IFPRI RESEARCH REPORT 1 1 6 Access to Credit and Its Impact on Welfare in Malawi Aliou Diagne Manfred Zeller INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE Access to Credit and Its Impact on Welfare in Malawi Aliou Diagne Manfred Zeller Research Report 116 International Food Policy Research Institute Washington, D.C. Access to Credit and Its Impact on Welfare in Malawi Aliou Diagne Manfred Zeller International Food Policy Research Institute Washington, D.C. Access to Credit and Its Impact on Welfare in Malawi Aliou Diagne Manfred Zeller International Food Policy Research Institute Washington, D.C. Copyright © 2001 International Food Policy Research Institute All rights reserved. Sections of this report may be reproduced without the express permission of but with acknowledgment to the International Food Policy Research Institute. Library of Congress Cataloging-in-Publication Data Diagne, Aliou. Access to credit and its impact on welfare in Malawi / Aliou Diagne, Manfred Zeller. p. cm. — (Research report ; 116) Includes bibliographical references. ISBN 0-89629-119-7 (pbk.) 1. Rural poor—Malawi. 2. Agricultural credit— Malawi. I. Zeller, Manfred. II. Title. III. Research report (International Food Policy Research Institute) ; 116. HC935.Z9 P614 2001 332.7′1′096897—dc21 00-054679 Contents List of Tables iv List of Figures vi Foreword vii Acknowledgments viii Summary x 1. Introduction 1 2. The Rural Economy and Microfinance Institutions in Malawi 6 3. Survey Design and Description of the Data 16 4. Econometric Analysis of the Impact of Access to Credit on Household Welfare 62 5. Results of the Econometric Analysis 81 6. Conclusions and Implications for Policy 123 Appendix: Econometric Methodology 130 References 143 iii Contents Tables 1. Loan disbursements and recovery rates of the Malawi Rural Finance Company 12 2. Demographic characteristics of households 20 3. Asset ownership, composition, and distribution 21 4. Asset ownership, composition, and distribution by credit program membership 24 5. Loan transactions and their characteristics 27 6. Distribution of formal and informal credit limits and unused credit lines, October 1993–December 1995 28 7. Households with access to credit, by program membership and sector of the credit market 34 8. Major rainfed crops grown, by household 37 9. Household cultivated land and its allocation among crops in the 1994/95 season, by credit program membership 38 10. Fertilizer acquisition and relative importance of different methods of acquisition and source of financing of inputs in the 1994/95 season, by program membership 40 11. Distribution of fertilizer among crops in 1993/94, 1994/95, and 1995/96 seasons, by program membership and type of farm 42 12. Average yield and net income per hectare for major rainfed crops, 1994 production year, by program membership 44 13. Average yield and net income per hectare for major rainfed crops, 1995 production year, by program membership 45 14. Fertilizer recommendations for maize and tobacco in Malawi 47 15. Total household farm and nonfarm income, 1994 and 1995, by credit program membership 54 16. Consumption expenditures, calorie intake, and nutritional status, by credit program membership, 1995 58 17. Regressors used in equations 75 18. Definition and summary statistics of variables used in the model 82 iv Tables 19. Predicted conditional probability choices 85 20. Determinants of program participation: Parameter estimates and partial changes in probability of participation resulting from marginal changes in selected independent variables 86 21. Formal credit limit equation: Estimated parameters and partial effects of marginal changes in selected independent variables 89 22. Informal credit limit equation: Estimated parameters and partial effects of marginal changes in selected independent variables 91 23. Formal credit demand equation: Estimated parameters and direct and indirect partial effects of marginal changes in selected independent variables 94 24. Informal credit demand equation: Estimated parameters and direct and indirect partial effects of marginal changes in selected independent variables 96 25. Annual income equation: Estimated parameters and partial effects of marginal changes in selected independent variables 102 26. Crop income equation: Estimated parameters and partial effects of marginal changes in selected independent variables 104 27. Nonfarm seasonal income equation: Estimated parameters and partial effects of marginal changes in selected independent variables 106 28. Food expenditure equation: Estimated parameters and partial effects of marginal changes in selected independent variables 112 29. Daily calorie intake equation: Estimated parameters and partial effects of marginal changes in selected independent variables 114 30. Daily protein intake equation: Estimated parameters and partial effects of marginal changes in selected independent variables 116 31. Weight-for-age Z-score equation: Estimated parameters and partial effects of marginal changes in selected independent variables 118 32. Height-for-age Z-score equation: Estimated parameters and partial effects of marginal changes in selected independent variables 120 v Figures 1. Location of the DRD/IFPRI Rural Finance Survey sites 17 2. Distributions of formal and informal credit limits and unused credit lines for all respondents, October 1993–December 1995 30 3. Distributions of formal and informal credit limits and unused credit lines when a formal loan was granted, October 1993–December 1995 31 4. Distributions of formal and informal credit limits and unused credit lines when an informal loan was granted, October 1993–December 1995 32 5. Distributions of formal and informal credit limits and unused credit lines when a loan demand was rejected, October 1993–December 1995 33 6. Distribution of formal and informal credit limits when no loan was requested, October 1993–December 1995 36 7. Yields of local maize, hybrid maize, and tobacco versus fertilizer use 48 8. Gross margins of local maize, hybrid maize, and tobacco versus fertilizer use 49 9. Yields of local maize, hybrid maize, and tobacco versus total input cost 50 10. Gross margins of local maize, hybrid maize, and tobacco versus total input cost 51 vi Figures Foreword For decades the poor in developing countries (and elsewhere) were essentially shut out of credit and savings services. Because the poor did not meet the tradi- tional criteria for borrowing, financial institutions perceived them as bad credit risks. More recently, development practitioners have come to see that the poor can indeed make effective use of credit to raise their incomes and get access to more food and other necessities. In fact, in some quarters microcredit is now seen as the solution to poverty. Research conducted at IFPRI shows, however, that although credit can be an important tool in the fight against poverty, credit alone cannot be guaranteed to raise incomes, increase food security, and improve nutrition. In this research report, Aliou Diagne and Manfred Zeller examine the case of Malawi, where several institutions offer credit to poor, smallholder farmers to allow them to buy fertilizer, seeds, and other inputs for growing maize and tobacco as a way of helping raise incomes. Surprisingly, they find that farmers who participated in these credit programs ended up with less net crop income than those who did not. Their results make clear that the conditions surrounding credit programs must be right—that is, they must reflect the actual opportunities and constraints faced by poor farmers—for credit to work effectively. For example, credit is not of much use in sit- uations in which farmers have little access to roads, markets, health care, and com- munications infrastructure and are subject to drought that can wipe out their crops, as is the case in Malawi. This research report reveals how complicated the task of effective rural develop- ment can be, but it also points to concrete steps, in addition to offering credit ser- vices, that governments and development organizations can take in their efforts to eradicate poverty and food insecurity. This research report should be of great signif- icance to anyone interested in how rural finance can be made to work best for those in the most need—the poor and food insecure in developing countries. Per Pinstrup-Andersen Director General vii Foreword Acknowledgments Our special gratitude goes to the members of the survey households, who dur- ing three survey rounds in 1995 gave of their precious time and who responded to numerous questions, some of which touched on very sensitive issues, such as their possession of assets, access to credit, and level of debt. We thank them for their trust and their contribution to what is essentially a public good that does not create any direct and immediate benefit for them. It is our hope that this report—in conjunction with prior reports, papers, policy summaries, and workshop proceedings dissemi- nated in Malawi by the rural finance research program of Bunda College and IFPRI—will be effectively used by policymakers to improve the economic opportu- nities for and therefore the welfare of rural households in Malawi. This research report and the underlying field research and data processing would not have been feasible without the essential and invaluable contribution of the re- search staff of the Bunda College of Agriculture, University of Malawi, and without the contribution of many others in Malawi, at IFPRI, and at other institutions. Fore- most, we are grateful for the assistance of the staff of the Department of Rural De- velopment (DRD) who contributed to the successful implementation of the field sur- vey, data cleaning, and data analysis for the DRD/IFPRI Rural Finance Study. We thank Karid Chirwa, Tyme Fatch, Swalley Lamba, Samson Manda, and Franklin Simtowe, who provided invaluable research and administrative assistance. We espe- cially thank Franklin Simtowe for his excellent research contribution to the in-depth descriptive analysis for this report, Dr. Alexander Phiri for helpful discussions dur- ing all phases of the research project, and Dr. Todd Benson for contributing critical comments and questions that sharpened the analysis presented here. We also enjoyed working with a number of students at Bunda College, notably Vinda Kisyombe, Mary Mandambwe, and Hardwick Tchale, who used the DRD/IFPRI Rural Finance data set for their M.Sc. research and who provided additional insights for the role of credit in rural development. Our utmost gratitude goes to Dr. Charles Mataya, whose sup- port as head of the Department of Rural Development made this collaboration pros- per over time. We thank Dr. Malcolm Blackie and Dr. Bharati Patel of the Rockefeller Founda- tion in Malawi for their encouragement during the course of the project. At IFPRI viii Acknowledgments we thank Tina Abad, Lynette Aspillera, Almaz Beyene, and Ginette Mignot for their administrative support. The guidance of Lawrence Haddad and Sudhir Wanmali in providing an enabling research environment deserves special recognition. We thank John Pender, the IFPRI internal reviewer, and two anonymous external reviewers for their critical but very constructive comments, which helped us significantly improve the analysis in and presentation of the report. Particular sections of this report have also benefited from the comments of Alain de Janvry, Andrew Foster, Lawrence Had- dad, Soren Hauge, Hanan Jacoby, Manohar Sharma, John Strauss, and participants in the Bunda/IFPRI workshop on rural finance held in October 1996 at Bunda, in seminars at IFPRI, and in various conferences at which papers emanating from this research were presented during 1996–98. Last but not least, we gratefully acknowl- edge the financial support of the Rockefeller Foundation; UNICEF Malawi; the Min- istry of Women and Children’s Affairs and Community Services (MOWCACS); the German Agency for Technical Cooperation (GTZ) in Malawi; and the United States Agency for International Development (USAID) in Malawi. ix Summary As in many countries in Sub-Saharan Africa, the majority of poor smallholders in Malawi are left out of the agricultural extension and credit systems. These households, characterized by landholdings of less than 1 hectare and very low crop yields, are unable to grow enough food to feed themselves even though they focus much effort on producing food crops, especially maize. It has been argued that most of these farmers are too poor and cash-strapped to be able to benefit from any kind of access to credit and that, even if they received adequate supplies of the right in- puts, their land constraints are so severe that any increase in productivity would still fall short of guaranteeing their food security. For these households, credit to support nonfarm income-generating activities has been suggested as a policy alternative for alleviating their food insecurity. To gain a better understanding of the possible role of credit in improving house- hold food security and alleviating poverty in Malawi, in November 1994 the Inter- national Food Policy Research Institute and the Department of Rural Development, Bunda College of Agriculture, University of Malawi, initiated a research program on rural financial markets and household food security in Malawi. The main objective of the research program was to analyze the determinants of access to credit in Malawi and its impact on farm and nonfarm income and on household food security. The study also sought to quantify the relationship between the demand for formal loans and that for informal loans. From a policy perspective, such an analysis is important for at least two reasons. First, by quantifying the welfare impact of access to finan- cial services, it can inform policymakers about the social benefits (if any) of policy strategies to promote the formation and expansion of microfinance institutions in ru- ral areas. Second, the analysis can provide knowledge about the relative importance of the various socioeconomic factors within or beyond the control of policy that de- termine whether or not some households will benefit from access to formal credit. This latter information can guide the design of institutional arrangements and the choice of financial services to be offered to different target groups. The research emanating from this program was published during 1996–98 in a number of reports and papers disseminated by IFPRI and the Bunda College of Agri- culture, following an October 1996 workshop held at the college at which the major x Summary research results were shared and discussed with policymakers, microfinance practi- tioners, and researchers. This research report presents an in-depth analysis address- ing the research objectives described. The study analyzed the determinants of access to formal and informal credit and the demand for loans. It found that formal lenders in Malawi—such as rural banks, savings and credit cooperatives, and special credit programs supported by the gov- ernment and nongovernmental organizations—prefer to give loans to households with diversified asset portfolios and therefore more diversified incomes. This is pre- sumably done to increase and stabilize repayment rates. It also found that households in Malawi are generally credit constrained in both the formal and informal sectors of the credit market. For example, close to half of the households participating in for- mal credit programs still have binding credit constraints. However, Malawian house- holds would borrow on average only about half the amount of any increase in their credit limits. The level of interest rates charged on loans seems not to be an important factor for households in deciding in which microfinance institution to participate. Nonprice attributes of credit institutions and their services play a larger role. These attributes include the types of loans provided and the restrictions on their use, as well as the types of nonfinancial services provided by the programs, such as training in the man- agement of microenterprises. This result suggests that the acceptance of an institu- tion by its clientele, and therefore its prospects for growth and sustainability, are de- termined by a range of characteristics of both its financial and its nonfinancial services. The main findings of the study regarding the impact of access to credit on house- hold welfare outcomes do not support the notion that improving access to micro- credit is always a potent means for alleviating poverty—an opinion voiced, for ex- ample, at the Microcredit Summit in Washington, D.C., in February 1997. Both the tabular and the econometric analysis shows that when households choose to borrow they realize lower net crop incomes than nonborrowers. Although this result is not statistically significant, it nonetheless points out the risk of borrowing: that bor- rowers can be worse off after repaying the principal and interest. Two main reasons for the negative (albeit insignificant) relationship between bor- rowing and net crop incomes are identified. Both have important implications for fi- nancial sector policy and the conduct of rural financial institutions in Malawi. The first reason is the focus of the loan portfolio on one loan product, which provides farmers too much costly fertilizer for hybrid maize. Three of the four institutions in- vestigated in this study provided agricultural credit, focusing mainly on an input package for hybrid maize. The second reason is the below-average rainfall in the two survey years and the concentration of the loan portfolios of the formal lenders on maize, a drought-sensitive crop. Consistent with the insignificant results for crop income, we find no significant impact of access to credit on the per capita incomes, food security, and nutritional status of credit program members. As the credit services of the formal institutions are mostly geared toward income generation, and in particular toward the growing xi of fertilized hybrid maize and tobacco, access to the type of credit products offered in Malawi is expected to have mostly indirect effects on consumption and nutrition through its potential effect on income. The rural financial institutions in Malawi cov- ered in this study do not offer financial products, such as consumption credit and pre- cautionary savings options, that could eventually have a direct effect on consump- tion or on nutritional status. Growing tobacco is found to be the most important determinant of household crop income. Another finding of the study, however, is the fact that households that grow tobacco are less food secure, with significantly lower per capita daily calorie intake and a higher prevalence of both chronic and acute malnutrition compared with households that do not. The food insecurity and malnutrition of tobacco households may be traced to the combination of larger than average household sizes because of the labor-intensive nature of tobacco growing and the high relative cost of buying maize for consumption. The study also found that the price of maize has a significant and negative direct impact on household per capita calorie intake, while its indirect effect on the latter through household income is positive but statistically insignificant. This finding is consistent with two other findings of the study: that the marginal impact of the price of maize on household income, although sizable, is not statistically different from zero and that smallholder farmers in Malawi are, on average, net buyers of maize be- cause of their 59 percent average maize self-sufficiency. Therefore any increase in the price of maize is likely to have a negative impact on the food security of the av- erage smallholder farm household. A major conclusion of this study is that the contribution of rural microfinance in- stitutions to the income of smallholders can be limited or outright negative if the de- sign of the institutions and their services does not take into account the constraints on and demands of their clients. Developing attractive credit services requires both identifying farm and nonfarm enterprises and technologies that are profitable under the conditions experienced by subsistence-oriented farmers and responding to the numerous constraints of resource-poor rural households. The results suggest that a strategy of expanding financial institutions in rural, drought-prone areas with inad- equate market and other infrastructure may—at least in below-average rainfall years—have no significant positive welfare effects. The risk of drought in Malawi, as in much of rainfed Sub-Saharan Africa and other countries, constitutes a consid- erable challenge for developing sustainable rural financial institutions. In such envi- ronments, a strategy providing for greater diversification of the portfolio of assets and liabilities of the rural financial institutions, as well as adequate provisions for loan defaults and the building up of reserves for rescheduling loans, is a necessary precondition for rural financial institutions to prosper and to be able to offer their clientele reliable access to future credit and savings services. The necessary resources, infrastructure, and socioeconomic environment are not yet in place for access to formal credit to realize its full potential benefits for Malawi’s rural population. Therefore—considering that the formation of sustainable rural financial institutions is a difficult task to achieve in rural economies that lack xii irrigation, exhibit insufficient hard and soft infrastructure, and support a poorly ed- ucated rural population adversely affected by malnutrition and disease, and consid- ering that the benefits at the household level may not materialize in drought years— the report recommends a cautious and gradual strategy for expansion of rural financial institutions in Malawi. This strategy would require direct support by the state through an adequate legal and regulatory framework, through the support of in- stitutional innovations and pilot programs in rural areas that may have the potential to reduce transaction costs in providing savings, credit, and insurance services to ru- ral clientele. Adoption of a cautious strategy would also imply that the formation and initial expansion of rural financial institutions should focus on high-potential agricultural areas that allow for lending to those growing a diversified array of cash and food crops as well as offering financial services for off-farm enterprises at low transaction costs. This does not mean that low-potential and drought-prone agricultural areas should be neglected, because credit may be the best or only option for the small- holder farmers to finance their input acquisitions after experiencing a crop failure. Indeed the evidence showed that without access to credit the ability of smallholder farmers to recover from a crop failure is extremely limited. The mere knowledge that credit will be available in case of crop failure can be beneficial to poor farmers by inducing them to adopt new and more risky but potentially profitable crops or tech- nologies. The econometric analysis has confirmed the positive and quite sizable (though not statistically significant) impact of merely having the option to borrow, even if it is not exercised. However, the expansion of microfinance into marginal ar- eas with insufficient market and other infrastructure should be coupled with a greater emphasis on other growth- and welfare-enhancing investments (such as those in transport, health, and communications infrastructure) and with targeted safety-net interventions for the very poor. In summary, the benefits of access to credit for smallholder farmers depend on a range of agroecological and socioeconomic factors, some of which are time-variant and subject to shocks such as drought. Access to credit is therefore no panacea for poverty alleviation. The full potential of credit access in increasing the welfare of the poor can only be realized if coupled with adequate investments in hard and soft in- frastructure as well as investments in human capital. xiii CHAPTER 1 Introduction As is the case in many African countries, the majority of smallholders in Malawi are left out of the rural financial system. These households, characterized by average landholdings of less than 1 hectare, do not grow enough food to feed them- selves even though they concentrate almost exclusively on the production of maize, the major staple food in Malawi. Consequently, as land is a binding constraint in most areas of Malawi, increases in agricultural productivity, in particular in the growing of maize, and increased diversification into other food and cash crops as well as non- farm enterprises are key requirements for poverty alleviation. Such changes in the production and consumption strategies of households require capital, and they are risky to implement for households that produce maize for subsistence with low-in- put, low-output technology in a highly drought-prone environment. It has been argued that most of Malawi’s smallholder farmers are too poor to be able to benefit from any kind of access to credit, and that, even if they had access to adequate credit and inputs, their land constraints are so severe that any increase in productivity would still fall short of guaranteeing their food security (Government of Malawi 1995). For these households, credit for nonfarm income-generating ac- tivities has been suggested as a policy alternative to address their food insecurity and malnutrition. To gain a better understanding of the possible role of credit in improv- ing income and household food security and in alleviating poverty in Malawi, in No- vember 1994 the International Food Policy Research Institute (IFPRI) and the De- partment of Rural Development (DRD) of the Bunda College of Agriculture, University of Malawi, initiated a research program on rural financial markets and household food security in Malawi. The objectives of the research program were to study the determinants of access to and participation in existing formal and informal credit and saving systems, and to analyze the effects of household access to credit on agricultural productivity, income generation, and food security. This report presents the major results of that research project. 1 CHAPTER 1 Introduction The Potential Contribution of Improved Access to Formal Credit in Poverty Alleviation It is generally agreed among researchers and policymakers that poor rural households in developing countries lack adequate access to credit. This lack of adequate access to credit is in turn believed to have significant negative consequences for various ag- gregate and household-level outcomes, including technology adoption, agricultural productivity, food security, nutrition, health, and overall household welfare. Access to credit affects household welfare outcomes through three pathways (Zeller et al. 1997). The first pathway is through the alleviation of the capital con- straints on agricultural households: expenditures on agricultural inputs and on food and essential nonfood items are incurred during the planting and vegetative growth periods of crops, whereas returns are received only after the crops are harvested sev- eral months later. Most farm households show a negative cash flow during the plant- ing season. Therefore, to finance the purchase of essential consumption and produc- tion inputs, the farm household must either dip into its savings or obtain credit. Access to credit can therefore significantly increase the ability of poor households with little or no savings to acquire agricultural inputs. Furthermore, easing potential capital constraints through the granting of credit reduces the opportunity costs of capital-intensive assets relative to family labor, thus encouraging the adoption of la- bor-saving, higher-yielding technologies and therefore increasing land and labor pro- ductivity, a crucial factor in encouraging development, in particular in many African countries (Delgado 1995; Zeller et al. 1997). The second pathway through which access to credit affects household welfare is by increasing a household’s risk-bearing ability and by altering its risk-coping strat- egy. The third pathway—enabling access to credit for consumption smoothing—is closely linked to the second, and we therefore discuss them together because they both affect the resilience of households in bearing production and consumption risks. The mere knowledge that credit will be available to cushion consumption against an income shortfall if a potentially profitable, but risky, investment should turn out badly may induce a household to bear the additional risk. The household may therefore be willing to adopt new, riskier technologies (Eswaran and Kotwal 1990). A household may also benefit from mere access to credit even if it is not borrowing, because with the option of borrowing it can avoid adopting such risk-reducing but costly strate- gies as the production of low-risk but less profitable food crops, such as local maize and cassava, and the accumulation of assets that mainly serve precautionary savings purposes but that may yield very poor or even negative returns (for example, keep- ing cattle or cash). Most rural financial sector strategies gave due recognition to the first pathway but often neglected or completely ignored the other two. The vast majority of credit pro- grams provided in-kind production credit at subsidized interest rates. And most of them failed both to serve the rural poor and to remain sustainable credit institutions (Adams, Graham, and von Pischke 1984; Adams and Vogel 1986; Braverman and Guasch 1986). For example, the agricultural credit system in Malawi used to be a 2 prime example within Africa of a successful government-supported credit program because it was enjoying average repayment rates of over 97 percent from 1968 to 1991. The system collapsed in 1992 owing to a combination of severe drought and political liberalization that caused the repayment rate to plummet to less than 25 per- cent (Msukwa et al. 1994; Murotho and Kumwenda 1996). In response to these failures and recognizing that traditional commercial banks typically have no interest in lending to poor rural households because of their lack of viable collateral and the high transaction costs associated with the small loans that are best suited to them, innovative credit delivery systems are being promoted throughout the developing world as a more efficient way of improving rural house- holds’ access to formal credit. Unlike commercial banks, these credit programs have as their guiding principles not profit but rather accessibility and sustainability. Many of them are group-based lending programs relying on joint liability and peer pres- sure as substitutes for collateral, along with community-based delivery systems that seek to exploit the social capital and information advantages of local communities in screening and monitoring borrowers. The Grameen Bank in Bangladesh is a well- known example with a proven record of reaching the poorest and simultaneously achieving very high repayment rates. Policy Relevance and Objectives of the Research Community- and member-based microfinance programs have enjoyed considerable political and financial support since the 1990s. Three basic premises explain the ren- aissance of “rural credit”; the first is relatively recent, but the other two are deeply rooted within development theory and strategy: 1. Member-based financial institutions have an advantage in transaction costs over traditional forms of banking characterized by reliance on land collat- eral and a large amount of paperwork. This perceived cost advantage can al- low innovative rural financial institutions to become financially sustainable in the long run. Initial subsidies by the state are deemed justified and are re- quired to finance the development of the institution and to allow it to achieve a scale at which it can cover its costs on its own. 2. With improved access to credit, poor rural households will be able to engage in more productive farm and nonfarm income-generating activities to raise their living standards. 3. The aggregate social benefits outweigh the opportunity costs of the public funds used for developing rural financial institutions. The research presented in this report focuses only on the investigation of the sec- ond premise. It addresses a number of questions related to the provision of institu- tional credit in the context of rainfed agriculture and poor market infrastructure: Do households who participate in credit programs improve their living conditions? If they do, in what ways does improved access to formal credit benefit these house- holds? In particular, does access to formal credit contribute to raising farm and off- farm income and household food security? For the particularly poor and disadvan- 3 taged among the rural population, such as women (who are the target group of some of these programs), does access to formal credit contribute to the desired goal of poverty reduction? Quantifying the impact of improved access to formal credit on different groups of households is important for policy purposes for at least two reasons. First, it can serve as guide for the allocation of scarce resources to the numerous development programs competing for the same funds. Second, it establishes the relative impor- tance of the various factors that permit certain households in a given socioeconomic environment to achieve greater benefits from access to formal credit than others (Zeller and Sharma 1998). Furthermore, despite the increasing importance of microcredit programs in de- veloping countries, most rural households continue to rely on the informal credit market for their intertemporal transfer of resources. They rely on complex strategies to increase their productive capacity, share risk, and smooth consumption over their life cycles. These strategies generally work through self-enforcing informal contracts among friends, neighbors, and members of extended families, and they are arranged within networks of informal institutions of diverse types (Fafchamps 1992; Coate and Ravallion 1993; Udry 1994, 1995b; Lund and Fafchamps 1997). One hypothe- sis often advanced by researchers and policymakers is that government- and non- governmental organization (NGO)–supported credit programs may crowd out the financial services offered by these informal financial institutions. Therefore under- standing how the informal institutions serve households’ demand for financial serv- ices and interact with the formal credit institutions set up by governments and NGOs is critical in identifying policies, institutional designs, and financial services that can expand and complement rather than substitute for the services offered by the exist- ing informal credit market. An important step in obtaining this information is to quantify the extent and determinants of households’ access to both informal and for- mal credit markets and the degree to which the two forms of credit are complements or substitutes. A Definition of Access to Credit Access to formal credit is often confused with participation in formal credit pro- grams. Indeed the two concepts are used interchangeably in many studies. However, to analyze satisfactorily the socioeconomic determinants of both access to credit and participation in formal credit programs and to assess their respective impacts on household welfare outcomes, one needs to make the distinction between access to credit (formal or informal), participation in formal credit programs or in the infor- mal credit market, and being credit constrained. A household has access to a particular source of credit if it is able to borrow from that source, although for a variety of reasons it may choose not to. The extent of ac- cess to credit is measured by the maximum amount a household can borrow (its credit limit). If this amount is positive, the household is said to have access. A household is said to be participating if it is borrowing from a source of credit. A household is 4 credit constrained when it lacks access to credit or cannot borrow as much as it wants. These distinctions are particularly important because, as discussed previously, a household living in a risky environment may benefit from mere access to credit even if it is not actually borrowing. Structure of the Report Chapter 2 gives a brief general description of the rural economy and the agricultural policy environment. The main part of this chapter describes the credit programs stud- ied. Chapter 3 covers the survey design and provides a descriptive and tabular analy- sis of the socioeconomic characteristics and behavioral and welfare outcomes of the households surveyed. This tabular analysis provides some indications of the effects of access to credit on the outcomes studied, but of course it falls short of providing statistically tested measurements. The chapter serves mainly to describe the observed outcomes and to disaggregate them according to membership in particular credit pro- grams and other socioeconomic characteristics. Chapter 4 describes the structure of the econometric model and presents the estimation procedure that we use to meas- ure access to credit and its effects on household welfare outcomes. Chapter 5 dis- cusses the results of the econometric analysis of the determinants of households’ access to and participation in informal and formal credit markets, as well as the mar- ginal impacts of access to formal credit on farm and nonfarm incomes, household food security, and nutritional status. Chapter 6 considers implications for policy and future research. 5 CHAPTER 2 The Rural Economy and Microfinance Institutions in Malawi This chapter briefly outlines the main features of the rural economy and recent changes in the agricultural policy environment. It then describes the credit pro- grams studied. The Rural Economy and Recent Policy Changes in Malawi Rural poverty in Malawi is pervasive (United Nations and Government of Malawi 1993). The country’s nominal per capita income level of US$140 in 1994 is one of the lowest in the world. Forty percent of gross domestic product and about 75 per- cent of export earnings were accounted for by the agricultural sector during 1989–94 (IMF 1995). About 90 percent of the population of 11 million lives in rural areas, and it is predominantly employed in small-scale farming activities (Chilowa and Chirwa 1997). Farms are very small. Seventy-two percent of smallholder farms cultivate less than 1 ha (World Bank 1995). A single rainy season with erratic rainfalls, coupled with a virtual absence of irrigation, makes crop production very risky. Malawi suf- fered two major droughts during the 1990s, one in 1991/92 and one in 1993/94, fol- lowed by a below-average maize crop in 1994/95. The latter two years were the re- call periods for crop income in the DRD/IFPRI survey. The Government of Malawi identifies drought risk as one of the major reasons for farmers’ failure to adopt agri- cultural innovations, as the profitability of these varies markedly with rainfall (Gov- ernment of Malawi 1995a). The Dualistic Structure of the Rural Economy in Malawi The rural economy in Malawi is characterized by the coexistence of estate and small- holder agriculture. Land cultivated by estates is privately owned (freehold land) or leased from the state on long-term leases for 99 years (leasehold land). Land culti- 6 CHAPTER 2 The Rural Economy and Microfinance Institutions in Malawi vated by smallholders is governed by customary laws that provide the farmer with user rights. These rights can be passed on to children, and only in exceptional cases do they deny traditional authorities the inheritance of user rights. The estate sector is characterized by relatively capital-intensive production that concentrates on lu- crative export crops, such as tobacco, sugar, tea, and cotton. In contrast, the small- holder sector is to a large extent oriented toward subsistence production. It employs and feeds most of the rural population. The share of land cultivated by estates has in- creased since independence in 1964, and it reached about 12 percent of the total arable area in the early 1990s (Harvard Institute for International Development 1994a,b). This trend was largely due to a policy framework that favored the estate sector over the smallholder sector, in particular the policy that only estates were al- lowed to grow tobacco, the major and most lucrative export crop of Malawi. Recent Reforms in Agricultural Policy Past policies in Malawi by and large favored the production of high-value cash crops in the estate sector while the smallholder sector was encouraged to produce and sell maize, the country’s food staple, through official market channels (Mtawali 1993). Economic and agricultural growth in the 1960s and 1970s was driven mainly by a prospering and expanding estate sector; however, external shocks, such as the dis- ruption of trade routes and deteriorating terms of trade during the late 1970s, led to a decline in gross domestic product and a serious economic crisis during the early 1980s. These problems also reflected basic structural weaknesses and policy distor- tions in the economy that could be attributed to an inefficient production sector, re- sulting from price controls and massive direct interventions by government in agri- cultural input and output markets that favored the estate over the smallholder sector. Since the early 1980s the Government of Malawi has gradually addressed these policy distortions (Kherallah and Govindan 1997). However, major changes directly affecting the smallholder agricultural sector were implemented only cautiously, be- ginning with the liberalization of output markets during 1987–93, the dismantling of credit subsidies in 1993/94, the abolition of fertilizer subsidies in 1995, and the grad- ual relaxation of the tobacco quota system that eventually allowed smallholders (in 1996/97) to produce and market tobacco without any restrictions for the first time. Developments in Smallholder Maize and Tobacco Production during the 1990s Maize is the major food and crop in Malawi. Tobacco is the major cash and export crop. The reforms enacted during the 1990s brought about significant changes in the production of these crops by smallholders (Chilima, Chulu, and Mataya 1998). To- bacco and hybrid maize are also the major crops for which agricultural credit has been given in Malawi during the 1990s. The recent developments in the two sectors therefore have a direct bearing on the effect of access to credit on farm income. 7 High (but Recently Declining) Reliance on Maize During the 1980s about three-quarters of smallholders’ acreage was planted to maize. This share declined somewhat during the 1990s. Other food crops include cassava, sweet potatoes, groundnuts, and rice. Many of the two million smallholder households are chronically food deficient because of small farm size and low yields of the dominant local maize varieties. About 50 percent of smallholder households are food insecure, and 60 percent of the rural and 65 percent of the urban popula- tion earn incomes below the poverty line of US$40 per capita per year (Government of Malawi 1994a). Although the objective of macroeconomic reform and the liberalization of agri- cultural and financial markets was to reduce discrimination against the smallholder agricultural sector and to provide more opportunities for diversification of rural in- comes, until 1992/93 the agricultural credit, input, and extension policy continued to focus on the dissemination of a fixed input package of hybrid maize seed and fertilizer that was delivered at subsidized interest rates and input prices to small- holders. The policy of massive distribution of maize credit to smallholders was success- ful in increasing the share of higher-yielding hybrid maize in total smallholder hec- tarage planted to maize from about 8 percent in 1985 to 25 percent in 1992, while the overall share of maize in smallholder acreage increased from 73 percent to 80 percent. However, the concentration of the loan portfolio on one drought-sensitive crop, combined with the droughts in 1992 and political promises to write off loan debt during the election year, led to widespread loan defaults and eventually to the collapse of the parastatal Smallholder Agricultural Credit Administration (SACA) in 1994. Although 400,000 farmers received credit in 1992, only 34,000 did so in 1994, from the newly formed Malawi Rural Finance Company (MRFC), a state-owned fi- nancial institution that seeks to offer agricultural credit on a national scale. Following the major drought in 1992, the share of smallholder hectarage planted to nonmaize crops, in particular cassava and pulses, increased. Farmers’ response to the perceived advantages of drought-resistant crops, the sudden collapse of the public system for distributing credit for maize production, and the policy reorien- tation toward diversifying smallholder crop production may all have played a role in this. Following a second drought in 1993/94, large-scale distribution of free fer- tilizer and hybrid maize seed to drought-affected areas during 1994/95 and 1995/96 seems to have contributed to a revival of hybrid maize in smallholder farms despite the unfavorable ratio between maize price and fertilizer price after the abolition of fertilizer subsidies in 1995 and the devaluation of the Malawi kwacha during 1994/95 by about 300 percent. However, as Chilima, Chulu, and Mataya (1998) point out in their analysis of smallholder maize production, the area cultivated under drought-prone maize is slowly losing ground as the hectarage planted to more lu- crative crops (mainly tobacco) and more drought-resistant crops (such as cassava) expands. 8 The Booming Smallholder Tobacco Sector From the time that Malawi achieved independence until the early 1990s, smallhold- ers cultivating customary land were squeezed out of the lucrative export market in tobacco by a particular set of policies. The Special Crops Act of the Government of Malawi allowed for cultivation of tobacco and other export crops only on leasehold and freehold land (Sahn and Arulpragasam 1993; Sijm 1997). The production of bur- ley and flue-cured tobacco on customary smallholder land was illegal until 1990. Moreover, the system of allocation of tobacco production quotas to estates created economic rents for the powerful landed elite and reinforced the will of political forces to safeguard the country’s dualistic agricultural structure. In 1990 the Government of Malawi initiated a policy of gradual liberalization of the tobacco subsector in order to mitigate the structural constraints that had for so long prevented smallholders from contributing to and earning their share from the overall development of the agricultural sector. The production of burley to- bacco by smallholders on customary land was first permitted on a pilot basis during the 1990/91 growing season, when a total of 7,600 growers were registered to grow burley tobacco with a quota of 3.0 million kilograms. Quantities allocated to each grower were limited to a maximum of 300 kilograms. Smallholder tobacco had to be sold initially to the Agricultural Development and Marketing Corporation (ADMARC), a parastatal, at below-market prices. The evident success of the pilot scheme, combined with the democratic election of a new government and the related review of all policies implemented during the past three decades, led to a gradual in- crease of the quota allocated to smallholders. By 1996 the size of the smallholder quota had increased more than tenfold from its initial level. In view of the success of the tobacco market reforms in encouraging widespread participation of smallholders in direct competition with the estates, the Government of Malawi repealed the Special Crops Act in 1996 and opened up the production of burley tobacco to any grower in Malawi, regardless of whether or not he or she was formally registered to produce the crop. The repeal abolished the system of produc- tion quotas and special marketing rights, and thereby eliminated the rents of the es- tates that for decades had benefited from them. Rural Microfinance Institutions in Malawi In common with many other developing countries, Malawi has over the past few years seen the emergence of various rural credit programs. The four that are the fo- cus of this research are MRFC, a state-owned and nationwide agricultural credit program; Promotion of Micro-Enterprises for Rural Women (PMERW), a micro- credit program targeted at women in support of nonfarm income-generating activ- ities; the Malawi Mudzi Fund (MMF), a replica of the Grameen Bank; and the Malawi Union of Savings and Credit Cooperatives (MUSCCO), a union of locally 9 based savings and credit associations. Except for MUSCCO, all programs rely on group lending.1 There are numerous other small credit programs run by various national and in- ternational NGOs, which are often—but not always—implemented in collaboration with a Malawi government institution (see Evans 1993, or more recently Chirwa et al. 1996). However, of all these credit programs, only MRFC and MUSCCO can claim to have national coverage. All the other programs operate in a few districts and, in general, in conjunction with other noncredit developmental programs (Evans 1987, 1993; Government of Malawi 1994b). In the research, we have focused on these four microfinance institutions as rep- resentative of the spectrum of formal credit and savings options available to rural households in Malawi. Furthermore, the structures, target clienteles, rules, and types of loans of the four microfinance institutions are different enough to allow for a com- parative study of the effects of alternative design characteristics on their performance in terms of participation and effects on the livelihood of their clienteles. Malawi Rural Finance Company (MRFC) MRFC is a recent creation of the Government of Malawi, with funding from the World Bank, following the collapse of SACA. SACA was a department of the Min- istry of Agriculture that had provided seasonal agricultural loans to smallholder farmers since its establishment in 1987 (on the history, operations, and performance of SACA, see Murotho and Kumwenda 1996). Although MRFC inherited the operations of SACA in October 1994 and absorbed many of its staff, MRFC seeks to operate under commercial principles under a board of directors independent of the Government of Malawi. In fact plans call for the com- pany eventually to be privatized and transformed into a licensed rural bank (World Bank 1993; Government of Malawi 1994b). Apart from its portfolio of loans to es- tates, the target clientele of MRFC is smallholder farmers organized into joint lia- bility credit groups of 5 to 10 members. MRFC provides mostly in-kind seasonal agricultural loans for seed, fertilizer, and pesticides for hybrid maize and tobacco. It also offers short-term (two-year) and medium-term (five-year) loans for farm equip- ment, although these services play a negligible role in its overall loan portfolio to smallholders. The DRD/IFPRI survey data cover the 1993/94 and 1994/95 seasons. As such, most of the smallholder loans from MRFC in our sample are for hybrid maize and relatively few are for tobacco.2 The data do not capture some of the more recent shifts that MRFC has undertaken to develop credit services for off-farm micro-, small-, and 10 1 PMERW and MMF have since been incorporated into the MRFC. 2 For example, 50 percent of the chemical fertilizer acquired in 1994/95 by MRFC smallholder customers in our sample was used on hybrid maize, compared with 11 percent used on tobacco and 39 percent used on local maize (see Table 11). Most of the loans may have been for tobacco, but these figures indicate that MRFC smallholder cus- tomers diverted most of their loan packages toward their food crops. medium-scale enterprises in rural areas and its shift away from hybrid maize loans to tobacco loans. The latter shift, according to MRFC, was motivated by the below- average loan repayment rates for hybrid maize loans in 1994/95 and 1995/96. In turn, the low repayment rates appear to be strongly linked to the apparent risk and decline in profitability of hybrid maize production because of the devaluation of the Malawi kwacha and the abolition of fertilizer subsidies in 1994/95. The low profitability of hybrid maize compared with tobacco and the high downside risk of achieving nega- tive gross margins compared with the growing of low-input local maize varieties are also confirmed in the DRD/IFPRI survey data, as will be shown in the next chapter. During the 1994/95 season MRFC serviced 2,343 credit clubs, made of 81,075 smallholder farmers, and 4,394 estates, for a total of about 35 million Malawi kwacha (MK) and an average loan size of MK 5,600 or approximately US$370 (MRFC 1994, 1995).3 Table 1 shows that the total amount of loans disbursed by MRFC reached a total value of about MK 241 million in the 1995/96 season before declining to about MK 166 million in the 1996/97 season. The share of the disbursed loans going to to- bacco clubs has steadily increased during the three lending seasons: 41 percent in 1994/95, 45 percent in 1995/96, and 47 percent in 1996/97. The share of the loans disbursed to the estates has also experienced similar growth (29 percent, 33 percent, and 43 percent, respectively). In contrast, the share of the loans received by the other clubs (mostly for maize) and individual customers has been declining (31 percent, 22 percent, and 10 percent, respectively). As a consequence of its adherence to commercial lending practices, MRFC has been charging relatively high interest rates. Indeed its loans carried annual interest rates of 40 percent in 1994/95, 54 percent in 1995/96, and 37 percent in 1996/97. These high rates have been justified by the fact that MRFC obtains its funds at mar- ket rates from the Reserve Bank of Malawi and by the inflationary environment that characterized Malawi during these three seasons. The inflation rate, for example, was 83 percent in 1995 (Reserve Bank of Malawi 1996). MRFC’s loan recovery rate in 1994/95 was good (95 percent), but it deteriorated sharply in 1995/96 (76 percent) before rising again in 1996/97 (87 percent). The sharp increase in the size of the loan portfolio in the 1995/96 season likely played a role in the deterioration that year of the recovery rate, which, as shown in Table 1, was due to some extent to the very low recovery rate of the loans given to the other clubs (about 54 percent compared with about 91 percent in 1994/95). The 1995/96 recovery rates for tobacco clubs (82 per- cent) and the estates (84 percent) were also significantly lower than those in 1994/95 (96 percent and 98 percent, respectively).4 The loan recovery rate has improved in 1996/97 for all cases (92 percent, 84 percent, and 82 percent, respectively). Accord- 11 3 The exchange rate was US$1 for MK 15 in 1995. The average loan size for a smallholder farmer is much smaller than the figure given because the loan is for the whole group. 4 When we adjust for their respective shares of the loan portfolio, tobacco clubs, estates, and other clubs are re- sponsible for, respectively, 33 percent, 24 percent, and 43 percent of the 20 percent decline in the overall loan re- covery rate (from 95 percent to 76 percent). ing to its internal records, the company earned profits from its lending in all lending seasons (MRFC Corporate profile 1998). Malawi Mudzi Fund (MMF) MMF was created in 1987 as a pilot credit program and a separate component of the World Bank–funded agricultural credit project that also supported SACA. The com- ponent for MMF has been supported by the International Fund for Agricultural De- velopment (IFAD). Its design was guided by the experience of the Grameen Bank in Bangladesh. The objective of the MMF was to provide loans for nonfarm income-generating activities to poor rural households with less than 1 hectare of land in two districts of Malawi (Chiradzulu and Mangochi) during a pilot phase of five years (World Bank 1987). From the start of its lending operations in 1990 to April 1995 (the point at which it was absorbed by MRFC), MMF granted 2,676 loans. The mostly female borrowers (95 percent of the loans) were organized into 561 credit groups, each with five members. The members were held individually and jointly responsible for the repayment of all loans obtained by those in the group. A cumulative total of MK 841,000 was disbursed by MMF, with an average loan size of MK 300 or US$20 (Murotho and Kumwenda 1996). Most of the MMF loans were given for the sale of produce (fish, maize, beans, and so forth) and other small-scale trading activities. Few loans were given for crop production, and of those most were for growing hy- brid maize (MMF 1994). 12 Table 1—Loan disbursements and recovery rates of the Malawi Rural Finance Company Total amount of loans Type of Number of borrowers disbursed Percentage credit recovered borrowers 1994/95 1995/96 1996/97 1994/95 1995/96 1996/97 1994/95 1995/96 1996/97 (1,000 MK)a (percent) Total 6,207 13,946 11,003 34,941 240,882 166,203 95.13 76.29 87.16 Tobacco clubsb 1,407 3,476 2,968 14,452 107,491 77,638 96.24 81.74 91.71 (41%) (45%) (47%) Estates 3,305 7,931 6,424 10,243 80,383 71,532 98.02 83.63 83.50 (29%) (33%) (43%) Other clubsb and entities 1,495 2,539 1,611 10,245 53,008 17,032 90.69 54.11 81.82 (31%) (22%) (10%) Source: Malawi Rural Finance Company internal documents (various issues). a The exchange rate was 1 U.S. dollar for 15 Malawi kwacha (MK) in 1995. The share of the total amount disbursed is in parentheses. b Each club has on average between 15 and 20 smallholder members who share a single loan issued to the club. Most of the credit clubs are either tobacco or maize clubs. In the first two years of its operation MMF was lending to both male and female borrowers. A very high default rate among male borrowers has since led MMF to concentrate its lending on women only (MMF 1994). Owing to its pilot nature and the close supervision and intensive training afforded credit recipients, MMF has been characterized by high operating costs per borrower served. In April 1995, MMF’s operation and groups were incorporated into MRFC. Under MRFC plans call for the MMF program to receive national exposure and become MRFC’s tool for reaching the poorest 25 percent in Malawi by providing them with loans for both nonfarm in- come-generating activities and agricultural production (World Bank 1994; MRFC Annual report 1996). Malawi Union of Savings and Credit Cooperatives (MUSCCO) MUSCCO is a federation of locally based savings and credit cooperatives (SAC- COs). It was created in 1980 with financial and technical support from the United States Agency for International Development (USAID). Its objective is to provide credit and savings options to those low-income people not serviced by commercial banks. This goal was to be achieved by promoting, organizing, and expanding the number and membership of the very few savings and credit cooperatives that existed at that time in Malawi (Reeser et al. 1989). Originally MUSCCO operated only in rural areas, servicing the financial needs of the few relatively better-off farmers. However, in 1985, with the response to its savings products by its rural clientele deemed unsuccessful, it refocused its activities on urban areas. By 1993 160 SAC- COs with a total membership of 23,000 were affiliated with MUSCCO. Of these 41 percent are located in urban areas and the remainder in rural towns (Evans 1993). Following cooperative principles, MUSCCO members buy shares in their respec- tive societies. For a member to qualify for a loan, he or she must have accumulated MK 100 in shares and a minimum of MK 50 in savings (MUSCCO 1994). The loan policies of the SACCO also stipulate that some form of collateral is required before a loan can be given to a member. There were 12,750 borrowers in 1993 (over 80 percent of whom were males) for a total of MK 7 million disbursed. On average the SACCO loans ranged from MK 700 to MK 7,000 (that is, from US$50 to US$500), with a ma- turity of between one and two years. The loans were used both for agricultural pro- duction (43 percent) and for nonfarm income-generating activities (Evans 1993). The Nafisi SACCO of Dowa, which was selected for this study, was created in 1990, initially capitalized by a US$12,300 grant from the Trickle Up Program of New York. Its members are relatively poor farmers who obtain loans almost exclu- sively for seasonal agricultural inputs such as fertilizers and seeds (VEZA Interna- tional 1994). The functioning of the Nafisi SACCO was closely linked to a local NGO, the Hills of Dowa Enterprise Zones Association (HODEZA). This NGO has supported the SACCO through technical assistance and logistical support in its day- to-day operations and in the marketing of its members’ maize crop. HODEZA itself is the local counterpart of a Chicago-based NGO called Village Enterprise Zone Associations International. 13 Promotion of Microenterprises for Rural Women (PMERW) Credit Program The PMERW credit program was started in 1986 by the Ministry of Women and Chil- dren’s Affairs and Community Services (MOWCACS) with the technical and finan- cial support of the German Agency for Technical Cooperation (GTZ). The program began as a multiservice developmental project with a small and loosely structured credit component. It introduced small-scale nonfarm income-generating technolo- gies to rural areas and provided business training and technical advice to women or- ganized into group-owned enterprises. The program was initially targeted to rural poor women with landholdings of less than 0.5 hectare in rural growth centers in Dedza, Mangochi, Nkhotakota, and Rumphi (Evans 1993). The program relied on the cooperation of the district com- munity development officers (DCDO) and community development assistants (CDA) of MOWCACS, who—apart from their other duties—organized and super- vised the women’s groups and provided them with business training and advice. The CDAs were also in charge of delivering and recovering the loans given individually to the women. Owing to management and operational problems, which resulted mainly from tying credit to developmental interventions, coupled with lax loan de- livery and recovery procedures, the credit component did not meet its objectives dur- ing the first phase of the project, which ended in 1989 (Zingani 1991; Evans 1993). Learning from this failed experience, a new and well-structured group-based credit program, separated from the small-scale technology development and busi- ness training program, was designed and implemented in 1991 with the help of a Kenyan NGO, the Undugu Society. The society trained the DCDOs and CDAs as trainers in group-based lending and credit management concepts. This new credit program, identified in this report as PMERW1, is a revolving fund operated by MOWCACS that gives two-year loans of MK 1,000 (approximately US$70) to savings-and-credit clubs, each made up of 10 to 15 poor entrepreneurial women who have completed training courses, conducted by the CDAs, in credit rules, management, and responsibilities.5 In order to be eligible for the MK 1,000 loan, the savings-and-credit club must have the equivalent of 60 percent of the loan amount in a post office savings account. The MK 1,000 loan is in turn distributed to half of the club’s members in smaller loans of two months’ maturity not exceeding MK 300 and carrying an annual interest rate of 30 percent. The other half of the club’s members must wait until the first half have fully repaid their loans before they are eligible for their own loans. Thus at any time during the two years only half of the club can receive loans. In addition to this peer-pressure device, each member is re- quired to have MK 20 of savings and two guarantors within her group before getting a loan. The individual loans are exclusively for nonfarm income-generating activi- ties that consist mostly of produce selling and beer brewing. It is expected that after 14 5 All monetary figures regarding the loans quoted in this section are prior to the October 1994 devaluation of the Malawi kwacha. As a general rule the PMERW program doubled all the amounts given in this section after the de- valuation. For example, each savings-and-credit club received a MK 2,000 loan in 1996. two years the savings-and-credit club would reimburse the MK 1,000 loan and would have generated enough funds through the savings and interest charges on the indi- vidual loans to be self-financing thereafter, thus enabling the ministry to lend the re- leased funds to newly formed clubs (PWRA 1993b). At the end of July 1994 there were 34 savings-and-credit clubs operating in Mangochi (13), Nkhotakota (12), and Rumphi (9), with a total of 506 women. At that time, the clubs’ repayment rates were over 95 percent. The average amount saved per club was MK 500, and 11 of the 14 clubs that were supposed to pay back their MK 1,000 loans had doubled the initial amounts (Faltermeir 1994). A second credit program, identified in this report as PMERW2, was started by MOWCACS/GTZ in 1993 in collaboration with the Commercial Bank of Malawi (CBM). The PMERW2 program is made of credit groups with 5 to 10 woman mem- bers who are skilled in business activities (PWRA 1993a). The credit groups func- tion more or less like the savings-and-credit clubs except that they receive their loans directly from CBM and the individual members can borrow up to MK 1,000. Credit group members are selected, as part of a loan graduation process, from among those savings-and credit clubs members who have excellent credit and business manage- ment skills. Successful women with business investments in the range of MK 300–1,000 and who live in the areas covered by the program can also be admitted as credit group members even if they did not previously belong to a savings-and-credit club. The loans given to credit groups by CBM are guaranteed up to 70 percent by a MOWCACS/GTZ fund maintained in an account at CBM. As of October 1994 there were 28 credit groups operating in the districts of Mangochi (10), Nkhotakota (10), and Rumphi (8), with a total of 280 members (PWRA 1993a, 1995). 15 CHAPTER 3 Survey Design and Description of the Data To study the impact of household access to formal credit, one needs a sample containing a sufficient share of households participating in the credit programs operating throughout Malawi. The data used in this study come from a year-long, three-round survey of 404 households in 45 villages in five districts of Malawi where the four microcredit programs studied were operating. Figure 1 shows the location of the survey sites. Sampling Methodology The first round of the survey took place in February-April 1995, the second round in July-August 1995, and the last round in November-December 1995. Despite the fact that there are numerous credit programs operating in various parts of the country, credit program participation is still rare, occurring in only very few villages. Out of 4,699 households enumerated in the 45 villages covered in the village census un- dertaken for the survey, only 12 percent were current members of a credit program. Moreover, the 12 percent figure significantly overstates the likelihood of credit pro- gram membership in Malawi because it represents the percentage of membership in villages that are actually hosting the four credit programs studied, and the majority of villages in Malawi do not host any credit program. The very low density of program participation in Malawi alone rules out straight random sampling at any geographic level above the village level. Since it was nec- essary to include enough credit program participants for the study, the only feasible alternative was to stratify along the program membership status variable with ran- dom selection within each stratum. Thus about half of the sample members were se- lected from participants in the four credit programs. The second half of the sample was equally divided between past participants (mostly from SACA, the failed gov- ernment credit program) and households who had never participated in any formal credit program. To correct for the oversampling of credit program participants, the 16 CHAPTER 3 Survey Design and Description of the Data summary statistics in the tables have been weighted using the strata population weights from the village census.6 Description of the Data The information collected in the survey included data on household demographics, land tenure, agricultural production, and livestock ownership; asset ownership and transactions; food and nonfood consumption; credit, savings, and gift transactions; wages, self-employment income, and time allocation; and the anthropometric status 17 6 See Chapter 4 for the sample selection correction in the econometric analysis. Figure 1—Location of the DRD/IFPRI Rural Finance Survey sites of preschoolers and their mothers. The agricultural data cover the 1993/94 and 1994/95 seasons. Given the central importance of the credit limit variable for the methodology of the study, we describe in greater detail how the data for this variable were collected. The rationale behind the procedure described here and the issues involved in the in- terpretation of the credit limit variable thus collected are discussed in Chapter 4. The questionnaire on credit and savings was administered to all adult household members (those over 17 years of age) in the sample. In each round respondents were asked the maximum amount they could borrow during the recall period from both in- formal and formal sources of credit.7 If the respondent was involved in a loan trans- action as a borrower, the question was asked for each loan transaction (for both granted and rejected loan demands). In this case the credit limit refers to the time of borrowing and to the lender involved in that particular loan transaction. If the re- spondent did not ask for any loan, the question was asked separately for formal and informal sources of credit with no reference to particular formal or informal lenders. Respondents who were granted loans were also asked the same general question (that is, with no reference to particular formal or informal lenders) in a way that elicited the credit limit they would face if they wanted further loans, not just from the same lender but from the same sector of the credit market (formal or informal) within which they had previously borrowed. Consequently, for both formal and informal credit, the formal and informal credit limits of each adult household member were obtained in each round, even if the respondent was not involved in any loan transaction. Several other control questions were used to verify the consistency of the answers given by the respondents to this question. Such control questions included the fol- lowing: What was their program membership status? If they did receive a loan of the same type, were they given a lesser amount than they had asked for, and, if so, how much had they asked for? Had they asked for a loan and been rejected? Why did they not ask for any (or any further) loans? In addition the enumerators were instructed to use other control questions not included in the questionnaires whenever there seemed to be inconsistencies in a respondent’s answers (such as, where could they borrow a given amount). A good deal of time was further spent in the field and in the office checking the consistency of answers to these questions and their relation to an- swers given on other parts of the questionnaire. As a result of these checks, during the first round of the survey most of the respondents were visited at least twice, in order to verify their answers or clarify some of the apparent inconsistencies in their answers. Most of the inconsistencies occurred during the first days of the survey and resulted from some misunderstanding of a question that was often interpreted by ei- ther the enumerator or the respondent as asking about “the maximum you would like to borrow.” This misunderstanding was resolved by instructing the enumerators to 18 7 Loans received prior to October 1994 were also recalled (up to three years prior to the 1994/95 season for formal loans and up to 10 months prior to October 1994 for informal loans of more than MK 100). explain to respondents, before they answered the question, the difference between the two questions.8 Demographic Characteristics of Households We begin by presenting selected demographic characteristics of sample households. Table 2 shows that 28 percent of the households in the sample are headed by women. This figure is close to the widely cited figure of 30 percent for Malawi as a whole. The table also shows that the average household size in the survey areas is 5 persons, and that it is the same for both male- and female-headed households. However, with an average dependency ratio of 0.5, female-headed households have slightly more dependents than male-headed ones (0.4).9 The average age of household heads in the sample is 42, with female heads of households being, on average, four years older than male ones (44 versus 40). Some 68 percent of household heads attended pri- mary school, but only 17 percent of them have a primary school diploma. Overall, female household heads tend to have a lower primary school attendance rate com- pared with male heads (63 percent versus 71 percent). Table 2 also shows the main occupation of household heads in the sample. Farm- ing dominates as first occupation of most household heads (66 percent). This is true for both male heads and female ones (62 percent and 74 percent, respectively). As separate categories, wage laborer and trader come second as first occupation (8 per- cent), while all the other self-employment income-generating activities grouped to- gether constitute the first occupation for 10 percent of household heads. However, female household heads are four times more likely than male heads to list trader as their first occupation (16 percent compared with 4 percent). The opposite is true for wage laborer (11 percent of male household heads versus 2 percent of female heads). Only 7 percent of female household heads list household work as their first occupa- tion. More than two-thirds of household heads have a second occupation, but fewer than a quarter of them have a third occupation. Many of those with a third occupa- tion are female household heads doing farming, household work, and trading. Household Asset Ownership, Composition, and Distribution Asset ownership is arguably an important determinant of access to credit, especially if creditworthiness is judged on the basis of wealth or landed collateral alone. Land, traditionally the most important form of collateral, has been recognized as one of the major constraints in the agricultural sector of Malawi, one of the most densely pop- ulated countries in Africa. Therefore we present data on households’ ownership of 19 8 Further details on the survey and the data collection methodology are reported in Diagne, Zeller, and Mataya (1996) and Simtowe and Diagne (1998). 9 The household dependency ratio was calculated as the ratio of the household population younger than 15 or older than 64 to the household size. various types of assets, including land, livestock, farm and nonfarm productive equipment, and other nonproductive assets. Assets classified as nonproductive con- sist of noncultivable land, buildings, furniture, and household utensils. The intra- household distribution of ownership of assets and differences among credit program participants and nonparticipants are also discussed because of their influence on the control and allocation of household income. Table 3 shows that the average total value of all household assets is approximately MK 6,700 or approximately US$450. The average values of land and livestock are, re- 20 Table 2—Demographic characteristics of households Male Female All Sample size 291 (72%) 111 (28%) 402 Household size 5 5 5 Adult equivalent population 3.6 3.6 3.6 Dependency ratio 0.4 0.5 0.4 Mean age of head 40 44 42 (percent) Head attended primary school 71 63 68 Head has primary school diploma 21 10 17 First occupation of head Farming 62 74 66 Household work 0 7 3 Wage laborer 11 2 8 Trader 4 16 8 Other self-employment 14 1 10 Student 1 0 0 Unemployed 3 0 2 Other 5 0 3 Second occupation of head Farming 37 9 28 Household work 27 13 22 Wage laborer 11 74 32 Trader 5 0 3 Other self-employment 6 3 5 Student 14 2 10 Unemployed 0 0 0 Other 1 0 0 Third occupation of head Farming 79 75 78 Household work 2 5 3 Wage laborer 4 7 5 Trader 2 0 1 Other self-employment 4 10 6 Student 6 0 4 Unemployed 2 0 2 Other 2 3 2 Source: DRD/IFPRI Rural Finance Survey. 21 Table 3—Asset ownership, composition, and distribution Male-headed Female-headed Total Sample size 291 111 402 Average value of: (MK) All assets 7,551 4,841 6,681 Land 3,866 2,148 3,306 Productive assetsa 4,537 3,343 4,154 Livestock (total) 1,440 1,848 1,571 Nonproductive assetsb 3,014 1,498 2,528 Share of assets held in the form of: (percent) Productive assets 57 58 57 On-farm assets 43 46 44 Livestock 11 12 11 Land 56 59 57 (hectares) Average size of land holdings 1.8 1.4 1.7 Household with: (percent) Less than 0.5 hectare 4 9 6 0.5–1.0 hectare 17 21 18 1–1.5 hectares 24 35 28 1.5–3 hectares 42 30 38 Over 3 hectares 12 5 10 Share of assets owned by: Head 82 85 83 Spouse 14 3 11 Joint (head and spouse) 2 . . . 2 Other 2 12 4 Hectares of land owned by: Head 76 90 80 Spouse 21 6 17 Joint (head and spouse) 2 0 2 Other 1 4 1 Share of on-farm assets owned by: Head 68 92 73 Spouse 27 6 22 Joint (head and spouse) 1 . . . 1 Other 4 2 4 Share of cultivable land owned by: Head 63 91 69 Spouse 29 6 24 Joint (head and spouse) 3 0 2 Other 5 2 4 Share of livestock owned by: Head 85 73 81 Spouse 3 0 2 Joint (head and spouse) 12 0 7 Other 0 27 10 Share of cattle owned by: Head 99 100 99 Spouse 1 0 1 Joint (head and spouse) 0 0 0 Other 0 0 0 Source: DRD/IFPRI Rural Finance Survey. a Noncultivable land, buildings, furniture, and household utensils. b On-farm assets (cultivable land, farm equipment, and oxen) and livestock. spectively, MK 3,300 and MK 1,600. In total the productive assets, including the ones for off-farm income-generating activities, make up 57 percent of the value of house- hold assets. The on-farm assets (cultivable land, farm equipment, and oxen) and live- stock constitute, respectively, 44 percent and 11 percent of the total value of household assets. There are noticeable differences between male- and female-headed households. In particular the total value of all assets is significantly higher for male-headed house- holds (MK 7,600) compared with female-headed ones (MK 4,800). Female-headed households also own noticeably less land than male ones (an average of 1.4 hectares versus 1.8 hectares, respectively). Overall, land is very scarce; more than half of all households in the survey areas (52 percent) have landholdings of less than 1.5 hectares. With regard to the intrahousehold ownership of assets, Table 3 shows that house- hold heads own more than 80 percent of the total value of all household assets com- pared with only 11 percent for spouses. The disaggregated figures show that, on av- erage, 80 percent of households’ land is owned by the household heads. Spouses own only 17 percent of land, and only 2 percent of land is jointly owned by heads and their spouses. Overall, spouses own 22 percent of households’ on-farm assets, with their shares for the different types of household assets being highest for cultivable land (24 percent). On the other hand, they own, on average, only 2 percent of the value of household livestock.10 The intrahousehold distributions of other household assets show more or less the same pattern as that for land. Table 4 differentiates the household asset ownership by participation in credit programs. The average total value of household assets of current credit program par- ticipants (MK 13,000) is more than twice the values for past participants and non- participants, which are about MK 5,000. Nonparticipants also have noticeably lower average landholding sizes (1.4 hectares) compared with current participants (2.3 hectares) and past participants (1.9 hectares). Moreover, 31 percent of nonpartici- pants have landholdings of less than 1 hectare. However, household members of PMERW and MMF are noticeably more likely to have landholdings of less than 1 hectare (about 20 percent of members) than those of MRFC (3 percent of members) or of MUSCCO (10 percent of members). Hence, even if their members are relatively wealthy in terms of assets compared with nonparticipants, these two programs still have the highest proportion of landless among the programs studied. Table 4 also shows that land ownership tends to be more evenly distributed be- tween heads and spouses in MRFC member households than in households belong- ing to any one of the other groups (including past participants and nonparticipants). Heads and spouses of MRFC member households own, respectively, 50 percent and 43 percent of household total land, whereas in the other programs and for nonpar- ticipants spouses own no more than 22 percent. In all cases joint ownership of land does not exceed 4 percent except for PMERW1 members, for whom it reaches 16 percent. In MRFC member households spouses own even significantly more cul- 22 10 The livestock ownership figures reflect cultural practices preventing women from owning cattle. Indeed virtually all the cattle in male-headed households (99 percent) are owned by heads, and female-headed households have vir- tually no cattle, although in terms of value they have more livestock (mostly poultry) than male-headed households. tivable land and on-farm assets than their husbands (53 percent and 50 percent, re- spectively, for spouses versus 29 percent and 35 percent, respectively, for the hus- bands). MMF member spouses in male-headed households also own up to 41 per- cent of their households’ cultivable land and value of on-farm assets. On the other hand, PMERW2 member spouses have the lowest shares of cultivable land (4 per- cent) and value of on-farm assets (4 percent). Three main conclusions can be drawn from the discussion of descriptive statis- tics for asset ownership: 1. Even when credit is targeted to the poorest segment of rural households— the approach taken in theory by the PMERW1 and MMF programs—the value of household assets and household landholding size seem to be posi- tively correlated with participation in formal credit programs. 2. The intrahousehold ownership distribution of assets, especially with regard to land, confirms the widespread belief that women in general are in a very weak position in terms of control of household resources. Furthermore, since land is the most common asset pledged as collateral for credit (when it is re- quired), one can conclude from these figures that women’s access to credit may strongly depend on the will and priorities of their husbands. Therefore PMERW, and to a lesser extent MMF, seem to have given access to credit to a class of women living in relatively wealthy households in terms of assets compared with nonparticipants, but who are in a very weak position in terms of control of their households’ resources. By providing loans for only non- farm income-generating activities, the two programs are focusing on invest- ment opportunities that are appropriate for their target clientele. 3. The apparent gender differences in the membership composition of MRFC and the MUSCCO-affiliated Nafisi SACCO of Dowa (38 percent and 3 per- cent of whose members, respectively, are women), both providing almost ex- clusively seasonal agricultural loans, seem to be the result of the significant differences in the intrahousehold land distribution figures between the two programs. Hence, by giving out loans exclusively for agricultural production purposes in an area characterized by a very unequal distribution of land be- tween male heads of households and their spouses, the Nafisi SACCO is de facto discriminating against women. Structure of the Formal and Informal Credit Markets in Malawi In this section we present evidence on the level of rural households’ access to formal and informal credit in Malawi. As discussed in the introduction, the credit limit is used to assess the extent of that access as well as the proportion of households having a binding credit constraint. Before presenting evidence on the level of access, we de- scribe briefly the structure of the formal and informal credit markets and some of the main attributes of the loan transactions recorded. The analysis distinguishes the for- mal and the informal sectors of the credit market because they provide different types of credit services. The formal sector comprises the government- and NGO-supported 23 24 Table 4—Asset ownership, composition, and distribution by credit program membership Current members Past Never been MRFC MMF MUSCCO PMERW1 PMERW2 Other All members members Sample size 86 32 29 62 37 15 231 65 106 Average value of: (MK) All assets 14,379 15,720 7,094 14,075 16,704 32,238 13,168 5,096 4,987 Land 9,578 8,213 3,263 7,974 6,616 35,733 7,840 2,857 1,919 Livestock (total) 1,381 169 1,539 3,120 3,861 2,399 1,720 1,199 1,608 Productive assetsa 7,936 7,003 4,861 6,806 9,875 17,125 7,454 3,442 3,270 Nonproductive assetsb 6,443 8,717 2,233 7,269 6,829 15,113 5,715 1,654 1,717 Share of assets held in the form of: (percent) Productive assets 58 42 68 53 54 53 58 65 55 On-farm assets 40 32 52 37 32 39 41 49 44 Livestock 15 1 15 15 16 13 14 14 9 Land 54 61 60 61 43 58 54 60 57 (hectares) Average 2.5 2.3 2.2 1.9 2.6 3.6 2.3 1.9 1.4 Household with size of landholdings: (percent) Less than 0.5 hectare 0 6 0 0 13 1 2 1 8 0.5–1.0 hectare 3 14 10 23 6 20 8 14 23 1.0–1.5 hectare 34 11 22 22 26 14 32 29 26 1.5–3 hectares 28 36 51 40 27 43 32 47 38 Over 3 hectares 36 33 17 16 28 21 27 9 5 Share of assets owned by: Head 69 81 94 66 81 49 71 84 93 Spouse 21 18 6 7 10 29 21 16 7 25 Joint (head and spouse) 4 0 1 26 9 10 4 . . . 0 Other 5 0 0 0 0 12 4 0 0 Hectares of land owned by: Head 50 75 87 72 88 48 56 81 87 Spouse 43 22 12 12 10 25 38 19 11 Joint (head and spouse) 4 3 0 16 2 12 4 1 2 Other 3 0 0 0 0 15 2 0 0 Share of on-farm assets owned by: Head 35 59 93 73 95 7 41 78 89 Spouse 50 41 7 9 4 64 46 22 10 Joint (head and spouse) 2 0 0 18 2 3 2 . . . 1 Other 14 0 0 0 0 26 11 0 0 Share of cultivable land owned by: Head 29 57 92 63 94 6 35 76 87 Spouse 53 41 8 9 4 62 49 23 11 Joint (head and spouse) 4 2 0 27 2 7 5 1 2 Other 15 0 0 0 0 26 11 0 0 Share of livestock owned by: Head 92 9 96 78 94 93 89 98 81 Spouse 4 83 2 12 6 4 6 2 1 Joint (head and spouse) 4 8 2 10 1 3 5 0 18 Other 0 0 0 0 0 0 0 0 0 Share of cattle owned by: Head 100 . . . 100 100 100 100 100 100 100 Spouse 0 . . . 0 0 0 0 0 0 1 Joint (head and spouse) percent) 0 . . . 0 0 0 0 0 0 0 Other 0 . . . 0 0 0 0 0 0 0 Source: DRD/IFPRI Rural Finance Survey. a Noncultivable land, buildings, furniture, and household utensils. b On-farm assets (cultivable land, farm equipment, and oxen) and livestock. credit programs, the microfinance institutions, and the commercial banks. The infor- mal sector is made up of professional moneylenders, traders, and friends and relatives. Table 5 shows that there were 364 formal and 212 informal loans granted during the recall period. For formal loans, the recall period was chosen to begin with mem- bership in the credit program (for SACA, only since 1992). Larger informal loans (more than MK 100) were recalled from October 1993 until the time of the round, with the last round ending on December 20, 1995. For informal loans of less than MK 15 and ones between MK 15 and MK 100, the recall period in each of the three rounds was chosen as eight weeks and three months, respectively. A total of 121 demands for loans were rejected, 56 percent of which were rejected by informal lenders. In total 79 percent of adults in the sample did not ask for any loan during the three rounds of the survey. The most common reason for not asking for a formal or infor- mal loan was dislike or no need of borrowing (48 percent and 27 percent for infor- mal and formal loans, respectively). Informal loans were mostly between friends and relatives (93 percent). The majority of them did not have any due date (57 percent). Virtually all informal loans were interest-free loans (98 percent) with an average size of MK 76. In contrast, formal loans carried an average annual interest rate of 24 per- cent before October 1993 and 39 percent after October 1993, and their average size was MK 377 before October 1993 and MK 520 after October 1993. These figures show that the credit market in Malawi is not as active as those in other Asian and African countries.11 Distribution of Credit Limits and Unused Credit Lines Table 6 presents the average informal and formal credit limits and unused credit lines that were observed in the three rounds. This information is shown for the whole pop- ulation and separately when a formal loan was granted, rejected, and not requested. The average formal and informal credit limits for the population as a whole are MK 44 and MK 46, respectively (about US$3).12 To put these figures into perspective, Malawi’s 1995 per capita GNP was US$170 or MK 2,550 (Word Bank 1997), and the average per capita 1995 income in the sample was MK 1,190. The average for- mal credit limit is significantly higher for cases in which formal loans were granted (MK 679 on average) compared with cases in which informal loans were granted (MK 35). One also notes that some rejected borrowers and respondents who did not ask for loans could nevertheless borrow some positive amounts from both sectors.13 26 11 For comparison, 2,233 informal and 338 formal loans were recorded in Bangladesh in a similar IFPRI survey in 1994 involving 350 households (Zeller, Sharma, and Ahmed 1996). In another similar IFPRI survey of 189 house- holds in Madagascar in 1992, there were 1,375 and 245 informal and formal loans, respectively (Zeller et al. 1994). 12 The exchange rate was US$1 for MK 15 at the time of the survey. 13 A small number of borrowers whose loan demands were rejected could borrow a lesser but positive amount from the same lender but chose not to do so. The main reason why a rejected borrower chooses to forego a loan instead of accepting a lesser amount is that the lesser amount is usually too small for the intended purpose of the loan. In addition, when a loan demand is rejected by one sector of the credit market (formal or informal), the potential bor- rower can often borrow at least some amount from another sector. 27 Table 5—Loan transactions and their characteristics Male Female All Sample size 1,087 1,361 2,448 Informal credit granted 113 (60%) 99 (59%) 212 (59%) Loan size (MK) 82 67 76 Loan maturity (weeks) 13 9 11 Loans with due date (percent) 48 36 43 Loans with no due date (percent) 52 64 57 Percent annual interest rate (percent) 6 2 4 Interest-free loans (percent) 97 98 98 Relation with informal lender Friend or relative (percent) 95 92 93 Neighbor (percent) 2 3 3 None of the above (percent) 3 5 4 Formal credit granted 143 (40%) 221 (41%) 364 (41%) Before October 1993 Loan size (MK) 456 161 377 Loan maturity (weeks) 42 41 42 Loans with due date (percent) 100 100 100 Percent annual interest rate (percent) 24 23 24 Loans with positive interests (percent) 96 100 97 After October 1993 Loan size (MK) 670 449 520 Loan maturity (weeks) 38 22 27 Loans with due date (percent) 100 100 100 Percent annual interest rate (percent) 37 40 39 Loans with positive interests (percent) 94 98 97 Loans rejected Informal loans 49 (55%) 27 (58%) 76 (56%) Formal loans 31 (45%) 14 (42%) 45 (44%) No loan requested 751 (72%) 1,000 (85%) 1,751 (79%) Reasons for not asking for informal loans I did not need credit 27 23 24 I dislike any borrowing 29 20 24 Other loans are too expensive 14 16 15 I felt that lender would refuse because of: My age 3 8 6 My health problem 5 6 6 Reasons other than above 16 18 17 Other 5 9 7 Reasons for not asking for formal loans (percent) I did not need credit 19 19 19 I dislike any borrowing 8 8 8 Other loans are too expensive 16 14 15 I felt that lender would refuse because of: My age 8 14 12 My health problem 7 6 6 Reasons other than above 23 24 24 Other 19 14 16 Source: DRD/IFPRI Rural Finance Survey. Note: The percentage figures in the table are weighted using the strata population weights from the vil- lage census (the count figures are not weighted). 28 Table 6—Distribution of formal and informal credit limits and unused credit lines, October 1993–December 1995 Credit limit and unused credit linea Formal Informal Standard Standard Mean Median deviation Minimum Maximum Mean Median deviation Minimum Maximum All respondents 44 0 248 0 10,000 46 0 188 0 12,000 (19) (0) (137) (0) (6,575) (36) (0) (112) (0) (5,000) When formal loan was granted 679 500 911 13 10,000 95 20 500 0 12,000 (148) (0) (474) (0) (6,575) (69) (10) (202) (0) (4,000) When informal loan was granted 35 0 149 0 1,000 127 50 369 5 12,000 (13) (0) (76) (0) (1,000) (52) (12) (134) (0) (4,000) When loan demand was rejected 72 0 254 0 4,000 46 0 89 0 400 (53) (11) (215) (0) (4,000) (34) (0) (69) (0) (300) When no loan was requested 12 0 88 0 5,000 32 0 104 0 5,000 (12) (0) (88) (0) (5,000) (32) (0) (104) (0) (5,000) Source: DRD/IFPRI/Rural Finance Survey. Notes: The exchange rate is US$1 = MK 15. Malawi’s 1995 per capita GNP is $170 (that is, MK 2,550; World Bank 1997). a Unused credit line in parentheses. The distributions of the credit limits and unused credit lines presented in the Table 6 and in the box plot diagrams in Figures 2–6 give a better picture of the extent of access to credit in Malawi. The median formal and informal credit limits in the pop- ulation as a whole are both zero. Over 75 percent of the population can borrow at most MK 50 (about US$3) from either sector of the credit market, and most often they could obtain this amount only from the informal sector. The distributions of the unused credit lines show that more often borrowers exhaust their credit lines in the formal sector but not in the informal sector. This finding, taken together with the fact that informal loan sizes and credit limits are significantly lower than the correspon- ding formal values, suggests that the two types of credit are not perfect substitutes for one another. Otherwise, since almost all informal loans do not carry any interest rate, one would expect to see households reach their credit limits more frequently in the informal sector than in the formal sector. One also notes that women in general have lower formal and informal credit limits compared with men. They also appear more likely to exhaust their formal credit lines than men. This finding provides some justification for the targeting of formal credit to women. Access to Credit and Participation in Formal Credit Programs A household is said to have access to a type of credit if at least one of its members has a strictly positive credit limit for that type of credit. Similarly a household is classified as credit constrained for a type of credit if at least one of its members is constrained for that type of credit. How do access to the two types of credit and the likelihood of having a binding credit constraint differ between participants and nonparticipants in credit programs? To answer this question, the households have been classified according to the types of access to credit and the binding of the credit constraints of their individual members. Table 7 tabulates the credit limits and occurrence of credit constraints by program membership. Consistent with our conceptual distinction between access to credit and participation in a credit program, the table shows that 8 percent of households who never participated in any credit program did have access to formal credit during the first-round recall period (that is, they said they could obtain a formal loan if they wanted to). Of households who never participated in a credit program 28 percent do not have access to any type of credit, while 64 percent have access only to informal credit. A different pattern of access to credit is shown for households who are no longer participating in credit programs. Indeed, they are four times more likely to have access to formal credit than those who never participated (32 percent compared with 8 percent). Interestingly, up to 40 percent of the households currently partici- pating in formal credit programs did not have access to formal credit during the first- round recall period. This means that not only did they not receive any formal loan during that period, they also could not borrow anything from a formal lender. Table 7 also shows that close to half of households participated in formal credit programs during that recall period (with 15 percent having their formal and informal binding and 34 percent having only their formal binding). This indicates that these house- 29 30 2,2582,258N = Credit limit (MK) 1,000 800 600 400 200 0 Formal Informal 1,312945 1,312945N = Credit limit (MK) 1,000 800 600 400 200 0 Male Female 2,258 Informal 2,258 Formal N = Unused credit line (MK) 1,000 800 600 400 200 0 1,312945 1,312945N = Unused credit line (MK) 1,000 800 600 400 200 0 Formal Informal Male Female Figure 2—Distributions of formal and informal credit limits and unused credit lines for all respondents, October 1993–December 1995 A. Formal and informal credit limits for all respondents B. Formal and informal unused credit lines for all respondents Notes: The box plot diagrams are interpreted as follows. For each box, 50 percent of cases have values within the box and the solid horizontal line inside it is the median. The length of the box is the interquartile range and the lower boundary (resp upper boundary) of the box is the 25th (resp 75th) percentile. The cir- cles are outliers and the stars are extreme values. The exchange rate is US$1 = MK 15. Malawi’s 1995 per capita GNP was US$170 or MK 2,550 (World Bank 1997). 31 90 Informal 90 Formal N = Credit limit (MK) 2,000 1,600 1,200 800 400 0 5930 5930N = Credit limit (MK) 2,000 1,600 1,200 800 400 0 Formal Informal FemaleMale Figure 3—Distributions of formal and informal credit limits and unused credit lines when a formal loan was granted, October 1993–December 1995 A. Formal and informal credit limits when a formal loan was granted B. Formal and informal unused credit lines when a formal loan was granted 90 Informal 90 Formal N = Unused credit line (MK