world of microenterprises finance: building healthy financial institutions for the poor, Kumarian Press, West Hartford, CT. Priberny, C., Dorfleitner, G. (2013): Risk perception and foreign exchange risk management in microfinance, Journal of Management & Sustainability 3, 68–78. Quayes, S. (2012): Depth of outreach and financial sustainability of microfinance institutions, Applied Economics 44, 3421–3433. Quayes, S. (2015): Outreach and performance of microfinance institutions: a panel analysis, Applied Economics 47, 1909–1925. Renneboog, L., Ter Horst, J., Zhang, C. (2008): Socially responsible investments: institutional aspects, performance, and investor behavior, Journal of Banking & Finance 32, 1723–1742. Rhyne, E., Guimon A. (2007): The Banco Compartamos initial public offering, Accion Insight, 23. Rhyne, E., Otero, M. (1992): Financial services for microenterprises principles and institutions, World Development 20, 1561–1571. Rhyne, E., Otero, E. (2006): Microfinance through the next decade: Visioning the who, what where, when and how, Paper commissioned by the Global Microcredit Summit 2006, Boston MA: ACCION International.

270  Handbook of microfinance, financial inclusion and development

Robinson, M (2001): The microfinance revolution: sustainable banking for the poor, The World Bank, Washington, DC. Rozas, D. (2014): The art of the responsible exit in microfinance equity sales, Forum 9, CGAP and Center for Financial Inclusion, Washington, DC. Sinha, F., Greenberg, A. (2017): Making the case for outcomes management to financial service providers, Social Performance Task Force, https://sptf​.info​/images​/ Making​_the​_Case​_for​_ Outcomes​_ Management​_to​_ Financial​_ Service​_ Providers​.pdf. Sinha, F., Sinha, S., M-Cril (2010): Rating of microfinance investment vehicles: a pilot initiative by M-CRIL, European Dialogue n°3, European Microfinance Platform, 27–41, https://www​ . findevgateway​.org​/sites​/default​/files​/mfg​- en​-paper​-making​-microfinance​-investment​-responsible​state​-of​-the​-practice​-in​-europe​-nov​-2010​.pdf. SMART Campaign (2019): Campaign endorsers, https://www​.smartcampaign​.org​/about ​/campaign​endorsers. Spaggiari, L. (2016): Guidelines on outcomes management for investors, European Dialogue Number 10, European Microfinance Platform. Spaggiari, L. (2019): More data for a more accountable financial inclusion industry, FinDev Gateway, https://www​.findevgateway​.org​/ blog​/2019​/nov​/more​- data​-more​-accountable​-financial​-inclusion​industry. SPFT (2017): The universal standards for social performance management implementation guide, Social Performance Task Force, https://sptf​.info​/images​/usspm​_impl​_guide​_english​_20171003​.pdf. SPFT (2019): Universal standards for responsible inclusive finance, https://sptf​.info​/images​/ USSPM​_ EnglishManual​_v2​.1​_2019​.pdf. Symbiotics Group (2011): 2011 symbiotics MIV survey, market data & peer group analysis, https:// www​.syminvest​.com​/download​/Symbiotics​-2011​-MIV​-Survey​.pdf. Symbiotics Group (2016): 2016 symbiotics MIV survey, market data & peer group analysis, https:// symbioticsgroup​.com ​/wp​-content​/uploads​/2016​/09​/Symbiotics​-2016 ​-MIV​-Survey​-Report​.pdf. Symbiotics Group (2018): 2018 symbiotics MIV survey, market data & peer group analysis, https:// symbioticsgroup​.com ​/wp​-content​/uploads​/2018​/10​/Symbiotics​-2018​-MIV​-Survey​.pdf. TCX (2018): Theory of change, championing sustainable and innovative finance for development, https://www​.tcxfund​.com ​/wp​-content​/uploads​/2018​/11​/ TCX​-Theory​-of​-Change​.pdf. UN PRI (2016): PIIF report on progress. Assessing the impact of responsible investors in inclusive finance, https://www​.unpri​.org​/download​?ac​= 4533. Wardle, L. (2015): Social performance task force - guidance, SPTF guidance issue 5, https://sptf​.info​/ images​/sptf​-guidance​-note​-issue​-5​-guiding​-your​-board​-on​-spm​-final​.pdf. Waterfield, C. (2015): Advocating transparent pricing in microfinance: a review of MF Transparency’s . work and a proposed future path for the industry, MicroFinance Transparency, http://www​ mftransparency​.org​/wp​- content​/uploads​/2015​/08​/ MFTransparency​-Advocating​-Transparent​-Pric ing​-in​-Microfinance​.pdf. White, V., Campion, A. (2002): Transformation: journey from NGO to regulated MFI, in Drake, D., Rhyne, E. (Ed.), The commercialization of microfinance, Kumarian Press, West Hartford, CT. Wisniwski, S., Maurer, K. (2004): Impact of financial sector projects in Southeast Europe - Effects on financial institutions and the financial sector, in: Matthäus-Maier, I., von Pischke, J. D. (Eds), The development of the financial sector in Southeast Europe: innovative approaches in volatile environments, Springer, Berlin, Heidelberg, 195–221.

PART V EVIDENCE FROM REGIONS AND COUNTRIES

15. Financial inclusion in high-income countries: gender gap or poverty trap? Anastasia Cozarenco and Ariane Szafarz

INTRODUCTION The microfinance literature has extensively informed the economic profession about financial exclusion in low-income countries and how to address it with financial services, such as credit, savings accounts, insurance products, and money transfers (Cull and Morduch, 2018). Strikingly, less is known about the (lack of) financial services provided to disadvantaged populations in high-income countries,1 suggesting misleadingly that financial inclusion in these countries is homogeneous (Lapavitsas and Powell, 2013). This chapter fills the gap by investigating and comparing financial inclusion in two major groups of developed countries: the Euro area (i.e. the European single-currency zone)2 and North America (the United States and Canada). Europe and North America are markedly different in various aspects of the economic, cultural, and social environments, including social security, job market, financial markets, etc. These differences in objective characteristics can in turn lead to significant differences in financial inclusion. To scrutinize access to bank accounts, credit, deposits, and access to emergency funds, we combine OECD data and the Global Findex database (https:// globalfindex​.worldbank​.org/), published every three years since 2011. Findex has been set up by the World Bank to provide a representative picture of financial inclusion worldwide. The purpose of this chapter is threefold. First, we focus on three geographical areas for which the Findex database provides ready-to-use data:3 the Euro area, North America, and the developing countries.4 We provide an overview of financial inclusion in these regions and compare them to get a sense of the amplitude of the issues at stake. We also check whether financial precariousness in developing countries has common characteristics with those in the developing world, as Ogden’s (2020) hypothesis of “great convergence” suggests. Second, beyond the global picture, we pay special attention to the gaps in access to financial services that prevail between men and women, and between rich and poor individuals. In general, the figures suggest that the European gender gap is wider than its North American counterpart. In contrast, the US exhibits a poverty trap in access to financial services that is deeper than in both Europe and developing countries. Last, our exploratory analysis offers potential explanations for the uncovered differences. The main contributions of this chapter are the following. First, we show that access to bank accounts and emergency funds are similar in the Euro area and North America. By contrast, access to credit and, to a lesser extent, access to deposit accounts are more developed in North America than in the Euro area. But in all aspects, the differences with respect to developing countries are still huge. Regarding the gender gaps and the poverty traps in financial inclusion, we find that the Euro area and North America face different challenges. Overall, the gender gaps are larger in the Euro area, with a single exception for financial resilience, while as expected, the financial exclusion of the poor is worse in North America. Gender differences in 272

Financial inclusion in high-income countries  273

access to finance matter not only because women are poorer than men on average (D’Espallier et  al., 2013; Agier and Szafarz, 2013; Hartarska et  al., 2014; Périlleux and Szafarz, 2015), but also because poor women are more likely to head their households (Worobey and Angel, 1990) and impoverished women tend to remain poor for longer periods than men (Wilson, 1987). In line with Ogden (2020), we find that the Euro area and North America significantly underperform developing countries in bridging the gender and poverty gaps for saving and borrowing opportunities offered to the poor. Overall, the figures suggest that the worldwide unmet demand for access to financial services originates primarily from women and poor households, but with great geographical variations.

FINANCIAL INCLUSION IN THE EURO AREA, NORTH AMERICA, AND DEVELOPING COUNTRIES By bridging the gap between borrowers and lenders, financial institutions actively participate in allocating resources in the economy. Intertemporal financial transactions are used by households to smooth consumption over the life cycle and so address mismatches between revenues and expenses (Deaton, 1992). Banks also provide a supposedly safe place to keep savings and liquidities. For all these reasons, banks and other financial intermediaries play an important role in improving the lives of individuals and businesses (Armendariz and Morduch, 2010), and inaccessibility or partial access to financial services can harm people (Allen et al., 2016). Yet, despite superficial similarities, citizens from different countries tend to use financial markets differently. Likewise, banking crises are addressed differently by local regulators (Rochet, 2009; Anginer et  al., 2019). The syndicated loan pricing puzzle (Carey and Nini, 2007) stems from the fact that, all else equal, credit in the European syndicated loan market is significantly cheaper than in its US counterpart. Burietz et al. (2017) suggest that the puzzle is due to region-specific accounting standards triggering region-specific credit conditions. Financial habits differ across the Atlantic Ocean, perhaps because Americans have lower risk aversion and gather less precautionary savings.5 The most common vehicle for shortterm borrowing is the credit card in both North America and Europe, but the prevalence of credit card debt is larger in North America. According to the 2017 Findex, 29 percent of North American adults and 17 percent of European adults borrow from a financial institution, whereas the rate reaches 41 percent and 29 percent for borrowing through a credit card, respectively. According to Pew Charitable Trusts (2015), about 80 percent of Americans are indebted, with mortgages being the most common type of liability. Morduch and Schneider (2017) confirm that credit scores are vital in North America as they are used by employers, insurers, landlords, and lenders. In contrast to other countries, in the US taking on debt is the only way to build a credit score. This could explain why so many people in the US borrow from a financial institution or through a credit card. Using the 2017 Findex database, Ogden (2020) points out that, in several aspects, poverty and income uncertainty in the US look like the situation prevailing in emerging and developing countries, such as South Africa, Brazil, Kenya, and Malaysia. The author argues that the worldwide financial inclusion landscape evolves toward a “great convergence.” In this section, we revisit the evidence and nuance the conclusions by bringing the Euro area into the picture. We focus on four main dimensions of financial inclusion: (1) the share of adults with an account at a financial institution or through a mobile money provider, (2) the share of adults borrowing from a financial institution or through a credit card, (3) the share of adults saving at

274  Handbook of microfinance, financial inclusion and development

a financial institution, and (4) the share of adults able to come up with emergency funds. Even though financial exclusion is sometimes defined with respect to access to an account only, economists contend that the three other dimensions are vital for financial inclusion as well (Demirgüç-Kunt et al., 2015). Hence, the rest of this section will examine the four dimensions and compare them across the Euro area, North America, and developing countries, whereas the next sections will discuss how gender and poverty gaps interact with financial inclusion and contrast the situations of the Euro area and North America. Account Ownership The literature on financial inclusion tends to focus on account ownership (Demirgüç-Kunt et al., 2017). In the Findex database, account ownership in a region is captured by the percentage of respondents in that region who report having an account—be it private or shared—at a bank or at another regulated financial institution, such as a credit union, a microfinance institution, a cooperative, or the post office. The definition is extended to having had access to mobile-money services in the past 12 months. Figure 15.1 shows the evolution of the share of adults with an account from 2011 to 2017 in the Euro area, North America, and developing economies. Account ownership in Europe and North America have similar evolutions with a steady increase from 90 percent in 2011 to around 94 percent in 2017. By contrast, developing countries experienced strong growth from 40 percent in 2011 to 63 percent in 2017. The comparison between the Euro area and North America confirms that both regions are financially well developed with a bancarization rate surpassing 90 percent in 2017. In some countries, such as Australia, Denmark, and the Netherlands, account ownership is even universal (Demirgüç-Kunt et  al., 2018). Yet, there is still a fraction of around 6 percent of

Figure 15.1 Account ownership. The figure shows the percentage of adults with an account

Financial inclusion in high-income countries  275

adults in the Euro area and North America who have no account. To assess the importance of this issue and understand the situations of those excluded from holding a formal account, we refer to the statistics collected by Findex in the four Euro-area countries (Cyprus, Greece, Lithuania, and the Slovak Republic) surveyed on this particular issue. Unbanked adults in the four countries typically claim that they chose unbankedness for reasons pertaining to lack of necessity or lack of trust in the financial system. By contrast, households in developing countries are often unbanked by necessity, because of the distance to financial institutions or insufficient funds to get an account. Formal Borrowing Formal borrowing means holding at least one loan from a financial institution or through a credit card. Formal borrowing in a region is measured by the percentage of respondents who report borrowing any money from a bank or another financial institution or using a credit card in the past 12 months. In 2017, 47 percent of adults worldwide reported having borrowed money in the past 12 months, including with a credit card (Demirgüç-Kunt et al., 2018).​ Figure 15.2 shows the evolution between 2014 and 20176 of the share of adults borrowing from a financial institution or through a credit card in the Euro area, North America, and developing economies. This share increased slowly in North America from 66 percent in 2014 to 70 percent in 2017. In the Euro area, it grew from 42 percent to 46 percent, while there is a small decline in developing economies, from 16 percent to 15 percent. The gap in 2017 between North America and the Euro area is surprisingly large (24 percentage points), but smaller than the difference between the Euro area and developing economies (31 percentage points).

Figure 15.2 Formal borrowing. The figure shows the percentage of adults borrowing from a financial institution or through a credit card

276  Handbook of microfinance, financial inclusion and development

Even though the European legal framework guarantees access to credit (Comparato, 2015),7 the use of credit cards as a borrowing tool is less prevalent in Europe than in North America. This difference contributes to explaining why overall indebtedness in the Euro area is lower than in North America. Formal Savings Formal savings are measured by the percentage of respondents who report saving or setting aside any money at a bank or at another financial institution in the past 12 months. In 2017, 27 percent of adults worldwide reported having saved formally (Demirgüç-Kunt et al., 2018). Figure 15.3 shows the evolution of this percentage by region. Akin to formal borrowing, the largest share of adult savers is found in North America (63 percent in 2017), followed by the Euro area (49 percent) and developing countries (21 percent). Financial Resilience Financial resilience is proxied by the percentage of respondents who report that, in case of emergency, they can come up with one-twentieth of gross national income (GNI) per capita in local currency within the next month. In 2017, 54 percent of adults worldwide claimed to be financially resilient (Demirgüç-Kunt et al., 2018). As Figure 15.4 shows, the ratio increased from 2014 to 20178 in the two developed regions but decreased in developing countries. In 2017, it reached 76 percent in the Euro area, 73 percent in North America, and 50 percent in developing countries. Figure 15.5 summarizes the main sources of emergency funds. In the Euro area and in North America, most of the emergency funds are savings. By contrast, in developing countries, they are most likely originating from family or friends or from working.

Figure 15.3 Formal savings. The figure shows the percentage of adults saving at a financial institution

Financial inclusion in high-income countries  277

Figure 15.4 Financial resilience. The figure shows the percentage of adults able to raise emergency funds

GENDER GAP AND POVERTY TRAP At the individual level, financial inclusion correlates with income (Nino-Zarazua and Copestake, 2008). Since women are statistically poorer than men, it is not surprising that the literature finds consistently that access to formal financial services is more difficult for women than for men (Garikipati et  al., 2017). However, the gender and poverty gaps in financial inclusion are not necessarily aligned. There can be additional factors, such as institutional discrimination (Demirgüç-Kunt et al., 2013) and social norms (Johnson, 2004; Guérin, 2011), which make financial inclusion more challenging to women than to sameincome men. From the second section, we know that, except for formal borrowing, the Euro area and North America share similar financial inclusion indicators. We will show here that the picture is strikingly different when one focuses on the interaction of financial inclusion with gender and poverty.9 The fourth section will then examine potential explanations for the discrepancies. Account Ownership Globally 72 percent of men and 65 percent of women own a bank account. In 2017, the gender gap was negligible in North America (0.7 percentage points) and reached 2.7 percentage points in the Euro area. The poverty gaps are estimated from household income quintiles. The largest gap between adults in the richest 60 percent of households and those in the poorest 40 percent is obtained for North America with 11.8 percentage points, while the European score is 2.5 percentage points. The American gap testifies to the extent of income inequalities in the region. Figure 15.6 features the evolutions of account ownership by gender while Figure 15.7 shows these evolutions for the two income-based segments (richest 60 percent and poorest 40 percent). In addition, Tables A15.1 and 16.2 provide the gaps computed with the 2017 absolute and relative data, respectively.

278  Handbook of microfinance, financial inclusion and development

Figure 15.5 Main sources of emergency funds

Figure 15.6 Account ownership by gender

Financial inclusion in high-income countries  279

Figure 15.7 Account ownership for the richest 60 percent and the poorest 40 percent Formal Borrowing Figure 15.8 presents the gender gaps in formal borrowing. The region with the larger gap in 2017 is the Euro area (9.7 percentage points), with a world record for this item held by Italy with 21 percentage points, followed by Bahrain (19.4 percentage points) and Saudi Arabia (18.6 percentage points). These three countries are classified as high-income by the World Bank. By contrast, the gender gap between men and women is smaller in North America (2.6 percentage points) where it even changed signs between 2014 and 2017. In North America in 2014, women were slightly more likely than men to borrow formally. Figure 15.9 reports access to formal borrowing for the richest 60 percent and the poorest 40 percent. In North America, the spectacular gap reached 30.3 percentage points in 2017, which is twice as large as the gap in the Euro area (15.3 percentage points), signaling that North America faces important challenges in terms of income inequality and financial inclusion. Singapore has the largest gap in the world (37 percentage points), followed by the United States (32 percentage points) and Hong Kong (31 percentage points). Again, these countries are high-income by the World Bank standards. Formal Savings Figures 15.10 and 15.11 bring some nuances to the claim by Demirgüç-Kunt et al. (2018) that, in high-income economies, men and women save formally alike. Considering the Euro area and North America separately, one sees that formal savings exhibit gaps like those observed for formal borrowing. In the Euro area, men are 7.2 percentage points more likely than women to save at a financial institution, which is a much larger gap than in North America where the difference is 1.3 percentage points.

280  Handbook of microfinance, financial inclusion and development

Figure 15.8 Formal borrowing by gender

Figure 15.9 Formal borrowing for the richest 60 percent and the poorest 40 percent

Financial inclusion in high-income countries  281

Figure 15.10 Formal savings by gender

Figure 15.11 Formal savings for the richest 60 percent and the poorest 40 percent

282  Handbook of microfinance, financial inclusion and development

Likewise, even though the worldwide poverty gap in formal savings is 23 percentage points in 2017, the US reached in 2017 a spectacular world record of 35 percentage points, followed again by Hong Kong (31 percentage points). Financial Resilience Figures 15.12 and 15.13 show the gaps in financial resilience between men and women, and between richer and poorer households, respectively. Here again, splitting European and American data apart adds nuances to the global picture of financial resilience in high-income economies being similar across genders (Demirgüç-Kunt et al., 2018). While the claim holds reasonably well for the Euro area where the gap is 1.8 percentage points, it is less credible for North America where men are 7.6 percentage points more likely than women to come up with emergency funds. Likewise, the 39.1 percentage-point gap between the richer and the poorer households in North America is the largest gap observed for any indicator of financial inclusion. Table A15.1 summarizes the gaps measured in absolute terms while Table A15.2 provides their relative values. Comparing the two tables shows that absolute and relative measures provide a consistent picture, suggesting that the overall gender gap in financial inclusion is larger in the Euro area whereas the corresponding poverty gap is larger in North America. The Appendix extends the two tables to the developing countries. In that case, the comparative results depend on the absolute vs. relative measurement of the gaps. In absolute terms, the larger gender gaps are split evenly between Europe and the developing countries, while North America keeps its leading position for the financial exclusion of the poor, with the notable exception of account ownership. These findings can be seen as a confirmation of the “great convergence” movement suggested by Ogden (2020). In relative terms, however, the developing countries still show worse performances for all gaps without any exception.

Figure 15.12 Financial resilience by gender

Financial inclusion in high-income countries  283

Figure 15.13 Financial resilience for the richest 60 percent and the poorest 40 percent

POTENTIAL EXPLANATIONS FOR THE FACTS The previous section shows striking differences between financial exclusion in the Euro area and in North America. The most puzzling differences relate to the gender gaps in access to account ownership, formal borrowing, and formal saving. These gaps are larger in Europe even though common wisdom does not suggest any major differences in gendered financial attitudes across the Atlantic. In this section, we speculate about mechanisms that could help rationalize the evidence in the third section. This preliminary exploration, limited to descriptive figures, is intended to open the way to further, more rigorous analysis into the determinants of the gaps and traps that can damage equal access to financial products for all in developed countries. Financial exclusion is a social threat that targets the most vulnerable segments of the population. This section follows this intuition and seeks to identify variables that could potentially be linked to financial inclusion in developed countries, with special attention to variables that are particularly sensitive to gender and poverty. The data used in this section are mostly retrieved from the 2018 country-based OECD dataset, but we also use other sources as specified for each case. The first subsection focuses on social spending and job-market risks, the second relates to poverty and inequality, and the third scrutinizes gender gaps. Social Spending and Job-Market Risks in North America and the Euro Area To put financial inclusion into perspective, we first consider the social spending and job market risks in North America and in the Euro area. The OECD 2018 data presented in Figure 15.14 show that Canada and the US allocate 17.3 percent and 18.7 percent of their GDP to their social spending,10 respectively. The Euro area average11 is 22.4 percent.

284  Handbook of microfinance, financial inclusion and development

Note:  Except for social spending in Canada (collected in 2017)

Figure 15.14 Social spending (percent of GDP): OECD data for 2018

Figure 15.15 shows that, in 2015, the US had the lowest GDP share (0.6 percent) dedicated to family benefits.12 Canada’s share was 1.2 percent, while the average value in the Euro area was 2.2 percent, with Luxembourg, Finland, and France having the largest shares (3.4 percent, 3.2 percent, and 3 percent, respectively). Overall, in line with common knowledge, we observe that North America dedicated substantially fewer resources than Europe to social goals. Another interesting variable that can interact with the use of financial services concerns the risks related to the labor market. The volatility of labor income can drive the need for income smoothing, which is a typical purpose of financial services to households (Morduch, 1995). The 2015 OECD dataset provides three indicators that can usefully help us investigate this topic: labor market insecurity, unemployment risk, and unemployment insurance. To gain space, we present the three variables together in Figure 15.16. Labor-market insecurity is the expected earnings loss associated with unemployment.13 In 2015, it reached 4.6 percent and 3.9 percent in the United States and Canada, respectively. The average value for the Euro area was 6.6 percent, with Greece, Spain, and the Slovak Republic having the largest insecurity. The unemployment risk is the monthly unemployment inflow probability times the expected average duration of unemployment spells. It was particularly low in Canada and the United States with values of 7.5 percent and 5.5 percent, respectively. Europe’s average was 10.7 percent, with large disparities and peaks for Greece, Spain, and Portugal. The indicator for unemployment insurance is a coverage rate that captures the mitigating effect of government transfers on individuals’ exposure to unemployment risk.14 The figures for unemployment insurance in 2015 corroborate those related to social spending, as the

Financial inclusion in high-income countries  285

Figure 15.15 Family benefits public spending (percent of GDP): OECD data for 2015

Note:  Except for unemployment insurance in Germany (collected in 2013)

Figure 15.16 Labor market insecurity, unemployment risk, and unemployment insurance: OECD data for 2015

286  Handbook of microfinance, financial inclusion and development

US had one of the lowest values (25.6 percent). Canada had a 48.4 percent coverage, which sits slightly above the Euro area average of 44 percent, but the American average was still below the European one. The countries with the most generous unemployment insurance are Finland, Belgium, and the Netherlands. Overall, the OECD data reveal that market insecurity and unemployment risk in North America are below the Euro area averages. The US has particularly low unemployment insurance, meaning that even if unemployment in the US is relatively infrequent, its financial consequences for households can be dramatic. Poverty and Inequality in North America and the Euro Area The Euro area includes richer countries in Europe, its 2018 GDP per capita is USD 40,000, to be compared with USD 38,200 in the European Union and USD 30,700 in the world region that gathers Europe and Central Asia.15 By comparison, the corresponding figures are USD 54,500 for North America, USD 11,500 in middle-income countries, and USD 2,000 in lowincome countries. Inequality is measured by the Gini coefficient, which compares the cumulative proportions of the population to the cumulative proportions of income. Its values range from 0 for equality and 1 for maximal inequality. Inequality is much larger in the US than in Europe (see Figure 15.17). By contrast, Canada sits close to the European average. The 2016 OECD values are 0.39 for the US, 0.31 for Canada, and 0.31 for the Euro area.16 The Slovak Republic has the smallest coefficient (0.24). The US has also the largest poverty rate measured by the ratio of the number of persons whose income falls below the poverty line, where the poverty line is half the median household income of the total population (see Figure 15.18). This measure mixes poverty and inequality since it is relative to the wealth of a country’s population. The US poverty rate is 0.18 percent, while the Euro area average is 0.11 percent. Remarkably, Finland reaches the smallest poverty rate with a value that is low enough (0.06 percent) to contrast with its neighboring

Figure 15.17 GINI coefficients: OECD data for 2016

Financial inclusion in high-income countries  287

Figure 15.18 Poverty rate: OECD data for 2016 France (0.08 percent). Canada (0.12 percent) has a score slightly higher than that observed for the Euro area. In addition, the World Bank uses an absolute measure for extreme poverty, namely the share of the population living with less than USD 1.9 a day. Figure 15.19 shows that extreme poverty is unequally spread across the Euro area where the average extreme poverty share is equal to 0.5 percent, where eight countries have a zero rate of extreme poverty. At the same time, the highest rates are observed in Southern Europe, and more precisely in Italy (2 percent) and Greece (1.5 percent). The US has the third highest ratio (1.2 percent) and Canada sits again in the middle with a 0.5 percent recorded extreme poverty. Gender Gaps in North America and the Euro Area Gender gaps can be explored in various directions. First, at the institutional level, we use the SIGI (Social Institutions and Gender Index) measured in 2019. This index is built on 27 variables gauging institutional discrimination; it ranges from 0 for no discrimination to 100 percent for maximal discrimination. In addition, the global SIGI is split into four subindices focusing on the family, physical integrity, access to productive and financial resources, and civil liberties, respectively. Second, we explore gendered differences in the labor market: unemployment rate and involuntary part-time employment. Both institutional discrimination and labor-market restrictions can impact female access to financial services. Interestingly, our exploratory analysis suggests

288  Handbook of microfinance, financial inclusion and development

Note:  Except for Canada (2013) and the US (2016)

Figure 15.19 Extreme poverty: share (in percent) of population living with less than $1.90 a day (2011 PPP): World Bank data for 2015 that, in developed countries, labor-market gendered characteristics relate to gender gaps in financial inclusion more consistently than institutional limitations imposed on women. Figure 15.20 shows the values obtained for the 2019 global SIGI. Except for Greece, the US and Canada are the worst performers. The particularly high index for Greece might be explained by the fact that, until 2018, the Greek Muslim minority used the Islamic legal system (Sharia) to address family matters, such as divorce, child custody, and inheritance rights, in a way that is often unfavorable for women.17 This is the only such situation in Europe. Overall, the SIGI worldwide picture contrasts with the fact that, financial resilience aside, female access to financial services is more developed in North America than in the Euro area. Figure 15.21 shows the gender gaps in unemployment rate computed as the differences between the same-country male and female unemployment rates. In terms of cross-country comparisons, the picture departs significantly from that provided by the SIGI: all the countries with a positive unemployment gender gap—i.e. where the female unemployment rate is higher than the male one—are located in Europe, with Greece, Spain, and Italy having the largest positive gaps. Both the US and Canada exhibit negative values. Yet, the largest negative values are found in Europe, in two Baltic states (Latvia and Lithuania) and in Germany. The average gap in the unemployment rate is 0.8 percentage points in the Euro area18 which is relatively large compared to the US and Canada negative gaps. To refine the comparison, Figure 15.22 presents the gender gap in involuntary part-time employment. As involuntary part-time jobs are dominantly female, the gap is systematically positive. However, there are significant differences between the small gap of 0.5 percentage points observed for the US and the large 13.2 percentage point gap reached by Italy. The average gap in the Euro area is 3.4 percentage points, which dominates both the US and Canadian

Financial inclusion in high-income countries  289

Figure 15.20 Social Institutions and Gender Index (SIGI): OECD data for 2019

Figure 15.21 Gender gap in unemployment rates: OECD data for 2018

290  Handbook of microfinance, financial inclusion and development

Figure 15.22 Gender gap in involuntary part-time employment: OECD data for 2018 figures of 0.5 and 2.5 percentage points, respectively. The particularly large gender gaps in Italy, Spain, and France are 13.2, 9, and 7.9 percentage points, respectively. Another interesting indicator relates to gender social norms (UNDP, 2020). Unfortunately, as Figure 15.23 shows, the Gender Social Norms Index (GSNI)19 is available only for onehalf of the countries in the Euro areas. Both the scores of the US (57.3 percent) and Canada (51.5 percent) are smaller than the Euro area average (60.2 percent). The representativeness of the Euro-area countries for which a score is available is however questionable. In addition, the data are not all measured during the same period and the scores are rather unstable: the German score increased from 59.1 percent for 2005–2009 to 62.6 percent for 2010–2014, while the US score passed from 60.6 percent for 2005–2009 to 57.3 percent for 2010–2014. In sum, the GSNI seems relatively uniform within high-income countries, so a more refined analysis is necessary to assess whether (part of) the gender gap in financial inclusion in these countries is driven by gender social norms. Overall, basic statistical facts seem to indicate that the large poverty gap in financial inclusion observed in North America is well aligned with the usual indicators for poverty and inequality.20 North America is indeed less generous than Europe on social spending and unemployment coverage, and its wealth distribution is more unequal than in the Euro area. The factors related to the gender gap in financial inclusion are less clear-cut. SIGI, the typical indicator for the level of gender discrimination at the institutional level, seems to have little connection with the evidence that in the Euro area, women—compared to men—have less access to financial services than their North American counterparts. By contrast, the figures relating to gender gaps in unemployment risks, especially in involuntary part-time employment, match well the evidence on the gender gap in financial inclusion. Overall, our preliminary investigation suggests that labor-market characteristics offer a promising avenue to analyze the determinants of the gender gap and the poverty trap in financial inclusion.

Financial inclusion in high-income countries  291

Note:  Except for Canada, Finland, France, and Hungary with most recent data for 2005–2009

Figure 15.23 Gender Social Norms Index: UN data for 2010–2014 period

CONCLUSION AND AVENUES FOR FUTURE RESEARCH The release of the Findex dataset represented a major step toward understanding financial inclusion over the world. Yet, the situation in high-income countries received less attention, probably due to the existence of well-developed, and supposedly inclusive, financial markets and institutions. So far, scholars hardly discuss correlations between the characteristics of financial inclusion in high-income countries and other meaningful social and macroeconomic variables, let alone causal linkages. To fill the gap, we provide inputs to stimulate the conversation. This goal explains the exploratory nature of this chapter, and its interest in the four dimensions of financial inclusion, with special attention to the gender and poverty gaps in the Euro area and North America. Financial exclusion has been scrutinized mainly in low-income economies (Allen et  al., 2016). Using the Findex dataset, we confirm that both North America and the Euro area perform better than developing countries in terms of financial inclusion. Less expectedly, we also observe that high-income economies are far from homogenous when it comes to access to financial services. Comparing data from the Euro area and North America, we detect important discrepancies in the financial inclusion of women and poor households. In the Euro area, financial inclusion appears to be challenging for women, while the poor North American households are particularly underserved. Further, our exploratory search for factual indicators reveals that region-wise poverty gaps in financial inclusion run parallel to the usual measures of inequality, which corroborates the intuition. Regarding the gender gap, our findings suggest that labor-related variables, such

292  Handbook of microfinance, financial inclusion and development

as the gendered unemployment rate, are more promising explanations than institutional gender gaps and gendered social norms. These preliminary observations confirm the key role of financial products in consumption smoothing, so favoring demand-side arguments over supply-side ones in explaining financial inclusion in industrialized countries. If so, this would represent a notable difference from developing countries, where financial inclusion is understood as a market gap due to supply shortage (Armendariz and Morduch, 2010). In that regard, our findings are reminiscent of those reported by Fungáčová and Weill (2015) about China, where financial coverage is wide and financial exclusion is mostly voluntary. Our findings are still too preliminary to be used for policy recommendations. Further work is needed not only to assess their validity but also to inform policymakers about the fate of specific groups at risk of financial exclusion, such as the unemployed and the migrants. Generally, scrutinizing differences in access to financial services with respect to race, age, marital status, household composition, citizenship, and location,21 could reveal interesting features of financial inclusion in high-income countries and so help scholars develop fruitful models in this new field of economic investigation. Overall, this paper shows that, despite the strong presence of financial institutions in high-income countries, their universal financial inclusion should not be taken for granted.

ACKNOWLEDGMENTS The authors thank Jonathan Morduch, Petros Sekeris, and Ilan Tojerow for useful comments and discussions. This study was carried out within the framework of the Montpellier Business School (MBS) Social and Sustainable Finance Chair funded by the Caisse d’Epargne Languedoc Roussillon.

NOTES 1. See Cozarenco and Szafarz (2018, 2019) and Cozarenco and Bourlès (2017) on microfinance in developed countries. 2. The Euro area includes 19 European countries, mainly located in Western and Northern Europe: Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Portugal, Slovakia, Slovenia, and Spain. See Corrado and Corrado (2015) for more information about financial inclusion in Eastern Europe vs. Western Europe. 3. The data are adjusted to ensure sample representativity. The adjustments include base sampling weighting (to correct for unequal probability of selection with respect to household size) and poststratification weighting (to correct for sampling error and non-response error). In 2017, the Findex team collected interviews from around 150,000 respondents located in 144 countries, with approximately 1,000 randomly selected representative individuals in each country. See Demirgüç-Kunt et al. (2018) for more details. 4. According to the World Bank classification, the developing countries are defined as having low and middle income. As of 2017, low-income countries were characterized by a gross national income (GNI) per capita below USD 1,006 while middle-income countries have a GNI per capita between USD 1,006 and USD 12,235. The 100 developing countries surveyed by Findex 2017 include 24 low-income and 76 middle-income countries. 5. Skinner (1988) and Dynan (1993) find little evidence that US households save with a precautionary motive. 6. Findex data for formal borrowing are available only for 2014 and 2017. 7. However, UK banks keep denying services to lower-income customers (Sinclair, 2013). 8. For this indicator, data is available only for 2014 and 2017.

Financial inclusion in high-income countries  293

9. 10.

11. 12. 13. 14.

15. 16. 17. 18. 19.

20. 21.

This section compares only the Euro area and North America. The Appendix provides a summary of the gaps in the three regions analyzed in the second section. The social expenditures comprise cash benefits, direct in-kind provision of goods and services, and tax breaks with social purposes. Benefits may be targeted at low-income households, the elderly, disabled, sick, unemployed, or young persons. To be considered “social,” programs must involve either redistribution of resources across households or compulsory participation. Social benefits are classified as public when controlled by the government. Net total social expenditures include both public and private expenditure. They also account for the direct and indirect tax effects. Unless stated otherwise, we compute the figures for the Euro area as unweighted averages of country data for the OECD members in the Eurozone, thus excluding Cyprus and Malta. Family benefits spending refer to public spending on family benefits, including financial support that is exclusively for families and children. The most recent figures are dated 2015. See OECD (2020) doi: 10.1787/8e8b3273-en (accessed on January 17, 2020). It is computed from the risk of becoming unemployed, the expected duration of unemployment and the government transfers to the unemployed. According to the OECD definition, the effective unemployment insurance is the coverage rate of unemployment insurance (UI) times its average net replacement rate among UI recipients plus the coverage rate of unemployment assistance (UA) times its net average replacement rate among UA recipients. The average replacement rates for recipients of UI and UA takes into account family benefits, social assistance, and housing benefits. Source: https://stats​.oecd​.org​/Index​.aspx​? DataSetCode​=JOBQ#. The data are extracted from the World Bank (https://data​.worldbank​.org​/indicator​/ NY​.GDP​.PCAP​. PP​.KD), accessed on January 17, 2020. The amounts are expressed in 2011 PPP-converted USD. The Euro area average is retrieved from Eurostat. (http://appsso​.eurostat​.ec​.europa​.eu​/nui​/show​.do​ ?lang​=en​&dataset​=ilc​_di12), accessed on March 3, 2020. www​.theguardian​.com ​/world ​/2018​/jan ​/10​/ historic​-step​-greek​-pm​-hails​-change​-to​-limit​-power​-of​sharia​-law. OECD (2020), “Unemployment rate (indicator).” doi: 10.1787/997c8750-en (accessed on March 2, 2020). The GSNI aims to capture how social beliefs can harm gender equality. We use the main GSNI, which denotes a country’s share of people surveyed with at least one bias among seven possibilities (two political biases, one educational bias, two economic biases, and two biases referring to physical integrity). Park and Mercado (2015) and Zins and Weill (2016) reach similar conclusions for Asia and Africa, respectively. Location refers mainly to urban vs. rural differences (Summers, 2019).

REFERENCES Agier, I. and A. Szafarz (2013). Microfinance and Gender: Is There a Glass Ceiling on Loan Size? World Development, 42, 165–181. Allen, F., A. Demirgüç-Kunt, L. Klapper, and M.S.M. Peria (2016). The Foundations of Financial Inclusion: Understanding Ownership and Use of Formal Accounts. Journal of Financial Intermediation, 27(5), 1–30. Anginer, D., A.C. Bertay, R.J. Cull, A. Demirgüç-Kunt, and D.S. Mare (2019). Bank Regulation and Supervision Ten Years after the Global Financial Crisis. Policy Research Working Paper WPS 9044, World Bank Group. Armendáriz, B. and J. Morduch (2010). The Economics of Microfinance. MIT Press. Bourlès, R. and A. Cozarenco (2017). Entrepreneurial Motivation and Business Performance: Evidence from a French Microfinance Institution. Small Business Economics, 51, 943–963. Burietz, A., K. Oosterlinck, and A. Szafarz (2017). Europe vs. the US: A New Look at the Syndicated Loan Pricing Puzzle. Economics Letters, 160, 50–53. Carey, M. and G. Nini (2007). Is the Corporate Loan Market Globally Integrated? A Pricing Puzzle. Journal of Finance, 62(6), 2969–3007.

294  Handbook of microfinance, financial inclusion and development

Comparato, G. (2015). The Rationales of Financial Inclusion in the Changing European Private Law. European Review of Contract Law, 11(1), 22–45. Corrado, G. and L. Corrado (2015). The Geography of Financial Inclusion across Europe during the Global Crisis. Journal of Economic Geography, 15(5), 1055–1083. Cozarenco, A. and A. Szafarz (2018). Gender Biases in Bank Lending: Lessons from Microcredit in France. Journal of Business Ethics, 147(3), 631–650. Cozarenco, A. and A. Szafarz (2019). Microfinance in the North: Where Do We Stand? In M. Hudon, M. Labie, and A. Szafarz (Eds), Research Agenda for Financial Inclusion and Microfinance. Edward Elgar Publishing, pp. 125–137. Cull, R. and J. Morduch (2018). Microfinance and Economic Development. In T. Beck and R. Levine (Eds), The Handbook of Finance and Development. Edward Elgar Publishing, pp. 550–572. Deaton, A. (1992). Understanding Consumption. Oxford University Press. Demirgüç-Kunt, A., L. Klapper, and D. Singer (2013). Financial Inclusion and Legal Discrimination Against Women: Evidence from Developing Countries. The World Bank. Demirgüç-Kunt, A., L. Klapper, and D. Singer (2017). Financial Inclusion and Inclusive Growth: A Review of Recent Empirical Evidence. The World Bank. Demirgüç-Kunt, A., L. Klapper, D. Singer, S. Ansar, and J. Hess (2018). The Global Findex Database 2017: Measuring Financial Inclusion and the Fintech Revolution. The World Bank. Demirgüç-Kunt, A., L. Klapper, D. Singer, and P. Van Oudheusden (2015). The Global Findex Database 2014: Measuring Financial Inclusion around the World. The World Bank. D’Espallier, B., I. Guerin, and R. Mersland (2013). Focus on Women in Microfinance Institutions. Journal of Development Studies, 49(5), 589–608. Dynan, K.E. (1993). How Prudent Are Consumers? Journal of Political Economy, 101(6), 1104–1113. Fungáčová, Z. and L. Weill (2015). Understanding Financial Inclusion in China. China Economic Review, 34, 196–206. Garikipati, S., S. Johnson, I. Guérin, and A. Szafarz (2017). Microfinance and Gender: Issues, Challenges and the Road Ahead. Journal of Development Studies, 53(5), 641–648. Guérin, I. (2011). The Gender of Finance and Lessons for Microfinance. In B. Armendariz and M. Labie (Eds), The Handbook of Microfinance. World Scientific, pp. 589–612. Hartarska, V., D. Nadolnyak, and R. Mersland (2014). Are Women Better Bankers to the Poor? Evidence from Rural Microfinance Institutions. American Journal of Agricultural Economics, 96(5), 1291–1306. Johnson, S (2004). Gender Norms and Financial Markets: Evidence from Kenya. World Development, 32(8), 1355–1374. Lapavitsas, C. and J. Powell (2013). Financialisation Varied: A Comparative Analysis of Advanced Economies. Cambridge Journal of Regions, Economy and Society, 6(3), 359–379. Morduch, J. (1995). Income Smoothing and Consumption Smoothing. Journal of Economic Perspectives, 9(3), 103–114. Morduch, J. and R. Schneider (2017). U.S. Financial Diaries: How American Families Cope in a World of Uncertainty. Princeton University Press. Nino-Zarazua, M.M. and J. Copestake (2008). Financial Inclusion, Vulnerability and Mental Models: From Physical Access to Effective Use of Financial Services in a Low-Income Area of Mexico City. Savings and Development, 32(4), 353–379. Ogden, T. (2020). The Great Convergence: Toward a Global Strategy for Financial Inclusion, The Aspen Institute Financial Security Program. Park, C.-Y. and R. Mercado (2015). Financial Inclusion, Poverty, and Income Inequality in Developing Asia. Asian Development Bank Economics Working Paper Series No. 426. Périlleux, A. and A. Szafarz (2015). Women Leaders and Social Performance: Evidence from Financial Cooperatives in Senegal. World Development, 74, 437–452. Pew Charitable Trusts (2015). The Complex Story of American Debt. July 2015. http://www​.pewtrusts​ .org/~​/media ​/assets​/2015​/07​/reach​-of​-debt​-report​_artfinal​.pdf. Rochet, J.C. (2009). Why Are There so Many Banking Crises? The Politics and Policy of Bank Regulation. Princeton University Press. Sinclair, S. (2013). Financial Inclusion and Social Financialisation: Britain in a European Context. International Journal of Sociology and Social Policy, 33(11/12), 658–676.

Financial inclusion in high-income countries  295

Skinner, J. (1988). Risky Income, Life Cycle Consumption, and Precautionary Savings. Journal of Monetary Economics, 22(2), 237–255. Summers, G.F. (2019). Persistent Poverty in Rural America. Routledge. United Nations Development Programme (UNDP) (2020). Tackling Social Norms: A Game Changer for Gender Inequalities. Human Development Perspectives, https://doi​.org​/10​.18356​/ff6018a7​-en. Wilson, J.B. (1987). Women and Poverty: A Demographic Overview. Women & Health, 12(3–4), 21–40. Worobey, J.L. and R.J. Angel (1990). Poverty and Health: Older Minority Women and the Rise of the Female-Headed Household. Journal of Health and Social Behavior, 31(4), 370–383. Zins, A. and L. Weill (2016). The Determinants of Financial Inclusion in Africa. Review of Development Finance, 6(1), 46–57.

296  Handbook of microfinance, financial inclusion and development

APPENDIX 15.1 Table A15.1  Absolute gaps in 2017 (larger gaps in bold) The absolute gap is computed as: • percentage for men – percentage for women • percentage for the richest 60 percent – percentage for the poorest 40 percent Men vs. women

Richest 60% vs. poorest 40%

Euro area

North America

Developing economies

Euro area

North America

Developing economies

Account ownership

2.7

0.7

8.8

2.5

11.8

14.8

Formal borrowing

9.7

2.6

3.8

15.3

30.3

8.2

Formal saving

7.2

1.3

5.9

22.0

33.9

15.2

Financial resilience

1.8

7.6

10.8

21.0

39.1

27.2

Table A15.2  Relative gaps in 2017 (larger gaps in bold) The relative gap is computed as: • (percentage for men – percentage for women) / percentage for men • (percentage for the richest 60 percent – percentage for the poorest 40 percent) / percentage for the richest 60 percent

​​

 

Men vs. women

Richest 60% vs. poorest 40%

Euro area

North America

Developing economies

Euro area

North America

Developing economies

Account ownership

2.8

0.8

13.1

2.6

11.9

21.4

Formal borrowing

19.0

3.6

22.2

29.4

36.9

43.9

Formal saving

13.7

2.0

25.2

38.2

44.5

57.2

Financial resilience

2.3

9.9

19.3

24.9

44.1

44.4

16. Financial literacy and the use of financial services by US households James R. Barth, Valentina Hartarska, Jitka Hilliard, and Nguyen Nguyen

INTRODUCTION There is a substantial and growing interest in the financial literacy of individuals and their households. As new and more complex financial services have become more readily available in the marketplace, a serious concern is that compared to a financially literate individual, less financially literate individuals will use services from financial firms that turn out to be less appropriate and more costly. The use of such services would make individuals worse off from a financial well-being standpoint. In particular, individuals who rely mainly on traditional financial services from banks typically benefit from less costly services, while those relying more heavily on alternative services provided by payday lenders and pawnshops, among other providers, typically incur higher costs for the same or similar services. To address the concern over the disparity in the use of services with quite different costs, there is an effort in many countries to promote “financial inclusion,” which refers to the notion of making more widely available appropriate and affordable financial services. However, even though such services may already be available, the degree of financial literacy of individuals and their households may differ so widely in a country that the use of such services is quite limited. Therefore, it is important to measure financial literacy and examine the relationship between financial literacy and the use of various financial services provided by banks and other financial firms. In this way, one can better determine the association between financial literacy and financial well-being among individuals. The purpose of this paper is to examine the relationship between the use of different financial services by individuals and households and their degree of financial literacy, controlling for a variety of individual and household characteristics. Although there is substantial literature on financial literacy, as far as we know, our chapter is the first to examine whether financial literacy is associated with the probability and the extent of the use of different types of financial services in each state and nationwide of the USA. The remainder of the chapter proceeds as follows. In the next section, we provide a review of the literature pertaining to the importance of financial literacy and the way in which it relates to various outcomes and, thereby financial well-being. The third section discusses the various datasets used to examine the linkages between financial literacy and the use of different financial services. The fourth section discusses our empirical model and the empirical results. The last section contains a summary and suggestions for future research.

LITERATURE REVIEW The literature on financial literacy is growing. It deals with defining and measuring the term financial literacy and uses various measures of financial literacy to examine whether financial 297

298  Handbook of microfinance, financial inclusion and development

literacy matters for various decisions individuals make, for example, regarding saving and investing. Lusardi and Mitchell (2014) review the relevant literature and provide an excellent assessment of what we know about financial literacy. Indeed, they summarize the information provided in more than 190 papers, with the earliest paper published in 1967 and the latest in 2013. Importantly, they conclude that the literature shows that large numbers of individuals around the world are financially illiterate. In addition, they point out that econometric models and experiments contributed to establishing a causal link between financial literacy and economic decision-making. Given the information provided in this paper, we focus on selected papers published after 2013, all of which examine the relationship between financial literacy and various outcomes associated with financial inclusion and well-being. In this regard, Almenberg and Dreber (2015) point out that women participate less than men in the stock market. Based on measures of financial literacy and a random sample of 1,300 adults living in Sweden in 2010, they find that gender differences in financial literacy explain a significant part of the gender gap in stock market participation. In a related paper, Bannier and Neubert (2016) examine the role that financial literacy and risk tolerance play in investment decisions by men and women based on 2,047 German households. They conclude that it is important to raise the financial literacy and risk tolerance of women to close the gap between men and women with regard to participation in standard investments. More recently, Potrich, Vieira, and Kirch (2018) develop an indicator to assess financial literacy levels and analyze gender differences. Based on a survey of 2,485 individuals in Brazil, they find that both men and women have a low level of financial literacy. Yet, they also find that among individuals with high levels of financial literacy, the proportion of males is higher than females. They state that further results indicate that single women with low levels of education and low family and own incomes have the highest tendency toward membership in the group with low financial literacy levels. In a study focused on student debt, Artavanis and Karra (2020) examine the financial literacy of college students and its implications for the repayment of student debt. Their examination is based on 1,040 students at a public university in Massachusetts in 2017. Their study indicates that students with a deficit in financial literacy are more likely to significantly underestimate future loan payments, which could impair their ability to repay their student debt. In addition, they find that low-literacy students systematically expect lower starting salaries than their high-literacy peers. They conclude that future research is needed to determine whether the higher default rate of student borrowers, despite their lower debt burden, is due to a lack of financial literacy, poor labor outcomes, or both. Focusing on a different issue, Von Gaudecker (2015) examines the relationship between financial literacy and portfolio choice for 381 Dutch households in 2005 and 2006. He finds that households that score high on financial literacy or rely on professionals or private contacts for advice achieve better investment outcomes than those with below-median financial literacy that trust their own decision-making capabilities. In a related paper, Bianchi (2018) examines mechanisms connecting financial literacy and portfolio returns. He obtains data on portfolio choices from a French financial institution and the resulting returns as well as a measure of financial literacy for 511 clients from September 2002 to April 2011. His findings indicate that the most literate households experience 0.4 percent higher annual returns than the least literate households. In addition, he finds that more literate households hold riskier positions when expected returns are higher and more actively rebalance their portfolios, doing so in a way that holds their risk exposure relatively constant over time.

Financial literacy and the use of financial services by US households  299

In contrast to these papers, Allgood and Walstad (2016) examine the relationship between a measure of financial literacy that combines actual (objective) financial literacy and perceived (self-assessed) financial literacy and financial behavior for 28,146 US individuals and households in 2009. Although they are unable to find a causal relationship, they do find that financial literacy affects financial behavior regarding credit cards, investments, loans, insurance, and financial counseling. In addition, they find that perceived financial literacy may be as important as actual financial literacy when considered separately. In a related paper, Lusardi and Mitchell (2017) examine whether individuals are financially literate and whether self-assessed and objective measures of financial literacy result in enhanced retirement planning. Based upon a sample of 989 US individuals who average 45 years of age, they find that financial literacy is a key determinant of retirement planning and that financial literacy is higher when individuals are exposed to economics in school and employer-sponsored retirement programs. In a novel and important paper, Lusardi, Michaud, and Mitchell (2017) develop a multiperiod theoretical model to explore the forces that shape financial knowledge accumulation over the life cycle and the extent to which differences in financial knowledge can explain wealth inequality. In addition, they evaluate the types of individuals that would benefit most from investment in financial knowledge and the use of sophisticated investment products. Their simulation results indicate that 30 to 40 percent of US wealth inequality may be attributed to financial knowledge. Interestingly, they show that some level of financial ignorance may actually be optimal. They explain this finding by pointing out that it is expensive to acquire financial knowledge and that not everyone benefits from greater financial sophistication. As a result, they argue, some individuals will remain financially ignorant. Turning to what is happening globally, Klapper and Lusardi (2019) examine financial literacy in more than 140 countries. They find that just one in three adults are financially literate. They also find that the percentage of financially literate adults ranges from a low of 14 percent in Afghanistan and Albania to a high of 71 percent in Norway, Sweden, and Denmark. Importantly, they rely on the first and only global survey on financial literacy to document that low levels of financial literacy occur not only in developing countries but also in countries with well-developed financial markets. One of their important findings is that there is a positive and significant relationship between the use of financial services and financial literacy. They argue that relatively low levels of financial literacy exacerbate individual and financial market risks in a marketplace with a growing number of complex financial services. Celerier and Matray (2019) examine the relationship between financial inclusion, wealth accumulation, and financial well-being. Specifically, they find that following US branching deregulation, both the supply of bank branches and financial inclusion increased. In addition, they find that financial inclusion positively affects low-income household wealth accumulation. Examining the channel behind this increase, they find that having a bank account allows households to not only accumulate liquid assets, but also permanent assets. More generally, their results indicate that improving access to a bank account increases wealth accumulation in low-income households. Dunham (2019) examines whether sociodemographic characteristics and mortgage lending variables have a predictive relationship on the presence of census tracts where alternative financial service providers like check-cashing outlets are more prevalent than banks in southeastern Pennsylvania. The empirical results indicate that the prevalence of more

300  Handbook of microfinance, financial inclusion and development

check-cashing outlets than banks is predicted based upon a lower median household income, a higher percentage of Black and Latin residents, and a lower percentage of residents age 65 or above, among other factors. On a global level, Allen et al. (2016) examine the individual and country characteristics associated with financial inclusion and the policies that are effective among those most likely to be excluded. Their analysis is based on data on over 124,000 individuals in 123 countries and is obtained from the 2011 Global Findex database. The focus is on three indicators of account use: (1) ownership of an account, (2) use of the account to save, and (3) frequent use of the account (defined as three or more withdrawals per month). They find that financial inclusion is associated with lower account costs, proximity to financial intermediaries, stronger legal rights, and more politically stable environments. The effectiveness of policies to promote inclusion, moreover, is found to vary depending on the characteristics of the individuals considered. In addition, they find that among those that report lack of money as the main barrier to account use, government policies to promote inclusion can increase the likelihood that individuals perceive financial services as being within their reach.

DATA Data used in our analysis is for individuals/households within a particular county. This includes information from the Financial Industry Regulatory Authority (FINRA) Foundation’s National Financial Capability Study (NFCS), which contains the results of four surveys of households throughout the US for the years 2009, 2012, 2015, and 2018. Each survey contains information on not only the financial literacy of individuals but also information on the individual’s household use of various financial services, including those from banks, payday lenders, pawnshops, rent-to-own firms, auto-title-loan companies, and revenue anticipation loans. The same dataset includes demographic characteristics of the surveyed individuals and their households, which serve as control variables. Additional control variables are the number of bank offices and credit union (or depository institution) offices and establishments of alternative financial services providers, which are obtained from the Federal Deposit Insurance Corporation (FDIC), National Credit Union Administration (NCUA), and Bureau of Labor Statistics. Lastly, such control variables as the poverty rate, unemployment rate, and median household income are included as well as the population of each county to scale the number of offices and establishments of the depository institutions and alternative financial services providers. These data are obtained from the American Community Survey. The data on an individual and an individual’s household is available at the zip code level. Matching this data to counties relies on the Department of Housing and Urban Development’s Zip Code Crosswalk Files. Our analysis is therefore based on the relationship between the financial literacy of individuals within households in counties and their use of financial services provided by a bank, a bank and an alternative financial services provider, or only an alternative financial services provider. In addition, we consider the relationship between financial literacy and the use of the different types of alternative financial services. The key explanatory variable in our analysis is financial literacy. In the NFCS surveys, the answers to five questions are the basis for measuring financial literacy.1 Based on the number of correct answers to these questions, financial literacy ranges from zero to five. Individuals, however, may answer that they do not know the answer or prefer not to say to each of the questions. Individuals who answer any of the questions with a “do not know” are assigned a

Financial literacy and the use of financial services by US households  301

score of zero for that question, while individuals who answer any of the five questions with a “prefer not to say” to any question are excluded from the dataset. As a result, 1 percent of the individuals surveyed are excluded from our sample. The specific questions used in the NCFS survey are included in Appendix 16.1. In total, there are five questions. The questions focus on the interest rate, inflation, bond price, mortgage, and risk. Figure 16.1a shows the average financial literacy of individuals across states in 2018. Financial literacy ranges from a high of 3.08 in New Hampshire to a low of 2.50 in Mississippi. Of all the states, 29 states have scores exceeding the US average, or 57 percent. In contrast, Figure 16.1b shows the ranking of states in terms of financial literacy in 2009. The state with the highest score is still New Hampshire, at 3.34, but Louisiana now has the lowest score, at 2.82. In 2009, 34 states, i.e. five more states than in 2018, have scores exceeding the US average, or 67 percent. The average score for the entire US in 2009 is 3.04, which is higher than that in 2018, by 0.30, or roughly 10 percent. The US average was 60 percent of a perfect score in 2009 but declined to 55 percent in 2018. Table 16.1 provides information on the characteristics of the individuals and their households for each of the five questions used to measure financial literacy. In addition, it indicates the distribution of the individuals according to various characteristics and degrees of financial inclusion. The table also includes the percentage of individuals answering each of the questions correctly and answering four or more questions correctly. The table, moreover, includes the percentage of individuals answering all five questions correctly. As may be seen, the question the highest percentage of individuals answered correctly relates to mortgage, at 74 percent, while the lowest percentage of individuals answered correctly question related to bond price, at 26 percent. Overall, only 35 percent of the individuals answered four or more questions correctly, and a much lower 13 percent answered all correctly. In addition, the households of the individuals answering the questions with a bank account have the highest percentage of correct answers for four or more questions, at 45 percent, as well as all five questions, at 18 percent. Other interesting information in Table 16.1 is that males score higher than females overall, both for correctly answering four or more questions and all five questions. As regards ethnicity, white non-Hispanic individuals tend to score higher than other groups on the individual questions, with the exception of Asian non-Hispanic for a few questions. Overall, however, the percentage of Asian non-Hispanics answering four or more and all five questions correctly is the highest, at 41 and 19 percent, respectively, while Black non-Hispanic has the lowest percentages, at 15 and 4 percent, respectively. Other interesting information in the table indicates that the percentages of correct answers tend to increase with age, income, and education. There are four key outcome variables in our empirical analysis. These are as follows: (1) banked without AFSs, (2) underbanked, i.e. banked with AFSs, (3) unbanked with AFSs, and (4) unbanked without AFSs. In 2018, there are 120 million households, with 70 percent being banked without AFSs, 24 percent being underbanked, 3 percent being unbanked with AFSs, and 2 percent being unbanked without AFSs households. Figure 16.2 shows these percentages are fairly constant over the four survey years. Focusing more directly on the use of AFSs, Figure 16.3 shows the percentage of households using one to five different types of these services for the four survey years. Figure 16.3a shows the percentages for the underbanked, while Figure 16.3b shows the percentages for unbanked with AFSs. As may be seen, the largest percentage of underbanked use only one type of AFS, ranging from just under 50 percent to 60 percent. In second place is the use of two types of

302

Figure 16.1 Financial literacy by state, 2009 and 2018

Source:  National Financial Capacity Survey

303

100

All individuals

3

24

70

  Unbanked with AFSs

 Underbanked

  Banked without AFSs

51

 Female

12

16

6

2

  Black alone NH

  Hispanic (any race)

  Asian alone NH

  Other NH

12

8

10

9

7

8

9

9

 18–24

 25–29

 30–34

 35–39

 40–44

 45–49

 50–54

 55–59

Age

64

  White alone NH

Ethnicity

49

 Male

Gender

3

  Unbanked without AFSs

Financial inclusion

Weighted percent

Variable

76

76

76

74

68

67

64

62

78

76

68

61

77

70

77

81

66

52

56

73

Interest rate

68

64

60

50

43

37

34

36

57

58

45

35

62

50

62

66

40

32

34

56

Inflation

30

26

25

22

21

19

19

22

23

34

22

19

28

21

32

30

23

15

20

26

Bond price

80

78

76

74

70

70

61

54

71

70

66

58

79

72

76

81

69

45

45

74

Mortgage

50

49

46

43

38

34

31

30

43

51

34

28

49

35

53

52

32

22

24

44

Risk

44

41

36

32

26

21

16

16

33

41

22

15

41

26

43

45

19

12

8

35

Four or more correct answers

Table 16.1  Measuring financial literacy: percentage of individuals correctly answering questions, 2018

17

14

13

10

8

6

4

4

11

19

7

4

16

8

19

18

6

2

2

13

(Continued)

All five correct answers

304

9

19

 60–64

 65+

11

11

15

19

14

12

6

  $15,000 to $25,000

  $25,000 to $35,000

  $35,000 to $50,000

  $50,000 to $75,000

  $75,000 to $100,000

  $100,000 to $150,000

  $150,000 or more

14

  College graduate

  Postgraduate education

Source: National Financial Capacity Survey

37

38

  Some college

11

  High school graduate

Education

12

  Less than $15,000

Income level

Weighted percent

Variable

Table 16.1  (Continued)

86

80

75

59

87

84

77

77

73

69

65

56

83

79

Interest rate

Inflation

76

64

54

41

77

70

59

60

55

49

46

37

78

72

Bond price

42

32

23

17

44

37

31

27

23

21

19

16

36

33

Mortgage

86

80

74

59

87

88

83

79

74

69

62

49

85

83

Risk

69

53

40

30

70

63

49

47

40

35

30

25

57

52

Four or more correct answers

62

44

31

16

65

55

42

38

30

24

19

14

52

47

30

18

10

5

33

26

18

14

9

7

5

3

24

21

All five correct answers

Financial literacy and the use of financial services by US households  305

Notes:  Alternative financial services (AFSs) include auto title loans, payday loans, refund anticipation loans, pawnshops, and rent-to-own stores. In 2015, refund anticipation loans are not included in AFSs Source:  National Financial Capacity Survey

Figure 16.2 Percentage of households banked, underbanked, and unbanked, 2009–2018 AFSs, at about 20 percent. Interestingly, the use of five types of AFSs was close to zero in 2009 but increased to 16 percent in 2018. In terms of the different types of AFSs used by the unbanked, Figure 16.3b shows that the pattern is fairly similar across the four surveys, with the use of one type of AFSs accounting for the largest percentage. Focusing on only AFSs, Figure 16.4 shows the percentages of households using the five different types in 2018. Clearly, households use pawnshops the most frequently, followed by payday loans, auto title loans, rent-to-own, and tax anticipation loans, in that order. The percentage of households using the AFSs ranges from a low of 10 percent to a high of 19 percent. In Figure 16.5, states are ranked by the percentage of households that are banked without AFSs. The state with the lowest percentage is Louisiana, at 58 percent, while the state with the highest percentage is Hawaii, at 82 percent. At the same time, the states with the lowest percentages of households that are banked without AFSs tend to have the highest percentages of households that are underbanked. Perhaps not surprisingly, the figure also shows that there is a negative and significant relationship between the poverty rate and the percentage of households that are banked without AFSs.

EMPIRICAL MODEL AND RESULTS As already discussed, we examine the relationship between financial literacy and whether households are: (1) banked without AFSs, (2) underbanked, (3) unbanked with AFSs, and (4) unbanked without AFSs. Specifically, we estimate the following model:

306

Figure 16.3 Underbanked and unbanked: percentage of households using different types of AFSs, 2009–2018

Source:  National Financial Capacity Survey

Notes:  Alternative financial services (AFSs) include auto title loans, payday loans, refund anticipation loans, pawnshops, and rent-to-own stores. In 2015, refund anticipation loans are not included in AFSs

Financial literacy and the use of financial services by US households  307

Source:  National Financial Capacity Survey

Figure 16.4 Percent of households using different types of AFSs, 2018

Yi ,c = a + b1 ´ Financial Literacyi ,c

n

+

å

bk ´ HHControlsi , c +

k =1

m

å

b j ´ CountyControlsc + ei ,c ,

(16.1)

j =1

where Yi ,c represents the four outcome variables described above, Financial Literacyi , c � is measured as described earlier, HHControlsi ,c refer to the characteristics of individuals in households responding to the NFCS survey, CountyControlsc refer to county-level variables, i refers to individuals in households, c refers to county, and ei , c � is a random error term. A logit regression model and an ordered logit model are used to estimate the model. The empirical results are presented in Tables 16.2 and 16.3. Table 16.2 shows the relationship between financial literacy and banked without AFSs, underbanked, unbanked with AFSs, and unbanked without AFSs, as well as whether a household is simply unbanked or not. The results indicate that a higher level of financial literacy is positively and significantly associated with a household being banked without AFSs, while negatively and significantly associated with a household being underbanked, unbanked with AFSs, and unbanked without AFSs. In addition, the table shows that the likelihood of a household being unbanked is negatively and significantly associated with a higher level of financial literacy. This means that financial literacy does indeed matter with respect to the type of financial services used by households. In addition, the likelihood of households being banked without AFSs is positively and significantly associated with individuals in households having attended higher levels of education. Moreover, compared to white, both Black non-Hispanic and Hispanic individuals in households have a lower likelihood of being banked without AFSs. Furthermore, the results for older individuals and individuals with higher income levels are essentially the same as

308

Figure 16.5 Percentages of households banked, underbanked, and unbanked by state, 2018 (dotted line represents relationship between percentage of household banked without AFSs and poverty rate)

Source:  National Financial Capacity Survey and American Community Survey

309

0.240***

Financial literacy (0 (Min) to 5 (Max))

–0.179** –0.302*** 0.796*** 0.236***

0.193*** 0.190*** 0.146* 0.466*** 0.542*** –0.852*** –0.255*** 0.107

  Dependent children (Number of children)

 Married (Yes or no)

  Some college (Yes or no)

  College graduated (Yes or no)

 Postgraduate (Yes or no)

  Black non-Hispanic (Yes or no)

 Hispanic (Yes or no)

  Asian non-Hispanic (Yes or no)

0.000 0.143***

 Unemployed (Yes or no)

  Income level (1 (Min) to 8 (Max)) 0.092 –0.906*** 0.412**

 Bank_density (Number of offices per 1,000 people)

 AFS_density (Number of establishments per 1,000 people)

  Ln(1+median household income)

County control variables

–0.031***

 Age (Years)

–0.317*

0.533**

–0.258

–0.060***

–0.430***

0.119

–0.221** 0.031***

  Other races (Yes or no)

–0.124

0.067

–0.126**

–0.169***

0.340***

–0.315***

–0.196***

(2) Underbanked

 Female (Yes or no)

Personal and household control variables

(1) Banked without AFSs

A: Dependent variable

Table 16.2  Banked, underbanked, and unbanked: logistic regression results, 2018

–1.619***

1.090*

0.431

–0.415***

0.359*

–0.014***

0.416

0.142

0.043

0.296*

–1.467***

–1.052***

–0.412***

–0.255

–0.201***

–0.129

–0.203***

(3) Unbanked with AFSs

–0.149

1.470*

0.512

–0.425***

0.784***

0.001

0.538*

–0.121

0.435*

0.167

–0.784**

–0.595**

–0.386**

–0.437**

–0.032

–0.103

–0.299***

(4) Unbanked without AFSs

(Continued)

–0.927**

1.212**

0.393

–0.419***

0.651***

–0.010***

0.444**

0.077

0.167

0.21

–1.120***

–0.934***

–0.453***

–0.290**

–0.130***

–0.094

–0.276***

(5) Unbanked

310

0.023 19,949

 Ln(1+population)

Observations –0.030***

19,949

–0.722

Notes:  Statistical significance is denoted at * 10 percent, ** 5 percent, and *** 1 percent levels

Financial literacy

0.038***

–0.052**

–1.961***

  Less than high school education (%)

B: Average marginal effects

2.189***

0.518

  Unemployment rate (%)

–0.219

0.299

  Poverty rate (%)

(2) Underbanked

(1) Banked without AFSs

A: Dependent variable

Table 16.2  (Continued)

–0.004***

19,949

0.185***

–0.664

–6.811

–2.09

(3) Unbanked with AFSs

–0.005***

19,949

0.051

0.023

8.874**

–1.841

(4) Unbanked without AFSs

–0.010***

20,450

0.132***

0.168

0.217

–1.688

(5) Unbanked

Financial literacy and the use of financial services by US households  311

Table 16.3  AFS use: logit and ordered logit regression results, 2018

A: Dependent variables

(1) AFS dummy

(2) Number of AFS types used

Financial literacy

–0.219***

–0.267***

–0.346***

–0.535***

Personal and household control variables Female Dependent children

–0.195***

–0.206***

Married

–0.167***

–0.162***

Some college

–0.100

–0.120*

College graduated

–0.396***

–0.419***

Postgraduate

–0.526***

–0.536***

Black non-Hispanic

0.816***

0.862***

Hispanic

0.221***

0.139**

Asian non-Hispanic

–0.074

–0.078

Other races

0.169

0.132

Age

–0.033***

–0.036***

Unemployed

–0.263**

–0.354***

Income level

–0.103***

–0.060***

County control variables Bank_density

–0.147

–0.153

AFS_density

0.700***

0.646***

Ln(1+median household income)

–0.417**

–0.409**

Poverty rate

–0.169

–0.029

Unemployment rate

–1.525

–1.245

Less than high school education

2.019***

2.111***

Ln(1+population)

–0.023

–0.025

Observations

20,364

20,364

Model

Logit

Ordered Logit

B: Average marginal effects Financial literacy

–0.035***

Financial literacy: No AFSs

0.042***

Financial literacy: 1 AFS type

–0.013***

Financial literacy: 2 AFS types

–0.008***

Financial literacy: 3 AFS types

–0.005***

Financial literacy: 4 AFS types

–0.003***

Financial literacy: 5 AFS types

–0.012***

Notes:  All variables are defined in Table 16.3. Statistical significance is denoted at * 10 percent, ** 5 percent, and *** 1 percent levels

312  Handbook of microfinance, financial inclusion and development

those for financial literacy in terms of both sign and significance. Lastly, the likelihood of a household being banked without AFSs is significantly and negatively related to AFS_density, which is the number of AFS establishments per 1,000 population at the county level. At the same time, AFS_density is positively and significantly associated with the other four dependent variables in Table 16.2. Panel B in Table 16.2 shows the average marginal effects of financial literacy on the probability of a household being banked without AFSs, underbanked, unbanked with or without AFSs, or simply unbanked. An increase in financial literacy by one unit2 increases the probability of a household being banked without AFSs on average by 3.8 percent. This translates into 4.6 million more households banked without AFSs. In contrast, an increase in financial literacy by one unit decreases the probability of a household being underbanked by 3.0 percent (or 3.6 million households), unbanked with AFSs by 0.4 percent (or 0.5 million households), unbanked without AFSs by 0.5 percent (or 0.6 million households), and simply unbanked by 1 percent (or 1.2 million households). Table 16.3 focuses on the relationship between financial literacy and the use of AFSs. Column (1) in the table shows the relationship between financial literacy and whether any AFSs are used, while column (2) shows the relationship between financial literacy and the number of AFS types used. The results indicate that increases in financial literacy negatively and significantly decrease the likelihood of using AFSs as well as the likelihood of using a greater number of AFS types. Results for the various control variables are similar to those reported in Table 16.2. In particular, Black non-Hispanics and Hispanics are associated with a significantly higher likelihood of using AFSs as well as a greater number of AFS types. Panel B in Table 16.3 shows the average marginal effects of financial literacy on the probability of a household using alternative financial services or using multiple types of these services. An increase in financial literacy by one unit decreases the probability of a household using AFSs by 3.5 percent (or 4.2 million households). The marginal effects of financial literacy on a number of types of AFSs are neither linear nor monotonic. The strongest marginal effects are associated with the probability that a household does not use any AFSs or uses all five types of AFSs. Specifically, if financial literacy increases by one unit, the probability of a household not using any AFSs increases by 4.2 percent (or 5.0 million households). At the same time, a one-unit increase in financial literacy decreases the probability that a household uses all five types of AFSs by 1.2 percent (or 1.4 million households).

CONCLUSIONS There is widespread research supporting the view that financial literacy matters for important financial and economic outcomes. In particular, the overwhelming evidence from this research indicates that more financially literate individuals and households make better saving and investment decisions, thereby improving their financial well-being. This research also indicates that an increase in financial literacy among more individuals and households is associated with significant increases in financial inclusion or better use of traditional financial markets. All these findings are clearly important because they document that with more individuals that are financially literate, there are clear social benefits as it becomes less expensive and easier for such individuals to manage their financial lives. Our contribution to the existing literature on financial literacy has focused on the relationship between the financial literacy of individuals and the mix of their use of traditional

Financial literacy and the use of financial services by US households  313

banking services and alternative financial services. We find that financial literacy, indeed, matters for which types of financial services are used by individuals and households in the financial marketplace. Specifically, our results indicate that greater financial literacy is associated with using more traditional financial firms and services than alternative financial firms and services. This is a key finding because improved financial literacy among a greater number of individuals and households could contribute to better financial inclusion. In other words, it may be less costly and less difficult for more individuals to obtain an appropriate mix of financial services. Of course, our findings are not the final word on the importance of financial literacy. Still, more research is needed on alternative measures of financial literacy, the effect of financial literacy on far more financial and economic outcomes, and the most effective ways to increase the financial literacy of more and different groups of individuals and households and in countries around the world.3

NOTES 1.

We use five questions in our empirical work because these questions appear in all four surveys. The survey in 2018 adds one additional question relating to compound interest. As will be seen, the use of five questions easily enables us to compare the measure of financial literacy in different years. 2. The measure of financial literacy is based on five questions. Therefore, a one-unit increase refers to one more question answered correctly. 3. As regards the relationship between financial education and financial literacy, see Fernandes, Lynch, and Netemeyer (2014), Xiao and O’Neill (2016), Wagner and Walstad (2019), Agarwal et al. (2020), and Lusardi, Michaud, and Mitchell (2019).

REFERENCES Agarwal, S., Amromin, G., Ben-David, I., Chomsisengphet, S., & Evanoff, D. D. (2020). Financial education versus costly counseling: How to dissuade borrowers from choosing risky mortgages? American Economic Journal: Economic Policy, 12(1), 1–32. Allen, F., Demirgüç-Kunt, A., Klapper, L., & Peria, M. S. M. (2012). The foundations of financial inclusion: Understanding ownership and use of formal accounts. Journal of Financial Intermediation, 27, 1–30. Allgood, S., & Walstad, W. B. (2016). The effects of perceived and actual financial literacy on financial behaviors. Economic Inquiry, 54(1), 675–697. Almenberg, J., & Dreber, A. (2015). Gender, stock market participation and financial literacy. Economics Letters, 137, 140–142. Artavanis, N., & Karra, S. (2020). Financial literacy and student debt. The European Journal of Finance, 26(4–5), 382–401. Bannier, C. E., & Neubert, M. (2016). Gender differences in financial risk taking: The role of financial literacy and risk tolerance. Economics Letters, 145, 130–135. Bianchi, M. (2018). Financial literacy and portfolio dynamics. The Journal of Finance, 73(2), 831–859. Célerier, C., & Matray, A. (2019). Bank-branch supply, financial inclusion, and wealth accumulation. The Review of Financial Studies, 32(12), 4767–4809. Dunham, I. M. (2019). Landscapes of financial exclusion: Alternative financial service providers and the dual financial service delivery system. Business and Society Review, 124(3), 365–383. Fernandes, D., Lynch Jr, J. G., & Netemeyer, R. G. (2014). Financial literacy, financial education, and downstream financial behaviors. Management Science, 60(8), 1861–1883. FINRA Investor Education Foundation (2019). National Financial Capability Study, Data File Information: 2009, 2012, 2015, and 2018 State-by-State Surveys.

314  Handbook of microfinance, financial inclusion and development

Klapper, L., & Lusardi, A. (2019). Financial literacy and financial resilience: Evidence from around the world. Financial Management, 49(3), 589–614. Lusardi, A., Michaud, P. C., & Mitchell, O. S. (2017). Optimal financial knowledge and wealth inequality. Journal of Political Economy, 125(2), 431–477. Lusardi, A., Michaud, P. C., & Mitchell, O. S. (2019). Assessing the impact of financial education programs: A quantitative model. Economics of Education Review, 78(C). Lusardi, A., & Mitchell, O. S. (2014). The economic importance of financial literacy: Theory and evidence. Journal of Economic Literature, 52(1), 5–44. Lusardi, A., & Mitchell, O. S. (2017). How ordinary consumers make complex economic decisions: Financial literacy and retirement readiness. Quarterly Journal of Finance, 7(3), 1750008. Potrich, A. C. G., Vieira, K. M., & Kirch, G. (2018). How well do women do when it comes to financial literacy? Proposition of an indicator and analysis of gender differences. Journal of Behavioral and Experimental Finance, 17, 28–41. Von Gaudecker, H. M. (2015). How does household portfolio diversification vary with financial literacy and financial advice? The Journal of Finance, 70(2), 489–507. Wagner, J., & Walstad, W. B. (2019). The effects of financial education on short‑term and long‑term financial behaviors. Journal of Consumer Affairs, 53(1), 234–259. Xiao, J. J., & O’Neill, B. (2016). Consumer financial education and financial capability. International Journal of Consumer Studies, 40(6), 712–721.

Financial literacy and the use of financial services by US households  315

APPENDIX 16.1: LIST OF FINANCIAL LITERACY QUESTIONS 1. Interest rate. Suppose you had $100 in a savings account and the interest rate was 2 percent per year. After five years, how much do you think you would have in the account if you left the money to grow? a. More than $102 b. Exactly $102 c. Less than $102 d. Don’t know e. Prefer not to say 2. Inflation. Imagine that the interest rate on your savings account was 1 percent per year and inflation was 2 percent per year. After one year, how much would you be able to buy with the money in this account? a. More than today b. Exactly the same c. Less than today d. Don’t know e. Prefer not to say 3. Bond price. If interest rates rise, what will typically happen to bond prices? a. They will rise b. They will fall c. They will stay the same d. There is no relationship between bond prices and the interest e. Don’t know f. Prefer not to say 4. Mortgage. A 15-year mortgage typically requires higher monthly payments than a 30-year mortgage, but the total interest paid over the life of the loan will be less. a. True b. False c. Don’t know d. Prefer not to say 5. Risk. Buying a single company’s stock usually provides a safer return than a stock mutual fund. a. True b. False c. Don’t know d. Prefer not to say

17. Financial inclusion, microfinance, and financial education in Latin America* Alejandro Javier Micco Aguayo and Patricio Andrés Valenzuela Aros

INTRODUCTION The promotion of financial inclusion and microfinance—that is, the use of formal financial services at affordable costs by all segments of the population and by small firms—has become a subject of growing interest among policymakers, academics, and other financial sector stakeholders. This interest has its roots in empirical evidence that emphasizes financial inclusion as an effective tool to improve households’ well-being while supporting economic development and growth. Recent academic papers have demonstrated that having a bank account increases savings, productive investment, and private expenditures by small firms and aspiring entrepreneurs (Dupas and Robinson, 2013), as well as female empowerment (Ashraf et al., 2010). Additionally, more inclusive financial markets mitigate the negative effects of wealth and income inequality on economic growth, because financial inclusion allows higher investment by poor agents (Braun et al., 2019). Given the importance of financial inclusion, it is now on the public agenda of most emerging economies. However, the objectives of financial inclusion policies vary across countries with different degrees of economic development. In most Latin American countries, a significant share of the population is still unbanked. Therefore, in these countries, both informal and alternative sources of capital still play a crucial role. As the financial sector develops, it is expected that formal financial institutions should become increasingly important. In more advanced economies, on the other hand, efforts often focus on two fronts: first, to promote informed access to a large set of better-quality and lower-priced financial products; and second, to target the low-income and still excluded specific groups of the population. For example, Chile—a Latin American country in which immigration grew at a faster rate than anywhere else in the region between 2010 and 2015—should consider a policy to promote financial inclusion among immigrants without access to formal financial services. Although financial development in Latin America has exhibited significant progress over the past decade, several countries in the region still face severe financial-inclusion gaps not only in the advanced economies but also in other peer-developing ones. A key issue is the lack of access the disadvantaged have to finance, which would promote economic growth at the broadest level. According to the World Bank’s Global Findex database, in the median Latin American country, more than half of the adult population lacked a bank account in 2017. The fact that an important share of the financially included population lacks financial education is even more worrisome. Although Latin America has little information about financial literacy, the few data available show the region has serious problems. For example, the OECD/INFE International Survey of Adult Financial Literacy Competencies, which covers 14 developed 316

Financial inclusion, microfinance, and financial education in Latin America  317

and developing countries, shows that 60 percent of Brazilians cannot exactly quantify their debt level, and only 18 percent are able to understand simple and compounding interest rates. In light of the ample empirical evidence of a positive linkage between access to finance and economic development (Burgess and Pande, 2005; Levine, 2005), an understanding of the level and evolution of financial inclusion—as well as its benefits—is crucial. This chapter takes stock of the current state of financial inclusion and microfinance across Latin American countries and discusses recent developments including innovations that could help the region leapfrog more traditional models. We use a comprehensive set of different databases to document different dimensions of the development of financial inclusion and microfinance in Latin America, highlighting variation within the region and changes over time. We compare the levels of financial inclusion in Latin American countries with those of comparable middleincome countries outside the region. We also explore how concentration is correlated with financial access around the world and in Latin America. When talking about financial inclusion in Latin America, one has to consider the heterogeneity within the region and within the countries. On the one hand, countries such as Chile have relatively developed banking systems and capital markets. On the other hand, smaller and poorer countries, such as Nicaragua, have shallow banking systems offering only the most rudimentary financial services. These heterogeneities make financial inclusion in Latin America challenging and increase the need for innovative solutions. Technology has reduced transaction costs and risks, thus enabling the processing of smaller transactions, and turning more households and small enterprises into commercially viable clients. Innovative products and delivery channels can address the constraints discussed above. Critically, these interventions and policy reforms have to work on both the supply and demand sides. In this chapter, we discuss several examples of such innovative approaches to financial inclusion. Specifically, we discuss the effects of specific innovations designed to reach out to previously unbanked population segments, such as branch and bank account expansion programs, banking agents, mobile banking, and new financial products (the so-called fintech industry). Finally, we highlight that financial education represents an important challenge to exploit the net benefits of financial inclusion and microfinance. Finally, we compare the level of financial education. Studies recognize financial education, which we relate also to financial consumer protection, and financial inclusion as two essential ingredients for the financial empowerment of individuals and the overall stability of the financial system. These two ingredients interact: financial education for financial inclusion (Atkinson and Messy, 2013). Financial inclusion requires financial education to reduce the possible cost to consumers from the use of financial instruments (e.g., short-term credit). Financial education and access to simple information not only help financial inclusion but may also help competition in this market. Empirical studies explore the impact of bank competition on firms’ access to credit. Beck et al. (2004) show that in countries with lower competition, measured by concentration, firms of all sizes face higher financing obstacles. The remainder of this chapter is structured as follows. The second section documents the degree of household and enterprise access to finance in Latin America by illustrating variation within the region and over time. It also describes the relationship between financial inclusion and market structure in Latin America. The third section discusses recent policies and interventions that have helped make Latin America’s financial markets more inclusive. The fourth section focuses on challenges in terms of financial education. The fifth section concludes and looks ahead.

318  Handbook of microfinance, financial inclusion and development

FINANCIAL INCLUSION AND MICROFINANCE IN LATIN AMERICA Household Access to Finance Table 17.1 reports account penetration across the three waves of the World Bank’s Global Findex dataset (2011, 2014, and 2017). The table reports a significant increase in the share of households with a bank account in the median country in Latin America, increasing from 28 percent in 2011 to 43 percent in 2014 to 47 percent in 2017. As reported in the table, all countries in Latin America exhibited significant increases in account penetration during the analyzed period. On the one hand, countries such as Chile and Uruguay have led the increase in Latin America, with increases above 30 percentage points in the 2011–2017 period. On the other hand, Argentina, Colombia, Nicaragua, Brazil, Ecuador, Paraguay, and Mexico have experienced increases below 15 percentage points. In view of the uneven progress in financial inclusion across Latin American countries, account

Table 17.1  Account penetration over time across countries Account penetration

2011

2014

2017

2017–2011

Argentina

33.1

50.2

47.9

14.8

Bolivia

28.0

40.7

51.2

23.2

Brazil

55.9

68.1

70.0

14.2

Chile

42.2

63.2

73.8

31.7

Colombia

30.4

38.4

44.9

14.5

Costa Rica

50.4

64.6

67.8

17.5

Dominican Republic

38.2

54.0

54.8

16.6

Ecuador

36.7

46.2

50.9

14.1

El Salvador

13.8

34.6

29.3

15.6

Guatemala

22.3

40.8

43.5

21.2

Honduras

20.5

30.0

42.9

22.4

Mexico

27.4

38.7

35.4

8.0

Nicaragua

14.2

18.9

28.4

14.2

Panama

24.9

43.4

45.8

20.9

Paraguay

21.7

— 

31.1

9.4

Peru

20.5

29.0

42.2

21.7

Uruguay

23.5

45.4

63.9

40.3

Venezuela

44.1

56.9

73.2

29.1

Minimum

13.8

18.9

28.4

8.0

Maximum

55.9

68.1

73.8

40.3

Median

27.7

43.4

46.9

17.0

Source:  Global Findex, World Bank

Financial inclusion, microfinance, and financial education in Latin America  319

penetration across Latin America varies widely, ranging from 28 percent in Nicaragua to 74 percent in Chile in the year 2017. Although this variation partly reflects the variation in income levels, it is certainly not the only factor. For example, the account penetration in Brazil is double that in Mexico despite the fact both countries have similar per-capita incomes. Table 17.2 reports information on credit penetration. In contrast to access to bank accounts, access to credit exhibited smaller progress between 2011 and 2017. The share of households with a loan from a formal financial institution in the median country in Latin America only increased from 10 percent to 13 percent during the 2011–2014 period. Then, this value dropped to 12 percent in 2017. While some countries in Latin America exhibited increases in credit penetration during the 2011–2017 period, others exhibited decreases. On the one hand, in Costa Rica, El Salvador, Honduras, Chile, Venezuela, and the Dominican Republic, the share of households with a loan from a financial institution increased by more than 4 percentage points. On the other hand, in Guatemala, Mexico, Panama, and Bolivia, the share

Table 17.2  Credit penetration over time across countries Loan penetration

2011

2014

2017

6.6

8.3

7.3

0.7

Bolivia

16.6

19.7

16.3

–0.3

Brazil

6.3

11.9

8.6

2.3

Chile

7.8

15.6

13.4

5.6

Argentina

2017–2011

Colombia

11.9

15.6

14.5

2.5

Costa Rica

10.0

12.7

14.1

4.1

Dominican Republic

13.9

18.2

22.7

8.8

Ecuador

10.6

13.4

11.8

1.2

El Salvador

3.9

17.2

8.5

4.6

Guatemala

13.7

12.3

9.6

–4.2

Honduras

7.1

9.7

12.4

5.3

Mexico

7.6

10.4

5.7

–1.8

Nicaragua

7.6

14.3

11.0

3.4

Panama

9.8

11.8

8.3

–1.5

Paraguay

12.9

— 

13.3

0.4

Peru

12.7

11.2

14.7

2.0

Uruguay

14.8

21.0

18.3

3.5

Venezuela

1.7

2.0

7.6

5.9

Minimum

1.7

2.0

5.7

4.0

Maximum

16.6

21.0

22.7

6.1

9.9

12.7

12.1

2.2

Median Source:  Global Findex, World Bank

320  Handbook of microfinance, financial inclusion and development

of households with a loan decreased during the analyzed period. As a consequence of these different evolutions, credit penetration across Latin America varies widely, from 6 percent in Mexico to 23 percent in the Dominican Republic in 2017. Again, although this variation may partly reflect the variation in income levels, it is certainly not the only factor. For instance, the Dominican Republic is the Latin American country with the highest credit penetration in the region, ahead of Chile, Mexico, and Uruguay, which have higher per-capita incomes. The Global Findex survey allows not only an aggregate picture of the share of households using formal financial services but also a more detailed look into which population segments have access to formal financial services. Table 17.3 reports account penetration across males versus females, the richest 60 percent versus the poorest 40 percent, and people in and outside of the labor force. By 2017, we still observe a significant gender gap. Roughly 51 percent of males in the median country have a formal financial account, compared with 44 percent of females. The gender gap is particularly large in El Salvador, Nicaragua, Peru, and Ecuador;

Table 17.3  Within-country gaps in account penetration in Latin American countries Account penetration

Male

Female

Richest 60%

Poorest 40%

In labor force

Out of labor force

Argentina

46.5

50.8

55.8

38.1

48.6

49.0

Bolivia

55.0

53.9

62.1

42.8

59.1

36.7

Brazil

72.8

67.5

79.0

56.6

74.1

61.4

Chile

77.8

71.3

79.3

66.9

83.9

58.5

Colombia

49.4

42.5

52.9

35.0

51.3

30.8

Costa Rica

75.5

60.9

74.3

58.0

71.5

61.2

Dominican Republic

58.4

54.1

65.6

42.1

62.7

37.9

Ecuador

60.2

42.6

63.1

33.4

58.5

33.1

El Salvador

37.6

24.4

37.7

19.3

37.2

17.7

Guatemala

46.4

42.1

53.1

30.4

50.1

30.7

Honduras

50.2

41.0

53.5

33.1

52.7

30.9

Mexico

41.1

33.3

44.0

25.8

42.9

26.0

Nicaragua

37.4

24.8

38.1

19.9

35.4

20.9

Panama

50.9

42.3

55.6

32.9

50.2

39.3

Paraguay

51.2

46.0

55.5

38.3

50.7

43.6

Peru

51.0

34.4

53.0

27.0

49.7

25.3

Uruguay

67.6

60.6

74.0

48.6

69.0

55.1

Venezuela

77.3

70.0

82.3

60.2

78.3

57.6

Minimum

37.4

24.4

37.7

19.3

35.4

17.7

Maximum

77.8

71.3

82.3

66.9

83.9

61.4

Median

51.1

44.3

55.7

36.5

52.0

37.3

Source:  Global Findex, World Bank

Financial inclusion, microfinance, and financial education in Latin America  321

it is relatively small in Chile, Brazil, the Dominican Republic, and Bolivia; and it is slightly negative in Argentina, with a higher proportion of females having a bank account than males. Note, however, that these comparisons are unconditional. The lower use of formal financial services by women may therefore be explained by gender gaps in other dimensions related to the use of financial services, such as their lower level of education and income, and by their employment status. The gap between the richest and the poorest is also significant. Whereas the share of the richest 60 percent of households having a bank account was 56 percent in 2017, it was only 37 percent for the poorest 40 percent. The gap was larger in Peru, El Salvador, Nicaragua, and Ecuador; while it was relatively small in Chile, Costa Rica, Brazil, and Venezuela. Finally, the table reports significant differences in the degree of financial inclusion between those individuals in the labor force and those individuals out of the labor force. Ecuador, Peru, and El Salvador exhibited the largest gaps. Argentina, Paraguay, Costa Rica, and Brazil exhibited smaller ones. Despite extensive financial development over the last few decades, several Latin American countries not only face within-country gaps but also gaps relative to other developing countries. Cross-country regressions to benchmark Latin America financial inclusion based on its correlates in other middle-income countries reveal a substantial gap between predicted and actual levels of Latin America financial inclusion.1 Figure 17.1 shows for half of the Latin American countries, predicted 2017 levels of the percentage of households having an account at a formal financial institution tended to be higher than their actual levels. Argentina, Colombia, Dominican Republic, Mexico, Panama, Peru, Paraguay, and Uruguay have levels of financial inclusion that are significantly below their predicted levels, whereas only Bolivia, Brazil, and Chile have levels of financial inclusion significantly higher than their predicted levels. Several obstacles prevent the reduction of financial-inclusion gaps within countries and between countries. The 2017 Global Findex data allow us to explore the reasons why some households within a country do not have accounts with formal financial institutions. In the median Latin American country, 59 percent of respondents without an account with a formal financial institution cite lack of money as a reason, 51 percent cite high costs, and 27 percent cite lack of trust in financial institutions; 25 percent of respondents point to someone else in the family having an account, whereas 24 percent report lack of the necessary documentation as an impediment. Finally, 23 percent point to geographic barriers and 5 percent report religious reasons for not having a formal bank account. Figure 17.2 displays the six most important reasons households do not have accounts with formal financial institutions by country. A number of macroeconomic factors can explain the financial-inclusion gaps between countries. Rojas-Suárez (2016) lists a few country-level obstacles that prevent Latin American countries from closing their financial-inclusion gaps with other countries with a similar degree of development. These obstacles are associated with low social development, high income inequality, macroeconomic instability, institutional weaknesses, and inefficiencies and inadequacies in the financial sector. Enterprise Access to Finance Although financial inclusion is an effective tool to improve households’ well-being, especially those excluded from the formal financial system, access to formal finance has broader positive implications. Financial inclusion is crucial for the creation and growth of small and

322  Handbook of microfinance, financial inclusion and development

Note:  Predicted values of the percentage of people over 15 years old that have an account from a financial institution in the past year, coming from OLS regressions that control for a set of country-level variables including total population, population density, natural resources rents/GDP, GDP per capita, economic growth, and manufacturing value added/GDP Source:  Authors’ calculations

Figure 17.1 Account from a financial institution in Latin American countries 2017 actual versus predicted values medium-sized enterprises (SMEs). Empirical evidence suggests firms’ access to financing can translate into formalization, productivity gains, higher sales, and job creation (see e.g. Heil, 2018). The financing sources of SMEs pose a significant challenge in Latin America and other emerging regions, which are different for enterprises with different profiles and financing needs. A large share of micro and small enterprises in Latin America consists of informal microenterprises whose establishment often stems from the lack of economic opportunities. Without access to formal financial products or formal guarantees, seeing this segment of the enterprise population becoming bankable over the medium to long term is hard, at least for credit services. They seem a natural target group for microcredit institutions and rely more heavily than other enterprises on alternative sources of finance such as informal finance providers or trade credit (Allen et al., 2014). A second segment of enterprises comprises small formal enterprises, some of which might have high growth potential. These firms are usually too big to be serviced by microfinance institutions, but not formal or established enough to get formal credit from traditional commercial banks. Finally, a group of medium-sized firms exists, often well-established and export-oriented companies. In most cases, they have access to bank finance but struggle to get access to market (equity and bond) financing. The World Bank’s Enterprise Surveys allow us to dig deeper into enterprise access to finance across firms of different sizes: small firms are firms with 5–19 employees, medium firms

Financial inclusion, microfinance, and financial education in Latin America  323

Source:  Global Findex, World Bank

Figure 17.2 Reasons households do not have accounts with formal institutions

are firms with 20–99 employees, and large firms are firms with more than 100 employees. Table 17.4 reports a large difference in access to bank accounts and the use of formal loans between small, medium, and large firms in Latin America. The table reports that across most Latin American countries, larger firms are more likely to have an account and/or a loan from

324  Handbook of microfinance, financial inclusion and development

Table 17.4  Firms with access to banking Country

Year

Percent of firms with a checking or savings account Small Medium Large (5–19) (20–99) (100+)

Percent of firms with a bank loan/line of credit Small Medium Large (5–19) (20–99) (100+)

Argentina

2017

99.1

99

97

35.1

52.9

64.4

Bolivia

2017

78.4

92.8

97.2

41.5

59.8

86

Brazil

2009

98.9

99

99.9

57.5

59.1

80

Chile

2010

95.2

99.8

97.8

72.5

83.5

81.7

Colombia

2017

99.1

98.1

100

50.3

80.7

79.5

Costa Rica

2010

95.8

99.3

99.7

41.2

71.9

79.3

Dominican Republic

2016

77.7

88.7

79

48.8

60.2

39.1

Ecuador

2017

97.8

93.1

99.8

54.1

61.9

87.4

El Salvador

2016

79.2

93.5

97.2

29.8

60.9

79.2

Guatemala

2017

79.1

87.8

93.9

34.1

53.3

84.1

Honduras

2016

75.3

97

95.5

35.1

64.6

75

Mexico

2010

58.6

66.9

69.6

26.8

36.7

54

Nicaragua

2016

77.7

94.4

97.9

38.7

65.6

63.2

Panama

2010

74.5

56.7

77.9

18.7

26.6

10.1

Paraguay

2017

86.3

97.3

99.4

49.6

68.2

83.6

Peru

2017

94.3

97.3

99.7

70.1

82

93.8

Uruguay

2017

95

99.7

100

46.2

61.4

87.5

Venezuela, RB

2010

95.6

99.7

97.7

25.6

58.1

85.4

Minimum

 

58.6

56.7

69.6

18.7

26.6

10.1

Maximum

 

99.1

99.8

100.0

72.5

83.5

93.8

Median

 

90.3

97.2

97.8

41.4

61.2

79.8

Source:  Enterprise Surveys, World Bank

a formal financial institution. Small firms’ enterprise access to credit in Latin America is more worrisome than access to bank accounts. The table reports that in the median Latin American country, a high proportion of firms have a checking or savings account in a formal financial institution. However, significant differences exist between small firms (90 percent), medium firms (97 percent), and large firms (98 percent). The variation in small enterprises’ access to accounts across Latin American countries is more pronounced. It ranges from 59 percent in Mexico to 99 percent in Colombia. Regarding access to credit, more than half of the small firms in the median country do not have access to credit from any financial institution. As reported in Table 17.4, only 41 percent of the small firms from the median country have bank credit, as compared to 80 percent of the large firms. The table also reports a wide variation in enterprises’ access to credit in Latin America by small firms, which ranges from 19 percent in Panama to 73 percent in Chile.

Financial inclusion, microfinance, and financial education in Latin America  325

Financial Inclusion and Market Structure The market power hypothesis, on the one hand, argues that in a competitive financial market, banks expand their outreach, raise efficiency, become more client-driven, and enhance the supply of financial services (Boot and Thakor, 2000). On the other hand, the information hypothesis highlights that bank competition may not favor financial inclusion. Information externalities imply strong competition may reduce banks’ incentives to invest in the screening process of loan applicants (Petersen and Rajan, 1995; Hauswald and Marquez, 2006). Increasing empirical evidence suggests higher competition is related to higher levels of financial inclusion. Using a survey of 74 countries, Beck et al. (2004) find bank concentration increases financing obstacles and decreases the probability of receiving bank finance. This negative effect is especially strong for small and medium firms. Carbo-Valverde et al. (2009) show that bank competition relieves financing constraints among Spanish firms. Chauvet and Jacolin (2017) show the positive impact of financial inclusion—measured by the share of firms at the industry level that has access to bank credit—on firm growth is strengthened under a less concentrated bank market. Figure 17.3 (top panel) presents the correlation between bank branches per 1,000 adults and net interest margin—a proxy for competition in the banking industry—across 127 countries in 2010. Figure 17.3 (bottom panel) presents the correlation between bank accounts per 1,000 adults and net interest margin across 68 countries for the same year. In both cases, we can see a negative correlation between low competition and bank accounts and branches. Latin American countries, marked in both figures, present a pattern similar to the one presented by the sample of all countries in the sample. Using municipal-level data, Marín and Schwabe (2019) explore the relationship between bank competition and the penetration of bank accounts in Mexico. The effect is economically meaningful. The authors estimate that moving from a monopoly to a duopoly leads to an increase of one account per ten adults, a 42 percent increase over the cross-municipality mean. The authors claim this increase is comparable to the effect of large increases in per-capita income and years of schooling, or the establishment of an additional branch by a bank that is already present in the local market.

OVERCOMING BARRIERS TO FINANCIAL INCLUSION Latin America has made substantial progress in terms of financial development and depth over the last decade. However, although financial depth has thus shown a certain degree of improvement, several barriers to financial inclusion persist as highlighted in the previous sections. Technological innovations, new delivery channels, and new players and products have helped to overcome these barriers, especially the geographic ones. Latin America has seen substantial innovation on this front over the past decade. Much of it comes from formal commercial banks. In many countries, regulators have reacted flexibly, opening space for innovation within existing regulatory frameworks or adjusting them where necessary. In this section, we report different forms of financial innovations, summarizing research findings on the recent expansions by some banks into traditionally underserved segments of the population as well as the banks’ use of agent networks to further extend their reach. Agents are typically owners of small retail businesses who are trained by a formal financial institution to collect deposits and process payments, including payments on small-scale loans (Lyman et al., 2006; Flaming et al., 2011). Finally, we explore the role of technological

326  Handbook of microfinance, financial inclusion and development

Source:  Financial Development and Structure Dataset and Global Financial Development. Authors’ calculations

Figure 17.3 Competition and financial inclusion sample of 127 countries in 2010 innovation in bridging some of the geographic divides that currently characterize the Latin American financial landscape. Financial Inclusion through Bank Accounts An interesting case of access to banking through the expansion of bank accounts is the case of Banco Estado in Chile. Banco Estado is a government commercial bank that has devised a

Financial inclusion, microfinance, and financial education in Latin America  327

banking-service strategy targeting low-income clients and underserved geographic areas. The record of Banco Estado in improving financial inclusion has been impressive. Thanks to the Chilean ID system, Banco Estado has been able to offer bank accounts (the so-called “Cuenta RUT”) to every citizen born in the country and to immigrants with a taxpayer number (RUT). This taxpayer number is also the number of the bank account. These accounts are being used as a tool to promote financial inclusion, often serving as an easy channel to acquire other financial services. They are also facilitating transfers to and from the government or other banks as well as retail payments or transactions at ATMs. As Figure 17.4 shows, by the end of 2018, more than 11 million Chileans had a Cuenta RUT account, up from 0.6 million with an active Cuenta RUT at the end of 2007. Cuenta RUT accounts represent a significant share of all accounts in Chile. In fact, 57 percent of total debit cards in Chile corresponded to Cuenta RUT accounts in 2018. Cuenta RUT accounts have also played a crucial role in promoting financial inclusion among immigrants. In fact, the number of Cuenta RUTs granted to immigrants increased 60 percent from 444,878 accounts in 2017 to 713,847 accounts in 2018. This increase is important for a country like Chile, where immigration has grown at a faster rate than anywhere else in the region. Branch Expansion The branch-expansion strategy of Banco Azteca in Mexico represents a good example of a Latin American private domestic bank that has recently been successful in reaching segments of the population that were previously considered “unbankable” by traditional commercial banks. The branch expansion of Banco Azteca was striking. The number of branches of the bank grew from 815 in 2002 to 1,751 in 2017, becoming the second largest bank in Mexico in terms of branches.

Source: BancoEstado

Figure 17.4 Cuenta RUT (millions of accounts)

328  Handbook of microfinance, financial inclusion and development

In 2002, one of the largest retailers of electronics and household goods in Mexico (Grupo Elektra) received a banking license and opened branches of the new Banco Azteca in all its retail stores. From the beginning, Banco Azteca targeted low- and middle-income households. To achieve this goal, Banco Azteca required softer eligibility requirements than traditional commercial banks, often taking collateral and co-signers instead of valid documents. Moreover, the bank benefited from synergies with its retail operations as it provided information about the financial needs of approximately four million financially underserved customers. Within the first month of operation of Banco Azteca in 2002, 157,000 accounts were opened, increasing to 250,000 accounts by the end of the same year. Banco Azteca also took over the issuance of installment loans, which were previously issued by Elektrafin, the financing unit of Grupo Elektra’s retail stores. The average amount of these loans was US$250. Although these loans were tied to merchandise, they could be used for business purposes. At the end of 2003, Banco Azteca also expanded into the mortgage and insurance business. Although Grupo Elektra was disbursing installment loans before opening Banco Azteca, Banco Azteca’s new savings accounts allowed its loan portfolio to expand significantly. The loan portfolio increased from around two billion Mexican pesos at the time of the bank’s opening to ten billion Mexican pesos by the end of 2004. Although this portfolio size is small compared to the total commercial bank credit to the private sector, it is large compared to the credit disbursed by smaller institutions that cater to low-income households. Some similarities as well as some disparities exist between Banco Azteca and most microfinance institutions (MFIs). On the one hand, Banco Azteca is similar to MFIs in terms of the characteristics of their customers and the size of their loans. On the other hand, Banco Azteca accepts deposits that reduce its cost of funds and enhances its growth opportunities, it is regulated and supervised as any other commercial bank, and it is uniquely positioned to take advantage of economies of scale due to the synergies with its retailer parent company. These differences have allowed Banco Azteca to reduce operating costs despite the small loan size. Bruhn and Love (2014) exploit the entry of Mexico’s Banco Azteca in Mexico to examine the impact of access to finance on the incomes of the poor. The study finds a statistically significant and economically meaningful effect of access to finance on labor market activity and income levels, especially among low-income individuals and those individuals located in areas with lower preexisting bank penetration. Thus, Banco Azteca’s business model proves to increase financial inclusion and households’ well-being, while reducing transaction costs, acquiring effective information, and enforcing loan repayment. Banking Agents Another innovation for promoting financial inclusion in sparsely populated areas is banking agents. Banking agents have allowed financial institutions to leapfrog not only geographic but also socio-cultural barriers that might prevent low-income population segments from entering formal bank offices. A review of the Latin American experience suggests agents have been effective in reaching the unbanked (Alliance for Financial Inclusion, 2012). The expansion strategy of Banco Estado in Chile has been accompanied by infrastructure through banking agents that promote the use of the bank’s accounts. Banco Estado’s Caja Vecina network in Chile has played a key role. Caja Vecina is the country’s largest network of retailers acting as correspondents. As displayed in Figure 17.5, Banco Estado has significantly expanded its number of agents from 300 in 2006 to more than 25,000 in 2018. The

Financial inclusion, microfinance, and financial education in Latin America  329

Source: BancoEstado

Figure 17.5

Caja Vecina

banking agent model implemented in Chile by Banco Estado has proved successful, due in part to its low costs. At the same time, it provides a boost to the activity of affiliated retailers. Additionally, Caja Vecina retailers are part of Banco Estado’s more than 600,000 SME clients. Fintech and Technological Innovations By revolutionizing electronic payment systems around the world, technology has played a crucial role in the promotion of financial inclusion in recent years, especially in less developed countries where access to finance is more difficult. In particular, mobile devices have become an efficient channel in transactional systems. Mobile devices have facilitated financial transactions, allowing users to deposit and withdraw cash from an account that is accessible by mobile handset. Using their mobile devices, users are also able to store value in the account and transfer value between users via text messages, menu commands, and personal identification numbers (Aker and Mbiti, 2010). Technology has enabled users to make payments and transfer funds at relatively low cost across much wider geographic areas than is possible using localized informal payment solutions. Aker et al. (2011) report that in the 2005–2010 period, money-transfer systems were established in 80 developing countries in Africa, Asia, and Latin America. Technology has also reduced banks’ transactional and operational costs, allowing them to pass these lower costs on to clients. By lowering the cost of supplying products, technology makes expanding access to financial products and services to the low-income and underserved segments of the population easier, while ensuring that financial inclusion is profitable for the financial institutions adopting them. Table 17.5 shows that technology has played a significant role in promoting financial inclusion in some Latin American countries, such as Chile, Paraguay, and Venezuela. The World

330  Handbook of microfinance, financial inclusion and development

Table 17.5  Mobile phone use in financial transactions in Latin America, 2017

 

Mobile money Used mobile phone or internet Used mobile phone or internet to account to access an account access a financial institution account (% age 15+) (% age 15+) (% age 15+)

Argentina

2

10

21

Bolivia

7

9

12

Brazil

5

13

18

Chile

19

28

34

Colombia

5

9

16

Costa Rica

 —

18

26

Dominican Republic

4

8

12

Ecuador

3

5

9

El Salvador 4

6

15

Guatemala

2

4

8

Honduras

6

8

10

Mexico

6

7

15

Nicaragua

4

4

6

Panama

4

6

12

Paraguay

29

28

10

Peru

3

5

10

Venezuela

11

30

40

Uruguay

— 

16

25

Source:  Global Findex, World Bank

Bank Global Findex dataset reveals that by 2017, between 2 percent (Argentina) and 29 percent (Paraguay) of the adult population in Latin American countries had a mobile money account, while between 4 percent (Guatemala) and 30 percent (Venezuela) of the adult population used a mobile phone or the internet to access an account within the past year. Additionally, mobile banking has taken off as a means of accessing accounts and performing financial transactions in several Latin American countries. For example, the percentage of the adult population that used mobile phones or the internet to access a financial institution account was 34 percent in Chile, 26 percent in Costa Rica, 40 percent in Venezuela, and 25 percent in Uruguay. Another manifestation of the introduction of fintech in Latin America has been the number of new fintech enterprises in the region. According to Finnovista and IADB (2018), 1,166 fintech companies are in Latin America today. Figure 17.6 shows the distribution of fintech enterprises by country in Latin America. Most of these companies are located in Brazil, Mexico, Colombia, Argentina, and Chile, accounting for 86 percent of the total enterprises in the region. The three most recurrent areas of these fintech startups are (1) payments and remittances, (2) lending, and (3) enterprise financial management.

Financial inclusion, microfinance, and financial education in Latin America  331

Note:  Others group including Costa Rica, Dominican Republic, Guatemala, Panama, El Salvador, Paraguay, Bolivia, Honduras, and Nicaragua that accounts for less than 3 percent of the total number of fintechs in the region Source:  Fintech Latin America 2018. BID and Finnovista

Figure 17.6 Fintech distribution by country in Latin America

THE ROLE OF FINANCIAL LITERACY To enhance the well-being of citizens by empowering them to create economic opportunities and protect them against negative shocks, jointly addressing financial inclusion and literacy— especially among populations that have greater difficulty understanding innovative financial products and services as well as their rights and responsibilities—is necessary. Generally, these segments are the youth, the poor, and the rural populations that have usually been excluded from formal financial services and products. The OECD International Network on Financial Education (OECD/INFE) defines financial literacy as a combination of knowledge, attitudes, and behaviors. Financial knowledge captures different aspects of knowledge that are widely considered useful to individuals when making financial decisions, that is, basic knowledge of financial concepts, and the ability to apply numeracy skills in a financial context. Financial attitudes capture perspectives on money and planning for the future. Finally, financial behaviors capture behaviors such as budgeting, thinking before making a purchase, paying bills on time, and saving and borrowing to make ends meet. Unregulated access to credit and credit to uninformed consumers can lead to consumer detriment, affecting the most vulnerable sectors of the population. In view of the costs associated with a lack of financial literacy, governments around the world have started to implement policies with the objective of expanding financial education. For the International Network on Financial Education, a network of public experts on financial education established by the OECD, literacy is a core life skill for participating in modern society. Therefore, they

332  Handbook of microfinance, financial inclusion and development

Table 17.6  The rationale for financial education Very much

A little

In the future

Financial miss-selling, fraud, or abuse

12

5

—

High complexity of financial services

11

6

—

FL needs of specific population subgroups

11

1

2

Personal/household over-indebtedness

10

4

—

Spread of DFS

8

6

1

Recent pension reform

7

2

2

Low financial inclusion

5

4

—

Growing entrepreneurship/low FL of entrepreneurs

4

5

—

Lack of trust in formal financial institutions

4

2

1

High migration/difficulties with remittances

3

1

3

Source: OECD/INFE Report on Financial Education in APEC Economies (2019)

recommend that financial education start as early as possible and be taught in schools. To improve financial knowledge, many countries have included financial education in school curricula. Table 17.6 reports the most important reasons for the implementation of financial education policies in a sample of 17 APEC (Asian Pacific Economic Cooperation) governments are as follows:2 (1) financial mis-selling, fraud, or abuse; (2) the high complexity of financial services; and (3) personal/household over-indebtedness. Although the academic literature has not been completely conclusive, the OCED claims (OECD, 2016): “there is a wide consensus across jurisdictions that low levels of financial literacy contribute to possible consumer detriment from the use of financial instruments, in particular, short-term credit.” This issue is an important one in Latin America. For example, 60 percent of Brazilians cannot exactly quantify their debt levels or indicate their creditors. In the case of Chile, Bernstein, and Ruiz (2005) present evidence that 97 percent of the contributors to the fully funded, individual-account pension system do not know the fee charged by their pension fund administrator. Spending on middle-income households has increased relative to income in emerging markets, increasing financial vulnerability, and over-indebtedness. As reported by OCDE (2019), this over-indebtedness has become a growing phenomenon in a number of countries and seems to be a problem that affects the middle class in Latin American countries with medium and high financial penetration. Several factors explain the high level of indebtedness. First, in the general economic context, decreases in aggregate economic activity increase unemployment and decrease household income, increasing the level of indebtedness. The second factor is low or unpredictable levels of disposable income. The third is life events such as accidents and illness without a proper and global insurance system. The fourth factor is a lack of financial education. Moore (2003) and Mottola (2013) report that households with lower financial education access more expensive mortgage loans and present a higher probability of using more expensive financing (e.g. credit cards). Although the supply-side interventions described earlier have indeed increased financial access among lower-income households, in several cases, the lack of financial literacy has

Financial inclusion, microfinance, and financial education in Latin America  333

hindered the optimal use of financial services, leaving households vulnerable to predatory services. Some examples in emerging economies are the microcredit over-indebtedness crises in Nicaragua and India, and a good example in an advanced economy is the subprime mortgage crisis in the United States (Mader, 2015; Bateman et al., 2018). Therefore, merely banking the unbanked population is not enough (Dupas et al., 2018). Financial inclusion needs to be complemented with financial education, which is a widely used tool for making sound decisions that foster financial health and protect households and small firms from predation. Unfortunately, recent evidence indicates three out of four newly banked global poor have never received any form of financial training (Deb and Kubzansky, 2012). The role of financial education in enhancing the benefits of financial services is crucial. Financial education has demonstrated the ability to increase knowledge and savings (Miller et  al., 2015; Kaiser and Menkhoff, 2017). However, a consistent impact has been difficult to find among lower-income participants in developing countries as well as among higherincome users in more developed countries (Kaiser and Menkhoff, 2017). Empirical evidence reveals that the intensity of the financial education programs relates to the impact (Miller et al., 2015; Kaiser and Menkhoff, 2017). Moreover, individuals with greater self-efficacy or locus of control appear to benefit more from financial education (Fernandes et al., 2014; Miller et al., 2015). This finding suggests financial education programs should also target psychological components. Recent studies have started to explore the impact of some financial education programs in Latin American countries. For example, Attanasio et al. (2019) conduct a randomized impact evaluation of a tablet-based financial education program on women in Colombia. The study finds significant positive impacts on financial knowledge, attitudes, practices, and performance, translating into increased financial health 25 months after the end of the intervention. Specifically, women who received the tablets demonstrated a better understanding of savings and budgeting concepts, expressed stronger preferences for saving formally rather than informally, reported more trust in banks, and professed more optimism in general. The participants were also more likely to set savings goals, felt more capable of teaching others how to use ATMs, and reported more informal savings. Unfortunately, the levels of financial literacy in Latin American countries are low according to the 2015 financial literacy module of the Programme for International Student Assessment (PISA) test. The continuous scale of the module of the PISA test is divided into five levels. Questions at Level 1 are considered the easiest. At best, students performing at Level 1 can recognize the difference between needs and wants, can make simple decisions on everyday spending, and can recognize the purpose of everyday financial documents, such as invoices. Level 2 is considered the baseline level of proficiency in financial literacy that is required to participate in society. On the other hand, Level 5 questions are considered the most challenging for 15-year-old students at the end of compulsory education. Students performing at Level 5 can look ahead to solve financial problems or make the kinds of financial decisions that will be relevant to them in the future. They can take into account features of financial documents that are significant but unstated or not immediately evident, such as transaction costs, and they can describe the potential outcomes of financial decisions, showing an understanding of the wider financial landscape, such as income tax. As reported in Table 17.7, the results of the 2015 PISA test indicate that across the ten participating OECD countries, 22 percent of students score below the baseline level of proficiency in financial literacy. In all Latin American participating countries, the percentage of

334  Handbook of microfinance, financial inclusion and development

Table 17.7  Performance in financial literacy PISA 2015 results Country China

% share Low performers

Top performers

Score  

9

33

566

Belgium

12

24

541

Canada

13

22

533

Russia

11

11

512

Netherlands

19

17

509

Australia

20

15

504

OECD average

22

12

489

United States

22

10

487

Poland

20

8

485

Italy

20

6

483

Spain

25

6

469

Lithuania

32

4

449

Slovak Republic

35

6

445

Chile

38

3

432

Peru

48

1

403

Brazil

53

3

393

Source:  PISA 2015 results, OECD

students who scored below the baseline level is worrisome: 38 percent in Chile, 53 percent in Brazil, and 48 percent in Peru. In Russia, a country with a more similar per-capita GDP to Latin American countries, only 11 percent of students perform at this level. The fact that more students in Latin America score at Level 1 than at any other of the five proficiency levels is an important concern. The poor performance of the Latin American countries is not a surprise, because financial education policies were practically inexistent until recently. For example, financial education in Chile—the country with the best performance in Latin America—was introduced in the curriculum for grades 7 and 8 in 2016 and for the first year of secondary (high school) education in 2017. In Mexico, each school can decide what content it will impart to its students. Only 20 percent of the schools chose the financial education course. Finally, Table 17.8 reports the average level of financial literacy for OECD countries, six developing countries, and Brazil and Chile. The table also reports a slightly different method for measuring the same three components of financial literacy for Bolivia, Peru, Chile, and Colombia. The average score for financial literacy in OECD countries is 13.7, with a standard deviation of 1. Brazil and Chile, on average, present a one-standard-deviation-lower score than the average OECD country (12.7). The main difference between OECD countries and Brazil and Chile is in financial knowledge. This finding is not surprising because this component captures the level of knowledge of financial mathematics (e.g. how to compute compound

Financial inclusion, microfinance, and financial education in Latin America  335

Table 17.8  Financial knowledge, attitudes, and behavior OECD measure Knowledge score

Behavior score

Attitude score

Total score

OECD

4.9

5.4

3.4

13.7

Thailand

3.9

5.8

3.1

12.8

Albania

4.2

5.2

3.4

12.8

Belarus

3.8

5.0

2.9

11.7

Russia

4.1

5.1

2.9

12.1

Turkey

4.6

4.8

3.1

12.5

Jordan

4.3

5.7

2.6

12.6

Average

4.2

5.3

3.0

12.4

Brazil

4.3

4.6

3.1

12.0

Chile

4.4

5.8

3.0

13.3

Average

4.4

5.2

3.1

12.7

Alternative measure Chile

5.1

5.8

3.0

14.0

Bolivia

4.8

5.4

3.6

13.8

Colombia

5.1

5.2

3.3

13.6

Ecuador

5.1

5.1

3.2

13.5

Peru

4.6

4.7

3.6

12.9

Source:  OECD (2016), SBIF and CAF (2016), and Mejía et a.l (2015)

interest on savings). Using an alternative measure, the table shows Chile presents the highest level of financial literacy in the Andean Region. Bolivia presents the second-best score and Peru the lowest score.

CONCLUSIONS The promotion of financial inclusion and microfinance has become a subject of growing interest among policymakers, academics, and other financial sector stakeholders. This interest has its roots in evidence that emphasizes financial inclusion as an effective tool to improve households’ well-being while supporting economic development. This chapter contributes to the exciting, policy-relevant research agenda going forward on these issues. As reported in this chapter, Latin America has made substantial progress in terms of financial development in recent years. However, important barriers to financial inclusion persist. As a consequence, most countries in the region exhibit important within-country gaps between men and women, the richest and the poorest, and the employed and unemployed. Also, several countries in the region still face severe financial-inclusion gaps to not only the advanced economies but also other peer developing ones.

336  Handbook of microfinance, financial inclusion and development

We explore different forms of financial innovations, bank branch expansions as well as the banks’ use of agent networks to further extend their reach. Success stories of financial inclusion in Latin America have demonstrated the importance of bank strategies targeting poorer households, small firms, and areas largely ignored by traditional commercial banks. Some examples are the cases of the Banco Estado in Chile and the Banco Azteca in Mexico. We also explore the role of technological innovation in bridging some of the geographic divides that currently characterize the Latin American financial landscape as technology has reduced banks’ transactional and operational costs, allowing them to pass these lower costs on to clients. Although the supply-side interventions in Latin America have indeed increased financial access among lower-income households, in several cases, the lack of financial literacy has hindered the optimal use of financial services, leaving households and small firms vulnerable to predatory services. Thus, merely banking the unbanked population is not enough. To enhance the well-being of citizens by empowering them to create economic opportunities and protect them against negative shocks, jointly addressing financial inclusion and literacy is necessary. In view of the low levels of financial literacy in Latin America, financial education needs to go hand in hand with the ongoing progress in financial inclusion. Only in this way the region will be able to develop a sound and secure financial system with responsible and informed consumers. Overall, this chapter highlights that new banking business models and financial innovations have helped broaden the share of the population with access to basic formal financial services, and technology has helped Latin America’s financial systems leapfrog traditional delivery channels. If the policy objective is to continuously increase access to financial products, one of the main lessons learned is that it is crucial to identify the principal barriers to their use from the very beginning as well as provide financial inclusion to the most vulnerable sectors of the population. To achieve this goal, close cooperation between researchers, regulators, and practitioners can help produce critical insights into what practices and policies work to improve financial inclusion and education in Latin America.

NOTES *

We particularly thank the editors Robert Cull and Valentina Hartarska for their extensive comments and suggestions. We also wish to thank the Chilean Association of Banks and Financial Institutions (ABIF) for its financial support. 1. Predicted values of the percentage of people over 15 years old who have an account from a financial institution in the past year, coming from OLS regressions that control for a set of country-level variables including total population, population density, natural resources rents/GDP, GDP per capita, economic growth, and manufacturing value added/GDP. 2. Although the ranking of the reasons for financial education of the Asian economies may not translate perfectly to the Latin American context, the benefits of financial education should be universal.

REFERENCES Aker, J. C., Boumnijel, R., McClelland, A., and Tierney, N. (2011). “Zap It to Me: The Short-Term Impacts of a Mobile Cash Transfer Program,” Center for Global Development Working Paper No. 268. Aker, J. C. and Mbiti, I. M. (2010). “Mobile Phones and Economic Development in Africa,” Journal of Economic Perspectives 24, 207–232.

Financial inclusion, microfinance, and financial education in Latin America  337

Allen, F., Carletti, E., Qian, J., and Valenzuela, P. (2014). “Does Finance Accelerate or Retard Growth? Theory and Evidence,” in F. Allen, J. R. Behrman, N. Birdsall, S. Fardoust, D. Rodnik, A. Steer, and A. Subramanian (Eds), Towards a Better Global Economy: Policy Implications for Citizens Worldwide in the 21st Century, 289–380, Oxford University Press, Oxford. Alliance for Financial Inclusion (2012). “Agent Banking in Latin America,” AFI Discussion Paper, March. Ashraf, N., Karlan, D., and Yin, W. (2010). “Female Empowerment: Impact of a Commitment Savings Product in the Philippines,” World Development 38, 333–344. Atkinson, A. and Messy, F. (2013). “Promoting Financial Inclusion through Financial Education: OECD/ INFE Evidence, Policies and Practice,” OECD Working Papers on Finance, Insurance and Private Pensions, No. 34, OECD Publishing, Paris. Attanasio, O., Bird, M., Cardona-Sosa, L., and Lavado, P. (2019). “Freeing Financial Education via Tablets: Experimental Evidence from Colombia,” NBER Working Paper 25929. Bateman, M., Blankenburg, S., and Kozul-Wright, R. (2018). The Rise and Fall of Global Microcredit, Routledge, New York. Beck, T., Demirgüç-Kunt, A., and Maksimovic, V. (2004). “Bank Competition, Financing and Access to Credit,” Journal of Money, Credit and Banking 36, 627–648. Bernstein, S. and Ruiz, J. L. (2005). “Sensibilidad de la Demanda Con Consumidores Desinformados: El Caso de las AFP en Chile,” Documento de Trabajo 4, Superintendencia de Administradora de Fondos de Pensiones. Boot, A. and Thakor A. (2000). “Can Relationship Banking Survive Competition?” The Journal of Finance 55, 679–713. Braun, M., Parro, F., and Valenzuela, P. (2019). “Does Finance Alter the Relation between Inequality and Growth?” Economic Inquiry 57, 410–428. Bruhn, M. and Love, I. (2014). “The Real Impact of Improved Access to Finance: Evidence from Mexico,” Journal of Finance 69, 1347–1376. Burgess, R. and Pande, R. (2005). “Can Rural Banks Reduce Poverty? Evidence from the Indian Social Banking Experiment,” American Economic Review 95, 780–795. Carbo-Valverde, S., Rodriguez-Fernandez F., and Udell, G. (2009). “Bank Market Power and SME Financing Constraints,” Review of Finance 13, 309–340. Chauvet, L. and Jacolin, L. (2017). “Financial Inclusion, Bank Concentration, and Firm Performance,” World Development 97, 1–13. Deb, A. and Kubzansky, M. (2012). Bridging the Gap: The Business Case for Financial Literacy, Citi Foundation, New York. Dupas, P. and Robinson, J. (2013). “Savings Constraints and Microenterprise Development: Evidence from a Field Experiment in Kenya,” American Economic Journal: Applied Economics 5, 163–192. Dupas, P., Karlan, D., Robinson, J., and Ubfal, D. (2018). “Banking the Unbanked? Evidence from Three Countries,” American Economic Journal: Applied Economics 10, 257–297. Fernandes, D., Lynch Jr, J., and Netemeyer, R. (2014). “Financial Literacy, Financial Education, and Downstream Financial Behaviors,” Management Science 60, 1861–1883. Flaming, M., McKay, C., and Pickens, M. (2011). “Agent Management Toolkit: Building a Viable Network of Branchless Banking Agents,” Consultative Group to Assist the Poor (CGAP) Technical Guide, CGAP, Washington, DC. Hauswald, R. and Marquez, R. (2006). “Competition and Strategic Information Acquisition in Credit Markets,” The Review of Financial Studies 19, 967–1000. Heil, M. (2018). “Finance and Productivity: A Literature Review,” Journal of Economic Surveys 32, 1355–1383. IDB and Finnovista. (2018). “Fintech Latin America. Growth and Consolidation,” https://publications​. iadb​.org​/ handle​/11319​/9234. Kaiser, T. and Menkhoff, L. (2017). “Does Financial Education Impact Financial Literacy and Financial Behavior, and if so, When?” Policy Research Working Paper 8161. Levine, R. (2005). “Finance and Growth: Theory and Evidence,” in P. Aghion and S. N. Durlauf (Eds), Handbook of Economic Growth, 865–934, Elsevier, Amsterdam. Lyman, T., Ivatury, G., and Staschen, S. (2006). “Use of Agents in Branchless for the Poor: Rewards, Risks, and Regulation,” Consultative Group to Assist the Poor (CGAP), Focus Note No.38, CGAP, Washington, DC.

338  Handbook of microfinance, financial inclusion and development

Mader, P. (2015). The Political Economy of Microfinance. The Financializaton of Poverty, Springer, New York. Marín, A. G. and Schwabe, R. (2019). “Bank Competition and Financial Inclusion: Evidence from Mexico,” Review of Industrial Organization 55, 257–285. Mejía, D., Pallotta A., and Egúsquiza E. (2015). “Encuesta de Medición de las Capacidades Financieras en los Países Andinos. Informe Comparativo 2014,” Documento de Trabajo Corporación Andina de Fomento 2015. Miller, M., Reichelstein, J., Salas, C., and Zia, B. (2015). “Can You Help Someone Become Financially Capable? A Meta-Analysis of the Literature,” The World Bank Research Observer 30, 220–246. Moore, D. L., (2003). “Survey of Financial Literacy in Washington State: Knowledge, Behaviour, Attitudes, and Experiences,” Washington State Department of Financial Institutions. Mottola, G. (2013). “In Our Best Interest: Women, Financial Literacy, and Credit Card Behaviour,” Numeracy 6, Article 4. OECD (2016). OECD/INFE International Survey of Adult Financial Literacy Competencies, OECD Publishing, Paris. OECD (2019). Under Pressure: The Squeezed Middle Class, OECD Publishing, Paris. Petersen, M. A. and Rajan, R. G. (1995). “The Effect of Credit Market Competition on Lending Relationships,” The Quarterly Journal of Economics 110, 407–443. Rojas-Suárez, L. (2016). “Financial Inclusion in Latin America: Facts, Obstacles and Central Banks’ Policy Issues,” Discussion Paper Nº IDB-DP-464. SBIF and CAF (2016). “Encuesta de medición de capacidades financieras en los países andinos: Chile 2016,” Documento de Trabajo. Corporación Andina de Fomento y Superintendencia de Bancos e Instituciones Financieras de Chile 2016.

18. Gender and financial inclusion in Latin America and the Caribbean Victor Motta

INTRODUCTION Financial inclusion, broadly defined as both access to and use of formal financial services (Demirgüç-Kunt et al., 2015), is an important determinant of economic development. At its most basic level, financial inclusion refers to owning an account at a formal financial institution, which is expected to be the first step toward the use of other financial services, such as credit and insurance (Khavul, 2010). Being financially included leads to economic benefits, as individuals with access to formal financial services are able to invest in education and entrepreneurial activities, contributing to poverty reduction and allowing them to increase their income (Bruhn & Love, 2014). Financial inclusion also allows individuals to save and borrow money formally, contract insurance, or use payment services. Therefore, it may foster financial stability by contributing to a more stable deposit base for banks in times of uncertainty (Han & Melecky, 2014). Not having a bank account can have detrimental effects on households as lack of access to an account not only can increase the probability of payment default and hamper liquidity levels, but cash transactions present a financial risk for unbanked individuals, including the risk of stolen funds (Rhine & Greene, 2006; Rhine et al., 2006). These negative effects are more pervasive among women since there exists gender disparities in both access to finance and the use of formal financial services and products. Focusing on gender is a critical issue since it is a significant determinant of broader macroeconomic issues, including economic development (Duflo, 2012). Financial inclusion is a particular concern for women in Latin America and the Caribbean (LAC) as financial systems are less inclusive and developed than those of more developed countries (Bebczuk, 2008). Worldwide, 47 percent of women own bank accounts compared with 55 percent of men (Ghosh & Vinod, 2017). Using data from the 2017 World Bank’s Global Findex database, the objective of this chapter was to examine the determinants of financial inclusion in the region, particularly the importance of gender. Based on recent investigations (Allen et al., 2016; Fungáčová & Weill, 2014; Zins & Weill, 2016), we assessed the impact of individual determinants, such as age, income, education, and employment status among others, on the three main indicators of financial inclusion: ownership of a bank account, saving in a bank account, and use of bank credit. Following the findings from Demirgüç-Kunt et al. (2013) of a gender gap in the use of financial services, we also examined gender differences in the use of financial services by focusing on the moderation effects of gender on the relationship between individual characteristics and financial inclusion. In addition, we examined the determinants of financial exclusion for unbanked women. The main contribution of this study relates to a more focused examination of the gender issue in access to finance in LAC. To the best of our knowledge, there has been little research 339

340  Handbook of microfinance, financial inclusion and development

elucidating gender issues in finance in the region. As a result, this is one of the earlier studies to systematically examine the issue of access to and use of finance among women-headed households in LAC. Existing studies in developed regions indicate a significant gender gap due to lower financial literacy among women, behavioral differences, and institutional discrimination (Beck & Brown, 2011; Browne, 2006; Fletschner, 2009; Lusardi & Tufano, 2015). The study proceeds as follows. The second section reviews the literature on financial inclusion, the third section describes the methodology used in the study, and the fourth section presents the results on the main determinants of financial inclusion. The fifth section discusses the findings and the sixth section concludes the study, providing limitations and suggestions for future research.

LITERATURE REVIEW Financial Inclusion and Development Currently, 1.7 billion people around the world are excluded from financial services in formal markets (Demirgüç-Kunt et al., 2018). Although excluded, this contingent of people, mostly low-income households, still save, borrow, and manage day-to-day expenses at informal markets. Family, friends, neighbors, and moneylenders are typical examples of a complex network that fills the void of the absence of formal markets (Banerjee et al., 2013). Financial inclusion mitigates the high costs of informality by allowing individuals to own an account at a financial institution, save and borrow money formally, and use payment services and contract insurance. Financial inclusion may mobilize savings and provide both households and firms with access to resources that are needed to finance consumption and investment as well as insure against economic shocks. It may lead to economic benefits in developing economies as individuals with access to formal financial services are able to invest in education and entrepreneurial activities, fostering labor and firm formalization and contributing to reductions in poverty and income inequality (Bruhn & Love, 2014). Recent studies have investigated financial inclusion in developing economies. Examining individual and country characteristics associated with ownership of a bank account, account use to save, and use of bank credit, three widely used indicators of financial inclusion, DemirgüçKunt and Klapper (2013) showed that differences in income among countries influence the level of financial inclusion. Fungacova and Weill (2014) found a higher level of financial inclusion in China but lower use of formal credit as compared to other emerging markets. The authors also found that higher levels of income and education as well as higher age and male gender have a positive and significant impact on financial inclusion. Similar results were also obtained for African economies (Zins & Weill, 2016). In addition, Anson et al. (2013) analyzed the influence of post offices on financial inclusion, while Demirgüç-Kunt, Klapper, and Randall (2014) study the use of Islamic finance and formal financial services in Muslim countries. Financial Inclusion in Latin America and the Caribbean Although the aforementioned studies focused on financial inclusion in developing economies, little attention has been paid to the key drivers of financial inclusion in Latin America and the Caribbean at the individual level. LAC has made substantial progress in financial development over the past decade. However, a significant share of the population is still unbanked. Several LAC countries still face financial inclusion gaps when compared to other developing economies.

Gender and financial inclusion in Latin America and the Caribbean  341

While firm financial inclusion in LAC countries is comparable to development levels in other emerging economies, household financial inclusion continues to lag due to low levels of account ownership, savings, and borrowing from formal financial institutions. The limited supply and demand for financial services reflect the region’s socio-economic constraints and vulnerabilities in the macroeconomic environment in terms of income per capita, gender differences, and education, combined with a low degree of export diversification and a large shadow and informal economy (Dabla-Norris et  al., 2015). In addition, institutional weakness plays a salient role in household financial inclusion as not only do the low quality of the regulatory environment and lack of rule of law enforcement reduce incentives to entrust depositors’ funds to formal financial institutions, but low institutional quality also strengthens the adverse impact of high bank concentration and insufficient bank competition on financial inclusion (Rojas-Suarez & Amado, 2014). Comparing account penetration across the three waves of the World Bank’s Global Findex dataset (2011, 2014, and 2017), Micco and Valenzuela report in a chapter of this handbook that the share of households with a bank account in the median country increased in the LAC region from 28 percent in 2011 to 43 percent in 2014 to 47 percent in 2017. Although all countries exhibited significant increases in account penetration during the full period, there is heterogeneity in the region as some countries grow at a faster pace than others. For instance, account penetration, as measured by the share of households with a bank account varies considerably, ranging from 28 percent in Nicaragua to 70 percent in Brazil and 74 percent in Chile according to the 2017 wave of the World Bank Global Findex database. In addition, access to credit displayed much smaller progress in the region. In addition, Global Findex also allows us to examine which population segments have access to formal financial services. There are still significant gaps in financial inclusion by gender and income levels. Despite recent progress in the usage of formal financial services by the adult population in LAC countries, the region’s financial inclusion gaps relative to other countries with a similar degree of development still persist. Although there were reductions in the account ownership gap, both savings and borrowing discrepancies increased in comparison with other emerging markets. Macroeconomic factors can explain these gaps. In a recent study, Rojas-Suarez (2016) found direct and indirect adverse effects of low institutional quality. For instance, the lack of contract enforcement between creditors and debtors directly reduces depositors’ incentives to entrust funds to formal financial institutions. In addition, financial systems in LAC still function with large operational inefficiencies, reflected by high administrative costs and a high degree of bank concentration, which may result in important constraints for financial inclusion and indirectly reinforce the negative impact of lack of adequate bank competition on the financial inclusion of households, reducing the usage of formal financial services. Inefficiencies in the financial system tend to both restrict the availability of financial products and increase the cost of access for depositors. High costs of opening and maintaining accounts and high minimum-balance requirements are outcomes of these operational inefficiencies. Gender and Access to Finance Among unbanked individuals, over one billion are women. There are various ways to examine the impact of gender on finance. Gender equality increases the stock of human capital since a more educated female labor force encompasses the accumulation of skills and raises the overall demand for finance (Amin & Islam, 2014). In addition, the dependency ratio may be

342  Handbook of microfinance, financial inclusion and development

lowered as more educated women join the workforce (Klyver, Nielsen, & Evald, 2013). Gender equality also impacts capacity enhancement. By improving the skill sets of the labor force, gender equality may raise labor productivity and promote investments (Ghosh & Vinod, 2017). In addition to investments, improving women’s income can raise domestic savings, increasing the propensity of women to enter the formal financial system (Seguino & Floro, 2003). There is extensive literature examining the influence of gender disparities in finance on several social and economic outcomes. Demirgüç-Kunt et al. ( 2018) found that the gender disparity in bank account ownership still persists although overall account ownership has increased over time. Studies on the gender gap in financial inclusion are also beginning to appear for emerging economies. Beck, Behr, and Madestam (2018) identified a gender bias in the interest rate on microcredit loans. The authors found that borrowers matched to a loan officer of the opposite gender pay higher rates. Malapit (2012) found that women are more credit constrained in the Philippines. In the case of Sub-Saharan Africa, Aterido, Beck, and Iacovone (2013) provide evidence in support of a gender gap in use of financial provide evidence in support of a gender gap in use of financial services. Agier and Szafarz (2013) exploit data provided by a Brazilian MFI encompassing over 34,000 loan applications. There is no evidence of gender bias when it comes to loan approval. However, the data analysis pointed to a glass ceiling in loan size, which disproportionately increases with respect to the scale of the borrower’s project. Therefore, while on one hand credit access seems fair, on the other hand, women with larger business projects face restrictions for getting larger loans.

METHODOLOGY We use the World Bank’s 2017 Global Findex database to conduct our econometric analyses. The database includes individual-level data originating from a survey of nearly 150,000 adults in 143 countries. Using randomly selected, nationally representative samples, the survey comprises approximately 1,000 adults in each economy. The Global Findex database provides several indicators of financial inclusion that enable the assessment of the amount of account penetration, use of financial services, reasons for not owning accounts, and alternatives to formal finance. It also provides micro-level individual information, such as gender, age, income, and education. We use a random sample of 16,504 adults in 17 countries in Latin America and the Caribbean for our analysis.1 Each LAC country has a representative sample of approximately 1,000 adults except for Haiti, with a representative sample of 504 adults. Consistent with previous literature (Demirgüç-Kunt & Klapper, 2013; Fungáčová & Weill, 2014; Zins & Weill, 2016), we performed a probit base estimation in order to evaluate the determinants of financial inclusion in Latin America through the following equation: financial inclusioni = a + b1iincomei + b2i educationi + b3i agei

+ b4i genderi + b5i workforcei + b6iincomei * genderi + b7 i educi * genderi + b8i agei * genderi + b9i workforcei * genderi + b10countryi + ei

(18.1)

Gender and financial inclusion in Latin America and the Caribbean  343

Financial inclusion, the first outcome dummy variable that measures financial inclusion, refers to the ownership of an account in a formal financial institution, such as banks, credit unions, cooperatives, or microfinance institutions, and i represents a given individual. The survey question used in this case is, “Do you currently have a bank account at a formal financial institution?” Since we only observe whether an individual uses a bank account to save and borrow if he or she owns an account at a formal financial institution, estimating the use of an account to both save and borrow involves estimating bivariate probit models (Corrado & Corrado, 2014; Goedecke et al., 2018; Mohieldin & Wright, 2000). These models are joint models for two binary outcomes that generalize the index function model from one latent variable to two latent variables that may be related via correlation of the error terms that appear in the index function model formulation of the binary outcomes. Specifically, the two outcomes are determined by two unobserved latent variables,

y1* = x1' b1 + e1



y2* = x2' b 2 + e 2

The errors ε1 and ε2 are jointly normally distributed with means of zero, variances of 1, and correlations of ρ, and we observe the two binary outcomes

* ïì1 if y1 > 0 y1 = í * îï0 if y1 £ 0



ìï1 if y2* > 0 y2 = í * ïî0 if y2 £ 0

As a result, the second financial inclusion indicator, formal saving, is based on saving behavior using an account at a formal financial institution in the past year. This is defined using the following question, “Have you saved or set aside money in a bank account in the past 12 months?” The third measure of financial inclusion, formal borrowing, considers the usage of bank credit and refers to the borrowing behavior of individuals from a formal financial institution, excluding credit card use. In order to understand individual borrowing behavior, the survey asks the question, “Have you borrowed from a bank or another type of financial institution in the past year?” All three variables are dummies equal to one if the person responded “yes” and zero otherwise. The explanatory variables comprise different groups of individual characteristics provided in the Global Findex dataset: income, education, age, and gender. Income includes four dummy variables that indicate an individual’s income in quintiles, ranging from the first (20 percent quintile) to the fourth (60–80 percent quintile). The omitted category is the fifth income quintile, > 80 percent. Education includes two dummies for individuals’ completion of zero to eight years of schooling (primary education) and nine to 15 years of schooling (secondary education). The omitted variable corresponds to the completion of a four-year university degree (tertiary education). In addition, we take into account gender by including a dummy variable equal to one if the individual is a woman (female), whether women are part of the

344  Handbook of microfinance, financial inclusion and development

labor force (workforce), as well as both age, defined in terms of number of years, and squared age in the estimations in order to consider potential nonlinearity in the relationship between age and financial inclusion. We also include interaction terms of education, income, age, and workforce with gender. Finally, we include country dummies to capture potential country unobserved heterogeneity.

RESULTS Table 18.1 displays the proportion of financial inclusion access and use for different LAC countries. The proportion of individuals with current accounts ranges from 29 percent to 76 percent, averaging 49 percent in the region. The proportion of savings and borrowing is much lower in the region, averaging 39 percent and 37 percent respectively. Table 18.2 displays the descriptive statistics by gender. Women represent 40 percent of the dataset (6,561 women compared to 9,943 men). In terms of financial inclusion, 52 percent of women have a current account at a formal financial institution (44 percent for men), 46 percent save formally (36 percent for men) and 39 percent borrow formally (33 percent for men).

Table 18.1  Financial inclusion access and use (percent)

Argentina

Account ownership

Savings

Borrowing

No

Yes

No

Yes

No

Yes

47

53

70

30

61

39

Bolivia

47

53

43

57

55

45

Brazil

29

71

71

29

62

38

Chile

32

68

55

45

59

41

Colombia

57

43

64

36

60

40

Costa Rica

33

67

46

54

67

33

Dominican Republic

48

52

53

47

52

48

Ecuador

50

51

69

32

70

30

El Salvador

69

31

63

37

76

24

Guatemala

59

41

61

39

68

33

Haiti

71

29

55

45

59

41

Honduras

56

44

61

39

67

33

Mexico

66

34

63

37

71

30

Nicaragua

71

29

56

44

69

31

Paraguay

66

34

74

26

66

34

Peru

57

43

61

39

63

37

Uruguay

40

60

67

33

49

51

Venezuela, RB

24

76

56

44

56

44

LAC

51

49

61

39

63

37

Gender and financial inclusion in Latin America and the Caribbean  345

Table 18.2  Descriptive statistics All

Women

Men

mean

mean

mean

0.52

0.44

Women

0.40

Account

0.47

Savings

0.40

0.46

0.36

Borrowing

0.36

0.39

0.33

Income: 20% below

0.18

0.13

0.21

Income: 20–40%

0.18

0.15

0.20

Income: 40–60%

0.19

0.18

0.20

Income: 60–80%

0.21

0.22

0.20

Income 80% above

0.24

0.32

0.18

Primary education

0.39

0.37

0.41

Secondary education

0.51

0.53

0.50

Tertiary education

0.10

0.10

0.09

In workforce

0.67

0.79

0.59

Age

41.07

41.79

40.60

N

16,504

6,561

9,943

Within the income category, 13 percent of the women are in the 20 percent below category (21 percent for men), 15 percent are in the 20–40 percent income category (20 percent for men), 18 percent are in the 40–60 percent category (20 percent for men), 22 percent of women are in the 60–80 percent income category (20 percent for men), and 32 percent of women are in the 80 percent and above income category (18 percent for men). In terms of education, 37 percent of the women completed primary education, 53 percent completed secondary education and 10 percent completed tertiary education (college degree). These percentages are very similar to the education of men. Forty-one percent of the men completed primary education, 50 percent completed secondary education, and 10 percent completed tertiary education. The average age is also similar (41.79 for women and 40.6 for men). In addition, 79 percent of the women are in the labor force (59 percent of the men). Table 18.3 displays the results of the model estimations for having access to a formal current account. Model 1 does not include interaction terms, whereas models 2–5 include interaction terms between female and the other explanatory variables individually. Model 6 includes all interaction terms together. Our findings indicate that greater income levels are associated with a higher probability of formal account ownership. Compared to the richest 20 percent, all other income quintiles are less likely to own accounts from formal financial institutions as well as withdraw money less frequently. The difference in probability is larger between the richest and the poorest 20 percent, although the poorest 20 percent are more likely to borrow than the second 20 percent. Likewise, higher levels of schooling are associated with a greater likelihood of being financially included as well as higher withdrawal frequency. Individuals who completed a four-year

346  Handbook of microfinance, financial inclusion and development

Table 18.3  Probit model: current account ownership at a formal financial institution

Income: 20% below

Income: 20–40%

Income: 40–60%

Income: 60–80%

Primary education

Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

–0.671

–0.610

–0.672

–0.673

–0.673

–0.632***

(0.035)

(0.044)

(0.035)

(0.035)

(0.035)

(0.045)

–0.536***

–0.460***

–0.537***

–0.537***

–0.536***

–0.481***

(0.034)

(0.044)

(0.034)

(0.034)

(0.034)

(0.044)

–0.379***

–0.390***

–0.377***

–0.379***

–0.380***

–0.405***

(0.032)

(0.043)

(0.032)

(0.032)

(0.032)

(0.044)

–0.259***

–0.242***

–0.257***

–0.258***

–0.259***

–0.250***

(0.031)

(0.043)

(0.031)

(0.031)

(0.031)

(0.043)

–0.904***

–0.910***

–0.825***

–0.901***

–0.898***

–0.808***

(0.042)

(0.042)

(0.053)

(0.042)

(0.042)

(0.054)

–0.554***

–0.479***

–0.547***

–0.545***

–0.488***

(0.039)

(0.039)

(0.050)

(0.039)

(0.039)

(0.050)

0.268***

0.271***

0.271***

0.267***

0.311***

0.310***

(0.024)

(0.024)

(0.024)

(0.024)

(0.029)

(0.029)

0.0790***

0.132***

0.248***

–0.0682

0.186***

0.215**

(0.022)

(0.043)

(0.073)

(0.053)

(0.043)

(0.106)

0.0318***

0.0314***

0.0317***

0.0305***

0.0328***

0.0307***

(0.003)

(0.003)

(0.003)

(0.003)

(0.003)

(0.003)

***

Secondary education –0.549***

In workforce

Female

Age

Age squared

20% below* female

20–40%* female

40–60%* female

60–80%* female

Primary* female

Secondary* female

***

***

***

***

–0.000268*** –0.000263*** –0.000267***

–0.000271*** –0.000280*** –0.000275***

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

–0.161**

–0.118

(0.070)

(0.072)

–0.199***

–0.152**

(0.068)

(0.070)

0.0552

0.0892

(0.065)

(0.066)

–0.0244

–0.00416

(0.062)

(0.063) –0.198**

–0.233***

(0.081)

(0.086)

–0.175**

–0.155*

(0.079)

(0.080) (Continued)

Gender and financial inclusion in Latin America and the Caribbean  347

Table 18.3  (Continued) Model 1

Model 2

Model 3

Age* female

Model 4

Model 6

0.00357***

0.00380***

(0.001)

(0.001)

Workforce* female

_cons

Model 5

–0.145***

–0.128**

(0.049)

(0.051)

0.0412

0.0233

–0.0254

0.0989

–0.00662

–0.0128

(0.086)

(0.088)

(0.090)

(0.088)

(0.087)

(0.096)

N

16,425

16,425

16,425

16,425

16,425

16,425

chi2

2,766.4

2,785.4

2,761.2

2,768.3

2,770.0

2,783.6

Country dummies

Y

Y

Y

Y

Y

Y

Notes:  * p < 0.10, ** p < 0.05, *** p < 0.01. The base income category is > 80 percent quintile. The base education category is tertiary education Robust standard errors in parentheses. Marginal effects as coefficients.

college degree are more likely to own a current account at formal financial institutions than those who complete either high school or primary school. Our results also suggest a non-linear relation between age and current account ownership from formal financial institutions. While age is positively related to financial inclusion, age squared has a negative coefficient. This indicates that older people are more likely to be financially included, although the probability of financial inclusion decreases after a certain age. In addition, individuals in the formal workforce are more likely to be financially included. Women are more likely to have access to a bank account from formal financial institutions than men. Once we include the interaction terms between women and the other explanatory variables, our results show a negative and statistically significant on the interaction between women and the lower two income quintiles, as well as the interaction between women and both primary and secondary education. The interaction between women and workforce participation also has a negative and statistically significant coefficient, while the interaction between women and age has a positive coefficient. Table 18.4 presents the predicted marginal effects for income and education among men and women. The predicted marginal effects are higher for women relative to men in the top three quintiles of income distribution but lower in the bottom two. In terms of education, the predicted marginal effects are higher for women relative to men in all categories. Tables 18.5 and 18.6 report the impact of individual characteristics on the likelihood of use of formal financial instruments, such as savings and borrowing. Overall, the result in Table 18.5 for the use of financial products and savings are similar to the ones found in Table 18.2 for access to financial inclusion. Compared to the richest 20 percent, all other income quintiles are less likely to save and borrow from formal financial institutions. Higher levels of schooling are also associated with greater likelihood of using formal financial instruments. Women are more likely to save and borrow from formal financial institutions. Similar to our findings for account ownership, individuals in the formal workforce are significantly more likely to be financially included. However, our results differ in the nonlinear

348  Handbook of microfinance, financial inclusion and development

Table 18.4  Predicted marginal effects of income and education by gender Women marginal effect

Men marginal effecct

  Income: 20% below

0.35***

0.36***

  Income: 20–40%

0.39***

0.41***

  Income: 40–60%

0.50***

0.44***

  Income: 60–80%

0.53***

0.49***

  Income 80% above

0.62***

0.58***

  Primary education

0.39***

0.38***

  Secondary education

0.52***

0.50***

  Tertiary education

0.73***

0.66***

Income quintiles

Education

Notes:  * p < 0.10, ** p < 0.05, *** p < 0.01.

relation between age and use of savings and borrowing instruments from formal financial institutions. We found a negative relationship between age and savings and a positive relationship between its squared term and savings, indicating that the probability of saving is greater at the extreme years of age. The results are not statistically significant. In terms of borrowing, our results suggest that age has a significant positive impact on the likelihood of borrowing from formal financial institutions, although its negative term has a significant negative impact on the probability of borrowing. Additional Tests We omitted the higher income quintile and tertiary education categories in our regression models. Since fewer women are in these categories as compared to men, and these fewer women may use formal financial services at a high rate, we estimated Table 18.3 again but used the lowest income quintile (bottom 20 percent) and primary education as the reference categories. Table 18.7 displays the results. Overall, we found a positive and significant effect of the interaction between women and the top three quintiles. Our results also show a positive and significant coefficient of the interaction between women and tertiary education. Once we include all interaction terms (model 6), the positive coefficients of the interaction terms between women and the third income quintile as well as women and tertiary education remain statistically significant.

DISCUSSION This study aimed to examine the key drivers of financial inclusion in Latin America and the Caribbean through the assessment of the impact of individual determinants with a special focus on gender issues through moderation effects of gender on the relationship between individual characteristics and financial inclusion. Our main findings suggest that individuals with

Gender and financial inclusion in Latin America and the Caribbean  349

Table 18.5  Bivariate probit models: formal savings

Income: 20% below

Income: 20–40%

Income: 40–60%

Income: 60–80%

Primary education

Secondary education

In workforce

Female

Age

Age squared

20% below* female

20–40%* female

40–60%* female

60–80%* female

Primary* female

Secondary* female

Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

–0.654

–0.658

–0.654

–0.654

–0.654

–0.667***

(0.035)

(0.045)

(0.035)

(0.035)

(0.035)

(0.045)

–0.489***

–0.500***

–0.489***

–0.489***

–0.489***

–0.508***

(0.033)

(0.044)

(0.034)

(0.033)

(0.033)

(0.044)

–0.393***

–0.383***

–0.392***

–0.393***

–0.392***

–0.389***

(0.032)

(0.043)

(0.032)

(0.032)

(0.032)

(0.043)

–0.246***

–0.293***

–0.245***

–0.246***

–0.246***

–0.298***

(0.031)

(0.043)

(0.031)

(0.031)

(0.031)

(0.043)

–0.657***

–0.657***

–0.603***

–0.657***

–0.659***

–0.609***

(0.040)

(0.040)

(0.051)

(0.040)

(0.040)

(0.052)

–0.362***

–0.362***

–0.315***

–0.362***

–0.364***

–0.314***

(0.037)

(0.037)

(0.048)

(0.037)

(0.037)

(0.048)

0.354***

0.354***

0.356***

0.355***

0.336***

0.340***

(0.025)

(0.025)

(0.025)

(0.025)

(0.029)

(0.029)

0.118***

0.0974**

0.226***

0.147***

0.0722*

0.177*

(0.022)

(0.042)

(0.068)

(0.052)

(0.044)

(0.102)

–0.0216***

–0.0217***

–0.0217***

–0.0213***

–0.0220***

–0.0219***

(0.003)

(0.003)

(0.003)

(0.003)

(0.003)

(0.003)

0.000158***

0.000160***

0.000159***

0.000159***

0.000163***

0.000165***

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

***

***

***

***

***

–0.000269

0.0188

(0.070)

(0.073)

0.0210

0.0358

(0.067)

(0.069)

–0.0348

–0.0204

(0.064)

(0.065)

0.104*

0.115*

(0.061)

(0.062) –0.129*

–0.122

(0.076)

(0.080)

–0.116

–0.120

(0.073)

(0.075) (Continued)

350  Handbook of microfinance, financial inclusion and development

Table 18.5  (Continued) Model 1

Model 2

Model 3

Age* female

Model 4

Model 6

–0.000736

–0.000537

(0.001)

(0.001)

Workforce* female

_cons

Model 5

0.0603

0.0543

(0.050)

(0.051)

0.485***

0.498***

0.441***

0.473***

0.505***

0.469***

(0.086)

(0.088)

(0.090)

(0.088)

(0.088)

(0.096)

N

16,425

16,425

16,425

16,425

16,425

16,425

chi2

2,073.3

2,078.1

2,074.3

2,076.8

2,078.1

2,082.6

Country Dummies

Y

Y

Y

Y

Y

Y

Notes:  * p < 0.10, ** p < 0.05, *** p < 0.01. The base income category is > 80 percent quintile. The base education category is tertiary education. Robust standard errors in parentheses. Marginal effects as coefficients.

greater income and education levels are more likely to own an account, save, and borrow from formal financial institutions. Consistent with previous results in China (Fungáčová & Weill, 2014), Sub-Saharan Africa (Allen et al., 2014; Zins & Weill, 2016), and the region comprising the Middle East and North Africa (Demirgüç-Kunt et al., 2014), the findings for our study point to the benefits of investments in education. Access to secondary and tertiary education may have positive effects on income at the household level. In addition, unlike previous literature (Fungáčová & Weill, 2014; Zins & Weill, 2016), women were found to be more likely to have access to and use of financial inclusion in terms of bank account ownership, savings, and borrowing from formal financial institutions. A potential explanation indicates that conditional cash transfer programs in the region may be important to the growth in access to both account ownership and use for women in the lower income stratum as they are the main beneficiaries of payments of social benefits (Bebczuk, 2008). In Brazil, for instance, women are more likely to use banking agents’ services, which seems to be linked to the monthly disbursement of conditional cash transfers (Sanford, 2014). In addition, Boehe and Barin Cruz (2013) find evidence that women develop a set of capabilities that distinguish them from male borrowers and that may be useful to promote, for instance, women’s inclusion in communities and economic networks that leverage incomegeneration activities. At the country level, however, the gender gap may persist in the access and use of bank accounts at formal financial institutions. For instance, women were found to be financially excluded in Venezuela, Mexico, Costa Rica, and Jamaica, countries where women may have lower levels of education and income (Dabla-Norris et al., 2015). As a result, policies to broaden financial inclusion access and use in the region should continue targeting less educated women in the bottom 40 percent income level. In LAC economies, poor and less educated working women are more likely to have informal jobs and receive regular payments in cash – for wages, domestic remittances, from the government, the sale of agricultural products, and proceeds from self-employment. Women having lower levels of education and income may benefit disproportionately when government and firms digitize

Gender and financial inclusion in Latin America and the Caribbean  351

Table 18.6  Bivariate probit models: formal borrowing

Income: 20% below

Income: 20–40%

Income: 40–60%

Income: 60–80%

Primary education

Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

–0.379

–0.370

–0.379

–0.379

–0.380

–0.373***

(0.035)

(0.044)

(0.035)

(0.035)

(0.035)

(0.045)

–0.237***

–0.255***

–0.237***

–0.237***

–0.237***

–0.258***

(0.033)

(0.044)

(0.033)

(0.033)

(0.033)

(0.044)

–0.226***

–0.261***

–0.226***

–0.226***

–0.227***

–0.264***

(0.032)

(0.043)

(0.032)

(0.032)

(0.032)

(0.043)

–0.121***

–0.154***

–0.121***

–0.121***

–0.121***

–0.155***

(0.031)

(0.043)

(0.031)

(0.031)

(0.031)

(0.043)

–0.463***

–0.465***

–0.453***

–0.463***

–0.460***

–0.444***

(0.040)

(0.040)

(0.051)

(0.040)

(0.040)

(0.052)

–0.246***

–0.238***

–0.245***

–0.244***

–0.232***

(0.036)

(0.036)

(0.047)

(0.036)

(0.036)

(0.048)

0.383***

0.383***

0.384***

0.383***

0.411***

0.412***

(0.025)

(0.025)

(0.025)

(0.025)

(0.029)

(0.029)

0.0367*

0.00769

0.0540

0.0158

0.111**

0.0923

(0.022)

(0.041)

(0.065)

(0.054)

(0.046)

(0.101)

0.0324***

0.0323***

0.0324***

0.0322***

0.0331***

0.0328***

(0.003)

(0.003)

(0.003)

(0.003)

(0.003)

(0.003)

–0.000411***

–0.000409***

–0.000410***

–0.000411***

–0.000418***

–0.000417***

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

***

Secondary education –0.246***

In workforce

Female

Age

Age squared

20% below* female

20–40%* female

40–60%* female

60–80%* female

Primary* female

Secondary* female

***

***

***

***

–0.0533

–0.0487

(0.071)

(0.073)

0.0362

0.0437

(0.067)

(0.069)

0.0793

0.0839

(0.064)

(0.065)

0.0680

0.0706

(0.061)

(0.062) –0.0232

–0.0389

(0.074)

(0.078)

–0.0169

–0.0274

(0.071)

(0.072) (Continued)

352  Handbook of microfinance, financial inclusion and development

Table 18.6  (Continued) Model 1

Model 2

Model 3

Age* female

Model 4

Model 5

Model 6

0.000521

0.000374

(0.001)

(0.001)

Workforce* female

–0.0958*

–0.0966*

(0.051)

(0.052)

–0.586***

–0.567***

–0.593***

–0.578***

–0.619***

–0.604***

(0.086)

(0.088)

(0.090)

(0.088)

(0.088)

(0.096)

N

16,425

16,425

16,425

16,425

16,425

16,425

chi2

1,359.5

1,365.7

1,359.9

1,358.4

1,358.2

1,364.4

Country Dummies

Y

Y

Y

Y

Y

Y

_cons

Notes:  * p < 0.10, ** p < 0.05, *** p < 0.01. The base income category is > 80 percent quintile. The base education category is tertiary education Robust standard errors in parentheses. Marginal effects as coefficients.

transfer payments. Digitizing such payments may increase account ownership and reduce the gender gap among poor and less educated individuals. Among women with an account in Brazil, for instance, about 10 percent got their first account to receive government transfers, while nearly a quarter of account owners in the poorest 40 percent of households in Argentina opened their first account for the same reasons (Demirgüç-Kunt et al., 2018).

CONCLUSION Financial inclusion is an important determinant of economic development as being financially included may lead to household investments in education and entrepreneurial activities as well as allowing individuals to save and borrow money. Lack of adequate access to formal financial services can have detrimental effects on households, such as hampering liquidity levels. Financial inclusion is a particular concern in LAC as banking systems are less inclusive than those from more developed countries. The objective of this chapter was to examine the determinants of financial inclusion in the LAC region, with a particular focus on understanding gender issues related to the ownership of a bank account, savings in a bank account, and use of bank borrowing. We also examined the impact of financial inclusion obstacles and investigated the main determinants of informal savings and informal credit. Overall results indicate that greater income and education levels are associated with a higher probability of financial inclusion. In addition, women are more likely to have access to a bank account, save, and borrow from formal financial institutions than men, and age has a non-linear relationship with financial inclusion. Future areas of research should further investigate the use of formal financial services among underrepresented groups in LAC countries and target specific actions to mitigate the main obstacles to financial inclusion. In addition, future research should investigate the opportunities for expanding financial inclusion through digital technology, particularly focusing on how digitizing payments from government and businesses may reduce the gender gap.

Gender and financial inclusion in Latin America and the Caribbean  353

Table 18.7  Probit model: current account ownership at a formal financial institution Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

0.135

0.151

0.135

0.136

***

0.137

0.152***

(0.035)

(0.043)

(0.035)

(0.035)

(0.035)

(0.043)

0.293***

0.220***

0.295***

0.294***

0.293***

0.227***

(0.035)

(0.043)

(0.035)

(0.035)

(0.035)

(0.043)

0.412***

0.368***

0.415***

0.414***

0.414***

0.383***

(0.034)

(0.043)

(0.034)

(0.034)

(0.034)

(0.043)

0.671***

0.610***

0.672***

0.673***

0.673***

0.632***

(0.035)

(0.044)

(0.035)

(0.035)

(0.035)

(0.045)

Secondary education 0.355***

0.356***

0.346***

0.353***

0.353***

0.321***

(0.026)

(0.026)

(0.031)

(0.026)

(0.026)

(0.033)

0.904***

0.910***

0.825***

0.901***

0.898***

0.808***

(0.042)

(0.042)

(0.053)

(0.042)

(0.042)

(0.054)

0.268***

0.271***

0.271***

0.267***

0.311***

0.310***

(0.024)

(0.024)

(0.024)

(0.024)

(0.029)

(0.029)

0.0790***

–0.0297

0.0501

–0.0682

0.186***

–0.136

(0.022)

(0.056)

(0.036)

(0.053)

(0.043)

(0.098)

0.0318***

0.0314***

0.0317***

0.0305***

0.0328***

0.0307***

(0.003)

(0.003)

(0.003)

(0.003)

(0.003)

(0.003)

–0.000268*** –0.000263*** –0.000267*** –0.000271***

–0.000280***

–0.000275***

(0.000)

(0.000)

(0.000)

Income: 20–40%

Income: 40–60%

Income: 60–80%

Income:80% above

Tertiary education

In workforce

Female

Age

Age squared

20–40%* female

40–60%* female

60–80%* female

80% above* female

Secondary* female

Tertiary* female

***

***

(0.000)

***

(0.000)

***

(0.000)

–0.0378

–0.0342

(0.077)

(0.077)

0.217***

0.207***

(0.074)

(0.074)

0.137*

0.114

(0.072)

(0.072)

0.161**

0.118

(0.070)

(0.072) 0.0229

0.0783

(0.045)

(0.051)

0.198**

0.233***

(0.081)

(0.086) (Continued)

354  Handbook of microfinance, financial inclusion and development

Table 18.7  (Continued) Model 1

Model 2

Model 3

Age* female

Model 4

Model 5

Model 6

0.00357***

0.00380***

(0.001)

(0.001)

Workforce* female

–0.145***

–0.128**

(0.049)

(0.051)

–1.534***

–1.497***

–1.522***

–1.474***

–1.578***

–1.454***

(0.081)

(0.083)

(0.082)

(0.083)

(0.083)

(0.089)

N

16,425

16,425

16,425

16,425

16,425

16,425

chi2

2,766.4

2,785.4

2,761.2

2,768.3

2,770.0

2,783.6

Country dummies

Y

Y

Y

Y

Y

Y

_cons

Notes:  * p < 0.10, ** p < 0.05, *** p < 0.01. The base income category is < 20 percent quintile. The base education category is primary education. Robust standard errors in parentheses. Marginal effects as coefficients.

NOTE 1. The countries included in our sample are: Argentina, Bolivia, Brazil, Colombia, Costa Rica, the Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, and Venezuela.

REFERENCES Agier, I., & Szafarz, A. (2013). Microfinance and gender: Is there a glass ceiling on loan size? World Development, 42(1), 165–181. Allen, F., Carletti, E., Cull, R., Qian, J. Q. J., Senbet, L., & Valenzuela, P. (2014). The African financial development and financial inclusion gaps. Journal of African Economies, 23(5), 614–642. https://doi​ .org​/10​.1093​/jae​/eju015 Allen, F., Demirgüç-Kunt, A., Klapper, L., & Martinez Peria, M. S. (2016). The foundations of financial inclusion: Understanding ownership and use of formal accounts. Journal of Financial Intermediation, 27, 1–30. https://doi​.org​/10​.1016​/j​.jfi​.2015​.12​.003 Amin, M., & Islam, A. (2014). Are there more female managers in the retail sector? Evidence from survey data in developing countries. Journal of Applied Economics, 17(2). https://doi​.org​/10​.1016​/ S1514​- 0326(14)60010-6 Anson, J., Berthaud, A., Klapper, L., & Singer, D. (2013). Financial inclusion and the role of the post office (Policy Research Working Paper No. 6630). Aterido, R., Beck, T., & Iacovone, L. (2013). Access to finance in sub-saharan Africa: Is there a gender gap? World Development, 47, 102–120. https://doi​.org​/10​.1016​/j​.worlddev​.2013​.02​.013 Banerjee, A., Chandrasekhar, A. G., Duflo, E., & Jackson, M. O. (2013). The diffusion of microfinance. Science (New York, N.Y.), 341(6144), 1236498. https://doi​.org​/10​.1126​/science​.1236498 Bebczuk, R. N. (2008). Financial inclusion in Latin America and the Caribbean: Review and Lessons (No. 68). Beck, T., Behr, P., & Madestam, A. (2018). Sex and credit: Do gender interactions matter for credit market outcomes? Journal of Banking and Finance, 87, 380–396. https://doi​.org​/10​.1016​/j​.jbankfin​. 2017​.10​.018

Gender and financial inclusion in Latin America and the Caribbean  355

Beck, T., & Brown, M. (2011). Use of banking services in emerging markets - household-level evidence. European Banking Center Discussion Paper, 2011–025(July), 1–32. Retrieved from http://papers​. ssrn​.com ​/abstract​=1889998 Boehe, D. M., & Barin Cruz, L. (2013). Gender and microfinance performance: Why does the institutional context matter? World Development, 47(Did), 121–135. https://doi​.org​/10​.1016​/j​.worlddev​.2013​. 02​.012 Browne, K. R. (2006). Evolved sex differences and occ segregation. Journal of Organizational Behavior, 162(February 2005), 143–162. Bruhn, M., & Love, I. (2014). The real impact of improved access to finance: Evidence from Mexico. Journal of Finance, 69(3), 1347–1376. https://doi​.org​/10​.1111​/jofi​.12091 Corrado, G., & Corrado, L. (2014). The geography of financial inclusion across Europe during the global crisis. Journal of Economic Geography, 15(5), 1055–1083. https://doi​.org​/10​.1093​/jeg​/ lbu054 Dabla-Norris, E., Deng, Y., Ivanova, A., Karpowicz, I., Unsal, F., VanLeemput, E., & Wong, J. (2015). Financial inclusion: Zooming in on Latin America. IMF Working Papers, 1–35. Demirgüç-Kunt, A., & Klapper, L. (2013). Measuring financial inclusion: Explaining variation in use of financial services across and within countries. Brookings Papers on Economic Activity, 2013(1), 279–340. https://doi​.org​/10​.1353​/eca​.2013​.0002 Demirgüç-Kunt, A., Klapper, L., & Randall, D. (2014). Islamic finance and financial inclusion: Measuring use of and demand for formal financial services among muslim adults. Review of Middle East Economics and Finance, 10(2), 177–218. https://doi​.org​/10​.1515​/rmeef​-2013​- 0062 Demirgüç-Kunt, A., Klapper, L., & Singer, D. (2013). Financial inclusion and legal discrimination against women: Evidence from developing countries. World Bank Policy Research Working Paper. https://doi​.org​/10​.1596​/1813​-9450​- 6416 Demirgüç-Kunt, A., Klapper, L., Singer, D., Ansar, S., & Hess, J. (2018). The Global Findex Database 2017: Measuring financial inclusion and the fintech revolution. Washington, DC. https://doi​.org​/10​ .1596​/978​-1​- 4648​-1259-0 Demirgüç-Kunt, A., Klapper, L., Singer, D., & Van Oudheusden, P. (2015). The Global Findex Database 2014: Measuring financial inclusion around the world. World Bank Policy Research Working Paper 7255, (April), 1–88. https://doi​.org​/10​.1596​/1813​-9450​-7255 Duflo, E. (2012). Women empowerment and economic development. Journal of Economic Literature, 50(4), 1051–1079. https://doi​.org​/10​.1257​/jel​.50​.4​.1051 Fletschner, D. (2009). Rural women’s access to credit: Market imperfections and intrahousehold dynamics. World Development, 37(3), 618–631. https://doi​.org​/10​.1016​/j​.worlddev​.2008​.08​.005 Fungáčová, Z., & Weill, L. (2014). Understanding financial inclusion in China. China Economic Review, 34, 196–206. https://doi​.org​/10​.1016​/j​.chieco​.2014​.12​.004 Ghosh, S., & Vinod, D. (2017). What constrains financial inclusion for women? Evidence from Indian micro data. World Development, 92, 60–81. https://doi​.org​/10​.1016​/j​.worlddev​.2016​.11​.011 Goedecke, J., Guérin, I., D’Espallier, B., & Venkatasubramanian, G. (2018). Why do financial inclusion policies fail in mobilizing savings from the poor? Lessons from rural south India. Development Policy Review, 36(S1), O201–O219. https://doi​.org​/10​.1111​/dpr​.12272 Han, R., & Melecky, M. (2014). Financial inclusion for financial stability access to bank deposits and the growth of deposits in the global financial crisis. World Development Report, Background (2014). https://doi​.org​/10​.1596​/1813​-9450​- 6577 Khavul, S. (2010). Microfinance: Creating opportunities for the poor? Academy of Management Perspectives, 24(3), 57–71. https://doi​.org​/10​.5465​/AMP​.2009​.43479265 Klyver, K., Nielsen, S. L., & Evald, M. R. (2013). Women’s self-employment: An act of institutional (dis)integration? A multilevel, cross-country study. Journal of Business Venturing, 28(4), 474–488. https://doi​.org​/10​.1016​/j​.jbusvent​.2012​.07​.002 Lusardi, A., & Tufano, P. (2015). Debt literacy, financial experiences and overindebtedndess. Journal of Pension Economics and Finance, 14(4), 332–368. https://doi​.org​/10​.1017​/S1474747215000232 Malapit, H. J. (2012). Are women more likely to be credit constrained? Evidence from low-income urban households in the Phillipines. Feminist Economics, 18(3), 81–108. Mohieldin, M. S., & Wright, P. W. (2000). Formal and informal credit markets in Egypt. Economic Development and Cultural Change, 48(3), 657–670. https://doi​.org​/10​.1086​/452614

356  Handbook of microfinance, financial inclusion and development

Rhine, S. L. W., & Greene, W. H. (2006). The determinants of being unbanked for U.S. immigrants. Journal of Consumer Affairs, 40(1), 21–40. https://doi​.org​/10​.1111​/j​.1745​-6606​.2006​.00044.x Rhine, S. L. W., Greene, W. H., & Toussaint-Comeau, M. (2006). The importance of check-cashing businesses to the unbanked: Racial/Ethnic differences. The Review of Economics and Statistics, 88(1), 146–157. https://doi​.org​/10​.1162​/rest​.2006​.88​.1​.146 Rojas-Suarez, L. (2016). Financial inclusion in Latin America: Facts and obstacles (No. 439). https:// doi​.org​/10​.2139​/ssrn​.2860875 Rojas-Suarez, L., & Amado, M. A. (2014). Understanding Latin America’s financial inclusion gap, (May 2014), 1–48. Retrieved from http://www​.cgdev​.org​/publication​/understanding​-latin​america’s- nancial- inclusion-gap-working-paper-367 Seguino, S., & Floro, M. S. (2003). Does gender have any effect on aggregate saving? An empirical analysis. International Review of Applied Economics, 17(2), 147–166. https://doi​.org​/10​.1080​/ 0269217032000064026 Zins, A., & Weill, L. (2016). The determinants of financial inclusion in Africa. Review of Development Finance, 6(1), 46–57. https://doi​.org​/10​.1016​/j​.rdf​.2016​.05​.001

19. Inclusive finance and agricultural development in Africa Calum G. Turvey and Apurba Shee

INTRODUCTION In this chapter, we explore the role of agricultural finance and inclusive finance as it relates to continental Africa, although the primary focus is on conditions in Sub-Saharan Africa. Meyers (2015) writes of the slow process of developing African rural financial markets. To speak of Africa as a whole is difficult since financial development with few exceptions (e.g. Africa Risk Capacity (ARC)) is on a country-by-country basis rather than a continental strategy. The results as characterized by Meyers (2015) are disappointing, including low levels of financial intermediation, high and varied interest rates resulting from currency and macroeconomic uncertainties, high government demand for loan funds, lack of competition, relatively small bank sizes, and contractual problems including weak creditor rights, compromised courts, a deficient insolvency framework, and a general disrespect for contracts (Meyers 2015, p. 7; cf. Honohan and Beck 2007). These raise significant barriers to financial development in the agricultural sector. Historically, several waves of intermediation interventions have been attempted, starting with interest rate ceilings that undercut the marginal costs and risk of credit delivery to farmers (Adams 1971; Adams et al. 1984; Gonzales-Vega 1982; Braverman and Guasch 1989). This was succeeded by a period that adhered to the “financial systems paradigm” which targeted financial institutions, markets and instruments, the legal and regulatory environment, and financial norms and behavior (Myers 2015, p. 8). These initiatives parroted the Washington Consensus which promoted financial sector deregulation of central banks in order to deepen financial markets and reduce economic frictions that had financially repressed markets, interest rates, and economic growth (McKinnon 1973; Shaw 1973; Townsend and Ueda 2006; Roubini and Sala-i-Martin 1992). Financial repression, it is argued, can have a direct impact on credit demand. Bencivenga and Smith (1992), for example, make the case that financial repression results in increased self-financing of investment and accessing of capital from informal means. The argument here is that financial repression places an undue restriction on financial reserves in the formal market which restricts the total supply of loanable funds for purposes of investment. Despite the economic reasoning for financial deepening, the presumption that deepening would create spillover effects to increase the supply of credit into rural markets was sorely overestimated. Capital controls and interest rate policies that might have repressed credit to agriculture before the 1989 Washington Consensus (Williamson 2000, 2004) could never overcome the stark reality that the combined effects of wide covariate risks and costly delivery led to continued credit rationing. The drivers of financial repression might have changed in form but not substance when it came to agricultural finance. As Williamson (2004; see also Kanbur 2009) points out, the problem with the Washington Consensus was that it failed to consider the realities of the economies on the ground and that different countries had different 357

358  Handbook of microfinance, financial inclusion and development

political and social agendas or were at different stages in development. Consequently, the institutions required to deal with large-scale global reforms were not in place. In fact, the combined effects of financial deepening in the urban/industrial complex coupled with continued repression in the rural/agricultural markets might have catalyzed rising inequality, albeit in complex ways (Kuznets 1955; Townsend and Ueda 2006). Continued repression of poorer populations, in general, was a motivator in the development of microfinance institutions (MFI). They emerged in an unregulated form to provide loans to the poor using the strength of social networks and group lending to make small loans (often secured by trust and savings rather than hard collateral assets) to previously unbanked populations (Meyers 2015). Value chain financing also took shape over this period, with some combination of higher-value market access, technical assistance (supporting higher-quality products), and better access to inputs and credit being highly valued (Ricketts et  al. 2014; Bellemare 2012). With value chains offering greater discipline in the use of inputs and direct access to broader domestic and international markets, the participating firms became, in essence, agents of the banking system. Lenders transferred risk from direct lending to farmers to meso-level lending to the chain that could provide greater security in terms of hard assets, contracts, and inventory.

FINANCIAL INCLUSION The concept of financial inclusion emerged more recently. It recognizes that for developing economies to proceed in a sustainable growth pattern, there must be an alignment between the rapidly growing industrial sector and uneven growth observed in agricultural production, productivity, and household income equality. It relates to the notion put forth by IFAD (2016) of “inclusive rural transformation” which is a “process in which rising agricultural productivity, increasing marketable surpluses, expanded off-farm employment opportunities, better access to services and infrastructure, and capacity to influence policy all lead to improved rural livelihoods and inclusive growth” (IFAD 2016, p. 12). Inclusive financial systems are critical to rural transformation in Africa because they offer the capital needed to generate widely based and equitable economic growth. Inclusive financial policies represent a more realistic paradigm that would allow for second-best solutions, or interventions, to undo ineffective policies of the past, recognize market imperfections and market failures for what they are, and attempt to meld those efforts toward a long-run sustainable growth path. This involves at times being both pro-poor and smart. Consequently, financial institutions and policies across Africa need to be adaptive to modern realities. Access and Usage The concept of financial inclusion was originally put forward by the United Nations in 2005. The policy concern was that simply focusing on financial development in terms of financial deepening and depth of financial services focused almost entirely on the factors, policies, and institutions that lead to effective intermediation and markets. Standard metrics, such as the ratio of financial institutions’ assets to GDP, the ratio of liquid liabilities to GDP, or deposits to GDP provided only macro-scale measurements of the financial sector. But these measures bypassed the distributional aspects of credit—particularly in the lower quantiles of the household frequency distribution—and said little about where financial institutions were concentrated, who had access to credit, and who borrowers were. Indeed, financial depth captures the

Inclusive finance and agricultural development in Africa  359

financial sector relative to the economy. It is the size of banks, other financial institutions, and financial markets in a country, taken together and compared to a measure of economic output. If there were particular groups, e.g. poor farm households that were excluded from credit markets by access or rationing, these would not be captured by macro metrics based on averages. To counter this, the concept of financial breadth deals with accessibility to financial services or the level of financial services (Beck et al. 2000; Beck and la Torre 2007). It is the channel with which financial intermediaries can put capital into the national economy and is often measured by the number of branches and, separately, deposit accounts per capita (DemirgüçKunt et  al. 2011), among other performance measures (Dev 2006; Sarma and Pais 2011; Chakravarty and Pal 2013). The difference between financial breadth and financial inclusion is in the focus on the individual borrower. As Beck et al. (2007) point out, increased access to credit does not imply increased use of credit. Financial inclusion, therefore, is aligned with both access and usage, thereby allowing individuals and firms to take advantage of business opportunities, invest in education, save for retirement, and insure against risks (DemirgüçKunt et al. 2008). In reality, financial inclusion as a concept requires the merging of breadth and depth into a single paradigm, suggesting that the two concepts are sufficiently entwined that one can be explained by the other; policies directed toward rural lending can only be achieved by building healthy rural financial institutions. These have to be done in a smart way because of the high costs and risks of underwriting small loans in rural areas. Imposing interest rate ceilings, as was recently done in Kenya, can lead to a flight of capital that dampens breadth and discourages depth. Inclusive financial policies are not directed only at agricultural credit, but also at savings, insurance, and other financial services, including mobile and electronic technologies. Figure 19.1 provides a schematic that relates policy to supply and demand forces. The catalyst is political will, and this will is becoming more evident with the global push for financial inclusion. But the drivers, channels, and solutions are complex. The upper branch is driven by demand forces that improve access to credit, while the lower branch is driven by supply forces to expand access to financial services. Demand is driven by a number of not mutually exclusive factors including demand characteristics, optimal use of inputs and input demand, consumption smoothing, risk balancing, risk aversion, and risk rationing. The supply side is comprised of formal and informal markets, other financial services (insurance), and financial education. The credit facilities include gender balancing, distance to access, collateral and guarantees, loan flexibility, and structured financial products such as bundled, linked, or riskcontingent credit. The government’s role in policy involves many aspects including regulation, oversight, and promoting of financial development, securing transactions, building infrastructure (e.g. roads and cellular), and the promotion of agricultural development through land laws, registration, outreach, and extension. This also involves removing exclusionary regulations that may limit access to certain groups, while promoting inclusive financial policies.

EVIDENCE ON THE RELATIONSHIP BETWEEN INCLUSIVE FINANCE AND AGRICULTURAL PRODUCTIVITY IN AFRICA There are many agricultural programs incorporating inclusive finance across Africa. For example, Abdallah et  al. (2019) identify numerous programs in Ghana including the block farms program (BFP) where land blocks are transferred in smaller units to beginning farmers

360  Handbook of microfinance, financial inclusion and development

Figure 19.1 Schematic of financial development with subsidized input costs and zero-interest loans; the Agricultural Mechanization Services Enterprise Centers (AMSECs) program that subsidizes prices and interest rates on tractor loans; the Stanbic/AGRA loan guarantee program which provides a facilitator/pro rata loss guarantee program; the Wienco Masara N’arziki input-credit project (input credit project to increase small and medium holder productivity); the AGRA/CARD credit program (to provide support for soil health); and institutionally, a collateral registry system (registering of assets under collateral to ease collections on default), the Credit Reference Bureau (centralizing credit scoring), and the USAID Financing Ghanaian Agriculture Project (financial facilities for agribusiness). Barrowclough et  al. (2020) investigate the relationship between financial inclusion and household welfare and develop an index to capture access, quality, and usage. Usage indicators included account ownership, saving propensity, credit availability, and insurance coverage. Account ownership has been linked to increased credit access, savings, and consumption as well as easing receipt of salaries, remittances, and government payments (Demirgüç-Kunt and Klapper 2013). Account ownership was captured by e-banking, mobile money, ATMs, e-zwich biometric smart cards, and a checking account. Savings were captured through savings accounts and fixed deposit accounts, and access was measured by having applied for a loan or receiving a loan. Insurance was drawn from medical, auto, business or long-term annuity, or life insurance. From a sample of 13,000 Ghanaian households (2017 GLSS7 survey),

Inclusive finance and agricultural development in Africa  361

56.4 percent of households were excluded from financial services entirely, 35.9 percent had low inclusion indicating the use of some form of account or transactions (mobile or ATM), 7.3 percent were found to be in medium inclusion, which included checkbook and/or savings accounts and insurance, and only 0.4 percent were in a high inclusive grouping which included obtaining credit. Multinomial logit results indicated that rural households were less likely to be included in the low to medium inclusion groups, but no differences were found in the high inclusion group. Further categorization indicated that respondents with wage labor were significantly more likely to be included in the financial market. This also affects rural youth. For example, Ankrah Twumasi, Jian, and Owusu (2019b), using 2018 data collected in Ghana comprised of 450 rural youth farmers, found that 211 (47 percent) of the respondents were credit constrained. Transaction costs in the form of cumbersome loan application procedures and loan disbursement times contributed to these credit constraints. Youth farmers facing these credit constraints (quantity and transactional) were found to have lower intensity of participation in agricultural activities than a random farmer from the sample. Jumpah et al. (2019) investigated smallholder farmers in Ghana and showed that distance, interest rate, experience, membership in a farmer-based organization, number of dependents, gender, and age were statistically significant farmer- and credit-specific characteristics that influence participation in microfinance programs. The direct and indirect costs of interest rates and distance negatively affected participation. Sackey (2018) finds that credit rationing persists and that applying for a relatively longer payment period, providing collateral and guarantor, being illiterate, being relatively older, and being in the agricultural sector increases the likelihood of being credit rationed, while having some relationship with the bank, having non-mandatory savings, and applying from a bank with relatively high interest rates reduce the likelihood of being credit rationed. Asante-Addo et al. (2017) find that farm households participate in credit programs because of improved access to savings services and agricultural loans, yet the fear of loan default (risk rationing) and lack of savings are reasons for non-participation in credit programs. Membership in farmer-based organizations (FBOs) and the household head’s formal education are positively associated with farmers’ participation in credit programs and credit rationing (i.e. their loan applications were either rejected or the amount of credit they applied for was reduced) was less likely among higher income farmers and members of FBOs such as farmer cooperatives and savings clubs. While gender bias against women is a common finding in agricultural credit, Sarworsi et al. (2016), using data on 9,710 farmers from Madagascar provided by AccèsBanque Madagascar, found that despite observations that female farmers had lower repayment performance, they had a higher rate of loan application approval compared to male farmers. This theme of credit access and credit usage has strong implications for agricultural productivity and general household welfare in Africa. Numerous papers including Nordjo and Adjasi (2019), Martey et al. (2019), Jumpah et al. (2019), Abdallah et al. (2019), Nkegbe (2018), Iddrisu et al. (2018), Sekyi et al. (2017), and Akudugu (2016) use a variety of techniques and small samples but provide strong meta-evidence that there is a direct and significant linkage between access/use of formal credit and, in some cases, informal credit and agricultural productivity. Abdallah (2016) finds that the relationship between credit and technology adoption is a one-way causal relation (i.e. credit access leads to technology adoption) as opposed to a two-way relation (i.e. mutually dependent relation) and that credit market inefficiency can be a major barrier to the adoption of yield-enhancing technologies in Sub-Saharan Africa. Tadesse (2014) uses 2005 and 2007 panel data on 278 Ethiopian households covering 5,700 field plots.

362  Handbook of microfinance, financial inclusion and development

He finds that only 22 percent of the plots actually had fertilizer applied but also found that wealthier farmers were more likely to purchase fertilizer from savings while poorer farmers were more dependent on credit.

THE STATUS OF FINANCIAL INCLUSION IN CONTINENTAL AFRICA Of the 762 MFIs globally tracked by MIX, 113, or 14.8 percent, are in Africa.1 The MIX Market 2017/2018 reports that these MFI financial service providers (FSP) lend to approximately 5.4 million borrowers with gross loans of 9.5 billion USD. These service providers have 26.7 million depositors for a borrower to depositor ratio of about 4.96 and deposits of 13 billion for a loan to deposit ratio of about 0.72. The makeup of FSP is 24 banks, 16 credit unions, 40 non-bank financial institutions, and 32 NGOs. In terms of overall activity, the entirety of borrowers in Africa is just slightly higher than the total number in the Philippines (5.2 million) and considerably less than the 38 million in India. The top five countries are Nigeria (1.89 million), Kenya (826,700), Benin (620,600), Uganda (296,800), and Ghana (246,800). Kenya dominates with 11.7402 million depositors followed by Nigeria with 5.593 million. Africa and its member countries have their own characteristics, which can limit the growth of microfinance institutions. For example, in 2016, the Kenyan government imposed an interest rate cap which in turn repressed lending activity in small enterprises and encouraged a flight to safer corporate borrowers. Drought and political unrest also had an impact on loan disbursements in Kenya. One of the most critical tracking datasets for financial inclusion is the Global Findex Database which is currently available for 2011, 2014, and 2017 (Demirgüç-Kunt et al. 2018).2 Figures 19.2 and 19.3 provide data on account holdings across Sub-Saharan Africa for the three sampling points. Between 2011 and 2017, the share of respondents with financial/mobile accounts increased by 83 percent across all borrowers but only 77 percent for females. On a percentage basis, the poorest 40 percent of households increased by 147 percent, compared to 69 percent for the richest 60 percent of households. The rapid rise for the poor is due to inclusive financial measures targeting the poor, and by this measure, policies appear to be effective. The share of rural households with an account nearly doubled from 19.4 percent to 39.5 percent between 2011 and 2017. Figure 19.3 looks only at respondents’ accounts at financial institutions (excluding mobile accounts), and this tells a different story. In 2011, mobile banking and mobile accounts were in their infancy. Across all respondents in Sub-Saharan Africa, the share with accounts at financial institutions increased from 23.25 percent in 2011 to 32.78 percent in 2017 representing a share increase of 40.99 percent. In comparison with Figure 19.2 which includes mobile accounts, the share increased from 23.25 percent to 42.61 percent for a share increase of 83.3 percent. In other words, there were more mobile accounts opened between 2011 and 2017 than institutional bank accounts. Women increased institutional accounts by 31.5 percent (20.79 percent to 27.34 percent) which is less of an increase than the 77.53 percent (20.79 percent to 36.91 percent) increase when mobile accounts are included. Perhaps most striking is the reliance on mobile accounts by the poor. In 2011, only 12.9 percent of the poor had institutional accounts and this increased to 22.7 percent in 2017. The difference in mobile accounts is the spread between 31.9 percent in Figure 19.2 and 22.7 percent in Figure 19.3, which is 9.8 percent. In other words, just over half (76 percent) of the total 147 percent increase in access to institutional/mobile accounts (12.9 percent to 31.9 percent) was due to mobile technology and

Inclusive finance and agricultural development in Africa  363

Source:  Global Findex Database (2017).

Figure 19.2 Accounts at financial institutions and mobile services, Sub-Saharan Africa

Figure 19.3 Holding accounts at financial institutions not institutional growth. The same pattern is observed for rural respondents whose share of institutional accounts increased from 19.41 percent to 29.87 percent, while the share including mobile accounts increased from 19.41 percent to 39.84 percent. Figure 19.4 pairs up two measures of borrowing. The top panel indicates the percentage of respondents who borrowed any money in 2014 and 2017. The lower panel identifies those that

364  Handbook of microfinance, financial inclusion and development

Figure 19.4 Borrowing access and usage, Sub-Saharan Africa borrowed from a financial institution or used a credit card. Since credit card usage is very low, this measure can reasonably represent the use of banks for loans. The first thing to note is that total borrowing decreased for all groups between 2014 and 2017. All respondents fell from 54.8 percent to 45.7 percent. While this can reasonably capture the total demand for credit, it is notable that only a fraction comes from formal financial institutions. Between 2014 and 2017, loans from banks increased from 7.5 percent to 8.4 percent, less than a 1 percent increase.

Inclusive finance and agricultural development in Africa  365

Loans to men, the wealthy, and rural households increased by over 1 percent, while loans to women and the poor increased by less than 1 percent. The differences between the upper and lower panel represent borrowing that is not satiated through formal lending. This would include familial borrowing from friends and relatives, savings clubs, moneylenders, pawnshops, and suppliers or value chains. In terms of financial inclusion, a comparison of the number of accounts opened represents increasing access, but the actual borrowing from financial institutions suggests that usage is low. With a 37.2 percent gap between total borrowing (45.67 percent) and borrowing from a bank/use credit (8.38 percent) in 2017, it is important to identify where borrowed funds are obtained. Figure 19.5 compares sources of borrowing for selected countries in Sub-Saharan Africa. In Kenya and Uganda, about 46 percent of borrowing is provided by familial lending between friends and relatives. Savings clubs are also popular in Uganda (24.1 percent), Malawi (23.3 percent), and Kenya (19.7 percent). Although these savings clubs are economically significant, they have evolved to different degrees of importance in different countries. For example, they are not well developed or popularized in Ethiopia, Ghana, or Nigeria. The share of respondents who borrowed from banks is considerably lower. In Cameroon, Ethiopia, Malawi, Nigeria, and Tanzania, fewer than 10 percent borrowed from banks. Only Kenya (16.5 percent) and Uganda (15.6 percent) exceed 15 percent and not by much. These observations suggest that degrees of financial inclusion in terms of bank lending have different levels of access and use, and these, in turn, are likely related to the varied pathways to economic development. The final row in Figure 19.5 provides the percentage of respondents that borrowed for farm or related business activities, which is relatively low, ranging from 11.6 percent in Cameroon to 22.1 percent in Uganda. Only one in five Kenyans or Ugandans

Figure 19.5 Sources of credit by country

366  Handbook of microfinance, financial inclusion and development

borrowed for farm business, and this exceeded the percentage of those that borrowed from a bank. The shortfall is made up with funds from family/friends or savings clubs. What is unclear is whether access to formal credit markets is limited by credit rationing, or whether usage is restrained by low and disinterested demand or risk rationing. However, as mentioned, expanded financial development throughout Africa may not so easily translate into broader access to credit services, and likewise, even if access is expanded in a way that is consistent with the goals of financial inclusion policy, this does not necessarily imply that usage will increase.

CREDIT RISK Grasping credit risk in Africa is difficult. However, data from the Council on Smallholder Agricultural Finance through 2018 (https://data​.csaf​.org/) provides some indication of the risk that lenders face. The CSAF consists of 12 private lenders to small and medium enterprises (SME) and value chains globally.3 These are not smallholder loans, but in many instances, the value chains act as lending agents to their member growers. Perhaps more critically, because members are concentrated into the same grower group (predominantly coffee (17 percent), cocoa (29 percent), and cashew (27 percent)), they have the same systemic and covariate risks. Figure 19.6 compares the PAR30 (portfolio at risk, 30 days past due) of the global portfolio and that of Sub-Saharan Africa. PAR 30 ranged from 14.1 percent in 2013 to a low of 4.5 percent in 2017. On average, 10.3 percent of loans in Sub-Saharan Africa were at risk when compared to 8.4 percent for the global portfolio. Figure 19.7 reports the portfolio at risk for a subset of African countries sourced from the MIX report. Average loan balance was 991 USD and deposits were 199 USD. The challenge across Africa is the variability in loan performance due to external factors. A severe drought in Malawi in 2016 that left many on the brink of starvation increased loan PAR90 to 58.2

Figure 19.6 Portfolio risk, Sub-Saharan Africa

Inclusive finance and agricultural development in Africa  367

Figure 19.7 Value at risk, PAR30 by country percent. Political unrest in Cameroon in 2016 corresponded with a significant increase in loan arrears. It is difficult to ascertain what the steady state loan delinquency rate would be, but it is generally low around 2–3 percent. It is the complexity of Africa that makes lending difficult.

SAVINGS Much has also been written about the role of savings in development. Figure 19.8 shows the percentage of respondents in Sub-Saharan Africa that saved in financial institutions between 2011 and 2017. The Findex results show a slight increase in savings between 2011 and 2014 but a decrease between 2014 and 2017 on average for females and the richest 60 percent. It also shows only slight increases for men, the poorest 40 percent, and virtually no change for rural households. Again, relative to increased access to financial services as provided in Figures 19.2 and 19.3, this does not appear to have increased usage by a material amount. Figure 19.9 shows the use of digital and mobile technologies in Sub-Saharan Africa. 29.1 percent of respondents indicated making some form of e-payment in 2017, up from 22.8 percent in 2011. Rural digital use increased from 20 percent to 26.2 percent. Related to this is the use of mobile phones or internet to access accounts. 20.8 percent of all respondents indicated that they had used mobile phones or the internet to access accounts, including 19.1 percent of rural households. Women and the poor tend to fall below the average. However, within Sub-Saharan Africa, there is considerable diversity in both access and use. For example, Figure 19.10 shows ownership of mobile phones in Kenya is 88.6 percent, but in Malawi, it is only 52.3 percent. With M-Pesa, 76.9 percent of Kenyans report having a mobile money account, but in Ghana, it is only 43 percent, Malawi 22.7 percent, and virtually non-existent in Ethiopia. Although many banks will have expanded the breadth of services to accommodate

368  Handbook of microfinance, financial inclusion and development

Figure 19.8 Savings at financial institutions in Sub-Saharan Africa (percent)

Figure 19.9 Digital and mobile phone use, Sub-Saharan Africa mobile and internet access, only 31.8 percent of Kenyans actually use mobile (or internet) technologies to access financial institution accounts. Usage is below 10 percent for Cameroon, Ethiopia, Malawi, Tanzania, and Uganda. In other words, despite access, usage is quite low which could be for a multitude of reasons ranging from the infrastructure of cellular networks and cellular access to whether an account is opened at an FI.​

Inclusive finance and agricultural development in Africa  369

Figure 19.10 Access and use of mobile technologies by country

RISK RATIONING, CREDIT RATIONING, AND COLLATERAL Collateral is a significant barrier to credit access in Africa. In times of stress, e.g. drought, when all savings and food reserves are depleted, farmers often resort to selling consumable, and later productive assets, depending on the severity of stress. The dynamics of risk leave many farmers with an initial position in poverty to be trapped in that poverty, while those with more means could transition into a poverty state and remain there until the collective liquid and productive assets accumulate beyond the poverty threshold (Barrett and Carter 2013). To smooth consumption across periods of adversity, traditional mitigation strategies have relied on a Joseph effect in which households accumulate liquid savings and food reserves during good years, and use these to smooth consumption over bad years. Morduch (1995) suggests that income smoothing is more likely to occur when households anticipate being unable to borrow or insure. In the alternative, farmers can access credit from informal or formal sources or employ other risk coping strategies, including savings clubs. There are two interrelated problems with this narrative. First, even though farmers appear to have improved access to formal credit, they do not use it, and second, this limits the ability to migrate out of low-scale farming. Limits to credit demand in turn remove incentives for FSP to add to the depth or breadth of credit supply. The failure of agricultural credit markets to meet the G20 financial inclusion goals is largely due to a complex endogenous relationship between supply and demand as depicted in Figure 19.1. Part of this complexity is illustrated in Castellani (2014) who finds that shocks that affect household assets in Ethiopia are important in explaining both the decision to borrow and the source of credit.

370  Handbook of microfinance, financial inclusion and development

Risk Rationing The barrier point for low-level equilibrium in African credit markets is risk and collateral. A convenient terminology which has been previously expressed in this chapter is ‘risk rationing’. Boucher, Carter, and Guirkinger (2008; see also Boucher et al. 2009) provide the first formal treatment of risk rationing in an economic and utility-centric context. In their view, risk rationing occurs when insurance markets are absent, and lenders facing asymmetric information shift so much contractual risk to the borrower that the borrower voluntarily withdraws from the credit market. This can arise even when the borrower has the collateral wealth needed to qualify for a loan contract. Formalizing risk rationing as an economic concept explains some puzzling observations by development economists including Binswanger and Siller (1983) who suggested that credit markets for small farmers may disappear because of lack of demand, despite the fact that small farmers might have available collateral in the form of unencumbered land. Bell, Srintvasan, and Udry (1997) found a credit demand relationship in which demand increases with liquid assets but decreases with fixed assets, which they find both puzzling and unsatisfactory. Eswaren and Kotwel (1990) observed that an inordinate degree of risk aversion to credit may be a reflection of their inability to sustain downswings in income Similar situations are echoed in Africa. Shee et al. (2015) find widespread risk rationing in the pastoral Marsabit region of Kenya. Pastoralists were explicit in their views that taking a loan would jeopardize their assets, and they were afraid of losing collateral. What is key is that they do have a demand for credit but do not act upon it thereby risk rationing themselves out of the market. As depicted in Figure 19.5 above, the reliance on familial lending is as much a risk coping strategy as one of convenience.4 When asked how much they would borrow without collateral requirement, they mentioned a need of 100,000–300,000 KSh for various entrepreneurial activities such as milk and meat trade, small shop, animal tracking etc. In Turbi, a pastoral region of Kenya, pastoralists indicated that although Equity Bank had a presence in the region, it was difficult for them to obtain or accept a loan because of collateral risk. To encourage credit access and use, an NGO deposited security in the bank, but borrowing still required a combination of savings and collateral. Table 19.1 summarizes published results from Kenya, Tanzania, and Malawi; China; and Mexico, Peru, Honduras, and Nicaragua. There is a significantly higher incidence of risk rationing in African countries than in China or Latin American countries. Price rationing is significant in all countries, but this simply means that respondent farmers are optimizing along their own credit demand curves. Balancing out the rationing measure is quantity rationing. Kenya and Tanzania are 10.3 percent and 13 percent respectively, but this pales in comparison to the risk-rationed group that may actually have a demand for credit but does not act upon that demand because of the collateral base. As a concept, risk rationing has provided some guidance towards the development of insurance products that could ostensibly substitute for collateral. Earlier concepts, such as Bester (1987) used a simple economic concept that lenders could add a risk premium to cover collateral and default risks to a point where they would be indifferent to offering a risk-free loan at risk-free rates and a risky loan with risk-adjusted rates. Shee and Turvey (2012) took this further by tying the risk premium to a specific contingent claim that weighed heavily on the most common exogenous economic risk facing the farmers (e.g. market prices or specific weather events). The idea of collateral-free lending by bundling insurance with credit is sound

Inclusive finance and agricultural development in Africa  371

Table 19.1  Risk rationing, quantity rationing, and price rationing Country

Risk rationed (%)

Quantity rationed(%)

Price rationed (%)

Kenya (a)

38.4

10.3

51.3

Tanzania (b)

57

13

40

Malawi (c)

38

12

50

China (d)

6.5

14

85

Mexico (d)

35

10

55

Peru 1992 (e)

8.6

36.6

54.9

Peru 2003 (e)

22.4

10.4

67.3

Peru (f)

19

37

46

Honduras (f)

16

23

62

Nicaragua (f)

12

48

40

Sources:  (a) Shee, Turvey, and You (2019); (b) Shee, Pervez, and Turvey (2018); (c) baseline survey data—Dual Cassava Project at NRI; (d) Verteramo-Chiu et al. (2014); (e) Boucher, Guirkinger, and Trivelli (2009); (f) Boucher, Carter, and Guirkinger (2008)

within the confines of a theoretical economic framework but, as will be discussed presently, not so sound in an actual lending environment. Nonetheless, the idea of risk-rationing behavior being distinct from risk aversion is an important one since it recognizes, at least implicitly, that farmers consider balancing business and financial risks in making their decisions to use agricultural credit, should it be available. Insurance A third tranche of inclusive financial policies is the provision of insurance and the opening of insurance markets. While progress has been made in the area of micro insurance for life, property, and casualty, the insurance of interest here is agricultural insurance. In this section we focus on two innovations in the agricultural insurance space that hold some promise for Africa, index-based livestock insurance (IBLI) and bundled credit or risk-contingent credit. The role of insurance is becoming increasingly important in agricultural development. Marr et al. (2016) reviewed 1,133 papers and reports on agricultural insurance and found several themes identifying key factors and indicators affecting insurance demand and uptake upon which most scholars and practitioners agree. The key factors and indicators are (1) risk (nature of risk, risk aversion, risk mitigation, basis risk, price risk); (2) behavior (understanding, trust, and education); and (3) credit and liquidity constraints (credit access, wealth, liquidity, and income). However, insurance markets for hedging production risk arising from adverse weather or market price risks do not typically exist in Africa—especially for small producers; when they do, the inconvenience of obtaining the products via traditional insurance markets may be unattractive to producers. The risk landscape within agriculture in Africa creates a selfreinforcing credit rationing/disinvestment trap in the market with banks unwilling to lend and borrowers unable to invest in more efficient technologies. In the insurance sector, all-risk crop

372  Handbook of microfinance, financial inclusion and development

insurance that has been available in developed economies for decades has not materialized for a number of reasons, including yield measurement, scale, covariate risks, moral hazard, and adverse selection. From a practical standpoint, crop insurance has not developed in Africa because the historical record of crop yields required to define risk probabilities is limited to non-existent. Adding to this, most farmers in Africa are small-scale farmers for which the administrative, monitoring, and underwriting costs are prohibitive. Covariate risks refer to common exogenous risks such as drought that affect all farmers in the risk pool at the same time. When these events happen, local insurers do not have the capital to indemnify the losses and must therefore involve global reinsurance markets. Even if these issues could easily be resolved, insurers are also cautious about moral hazard and adverse selection, although these effects could be moderated through random audits of insureds to ensure compliance (Turvey et al. 2002; Esuola et al. 2007). To overcome these issues, scholars and practitioners have turned to ‘index insurance’. Index insurance is a generic term used to describe insurance models with an indemnity linked to an index rather than crop yields. Among the first of these index insurance models was the area-yield insurance model proposed by Miranda (1991) and weather insurance to capture specific event risks (Turvey 2001). While targeted towards volumetric risk, index models brought about a new element called basis risk in which the event measurement (average county yield or rainfall at a weather station) differs materially from the conditions faced by the insured leading to significant Type I and Type II error (Norton et al. 2013; Woodard and Garcia 2008). Although the implementation of weather-based index insurance models has been criticized on implementation and measurement (Binswanger-Mkhize 2012), there are few alternatives to addressing small-holder insurance in Africa. Index-Based Livestock Insurance (IBLI) A variant index insurance model is the Index-Based Livestock Insurance (IBLI) which has been offered to pastoralist farmers in Kenya since 2010. IBLI is designed to protect livestock asset losses due to covariate rangeland conditions (rainfall) by providing uniform compensation based on signals and observations from the satellite-based Normalized Difference Vegetation Index (NDVI). NDVI, when used as a measure of vegetation required for grazing, is highly correlated with livestock mortality and was used to derive a mortality index. If the mortality index falls below a pre-stated threshold, farmers would receive an indemnity (Chantarat et al. 2013; Woodard et al. 2016). The economic driver behind IBLI is the dynamic asset poverty trap when failures in Sub-Saharan long and/or short rains decrease rangeland grazing, increasing livestock mortality. Post evaluation and simulations reported in Chantarat et al. (2017), however, find that IBLI does not perform well for the poorest whose small asset endowment (camels) will typically collapse in the presence of a drought. However, IBLI does seem to be effective for the vulnerable non-poor who might not see a total collapse in herd numbers and are then positioned to a more rapid re-accumulation of herds with IBLI indemnities in place. In addition to ex-ante assets (herd size) and asset dynamic being a key driver of IBLI uptake and benefits, they also find that the demand is quite elastic. As a commercial product, IBLI requires substantial loadings above the computed actuarial price, and this can dampen demand considerably. In willingness to pay (WTP), Chantarat et al. (2017) report that fully loaded insurance would find demand only with herd sizes above 15 tropical livestock units (TLU), and additionally that the demand is highly elastic. To increase uptake of IBLI, even small subsidies on loadings could increase demand/uptake substantially. Finally, and

Inclusive finance and agricultural development in Africa  373

in line with the Binswanger-Mkhize (2012) critique, they show that offering the maximum coverage is sub-optimal. Index insurance is designed to counter low-frequency, high-impact specific events that are highly correlated with loss (Turvey 2001). Insurance-Bundled Credit or Risk-Contingent Credit Considerable interest has also arisen in bundled or linked credit products. Bundled credit is a structured financial product with embedded collateral-like indexed-base risk transfer mechanisms in the form of insurance contracts or contingent claims (Skees and Barnett 2006). As discussed above, two overarching conclusions are that farmers can improve productivity and household income with improved access and use of credit; but collateral, moral hazard, and adverse selection result in the rationing (risk and quantity) of credit to farmers. The common element that bridges these two conclusions is risk, and the driving force behind bundled-credit products is to provide a mechanism that reduces business risk and collateral exposure to financial risks faced by both borrower and lender. Credit demand-supply endogeneity suggests that in balancing business and financial risks, borrower demand and lender supply could increase at a lower cost. There is an emerging literature on bundled credit. Skees and Barnett (2006) describe the Indian MFI BASIX’ purchase of rainfall insurance from the insurer ICICI-Lombard; how IBLI in Mongolia could be used to reduce default risk on loans; and how insurers and reinsurers could use El Nino measures of surface ocean temperature to protect lenders from exposure to excessive rainfall risk (see also Miranda and Gonzalez-Vega 2011 and Collier et al. 2011). Carter (2011) examined the impact of linked credit on financial deepening and its impact on farm households; In Africa, Giné and Yang (2009) investigated an operating loan product in Malawi in which the payoff was determined by rainfall; Karlan and Zinman (2011) investigated the adoption of price-protected loans in Ghana. Shee and Turvey (2012) outlined how risk-contingent credit (RCC) could be used to indemnify loans for Indian pulse crop farmers. Banerjee, Duflo, and Hornbeck (2014) deployed an RCT in India that encouraged a group of MFI borrowers to bundle life insurance with the credit, leading to an overall decrease in loan uptake. Shee et  al. (2015) report on games played with Kenyan pastoral and dairy farmers to uncover potential demand for risk-contingent credit. These ideas were operationalized in an RCT initiated in Machakos Kenya in 2017. Results of this RCT are reported in Shee et al. (2019) and Ndegwa et  al. (2020). The RCT involved 1,170 randomly selected farm households. Risk-contingent credit linked to accumulated rainfall over the October 15 to January 15 long rain season was bundled with a loan originating with Equity Bank. Random assignment was for no loan offered, a traditional loan, and risk-contingent credit (RCC). A baseline survey collected self-identified risk, quantity, price, and transactions costs rationing. They find that 48 percent of the households were price-rationed, 41 percent were risk-rationed, and 11 percent were quantity-rationed. The average credit uptake across the 819 farmers who were offered credit was 33 percent, with the uptake of bundled credit being significantly higher than that of traditional credit. If bundled credit is to become part of the inclusive finance platform for African farmers, there are several important lessons coming out of this RCT. The first was that the uptake of traditional credit was nearly as high as the RCC (30 percent versus 32 percent) suggesting that linking credit to insurance might not be as strong as theory might suggest. There were some mitigating factors including the fact that the lender did require a full credit application as well

374  Handbook of microfinance, financial inclusion and development

as a pledge of collateral, and the timing of the loan might have been late for some farmers. A second lesson related to risk-rationed farmers. As a conjecture, risk rationing suggests that uptake of RCC would be significantly higher than traditional credit. In fact, the uptake was only 6.8 percent higher, but when it came to uptake of either traditional or RCC, there was no significant difference in the uptake by risk-rationed farmers. Furthermore, a sub-experiment was conducted where 100 RCC borrowers were randomly assigned to 25 percent, 50 percent, and 75 percent subsidies on the insurance costs. The results showed that while uptake was positive for 25 percent and 50 percent subsidies, they were not significantly different from zero, and with a 75 percent subsidy, the uptake was actually negative. This important result, even on a small sample, suggests that the demand for credit generally, and risk contingent credit specifically, is quite inelastic. In comparison to results reported in Chantarat et al. (2017) who found an elastic insurance demand, this may not translate to bundled credit. There might also be a similar effect noted in Banerjee et al. (2014) who found a drop in credit demand when insurance was added to the product. The parallel here was that farmers offered RCC could only borrow RCC, and farmers offered traditional credit had no option to take RCC. Flexibility Inflexibility can also be a barrier to the use of financial services. Laureti and Hamp (2011) state the poor need flexible financial products (savings, credit, and insurance), but there is a tradeoff between flexibility and rigidity in the payment/repayment of financial products. Loan repayment provides grace periods that reduce the ordered discipline of rigid financial structures, or even the possibility of rescheduling loans in the face of exogenous natural adversity as is done by the Bank for Agriculture and Agricultural Cooperatives in Thailand. Weber and Musshoff (2013, 2017) speak to the issue of ‘flex loans’. In their study of Madagascar MFIs, they compare flex loans with repayment balanced to the agricultural production (and liquidity) cycle to standard loans. The loans weren’t perfectly tied to the liquidity cycle but were rather defined by repayment grace periods. They find a higher percentage of farmers accessed flex loans, but the loan amounts were lower than those accessing standard loans. They also find that flex loans had a higher delinquency rate than standard loans, but not materially so. If innovative risk transfer products, whether linked to credit or not, are to succeed in SubSaharan Africa, farmers ought to be offered choice and flexibility. Another example can be found in bundled credit. The original Kenya RCT discussed above was designed around a higher interest rate – a risk premium above the base rate – to absorb the costs of insurance. The lender would then transfer the insurance to the insurer on a pro-rata basis. Ultimately, the insurer required the payment be paid in full upfront and the insurance premium (about 13 percent) was added to the loan principal of 10,000 KSh. The flexibility of adding the insurance premium to the loan avoids problems raised by Casaburi and Willis (2018) who showed that liquidity constraints that prohibited Kenyan contract farmers from purchasing insurance could be alleviated if payment was deducted from the final contract payment. Chantarat et al. (2017) issue a similar warning noting that insurance that consumes scarce resources, and fails to protect the household from catastrophic shocks, can do damage (p. 125). Finally, flexibility should be added to index insurance design. Turvey et al. (2019) describe the failure of the Kenya RCC design to capture within-season basis risk. In the fall of 2017, early season rains exceeded the seasonal rainfall insurance threshold so that despite a late season drought, which caused significant crop losses, no insurance was paid out under the

Inclusive finance and agricultural development in Africa  375

insurance terms. Ultimately, some Indemnities were paid by project funds, but it became evident that designing insurance products based on historical seasonal averages was insufficient. In response, a new and flexible dynamic trigger was implemented with an indemnity paid if the accumulated rainfall in up to four non-overlapping 21-day periods fell below a percentage of the historical average. This new, flexible, design ensures that farmers receive some compensation if severe drought conditions arise across multiple stages of the production cycle. Savings, Credit, and Vulnerability As introduced in section 6.0 the role of savings in agricultural development is an important aspect of inclusive financial policies. Interest-earning savings accounts can increase household income, provide security against loss or robbery, and provide financial collateral for obtaining credit. This is particularly important in savings associations or self-help groups which can receive bank or MFI funds for group lending activities if a threshold of savings is met. The dynamic for individual borrowers is different from that of self-help groups since accumulated non-precautionary savings can substitute for credit. For example, Ankrah Twumasi et al. (2019a) investigate the Birim central municipality in Ghana. In a small sample of 141 households with access to credit and 75 that did not, they use an IV-Probit model and find that savings mobilization has a positive significant impact on access to credit and the total amount of credit one can borrow as well. With an overarching goal of addressing issues of vulnerability, the institutional structures of inclusive financial policies rely primarily on savings mobilization and access to credit to reduce farm/non-farm disparities. The highly endogenous nexus of savings–credit–vulnerability is not so clear cut, yet it is important for understanding the efficacy of inclusive financial policies. Using 2014 Findex data for Kenya, Olson (2018) investigated the credit–savings– vulnerability paradigm for farm and non-farm groups.5 Because arguments can be made that credit–savings–vulnerability are self-endogenous, he used a 3SLS approach with each being treated as an endogenous variable. The reasoning is that there are certain causal arguments that need to be made and investigated. For example, does savings drive credit demand, or does credit demand drive savings? What is the propensity of vulnerable households to save or borrow? These relationships are provided in the three panels of Table 19.2. Although other control variables were used in the 3SLS regressions, only the endogenous variables and income quintile are reported, and where the borrower holds an account at a FI. Groupings include all credit, formal credit, and informal credit, as well as farmer and non-farmer. The endogenous relationships were relevant and valid, and robust to different specifications of IV models. The upper panel in Table 19.2 has credit as the dependent variable (savings and vulnerability endogenous). None of the variables was significant at the 5 percent level of significance or better. This is a surprising result because it suggests that neither savings nor vulnerability are unique drivers of either formal or informal credit, and this holds for farmers and non-farmers alike. Credit use appears to be independent of income quintile status, and perhaps more significantly, having a formal account (access) does not necessarily imply increased use of formal credit (usage). The middle panel in Table 19.2—with savings as the dependent variable—tells a different story. It was found that households that are more vulnerable are less likely to save, and this is significant for non-farmers who indicate that they borrow informally. The savings behavior of farmers who indicated the use of informal credit and non-farmers and farmers who indicated borrowing from a financial institution does not appear to be affected by their vulnerability

376  Handbook of microfinance, financial inclusion and development

Table 19.2  Savings, vulnerability, and credit All credit Non-farmer

Farmer

Informal credit

Formal credit

Non-farmer

Farmer

Non-farmer

Farmer

Credit use Savings

0.491

0.549

0.471

0.419

0.0643

0.374

Vulnerability

2.872

1.193

1.899

1.965

2.384

–0.514

Quintile

0.191

0.0209

0.137

0.02014

0.143

31

Account

0.275

0.191

0.244

0.545

289

–0.019

–0.58

–0.224

–1.98

0.621

0.59

–3.083

–0.00653

–0.0299

–0.00251

0.147

0.242

0.476

***

*

–0.612

–1.338

Savings Vulnerability Credit Use Quintile

–0.749

–0.588

–0.844

***

***

***

0.499

1,091

0.526

*

**

*

–0.669

–0.0207

–0.0755

** Account

0.178

** 0.0682

**

0.165 **

Vulnerability Credit Use

0.586

1.199

* Savings Quintile

1.086

*

**

–1.104

–1.752

–1.053

–1.433

***

***

***

***

–0.0846

–0.0282

–0.0869

–0.00822

*** Account

0.597

0.184 *

*** 0.272

0.16

* –0.492

–0.635 *

–0.0695

–0.00294

*** 0.167

0.134

0.276 *

status. However, a positive relationship between actual credit use and savings for all respondents and non-farmers using informal credit was found. This does not seem to be the case for farmers using informal credit or those using formal credit. It was also found that higher quintile non-farm households use less credit generally, and this is also reflected in use of informal credit. Informal credit by farmers and formal credit does not appear to be related to savings. In terms of access to a bank account, this appears to translate into greater savings for nonfarmers, but not for farmers. In other words, an access-to-usage linkage for savings appears only to hold for non-farm households. In the top panel, there was no statistical relationship between increasing vulnerability and credit use; however, in the lower panel, positive effects were found for non-farmers’ overall

Inclusive finance and agricultural development in Africa  377

credit use and non-farmers’ and farmers’ use of informal credit. This suggests that, for at least these groups, those that borrow tend to be more vulnerable. However, it was also found that Kenyan farmers who borrow formally tend to be less vulnerable. The dominant effect here is the role that informal lending plays. The combined results suggest that just because a household is vulnerable does not imply that they are more likely to borrow in the informal market, but there is an asymmetry in the sense that those who do borrow informally are more likely to be vulnerable. Both non-farm and farm are negative for formal credit (only farmer is significantly different from zero) which suggests a possible substitution of formal for informal credit for the more vulnerable group. This could be familial, savings groups, or moneylenders. Across all groupings, it was found that those who save are less likely to be vulnerable, and this is significant for all but the non-farm groups that tend to borrow from financial institutions. Across all groupings, a negative relationship was found between economic quintile and vulnerability, which relates income to vulnerability, but not in all instances. In fact, the results suggest that for the farmer group, vulnerability is independent of income. A meaningful relationship was not found between holding an account and vulnerability, although they are positive and significant in two instances.

CONCLUSIONS This chapter provided an overview of agricultural credit and financial inclusion in Africa, but with limited focus. The reality is that each country has unique financial structures and programs with varying degrees of access and usage. There is significant variability between West and East Africa and Northern and Southern Africa. This is evidenced by the heterogeneity in the common elements such as deposits, credit, mobile use, or insurance discussed herein. Financial inclusion in continental Africa has mixed impacts depending on the country and region. A significant number of micro-studies across Africa conclude that there is a relationship between agricultural credit and agricultural productivity, and with respect to insurance, outcomes look promising except, perhaps, for the very poor. The very poor need specifically targeted policies that could include subsidies. These observations are in line with the meta-review by Van Rooyen et  al. published in 2012 who found up to that point in time that specific elements of microfinance seem to work in specific contexts, but the complexity of poverty and the various types of interventions make broad generalizations difficult. Ultimately because of certain findings that microfinance can at times increase poverty, reduce levels of children’s education, and disempower women, they advise against the promotion of microfinance to meet the Millenium Development Goals. The present analysis is not positioned to reach such a conclusion. Indeed, if inclusive financial policies targeted at agriculture in continental Africa are viewed more broadly than microfinance as it is traditionally used, specific targeting of insurance, credit products, flexibility, regulatory policy and oversight, and the adoption and use of cellular and mobile technologies do appear to be inclusive, at least at the meso- or macro-level. Flexibility appears to be important and some sense of balance between farmers’ demand for credit and lenders’ willingness to supply. For example, in choice experiments run by Shee et al. (2021) in Machakos, Kenya, for linked credit, it was found that there were conflicting demand- and supply-side preferences for credit terms, collateral requirements, and loan use flexibility. For example, while longterm loans were preferred by farmers, they were not preferred by finance providers. Farmers preferred medium-term credit while suppliers preferred short-term credit; no collateral loans

378  Handbook of microfinance, financial inclusion and development

were preferred by farmers, whereas collateral was strongly preferred by suppliers; farmers preferred loans to be used for any purpose (fungibility) while suppliers preferred loans only for agricultural production purposes. Inclusive finance is therefore not an absolute that can be dictated in market economies, but a balance, or equilibrium, between borrowers and lenders with competing interests. What becomes clear is that the disparate economic and political characteristics of African countries, combined with natural resource endowments, topography, and climate zone make the notion of a one-size-fits-all financial policy all but impossible, at least for credit and insurance. Mobile and cellular technologies should avoid problems of space, climate, and topography, but the high adoption and usage of mobile technologies in Kenya relative to other Sub-Saharan counties is anomalous.

NOTES 1. 2. 3. 4.

5.

www​.themix​.org​/mix​-market, a microfinance data service operator. The numbers represent those tracked by MIX. There are many more MFIs across Africa outside of this sample of 113 FSPs. https://microdata​.worldbank​.org​/index​.php​/catalog​/global​-findex. AgDevCo, AlterFin, Global Partnerships, Incofin, Oiko Credit, Rabo Rural Fund, ResponseAbility, Root Capital, Shared Interest, SME Impact Fund, and Triodos. As an example of familial lending, risk coping, and income smoothing, Shee, Turvey, and Woodard (2015) report on a post-disaster usufruct loan in which camel herders with a surviving herd would loan camel cows to one who lost the herd. The “borrower” was obligated to return all cows to the owner, but were able to keep, raise, and sell bull calves. The arrangement was assumed to be reciprocal and contributed to the communal recovery of the herd and lost asset accumulation. The econometric work presented here is drawn from Olson (2018).

REFERENCES Abdallah, A. (2016), “Does credit market inefficiency affect technology adoption? Evidence from SubSaharan Africa”, Agricultural Finance Review, Vol. 76, No. 4, pp. 494–511. Abdallah, A., Ayamga, M. and Awuni, J. (2019), “Impact of agricultural credit on farm income under the Savanna and Transitional zones of Ghana”, Agricultural Finance Review, Vol. 79, No. 1, pp. 60–84. https://doi​.org​/10​.1108​/AFR​- 02​-2018​- 0009 Adams, Dale W. (1971), “Agricultural credit in Latin America: A critical review of external funding policy”, American Journal of Agricultural Economics, Vol. 53, No. 2, pp. 163–172. Adams, D.W., Graham, D.H. and Von Pischke, J.D. (1984), Undermining Rural Development with Cheap Credit. Westview Press, Boulder and London. Akudugu, M. (2016), “Agricultural productivity, credit and farm size nexus in Africa: A case study of Ghana”, Agricultural Finance Review, Vol. 76, No. 2, pp. 288–308. Ankrah Twumasi, M., Jiang, Y., Osei Danquah, F., Chandio, A. and Agbenyo, W. (2019a), “The role of savings mobilization on access to credit: A case study of smallholder farmers in Ghana”, Agricultural Finance Review, Vol. 80, No. 2, pp. 275–290. https://doi​.org​/10​.1108​/AFR​- 05​-2019​- 0055 Ankrah Twumasi, M., Jiang, Y. and Owusu Acheampong, M. (2019b), “Capital and credit constraints in the engagement of youth in Ghanaian agriculture”, Agricultural Finance Review, Vol. 80, No. 1, pp. 22–37. https://doi​.org​/10​.1108​/AFR​-11​-2018​- 0100 Asante-Addo, C., Mockshell, J., Zeller, M., Siddig, K. and Egyir, I. (2017), “Agricultural credit provision: What really determines farmers’ participation and credit rationing?”, Agricultural Finance Review, Vol. 77, No. 2, pp. 239–256. Banerjee, Abhijit, Duflo, Esther and Hornbeck, Richard. (2014), “Bundling health insurance and microfinance in India: There cannot be adverse selection if there is no demand”, American Economic Review, Vol. 104, No. 5, pp. 291–297. Barrett, C.B. and Carter, M.R. (2013), “The economics of poverty traps and persistent poverty: Empirical and policy implications”, The Journal of Development Studies, Vol. 49, No. 7, pp. 976–990.

Inclusive finance and agricultural development in Africa  379

Barrowclough, Michael, Mohammed, Fatima, Kibler, Michelle and Boerngen, Maria (2020), “Measuring usage of formal financial services as a proxy of financial inclusion: A case of agricultural households in Ghana”, Agricultural Finance Review, Vol. 80, No. 4, pp. 471–489. Beck, T. and de la Torre, A. (2007), “The basic analytics of access to financial services”, Financial Markets, Institutions and Instruments, Vol. 16, No. 2, pp. 79–117. Beck, T., Demirgüç-Kunt, A. and Peria, M.S.M. (2007), “Reaching out: Access to and use of banking services across countries”, Journal of Financial Economics, Vol. 85, No. 1, pp. 234–266. Beck, T., Ross, L. and Loayza, N. (2000), “Finance and the sources of growth”, Journal of Financial Economics, Vol. 58, pp. 261–300. Bell, C., Srinivansan, T. and Udry, C. (1997), “Rationing, spillover, and interlinking in credit markets: The case of rural Punjab”, Oxford Economic Papers, Vol. 49, No. 4, pp. 557–585. Bellemare, M. (2011), “As you sow, so shall you reap: The welfare impacts of contract farming”, World Development, Vol. 40, No. 7, pp. 1418–1434. Bencivenga, V.R. and Smith, B.D. (1992), “Deficits, inflation, and the banking system in developing countries: The optimal degree of financial repression”, Oxford Economic Papers, Vol. 44, No. 4, pp. 767–790. Bester, H. (1987), “The role of collateral in credit markets with imperfect information”, European Economic Review, Vol. 31, No. 4, pp. 887–899. Binswanger-Mkhize, H.P. (2012), “Is there too much hype about index-based agricultural insurance?”, Journal of Development Studies, Vol. 48, No. 2, pp. 187–200. Binswanger, H.P. and D.A. Sillers (1983), “Risk aversion and credit constraints in farmers’ decision‑making: A reinterpretation”, The Journal of Development Studies, Vol. 20, No. 1, pp. 5–21. Boucher, S.R., Carter, M.R. and Guirkinger, C. (2008), “Risk rationing and wealth effects in credit markets: Theory and implications for agricultural development”, American Journal of Agricultural Economics, Vol. 90, No. 2, pp. 409–423. Boucher, S.R., Guirkinger, C. and Trivelli, C. (2009), “Direct elicitation of credit constraints: Conceptual and practical issues with an application to Peruvian agriculture”, Economic Development and Cultural Change, Vol. 57, No. 4, pp. 609–640. Braverman, A. and Guasch, J.L. (1989), Rural Credit in Developing Countries. Policy, Planning and Research Working Papers, Agricultural Policies, WPS 219 June 1989. The World Bank, Washington, DC. Carter, M.R., Cheng, L. and Sarris, A. (2011), “The impact of interlinked index insurance and credit contracts on financial market deepening and small farm productivity”, in Annual Meeting of the American Applied Economics Association, Pittsburgh PA, July (pp. 24–26). Casaburi, L. and Willis, J. (2018), “Time versus state in insurance: Experimental evidence from contract farming in Kenya”, American Economic Review, Vol. 108, No. 12, pp. 3778–3813. Castellani, D. (2014), “Shocks and credit choice in Southern Ethiopia”, Agricultural Finance Review, Vol. 74, No. 1, pp. 87–114. Chakravarty, S.R. and Pal, R. (2013), “Financial inclusion in India: An axiomatic approach”, Journal of Policy Modeling, Vol. 35, No. 5, pp. 813–837. http://dx​.doi​.org​/10​.1016​/j​.jpolmod​.2012​.12​.007 Chantarat, S., Mude, A.G., Barrett, C.B. and Carter, M.R. (2013), “Designing index-based livestock insurance for managing asset risk in northern Kenya”, Journal of Risk and Insurance, Vol. 80, No. 1, pp. 205–237. Chantarat, S., Mude, A.G., Barrett, C.B. and Turvey, C.G. (2017), “Welfare impacts of index insurance in the presence of a poverty trap”, World Development, Vol. 94, pp. 119–138. Collier, B., Katchova, A.L. and Skees, J.R. (2011), “Loan portfolio performance and El Niño, an intervention analysis”, Agricultural Finance Review, Vol. 71, No. 1, pp. 98–119. Demirgüç-Kunt, A. and Klapper, L. (2013), “Measuring financial inclusion: Explaining variation in use of financial services across and within countries”, Brookings Papers on Economic Activity, Vol. 2013, No. 1, pp. 279–340. Demirgüç-Kunt, A., Beck, T. and Honohan, P. (2008), Finance for All? Policies and Pitfalls in Expanding Access. World Bank Research Report. World Bank, Washington, DC. Demirgüç-Kunt, A., Córdova, E.L., Pería, M.S.M. and Woodruff, C. (2011), “Remittances and banking sector breadth and depth: evidence from Mexico”, Journal of Development Economics, Vol. 95, No. 2, pp. 229–241.

380  Handbook of microfinance, financial inclusion and development

Demirgüç-Kunt, A., Klapper, Leora, Singer, Dorothe, Ansar, Saniya and Hess, Jake (2018), The Global Findex Database 2017: Measuring Financial Inclusion and the Fintech Revolution. World Bank, Washington, DC. Ref: AFG_2017_FINDEX_v02_M. Accessed at Ithaca NY on 2/24/2020. Dev, S.M. (2006), “Financial inclusion: Issues and challenges”, Economic and Political Weekly, Vol. 41, No. Oct 14–16, pp. 4310–4313. Esuola, A., Hoy, M., Islam, Z. and Turvey, C.G. (2007), “Evaluating the effects of asymmetric information in a model of crop insurance”, Agricultural Finance Review, Vol. 67, No. 2, p. 341. Eswaran, M. and Kotwal, A. (1990), “Implications of credit constraints for risk behaviour in less developed economies”, Oxford Economic Papers, Vol. 42, No. 2, pp. 473–482. Giné, X. and Yang, D. (2009), “Insurance, credit, and technology adoption: Field experimental evidence from Malawi”, Journal of Development Economics, Vol. 89, No. 1, pp. 1–11. Gonzales-Vega, C. (1982), “Cheap agricultural credit: Redistribution in reverse”, Discussion Paper 10, Colloquium on Rural Finance, September 1–3, 1981. Economic Development Institute, World Bank, Washington, DC (Revised January 11, 1982). Honohan, P. and Beck, T. (2007), Making Finance Work for Africa. The World Bank. Iddrisu, A., Ansah, I. and Nkegbe, P. (2018), “Effect of input credit on smallholder farmers’ output and income: Evidence from Northern Ghana”, Agricultural Finance Review, Vol. 78, No. 1, pp. 98–115. https://doi​.org​/10​.1108​/AFR​- 05​-2017​- 0032 IFAD (International Fund for Agricultural Development) (2016), Rural Development: Report 2016: Fostering Inclusive Rural Transformation. Rome. Accessed at Rural Development Report 2016: Fostering inclusive rural transformation (ifad​.o​rg). Jumpah, E., Osei-Asare, Y. and Tetteh, E. (2019), “Do farmer and credit specific characteristics matter in microfinance programmes’ participation? Evidence from smallholder farmers in Ada west and east districts”, Agricultural Finance Review, Vol. 79, No. 3, pp. 353–370. https://doi​.org​/10​.1108​/AFR​05​-2018​- 0044 Kanbur, R. (2009), “The co-evolution of the Washington consensus and the economic development discourse”, Macalester International, Vol. 24, No. 1, Article 8. Karlan, Dean and Zinman, Jonathan (2008), “Credit elasticities in less developed countries: Implications for microfinance”, American Economic Review, Vol. 98, No. 3, pp. 1040–1068. Karlan, D. and Zinman, J. (2011), “Microcredit in theory and practice: Using randomized credit scoring for impact evaluation”, Science, Vol. 332, No. 6035, pp. 1278–1284. Kuznets, S. (1955), “Economic growth and income inequality”, American Economic Review, Vol. 45, No. 1, pp. 1–28. Laureti, C. and Hamp, M. (2011), “Innovative flexible products in microfinance”, Savings and Development, Vol. 35, No. 1, pp. 97–129. Marr, A., Winkel, A., van Asseldonk, M., Lensink, R. and Bulte, E. (2016), “Adoption and impact of index-insurance and credit for smallholder farmers in developing countries: A systematic review”, Agricultural Finance Review, Vol. 76, No. 1, pp. 94–118. Martey, E., Wiredu, A., Etwire, P. and Kuwornu, J. (2019), “The impact of credit on the technical efficiency of maize-producing households in Northern Ghana”, Agricultural Finance Review, Vol. 79, No. 3, pp. 304–322. https://doi​.org​/10​.1108​/AFR​- 05​-2018​- 0041 Mckinnon, Ronald I. (1973), Money and Capital in Economic Development, Brookings Institution, Washington, DC. Meyer, R.L. (2015), “Financing agriculture and rural areas in sub-Saharan Africa: Progress, challenges and the way forward”, IED Working Paper, IED London, March 2015. Miranda, M.J. (1991), “Area-yield crop insurance reconsidered”, American Journal of Agricultural Economics, Vol. 73, No. 2, pp. 233–242. Miranda, M.J. and Gonzalez-Vega, C. (2011), “Systemic risk, index insurance, and optimal management of agricultural loan portfolios in developing countries”, American Journal of Agricultural Economics, Vol. 93, No. 2, pp. 399–406. Morduch, J. (1995), “Income smoothing and consumption smoothing”, The Journal of Economic Perspectives, Vol. 9, No. 3, 103–114. Ndegwa, M.K., Shee, A., Turvey, C.G. and You, L. (2020), “Uptake of insurance-embedded credit in presence of credit rationing: Evidence from a randomized controlled trial in Kenya”, Agricultural Finance Review, Vol. 80, No. 5, pp. 745–766.

Inclusive finance and agricultural development in Africa  381

Nkegbe, P. (2018), “Credit access and technical efficiency of smallholder farmers in Northern Ghana: Double bootstrap DEA approach”, Agricultural Finance Review, Vol. 78, No. 5, pp. 626–639. https:// doi​.org​/10​.1108​/AFR​- 03​-2018​- 0018 Nordjo, R. and Adjasi, C. (2019), “The impact of credit on productivity of smallholder farmers in Ghana”, Agricultural Finance Review, Vol. 80, No. 1, pp. 91–109. https://doi​.org​/10​.1108​/AFR​-10​2018​- 0096 Norton, M.T., Turvey, C. and Osgood, D. (2013), “Quantifying spatial basis risk for weather index insurance”, The Journal of Risk Finance, Vol. 14, No. 1, pp. 20–34. Olson, S. (2018), Credit Rationing in Kenyan Agricultural Households and Uptake of Risk Contingent Credit: Evidence from the Field. Unpublished MS Thesis, The Graduate School, Cornell University. https://ecommons​.cornell​.edu​/ handle​/1813​/59347 Ricketts, K.D., Turvey, C.G. and Gómez, M.I. (2014), “Value chain approaches to development”, Journal of Agribusiness in Developing and Emerging Economies, Vol. 4, No. 1, pp. 2–21. Roubini, N. and Sala-i-Martin, X. (1992), “Financial repression and economic growth”, Journal of Development Economics, Vol. 39, No. 1, pp. 5–30. Sackey, F. (2018), “Is there discrimination against the agricultural sector in the credit rationing behavior of commercial banks in Ghana?”, Agricultural Finance Review, Vol. 78, No. 3, pp. 348–363. Sarma, M. and Pais, J. (2011), “Financial inclusion and development”, Journal of International Development, Vol. 23, pp. 613–628. Sarwosri, A., Römer, U. and Musshoff, O. (2016), “Are African female farmers disadvantaged on the microfinance lending market?”, Agricultural Finance Review, Vol. 76, No. 4, pp. 477–493. Sekyi, S., Abu, B. and Nkegbe, P. (2017), “Farm credit access, credit constraint and productivity in Ghana: Empirical evidence from Northern Savannah ecological zone”, Agricultural Finance Review, Vol. 77, No. 4, pp. 446–462. Shaw, Edward S. (1973), Financial Deepening in Economic Development, Oxford University Press, New York. Shee, A. and Turvey, C.G. (2012), “Collateral-free lending with risk-contingent credit for agricultural development: Indemnifying loans against pulse crop price risk in India”, Agricultural Economics, Vol. 43, No. 5, pp. 561–574. Shee, A., Pervez, S. and Turvey, C.G. (2018), “Heterogeneous impacts of credit rationing on agricultural productivity: evidence from Kenya.” Available at AgEcon Search https://ageconsearch​.umn​.edu​/ record​/274224​/files​/Abstracts​_18​_05​_21​_15​_37​_ 56​_13_​_80​_41​_92​_149​_0​.pdf Shee, A., Turvey, C.G. and Woodard, J. (2015), “A field study for assessing risk-contingent credit for Kenyan pastoralists and dairy farmers”, Agricultural Finance Review, Vol. 75, No. 3, pp. 330–348. Shee, A., Turvey, C.G. and You, L. (2019), “Design and rating of risk-contingent credit for balancing business and financial risks for Kenyan farmers”, Applied Economics, Vol. 51, No. 50, pp. 5447–5465. Shee, A., Turvey, C.G. and Marr, A. (2021), “Heterogeneous demand and supply for an insurance‑linked credit product in Kenya: A stated choice experiment approach”, Journal of Agricultural Economics, Vol. 72, No. 1, pp. 244–267. Skees, J.R. and Barnett, B.J. (2006), “Enhancing microfinance using index-based risk-transfer products”, Agricultural Finance Review, Vol. 66, No. 2, p. 235. Tadesse, M. (2014), “Fertilizer adoption, credit access, and safety nets in rural Ethiopia”, Agricultural Finance Review, Vol. 74, No. 3, pp. 290–310. Townsend, R.M. and Ueda, K. (2006), “Financial deepening, inequality, and growth: A model-based quantitative evaluation”, The Review of Economic Studies, Vol. 73, No. 1, pp. 251–293. Turvey, C.G. (2001), “Weather derivatives for specific event risks in agriculture”, Applied Economic Perspectives and Policy, Vol. 23, No. 2, pp. 333–351. Turvey, C.G., Hoy, M. and Islam, Z. (2002), “The role of ex ante regulations in addressing problems of moral hazard in agricultural insurance”, Agricultural Finance Review, Vol. 62, pp. 103–116. Turvey, C.G., Shee, A. and Marr, A. (2019), “Addressing fractional dimensionality in the application of weather index insurance and climate risk financing in agricultural development: A dynamic triggering approach”, Weather, Climate, and Society, Vol. 11, No. 4, pp. 901–915. Van Rooyen, C., Stewart, R. and De Wet, T. (2012), “The impact of microfinance in sub-Saharan Africa: A systematic review of the evidence”, World Development, Vol. 40, No. 11, pp. 2249–2262.

382  Handbook of microfinance, financial inclusion and development

Verteramo-Chiu, L.J., Khantachavana, S.V. and Turvey, C.G. (2014), “Risk rationing and the demand for agricultural credit: Acomparative investigation of Mexico and China”, Agricultural Finance Review, Vol. 74, No. 2, pp. 248–270. Weber, R. and Musshoff, O. (2013), “Can flexible microfinance loans improve credit access for farmers?”, Agricultural Finance Review, Vol. 72, No. 3, pp. 416–435. Weber, R. and Musshoff, O. (2017), “Can flexible agricultural microfinance loans limit the repayment risk of low diversified farmers?”, Agricultural Economics, Vol. 48, No. 5, pp. 537–548. Williamson, J. (2000), “What should the world bank think about the Washington Consensus?”, The World Bank Research Observer, Vol. 15, No. 2, pp. 251–264. Williamson, J. (2004), “A short history of the Washington Consensus”, Paper presented at the conference ‘From the Washington Consensus to a new Global Governance’, September 24–25, Barcelona, Spain. Woodard, J.D. and Garcia, P. (2008), “Weather derivatives, spatial aggregation, and systemic risk: Implications for reinsurance hedging”, Journal of Agricultural and Resource Economics, Vol. 33, No. 1, pp. 34–51. Woodard, J.D., Shee, A. and Mude, A. (2016), “A spatial econometric approach to designing and rating scalable index insurance in the presence of missing data”, The Geneva Papers on Risk and InsuranceIssues and Practice, Vol. 41, No. 2, pp. 259–279.

20. Evaluating digital financial inclusion: a Kenyan perspective on morality and finance Susan Johnson and Silvia Storchi*

INTRODUCTION The digital financial inclusion revolution is gathering pace. In 2021 13 percent of those in developing countries had a mobile money account compared to 5 percent in 2017, and in Africa some 33 percent had a mobile money account in 2021, rising from 21 percent in 2017 (World Bank, 2021). Expectations of the potential contribution to the development of digital financial services are significant (UNSGSA et al., 2018; United Nations, 2018). Kenya has been the epicenter of this transformation. In the wake of its mobile money revolution, 81 percent of the adult population now have access to an e-wallet, and some 62 percent report having used it to save or keep money over the last 12 months (Central Bank of Kenya et al., 2021). Alongside this, much-cited impact research suggests that both consumption smoothing and poverty reduction impacts are evident (Jack and Suri, 2014; Suri and Jack, 2016). Moreover, short-term credit accessed via mobile money is also growing apace, seemingly overcoming the historic problem of formal loan access for those on low incomes. Of Kenya’s adult population, 9.5 percent report using mobile bank loans (such as M-Shwari), and 8.3 percent report borrowing from unregulated digital loan apps such as Branch and Tala (Central Bank of Kenya et al., 2019). Early enthusiasm suggested its importance for example for emergency loans: “the most effective friend [when] in need, it is the phone!” (Cook and McKay, 2015: 10). However, concern was raised regarding the negative listing of 3.8 million Kenyans in the Credit Bureaux—some 13 percent of the population (May 2018), of whom one million were defaulting on loans of under USD 10 (Wright, 2018). Although action has now been taken to ensure defaulters receive a 30-day warning of listing (Business Daily Africa, 2022), this would nevertheless seem a timely moment to question whether these metrics and the perspectives that underlie them are adequate in evaluating this phenomenon and the trajectory of financial inclusion it heralds. Papers in this field rarely have the word “morality” in their title. Ethical debate was stimulated in 2007 following the IPO offering of Compartamos in Mexico and the massive returns resulting from the investment of public and third-sector funds, giving rise to concerns about excessive interest rates charged to low-income clients (Hudon and Sandberg, 2013) and leading to increased emphasis on client protection. However, this has not led to a fundamental questioning of the underpinning values that guide the sector and how they may contrast with those held by users. While the rapid adoption of mobile money (MM) in Kenya—and other parts of East Africa (Demirgüç-Kunt et al., 2018)—has been understood as filling a gap in domestic remittances services with something safe, secure, convenient, and cheap, related research finds that services are valued for features that go far beyond product characteristics and suggests the phenomenon requires deeper investigation into the social networks and cultural repertoires that underpin its adoption. MM is a fundamentally networked phenomenon (Kusimba et al., 2015) 383

384  Handbook of microfinance, financial inclusion and development

and has been shown to build on the interpersonal relationships of exchange that are central to Kenyan culture (Johnson and Krijtenburg, 2018). MM’s success builds on cultural exchange repertoires which were found to be underpinned by values of reciprocity—which may be open-ended in relation to people and time—and “upliftment”, where people “boost” each other by supporting family and friends with resources, with such exchanges entailing values of interdependence and belonging. These were found to stand in stark contrast to the resourcefocused repertoire of the banking sector where savings accounts are a one-way exchange and rarely give rising to borrowing relationships, so not resonating with the nature of friendship and reciprocity. Only when loans are repeated and enable important achievements might they become part of this repertoire (Johnson and Krijtenburg, 2018: 588). This suggests the need for an evaluative approach to financial inclusion which goes beyond a focus on access, use, and poverty reduction to take account of the values that people hold in relation to how they manage their money. Economic anthropology has a long tradition of recognizing the meanings and relationships involved in money exchange and finance (Peebles, 2010; Maurer, 2006), offering perspectives that are rarely engaged in debates over financial inclusion policy. This chapter therefore proposes an alternative evaluative approach based on Amartya Sen’s well-known capability approach (CA) in order to propose the potential for evaluation based on an understanding of values. This perspective allows a shift away from conventional metrics toward a focus on what people value in the financial services that are available to them because they support their wellbeing; it then explores this perspective using research in Kenya. The next section presents the shift in theoretical perspective based on the capability approach. However, the CA does not offer a means to explore what people value other than a proposal that this be based on public reasoning. The chapter therefore adopts a relational approach to wellbeing that allows the exploration of the meaning behind what people value. It summarizes findings from qualitative research in two areas of Kenya on what it means to live a good life and draws out the values and wider moral framework underpinning this. It then investigates the financial practices that they also employ to support living well, which were predominantly interpersonal exchange and informal groups. The analysis shows how ensuring “money is always moving around” fulfills building relationships through money management within this moral framework. Before concluding, the penultimate section discusses how this perspective interacts with the promise and perils of the evolving digital financial ecosystem to raise concerns about the value framework within which it operates.

EVALUATING FINANCIAL INCLUSION USING SEN’S CAPABILITY APPROACH Sen’s CA (Sen, 1999) presented a paradigm shift to economics by proposing that interpersonal welfare comparisons should not focus on the bundles of commodities people possess or desire, but on the freedoms—or capabilities—people have to do and be the things they have reason to value. Thus, for example, a bicycle is valued for the capability of mobility that it offers. A commodity bundle that includes a bicycle does not offer valued mobility to a disabled person, so an evaluation needs to address the capability for mobility rather than the bundle of goods she has available to her. Sen uses the term capability set to refer to the combination of opportunities that people are able to achieve in a society and have reason to value. He argues that what is valued is not

Evaluating digital financial inclusion  385

based simply on each individual’s view but must be based on a process of public reasoning in a society. However, it is not easy to observe this set of capabilities because we can only observe what people actually do from the set of beings and doing—that is, what they are able to achieve. The actual achievements people reach are the observable outcomes that are chosen from the capability set—see Figure 20.1. This approach offers a rather different perspective on the term financial capability than the way it has been used in the financial services literature focusing on the individual’s skills and abilities to make effective financial decisions (Kempson et al., 2013). This formulation started with financial education and literacy as necessary in order to manage financial services effectively. This developed into financial capability and referred to a multitude of factors, such as knowledge, skills, attitudes, individual abilities, and behaviors, recognizing the social, cultural, and financial contexts in which people take their financial decisions (Kempson et al., 2013). However, in practice, it has focused on measuring and evaluating financial behaviors, such as day-to-day money management, budgeting, and planning for the future, without considering that such financial strategies can assume different meanings and forms based on the social and cultural context. The latest incarnation is the concept of financial health which is proposed as a step beyond financial inclusion to focus on whether people’s “daily financial systems help build resilience from shocks and create opportunities to pursue one’s dreams” (Rhyne et  al., 2017: 3). While starting to converge much more on the importance of how people are able to manage their money for their survival and reaching their goals, Sen’s perspective more substantively frames the capability set as the bundle of reachable financial opportunities that people have reason to value in light of their valued “beings” and “doings”. The focus of development then is the expansion of this capability set. Many studies show that low-income people have a range of ways to manage their money and assets (Collins et al., 2009; Zollmann, 2014)—some of which may be valued more than others. They choose from this set on the basis of their wellbeing goals, thus developing a certain set of financial practices—their financial functionings (Figure 20.2). The CA recognizes that individuals differ in their personal, social, and geographical characteristics and these form a set of so-called conversion factors which influence the ways in which people convert assets and resources into valued financial functionings. The availability of financial services, together with individual conversion factors, act as filters to define the individual’s financial capability set. The critical point about this perspective is that it shifts attention from the individual’s ability to take appropriate financial decisions and manage their financial lives to the (in)adequacy of the financial system to offer financial solutions that are in line with people’s values and wellbeing goals. In fact, the extent of opportunities that people have to manage their money (their financial capability set) is not only influenced by their level of knowledge and skills, but

Figure 20.1 Capabilities and functionings in Sen’s capability approach

386  Handbook of microfinance, financial inclusion and development

Supply/availability of financial services

(formal and informal)

Individual conversion factors Personal (e.g. age,

gender, education, financial literacy, attitudes, wealth/income) Social (e.g. social norms, cultural practices, moral values, gender roles, power dynamics)

Geographical

Financial capability Set of achievable opportunities for money and asset management that people have reason to value for their contribution to wellbeing

Choice

Financial functionings Observed financial strategies

Valued wellbeing outcomes

(e.g.climate, geographical location)

Figure 20.2 A wellbeing perspective on financial capability also by the social, political, and economic structures in which they live and by how the offer of financial services fits within this broader context. By looking at the individual within his or her context, rather than only the individual’s abilities and knowledge to take financial decisions, a new space for understanding the financial practices of poor people opens up. Indeed, only by understanding what people value pursuing in their life, can the appropriateness of their financial decisions be evaluated. The CA stresses the importance of people’s freedom to choose the type of life they have reason to value among different opportunities and considers choice as being intrinsically important for wellbeing (Sen, 1999). Because of this, Sen argues that policies should focus on the improvement of the capability set (Sen, 1988). Similarly, this suggests that the goal of financial inclusion is that of increasing people’s financial opportunities to pursue goals they value. This means that the value of financial inclusion does not exist in the services per se but in what those services allow people to pursue in terms of valued beings and doings. This approach therefore highlights the potential for a much wider range of valued goals that go beyond standard impact measurements on consumption, assets, or poverty reduction. This approach may lead to a consideration of other important aspects of people’s lives such as how services enable funds to be circulated within communities or how they support the development of meaningful relationships with others—even if not apparently profitable from an economic perspective. How money is managed may be valued for reasons other than immediate economic or financial gain, or the frequently cited needs of convenience, safety, security, and efficiency. It also suggests that there is no “optimal” set of financial behaviors that is the same for everyone across contexts. To summarize, applying the CA allows for a shift in the perspective on financial capability and inclusion in two ways. First, it permits a redefinition of financial capability as a set of financial and economic opportunities that people have reason to value in light of their conception of the good life, rather than a set of optimal skills, attitudes, and behaviors. This shifts the focus from the individual to the sector, moving away from the focus on individuals and their responsibility for economic and financial decisions. This suggests a new angle

Evaluating digital financial inclusion  387

on financial inclusion which evaluates how the financial sector is able to both meet people’s needs and reflect their values through services that can enhance people’s wellbeing, rather than through levels of adoption, use, mobile capacity, country commitments, and regulatory frameworks (Lewis et al., 2017). This does not imply that there is no role for financial education programs, but rather points out that financial decision-making processes are complex and need to be evaluated in relation to the cultural and social context in which they are taken. In the financial inclusion arena, there is a lot of discussion about adopting user-centered approaches and understanding clients’ needs in order to meet them better (Seltzer and McKay, 2014). However, this tends to focus on needs that are economic and financial, whereas shifting the focus onto goals people have reason to value allows the perspective to become more holistic and recognizes that what poor people regard as valuable is more than simply economic. In this way, any attempt to improve financial inclusion would be attuned to the social and cultural environment, and the financial sector would be more aligned with people’s values and goals. Second, it enables a shift away from a focus on income poverty reduction to a more holistic view of financial inclusion for living well. From the normative perspective of the CA, financial services may be valued for more than claims of poverty reduction and income growth. Hence its evaluation must move beyond these standard indicators to an emic—that is, people’s own perspectives—of what people consider important to live well and point to the importance of evaluating financial practices for their ability to promote wellbeing.

UNDERSTANDING AND RESEARCHING WELLBEING A range of studies in developing countries have explored what it means to live well and moved beyond quality-of-life and happiness studies (Gough et al., 2007; Clark, 2002). These seek to understand wellbeing in a holistic way. On the one hand, they move away from an understanding of wellbeing solely based on objective measures, based on a conventional assumption that what poor people need most “is self-evidently more income, security, basic needs, human rights, and so on” (Copestake, 2008: 218). On the other hand, these studies move away from subjective wellbeing approaches that use global measures of happiness and satisfaction that may not be meaningful and valid across cultures (Camfield and McGregor, 2005). Instead, these studies move toward a more substantive understanding of wellbeing, paying attention to understanding subjective wellbeing through the experiences and perspectives of local people. White (2006) argues for a cultural construction of wellbeing itself because culture is not just an influencing factor, separate from everyday life. Rather, it “structures material and relational desires through a cascade of associations that makes them meaningful and designates some as pressing” (White, 2006: 9). This is why, she notes, people do not only aspire to have a shelter but a specific kind of house. As White argues: “[T]he material and cultural are not separable, such that one can separate ‘objective reality’ from ‘cultural values’, but fundamentally intertwined” (2006: 10). Many studies find that relatedness is important for wellbeing, and as Sen says, “taking part in the life of the community and having self-respect” are both important capabilities (Sen, 1999: 75). However, most approaches to wellbeing, such as Ryff’s model of psychological wellbeing (Ryff, 1989), Ryan and Deci’s self-determination theory (STD) (Ryan

388  Handbook of microfinance, financial inclusion and development

and Deci, 2001) and Sen’s CA, consider relationships with others important only is as far as they contribute to individual wellbeing (White, 2017). Hoffmann and Metz (2017) confirm the importance of a deeply relational perspective in examining the African concept of ubuntu. Ubuntu is the isiNguni word for humanness or virtue and the central maxim is that “a person is a person through other persons” which can be understood as that developing personhood involves “priz[ing] communal relationships with them” (157). These relationships in turn involve thinking of “we” rather than “I”; celebrating the accomplishments of others; cooperating with them; mutual aid; engaging with them as a form of mutual identity; recognizing how others feel; and treating them with compassion and as having dignity. The extent of being other-regarding reflects the degree of humanness and hence excellence. White goes on to present a view in favor of a relational ontology of wellbeing that overcomes the myth of the autonomous individual and “regards relationality not as an external ‘social determinant’ or ‘social support’ (or constraint) to individual subjects, but as fundamentally constitutive of subjectivity” (2016: 129), that is, how people create and understand the meaning of living well. Wellbeing is conceived as created through the common and shared experiences of people living in relationships with others: “wellbeing is not seen as the property of individuals but as something that belongs to and emerges through relationships with others” (Christopher, 1999 in White, 2016: 29). Relationships are therefore central to the formation and understanding of wellbeing, as they are not only instrumental means through which wellbeing is achieved but also intrinsically constitutive of how people experience it (White, 2017). The relational wellbeing model (RWB), adopted here, when seen in terms of three dimensions, suggests that wellbeing emerges through the interaction between the “objective” of people’s situations and their “subjective” experiences and perceptions of them (Figure 20.3). The position of the subjective dimension at the top of the model signifies the importance of grounding wellbeing in cultural meanings, while recognizing its material and relational dimensions. The subjective is not only made up of individual perceptions; but people’s ideas, perceptions, and values are grounded within a certain culture, forming a lens through which every dimension of wellbeing assumes particular meanings and values (White, 2010). In this conceptual framework, all aspects of wellbeing are simultaneously material, relational, and subjective. As a result, this approach enables underlying dimensions of wellbeing to be

Subjecve

Material Source:  White, 2010

Figure 20.3 Dimensions of relational wellbeing

Relaonal

Evaluating digital financial inclusion  389

revealed, and this framework was adopted to enquire into what values underlie what it means to live well in Kenya.

RESEARCH METHODOLOGY Operationalising the RWB approach involved qualitative research in the form of two (or more) semi-structured interviews with each of 34 respondents in two counties in Kenya. These respondents were a purposefully drawn sub-sample from a wider research project choosing two of the financial market hubs centered on the towns of Nyamira and Kitui—which are in the bottom two terciles of Kenya’s district poverty rankings (Johnson et al., 2016). Purposeful sampling captured age (above/below 35), gender, and a range of financial service use and differential physical access to formal financial services with respondents being located in villages either very close or some distance from the towns. A first round of interviews focused on respondents’ life events, and their understanding and experiences of living a good life. The second round explored how respondents managed their money and how these strategies related to the aspects of a good life they had previously discussed. The interviews were recorded, transcribed, and then analyzed thematically to identify, first, descriptive themes regarding what was important for wellbeing, their relationship to the relational wellbeing framework, and how financial service use related to these. Thematic analysis operates within an interpretivist epistemology to reveal the underlying meanings revealed by respondents’ narratives which become clearer through iterative analysis. Thus qualitative analysis proceeds through highlighting themes that cohere and offer overall consistency across respondents’ narratives about the meaning of living a good life and the use of financial services in relation to it. While negative experiences were also encountered, what these show through an interpretive lens is not the negative experience in itself but what it also reveals about what is important for wellbeing.

LIVING WELL IN KENYA Interviews showed that material dimensions regarding food, housing, land, and education, for example, were central to respondents’ narratives of wellbeing. However, none of these narratives concerned the material conditions of the individual respondent alone but how they provided for, cared for, and related to others. For example, men are expected to provide food for their families, send their children to school, and improve their houses to signal their adulthood status. It is important for a man to be seen as a breadwinner and responsible father who takes care of his family. In order to develop and maintain these identities, a man “works hard” to support his family materially, while also gaining status and respect from other family members and men in the community. A responsible man does not drink and is truthful and God-fearing. Following local traditions such as paying bridewealth on marriage is also important for a man to be considered “man enough” and worthy of participating in the family’s decisions. As a male respondent reported, not having been able to pay bridewealth to the in-laws caused him stress and he wished to be able to do so in order to be included in family decisions and so that “the in-laws could have much respect for me”. Moreover, while a respondent might express the need for good housing in the form of a house built of burnt bricks, having a number of rooms and a corrugated iron roof, the journey to achieving such a house involved the meaning of and relationships involved in such an

390  Handbook of microfinance, financial inclusion and development

achievement. For a young man, constructing a house involves the meaning of becoming an adult. The house also represents the role of hard work in earning the money to buy corrugated iron and the contributions of friends and family to funding or directly building the house. Some young men talked of assisting their parents to build better houses and similarly, for elders whose children may have supported them to build, the house symbolizes these relationships. On the other hand, living in a poor house of mud and sticks for one respondent symbolized a lack of intergenerational support and broken relationships with her children; she hoped that her six grandchildren with whom she lived would later help her build a better house so that she could “later die in peace”. Hence the house is not simply about meeting the material need for shelter, but the type of house needed to live a good life depends on local culture and norms which also reflect valued family relationships. At the same time, it is symbolic of status within the community regarding being hardworking or achieving adulthood or elderhood. In this way, achievements and aspirations concerned the respondent in relation to other people, either family members, community members, or friends. These findings confirmed that wellbeing is not the property of individuals but rather emerges through relationships and is key to an understanding of the formulation of the “good life” and how people can experience it (White, 2016, 2017). Moreover, relationships were expressed and lived based on a local understanding of morality and identities. Wellbeing is experienced when respondents develop relationships in which both parties behave according to social expectations and moral values of care, love, generosity, mutual support, and honesty, and thus feel respected and accepted by others on the basis of their identities. Important identities for respondents are those of hardworking breadwinner and caring father, “good” daughter-in-law and mother, successful elderly parent and adult children, supportive friend, and respected community member. These identities are constructed according to life stages and gender norms as well as moral values that underpinned the conception of the “good life”. Further, analysis revealed a hierarchy of moral levels which fits with Appadurai’s (2004) analysis of aspirations. He identifies a set of immediate wants and choices but argues that these are representative of an intermediate set of norms and beliefs about what it means to live well. These in turn are set within an understanding of a deeper cosmological order. Respondents’ moral narratives of wanting to behave well and help others were underpinned by a belief in God and religious teachings. This higher moral understanding was expressed by one respondent through these words: “assisting someone is good, and giving is good [. . .] because even the teachings say that blessed is the hand that gives than the one which takes”. A good life was therefore conceived by respondents in terms of relational and moral life. Their narratives depicted a life that allows individuals to develop and nurture relationships— with people in the present, people who have died, future generations, and with God—and at the same time construct their social identities as driven by moral values. In fact, the identities reviewed here show that respondents presented themselves as moral subjects and that a good life was lived through relationships that operate in accordance with a set of moral values. Mutual support, and being God-fearing were strongly valued, as well as practicing respect and appreciation and being hardworking, honest, and generous. “Doing Good” refers to people’s perception of what it means to “act right” and this is based on a shared understanding of how the world should be and goes beyond individualistic views of living well (White, 2010). Such conceptions prevail even when relationships experience conflicts and may not appear to be guided by moral values.

Evaluating digital financial inclusion  391

FINANCIAL PRACTICES FOR LIVING WELL: “MY MONEY IS ALWAYS MOVING AROUND” These relationships and social identities—fundamentally characterized by a moral dimension—then guided the analysis of how respondents manage their money in order to explore the relationship between living a good life and financial practices. The practices most discussed by respondents in this respect were interpersonal resource exchange and informal financial groups—chamas. The findings showed that what might appear to be rather small-scale informal exchanges involving a few hundred Kenyan shillings (frequently less than KSh 500 or USD 5) acquired meaningful relational and subjective dimensions. It is through these exchanges that respondents constructed and confirmed their social identities. For example, the identity of the “good” friend who is supportive in times of need, or the “good” son who is appreciative and respectful toward his parents. The practice of seeking and constructing new friendships through exchange relationships was an additional confirmation of the intrinsic role of these practices for wellbeing. A young respondent commented that “a real friend is the one who helps you when you are in need” and another reported that she had tested a new friend by asking for material support: “I wanted to find out if she could help me when in need or [she] is just a talk kind of friend”. The ways in which money is used in relationships present it as relational at its core. Indeed, this research suggests that relationships cannot be developed and nurtured without the exchange of resources. For instance, a united family is symbolized by the support that circulates across the family: a sibling who receives may reciprocate in the future by giving a different amount to a different sibling. These relationships, as well as support in the form of charity, are a way of acknowledging, respecting, and appreciating others, as well as practicing love and generosity in accordance with religious teachings. Morality emerges through interpersonal material support. Informal financial groups are extensively used in Kenya—41 percent of the adult population are members (FSD Kenya and Central Bank of Kenya, 2016). ROSCAs and ASCAs have been extensively studied as social mechanisms which offer effective means for organizing savings (Gugerty, 2007; Anderson and Baland, 2002) and which offer strengths in flexibility due to close relationships but these in turn can also give rise to governance weaknesses (Johnson and Sharma, 2007). This analysis of chamas (a term that was also used by respondents to refer to small-scale local SACCOs) revealed that by participating in groups with family members, friends, and colleagues, respondents develop a strong sense of identification and belonging with the group. Thus, personal achievements achieved with group financial support are rarely seen in individual terms but are usually attributed to the group. For example, a cow purchased using group funds was identified with those relationships. In all cases reviewed, groups were perceived as a way to develop together, and not taking part in groups may be perceived to be selfish and individualistic. This was highlighted in the way a women’s group reported that “unity is power” and a respondent that “you cannot prosper on yourself alone”. Progressing together and experiencing a sense of acceptance and unity with other people were therefore seen as vital for living a good life. In fact, these moral dimensions of money management align with a higher cosmological order where God comes first and is the ultimate decision-maker and the respondents’ biggest supporter. Although not everyone might be a believer, many expressed the importance

392  Handbook of microfinance, financial inclusion and development

of being God-fearing for wellbeing, and both Christians and Muslims thanked God for their material and relational achievements and handed over their aspirations for future achievements to God’s will. This signals the aspiration to seek and maintain a connection with a higher cosmological order, something beyond themselves, and how financial practices of resource exchange and chamas are aligned with this cosmological order while also responding to short-term material needs. Thus, using money “well” is a way in which respondents nurture their relationship with God and with others. This includes giving money to the church or Islamic practices of giving alms on Fridays, helping the needy, and using money for community progression. This again confirms that living a good life was not an individual affair but a relational and collective experience. It is when everyone behaves well and uses money well—instead of in a selfish way—that people can construct and live according to their social identities, meeting social expectations and the moral requirements they entail, and being perceived by others as a moral subject (White, 2018). However, of course, some respondents are unable to maintain relationships of exchange when they do not have a sufficiently reliable income source or are spending most of their money on school fees. For example, Jacqueline had to sell her assets when her husband was ill, including the sewing machine that she used for her tailoring business. When she opened another shop she found that she was not able to ask for support from her friends in the same way because she was less able to reciprocate. She reported “When you have [money], people will consider you very important” and will ask you for support; on the other hand, when you are poorer “they consider you to be of less value [and] you may not help them so much”. Since being asked for support is a sign of being a respected member of the community, not being asked signals that one is perceived as unable to play this role and is experienced as a failure at a personal level—in Jacqueline’s case a somewhat painful revision to her status and her sense of loss of good friends. On the other hand, if someone fails to give when asked and they are perceived as having the means to assist, then it is seen understood as a lack of care and a breach of the relationship, and so seen as symbolizing the breakdown of the moral order with such individuals being seen as out of sync with morality. Considering financial practices in light of the pursuit of a good life shows that respondents create their wellbeing by keeping their money in circulation. While this movement is extremely clear in the case of chamas where money passes from one member to the next, the evidence on informal exchanges also clearly shows that respondents move their money through relationships in order to build new ones, and maintain and nourish existing ones. Indeed, the practice of supporting the education of another’s child “does not stop there and goes round and round” (see also Shipton, 2007). These exchanges are intrinsic to social relationships and they are a daily practice in the lives of respondents. Nevertheless, the findings show that this practice of keeping money circulating through relationships is part of a deeper understanding of how to create and nourish relationships, based on respect, appreciation, mutual support, honesty, and generosity. It is also about constructing social identities and living in accordance with them. The example of Jonathan—a 61-year-old father of ten—illustrates the significant role of a chama in supporting wellbeing in its relational and subjective as well as material dimensions, and also the operation of interpersonal exchanges. With the profits from the butchery he has run since 2003 in Mjini, a neighborhood of Kitui town, Jonathan supports his children and grandchildren, especially with school fees. Jonathan says that he tries to instill all his children

Evaluating digital financial inclusion  393

with “good morals” and respect for their father. He is proud of the unity and harmony among his children and he describes himself as the “president” of his family. Since the time he opened his butchery, Jonathan has been part of a chama that involves owners and employees of local butcheries as a merry-go-round (ROSCA). He is now the oldest member and chair of the group and does not have a bank account. The members meet every day at the slaughterhouse which they visit very early in the morning to purchase meat. Each day, each member contributes KSh 100 (USD 1 at the time of the research) and the pot is taken by one of them. Jonathan decided in 2003 to start contributing in the name of other family members such that he has 20 names in the group as a way to show symbolic support for his family and so contributes KSh 2,000 daily. Since there are a total of 50 names in the group, he receives the group payout 20 times during every savings cycle of 50 days. While contributing KSh 2,000 daily can be a challenge, Jonathan reports that putting his money in the chama works for him and enables him to manage his money better. He reports that he would rather not eat for a day than fail to pay his contribution and marks the pay-out days on his calendar in order to plan his expenses accordingly. This resonates with findings from research that suggests that managing money within a group creates a sense of accountability about how money is spent (Storchi and Rasulova, 2017). Through the chama he has established a good relationship with his goat supplier: when his business does not do very well he can get goats on credit to repay later in the day or when he receives his payout. However, there are times when the supplier also gives him loans in order to contribute to the chama: he says “this money from the group is for paying debt if I had borrowed money”. In this way, Jonathan has created a system in which his money “is always moving around” and “always busy” and is understood to be working for him. He keeps his money “circulating, from the goats to the chama where it is collected in a lump sum and to school or whatever need”. Other studies on savings groups in Kenya and South Africa also demonstrate the importance respondents attach to “making money work” for their needs and that they perceive group-based money management as an effective way to do so (Storchi, 2018; Storchi and Rasulova, 2017). The chama also supports him in playing his role as a head of household and caring and respected father of an extended family. He has used the chama to educate his children and he is now also supporting his grandchildren’s education. He has used payouts to buy water tanks for his farm and says he is always able to feed his family. Whenever his children ask for his financial assistance, he plans how to help them based on his chama payout schedule. In this way, Jonathan provides material as well as symbolic support to his family and ensures that his chama membership reflects his family responsibilities and his commitment towards all his children. He was satisfied that he had been able to continue supporting all his children even though two of his wives have died. Managing his money through the relational dynamics of the group not only allows him to plan for and regulate expenditure but the money is seen as working for everyone and contributing to communal “upliftment” and development. By keeping it in circulation, the money is seen to be benefitting not only its original owner, but other group members and their families. This creates a bond among members, such that Jonathan is able to say of his members: “They have become my friends”. He feels he can relate to the other members “on a personal level”, beyond the fact that they are part of a group. Because of this bond, they “always help each other”, again showing the connection between friendship and mutual support and the importance for respondents of managing money through relationships.

394  Handbook of microfinance, financial inclusion and development

On the other hand, money put into a bank account was understood by respondents as not directly working for them because it does not help them to maintain and build relationships and construct their social identities (see also Johnson and Krijtenburg, 2018). This is because the majority of respondents using a bank account were not able to create a relationship of support with the bank, according to their own value systems of reciprocity, appreciation, respect, and acknowledgment, primarily because they were not usually able to take loans. As a consequence, the majority of respondents, and even those who did secure loans, did not develop an identification or sense of belonging with the bank—something that respondents nonetheless developed towards their chamas and SACCOs. In addition, participation in chamas and help for other people in the community were seen as a form of collective development which was never evident when respondents described their relationships with the bank. Instead, they saw the bank as the primary beneficiary in their relationship, thus confirming a mismatch with the reciprocity of the “relational repertoire” (Johnson and Krijtenburg, 2018) and the underlying moral framework. Indeed, money in the bank is termed “sitting down” and outside of respondents’ social relationships. This means that managing money through a bank account does not directly support people in developing and maintaining their social relationships, while simultaneously constructing their social identities and moral standing. For example, Thomas was a mechanic and car wash owner who—having already adopted two nieces some years earlier as a result of a sister’s death—had recently taken on supporting the education of his nephew after the death of a brother-in-law. For him, leaving money he had saved in the bank when he could support a nephew’s schooling did not feel appropriate. He closed his bank account when it refused him a loan and joined the local traders’ SACCO. He felt it would be hard to justify to himself and his family his role as an uncle if he did not offer this support. Money saved in a bank account, associated with values of individual accumulation, does not therefore conform to values of mutual support and generosity that are the foundation of how and why respondents circulate their money. Putting money in the bank does not support respondents in the creation of their wellbeing by building relationships, or constructing identities and a sense of belonging. Thus, financial practices are an essential element in achieving wellbeing goals, and informal financial practices particularly so. While using a bank account can make someone proud to be able to save and increase their social status, bank services are often not respondents’ first option for managing their money, even when they have access to them. Rather, respondents keep their money in the bank only when such money is regarded as a surplus and after they have already taken care of their relationships—or indeed have ensured the right balance of these obligations. This evidence suggests that respondents search for wellbeing through the creation and maintenance of their relationships and identities, by conforming to the values of mutual support, generosity, honesty, appreciation and acknowledgment, hard work, and respect, and they do this through informal ways of managing money. Thus interpersonal resource exchange and chama membership are intrinsic to respondents’ social life and their search for inclusion in meaningful relationships: it is by circulating money that they circulate wellbeing. While it is certainly the case that interviews framed around living well brought out more idealistic perspectives in contrast to more negative experiences—it clearly offers an insight into a morality that is normally hidden. While the ideal may not be continuously achievable, an understanding of these underlying motivations and values suggests a starkly different set of concerns regarding how respondents are interacting with the financial services offered.

Evaluating digital financial inclusion  395

In particular, this research has built on previous research highlighting the relational dimension of financial practices and financial service use (Johnson and Krijtenburg, 2018; Kusimba et al., 2016). This highlights the importance of the way funds are used to support relationships and recognizes the role of collective development which is not achieved in isolation from the endeavors and support of family, friends, and local community. The underpinning moral order aligns these values in ways that present respondents with the need to negotiate across them. In this context, the prospect of individualistic financial gain must always be understood in this light. Such an understanding challenges the claims of economists who argue that the demands of social connections undermine the ability to save and take up economic opportunities (Karlan et al., 2014). Jakiela and Ozier, for example, ask “Does Africa need a rotten kin theorem?” regarding social pressure from relatives to share income (Jakiela and Ozier, 2016). Such economistic preferences for individual optimizing decisions to maximize savings, investment, and returns operate with an assumed welfare function that is in tension with this alternative moral and relational framework regarding what it means to live well. What is being optimized here is quite different. This is not to say that there are no tensions or conflicts in how funds are managed and used intra-family or intra-household. But as a respondent in related research indicated “Even if you are asked, you send it willingly, you are not forced [laughing] . . . it is not a must that you should have to send . . . you send because you have analysed the problem and you know he deserves to be helped” (Johnson and Krijtenburg, 2018: 586). Moreover, this evidence regarding the value framework in Kenya is also highly consistent with the concept of ubuntu outlined above, and in stark contrast to ontologically individualistic perspectives.

THE PROMISE AND PERILS OF DIGITAL FINANCE How then do insights from Sen’s theoretical framework, and this evidence of the value frameworks regarding the relational and moral dimensions of wellbeing, relate to the emergence of the digital financial landscape? Does digital credit indeed offer a “friend in the pocket” who conforms to the values and moral norms described above, or does a negative listing in the credit agency quickly disabuse regarding the nature of such a “friendship”? What are the implications for evaluating digital finance through a framework where the role of services is in supporting valued beings and doings and in which values matter? As indicated above, the aspirations for digital finance are considerable. With respect to lending, digital trails appear an obvious extension—or alternative—to the use of credit bureaus using past credit performance to reduce information asymmetry and determine access. Digital apps seek to exploit the proliferating data available to make this determination. Hence, M-Shwari—Kenya’s first mobile lending product arising from a partnership between Safaricom and the Commercial Bank of Africa (CBA)—uses data on calls, data use, airtime purchases, mobile money use, as well as deposits and prior borrowing experience with the product to determine loan limits. Branch—a loan app based in California—requires users to sign up via a Facebook account. The ways in which such apps invasively exploit users’ contacts in order to track people’s social networks are now becoming better understood by the public. In China, private credit rating systems such as Sesame Credit, built on the back of Alibaba’s payment system and related shopping websites, already offer credit rating systems that have been used to assess partners in marriage negotiations (Botsman, 2014). While Sesame Credit operates on an opt-in basis, the government’s intention is to build a social credit

396  Handbook of microfinance, financial inclusion and development

system in which all businesses’ and citizens’ records are available and which will likely draw on these private systems. The system will monitor behavior that is scored as good or bad in a whole range of behaviors—from playing video games to not paying fines or jaywalking. Consequences of blacklisting as a result of, for example, a court decision for being a journalist on corruption are already evident—such as not being able to buy a plane ticket (Kobie, 2019). Even in the context of Western legal systems, understanding of how algorithms underpinning digital systems influence outcomes is now better understood and raising concerns. O’Neil points out how algorithms entrench the past in the ways that they use big data and that they are “opaque, unregulated and incontestable, even when they are wrong” (O’Neil, 2016). In her analysis of algorithms—whether for advertising or for public policy to assess teacher performance or improve policing—she concludes that they are understood by only a few people who design them and embody values that entrench their normative values and perspectives. Moreover, contesting their results is extremely difficult; there is little in the way of an appeals process when such an algorithm draws conclusions about an individual’s work performance or credit status, which can have serious consequences for their lives. Big data processes codify the past. They do not invent the future. Doing that requires moral imagination, and that’s something only humans can provide. We have to explicitly embed better values into our algorithms, creating Big Data models that follow our ethical lead. Sometimes that will mean putting fairness ahead of profit. (O’Neil, 2016: 204)

The role of algorithms in marketing and so-called “mass predictive personalization” endeavors to predict what we will want to do or buy and put it in front of us. Zubhoff (2019) uses the term surveillance capitalism to capture the dynamics of this new order. Her argument is not simply that Google moved from a search engine that just happened to collect and store user search data and then discovered how to exploit this for marketing by seeking to predict behavior. More than this, the cognitive behavioralist understanding that has developed over the last 15 years is now being used to exploit cognitive weaknesses rather than overcome them. Nudge tactics adopted by public policymakers sought to engage the unconscious “thinking fast” reactions highlighted by Kahneman to enhance the potential to persuade citizens of what is good for them by crafting messages regarding eating healthily, saving, and so on (Kahneman, 2012; Thaler and Sunstein, 2009). By contrast, it is precisely this understanding of reactive responses that was used by the technology companies to spawn developments such as Facebook’s “like” button and notification feeds in order to increase users’ engagement with the technology. Going further, Zubhoff argues that the tech giants deliberately exploit reactive “thinking fast” cognitive weaknesses to undermine any “thinking slow” rational response about how our attention is best allocated. Techniques for “‘tuning’, ‘herding’ and conditioning the behavior of individuals, groups and populations” (Zubhoff, 2019: 17) create the very behavior that the algorithms predict, so ensuring algorithmic products are of the greatest commercial value. Zubhoff argues that these companies have deliberately sought to be as secretive as possible about the ever-widening array of data being collected through their apps and metadata on device use, keep the public in the dark about how data is collected, stored, and used, and offer no exit from their accumulation of these surveillance assets. Yeung (2018) raises a number of fears arising from mass personalization that go further to their societal impacts. First, is the distributive injustice that it entails in the market as consumers are divided into high and low value segments. While, of course, this is already retail practice, the difference in the online world is that personalization makes this invisible

Evaluating digital financial inclusion  397

and consumers are unable to access the offers available to other customers that, in the offline world, are accessible simply by walking into a shop. Second, she raises concerns related to the loss of solidarity and community which can arise as a result of personalization exacerbating the “narcissistic mindset” which makes difference and particularity the dominant logic, rather than generality. In particular, she highlights that financial and economic logics are prioritized over “values associated with community, particularly the value of equality, in ways that may weaken and marginalize our commitment to equality as both a matter of principle and social practice” (Yeung, 2018: 38; see also Williams, 2018). Such concerns may seem a long way from the new digital apps being developed to improve credit availability to low-income people in Kenya and their potential to undermine the relationships and moral values highlighted above. While concerns about market segmentation and exclusion have been central to financial inclusion, these concerns raise this further in terms of the potential for distributive injustice. More profoundly, the evidence above presents a stark contrast to the conventional narrow perspective of financial inclusion policy regarding income poverty reduction and economic growth. It demonstrates that people’s financial practices contain an underpinning moral framework that values a particular way of doing finance through a set of conceptualizations about how moving money around builds connections between people and seeks their wellbeing through the outworking of the identities and values that this makes possible. M-Pesa is an example of a financial technology that was successful because it built on the underpinning moral logic and engaged the relational repertoire of these practices. As its capacities are deployed in further innovations, they have the scope to erode and undermine these moral and relational underpinnings of living well. The evidence suggests that there is an inherent social inclusionary logic in informal financial practices—of course, some people are excluded when they are less able to reciprocate; at this point, charity kicks in and this is not necessarily sufficient to meet needs. The point to be made is that financial sector development can seek to uphold and develop indigenous logics or pursue mechanisms that fragment and undermine them. Indeed, the (offline) success of Kenya’s Equity Bank in expanding the proportion of banked people can be seen as an example of sensitivity to this deeper logic.1 Apart from changing the structure of bank account charging (Stone et al., 2010) and inclusion also being driven by a phase of formal economy expansion in the mid to late 2000s, it also keyed into the underlying value framework. First, when Equity developed its business in Central Kenya this was in the context of the poor economic performance of the late 1990s and a time when governmentowned banks were hardly lending to ordinary people.2 It focused on ensuring that money was lent out as well as deposited. In this way, it presented itself as a bank with a reciprocal proposition for savers and keyed into this underlying value. Second, in giving loans it identified itself as investing in local communities—if one stood in the queue at an Equity branch, one would be shown a video of all the projects in which Equity was investing, so identifying its lending with the value of upliftment and enabling customers to see where deposits were going. Third, its CEO—James Mwangi—took a lead role in the development of Vision 2030, the national development strategy, thus identifying Equity with the value of community development at the national level. Further, its “I’m a member” campaign in the late 2000s demonstrated a concern with identity and belonging. Finally, the Equity Foundation has supported education scholarships for disadvantaged children (Wings to Fly) in every county, an initiative that is widely known and appreciated and keys into the underlying value of mutual support and not least leading low-income people to open bank accounts to seek eligibility. This perspective

398  Handbook of microfinance, financial inclusion and development

takes us beyond the narrow characteristics of the products and services to the deeper way in which Equity identified with underlying value frameworks as a means of explaining its rapid expansion. Even if the level of exploitation suggested by fears of digitization as yet appears some way off for Africa, the point of the discussion above is to demonstrate that financial practices have relational and moral logics and the digital revolution can seek to support or undermine them. There are ways to design financial practices that engage with or support these moral logics. For example, one app called M-Changa (changa means collect or gather in Swahili) builds on the underlying concept of harambee (let’s pull together) and collective development by providing a crowdfunding platform for this practice and taking a fee for its service. It uses an app to add value to the administration and functioning of a practice without seeking to exploit the data for additional personalization value. This differs from the intentions of an app such as MaTontine which seeks to digitize the transactions of ROSCAs in Senegal in order to lend to individuals and gain repayment when they receive their payout, thus seeking to build credit histories in ways that will enable them to offer larger loans (Murthy et al., 2019). While this indeed may have benefits of providing short-term liquidity when it is needed, it is not clear how the model seeks to maintain the collective value of the Tontine as a socio-cultural tool whose benefits of sociality are intertwined with the financial services involved. In Kenya, it is not only the sociality of the ROSCA that provides flexibility since people can negotiate their needs within the group to receive the payout at a different time or share the liquidity (Johnson, 2004) but also—as shown above—the inextricable way in which money and relationships are intertwined. Seeking to move people into digital lending removes people from the relational repertoire to the resource-focused repertoire of exchange (Johnson and Krijtenburg, 2018) and neglects the moral logics of shared development. Models of evaluation that do not recognize this shift in social relations have the long-run potential to undermine the dimensions of wellbeing that are most valued. In fact, the microfinance sector itself has historically been the terrain of similar developments. For example, the international MFI FINCA started out with a model that involved group-based lending with an internal fund with the vision that this would develop over time and make external borrowing from FINCA itself unnecessary, so creating autonomous village banks circulating their own resources. However, the logic of MFI sustainability led to the opposite happening—the internal fund was undermined because it was used by FINCA to ensure repayment, and users therefore no longer wished to contribute to it, thus undermining the potential for the development of local independent and self-governing entities and becoming dependent on the MFI—also at high-interest rates (Johnson, 2005).

CONCLUSION This chapter has proposed that Sen’s approach to evaluating wellbeing offers an alternative perspective on the evaluation of financial inclusion policy which requires establishing people’s valued beings and doings for living a good life. Evidence from research into what it means to live a good life in Kenya demonstrates that this has relational and moral dimensions which are integral to how material—including financial—resources are understood and used. Informal financial practices of interpersonal exchange and financial groups build relationships and enable valued identities to be constructed, while people strive to achieve behavior that fits a moral framework involving doing good and supporting intermediate norms such

Evaluating digital financial inclusion  399

as mutual support and honesty. This is not in any way to suggest that they always achieve this, rather this research reveals an underpinning value framework that plays a role in guiding action which highlights that relationships cannot be developed and nurtured without the exchange of resources. The implications of this are: first, from a practical point of view, it is financial services that fit with these deeper logics that are likely to have greater take-up and contribution to wellbeing—as mobile money and the practice of interpersonal exchange have shown. Second, a discussion of the value frameworks which underpin the logics of financial practices provides a basis for how financial inclusion policy is implemented and evaluated, and in the age of digital finance, this is an ever more pressing need if services are to contribute to wellbeing.

NOTES * 1. 2.

We are grateful to the Financial Sector Deepening Trust, Kenya, for support to undertake this research. We are particularly grateful to Severine Deneulin, Amrik Heyer, Sibel Kusimba, and Sarah White for feedback on earlier versions of these ideas. In 2013, 50 percent of the banked population had an account with Equity. The political economy of Equity’s rise in the late 1990s and early 2000s is an important part of this story as its heartland of Central Kenya was home to the political opposition to President Moi.

REFERENCES Anderson, S. & Baland, J.-M., 2002. The Economics of Roscas and Intrahousehold Resource Allocation. Quarterly Journal of Economics, CXVII(3), pp. 963–995. Appadurai, A., 2004. The Capacity to Aspire: Culture and the Terms of Recognition. In: V. Rao & M. Walton, eds. Culture and Public Action. Stanford: Stanford University Press, pp. 59–84. Botsman, R., 2014. Who Can You Trust? How Technology Brought Us Together and Why It Might Drive Us Apart. New York: Portfolio Penguin. Business Daily Africa, 2022. Digital Loan Defaulters to Get Notice before CRB Listing. Available at: www.businessdailyafrica.com/bd/economy/digital-loan-defaulters-to-get-notice-before-crb-listing3773934. Accessed 30 October 2022. Camfield, L. & McGregor, A., 2005. Resilience and Well-Being in Developing Countries. In: M. Ungar, ed. Handbook for Working With Children and Youth: Pathways to Resilience Across Cultures and Contexts. London: Sage. Central Bank of Kenya, Kenya National Bureau of Statistics & FSD Kenya, 2021. 2021 FinAccess Household Survey: Access, Usage, Quality, Impact Nairobi: Central Bank of Kenya, Kenya National Bureau of Statistics. Nairobi: FSD Kenya. Clark, D.A., 2002. Visions of Development: A Study of Human Values. Cheltenham: Edward Elgar. Collins, D., Morduch, J., Rutherford, S. & Ruthven, O., 2009. Portfolios of the Poor: How the World’s Poor Live on $2 a Day. Princeton and Oxford: Princeton University Press. Cook, T. & McKay, C., 2015. How M-Shwari Works: The Story So Far. Washington, DC: CGAP. Copestake, J.G., 2008. Wellbeing and Development in Peru: Local and Universal Views Confronted. New York: Palgrave Macmillan. Demirgüç-Kunt, A., Klapper, L., Singer, D., Ansar, S. & Hess, J., 2018. The Global Findex Database 2017: Measuring Financial Inclusion and the Fintech Revolution. Washington, DC: World Bank. FSD Kenya & Central Bank of Kenya, 2016. 2016 FinAccess Household Survey on Financial Inclusion. Nairobi: FSD Kenya and Central Bank of Kenya. Gough, I., McGregor, A. & Camfield, L., 2007. Introduction: Conceiving Wellbeing in Development Contexts. In: I. Gough & A. McGregor, eds. Researching Wellbeing. Cambridge: Cambridge University Press. Gugerty, M.K., 2007. You Can’t Save Alone: Commitment in Rotating Savings and Credit Associations in Kenya. Economic Development and Cultural Change, 55(2), pp. 251–282.

400  Handbook of microfinance, financial inclusion and development

Hoffmann, N. & Metz, T., 2017. What Can the Capabilities Approach Learn from an Ubuntu Ethic? A Relational Approach to Development Theory. World Development, 97, pp. 153–164. Hudon, M. & Sandberg, J., 2013. The Ethical Crisis in Microfinance: Issues, Findings, and Implications. Business Ethics Quarterly, 23(4), pp. 561–589. Jack, W. & Suri, T., 2014. Risk Sharing and Transactions Costs: Evidence from Kenya’s Mobile Money Revolution. American Economic Review, 104(1), pp. 183–223. Jakiela, P. & Ozier, O., 2016. Does Africa Need a Rotten Kin Theorem? Experimental Evidence from Village Economies. Review of Economic Studies, 83(1), pp. 231–268. Johnson, S., 2004. “Milking the Elephant”: Financial Markets as Real Markets in Kenya. Development and Change, 35(2), pp. 249–275. Johnson, S., 2005. Gender Relations, Empowerment and Microcredit: Moving on from a Lost Decade. European Journal of Development Research, 17(2), pp. 224–248. Johnson, S. & Krijtenburg, F., 2018. “Upliftment”, Friends and Finance: Everyday Exchange Repertoires and Mobile Money Transfer in Kenya. The Journal of Modern African Studies, 56(4), pp. 569–594. Johnson, S., Muruka, G., Genga, K., Ochieng, S., Adero, J. & Storchi, S., 2016. A Promise Fulfilled? Financial Market Development in Kenya 2011–2015. Nairobi: FSD Kenya. Johnson, S. & Sharma, N., 2007. “Institutionalizing suspicion”: The Management and Governance Challenge in User-Owned Microfinance Groups. In: T. Dichter & M. Harper, eds. What’s wrong with microfinance. Rugby: Intermediate Technology Publications Ltd. Kahneman, D., 2012. Thinking, Fast and Slow. London: Penguin. Karlan, D., Ratan, A.L. & Zinman, J., 2014. Savings by and for the Poor: A Research Review and Agenda. Review of Income and Wealth, 60(1), pp. 36–78. Kempson, E., Perotti, V. & Scott, K., 2013. Measuring Financial Capability: A New Instrument and Results from Low- and Middle-Income Countries. Washington, DC: World Bank. Kobie, N., 2019. The Complicated Truth about China’s Social Credit System. [Accessed 25/06/2019]. Kusimba, S.B., Yang, Y. & Chawla, N., 2016. Hearthholds of Mobile Money in Western Kenya. Economic Anthropology, 3(2), pp. 266–279. Kusimba, S.B., Yang, Y. & Chawla, N.V., 2015. Family Networks of Mobile Money in Kenya. Information Technologies & International Development, 11(3), pp. 1–21. Lewis, R.J., Villasenor, J.D. & West, D.M., 2017. The 2017 Brookings Financial and Digital Inclusion Project Report: Building a Secure and Inclusive Global Financial Ecosystem. Washington, DC: Centre for Technology and Innovation at Brookings. Maurer, B., 2006. The Anthropology of Money. Annual Review of Anthropology, 35, pp. 15–36. Murthy, G., Fernandez-Vidal, M., Faz, X. & Barreto, R., 2019. Fintechs and Financial Inclusion: Looking Past the Hype and Exploring Their Potential. Washington DC: CGAP Focus Note. O’Neil, C., 2016. Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. London: Allen Lane. Peebles, G., 2010. The Anthropology of Credit and Debt. Annual Review of Anthropology, 39, pp. 225–240. Rhyne, E., Kelly, S., Parker, S.R., Ladha, T. & Asrow, K., 2017. Beyond Financial Inclusion: Financial Health as a Global Framework. Boston: Centre for Financial Inclusion. Ryan, R.M. & Deci, E.L., 2001. On Happiness and Human Potentials: A Review of Research on Hedonic and Eudaimonic Well-Being. Annual Review of Psychology, 52(141), pp. 141–166. Ryff, C.D., 1989. Happiness Is Everything, or Is It? Explorations on the Meaning of Psychological WellBeing. Journal of Personality and Social Psychology, 57(6), pp. 1069–1081. Seltzer, Y. & McKay, C., 2014. Insights Into Action: What Human-Centered Design Means for Financial Inclusion. Washington, DC: CGAP. Sen, A., 1988.Freedom of Choice: Concept and Content. European Economic Review, 32(2), pp. 269–294. Sen, A., 1999. Development as Freedom. Oxford: Oxford University Press. Shipton, P., 2007. The Nature of Entrustment: Intimacy, Exchange and the Sacred in Africa. New Haven and London: Yale University Press. Stone, R., Johnson, S. & Hayes, J., 2010. Financial Sector Deepening Kenya Impact Assessment. Nairobi: FSD Kenya.

Evaluating digital financial inclusion  401

Storchi, S., 2018. Impact Evaluation of Savings Groups and Stokvels in South Africa. The Economic and Social Value of Group-Based Financial Inclusion. Johannesburg: FinMark Trust. Storchi, S. & Rasulova, S., 2017. Impact Evaluation of FSD Kenya’s Savings Groups Project: Final Report. Nairobi: Financial Sector Deepening Kenya. Suri, T. & Jack, W., 2016. The Long-Run Poverty and Gender Impacts of Mobile Money. Science, 354(6317), pp. 1288–1292. Thaler, R.H. & Sunstein, C.R., 2009. Nudge: Improving Decisions About Health, Wealth and Happiness. London: Penguin. United Nations, 2018. Financing for Development: Progress and Prospects 2018—Report of the InterAgency Task Force on Financing for Development. New York: United Nations. UNSGSA, Better than Cash Alliance, UNCDF & World Bank, 2018. Igniting SDG Progress Through Digital Financial Inclusion. www​.unsgsa​.org: UNSGSA, Better than Cash Alliance, UNCDF, The World Bank. White, S.C., 2006. The Cultural Construction of Wellbeing: Seeking Healing in Bangladesh. Working Paper No. 15. Centre for Development Studies, University of Bath. White, S.C., 2010. Analyzing Wellbeing? A Framework for Development Practice. Development in Practice, 20(2), pp. 158–172. White, S.C., 2016. Introduction: The Many Faces of Wellbeing. In S. White and C. Blackmore (Eds), Cultures of Wellbeing: Method, Place, Policy. Basingstoke: Palgrave Macmillan. White, S.C., 2017. Relational Wellbeing: Re-centring the Politics of Happiness, Policy and the Self. Policy & Politics, 45(2), pp. 121–136. White, S.C., 2018. Moralities of Wellbeing. Working Paper No. 58. Centre for Development Studies, University of Bath. Williams, J., 2018. Stand Out of Our Light: Freedom and Resistance in the Attention Economy. Cambridge: Cambridge University Press. World Bank, 2021. Global Findex. Washington, DC. Available at: www.worldbank.org/en/publication/ globalfindex. Accessed 30 October 2022. Wright, G., 2018. Technology’s Threat to MFIs – Digitise or Die. In: S. Mendelson, ed. European Microfinance Award 2018: Financial Inclusion through Technology – Digital Pathways in Financial Inclusion. Luxembourg: European Microfinance Platform. Yeung, K., 2018. Five Fears about Mass Predictive Personalization in an Age of Surveillance Capitalism. International Data Privacy Law, 8(3), pp. 258–269. Zollmann, J., 2014. Kenya Financial Diaries: Shilingi Kwa Shilingi – The Financial Lives of the Poor. Nairobi: FSD Kenya. Zuboff, S., 2019. Surveillance Capitalism and the Challenge of Collective Action. New Labor Forum (Sage Publications Inc.), 28(1), pp. 10–30.

21. Inclusive finance and inclusive rural transformation in China Calum G. Turvey

INTRODUCTION In transitioning and transformational economies such as China, what occurs in the rural, largely agricultural, region is of great economic significance. Central themes on the importance of credit in agriculture and economic development include (a) the importance of credit in agricultural production; (b) the role of institutions and credit policies on credit supply; and (c) the relationship between risk, collateral, production, and credit. From this literature, it is generally concluded that farmers can improve productivity and household income by using credit; however, collateral, moral hazard, and/or adverse selection can also result in the rationing of credit to farmers. In 2015, the China Banking Regulatory Commission opened an Office for Financial Inclusion. The following year, in 2016, the State Council of China issued “The Plan for Promoting the Development of Financial Inclusion (2016–2020)” with an ambitious goal that China’s financial inclusion would match or exceed that of upper-middle-level economies by 2020 (Liu et al., 2019). This chapter focuses on inclusive finance as it relates to agricultural credit in China by analyzing the interplay of demand and supply factors for rural finance, and in particular, greater inclusiveness in access to it. The chapter addresses three broad issues: 1. The growth in demand for financial services and how these are related to the broader processes of structural and rural transformation. 2. Understanding the demand for financial services in the processes of structural and rural transformations. 3. Understanding how innovations in rural finance contribute to making access to financial services and rural transformation more inclusive. To address these issues the next section explains what we mean by inclusive finance and inclusive rural transformation. The third section reviews past and present developments in China’s agricultural finance sector. The fourth section explores in greater detail the meaning of financial inclusion and how this relates to developments in China. The fifth section explores certain aspects of credit demand in China. The sixth section outlines a number of different credit-aligned structures that have evolved in China to deal with credit gaps, particularly in poorer, underserved credit markets and aligned credit relationships in value chain networks. The seventh section summarizes and concludes the chapter.

RURAL FINANCIAL INCLUSION Financial Inclusion Financial inclusion can be defined as the delivery of banking services at an affordable cost to the vast sections of disadvantaged and low-income groups (Dev, 2006), or a process that 402

Inclusive finance and inclusive rural transformation in China  403

ensures ease of access, availability, and usage of the formal financial system for all members of an economy (Sarma and Pais, 2011; Beck et al., 2000, 2007; Chakravarty and Pal, 2013) with reasonable-cost, fair, and safe financial services (credit, deposits, insurance, etc.). The concept of financial breadth deals with accessibility to financial services (Beck and de la Torre, 2007; Beck et al., 2000, 2007) or the level of financial services. It is the channel through which financial intermediaries can put capital into the national economy and is often measured by the number of branches and, separately, deposit accounts per capita (Demirgüç-Kunt et al., 2011).1 If, as in the Mckinnon-Shaw model, banks allocate credit not according to the expected productivity of the investment projects, but according to transaction costs and perceived risks of default (Fry, 1978), it is no wonder then that even with increased deregulation and relaxed capital controls, the high cost of servicing, combined with significant covariate risk, makes agricultural economies unattractive beneficiaries of spillover effects. In China, which has only recently evolved from a largely agrarian economy, rural-agricultural growth lags behind urban-industrial growth. As a matter of balance, inclusive finance policies are designed to balance growth across economic sectors and redistribute capital resources through policy interventions using the tax system, regulation, capital controls, guidance, and moral suasion. The difference between financial breadth and financial inclusion is in the focus on the individual borrower. As Beck et al. (2008) point out, increased access to credit does not imply increased use of credit. Financial inclusion, therefore, is aligned with both access and usage, thereby allowing individuals and firms to take advantage of business opportunities, invest in education, save for retirement, and insure against risks (Demirgüç-Kunt et al., 2008). Financial inclusion as a concept requires the merging of breadth and depth into a single paradigm suggesting that the two concepts are sufficiently entwined that one can be explained by the other. Put another way, financial depth is a necessary but not sufficient condition for financial inclusion, while financial inclusion is a sufficient but not necessary condition for financial depth. Inclusive Finance and Inclusive Rural Transformation Rural transformation needs to be considered in the overall structural transformation of an economy. Given the urban-industrial growth in China, policy is concerned with how the determinants of the rate, nature, and outcomes of rural transformation can be more inclusive. China, as with many transforming countries, has seen rapid economic growth, but underlying this growth is extraordinary inequality in the distribution of economic gains. A critical link to inclusive rural transformation is an inclusive financial system. Within the financial system, lending for rural development and enterprises, including agriculture, is crucial to the support of progressive rural transformation. In China, this has arisen through a rapid expansion of financial services and new-type financial institutions designed to expand rural and agricultural lending. Formal and informal financial institutions servicing agriculture are critical. Much has been written on the role of credit in China’s agricultural sector, and with few exceptions, the general finding is that access to credit improves household income and agricultural productivity2 and that credit constraints lead to suboptimal levels of investment in human, physical, and working capital, which may become a persistent cause of poverty and poverty traps.3 In direct responses to credit constraints, Feder et al. (1990) found that in China, a 17.82 yuan increase in liquidity from relaxed credit constraints would increase farm output by 201.08 yuan. More recently Kumar et al. (2013) find that in the presence of credit constraints, almost 90 percent

404  Handbook of microfinance, financial inclusion and development

of Chinese farmers would seek wage-employment substitution of farm labor for wage labor, 74 percent would have to reduce agricultural input use, 50 percent stated that credit constraints would cause a reduction in health care and education expenditures, and 21 percent stated that increased credit constraints would cause a reduction in food consumption.4 Examples such as these exemplify the role that credit plays in development by highlighting the effects that non-inclusive financial systems can have on China’s agricultural economy. Importantly, a lack of access to credit may not necessarily imply an unmet credit demand because of alternative informal systems such as friends and relatives, money lenders, pawn shops, and so on. Nor will lack of credit access affect growth if there are no productive investment opportunities for farmers to pursue. Past strategies of encouraging rural investment through supply-side initiatives (known in China as policy loans) might have done more harm than good by imposing investment in new technologies which recipients and markets were ill-prepared to use (Meyer, 2011). Nonetheless, access to credit can stimulate the production of commodities, encourage the use of inputs such as fertilizer and breeding stock, encourage investment in modern technologies and irrigation, and provide financial services to specific targets such as low-income households, cooperatives, or specific producer groups. China, for its part, has loosened the regulatory environment and expanded oversight that has led to the development of a host of financial institutions, including rural credit banks, village banks, agricultural banks, credit unions, cooperative banks, microfinance institutions, and related services, such as agency banking, postal savings, point-of-sale, full-function ATMs, and mobile and e-banking (Turvey et al., 2011). Financing agriculture involves not only extensive use of credit but also access to financial services such as savings accounts and insurance. To safeguard credit in rural areas, an inclusive financial system requires a flexible regulatory environment with oversight that does the following: protects borrowers and depositors; promotes financial education and literacy; is open to new types of financial institutions; and develops financial products to meet the timing and sequencing of cash flows in agriculture, while balancing business and financial risk system-wide (Turvey and Kong, 2009). For farmers to be transformative, they must borrow money if they are to purchase or improve/irrigate land for farming, adopt mechanized and labor savings technology, build out-buildings and storage facilities, etc. Many farmers find it necessary to borrow for current expenses in producing crops or livestock and/or consumption smoothing. However, some of the greatest problems facing farmers in China evolve from the difficulty of obtaining adequate credit, making proper use of it, and meeting financial obligations when they are due. At times there is a strong disconnect between how lenders view and understand agriculture and how borrowers view and understand lenders (Turvey et al., 2014; Kong et  al., 2014). Financial education is required by both parties and is a crucial part of inclusive financial policies.

THE HISTORICAL DEVELOPMENT OF AGRICULTURAL CREDIT IN CHINA China’s maturing finance sector is addressing the issue of financial inclusion which is the cornerstone of the 13th five-year plan (2016–2020) to eliminate poverty by 2020. While progress is surely being made in deepening rural financial markets and agricultural finance, this modern financial system is young—starting with the 1978 market reforms—relative to those

Inclusive finance and inclusive rural transformation in China  405

in other developed and developing countries. However, China has a significant history of progressive financial policies directed toward agriculture, and to understand agricultural finance in the present day, it is worthwhile to pause and review this 2,000-year history. Pre-Republican Dynastic Era The first record of dedicated financial programs for agriculture arose as early as the Western Han Dynasty (202 BC–8 AD) when leaders implemented several types of granaries which were used to store and accumulate surplus in years of plenty to be distributed during years of famine. Not only were granaries used to stabilize prices, but also to reduce credit risks to farmers who had to borrow between harvests. During the Northern Song Dynasty (906–1279 AD) emerged the Green Sprouts policy designed to increase access to credit for farmers. The government would loan wheat and millet to farmers during the planting and growing seasons, to be repaid in cash or grain following the harvest with 20–30 percent interest. Following the Green Sprouts policy, a variety of money-loan societies were formed. These operated much like ROSCAS (rotating savings and credit associations) in the present day. Simply put, a money-loan society was an informal microfinance group of individuals, who met for a predefined period in order to save and borrow together. The “foots” of the societies would deposit money for the benefit of the “head” of the society, with the depositor leaving funds in for the longest term receiving the highest return on money. The Republican Era, 1912–1948 In the 20th century, the first bank for agriculture—the Agricultural and Industrial Bank—was formed in 1915 with features that permitted the support of credit societies. However, the real push for credit societies did not emerge until the push by the China International Famine Relief Commission (CIFRC) following the 1921 drought and famine. The CIFRC pushed for a cooperative system of lending following the German Raiffeisen system. The first of these “rural credit societies” emerged around 1924, although there were several less formal attempts prior to that. By the late 1920s, cooperative credit societies became one of the building blocks of the Rural Reconstruction movement. This was formalized by the “Blueprint” for formal financial institutions by the government in 1929 which laid out the ideas for a Commission of Agricultural Finance and the formation of a Central Bank of Agriculture along with provincial and county-level banks. This included formal rules, regulations, and oversight as well as the formation of a cooperative treasury system to ensure that the supply and demand of funds could be met throughout China. Progressive agricultural finance policies generally survived the war years, although many credit societies in Japanese-held territory were forced to close. To deal with the war, the Kuomintang (KMT) formed the Farm Credit Bureau in June 1938, which was tasked with the procurement and distribution of agricultural commodities and related goods, as well as circulating agricultural capital. The financial side of the Farm Credit Bureau was short-lived as it conflicted with similar efforts by commercial banks. Nonetheless, as a government-sponsored enterprise (GSE), the Farmers Bank of China saw a decrease of only 17 branches from an original 87 through the first year of the war in 1937 and then recovered to 744 branches by 1945 (Fu and Turvey, 2018, p. 333). Cooperatives, which numbered 3,978 in 1932 (with 87 percent being credit cooperatives) peaked at 172,053 (with 44.2 percent being credit cooperatives) by 1945 (Fu and Turvey, 2018, p. 294).

406  Handbook of microfinance, financial inclusion and development

The Collective Era, 1950–1978 Much of the architecture of China’s efforts in agricultural finance was dismantled by the formation of the People’s Republic of China in 1949, although some remnants of the cooperative system remained. Many of the regional banks and their respective border currencies were rolled into the People’s Bank of China (PBC). In 1951 the PBC formed the Agricultural Bank of China (as the Agricultural Cooperative Bank), but it was dissolved in 1952. It was resurrected again in 1955 to issue loans for extremely poor households, finance irrigation and water conservancy, and build up a poverty cooperative fund. It was again dissolved during the Cultural Revolution and resurrected for a third time in 1963. Since then the ABC has played a pivotal role in delivering agricultural credit to farmers as well as making policy loans on behalf of the PBC. Because of land reforms and collectivization, the rural credit cooperatives, as developed during the 1930s, had little functionality in the initial stages of the 1950–1978 collective period. It was not until late 1953 that the Communist Party of China (CPC) started a push for agricultural production cooperatives, which in turn promoted the development of rural credit cooperatives (RCC). Loans were broadly classified into production loans (generally made to collectives) and farmer loans (generally made to farmers for improvements). By the end of 1955, 159,363 RCCs had been established. Neither the ABC nor RCCs had much flexibility in determining their own interest rates. For example, in order to increase grain output during the Great Leap Forward (1958–1965), annual interest rates of production loans were reduced from 6 percent to 4.8 percent and kept at 2.16 percent between 1965 and 1978 (Wang, 2019). Credit policies and financial development throughout the collective era are difficult to assess, largely because of the multiple periods of turmoil. This started in 1950 when land laws removed ownership rights of individual farmers, and hence, the incentives to make land improvements and optimize output. This was followed by the Great Leap Forward (1958– 1965) and consequential famine, and then the Cultural Revolution (1966–1971). Nonetheless, the structures in place in 1978, when the collective era ended and the household responsibility system began, laid the foundations for China’s credit system today. The Modern Era, 1978–Present The move toward a more market-oriented economy in 1978 heralded much growth in the agricultural sector. By 1984, agricultural incomes and wages increased by 73.2 percent while output from the rural industrial sector grew by 1188 percent (Shen et al., 2010). Growth was unsustainable given credit conditions of the day and the financial system had to quickly adapt to meet demand. The PBC developed the Bank of China, China Construction Bank, and Industrial and Commercial Bank of China. Much effort was directed toward financing township and village enterprises (TVE). The ABC took on a prominent role as a policy bank with an exclusive mandate to meet the needs of farmers and rural firms. As part of this mandate, the ABC was also directed to oversee the network of RCCs across China. However, Chinese banks could not operate as western banks did, because the issuance of land use rights (LUR) under the responsibility system removed the primary source of collateral required for loans. To compensate, banks and RCCs instituted a system of group guarantees (joint liability) amongst TVEs (Shen et al., 2010) and farmers. These are detailed in the following subsection. The ABC, and relatedly the RCCs, were also subject to governmental predation which included having to make loans based on policy rather than profits. To maintain profitability,

Inclusive finance and inclusive rural transformation in China  407

lenders established their own responsibility system providing front-line lenders bonuses if they made profitable loans while penalizing unprofitable loans. This system had the unintended consequence of increasing lender risk aversion and making lending to farmers more difficult (Cao et al., 2016). A major structural reform in 1996 saw the RCCs administratively untied from the ABC. In 1997, an on-lending window was opened which allowed RCCs to increase loanable funds above loan-to-deposit ratios. During the same time, RCCs within counties were amalgamated into an RCC union, or RCCU, that became the legal entity for the county. Experimentation also started in converting cooperatives to joint-liability-stock rural credit banks. In 2002 new policies allowed RCCs to set their own interest rates within upper and lower bounds to adapt to demand and supply conditions. With RCCUs in place, farmers within a county were restricted from borrowing from neighboring counties, so county RCCs could act as monopolists in some cases. Further reforms took place starting in 2006 with the New Countryside campaign. With the blessing of the PBC and the China Banking Regulatory Commission (CBRC), a suite of a new type of financial institutions was introduced. These included pilot microloan companies, guarantee companies, joint stock village or township banks, rural credit mutual help associations, rural mutual credit cooperatives, credit-only microcredit companies, and postal savings bank (Guo and Jia, 2010). The year 1994 also saw reforms that allowed the establishment of microfinance institutions (MFI). Initially, 17 Grameen-style NGO MFIs were established to make microloans of about 1,000 RMB (Park and Ren, 2001). In reality, Grameen-style MFIs did not emerge to any significant extent in China, largely because China’s vast network of RCCs also started issuing small microcredit loans in late 2001 under the guise of a social guarantee or trust-based lending. Indeed, an MFI in the Chinese context usually refers to a lending-only company that makes loans to TVEs and SMEs, and not poor farmers or individuals. NGO MFIs could not easily compete. Foreign NGOs were required to collaborate with the Chinese, their scale was generally small, and interest rates of about 12 percent were substantially higher than Chinese government poverty loans with interest rates of 2.88 percent. Furthermore, familial bonds provide a huge social network of friends and relatives that willingly lend to each other at zero interest rates. Turvey and Kong (2010; see also Turvey et al., 2010) surveyed 1,556 households in Shaanxi, Henan, and Gansu and found that only 48.7 percent held some form of informal or formal debt, and of the 757 households with debt, only 21.4 percent had formal debt, 53,5 percent had informal debt, and 24.6 percent had some combination of both. The empirical evidence shows a strong preference for familial lending over formal lending, and these informal relationships are part of the lending fabric in rural China, driven not only by convenience, but also trust, reciprocity, and altruism. Creditworthiness and Creditworthy Villages In the post-collective era, RCCs sought profitability by either limiting the supply of credit to agriculture or using non-price-rationing means. In 2001 the PBC issued guidelines to initiate microfinance through its web of RCCs. The loans, based on “social guarantee” and trust, removed collateral restrictions by designating specific farmers within a village as being “creditworthy.” In some cases, entire villages were deemed creditworthy. In the former case, a farmer would be deemed creditworthy after the RCC lender met with the village leader who would identify farmers who were generally trustworthy and non-alcoholic, did not gamble,

408  Handbook of microfinance, financial inclusion and development

and were not disabled. Creditworthy villages were usually tied to specific production types of value chains where farm households worked collectively on various parts and processes of a specific product for a TVE.5 Microcredit was offered to farm households regardless of whether the household had demand for credit and was issued as a line of credit through a debit card. The local RCC would determine the use of these funds. In some instances, the microcredit loans were restricted to agricultural use, while other RCCs permitted use for any purpose. Repayment terms were usually within one year, but it was not unusual for RCCs to extend the term by issuing subsequent loans to pay off the existing ones. Group Guarantees For farmers not deemed creditworthy, an alternative pathway to obtaining credit was based on the idea of the group guarantee. The understanding of group guarantees in the China microcredit landscape is not the same as group guarantees of the Grameen type. A group guarantee in China requires that a single borrower recruit between three and four friends or relatives to guarantee the loan on a joint basis. This requires the lender to do due diligence not only on the borrower but on the guarantors as well. While the approach avoids collateral requirements in most cases, it places an undue burden on both the borrower and the lender, increasing transaction costs considerably. Because the guarantors are voluntary, the likely motive is through kindness or reciprocity. From the lender’s perspective, it is far less costly to have group members monitor the loan and behavior of the borrower than the lender, who may be located at some distance from the village. Kong et al. (2015) investigated the group guarantee and found that, generally, farmers did not form group guarantees voluntarily but were required to do so by the lender. Older farmers were less likely to borrow under a group guarantee or voluntarily join a group guarantee as a guarantor; guarantors in a group guarantee are more likely to be risk-takers, or not too risk-averse, and farmers who mistrust lenders are less likely to join a group guarantee. They raised the possibility that the group-guarantee model may not be as beneficial as one would think. In addition to the costs of establishing the group, the group guarantee actually transfers collateral risk from the lender to the guarantors. For example, in the event of a catastrophic covariate event, it is the collateral of the jointly liable guarantors that is put at risk to satisfy the loan.

MEASURING FINANCIAL INCLUSION Economic measurement of financial inclusion can be a complex process. The first problem is in identifying the economic drivers of inclusion that capture depth, breadth, access, and usage and compiling them into a meaningful index for regression analysis. Following Sarma (2010), Xiong et al. (2013) used a linear dimensionless process with the measure for each dimension given by

di =

Ai - mi . (21.1) M i - mi

Here Ai is the actual value of each dimension; mi is the minimum value of each dimension; Mi is the maximum value of each dimension; and 0 ≤ di ≤ 1.0. Measuring dimensions according to equation (21.1) is the same as that used in other metrics such as the human development

Inclusive finance and inclusive rural transformation in China  409

index (HDI) (see, for example, Ravallion, 2012). Although Sarma (2010) argues that each dimension should be weighted, we followed the HDI protocol and assigned equal weights. The Index of Financial Inclusion is6

(1 - d ) + (1 - d ) 2



IFI = 1 -

1

2

2

n

+¼¼+ (1 - d n )

2

. (21.2)

Using a weight of 1.0 as the distance measure for each dimension is justified by the assurance that in doing so the IFI metric is confined between 0 and 1, with values approaching 0 indicating increasing financial exclusion and values approaching 1, increasing financial inclusion.7 Xiong et al. (2013) defined d1 as banking penetration; d2 as availability of the banking system; d3 as usage of the financial system; and d4 as the loan–deposit ratio. This latter measure was used to capture barriers to credit. For example, in China, the CBRC and the PBC often impose controls on the loan-to-deposit ratio to increase or decrease credit supply. In other instances, low loan-to-deposit ratios may signify excessive credit rationing or local (provincial or county) regulations or practices that discourage borrowing or lending. It can also potentially signify red-lining activities in which rural deposits are used or transferred to other financial institutions for non-farm or urban loans. In addition to the challenges of measuring financial inclusion, there is the challenge of determining the outcomes for financial inclusion. Because of endogeneity issues, Xiong et al. (2013) used a three-stage least squares (3SLS) panel model to investigate factors influencing and influenced by IFI. They treat the index of financial inclusion (IFI), non-performing loans (NPL), and the consumption-to-income ratio (Cons/Inc) as endogenous variables. A threeequation simultaneous equation model was used to account for endogeneity and to identify potential causal pathways. To avoid the identification problem, they used the proxy regional dummy variables to capture agricultural zones, approximately defined by China’s official agricultural map. The three models in Table 21.1 show good significance with R2 values ranging from 0.70 to 0.90. It was found that NPL (B = –0.254, p = 0.370) is negative and not statistically different from zero which indicates that nationally there is no statistical bias against provinces with greater credit risk. However, from model 2, it is notable that the effect of IFI on NPL is negative and significant (B = –0.198, p = 0.022). This suggests that improvements in financial inclusion may in fact lead to higher loan quality. The consumption–income ratio is also not significant in relation to IFI (B = 0.023, p = 0.842) suggesting that there is no bias against poorer or lower savings households. There is an asymmetric endogenous effect because model 3 IFI is highly and positively associated with increases in Cons/Inc (B = 0.221, p < 0.001). Positive consumption is a good outcome, especially for poorer households, but this also suggests a lower savings rate, which may not achieve certain policy goals. Furthermore, the Engel coefficient (food expenditures/total expenditures) in equation (21.3) is negative and significant (B = −0.412, p = 0.002), which suggests that as food expenditures rise relative to all household expenditures, consumption to income falls. This is not a surprising relationship when food expenditures substitute for nonfood expenditures. Income inequality, as measured by the income ratio (urban/rural) is negative in the IFI equation (B = –0.053, p = 0.047), which suggests that increased financial inclusion may not be targeted to higher-income urban centers. However, the negative Engel coefficient (B = –0.402,

410  Handbook of microfinance, financial inclusion and development

Table 21.1  Regression result of provincial-level data Model 1 R2 = 0.7009 P = 0.0000

Model 2 R2 = 0.8439 P = 0.0000

Model 3 R2 = 0.9015 P = 0.0000

IFI

NPL

Cons/Inc

Non-performing loans (NPL)

–0.254 (0.370)

IFI

–0.198 (0.022)

Ratio of consumption to income (Cons/Inc)

0.023 (0.842)

Cons/Inc –0.041 (0.782)

IncRat: the ratio of urban resident income to the rural

–0.053 (0.047)

IncRat

Engel coefficient (Eng)

–0.402 (0.044)

CityState –0.104 (0.011)

Employment (Emp): % of population employed

0.079 (0.001)

IFI

0.221 (0.000)

Eng –0.412 (0.002)

0.018 (0.302)

Medicare (Med): number of hospital and welfare –0.002 institutions beds divided by total population (0.000) Telecommunications (Tel): % of households owning telephones

–0.001 (0.016)

Urban (Urb): % of population that is urban

–0.104 (0.354)

CityState: = 1 if province is a city-state

–0.253 (0.000)

Constant

0.944 (0.000)

0.212 (0.075)

0.807 (0.000)

Ag region

Included

Excluded

Excluded

Province

Excluded

Included

Included

Year

Included

Included

Included

Fixed effects

Note:  Values in brackets are p-values Source:  Xiong and Turvey (2014)

p = 0.044) suggests that there may still be a bias against the poorest of farmers even though the focus is not entirely on higher-income urban centers. Other control variables include employment (B = 0.078, p = 0.001), which suggests that perhaps more industrialized regions are favored; Med (B = –0.0019, p < 0.001) indicates that more urban areas with greater medical facilities (#beds) are less favored; Tel (B = –0.0014, p = 0.16) indicates areas with greater telephone use are less favored; Urb (B = –0.104, p = 0.354) is not significant; and city-state (B = –0.253, p < 0.001), a dummy variable capturing Beijing, Shanghai, Chongqing, and Tianjin, indicates that final inclusion is not overly concentrated in the largest urban centers. These last four variables are encouraging. because the measure of financial inclusion, IFI, is measured relative to the rural population; it should not be surprising that these urban metrics

Inclusive finance and inclusive rural transformation in China  411

are negative. With a policy goal of making rural areas more inclusive, there does not appear to be an urban bias that centers on larger villages or towns located in rural areas. It is also notable that the coefficient for city-state in the NPL equation is negative (B = –0.104, p = 0.011), which indicates that rural loans issued through major city banks generally are of higher quality.8 Liu et al. (2019) explored whether the push toward inclusive rural transformation brought rural households into the economic mainstream and ensured that all individuals, particularly the rural poor, could exercise their economic and political rights, maximize their economic potential, and take advantage of the opportunities that came along with economic growth. China’s drive toward financial inclusion is to promote entrepreneurial activities, particularly in targeted poverty zones. Based on a sample of 988 households drawn from poverty and non-poverty villages and households that were entrepreneurial and non-entrepreneurial (traditional), they explored the relationships between entrepreneurial activity and financial inclusion. RCC policies of issuing “creditworthy” certificates to selected households without collateral or guarantees, provided a measure of credit access. Usage was measured by the extent that households draw on these credit lines. Of the 988 households, 47.57 percent had access to formal credit. Of these, only 57.45 percent used formal credit only; 21.49 percent used both formal and informal (familial) credit, and 18 percent used only informal credit. Using probit regression, propensity score matching, and endogenous switching models, these results are very much in line with other findings in Chinese academic literature. General results from these models indicate that simply having “access” to credit does not have any significance on entrepreneurship but “usage” does.9 Furthermore, results show only formal credit was significant in explaining entrepreneurship in households that borrow both formally and informally. In other words, formal and informal credit describe separate channels and only formal credit appears to be a driver of economic entrepreneurship. Nonetheless, there is strong evidence that inclusive financial policies causally benefit entrepreneurial activity, and entrepreneurial activity causally leads to an improvement in household income. Thus, deepening rural credit markets by increasing access to credit is sufficient but not necessary to promote entrepreneurial activity; promoting usage, however, is both necessary and sufficient. If the goals of inclusive finance are to reduce poverty in China, an important finding was that households in poverty-stricken villages were less likely to be entrepreneurial. In order for that to occur, greater efforts will be needed in financial education. Results from the PSM model confirm that properly implemented financial inclusive credit policies applied to poverty-stricken regions can have a substantial impact. Financial Education Financial education, as noted above, is a critical element of financial inclusion. Concerns about financial literacy were first raised by the Organization for Economic Cooperation and Development (OECD) in 2013. That same year the PBC, CBRC, China Securities Regulatory Commission, and China Insurance Regulatory Commission formulated the “National Strategy for China’s Financial Education” which included events such as the Jinhui Project, which involved a monthly series of financial popularization activities and investor education, etc., with a special interest in educating disadvantaged groups (Zhang and Xiong, 2020). A consequence of financial “illiteracy” is that it can lead to what Turvey, Xu, and Kong (2014) refer to as attitudinal asymmetries between lenders and borrowers, and a disconnect between how farmer borrowers perceive agricultural lenders and these lenders’ needs and how agricultural lenders perceive farmer borrowers and their needs. In their study, 394 farm

412  Handbook of microfinance, financial inclusion and development

households and 120 front-line lenders (from Shandong) were provided 83 attitudinal questions across a range of topic areas related to agricultural finance; it was found that on only 17 of these attitudinal questions did both borrower and lender agree. The results suggest that the onus of financial education should not simply be placed on the borrower to understand the lending process, but also on the lender to understand the borrower’s needs and understanding. In a related study, Kong et al. (2014) used psychographic clustering to identify four borrower clusters and two lender clusters. Two of the borrower clusters (N = 67) identified general dissatisfaction with the existing lender–borrower relationship, while the remaining two clusters (N = 165) were generally satisfied with the lender–borrower relationship. The two lender clusters were identified as “indifferent lenders” (N = 65) and “farmer-friendly lenders” (N = 55) and were segmented largely on their focal interests in making agricultural vs. nonagricultural loans. The upshot of this work is that financial literacy from the borrower side should be directed toward understanding not only interest rates but the lending process, while on the lender side, financial literacy should extend toward understanding the varied and flexible needs of farmer borrowers. These issues are part and parcel of the concerns raised by the OECD (2013) that despite efforts on financial education by many countries, the evidence did not necessarily lead to an increase in financial literacy. The academic literature is more positive when efforts are targeted to specific sub-groups (Walstad et al., 2010; Lusardi and Mitchell, 2007). The identification problem involves matching differences in financial literacy with a subgroup that has self-selected into financial education programs. Measuring financial literacy is not an easy task. Huston (2010) suggests that financial literacy be measured by how well an individual can understand and use personal finance-related information. However, what is deemed relevant is an open question since it must be contextualized to the targeted population. Xiong and Zhang (2020) adapt the Program for International Student Assessment (PISA) financial literacy measurement framework to conditions facing rural Chinese. Their PISA metrics include compound interest, inflation, and risk dispersion in addition to basic questions on loan products, rural pension insurance, pre-loan preparation, risk and reward, financial planning and responsibility, consciousness about savings, and investments. Sampling 1,565 individuals from Shandong, Henan, and Guizhou, they computed PISA scores and used logistic regression and propensity score matching (PSM) to identify the household characteristics/demographics which identify financial literacy. The pairing of subgroups showed that individuals with financial education had significantly higher financial literacy scores (5.16 points out of 26). This was also found in their regression and PSM results. On gender, they found that women in Henan and Guizhou had lower scores than men, but this was reversed in Shandong. Individuals with higher levels of education generally were more financially literate. As a practical matter, expanding the financial literacy of China’s farming communities is no easy task for the simple reason that the target population is immense, and the current demographic is largely uneducated. Xiong and Zhang (2020) for example found that only 20.4 percent of respondents had some form of financial education, and only 47.5 percent had at least completed high school. The intergenerational ability to transfer financial knowledge would have been hampered significantly since of the respondents’ parents only 9.9 percent had at least completed high school. With a greater proportion of Chinese students completing high school, the more obvious pathway to financial education would be through the formal educational system, including technical high schools. It is also clear, however, that in China the

Inclusive finance and inclusive rural transformation in China  413

RCC/RCBs and related agricultural lenders can also play a significant role and increase their social interactions with farmer-borrowers through outreach programs and in-house training programs to reduce the asymmetric information and understandings between borrowers and lenders (Turvey et al., 2014). Credit and Credit Demand in Rural China Microfinance exists because there is a segment of potential borrowers that commercial lending institutions see as being either too poor, too small, too risky, too costly, or all of the above. Yet, this market in general has a decided demand for credit even in small amounts, and microcredit fills this gap. The spread between conventional deposit rates at commercial institutions and interest charges on microloans to the poor has been attributed to the direct and opportunity costs of these small-borrower characteristics. On the demand side, micro-entrepreneurs require microcredit for two reasons. The first is the acquisition of working assets, including a simple inventory of goods to be resold through new marketing channels, and the second is as a liquidity reserve. Much has been written on the first, but the second has not been explored in detail even though the fundamental concept of a credit reserve as a source of liquidity has existed since at least 1981 (Barry et al., 1981). More recently, this concept has been reconstituted in the development literature to refer to the generation of cash from borrowing to avoid the sale of mobile or immobile working assets during times of crisis. At the level of entrepreneurship, it is reasonable to characterize a positive demand when the marginal revenue, net of direct costs, exceeds the marginal cost of borrowing. More generally, if the return on assets exceeds the cost of borrowing, a rational demand for credit exists. Randomized control studies by McKenzie and Woodruff (2008) and McKenzie et al. (2008) in Mexico and Sri Lanka (for small non-farm businesses) respectively found that the return on capital with a small loan amount returned 4.6–5.3 percent/month and 20–33 percent/month respectively. Banerjee and Duflo (2008) exploited a natural experiment in India when the upper limit on capital for which a subsidized loan could be received was later reversed. They find that newly subsidized firms saw greater returns to capital than those that were initially eligible. These results are not out of line with the findings in Turvey et al. (2012) for Chinese farmers and Verteramo-Chiu et al. (2014) for Mexican and Chinese farmers that the elasticity of demand increases as interest rates fall, or those in Bogan et al. (2015) for Dominican Republic micro-entrepreneurs that state that the demand for credit is more elastic (= 1) than previously believed. Inclusive financial policies need also to consider how the demand for agriculture lending is affected by interest rate changes and how risk affects demand. Farm households will borrow for consumption or production purposes, and in many situations, consumption loans exceed production loans for a variety of social and cultural reasons. There is also a strong relationship between farms’ demand for credit and idiosyncratic or personal shocks. It is generally assumed that credit demand is highly inelastic, but recent evidence suggests that this may not hold true generally (Turvey et al., 2012; Bogan et al., 2015 and citations therein). Some evidence suggests that demand becomes more elastic as interest rates decrease, which has important policy implications. First, it is often argued that because interest rates are highly inelastic, imposing interest rate ceilings or providing subsidies in support of transformative policies will have little effect on borrowing behavior. The effects of an interest rate subsidy can just as easily be accomplished with a decoupled lump-sum transfer. If credit demand is elastic then interest rate policies can in fact influence credit demand. This would

414  Handbook of microfinance, financial inclusion and development

be especially beneficial if policymakers wished to encourage specific types of technology or induce farmers to switch to different forms of agriculture. The effect is to increase the returns from the new investment to the farmer, thereby removing risk or cost barriers relative to the status quo. Turvey et al. (2012) used a multiple bounded discrete choice model to derive individual household demand elasticities in Shaanxi and Gansu provinces.10 They found a range of elasticities with average point estimates of –0.60, with nearly 20 percent having perfectly inelastic demand elasticities, 20 percent having elasticities above –0.75, and 15 percent having elasticities exceeding –1.0. These results suggest that significant numbers of farmers will increase credit demand if interest rates fall, but not all farmers will do so. They also found degrees of endogeneity in these elasticities. For example, higher-risk farms tended to reveal more inelastic demands, while more profitable farms revealed more elastic (or less inelastic) demands. Farmers with greater savings tended to have more inelastic demands. Finally, in terms of “usage,” they found that farmers borrowing for non-productive or consumption purposes (e.g. building a house, education, etc.) had a more elastic demand than borrowing for production purposes, suggesting the former is more flexible as a luxury than the latter, it is viewed more as a necessity. Risk Rationing Perhaps more important to understanding inclusive financial policies is in understanding the nature of constraints facing Chinese farmers. These can broadly be defined in terms of quantity rationing, price rationing, transaction costs rationing, and risk rationing (Boucher et al., 2008). Quantity rationing refers to external credit rationing by the lender who fails to supply the amount of loan required. Price rationing refers to increases or decreases along the credit demand curve, which as discussed in the previous section is determined largely by the credit demand elasticities. Transaction costs rationing refers to the costs of obtaining a loan, which may include service costs, transportation costs, collateral requirements, and guarantees. Risk rationing is an important behavioral consideration in rural transformation. Loosely speaking, risk rationing refers to situations in which farmers have a real demand for credit and have access to that credit but will not avail themselves of that credit because of an aversion to losing collateral in mobile or fixed productive assets (Binswanger and Sillers, 1983; Boucher et al., 2008; Verteramo-Chiu et al., 2014). For many farmers, even subsidized or reduced interest rates on a loan will not encourage borrowing when borrowing can generate positive returns. Risk rationing behavior has been observed in many field studies, although not explicitly recognized as such.11 Segregating the sources of rationing into its components is important. Verteramo-Chiu et  al. (2014) followed a direct elicitation approach on 730 Shaanxi farm households using a design structure similar to that suggested in Boucher et al. (2009) to investigate quantity, price, and risk rationing (with transactions costs rationing bundled with price rationing). They found that only 14 percent of Chinese farmers were quantity rationed, 6.2 percent were risk rationed, and 79.9 percent were price rationed.12 Fifty-two of the Chinese farmers were deemed creditworthy and offered a loan without requesting one, and of this group, 27.3 percent did not fully utilize the loan due to risk rationing effects. Another interesting finding is between credit demand elasticities and rationing typology. Generally speaking, they found no strong or conclusive relationship between demand elasticity and risk rationing, suggesting that even though a farmer might behave as if demand was perfectly or highly inelastic, the true demand might actually be more elastic, but just not acted upon.

Inclusive finance and inclusive rural transformation in China  415

INNOVATIONS IN RURAL FINANCIAL SERVICES AND INCLUSIVE RURAL TRANSFORMATION To address deficiencies in China’s credit market, certain types of aligned credit structures have been formalized in law. To overcome barriers to credit access, a number of innovative mechanisms to reduce or eliminate collateral requirements reduce exclusion and increase financial inclusion. Through collateral-free lending, public support combined with entrepreneurial RCCs or other financial institutions is delivering financial products and services that bypass the collateral restrictions faced by many farmers. The suite of innovative financial structures using aligned credit, when properly employed and monitored, brings much-needed credit to potentially millions of farm households. Aligned Credit Structures Aligned credit structures involve the conversion of a self-regulated mutual self-help group through interlinkage with an RCC, while the other is regulated as a bank. This has included a reformation of the old-style lending practices to newer customer-oriented approaches through government- and central bank–supported activities. Most interesting are the interlinked or aligned relationships between the lenders and the borrowers that tie the informal channels of government, donor or NGO development funds, and mutual fund associations to the formal channels. These mechanisms are varied and, in most cases, require some external support from the government or an NGO, and the participating lender can be an RCC, agricultural bank, or MFI. Most likely, the participating lender will be a deposit-taking institution or an MFI in partnership with a deposit-taking institution. One of these pathways is illustrated in Figure 21.1a. For many poor areas designated as poverty zones, the government makes money available through poverty alleviation programs. These funds are placed into a village development fund which is normally supported by a self-help group or placed under the leadership of a village committee or other organization. Farmers can buy shares into the village development fund which then loans money to the farmers. The capital available for lending is the share value plus the poverty alleviation fund contributions. Historically, these projects were minimally effective because the village development fund could not leverage the total capital available to it and therefore could not meet the demand for credit. However, a breakthrough innovation in financial inclusion is the alignment of the village development fund with rural credit cooperatives willing to leverage the fund’s capital by a multiple of eight. Thus, the village development fund, originally held on behalf of a mutual self-help group, is transformed into a mutual self-help cooperative formally aligned with a lender. Such credit alliances are currently taking shape throughout China, although some are quite informal. At the farm and village level, other innovations that lead to collateral-free or collateral-reduced lending are the organizations of guarantee groups at both the farm and SME levels, credit mutual aid associations, community credit councils (e.g. village committee or village credit committee, some with compensation others without), farmers credit mutual aid associations, and private credit unions for small and medium enterprises. An alternative structure widely adopted in China is the mutual self-help cooperative (MSHC). Figure 21.1b shows the basic structure of a mutual self-help cooperative in which the government provides indirect security to the aligned RCC. The MSHC provides a mutual guarantee for all RCC loans and takes on the role of credit assessor. The members also provide a guarantee based upon their capital contributions which are irrevocable savings and included

416  Handbook of microfinance, financial inclusion and development

Notes:  (a) (top left) village development funds; (b) (top right) depiction of mutual self-help cooperative with 5 percent of loan contribution; (c) (center left) depiction of mutual self-help cooperative with 5 percent of share capital as contribution; (d) (center right) depiction of cooperative aligned with RCC; (e) (lower left) aligned credit and value chain

Figure 21.1 Aligned credit structures

Inclusive finance and inclusive rural transformation in China  417

in the security pool. The government provides an initial seed guarantee which is ultimately paid back by the MSHC. The MSHC then acts as an intermediary between the RCC and farmers, although in most instances the RCC will lend directly to the farmer rather than funneling money through the MSHC. In this model, farmers are asked to contribute 5 percent of the loan to the mutual pool of funds. These funds are initially used to replace the initial guarantee capital of the government support funds using a scaled repayment schedule. An alternative structure, shown in Figure 21.1c, is similar. Unlike the previous model, farmers do not contribute 5 percent of loans to the MSHC, but 5 percent of share value at issue. Thus, the amount of capital available is eight times 95 percent of member contributions, with the remaining 5 percent being returned to the government sponsor. When that is repaid, it is retained as a surplus to the MSHC. Note that the scaled repayment does not eliminate the initial guarantee capital but simply replaces it. Thus, the initial guarantee is always available and is held in an account with the RCC as a permanent guarantee deposit. Interlinked or Aligned Cooperative Lending An alternative aligned credit mechanism is developed through a formal supplier or marketing cooperative. This can also be applied to marketing boards and other farmer or agribusiness industry organizations. Figure 21.1d shows the basic structure of an aligned marketing and/or supplier cooperative. The cooperative is normally started by a sponsor or sponsors who may provide a significant amount of their own capital to the cooperative. This capital contributes to the guarantee pool, and because it is private capital, the founders in return receive “Class A” shares which have special rights and special dividend rules. One of these rights is a veto right over member votes if the member vote puts undue risk on capital. Members contribute to the cooperative by purchasing shares, normally for about 100 RMB. Members have a one-member-one-vote privilege, but the vote is not binding if it adversely affects the principals. The capital contributed through Class A and Class B shares makes up the capital of the cooperative. The cooperative is aligned with an RCC which can advance funds to the cooperative to cover purchases paid on account or provide funds directly to members who wish to purchase from the cooperative. Some cooperatives have a member’s only rule, but many do not and will allow any farmer to purchase with special discounts and/or patronage going to the members. However, only members would be permitted to obtain a guarantee from the cooperative for an RCC loan. Typically, the cooperative will not lend on account or provide a guarantee to a farmer who is not deemed creditworthy by his member peers. Thus, the cooperative provides an initial screening which makes it easier for the RCC to justify a loan. The loan process has two guarantees. The first is a group guarantee by members where non-members are not allowed to be part of the group, and the second is a guarantee from the cooperative itself. The cooperative is the second line of defense. If the group guarantee fails, the cooperative will pay the amount of the loan outstanding from its own capital or reserves and will then attempt to collect from the group members. If collection is ultimately unsuccessful, the cooperative will extend the loan repayment against its income and absorb the loss. Aligned Farmer–Lender Relationships with SMEs An additional barrier to inclusive transformation is farmers’ ability to participate in global marketing schemes through vertically linked value chains. The critical barrier is the ability of the SME to obtain the required working capital to fund product and marketing operations. In the standard setup illustrated in Figure 21.1e, the SME might provide seed, fertilizer (or

418  Handbook of microfinance, financial inclusion and development

animals, feed, etc.), and other inputs to the farmer, who will in turn follow stated management practices to provide a uniform product to the SME for sale or export. Value chains require massive amounts of working capital to purchase the required inputs used by member farmers. When chain farmers deliver harvests to the SME, the value of the inputs supplied is deducted from the contracted or market value of the commodity delivered. Only when that commodity is sold into the marketplace and all receivables collected will the SME have the capacity to repay the loan. In this setup, the SME is an agent of the lender. Rather than the RCC or lender providing funds to each farmer in the value chain, the SME takes on this role. Working capital will be secured by contracts along the supply chain and the internal capital of the SME. Farmers enter into a forward contract with the SME and need not risk collateral; the burden of financial risk rests with the SME. Depending on the scale of operations and the working capital required, the SME might have to syndicate loans across many lenders. Key to this is a legal structure for contract enforcement with farmers, but from time to time, widespread systemic risks may result in the inability of farmers to fulfill contract terms. In most instances, “loan” repayment by farmers is on a first-in basis deducted from contract and delivered produce. While the SME bears the burden of financial risk, farmers ultimately bear the burden of production risk; however, under such a structure farmer assets are (mostly) preserved.

LAND USE RIGHT REFORMS AND RURAL LAND MORTGAGE LOANS The underdevelopment of agricultural credit in China is largely due to the small scale of farming and the historical limits on transferring land use rights (LUR) between farmers to increase the size and scale of farming operations. After 1978 farm household members were issued a LUR based on rural residence, and in aggregate amounted to about one acre, or six Chinese mou, for a family of six. Land use rights are usufruct rights issued to natural persons, legal persons, or other organizations that provide rights of production and management, but do not bestow rights of land ownership. In other words, agricultural production was decoupled from land ownership or title and the land is either owned by the state or a rural collective organization (e.g. a village).13 However, in recent years, the government of China has acted to loosen the attachment of land use rights and to permit open transactions between farmers through newly established land transfer centers. Reforms promulgated in 2015 are the most significant and will transform China’s agricultural economy in size and scale, and in doing so increase the demand for agricultural credit. Separating land use/management rights from land contract rights is an important amendment to China’s law on land contracts. Most significant is the splitting and division of land property rights. Farmers could keep their contract right unchanged and only transfer the management right when they rent out their contracted land. Aside from that, they can now use the management right to mortgage the loan or “invest” the right to exchange the shares in a cooperative (Peng and Zhou, 2020). The genesis of LUR reforms is owing to the issuance of the Property Law in 2007 which provided the property base for LUR through a national registry, which did not start until 2014. On October 12, 2008, the Chinese government issued a policy regarding the Advancement of Rural Reform and Development which made allowances for farmers to lease their contracted farmland or transfer their land use rights. At about the same time, the government issued policies concerning the Financial Advancement of Economic Development in 2008 to encourage financial institutions to expand the scope of

Inclusive finance and inclusive rural transformation in China  419

rural collateral and explore various credit products. The provisions to enable farm households to mortgage LUR in some locations was issued by the Central People’s Bank and China Banking Regulatory Commission in 2009. The clear intention was not only to improve the scale of existing operations, but by linking LUR to longer-term credit the broader effort was aimed at farm commercialization and advancing entrepreneurial activity (Peng and Kong, 2020). Subsequently, China’s first policy documents for 2014 and 2015—the No. 1 Central Policy—allowed farmers to mortgage LUR in certain locations and under certain conditions. These are known as rural land mortgage loans. On August 10, 2015, the document concerning the trial implementation of rural management rights over contracted land and farmers’ homes as collateral for bank loans was approved by the State Council confirming that Chinese farmers would be allowed to transfer LUR and use LUR and homes to raise mortgages, in addition to converting LUR into shares in large-scale farming entities (Peng and Kong, 2019). How changes to China’s LUR laws will affect overall credit demand and supply in China is not clear at this time. It is programmatically inclusive since farmers who rent out their contracting rights can use the contract to secure a loan. These loans can then be put to some other entrepreneurial use. At present, the rate of rural transformation is significant. Land transfer centers will bundle the land from 100 persons and transfer that land to a single farmer. In time, China’s agricultural economy will be comprised of large commercial farms rather than the hundreds of millions of small and poor, limited-resource farms that have defined China’s rural landscape to date. This will redefine credit markets and institutions. Borrowing against the rights immediately increased access to credit for both landlord and tenant, and in many locations, local rural credit banks or cooperatives established aligned credit relationships with the land transfer centers.

SUMMARY AND CONCLUSIONS This chapter provided an overview of inclusive finance and how this relates to inclusive rural transformation in China. China has a formal inclusive finance policy but its operation until recent years has been targeted at small-scale agriculture. To increase access to credit, China has improved regulatory oversight through the China Banking and Regulatory Commission. In addition, China has improved licensing of numerous new types of financial institutions including the conversion of rural credit cooperatives to joint-stock rural credit banks, the formation of village banks, postal savings, non-deposit joint-stock lending institutions, microfinance institutions, and loan guarantee companies. Through these new institutions, “agricultural” loans have increased substantially, but it is important to recognize that in China an agricultural loan is one made in an agricultural region and not necessarily to farmers. To encourage lending to farmers, the CBRC has allowed rural financial institutions to identify creditworthy farmers and creditworthy villages to receive microloans. To address the lack of collateral and asymmetric information, group guarantees are often required for a farmer to obtain a loan. In poverty zones, alternative mechanisms to improve credit access to agriculture include village development funds, mutual self-help, and aligned credit and value chains. The usage of credit in China still favors informal or familial lending. Familial lending should not be construed as a consequence of credit rationing. In fact, the opposite may be true; only 14 percent of Chinese farmers were quantity rationed, 6.2 percent were risk rationed, and 79.9 percent were price rationed. This dynamic, however, will undoubtedly change as new

420  Handbook of microfinance, financial inclusion and development

laws permitting the legal transfer of land use rights converts China from a limited resource to a commercially oriented agricultural economy.

NOTES 1. Financial deepening refers to the total liquidity in the market, e.g. the ratio of the money supply to GDP which indicates that an increase in GDP spins off more cash and contributes to economic growth (Mckinnon, 1973; Shaw, 1973; Galbis, 1977; Kapur, 1976; Khatkhate, 1988; Mathieson, 1980; Fry, 1978; Townsend and Ueda, 2006). “Financial repression” describes a set of policies that aim to extract revenue from the financial system and to use the financial system to funnel resources into specific sectors of the economy (Mckinnon and Mathieson, 1981; Feldman and Gang, 1990). In the absence of centralized monetary control, money will find its best economic use through the competitive bid for scarce resources (McKinnon, 1991). Financial repression affects financial deepening by distorting financial prices (the real interest rate) leading to low (or negative) real growth in the economy (Feldman and Gang, 1990; Demetriades and Luintel, 1997). 2. See, for example, Feder (1985), Binswanger and Sillers (1983), Carter (1988), Carter and Olinto (2003), Petrick (2004), Foltz (2004), Bernhardt and Bacuks (1990), Eswaran and Kotwal (1990), and Jia et al. (2010). 3. See, for example, Turvey et al. (2011) specifically for China, as well as Sahn and Alderman (1988), Schultz (1993), Carter and May (2001), Carter and Barrett (2006), Morduch (1994, 1995), Carter and Olinto (1993). 4. Related results outside of China include Foltz (2004) who finds that, in Tunisia, credit constraints have a negative impact on profits but not investment. Boucher et al. (2009) find a substantial loss in profits (–27 percent) and financial efficiency (–23 percent) in the presence of credit constraints in Peru. Baiyegunhi et al. (2010), examining household impacts of credit constraints in South Africa, find that the average value of assets for credit constrained households was estimated at 1,703 rand, while for unconstrained households the average asset value was 54,929 rand. 5. For example, one credit-worthy village manufactured coffins for the Italian market. Crop lands were converted from grains to wood, and the wood was harvested, cut, shaved, and formed into coffins with each household contributing in some way to the whole. 6. From a measurement point of view, indices of this sort have several desirable axiomatic features (see Sarma, 2010; Nathan et  al., 2008; and Chakravarty and Pal, 2013); first, equation (21.1) satisfies normalization (a measure of IFI should be bound between 0 and 1); second, homogeneity in d and w (i.e. for any arbitrary scaling by c the dimension and index are scaled, d i ( Ai , mi , M i ) = d i ( cAi , cmi , cM i ) and IFI ( w1 , w2 ...wn ) = IFI ( cw1 , cw2 ...cwn ) ); third, anonymity (identical IFI values will result if one dimension is swapped for another); fourth, monotonicity (a rise or fall in one dimension, holding all other dimensions constant will have a concomitant rise or fall in the IFI); fifth, concavity (as a dimension value rises, the rate at which IFI increases is diminishing); sixth, proximity (a greater or lower value of the IFI index indicates that it is closer to or further from the ideal measure of financial inclusion); seventh, uniformity (the greater the dispersion (standard deviation) about any one or all dimensions about the means will have a lower index value (because of the squaring)); and eighth, signaling (a measure of IFI should indicate a unique, optimal path to reach a higher value). 7. For example, absolute exclusion will occur when Ai = mi for all dimensions and absolute inclusion will occur when Ai = M i for all dimensions. We refer to this latter point as the ideal target since Ai = M i Þ d i = 1. In reality, 0 < IFI < 1 when mi < Ai < M i for at least one dimension. Sarma and Pais (2011), however, use a different weighting scheme. 8. These results are drawn from Xiong et al. (2013). In that paper they also provided results for Hubei province. These were generally similar except for the third equation in which IFI was negative and significant, suggesting that increases in IFI reduced the consumption-to-income ratio. This result makes sense if IFI increased income at a higher rate than consumption. However, the range of IFI across Hubei counties was much higher than across China’s provinces. 9. Many papers in the agricultural finance literature interchange the words access and usage. The word “accessed” is a verb and can be interchanged with “used,” but in the context of “having access” the word access is a noun and does not necessarily imply usage, although in many contexts

Inclusive finance and inclusive rural transformation in China  421

10.

11. 12.

13.

it is presumed to be the same. These results confirm Beck’s separation of the terms “access” and “usage.” Other studies with estimates of credit demand elasticities include Weersink et al. (1994, US Rural Farmers USDA survey data) with elasticity estimates between –0.84 and –0.69; Bell et al. (1997, Punjab Rural Farmers World Bank survey data) with an elasticity estimate of –0.22; Kochar (1997, India Rural Farmers Government of India survey data low demand for credit) who found credit demand inelastic; Gross and Souleles (2002, US credit card holders bankcard issuer account archives) who found a short-run elasticity estimate of –0.80; Dehejia et  al. (2012, Bangladesh Micro-Entrepreneurs Credit Cooperative data) with elasticity estimates between –1.04 and –0.73; Karlan and Zinman (2008, South Africa Working Poor RCT with loan contract data) found elasticity estimates between –0.51 and –0.14; and Bogan et al. (2015) (Dominican Republic microentrepreneurs with MFI, by survey and experiment) had mean elasticity of –1.0. Verteramo-Chiu et  al. (2014) list Binswanger and Siller (2008); Eswaran and Kotwal (1990); Morduch (1995); Bell et  al. (1997); Swain (2002); Bhattacharyya (2005); Castellani (2014); Binswanger and Sillers (1983). This study also compared results to 372 Mexican farmers using the same instrument; 9.94 percent of Mexican farmers were quantity rationed; and 55.37 percent were price rationed. Barham, Boucher, and Carter (1996) report that 32 percent of Guatemalan farmers surveyed did not apply for credit and were fully constrained in their credit choice due to either transactions costs (transaction cost rationing) or fear of risk leading to self-insure (but without using the term “risk rationing”). Boucher et  al. (2009) find 8.6 percent of surveyed Peruvian farmers in 1997 were risk rationed and (with) Fletschner et al. (2010) find 21 percent to 25 percent of a resample of Peruvian farmers in 2003 to be risk rationed. Boucher et al. (2008) report results from a number of surveys that 19 percent of Peruvian farmers, 16 percent of Honduran farmers, and 12 percent of Nicaraguan farmers were identified as risk rationed. Zhou et al. (2019) provide an overview of land reforms since 1949 and provide a broad overview of land tenure issues and policy initiatives in China. Khantachavana et al. (2013) review land tenure relationships in China and provide estimates of willingness to pay and willingness to accept the present value of transfer rights.

REFERENCES Baiyegunhi, L. J. S., Fraser, G. C. G. and Darroch, M. A. G. (2010), “Credit Constraints and household welfare in the Eastern Cape Province, South Africa,” African Journal of Agricultural Research, 5(16), 2243–2252. Banerjee, Abhijit and Duflo, Esther (2008), “Do Firms Want to Borrow More: Testing Credit Constraints Using a Targeted Lending Program,” BREAD Working Paper No. 005, 2004, revised 2008. Barham, B. L., Boucher, S. and Carter, M. R. (1996), “Credit Constraints, Credit Unions, and SmallScale Producers in Guatemala,” World Development, 24(5), 793–806. Barry, P. J., Baker, C. B. and Sanint, L. R. (1981), “Farmers’ Credit Risks and Liquidity Management,” American Journal of Agricultural Economics, 63(2), 216–227. Beck, T. and de la Torre, A. (2007). “The Basic Analytics of Access to Financial Services,” Financial Markets, Institutions and Instruments, 16(2), 79–117. Beck, T. and Demirgüç-Kunt, A. (2008), “Access to Finance: An Unfinished Agenda,” World Bank Economic Review, 22, 383–396. Beck, T., Ross, L. and Loayza, N. (2000), “Finance and the Sources of Growth,” Journal of Financial Economics, 58, 261–300. Beck, T., Demirgüç-Kunt, A. and Peria, M. S. M. (2007). “Reaching Out: Access to and Use of Banking Services Across Countries,” Journal of Financial Economics, 85(1), 234–266. Bell, C., Srinivansan, T. and Udry, C. (1997), “Rationing, Spillover, and Interlinking in Credit Markets: The Case of Rural Punjab,” Oxford Economic Papers, 49(4), 557–585. Bhattacharyya, S. (May 2005), “Interest Rates, Collateral and (de-) Interlinkage: A Micro-Study of Rural Credit in West Bengal,” Cambridge Journal of Economics, 29(3), 439–462. Binswanger, H. P. and Sillers, D. A. (1983), “Risk Aversion and Credit Constraints in Farmers’ DecisionMaking: A Reinterpretation,” The Journal of Development Studies, 20(1), 5–21.

422  Handbook of microfinance, financial inclusion and development

Bogan, V. L., Turvey, C. G. and Salazar, G. (2015), “The Elasticity of Demand for Microcredit: Evidence from Latin America,” Development Policy Review, 33(6), 725–757. Boucher, S. R., Carter, M. R. and Guirkinger, C. (2008), “Risk Rationing and Wealth Effects in Credit Markets: Theory and Implications for Agricultural Development,” American Journal of Agricultural Economics, 90(2), 409–423. Boucher, S. R., Guirkinger, C. and Trivelli, C. (2009), “Direct Elicitation of Credit Constraints: Conceptual and Practical Issues with an Application to Peruvian Agriculture,” Economic Development and Cultural Change, 57(4), 609–640. Cao, Y. J., Turvey, C., Ma, J., Kong, R., He, G. and Yan, J. (2016), “Incentive Mechanisms, Loan Decisions and Policy Rationing: A Framed Field Experiment on Rural Credit,” Agricultural Finance Review, 76(3), 326–347. Carter, M. R. (1988), “Equilibrium Credit Rationing of Small Farm Agriculture,” Journal of Development Economics, 28(1), 83–103. Carter, M. R. and Barrett, C. B. (2006), “The Economics of Poverty Traps and Persistent Poverty: An Asset-Based Approach,” Journal of Development Studies, 42(2), 178–199. Carter, M. R. and May, J. (2001), “One Kind of Freedom: The Dynamics of Poverty in Post-Apartheid South Africa,” World Development, 29, 1987–2006. Carter, M. R. and Olinto, P. (2003), “Getting Institutions ‘Right’ for Whom? Credit Constraints and the Impact of Property Rights on the Quantity and Composition of Investment,” American Journal of Agricultural Economics, 85(1), 173–186. Castellani, D. (2014), “Shocks and Credit Choice in Southern Ethiopia,” Agricultural Finance Review, 74(1), 87–114. Chakravarty, S. R. and Pal, R. (2013), “Financial Inclusion in India: An Axiomatic Approach,” Journal of Policy Modeling. http://dx​.doi​.org​/10​.1016​/j​.jpolmod​.2012​.12​.007. Dehejia, R., Montgomery, H. and Morduch, J. (2012), “Do Interest Rates Matter? Credit Demand in the Dhaka Slums,” Journal of Development Economics, 97(2), 437–449. Demetriades, Panicos O. and Luintel, Kul B. (1997), “The Direct Costs of Financial Repression: Evidence From India,” The Review of Economics and Statistics, 79(2), 311–320. Demirgüç-Kunt, A., Beck, T. and Honohan, P. (2008), Finance for All? Policies and Pitfalls in Expanding Access. World Bank Research Report. World Bank. Washington, DC. Demirgüç-Kunt, A., Córdova, E. L., Pería, M. S. M. and Woodruff, C. (2011), “Remittances and Banking Sector Breadth and Depth: Evidence from Mexico,” Journal of Development Economics, 95(2), 229–241. Dev, S. M. (2006), “Financial Inclusion: Issues and Challenges,” Economic and Political Weekly, 41(October 14–16), 4310–4313. Eswaran, M. and Kotwal, A. (1990), “Implications of Credit Constraints for Risk Behaviour in Less Developed Economies,” Oxford Economic Papers, 42(2), 473–482. Feder, G. (1985), “The Relation between Farm Size and Farm Productivity: The Role of Family Labor, Supervision and Credit Constraints,” Journal of Development Economics, 18, 297–313. Feder, G., Lau, L. J., Lin, J. Y. and Luo, X. (1990), “The Relationship between Credit and Productivity in Chinese Agriculture: A Microeconomic Model of Disequilibrium,” American Journal of Agricultural Economics, 72(5), 1151–1157. Feldman, David H. and Gang, Ira N. (1990), “Financial Development and the Price of Services,” Economic Development and Cultural Change, 38(2), 341–352. Fletschner, D., Guirkinger, C. and Boucher, S. (2010), “Risk, Credit Constraints and Financial Efficiency in Peruvian Agriculture,” The Journal of Development Studies, 46(6), 981–1002. Foltz, Jeremy D. (2004), “Credit Market Access and Profitability in Tunisian Agriculture,” Agricultural Economics, 30, 229–240. Fry, Maxwell J. (1978), “Money and Capital or Financial Deepening in Economic Development?” Journal of Money, Credit and Banking, 10(November), 464–475. Fu, H. and Turvey, C. G. (2018), The Evolution of Agricultural Credit during China’s Republican Era, 1912–1949. New York: Springer. Galbis, Vicente (1977), “Money and Finance in Economic Growth and Development, Essays in Honor of Edward S. Shaw,” Finance & Development, 14(3), 40–41. Gross, D. B. and Souleles, N. S. (2002), “Do Liquidity Constraint and Interest Rates Matter for Consumer Behavior? Evidence from Credit Card Data,” The Quarterly Journal of Economics, 117(1), 149–185.

Inclusive finance and inclusive rural transformation in China  423

Guo, P. and Jia, X. (2009), “The Structure and Reform of Rural Finance in China,” China Agricultural Economic Review, 1(2), 212–236. Huston, S. J. (2010), “Measuring Financial Literacy,” Journal of Consumer Affairs, 44(2), 296–316. Jia, X., Heidhues, F. and Zeller, M. (2010), “Credit Rationing of Rural Households in China,” Agricultural Finance Review, 70(1), 37–54. Kapur, Basant K. (1976), “Alternative Stabilization Policies for Less Developed Countries,” Journal of Political Economy, 84(4), Part 1, 777–795. Karlan, D. S. and Zinman, J. (2008), “Credit Elasticities in Less Developed Economies: Implications for Microfinance,” American Economic Review, 98(3), 1040–1068. Khantachavana, S. V., Turvey, C. G., Kong, R. and Xia, X. (2013), “On the Transaction Values of Land Use Rights in Rural China,” Journal of Comparative Economics, 41(3), 863–878. Khatkhate, Deena R. (1988), “Assessing the Impact of Interest Rates in Less Developed Countries,” World Development, 16, 577–588. Kochar, A. (1997), “Does Lack of Access to Formal Credit Constrain Agricultural Production? Evidence from the Land Tenancy Market in Rural India,” American Journal of Agricultural Economics, 79(3), 754–763. Kong, R., Turvey, C., Xu, X. and Liu, F. (2014), “Borrower Attitudes, Lender Attitudes and Agricultural Lending in Rural China,” The International Journal of Bank Marketing, 32(2), 104–129. Kong, R., Turvey, C. G., Channa, H. and Peng, Y. (2015), “Factors Affecting Farmers’ Participation in China’s Group Guarantee Lending Program,” China Agricultural Economic Review, XXXX. Kumar, C. S., Turvey, C. G. and Kropp, J. D. (2013), “The Impact of Credit Constraints on Farm Households: Survey Results from India and China,” Applied Economic Perspectives and Policy, 35(3), 508–527. Li, M. (2008), “What Affects Financial Breadth and Financial Depth? A Cross-Country Analysis,” South China Journal of Economics, 5, 56–67. Liu, T., He, G. and Turvey, C. G. (2019), “Inclusive Finance, Farm Households Entrepreneurship, and Inclusive Rural Transformation in Rural Poverty-stricken Areas in China,” Emerging Markets Finance and Trade, 57(9), 1–30. Lusardi, A. and Mitchell, O. S. (2007), “Baby Boomer Retirement Security: The Roles of Planning, Financial Literacy, and Housing Wealth,” Journal of Monetary Economics, 54(1), 205–224. Mathieson, Donald J. (1980), “Financial Reform and Stabilization Policy in a Developing Economy,” Journal of Development Economics, 7(3), 359–395. McKenzie, David and Woodruff, Christopher (2008), “Experimental Evidence on Returns to Capital and Access to Finance in Mexico,” The World Bank Economic Review, 22(3), 457–482. McKenzie, David, de Mel, Suresh and Woodruff, Christopher (2008), “Returns to Capital: Results from a Randomized Experiment,” Quarterly Journal of Economics, 123(4), 1329–1372. Mckinnon, Ronald I. (1973), Money and Capital in Economic Development. Washington, DC: Brookings Institution. Mckinnon, Ronald I. (1991), “Financial Control in the Transition from Classical Socialism to a Market Economy,” The Journal of Economic Perspectives, 5(4), 107–122. Mckinnon, Ronald I. and Mathieson, Donald J. (c1981), How to Manage a Repressed Economy: International Finance Section. Princeton, NJ: Princeton University Press. Meyer, R. L. (2011), “Subsidies as an Instrument in Agricultural Finance: A Review,” Joint Discussion Paper. Washington, DC: The International Bank for Reconstruction and Development/The World Bank, June. Morduch, J. (1994), “Poverty and Vulnerability,” The American Economic Review, 84(2, Papers and Proceedings), 221–225. Morduch, J. (1995), “Income Smoothing and Consumption Smoothing,” The Journal of Economic Perspectives, 9(3), 103–114. Nathan, H. S. K., Mishra, S. and Reddy, B. S. (2008), “An Alternative Approach to Measure HDI,” IGIDR Working Paper WP-2008-002. OECD (Organisation for Economic Co-operation and Development) (2013), Advancing National Strategies for Financial Education. A Joint Publication by Russia’s G20 Presidency and the OECD. Paris: OECD Publishing. Park, A. and Ren, C. (2001), “Microfinance with Chinese Characteristics,” World Development, 29(1), 39–62.

424  Handbook of microfinance, financial inclusion and development

Peng, Y. and Kong, R. (2020), “An Analysis of China’s Reforms on Mortgaging and Transacting Rural Land Use Rights and Entrepreneurial Activity,” Agricultural Finance Review, 80, Forthcoming. https://doi​.org​/10​.1108​/AFR​- 08​-2019​- 0094. Peng, P. and Zhou, L. (2020), “Reforming Land Mortgages in Rural China and the Incentives for Application of Environment-Friendly Formula Fertilizers on Farm,” Agricultural Finance Review, 80, Forthcoming. https://doi​.org​/10​.1108​/AFR​- 03​-2019​- 0026. Petrick, M. (2004), “A Microeconometric Analysis of Credit Rationing in the Polish Farm Sector,” European Review of Agricultural Economics, 31, 77–101. Ravallion, M. (2012), “Troubling Tradeoffs in the Human Development Index,” Journal of Development Economics, 99, 201–209. Sahn, D. and Alderman, H. (1988), “The Effect of Human Capital on Wages and the Determinants of Labor Supply in a Developing Country,” Journal of Development Economics, 29(2), 157–183. Sarma, M. (2010), “Index of Financial Inclusion,” Discussion Paper in Economics 10-05, CITD Jawaharlal Nehru University, India, November 2010. Sarma, M. and Pais, J. (2011), “Financial Inclusion and Development,” Journal of International Development, 23, 613–628. Schultz, T. P. (1993), “Investments in the Schooling and Health of Women and Men: Quantities and Return,” Journal of Human Resources, 28(4), 694–734. Shaw, Edward S. (1973), Financial Deepening in Economic Development. New York: Oxford University Press. Shen, M., Huang, J., Zhang, L. and Rozelle, S. (2010), “Financial Reform and Transition in China: A Study of the Evolution of Banks in Rural China,” Agricultural Finance Review, 70(3), 305–332. Swain, R. B. (2002), “Credit Rationing in Rural India,” Journal of Economic Development, 27(2), 1–20. Townsend, R. M. and Ueda, K. (2006), “Financial Deepening, Inequality, and Growth: A Model-Based Quantitative Evaluation,” The Review of Economic Studies, 73(1), 251–293. Turvey, C. G. and Kong, R. (2009), “Business and Financial Risks of Small Farm Households in China,” China Agricultural Economic Review, 1(2), 155–172. Turvey, C. G. and Kong, R. (2010), “Informal Lending amongst Friends and Relatives: Can Microcredit Compete in Rural China?” China Economic Review, 21(4), 544–556. Turvey, C. G., Kong, Rong and Hu, Xuexi (2010), “Borrowing Amongst Friends: The Economics of Informal Credit in Rural China,” China Agricultural Economic Review, 2(2), 133–147. Turvey, C. G., He, Guangwen, Kong, R., Ma, J. and Meagher, P. (2011), “The 7 Cs of Rural Credit in China,” Journal of Agribusiness in Developing and Emerging Economies, 1(2), 100–133. Turvey, C. G., He, G., Ma, J., Kong, R. and Meagher, P. (2012), “Farm Credit and Credit Demand Elasticities in Shaanxi and Gansu,” China Economic Review, 23, 1020–1035. Turvey, C. G., Xu, X., Kong, R. and Cao, Y. (2014), “Attitudinal Asymmetries and the Lender-Borrower Relationship: Survey Results on Farm Lending in Shandong, China,” Journal of Financial Services Research, 46(2), 115–135. Verteramo-Chiu, L. J., Khantachavana, S. V. and Turvey, C. G. (2014), “Risk Rationing and the Demand for Agricultural Credit: A Comparative Investigation of Mexico and China,” Agricultural Finance Review, 74(2), 248–270. Walstad, W. B., Rebeck, K. and MacDonald, R. A. (2010), “The Effects of Financial Education on the Financial Knowledge of High School Students,” Journal of Consumer Affairs, 44(2), 336–357. Wang, H. (2019), “An Economic Investigation on Developments in Agricultural Credit During China’s Collective Period: 1950–1984,” Unpublished M. S. Thesis, Cornell University Graduate School. https://ecommons​.cornell​.edu​/ handle​/1813​/67560. Weersink, A., Vanden-Dungen, M. J. and Turvey, C. G. (1994), “Estimating the Demand for Farm Operating and Term Credit,” Cahiers D’Economie et Sociologie Rurales, 33(4), 97–116. Xiong, Xiaoping, Turvey, C. G. and Li, C. (2013), “Financial Inclusion and Rural Financial Services in China,’” Unpublished mimeo, ageconsearch​.umn​.e​du. Zhang, H. and Xiong, X. (2020), “Is Financial Education an Effective Means to Improve Financial Literacy? Evidence from Rural China,” Agricultural Finance Review, 80(3), 305–320. https://doi​.org​/ 10​.1108​/AFR​- 03​-2019​- 0027. Zhou, Yang, Li, Xunhuan and Liu, Yansui (2019), “Rural Land System Reforms in China: History, Issues, Measures and Prospects,” Land Use Policy, 91, 104330. https://doi​.org​/10​.1016​/j​.landusepol​. 2019​.104330.

22. Does microfinance cause banking sector development and economic growth? An application to Mongolia Batkhuyag Myagmar, Robert Lensink, and Wim Heijman

INTRODUCTION The microcredit movement was originally based on a macroeconomic goal of poverty alleviation in developing countries through financing poor people. However, as one started to realize that credit only will not be sufficient to reduce poverty, microfinance institutions started to offer poor people not only credit but also other financial services, including savings, insurance, remittances, and non-financial services, such as financial literacy training and skills development programs (Armendáriz de Aghion & Morduch, 2010). Current trends in the microfinance sector, however, are characterized by increasing patterns of microfinance integration with the financial system. Even though it is difficult to determine the exact contributing factors to this integration, commercialization of the sector and increased transaction costs are inarguably among the main factors. In the last two decades, many microfinance institutions (MFIs) have been financially challenged when the aid and funding from development agencies were declined or even stopped. As a result, to sustain their operation MFIs have transformed themselves into commercial institutions and pursued profits. Transaction costs, another possible contributing factor, increase as the MFIs become bigger and have to manage larger portfolios. To reduce transaction costs, the MFIs tend to focus on the financially more capable population, i.e. the middle class, instead of the poor. This together with the reduced aid and funding from donors pushed the MFIs to deviate from their original purpose of serving the poor. Consequently, microfinance is no longer regarded as informal financial intermediation funded by aid and donor organizations; rather, it has become a low-end subsector of the broader financial system (Maksudova, 2010). As part of the overall financial system, microfinance is expected to be related, to a certain degree, to its associated sector, e.g. the banking sector. The evolution of microfinance has raised some interesting questions. Most importantly, does microfinance contribute to the well-being of the population? And what is the relationship between microfinance and banking sector development? Does microfinance stimulate banking sector development, or is banking sector development a prerequisite for microfinance development? In order to answer these questions, we aim to determine whether there is any long- and/or short-run relationship between microfinance sector development, banking sector development, and the economic well-being of people. There is a large number of theoretical and empirical papers that endeavors to define the relationship between financial and economic development (see e.g. King & Levine, 1993a; Demetriades & Hussein, 1996; Levine, 1997; Beck et  al., 2000). Additionally, there is ample literature that deals with the impact of microfinance on economic well-being (see e.g. Khandker et al., 1998; Copestake et al., 2001; Banerjee et al., 2015). However, there are only a few papers dealing with the interaction between banking and microfinance development 425

426  Handbook of microfinance, financial inclusion and development

(see e.g. Charitonenko et al., 2004; Vanroose & D’Espallier, 2013). The literature that investigates the causal relationship between banking development, microfinance development, and economic growth is even more restricted (but see e.g. Buera et al., 2012; Alimukhamedova, 2013). The main contribution of our paper is to provide new evidence on the short and longrun relationships between financial development, microfinance, and economic growth using cointegration analysis. Our analysis focuses on Mongolia, a country with a vibrant banking and microfinance sector. While the external validity of any study is unclear ex-ante, we believe that our study is of interest to a broader set of countries, especially countries with the same development stage as Mongolia. The rest of the chapter is organized as follows: the second section gives background information about Mongolian banking as well as microfinance sector development. The third section reviews the literature that is closely related to the topic of this chapter; the section also develops the main framework concerning the interactions between economic well-being and banking and microfinance sector development. The fourth section describes data sources and the measurement of the main variables. The fifth section deals with the main methodological issues, including the empirical model specification. The sixth section shows the main findings and discusses the causal relations between microfinance, the banking sector, and economic growth. Finally, the seventh section contains our concluding comments.

FINANCIAL SYSTEM IN MONGOLIA The underlying infrastructure of the Mongolian financial system is similar to that of many transitional economies, which are mostly based on the liberalization and privatization of financial institutions of a former centrally planned economy. One of the main features of the current financial system of Mongolia is that there are a large number of low-income individuals and small enterprises on the demand side, whereas a few commercial banks (i.e. 14 commercial banks as of 2016) dominate the supply side. Another characteristic is that its capital markets have developed poorly since its transition to a market economy. Only 258 out of around 55,000 companies are listed on the Mongolian Stock Exchange, and only 30 of the listed companies make up around 90 percent of the market capitalization as of 2016 (National Statistics Office of Mongolia, 2017). Moreover, almost all unlisted companies are categorized as micro, small, or medium enterprises; hence, all of their financial sources come from the banking sector. We, therefore, define the financial system of Mongolia as a bank-based system and use banking sector indicators to quantify the financial sector development. The establishment of savings and credit cooperatives (SCC) and nonbanking financial institutions (NBFI) in 1996 and 1999, respectively, can be considered as the starting point of the microfinance movement in Mongolia. But until the establishment of the Financial Regulatory Commission of Mongolia (FRCM) in 2006, there was no prudential regulation for these types of financial institutions. Since 2006, the FRCM has classified SSCs and NBFIs as MFIs and started monitoring all MFIs, except for pawnshops. During the last two decades, the microfinance sector of Mongolia has grown enormously. Some growth indicators between 2006 and 2016 are shown in Figure 22.1. Despite the expansion of the sector, the microfinance sector was affected by economy-wide, external shocks, such as the 2000s commodities bubble and the financial crisis of 2007–2009. Fortunately, the development of the Mongolian social welfare system during the shocks has improved the financial accessibility of the population. In a report of the World Bank (2015),

Does microfinance cause banking sector development and economic growth?  427

Figure 22.1 Microfinance sector growth in Mongolia for instance, it was stated that the bank account penetration rate of the population over the age of 14 years, an indicator for access to finance, in Mongolia is 92 percent, which is as high as that of high-income countries expressly 95 percent.

LITERATURE REVIEW Before studying whether microfinance is integrated with economic growth and/or the formal banking sector, we review the literature on the traditional nexus between economic growth and financial development. The connection between economic growth and financial development has been extensively studied since Schumpeter (1934)1 first emphasized the causal effects between them. His work is followed by many studies. The most notables are Robinson (1952) and Goldsmith (1969), who argue that finance does not cause economic growth, instead “finance follows the real sector.” In contrast to this “finance follows” view, Gurley and Shaw (1955), McKinnon (1973), Shaw (1973), and their disciplines contend that finance contribution to the real sector should not be neglected in economic growth research. Despite ongoing discussions, there is considerable empirical evidence that the financial sector provides a positive contribution toward economic growth and vice versa (see e.g. King & Levine, 1993a; Demetriades & Hussein, 1996; Levine, 1997; Rajan & Zingales, 1998; Beck et al., 2000). Yet, there is also some evidence that the causality between finance and growth differs across countries. Demetriades and Hussein (1996), for instance, apply cointegration and causality tests between financial development and economic growth to 16 economies separately. Although they find significant evidence of bidirectionality and some evidence of reverse causation, they highlight the fact that causality directions may vary across countries. Furthermore, many researchers explain the theoretical background to the nexus as the functions of the financial system in an economy (see e.g. King & Levine, 1993a; Merton, 1995; Levine, 1997). Levine (1997, 2005), for instance, categorizes the regulatory processes of the financial system, through which financial development affects economic growth, into five functions. His classification is principally in line with the six core functions of financial systems defined by Merton (1995). Besides the huge literature on finance and growth, there is ample literature on the impact of microfinance on poverty alleviation at the micro- and middle levels. The outcomes of this

428  Handbook of microfinance, financial inclusion and development

literature consistently suggest that microfinance has not only a direct impact on household well-being but also an indirect impact on economic growth via poverty reduction (Khandker, 2005; Maksudova, 2010; Buera et al., 2012; Alimukhamedova, 2013; Banerjee et al., 2015). For example, Khandker (2005) investigates the impact of microfinance on poverty reduction using panel data from Bangladesh and concludes that access to microfinance contributes to poverty reduction both at the participant level and in local economic growth. Buera et  al. (2012) theoretically model the relation between microfinance and economic growth and conclude that the vast majority of the population benefits from microfinance through an increase in wages. Conversely, there is also literature that tests whether the microfinance sector is influenced by its macroeconomic environment (Ahlin et al., 2011; Silva & Chávez, 2015; Xu et al., 2016). Ahlin et al. (2011), for instance, examine whether and how MFIs are influenced by the macroeconomic and macro-institutional characters using panel data on 373 MFIs from different economies. They conclude that MFIs are significantly affected by the macroeconomic and macro-institutional environment in which the MFI is situated. While there is some evidence supporting the interactions between microfinance and economic growth, much more research is needed (Armendáriz de Aghion & Morduch, 2010). This study is an attempt to connect the two large literatures on finance and growth and microfinance and growth by applying cointegration analysis to microfinance, financial development, and economic growth. To develop the theoretical background to the issue at hand, we scale down the functions of the financial sector for the microfinance sector as follows: (i) facilitating risk by introducing nontraditional but more affordable products for the poor that can reduce portfolio volatility,2 (ii) allocating funds to a larger population, especially to the poor, (iii) monitoring the imbursement of funds by implementing more flexible regulations, e.g. group lending, (iv) pooling funds from international donors as well as for-profit organizations, and (v) easing the exchange of goods through its immediate link to the formal banking sector. Through these functions, microfinance can contribute to economic growth directly by increasing household income as well as indirectly by supporting banking sector development. The relationships between microfinance, the banking sector, and economic growth are demonstrated in Figure 22.2. In Figure 22.2, the two arrows between economic growth and banking sector development denote the “finance follows” (Robinson, 1952) and “finance leads” (McKinnon, 1973) viewpoints of the finance-growth debate. The arrows linking economic growth and microfinance show the contribution of the microfinance sector to economic growth or vice versa. The connections between banking and microfinance sector development imply the directions of possible causality.

Figure 22.2

Possible interactions

Does microfinance cause banking sector development and economic growth?  429

DATA AND MEASUREMENTS Data Sources We use quarterly time series data on microfinance and banking sector performances and economic growth of Mongolia, covering the period between 2006 quarter 4 to 2016 quarter 3. The data on macroeconomic indicators, such as real GDP and population, are obtained from the National Statistics Office of Mongolia (NSOM). The main performance indicators of MFIs are obtained from the quarterly bulletins on the performances of NBFIs and SCCs issued quarterly by the Financial Regulatory Committee of Mongolia (FRCM). Indicators on the performance of the banking sector are downloaded from the website of the Bank of Mongolia (BoM) and calculated where necessary.3 As the data are quarterly and the main sectors of the Mongolian economy heavily depend on seasonality, some flow variables we use, for instance, real GDP, have a clear quarterly seasonal pattern.4 We therefore seasonally adjusted these series. We use a seasonal adjustment program called CAMPLET, which was developed by Abeln and Jacobs (2015).5 Table 22.1 summarizes the dataset after the seasonal adjustment to selected series. As shown in Panel A of Table 22.1, the banking sector outperforms the microfinance sector in terms of monetary performance, such as gross loan portfolio, total assets, and income, etc. From Panel B, it is seen that MFIs make up almost 30 percent of total financial service units. Similarly, the number of MFIs clients and members constitutes about 30 percent of total individual clients of both the banking and microfinance sectors. This intuitively implies that there might be a potential competition between the two sectors for individual borrowers or clients. But it also can be due to the segmentation of the market. Furthermore, we can simply estimate that the average loan amount of microfinance institutions, MNT 766,485, is almost six times smaller than commercial banks, MNT 4,394,593, indicating that MFIs mostly serve poorer people. Measurements GDP per capita is still the most popular measure for assessing the overall economic performance of countries. Hence, in this study, we follow the widely accepted standard practice and proxy economic well-being by the quarterly growth rate of real GDP per capita. Unlike the GDP, there is no commonly accepted single measure that quantifies financial sector development. Recent meta-analyses by Arestis et al. (2015) and Valickova et al. (2015) reveal that most of the empirical research on financial development only uses single measures to assess financial sector development, including the ratio of private credit by the banking sector to GDP, the ratio of broad money to GDP, and the ratio of stock market capitalization to GDP (see e.g. King & Levine, 1993b; Demetriades & Hussein, 1996; Levine, 1997; Cull et al., 2007). However, these measures, whether they are used independently or jointly, cannot fully capture all aspects of financial development as the financial sector consists of a variety of financial institutions, markets, and services. To overcome the drawback of using a single indicator, Čihák et al. (2012, 2013) propose to use the Global Financial Development Database (GFDD) which introduces more than 100 indicators categorized into four dimensions of financial development: (a) size of financial institutions and markets (financial depth); (b) degree to which individuals can and do use financial services (access); (c) efficiency of financial institutions and markets in intermediating resources and facilitating financial transactions

430  Handbook of microfinance, financial inclusion and development

Table 22.1  Summary statistics of the series Series

Obs. Mean

Std. dev.

Min

Max

42

3,519,641

1,837,204

978,724

6,382,144

42

2,964,168

808,476

1,755,340

4,314,965

Panel A: Monetary series (million MNT) Nominal GDP † Real GDP, 2010 price

†

Broad money

42

6,107,284

3,475,671

1,536,493

12,100,000

Domestic credit to private sector

42

3,529,878

2,258,683

607,338

6,805,825

Gross loan portfolio to individuals

42

2,796,438

1,982,357

461,242

5,779,645

Gross loan portfolio of NBFIs

41

157,877

131,661

28,324

447,716

Gross loan portfolio of SCCs

40

44,619

19,351

11,814

81,695

Total assets of banks

42

11,300,000

8,171,089

1,814,490

25,300,000

Total assets of NBFIs

41

259,617

206,338

57,818

724,154

Total assets of SCCs

40

60,838

23,375

16,545

108,775

Total equity of banks

42

1,038,196

868,968

230,212

2,936,043

Total equity of NBFIs

41

176,188

141,756

46,043

531,078

Total equity of SCCs

40

13,657

5,202

2,861

22,766

Net income of banks

41

30,527

54,874

–129,272

220,702

Net income of NBFIs

41

5,035

4,762

328

19,227

Net income of SCCs

40

696

436

–371

1,657

Total population‡

42

2,822,123

165,699

2,574,366

3,118,531

Adult population‡

42

1,869,855

126,522

1,650,393

2,012,983

Panel B: Non-monetary items

Number of individual borrowers

42

636,336

174,678

415,348

938,778

Number of clients of NBFIs

41

237,238

203,246

29,752

663,098

Number of members of SCCs

40

26,950

7,991

5,150

44,890

Number of bank branches‡

42

1,233

228

794

1,492

‡

Number of NBFIs

41

309

142

139

632

Number of SCCs

40

180

50

52

277

Note:  † denotes seasonally adjusted series. ‡ denotes interpolated and/or extrapolated series MNT is the currency abbreviation for the Mongolian tugrug (or tögrög), the currency for Mongolia

(efficiency); and (d) stability of financial institutions and markets (stability). Using this database, Sahay et al. (2015) and Svirydzenka (2016) construct a broad-based, composite index of financial development, using a methodology developed by Nardo et al. (2005, 2008). We follow their (Čihák et al., 2012; Sahay et al., 2015; Svirydzenka, 2016) methodologies and construct two separate broad-based indicators, one for banking sector development and the other for microfinance sector development. To build the broad-based indicators, we first select the required indicators from the GFDD. The selected indicators and their summary statistics are shown in Table 22.2.

Does microfinance cause banking sector development and economic growth?  431

Table 22.2  The summary statistics for the selected indicators Variables

Measures

GDP per capita growth

Obs

Mean

Std. dev.

Min

Max

40

0.0142

0.0514

–0.1287

0.1485

Financial sector (commercial banks)   Return on assets

Efficiency

40

0.0017

0.0082

–0.0354

0.0103

  Return on equity

Efficiency

40

0.0161

0.1063

–0.4055

0.1267

  Capital adequacy

Stability

40

0.0897

0.0246

0.0521

0.1408

 Z-score

Stability

40

11.1037

3.3086

4.4887

17.9039

  Bank branches per 100,000 adults

Access

40

66

8

50

74

  Individual borrowers per 100,000 adults

Access

40

33,579

7,099

22,979

46,796

  Ratio of broad money to GDP

Depth

40

1.6833

0.1494

1.3498

2.0031

  Ratio of private sector credit to GDP

Depth

40

0.9478

0.1671

0.5898

1.2136

  Ratio of total assets of banks to GDP

Depth

40

2.8604

0.7553

1.7000

4.2741

Efficiency

40

0.0160

0.0047

0.0060

0.0243

  Return on equity

Efficiency

40

0.0276

0.0078

0.0100

0.0392

  Capital adequacy

Stability

40

0.5768

0.0394

0.4979

0.6761

 Z-score

Stability

40

125.1545

8.6847

107.1168

146.3340

  Number of MFIs per 100,000 adults

Access

40

26

8

12

45

  Number of MFI clients per 100,000 adults

Access

40

13,829

10,179

2,314

34,499

  Ratio of total assets of MFIs to GDP

Depth

40

0.0860

0.0198

0.0657

0.1444

  Ratio of loan portfolio of MFIs to GDP

Depth

40

0.0530

0.0142

0.0374

0.0918

Microfinance sector (MFIs)   Return on assets

After calculating the indicators, we apply principal component analysis (PCA) to derive some broad-based indicators. The derivation of the indicators is demonstrated in Appendix 22.1.

METHODOLOGY In time series analysis, causality tests are preceded by cointegration testing because the existence of cointegration has important implications for causality among the selected variables (Sims et al., 1990). Moreover, carrying out cointegration tests before causality tests provides evidence concerning the existence of a long-run relationship between the variables. Although there are many cointegration techniques for different situations, most cointegration tests are traditionally performed using the Engle and Granger (1987) two-step approach and/or the Johansen (1988) maximum likelihood method. Both methods can be used in situations where a “dependent” variable and a set of “independent” variables cannot be specified explicitly. This is a common advantage of the methods if the variables are jointly specified and the interdependence among them is obscure (Enders, 2009). Cointegration Testing Approaches The Engle and Granger (1987) two-step approach is the simplest cointegration technique that estimates the long-run equilibrium relationship from a regression of Xt on Yt or vice versa.

432  Handbook of microfinance, financial inclusion and development

Yet it also has some drawbacks (Enders, 2009). First, the ordering of variables can affect the stationarity of the residuals for a small sample, which may bias the OLS estimates of the cointegration vector. Second, the technique is only valid for two variables that are integrated in the same order. Third, it relies on a two-step procedure.6 The Johansen (1988) maximum likelihood method, however, can detect more than one cointegrating relationship and give asymptotically efficient estimates of the cointegrating vectors and the speed of adjustment. But also the Johansen method has its own limitations. First, the result of the test is sensitive to lag length. Second, the results of the two test statistics offered by the method, i.e. the trace test and maximum eigenvalue test, may conflict in some circumstances. Third, the Johansen approach is subject to asymptotic properties (i.e. large samples) and simultaneity bias among the regressors. Due to the asymptotic or large sample restriction imposed by the two methods, we adopt the bounds testing approach that uses the autoregressive distributed lag (ARDL)7 model for cointegration proposed by Pesaran and Shin (1999) and Pesaran et al. (2001). This cointegration technique has many advantages over the conventional cointegration methods. First, it can be applied to cointegration analysis for a long-run relationship irrespective of whether the variables in the model are level stationary I ( 0 ), first difference stationary I (1) , or a mixture of both. Second, the ARDL model can have a single equation setup, allowing simpler implementation and interpretation. Since each of the variables stands as a single equation, the residuals are uncorrelated, which means that endogeneity is less of a problem. Third, the lagged variables in the model can have different lag lengths. Last but not least, the error correction form of the ARDL model is more efficient when there is a single long-run relationship among the underlying variables and the sample size is small or finite (Narayan & Smyth, 2005; Nkoro & Uko, 2016). The bounds testing approach for cointegration can be summarized in three steps: (1) Detect the presence of a long-run relationship for the variables under consideration. An unrestricted error correction model (ECM) is formulated for each of the variables under analysis. For instance, if there are two variables, y and x, the relevant unrestricted ECMs are formulated as follows: m



Dyt = a 0 +

å

a1Dyt -i +

i =1

m



Dxt = b 0 +

å

b1Dyt -i +

i =0

m

åa Dx

t- j

2

+ g 1 yt -1 + g 2 xt -1 + m1t (22.1)

j =0

m

åb Dx 2

t- j

+ d1 xt -1 + d 2 yt -1 + m2t . (22.2)

j =1

In equations (22.1) and (22.2), the selected maximum lag order, m, depends on the purpose of the analysis and the data frequency. For each model, the lagged levels of all variables, yt−1 and xt−1, are jointly tested with an F-test under the null hypothesis of no cointegration among the underlying variables. The asymptotic distribution for this F-test, however, is non-standard, and thus Pesaran et al. (2001) provide two sets of critical values for a different number of explanatory variables and a given significance level.8 One set is based on the assumption that the underlying variables of the ARDL model are I(0),

Does microfinance cause banking sector development and economic growth?  433

i.e. the lower bound, whereas the other assumes that the variables are I(1), i.e. the upper bound. If the result of the test statistic is greater than the upper critical bound value, then the null hypothesis of no cointegration is rejected, i.e. the variables are cointegrated. If it is smaller than the lower bound, then the null hypothesis cannot be rejected, i.e. there is no cointegration. If the computed F-statistic falls between the two bounds, then the cointegration test is inconclusive. (2) Once a long-run relationship is detected, the optimal lag length for each variable of the selected model needs to be determined. The optimal lag orders of the first differenced variables, Dxt - j and Dyt -i , are obtained by applying the Akaike Information Criterion (AIC) and the Schwartz Bayesian Criterion (BIC) to all ECMs. The number of models to be tested by the information criteria depends on the number of variables as well as the maximum lag length selected in the first step. As a result, an unrestricted ECM that has the optimal lag length is selected, from which the long-run coefficients are obtained. (3) In the last step, the conventional (or restricted) ECM is estimated, and the short-run dynamics and the adjustment towards the long-run equilibrium are obtained from it. Relevant diagnostic tests on the stability of the model and on the serial correlation among the residuals are applied in order to examine the robustness of the specified ARDL model. The Empirical Model Specification Our model is specified as follows:

GDPt = f ( FDt , MFt ) (22.3)



FDt = f ( MFt , GDPt ) (22.4)



MFt = f ( FDt , GDPt ) (22.5)

where GDPt is the economic well-being of the population measured by the growth rate of the quarterly GDP per capita; MFt and FDt denote the microfinance sector and the bank-based financial sector development indicators, respectively. The relationships can be expressed in level ARDL ( p, q1, q2 ) form as follows: p



GDPt = a 0 +

å

a1iGDPt -i +

i =1

FDt = b 0 +

MFt = g 0 +

q1

q2

å

p

q1

i =0



åa MF

å

b1iGDPt -i +

t -i

i =0

t -i

+ e1t (22.6)

åb MF 3i

t -i

+ e 2t (22.7)

i =0

i =1

1i

3i

i =0

b 2i FDt -i +

åg GDP + åg i =0

q2

å

a 2i FDt -i +

i =0

p



q1

2i

FDt -i +

q2

åg i =1

3i

MFt -i + e 3t (22.8)

434  Handbook of microfinance, financial inclusion and development

In equation (22.6), α0 is a constant, α1−α3 are respective regression coefficients, p and qi are the lag orders in the level ARDL model, t is the time period (t = 1, 2, ¼, T ), and εit is the white noise error term, where e t ~ IID ( 0, s 2 ) . The unrestricted ECM form of the level ARDL model (22.6), (22.7), and (22.8) is modeled as follows: p -1



DGDPt = J0 +

å

J1i DGDPt -i +

i =1

q1 -1

å

J2i DFDt -i +

q2 -1

åJ DMF 3i

t -i

i =0

i =0

(22.9)

+ JGDPGDPt -1 + JFD FDt -1 + JMF MFt -1 + m1t



DFDt = j0 +

q2 -1

q1 -1

p -1

åj DGDP + åj DFD + åj DMF 1i

2i

t -i

3i

t -i

t -i

i =0

i =1

i =0

(22.10)

+ jGDPGDPt -1 + j FD FDt -1 + j MF MFt -1 + m2t p -1



DMFt = w0 +

å

w1i DGDPt -i +

q1 -1

å

w2i DFDt -i +

åw DMF 3i

i =1

i =0

i =0

q2 -1

t -i

(22.11)

+ wGDPGDPt -1 + wFD FDt -1 + wMF MFt -1 + m3t In quation (22.9), ϑ 0 is a constant; ϑ1−ϑ3 represents the relative short-run coefficients; ϑ GDP, ϑFD, and ϑ MF are the long-run coefficients; Δ is the difference operator; p and qi are the lag orders in the level ARDL model; t is the time period (t = 1, 2, ¼, T ); and μt is the white noise error term, where mt ~ IID ( 0, s 2 ) . To detect a cointegrating relationship, we apply the bounds testing approach to the unrestricted ECMs, namely equations (22.9), (22.10), and (22.11). After implementing the bounds testing procedure, we modify the model by including different lag orders for each variable. Although the optimal lag selection (or lag modification) procedure can be done by employing either the AIC or the BIC, Pesaran and Shin (1999) highlight the superiority of the latter over the former. This lag selection procedure evaluates all possible versions for the given number of variables and lag lengths and gives us an appropriate ARDL model. We obtain the long-run coefficients from the selected model. Then we need to reparametrize the model in order to derive a conventional or restricted ECM. From the ECM, the short-run dynamics and the adjustment towards the long-run equilibrium are obtained. The conventional error correction equivalents of equations (22.9), (22.10), and (22.11) are as follows: p -1

DGDPt = J0 +

å

J1i DGDPt -i +

i =1



åJ DMF 3i

i =0

åJ DFD 2i

i =0

q2 -1

+

q1 -1

t -i

+ q1ECTt -1 + m1t

t -i

(22.12)

Does microfinance cause banking sector development and economic growth?  435 q1 -1

p -1

DFDt = j0 +

åj DGDP + åj DFD 1i

t -i

i =0



q2 -1

åj DMF

+

2i

t -i

i =1

3i

t -i

(22.13)

+ q 2 ECTt -1 + m 2t

i =0

q1 -1

p -1

DMFt = w0 +

åw DGDP + åw DFD 1i

t -i



q2 -1

+

åw DMF 3i

2i

i =0

i =0

t -i

t -i

(22.14)

+ q3 ECTt -1 + m3t

i =1

where ECTt−1 are the error correction terms and θ1−θ3 are the speed-of-adjustment parameters. The larger θ1, for instance, means the greater the response of GDP to the previous period’s deviation from long-run equilibrium. The rest of the variables are the same as described in the unrestricted ECMs. The dynamic stability of the models and the serial correlation and heteroscedasticity of the variables are evaluated at each step of the model selection procedure via relevant tests. We used Stata version 13 (StataCorp, 2013), and user-written Stata packages, namely kpss (Baum, 2000b), cumum6 (Baum, 2000a), and ardl (Kripfganz & Schneider, 2016) to complete our work.

RESULT AND DISCUSSION The descriptive statistics and correlation matrix of the variables are reported in Table 22.3. The correlation results suggest a strong and positive correlation between banking and microfinance Table 22.3  Descriptive statistics and correlation matrix GDP per capita growth

Banking sector development

Microfinance sector development

Mean

0.0142

0.0000

0.0000

Variance

0.0026

1.3895

1.8103

Standard deviation

0.0514

1.1788

1.3455

Skewness

–0.0644

–0.2542

0.8012

Kurtosis

3.6708

2.2473

2.8910

Correlation with GDP

1.0000

–0.2396

–0.2593

Correlation with FD

–0.2396

1.0000

0.7043*

Correlation with MF

–0.2593

0.7043*

1.0000

Note:  * denotes significance level at 10 percent

436  Handbook of microfinance, financial inclusion and development

sector developments. But the correlation between economic growth and the banking sector, as well as the microfinance sector is negative and insignificant. Results of Cointegration Tests We start the analysis by conducting unit roots tests. Even though the ARDL approach does not require pretesting for unit roots, the unit root tests provide guidance as to whether the ARDL bounds testing approach is applicable. The approach is not applicable to the analysis of variables that are integrated of order two, i.e. I ( 2 ) or above (Pesaran et al., 2001). The conventional stationarity tests, such as the Dickey-Fuller (DF) test and the PhilipPerron (PP) test, suffer from size and power problems (see Perron & Ng, 1996), and thus are not reliable for a small sample size (Harris & Sollis, 2003). We therefore use the augmented Dickey-Fuller (AFD) test and the Kwiatkowski-Phillips-Schmidt-Shin (KPSS). The AFD test considers a null hypothesis of non-stationarity. In contrast, the KPSS tests the null hypothesis that the series are stationary. The tests are applied with and without a trend, The degree of augmentation for the ADF test is based on the BIC. The bandwidth or the lag length in the KPSS tests is based on the Newey-West method coupled with the quadratic spectral kernel. The combination is preferable, due to its better small sample properties (Hobijn et al., 2004). The results of the two tests are reported in Table 22.4. Table 22.4 indicates that GDP is level stationary or I ( 0 ), while FD (bank-based financial development) and MF (microfinance sector development) become stationary after first differencing or I (1) . The results of both tests are consistent and confirm that the series is a mix of I ( 0 ) and I (1) variables. As none of the series were found to be I ( 2 ) , we can apply the ARDL bounds testing approach to detect cointegrating relations among the variables. This approach tests the null hypothesis of the non-existence of cointegration, defined as H 0 : JGDP = JFD = JMF = 0, against the alternative hypothesis of cointegration, expressed as H a : JGDP ¹ JFD ¹ JMF ¹ 0. We use relevant critical values for the upper and lower bounds from Narayan (2005), which are suitable for small samples. Table 22.5 presents the critical values for simulation Case III (Narayan, 2005). Table 22.4  The results of the stationarity tests Variable

Levels With trend

First differences Without trend

With trend

Without trend

Augmented Dickey-Fuller (ADF) test GDP

9.281***

9.301***

—

—

FD

2.410

1.256

9.086***

9.070***

MF

3.054

0.077

11.226***

10.500***

Kwiatkowski-Phillips-Schmidt-Shin (KPSS) test GDP

0.125**

0.289***

—

—

FD

0.234

1.130

0.069***

0.119***

MF

0.292

1.390

0.077***

0.484**

Note:  *, **, and *** denote stationarity at the 10 percent, 5 percent, and 1 percent significance levels, respectively

Does microfinance cause banking sector development and economic growth?  437

Table 22.5  The critical values for the F-statistic Critical values at

Lower bound, I(0)

Upper bound, I(1)

10%

3.373

4.377

5%

4.133

5.260

1%

5.893

7.337

Note:  Case III: unrestricted intercept and no trend. k = 2, the number of dependent variables. n = 40, the number of observations

Table 22.6  The results of the ARDL bounds testing approach Model

Optimal lag length

F-statistics

t-statistic

f (GDPt | FDt , MFt )

ARDL(1,0,0)

13.204***

–6.230***

f ( FDt | MFt , GDPt )

ARDL(2,0,3)

4.653*

–2.088

f ( MFt | FDt , GDPt )

ARDL(3,3,0)

4.903*

0.497

Note:  *, **, and *** denote stationarity at the 10 percent, 5 percent, and 1 percent significance levels, respectively

We apply the ARDL bounds testing to the models specified in equations (22.9), (22.10), and (22.11) independently. In each model, one of the variables acts as the dependent variable, whereas the other two stand as independent variables. Since our data is quarterly, we chose k +1 four lags as the maximum lag order and calculated 125 equations (or (1 + m ) = 53 ) with different combinations of lags for each model. To select the optimal lag order for each variable, we applied the BIC to the equations. The results of the lag selection and bounds testing procedures are reported in Table 22.6. The F-statistic for dependent variable GDP is 13.204, which is greater than the upper bound critical value of 7.337 at the 1 percent significance level reported in Table 22.5. In addition, the F-statistics for FD and MF models are 4.653 and 4.903, respectively, which are marginally greater than the upper bound critical value of 4.377 at the 10 percent significance level. The results, therefore, suggest that there is at least one cointegrated relation among the variables. ARDL Model Estimates With the variables being cointegrated, we can express the model in equations (22.9), (22.10), and (22.11) in a conventional error correction form, i.e. equations (22.12), (22.13), and (22.14), from which the long- and short-run coefficients are obtained. Table 22.7 reports the estimation results of the models. Panel A shows the long- and short-run coefficients, and their speed of adjustment. Panel B demonstrates the results of the required diagnostic tests, namely the Breusch-Godfrey test and Durbin’s alternative test for first and higher-order serial correlation, the Breusch Pagan test for heteroscedasticity, the Ramsey RESET test of model specification, and the mean variance inflation factor (VIF) for collinearity. The results of the GDP model in Table 22.7 indicate that microfinance and financial sector development have no statistically significant effect on economic well-being in the long and short run. Nonetheless, the statistically significant error correction term (ECT) confirms the

438  Handbook of microfinance, financial inclusion and development

Table 22.7  The results of the estimated ARDL models Models

GDP model

FD model

MF model

Dependent variable

DGDPt

DFDt

DMFt

Optimal lags

ARDL (1, 0, 0 )

ARDL ( 2, 0, 3)

ARDL ( 3, 3, 0 )

Panel A: Variables

Coefficient

t-stat.

FDt -1

–0.002

(–0.21)

MFt -1

–0.008

(–1.24)

Coefficient

t-stat.

Coefficient

t-stat.

–4.918

(–0.43)

–5.171

(–0.12)

–0.013

(–0.12)

–0.637***

(–4.70)

–0.353**

(–2.53)

DMFt -1

–1.004***

(–5.86)

DMFt - 2

–0.456***

(–3.15)

0.207

(0.12)

0.354***

(4.80)

Long-run results

GDPt -1

1.309***

(2.85)

78.486*

(1.77)

Short-run results DFDt

–0.002

(–0.21) –0.306*

DFDt -1

(–1.88)

DFDt - 2 DMFt

–0.009

(–1.23)

0.294***

(3.03)

DGDPt

1.153

(0.51)

DGDPt -1

–13.342***

(–3.82)

DGDPt - 2

–6.291***

(–3.29)

0.016**

(2.17)

–0.241*

(–1.97)

–1.120***

(–6.23)

–0.225**

(–2.09) 0.04

(0.50)

Diagnostic tests

Statistics

P-val.

Statistics

P-val.

P-val.

Breusch-Godfrey (χ2)

0.267

[0.605]

2.437

[0.088] 0.206

[0.650]

Durbin’s alternative (χ2)

0.232

[0.630]

1.897

[0.168]

0.150

[0.699]

Breusch-Pagan (χ2)

1.590

[0.207]

0.200

[0.659]

0.570

[0.451]

Ramsey RESET (F) 0.390

[0.760]

2.160

[0.118]

0.880

[0.465]

Mean VIF

—

3.600

—

2.090

—

Constant Adjustment term ECTt−1 Panel B:

1.740

Statistics

Note:  *, **, and *** denote stationarity at the 10 percent, 5 percent, and 1 percent significance levels, respectively

Does microfinance cause banking sector development and economic growth?  439

bounds testing result. The ECT of –1.12 implies that the relationship among the variables corrects its previous period disequilibrium at a speed of 112 percent.9 In other words, the discrepancy in the equilibrium is corrected at a substantial speed of adjustment in the following period, in which case, reaching long-run equilibrium may take a longer time to achieve a steady state. This implies that potential adverse effects on the economic growth of the microfinance sector and/or banking sector downturns will persist for a long time. In that case, a credit may become a burden to the population, instead of increasing their economic well-being. Thus, some form of government intervention, either implementing tighter prudential regulations in the financial system or expanding its welfare program to the most vulnerable section of the population, may be required. From the FD model results, we conclude that the microfinance sector has a statistically significant positive impact on the banking sector development, both in the short and long run. The model results further suggest that economic growth has a significant positive impact at the 10 percent significance level on the banking sector in the long run. This is in line with Robinson (1952) and Goldsmith (1969), who argue that “finance follows the real sector.” But in the short run, economic growth has a statistically significant and negative effect on the banking sector’s development. This reason may be that the banking sector could not cope with the economic growth spurt due to the commodity market boom of the 2000s. The significant ECT of –0.225 means that one unit deviation from the model equilibrium is corrected in the next period at a speed of 22.5 percent per quarter. The results of the MF model show that in the long run there is neither a significant effect of economic growth nor banking sector development on microfinance sector development. This is confirmed by the positive ECT meaning that the error correction process is not converging in the long run. However, in the short run, the lagged values of the banking sector development have significant and negative effects on the microfinance sector. Thus while the microfinance sector supports the banking sector, the banking sector negatively affects the microfinance sector. A potential explanation for this result is that the banking sector acts as wholesale bankers and lends money to the microfinance sector, but, on the other hand, the two are competitors in the same consumer credit market. The diagnostic test results reported in Panel B show that all three models pass the relevant tests for serial correlation, heteroscedasticity, and model specification. The dynamic stability of the models is analyzed by the cumulative sum of recursive residuals (CUSUM) and the cumulative sum of squares of recursive residuals (CUSUM squared). The plots for the CUSUM and the CUSUM squared are reported in Appendix 22.2. The plots indicate that the long- and short-run estimates of the model are dynamically stable at the 5 percent significance level. The slight deviation in the FD model plot may be an indication of the fact that the model should be specified in a non-linear form.

CONCLUSION This chapter tests whether there are statistically significant and economically meaningful relationships between economic well-being, banking sector development, and microfinance sector development in Mongolia. We employ a principal component analysis to derive indices for microfinance development and banking sector development, and we use the quarterly GDP per capita growth rate to quantify the economic well-being of people. The ARDL bounds testing approach to cointegration proposed by Pesaran et al. (2001) is used to examine interactions between microfinance, the banking sector, and economic growth. The analysis suggests

440  Handbook of microfinance, financial inclusion and development

that there is one cointegrated relationship among the variables at the 1 percent significance level. We also find evidence for a significant bidirectional causal relation between microfinance and banking sector development in the short run. The result suggests that in the short run the microfinance sector supports the banking sector while the banking sector adversely affects microfinance sector development. Microfinance development also affects the long-run development of the banking sector, and yet there is no significant impact of the banking sector on microfinance in the long run. This finding can be interpreted as that the microfinance sector is one of the clients of the banking sector at the wholesale banking level, and thus contributes to the banking sector development in the long run. But the sectors compete at the retail banking level in the short run. Neither the banking nor microfinance sector has an impact on economic growth both in the short and long run. We yet find evidence in favor of the “finance follows” view as there seems to be a unidirectional causal relation between economic growth and banking sector development both in the long and short run. A fruitful area for future research is to test the external validity of our results. While a priori we don’t have any reasons to believe that our results will not hold for other countries, additional research is needed to confirm this.

ACKNOWLEDGMENT We would like to acknowledge the ALFABET project under the European Commission’s Erasmus Mundus Partnerships with the grant number 2014-0857/001-001 for funding the research.

NOTES 1.

In his seminal work in 1911, J. A. Schumpeter highlights the services offered by financial intermediaries as a crucial element in innovation and development. 2. According to the financial concept of risk/return trade-off, a high risk investment is regulated by a high risk premium, which increases the rate of return; moreover, different mechanisms and techniques in microfinance, such as village banking and group lending, allow the sector to have a low default rate. 3. If the series, including the number of commercial bank clients and bank branches, are constantly increasing over the years, the missing data points are completed by applying interpolation and extrapolation methods. Since the methods being used in the derivation of quarterly data are based on annual data, the general pattern of the quarterly data points does not diverge much from that of the annual ones. Consequently, we assume that the application of the methods are justifiable in this case. 4. Even the additional information in the seasonal component can provide new information that will improve model performance. It may also complicate the model, especially in the presence of the seasonal unit root. Thus, identifying and removing the seasonal component from the series allow for a better analysis and can provide a clearer interpretation regarding the relationship between input and output variables. 5. Alternative method is X-12-ARIMA, developed at the US Census Bureau that is based on the Bureau’s earlier X-11 program and the X-11-ARIMA/88 program developed at Statistics Canada. 6. In step one, the residuals of the specified model are saved. In step two, the residuals are tested for stationarity using the Dickey-Fuller test. If the residuals are stationary, the error correction model is estimated. 7. In the ARDL model, lagged values of the dependent variables (i.e., autoregressive) as well as both current and lagged values for the independent variables (i.e. distributed lags) are included.

Does microfinance cause banking sector development and economic growth?  441

8. The critical values estimated in Pesaran et  al. (2001) are based on large sample sizes. Hence, Narayan (2005) provides a set of critical values for small sample sizes between 30 and 80. 9. The coefficient on the error-correction term should be negative and not lower than –2, that is, within the unit circle. 10. In winsorization, the extreme outliers are replaced by certain percentiles, instead of being excluded. The winsorization is done with a fifth and 95th percentile set at the cutoff levels. 11. The result of the KMO test can be between 0 and 1 and interpreted as: above 0.9 is “marvelous,” 0.80 as “meritorious,” 0.70 as “middling,” 0.60 as “mediocre,” 0.50 as “miserable,” and below 0.49 as unacceptable.

REFERENCES Abeln, B., & Jacobs, J. P. A. M. (2015). Seasonal adjustment with and without revisions: A comparison of X-13ARIMA-SEATS and CAMPLET. CAMA Working papers. Working paper. The Centre for Applied Macroeconomic Analysis, Crawford School of Public Policy. The Australian National University. Retrieved from https://cama​.crawford​.anu​.edu​.au​/sites​/default ​/files​/publication ​/cama ​_ crawford​_anu​_edu​_au​/2015​- 07​/25​_2015​_abeln​_ jacobs​.pdf. Ahlin, C., Lin, J., & Maio, M. (2011). Where does microfinance flourish? Microfinance institution performance in macroeconomic context. Journal of Development Economics, 95(2), 105–120. doi: 10.1016/j.jdeveco.2010.04.004. Alimukhamedova, N. (2013). Contribution of microfinance to economic growth: Transmission channel and the ways to test it. Business and Economic Horizons, 9(4), 27–43. doi: 10.15208/beh.2013.20. Arestis, P., Chortareas, G., & Magkonis, G. (2015). The financial development and growth nexus: A meta-analysis. Journal of Economic Surveys, 29(3), 549–565. doi:10.1111/joes.12086. Armendáriz de Aghion, B., & Morduch, J. (2010). The economics of microfinance (Second ed.). Cambridge, MA: MIT Press. Banerjee, A., Karlan, D., & Zinman, J. (2015). Six randomized evaluations of microcredit: Introduction and further steps. American Economic Journal: Applied Economics, 7(1), 1–21. doi: 10.1257/ app.20140287. Baum, C. F. (2000a). CUSUM6: Stata module to compute cusum, cusum^2 stability tests (Version S408601): Boston College Department of Economics. Retrieved from https://ideas​.repec​.org​/c​/ boc​/ bocode​/s408601​.html. Baum, C. F. (2000b). KPSS: Stata module to compute Kwiatkowski-Phillips-Schmidt-Shin test for stationarity (Version S410401): Boston College Department of Economics. Retrieved from https:// ideas​.repec​.org​/c​/ boc​/ bocode​/s410401​.html. Beck, T., Levine, R., & Loayza, N. (2000). Finance and the sources of growth. Journal of Financial Economics, 58(1–2), 261–300. doi: 10.1016/S0304-405X(00)00072-6. Buera, F. J., Kaboski, J. P., & Shin, Y. (2012). The macroeconomics of microfinance. Working Paper No. 17905. NBER Working Paper Series. National Bureau of Economic Research. The USA. Retrieved from http://www​.nber​.org​/papers​/w17905. Charitonenko, S., Campion, A., & Fernando, N. A. (2004). Commercialization of microfinance perspectives from south and southeast Asia. Manila: The Asian Development Bank. Čihák, M., Demirgüç-Kunt, A., Feyen, E., & Levine, R. (2012). Benchmarking financial systems around the world. Working Paper No. 6175. Policy Research Working Papers. World Bank. Washington, DC. Retrieved from https://openknowledge​.worldbank​.org​/ handle​/10986​/12031. Čihák, M., Demirgüç-Kunt, A., Feyen, E., & Levine, R. (2013). Financial development in 205 economies, 1960 to 2010. The Journal of Financial Perspectives, 1(2), 19. doi:10.3386/w18946 (for NBER Working Paper 18946). Copestake, J., Bhalotra, S., & Johnson, S. (2001). Assessing the impact of microcredit: A Zambian Case Study. The Journal of Development Studies, 37(4), 81–100. doi:10.1080/0022038041233132 2051. Cull, R., Demirgüç-Kunt, A., & Morduch, J. (2007). Financial performance and outreach: A global analysis of leading microbanks. The Economic Journal, 117(517), F107–F133. doi:10.1111/j.14680297.2007.02017.x.

442  Handbook of microfinance, financial inclusion and development

Demetriades, P. O., & Hussein, K. A. (1996). Does financial development cause economic growth? Time-series evidence from 16 countries. Journal of Development Economics, 51(2), 387–411. doi: 10.1016/s0304-3878(96)00421-x. Enders, W. (2009). Applied econometric time series (Third ed.). New York: John Wiley & Sons, Inc. Engle, R. F., & Granger, C. W. J. (1987). Co-integration and error correction: Representation, estimation, and testing. Econometrica, 55(2), 251–276. doi:10.2307/1913236. Goldsmith, R. W. (1969). Financial structure and development. New Haven: Yale University Press. Gorsuch, R. L. (1983). Factor analysis: New Jersey: Lawrence Erlbaum Associates. Gurley, J. G., & Shaw, E. S. (1955). Financial aspects of economic development. The American Economic Review, 45(4), 515–538. Harris, R., & Sollis, R. (2003). Applied time series modelling and forecasting. Hoboken: John Wiley & Sons, Inc. Hobijn, B., Franses, P. H., & Ooms, M. (2004). Generalizations of the KPSS-test for stationarity. Statistica Neerlandica, 58(4), 483–502. doi: 10.1111/j.1467-9574.2004.00272.x. Johansen, S. (1988). Statistical analysis of cointegration vectors. Journal of Economic Dynamics and Control, 12(2), 231–254. doi: 10.1016/0165-1889(88)90041-3. Khandker, S. R. (2005). Microfinance and poverty: Evidence using panel data from Bangladesh. The World Bank Economic Review, 19(2), 263–286. doi:10.1093/wber/lhi008, Khandker, S. R., Samad, H. A., & Khan, Z. H. (1998). Income and employment effects of micro‐credit programmes: Village‐level evidence from Bangladesh. The Journal of Development Studies, 35(2), 96–124. doi:10.1080/00220389808422566. King, R. G., & Levine, R. (1993a). Finance and growth: Schumpeter might be right. Quarterly Journal of Economics, 108(3), 717–737. doi: 10.2307/2118406. King, R. G., & Levine, R. (1993b). Finance, entrepreneurship and growth. Journal of Monetary Economics, 32(3), 513–542. doi:10.1016/0304-3932(93)90028-E. Kripfganz, S., & Schneider, D. C. (2016). D.C. ardl: Stata module to estimate autoregressive distributed lag models. Stata Conference, Chicago, July 2016. Levine, R. (1997). Financial development and economic growth: Views and Agenda. Journal of Economic Literature, 35(2), 688–726. Levine, R. (2005). Finance and growth: Theory and evidence handbook of economic growth (First ed., Vol. 1A, pp. 865–934). The Netherlands: Elsevier B.V. Maksudova, N. (2010). Macroeconomics of microfinance: How do the channels work? Working Paper No. 423. CERGE-EI Working Paper Series. CERGE-EI. Czech Republic. Retrieved from https:// papers​.ssrn​.com ​/sol3​/papers​.cfm​?abstract​_id​=1699982. McKinnon, R. I. (1973). Money and capital in economic development. Washington, DC: Brookings Institution Press. Merton, R. C. (1995). A functional perspective of financial intermediation. Financial Management, 24(2), 23–41. doi:10.2307/3665532. Narayan, P. K. (2005). The saving and investment nexus for China: Evidence from cointegration tests. Applied Economics, 37(17), 1979–1990. doi:10.1080/00036840500278103. Narayan, P. K., & Smyth, R. (2005). Electricity consumption, employment and real income in Australia evidence from multivariate Granger causality tests. Energy Policy, 33(9), 1109–1116. doi: 10.1016/j. enpol.2003.11.010. Nardo, M., Saisana, M., Saltelli, A., Tarantola, S., Hoffman, A., & Giovannini, E. (2005). Handbook on constructing composite indicators: Methodology and user guide. Working Paper No. 2005/03. OECD Statistics Working Papers. OECD Publishing. Paris, France. Retrieved from http://www​.oecd​ -ilibrary​.org​/economics​/ handbook​-on​-constructing​-composite​-indicators​_ 533411815016. Nardo, M., Saisana, M., Saltelli, A., Tarantola, S., Hoffman, A., & Giovannini, E. (2008). Handbook on constructing composite indicators: Methodology and user guide (p. 162). Retrieved from http:// www​.oecd​.org​/std​/42495745​.pdf. National Statistics Office of Mongolia. (2017). Number of active establishments, by employment size class, enterprise type. Retrieved from 2017-03-24 http://www​.1212​.mn​/tables​.aspx​?TBL​_ID​=DT​ _NSO​_2600​_009V1. Nkoro, E., & Uko, A. K. (2016). Autoregressive Distributed Lag (ARDL) cointegration technique: Application and interpretation. Journal of Statistical and Econometric Methods, 5(4), 63–91.

Does microfinance cause banking sector development and economic growth?  443

Perron, P., & Ng, S. (1996). Useful modifications to some unit root tests with dependent errors and their local asymptotic properties. The Review of Economic Studies, 63(3), 435–463. doi: 10.2307/2297890. Pesaran, M. H., & Shin, Y. (1999). An autoregressive distributed-lag modelling approach to cointegration analysis. In S. Strøm (Ed.), Econometrics and economic theory in the 20th century: The Ragnar Frisch Centennial Symposium (pp. 371–413). Cambridge: Cambridge University Press. Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289–326. doi:10.1002/jae.616. Rajan, R., & Zingales, L. (1998). Financial dependence and growth. The American Economic Review, 88(3), 559–586. Robinson, J. (1952). The rate of interest and other essays (First ed.). London: Macmillan and Co. Sahay, R., Cihak, M., N’Diaye, P., Barajas, A., Ayala Pena, D., Bi, R., . . . Yousefi, R. (2015). Rethinking financial deepening: Stability and growth in emerging markets. SDN/15/08. IMF Staff Discussion Notes. International Monetary Fund. Retrieved from https://www​.imf​.org​/external​/pubs​/ft​/sdn​/2015​ /sdn1508​.pdf. Schumpeter, J. A. (1934). The theory of economic development: An inquiry into profits, capital, credit, interest, and the business cycle. New Jersey: Transaction Books. Shaw, E. S. (1973). Financial deepening in economic development. Oxford: Oxford University Press. Silva, A. C., & Chávez, G. A. (2015). Microfinance, country governance, and the global financial crisis. Venture Capital, 17(1–2), 191–213. doi:10.1080/13691066.2015.1021032. Sims, C. A., Stock, J. H., & Watson, M. W. (1990). Inference in linear time series models with some unit roots. Econometrica, 58(1), 113–144. doi:10.2307/2938337. StataCorp. (2013). Stata statistical software (Version Release 13). College Station, TX: StataCorp LP. Svirydzenka, K. (2016). Introducing a new broad-based index of financial development. No. WP/16/5. IMF Working Papers. Strategy, Policy, and Review Department. International Monetary Fund. Retrieved from https://www​.imf​.org​/en​/ Publications​/ WP​/ Issues​/2016​/12​/31​/ Introducing​-a​-New​ -Broad​-based​-Index​-of​-Financial​-Development​- 43621. Valickova, P., Havranek, T., & Horvath, R. (2015). Financial development and economic growth: A meta-analysis. Journal of Economic Surveys, 29(3), 506–526. doi:10.1111/joes.12068. Vanroose, A., & D’Espallier, B. (2013). Do microfinance institutions accomplish their mission? Evidence from the relationship between traditional financial sector development and microfinance institutions’ outreach and performance. Applied Economics, 45(15), 1965–1982. doi: 10.1080/00036846.2011.641932. World Bank. (2015). Global financial development report 2015-2016: Long-term finance Global financial development report. Washington, DC: World Bank Group. Xu, S., Copestake, J., & Peng, X. (2016). Microfinance institutions’ mission drift in macroeconomic context. Journal of International Development, 28(7), 1123–1137. doi:10.1002/jid.3097.

444  Handbook of microfinance, financial inclusion and development

APPENDIX 22.1 PRINCIPAL COMPONENT ANALYSIS We use PCA and aggregate the results to construct composite indices for microfinance and banking sector developments. Construction of the indices using PCA involves several steps (Nardo et al., 2005, 2008). The first step includes several preliminary statistical analyses and data treatments, i.e. identifying outliers, normalization, determining the ratio of cases to variables, and running correlation analysis. Outliers are detected by various procedures, such as a histogram and Q-Q plot, and winsorization where appropriate.10 PCA also requires that the data be normalized if the variables are on different scales. Unless they are normalized, indicators whose numbers are larger in scale will have a much bigger variance. In our case, the variables are on vastly different scales. For example, the return on assets is expressed in percentages, whereas the number of clients is in an integer, and thus normalization must be carried out. Fortunately, Stata 13.0 (StataCorp, 2013), the software we used, applies PCA on a correlation matrix. Another issue that should be considered in PCA is determining a ratio of so-called cases to variables. There are different rules of thumb for the ratio. In our case, the ratios for financial development and microfinance development are 4.4 and 5, respectively, which are acceptable (Gorsuch, 1983) as we have only 40 quarters of observations on eight and nine variables. Last but not least, correlation analysis is done between the individual variables. If the correlation between two variables is strong, it is likely that they share common information, which implies that they can be in the same PCA. In addition to analyzing individual correlation tests, we use specific tests that check the appropriateness of applying PCA: (1) Bartlett’s test of sphericity and (2) the Kaiser-Meyer-Olkin (KMO) test. Bartlett’s test of sphericity tests whether the variables in the correlation matrix are uncorrelated, whereas the KMO test is a measure of sampling adequacy and determines how well the data is suited for PCA.11 The results of the tests are presented in Table A22.1. The results of the two tests show that the correlation matrix is not an identity matrix and also that the strength of the relationship among the variables is strong enough. Thus, we conclude that using the PCA is appropriate for the data. The second step is to extract the components that explain most of the variance of the original variables. We selected the components that have eigenvalues larger than one or contribute individually to the explanation of the total variance by more than 10 percent. The results of PCA in Table A22.2 indicate that the first three principal components explain relatively a large proportion of the total variances, 45.4 percent, 29.2 percent, and 17.5 percent, respectively. In

Table A22.1  The results of the KMO test and Bartlett’s test Banking sector variables

Microfinance sector variables

Bartlett test of sphericity (H0: variables are not intercorrelated) Chi-square

520.48

1,548.69

Degrees of freedom

36

28

p-value

0.000

0.000

Measure of sampling adequacy Kaiser-Meyer-Olkin

0.69

0.65

Does microfinance cause banking sector development and economic growth?  445

Table A22.2  Principal component analysis for banking sector development indicators Comp1

Comp2

Comp3

Comp4

Comp5

Comp6

Comp7

Comp8

Comp9

Eigenvalue

4.089

2.630

1.576

0.387

0.184

0.072

0.040

0.015

0.007

Difference

1.459

1.054

1.189

0.203

0.112

0.032

0.025

0.008

0.000

Proportion

0.454

0.292

0.175

0.043

0.020

0.008

0.004

0.002

0.001

Cumulative

0.454

0.747

0.922

0.965

0.985

0.993

0.998

0.999

1.000

Table A22.3  The obtained components before and after rotation—banking Unrotated components

Rotated components

Comp1

Comp2

Comp3

Comp1

Comp2

Comp3

Variance

4.089

2.630

1.576

4.062

2.232

2.001

Difference

1.459

1.054

1.189

1.830

0.232

—

Proportion

0.454

0.292

0.175

0.451

0.248

0.222

Cumulative

0.454

0.747

0.922

0.451

0.699

0.922

total, they capture about 92.2 percent of the information from the original data on banking sector development. The other six components are ignored since their marginal information content is small. The third step is to rotate the selected components to minimize the number of individual variables that have a high correlation with the same component. The three retained components were rotated using the oblimin rotation method. Table A22.3 reports the rotation result. The proportions in the total variance have slightly changed and become 45.1 percent, 24.8 percent, and 22.2 percent, respectively. After the rotation, the three retained components explain most of the variance of all the variables in the four different dimensions of the banking sector development. Table A22.4 demonstrates that the first component represents the accessibility and the depth of the banking sector development; the second and third components represent the efficiency and the stability of the sector, respectively. The fourth step involves the estimation of the weights on which the construction composite indicators are based. As only three components out of nine are obtained, we adjust the proportion percentages of variance. For example, the first component, which explains 45.1 percent of total variance after rotation, has a weight of 48.9 percent (= 0.451/0.922). The second and the third components have weights of 26.9 percent (= 0.248/0.922) and 24 percent (= 0.222/0.922), respectively. In the final step, the extracted components are combined using the weights calculated to derive a composite index for the bank-based financial sector development. The same methodologies are applied for constructing the microfinance sector index. The unrotated principal components explaining the total variance of all the microfinance variables are shown in Table A22.5. We retained the first three components as they can explain 95.43 percent of the total variance. We rotated the components using the orthogonal oblimin rotation method. Table A22.6

446  Handbook of microfinance, financial inclusion and development

Table A22.4  Eigenvectors of the components before and after rotation—banking Variables

Dimension

Unrotated

Rotated

Comp1

Comp2

Comp3

Comp1

Comp2

Comp3

roa_cb

Efficiency

0.037

0.548

–0.344

–0.040

0.642

0.076

roe_cb

Efficiency

0.078

0.488

–0.451

0.008

0.667

–0.041

capad_cb

Stability

0.050

0.393

0.605

0.000

–0.072

0.719

zscore_cb

Stability

0.072

0.500

0.447

0.007

0.113

0.665

acc1_cb

Accessibility

0.457

–0.061

–0.119

0.460

0.078

–0.094

acc2_cb

Accessibility

0.467

–0.026

0.129

0.467

–0.050

0.121

m2_gdp

Depth

0.379

0.112

–0.240

0.359

0.278

–0.087

pc_gdp

Depth

0.429

–0.188

0.159

0.451

–0.198

0.041

tacb_gdp

Depth

0.481

–0.074

0.004

0.487

–0.007

–0.005

Notes:  roa_cb = return on assets of banks; roe_cb = return on equity of banks; capad_cb = capital adequacy ratio; zscore_cb = Z-score of banks; acc1_cb = number of bank branches per 100,000 adults; acc2_cb = number of individual borrowers per 100,000 adults; m2_gdp = ratio of M2 to GDP; pc_gdp = ratio of private credit to GDP; tacb_gdp = ratio of total assets of banks to GDP A darker cell is more representative of the dimension or the variable

Table A22.5  Principal component analysis for microfinance sector development indicators Comp1

Comp2

Comp3

Comp4

Comp5

Comp6

Comp7

Comp8

Eigenvalue

5.1535

1.6859

0.7952

0.2522

0.0929

0.0197

0.0007

—

Difference

3.4676

0.8907

0.5430

0.1593

0.0732

0.0190

0.0007

—

Proportion

0.6442

0.2107

0.0994

0.0315

0.0116

0.0025

0.0001

—

Cumulative

0.6442

0.8549

0.9543

0.9858

0.9975

0.9999

1.0000

1.0000

Table A22.6  The Obtained components before and after rotation—microfinance Unrotated components

Rotated components

Comp1

Comp2

Comp3

Comp1

Comp2

Comp3

Variance

5.1535

1.6859

0.7952

3.4262

2.1938

2.0146

Difference

3.4676

0.8907

0.5430

1.2325

0.1792

—

Proportion

0.6442

0.2107

0.0994

0.4283

0.2742

0.2518

Cumulative

0.6442

0.8549

0.9543

0.4283

0.7025

0.9543

Does microfinance cause banking sector development and economic growth?  447

Table A22.7  Eigenvectors of the components after rotation—microfinance Variables

Dimension

Unrotated

Rotated

Comp1

Comp2

Comp3

Comp1

Comp2

Comp3

roa_mf

Efficiency

0.298

0.540

0.227

–0.012

0.650

0.095

roe_mf

Efficiency

0.231

0.645

0.135

–0.028

0.696

–0.052

capad_mf

Stability

0.342

–0.320

0.528

0.006

–0.027

0.706

zscore_mf

Stability

0.361

–0.244

0.532

0.004

0.049

0.686

acc1_mf

Accessibility

0.391

–0.237

–0.285

0.520

–0.119

0.075

acc2_mf

Accessibility

0.364

0.213

–0.270

0.403

0.273

–0.115

tamf_gdp

Depth

0.392

–0.134

–0.354

0.543

–0.043

–0.020

loanmf_gdp

Depth

0.414

–0.107

–0.299

0.521

0.003

0.022

Notes:  roa_mf = return on assets of MFIs; roe_mf = return on equity of MFIs; capad_mf = capital adequacy ratio; zscore_mf = Z-score of MFIs; acc1_mf = number of MFIs per 100,000 adults; acc2_mf = number of clients per 100,000 adults; tamf_gdp = ratio of total assets MFIs to GDP; loanmf_gdp = ratio of gross loan portfolio of MFIs to GDP A darker cell is more representative of the dimension or the variable

reports the rotation result. The proportions of the first three components in the total variance have become 42.8 percent, 27.4 percent, and 25.2 percent, respectively. The rotated component loadings in Table A22.7 have become more representative of certain dimensions of the microfinance sector. For instance, the first component represents the accessibility and depth of the microfinance sector, whereas components two and three represent the efficiency and stability of the sector, respectively. As three components out of eight are obtained, we adjust the proportion percentages of variance. For example, the first component, which explains 42.8 percent of total variance after rotation, has a weight of 44.9 percent (= 0.428/0.954). The second and the third components have weights of 28.7 percent (= 0.274/0.954) and 26.4 percent (= 0.252/0.954), respectively. Finally, the retained components are combined using the weights calculated to derive a composite index for the microfinance sector.

448  Handbook of microfinance, financial inclusion and development

APPENDIX 22.2 PLOTS OF THE CUSUM AND THE CUSUM SQUARED OF RECURSIVE RESIDUALS​​​

Figure A22.1 The GDP model CUSUM and CUSUM squared plots

Figure A22.2 The FD model CUSUM and CUSUM squared plots

Figure A22.3 The MF model CUSUM and CUSUM squared plots

23. Financial inclusion and poverty: evidence from Armenia Aleksandr Grigoryan, Knar Khachatryan, Knarik Ayvazyan, and Pundarik Mukhopadhaya

INTRODUCTION A body of theoretical and empirical literature suggests that a well-functioning financial sector benefits the poor directly by providing access to formal financial services to those who lack the resources to obtain a bank loan because of information asymmetries (Banerjee and Newman, 1993; Galor and Zeira, 1993; Beck, Demirgüç-Kunt and Levine, 2007; Cui and Sun, 2012; Chen and Zhang, 2018; Churchill and Marisetty, 2020). It is asserted that access to credit and savings facilities help individuals alter their production and employment choices and potentially exit poverty (Aghion and Bolton, 1997; Banerjee and Newman, 1993). Moreover, access to finance enables the poor to smoothen their consumption needs, generate stable income flows and savings, seek small business opportunities, and afford social services, such as health and education (Levine, 2008; Rosenzweig and Wolpin, 1993). Beck et al. (2004) in the same line find that the development of financial intermediaries raises the income share of the poor. Later, they confirm this relationship by employing cross-country as well as panel data analysis over the period 1960–2005 (Beck et al., 2007). Park and Mercado (2015) noted the positive effect of financial inclusion in Asian developing countries, which was supported by the studies of He and Kong (2017) on rural farmers.1 On the other hand, Bacarreza and Rioja (2008) find that the income of the poorest quintile has not been affected by the expansion of the financial system in Latin America and the Caribbean countries. Some other studies establish as well that financial inclusion may have different contributions to different poor groups. For example, Yiming and Wei (2017) show that inclusive financial development benefits (captured by increased income and reduced poverty) the higher-income rural poor more than the lower-income rural poor. At the same time, Kondo et al. (2008) find that the effect of financial inclusion on poverty reduction of extremely poor families is insignificant. Despite the evidence of uncertain relationship found in the literature, the international community and most industrialized countries (G20) have committed to promoting financial inclusion worldwide, considering that access to financial and social assets help youth to make their own economic decisions and escape poverty. Financial inclusion has also been featured among the Sustainable Development Goals (SDGs) that aim at ending poverty by 2030 (ICSU, 2015; Klapper et al., 2016). Moreover, the World Bank (WB) and the International Financial Corporation (IFC) set a target of achieving universal financial access (UFA) by 2020 as a poverty reduction tool. The core objective of the UFA that was developed in 2015 is to enable currently excluded adults to have access to a transaction account to store money and send and receive payments. The WB (World Bank, 2018) noted that the degree of financial inclusion varies across countries, and gender and regional disparities condition vulnerability to financial 449

450  Handbook of microfinance, financial inclusion and development

inclusion (about half of unbanked people include women, poor households in rural areas, or those who are out of the workforce).2 Financial development is seriously considered a vehicle for ameliorating poverty in a broader sense than income. In a recent paper, Russino (2018) investigates the concept of equality of opportunity as an effect of financial development on intergenerational mobility in 39 countries. Beyond development in education, existing literature recognizes the importance of financial development on women’s empowerment (see Hendriks, 2019 for the list of studies), energy consumption (see Ma and Fu, 2020 for the effect on a global perspective), and social security and general health of the population.3 Evidence shows that financial development can affect health (Kennedy et al., 2014; Orton et al., 2016) and access to education (Chliova et al., 2015; Steinert et al., 2018). The impact of financial inclusion on these two dimensions is interrelated and reduces intergenerational poverty transmission. More specifically, enhanced financial literacy, access, and behavior can boost incomes and enable households to access and use more health services and have better health knowledge and attitude change. However, a comprehensive study on the nexus between financial development and multiple dimensions of poverty is rare in the literature.4 Our study primarily contributes to the literature by exploring the importance of financial development in multidimensional poverty. In this context, we aim to shed light on a relatively less studied aspect of financial inclusion having different poverty reduction effects as well as social opportunities for different groups. Our study further explores the relationship between households’ financial commitments5 and over-indebtedness6 in multidimensional poverty. Literature evidenced that over-indebtedness may generate negative spillovers that affect borrowers, financial institutions, and society at large (Lascelles and Mendelson, 2012; Griffiths, 2000). Some of the consequences of over-indebtedness can lead to a reduced standard of living and social exclusion, thus limiting individuals’ ability to do certain basic things (Ntsalaze and Ikhide, 2018). Empirical evidence posits that over-indebtedness can be a key factor in creating monetary poverty, particularly among low-income, old-age households and single-parent households with young children (Ntsalaze and Ikhide, 2016; Betti et al., 2007). We study how financial inclusion and poverty7 are related in Armenia, which has a sound and well-regulated financial sector. Armenia is a small country with a population of about three million located in Eastern Europe. The country’s economy depends heavily on remittances and a significant part of its population (25.7 percent) is below the national poverty line of Armenian dram (AMD) 41,612 per month (1 USD equals about 480–500 AMD) as of 2017.8 According to the Asian Development Bank (ADB, 2019), the proportion of the employed population below $1.90 purchasing power parity a day in 2017 is 1.4 percent. Yet credit has been growing in Armenia and in 2019, as compared to the same period in 2018, consumer and mortgage loans increased by around 35 percent. Over the same period, the population income growth rate was less than half of the credit growth. This trend signals the need for policy interventions aimed at maintaining the balance between earnings and debt. One of the arguments that the literature provides on linking income and credit market participation is the tunnel effect, which specifies that individuals perceive their counterpart’s (reference group) faster progressions as a source of information to predict their own situations (Levy-Garboua and Montmarquette, 2001; Hirschman and Rothschild, 1973). This relationship is supported particularly by evidence from developing countries and transition economies, including Eastern Europe (Senik, 2008). Households in developing countries exhibit similar behavior because people have highly volatile incomes and uncertainty about employment,

Financial inclusion and poverty  451

thus they seek information from social network members to form expectations about future income and wealth status (Senik, 2004). In many countries in Central and Eastern Europe and Central Asia, adults do not tend to use their bank accounts to save. Fewer than one in six adults with a formal account report having saved using a formal account in the past year (Demirgüç-Kunt and Klapper, 2013). Adults in this region are using their bank accounts predominantly for receiving wages and government payments. To explore the relationship between multidimensional poverty and financial development in Armenia we use nationally representative data from a financial inclusion survey administered by the Central Bank of Armenia (CBA) for 2016–2018. First, we analyze the contributions of various dimensions of multidimensional poverty on household well-being. Second, we estimate an econometric model with the binary deprivation score, which captures multidimensional poverty at a household level. The variables of interest on the right-hand side are financial inclusion indicators, and we control for individual, household, and spatial characteristics. This study has a two-fold contribution to the existing literature. Firstly, it measures the multidimensional poverty in Armenia by employing a novel nationally representative dataset and explores the contributions of various dimensions of poverty. Secondly, based on a probit regression, it estimates the relationship of multidimensional poverty with two important aspects of financial development: over-indebtedness and access to finance controlling for various household traits. The remainder of the chapter is divided as follows. The second section provides an overview of the Armenian economy. The third section sets the method and the data. The fourth section presents regression analysis and discusses empirical estimations and results. Finally, the fifth section offers concluding remarks and policy implications.

OVERVIEW OF THE ARMENIAN MACROECONOMIC ENVIRONMENT AND FINANCIAL SECTOR Since the collapse of the Soviet Union (in 1991), Armenia has made substantial steps in liberalizing the economy via continuous reforms committed at the early stage of independence. However, its closed borders with two of its neighboring countries—Turkey and Azerbaijan— do not create favorable development conditions for the country’s small economy. While in the early 2000s the economy experienced sound economic growth, it was severely disrupted by the world financial crisis in 2008. Despite improvements in aggregate economic indicators, income poverty and inequality, rural development, and access to finance, peculiar to transition countries remain important challenges in Armenia. The country heavily depends on remittances, which effectively transfer external shocks to domestic markets. Figure 23.1 shows that monetary poverty dramatically decreased in the period 2004–2007, but this trend was disrupted in the first year of the global financial crisis, 2008. Since 2011, poverty has been decreasing again, but is still high (25.7 percent in 2017). Prior to the financial crisis period, income inequality was mitigated, but since 2008, the reversed trend became persistent. During the 2004 to 2007 period, economic growth was rather impressive and it led to substantial improvement in both inequality and poverty. In the post-crisis period, economic growth over 2013–2017 created prerequisites for improved living conditions and a reduction in monetary poverty. According to the Statistical Committee of the Republic of Armenia

452  Handbook of microfinance, financial inclusion and development

Note:  (i) GDP per capita is gross domestic product divided by midyear population. (ii) Poverty headcount ratio is the percentage of the population living below the national poverty line. National estimates are based on populationweighted subgroup estimates from household surveys, conducted in 2004. (iii) Gini index is the World Bank estimate. Source:  World Bank.

Figure 23.1 Evolution of growth, poverty, and inequality for the Armenian economy, 2004–2017 (SCRA), compared to 2012, the GDP of Armenia increased by 18.96 percent and poverty decreased by 20.7 percent in 2017, thus producing a negative poverty-to-GDP elasticity coefficient over the 2013–2017 period. For the first time in 2017, the poverty rate in Armenia has decreased to a level that is lower than the 2008 rate (27.6 percent) by 1.9 percentage points.9 SCRA reports that 1.85 percentage points decrease in total poverty over 2008–2017 was due to the impact of both the consumption and redistribution effects. As of 2017, among the 25.7 percent poor population, 1.4 percent are extremely poor (households below the food poverty line, AMD 24,269 per month), 9.2 percent are moderately poor (households between food and lower poverty lines, AMD 24,269–34,253 per month), and the remaining 15.1 percent are poor (households between lower and upper poverty lines, AMD 34,253–41,612 per month). In 2017, the difference between the poverty rates in urban (25.0 percent) and rural (26.8 percent) communities was small. However, the difference is large between the capital city Yerevan (22.4 percent) and other towns of the country (27.9 percent). The Armenian financial system has grown rapidly in recent years, with total financial assets increasing from about 64 percent of GDP at the end of 2012 to almost 91 percent by the end of 2017. The financial system is dominated by commercial banks, although banks’ share of total financial assets is slowly eroding. At the end of 2017, the assets of the banking sector were about 78 percent of GDP and accounted for 86 percent of total financial assets. A large proportion of these assets belongs to subsidiaries of foreign banks. The banking system in Armenia is privately owned with no government banks. Moreover, three of the 17 Armenian banks are open joint stock companies, and all banks are expected

Financial inclusion and poverty  453

to continue to strive to attract new shareholders. Credit organizations are the second largest segment of the financial system in Armenia. The share of credit organizations, which are credit-only, nearly doubled in size from 2011 to 2017 and accounted for about 8.5 percent of GDP at the end of 2017. Insurance companies are also important, mainly because third-party auto insurance became mandatory in 2010. Between 2007 and 2017, while the total assets of the banking sector increased from 24.3 to 78.2 percent of GDP and banking capital more than doubled, the total profits of Armenian banks remained broadly constant in absolute terms (about $60 million per year on average). In 2017, the average return on assets (ROA) of Armenian banks was less than 1 percent and the average return on equity (ROE) was about 6 percent.

THE METHOD AND DATA Computation of Multidimensional Poverty The unit of our analysis is households in Armenia. Our outcome variable is based on the poverty deprivation score, and the methodology for constructing the score is described in the following (see Alkire and Foster, 2008). Let Xi,j denote the achievement of household i in dimension j for all i = 1, 2, ¼, n and j = 1, 2, ¼, d . A dual cut-off framework is used to identify multidimensionally poor households. The first one is called the deprivation cut-off Zj, which denotes the poverty line in dimension or indicator j. If the achievement of household i is higher than the cut-off, X i , j ³ Z j , household i is not deprived in dimension j. Otherwise, the household i is deprived in this dimension. If household i is deprived in dimension j, then the deprivation status value is gij = 1, otherwise, gij = 0. The second cut-off is the overall poverty cut-off k ( 0 £ k £ 1), which is a pre-determined fraction of the total number of dimensions or indicators. That is, if we define poverty measure as the household being poor when it is deprived in 40 percent of the total number of indicators then we assign a value k = 0.4. In this process, there are two steps to identify a poor household. First, by giving weight wj to each dimension or indicator j such that å dj =1w j = 1, we obtain the weighted deprivation status value w j gij and the deprivation score,

ci =

å

d

w j gij . (23.1)

j =1

Second, we compare the deprivation score with the poverty cut-off for household i and identify the (multidimensional) poverty status. If ci ≥ k, household i is considered to be poor (gij ( k ) = 1, otherwise (ci