Janvry de and Sadoulet
Chapter 12 Financial services for the poor
Revised October 20, 2010 Take home messages for chapter 12 1. Because of asymmetrical information in time-delayed transactions, capital markets fail the poor who do not have collateral to pledge. The microfinance revolution is an effort at creating new institutions that can solve this problem without reliance on collateral. 2. The lender problem that needs to be solved has four components: avoiding adverse selection of potential borrowers, effective monitoring to reduce moral hazard in project implementation, reducing moral hazard in providing limited liability insurance, and preventing moral hazard in enforcing the repayment of loans. 3. Institutional alternatives to help the poor access lump sums of cash when needed while overcoming (at least partially) the lender problem include: moneylenders, ROSCAS, group lending, village banks, interlinked credit in value chain contracts, and individual proximity lending often with the use of credit bureau information. 4. Group lending has been the most successful MFI innovation, relying on self-selection by members, joint liability, mutual insurance, dynamic incentives, and collective sanctions in overcoming the lender problem. 5. Group lending is increasingly giving way to individual lending with entry of for-profit lenders, rising competition, and mission drift away from poorer borrowers. Dynamic incentives and informed and motivated credit officers are key for success of individual lending without collateral. 6. Effective financial services for the poor require bundling access to credit with savings, insurance, and low cost transfers. These institutional development are still in progress, with many notable experiments in particular in using mobile phone banking and index-based insurance.
I. The microfinance revolution Microfinance institutions (MFI) have been created to overcome a market failure on capital markets that is particularly detrimental to the poor: because of asymmetrical information that induces adverse selection and moral hazards in borrowing, commercial lenders require borrowers to provide collateral as a guarantee that the loan will be repaid. As a consequence, capital markets fail as they are wealth constrained by the collateral requirement. For the poor, the implication is stark: even a good entrepreneur with a brilliant business idea will not qualify for a loan for lack of ownership of collateralizable assets (i.e., of assets that can serve as collateral with the lender). MFIs consist in a broad range of institutional innovations that seek ways around this problem. Typically, they will offer mechanisms by which lending can happen with minimal risk to the lender in spite of lack of formal collateral to secure the transaction. This is the microfinance revolution, a set of major institutional breakthroughs, pioneered by the Grameen Bank in Bangladesh created by Professor Yunus who received the Nobel Peace Prize in 2006, that help the poor gain access to capital. Many other institutions have followed suit such as Banco Sol in Bolivia, BRAC based in Bangladesh with offices in 14 countries, and Compartamos in Mexico (Table 1). While MFIs in developing countries originated principally with the Grameen Bank extending small loans to poor women, now numbering over seven million, MFIs increasingly go beyond credit to provide a broader range of financial services to the poor including savings, insurance, and the transfer of funds (Sengupta and Aubuchon, 2008). These services are complementary. Insurance is for instance important to help poor
1 10/19/10
de Janvry and Sadoulet
people assume more risks in developing businesses and take more loans. Lump sum expenditures can be met through both savings and borrowing, and any combination of the two. Yet, these complementary financial services are rarely offered as a package, especially the insurance component. Interesting to note in Table 1 is that providers of MFI services can be non-profit or commercial banks, that they can have high levels of performance (very low shares of their lending portfolio at risk), and be potentially quite profitable, making business at the bottom of the pyramid potentially quite attractive (Prahalad, 2006).
Grameen Bank Bangladesh Year established 1983 Membership 6,950,000 Average loan balance (US$) 69 Percent female 97 Legal status Non-profit Services offered Loans Group lending contracts? Yes Collateral required? No Portfolio at risk > 30 days (%) 1.9 Return on equity (%) 2.0 Source: Based on Sengupta and Aubuchon, 2008 Banco Sol Bolivia 1992 104,000 1571 46 Commercial bank Savings and loans Yes No 2.9 22.8 Compartamos Mexico 1990 617,000 440 98 Commercial bank Savings and loans Yes No 1.1 57.4
Table 1. Characteristics of selected Microfinance Institutions
Lump sums of money are needed to (1) meet lifecycle needs such as dowries, school fees, and burial costs, (2) face emergencies such as doctor fees, (3) acquire indivisible assets such as livestock, land, housing, and durable goods, and (4) and invest in business activities such as capital goods and inventories (Rutherford, 2000). MFIs thus have the role of helping poor people transform a flow of earnings into these discontinuous lump sums of money. In this chapter we first analyze the traditional lending problem and explain why poor people are excluded from formal financial institutions. We then look at the traditional institutions for access to loans and savings: money lenders and ROSCAS (Rotating Saving and Credit Associations). We then analyze microfinance institutions practicing group lending, mobilizing savings in village banks, and engaging in proximity lending for individual loans. We finish with recent institutional innovations to improve financial services for the poor such as mobile phone banking, credit bureaus, microinsurance, and internet-based philanthropic lenders. Financial transactions are particularly complex because they involve a delay between the two sides of the transaction: a loan is given now, but will only be repaid later; a saving is deposited now and will be withdrawn later; an insurance contract is paid for now and will become effective later when disaster strikes. Much of the difficulty in these deferred-payment transactions comes from asymmetric information among the parties involved, inviting opportunistic behavior called adverse selection and moral hazard. Institutional innovations such as MFIs have the purpose of defeating these sources of market failure (Williamson, 1985). Adverse selection and moral hazard are defined as follows (Perloff, 2008):
2
10/19/10
de Janvry and Sadoulet
Adverse selection (AS) corresponds to hidden information about the characteristics of a person or a product that gives room for opportunistic behavior. In lending, there is an AS problem when the lender cannot fully know the quality of a potential borrower, and thus cannot properly select only good borrowers. In insuring, more risky people will tend to seek insurance, raising the cost of payments for the insurance company that cannot recognize them ex-ante. This creates a market failure by reducing the size of the market for lending and insurance transactions, eventually completely eliminating the market as explained by Akerlof (1990) in his theory of the market for lemons. Moral hazard (MH) corresponds to asymmetrical information allowing opportunism under the form of hidden actions. In lending, there is MH when the lender does not know what the borrower will do (Will borrowed funds be used productively? Will the loan be repaid when the borrower has the capacity to do so?) due to the fact that this cannot be observed or cannot be punished. In insuring, the insured may behave more recklessly knowing that he is now insured, increasing he likelihood of accidents. Here again, this creates market failures by reducing transactions and eventually preventing any market activity. II. The formal lender problem: Banks Why are the poor excluded from formal financial institutions such as commercial banks and public development banks? Why do financial markets fail for them? There are four problems that a formal lender needs to solve in order to successfully obtain repayment of a loan made to a borrower (Figure 1):
Figure 1. The lender problem
1. Selection problem: AS of borrower. The lender has difficulty in screening ex-ante good from bad borrowers due to incomplete information. Hence, he cannot fully know if the borrower will be able and willing to repay the loan. 2. Monitoring problem: MH in project implementation. The lender cannot closely monitor the borrowers behavior. Hence, he cannot be certain that the borrower will be making good use of the loan so that she will be able to repay. 3. Insurance problem: MH in limited liability. The borrower needs insurance (i.e., limited liability in repaying the loan), otherwise even loans for good projects will not be repaid in bad years. But the lender cannot easily provide insurance as he cannot distinguish genuine failures from false claims due to imperfect information.
3
10/19/10
de Janvry and Sadoulet
4. Enforcement problem: MH in loan repayment. The lender cannot easily force the borrower to repay. He does not know if the borrower will be willing to repay, even if he is able to do so having made successful use of the loan. The solution for a formal lender is to require collateral from the borrower in order to overcome the problems of selection, monitoring, and enforcement. Hence, access to credit is restricted to those with collateral. There is a credit market failure as credit is provided on a wealth-constrained market. This has: i) An efficiency cost: The allocation of credit is unrelated to the marginal productivity of capital. Many good projects are not funded simply because the entrepreneur is poor and unable to pledge collateral. ii) An equity cost: The poor with no collateralizable assets are excluded, creating sharp inequities across potential borrowers (World Bank, 2005). Allocation of credit to the wealthy thus contributes to reinforcing and reproducing inequality. iii) An insurance problem that remains largely unresolved: Poor entrepreneurs with collateral may not take loans as they cannot afford to put their collateral (land, house) at risk (Boucher, Carter, and Guirkinger, 2008). Hence, asset ownership by the poor, as championed by Hernando de Soto (2000), is necessary but not sufficient to access loans. Limited liability insurance provided by the lender is needed to shelter the collateral from unexpected shocks, or the borrower needs to be able to self-insure, or he needs access to other sources of insurance in order to take a loan. Financial services for the poor thus need to provide not only access to loans, but also to insurance. But because there is MH in insurance claims, the lender may not be able to offer insurance to the borrower. The formal lenders collateral solution to the lender problem thus tends to exclude most of the poor from access to bank loans. This is a serious problem in development if we believe that entrepreneurship can be an important pathway out of poverty. The microfinance revolution consists in a set of institutional innovations to give access to credit to the poor in spite of not being able to pledge collateral, and this under terms and conditions that are more favorable than offered by local money lenders. Each MFI has particular advantages and shortcomings in addressing the selection, monitoring, insurance, and enforcement problems. They are summarized in Table 2. We use this organizing principle to analyze the role of each type of financial institution. III. Local moneylenders or usurers Because there are profits to be made, local moneylenders are present in all contexts where people need to make lump-sum payments but are left without access to formal lenders. They are present in remote agrarian communities waiting to be repaid when the harvest comes, and they are waiting on the beach when fishermen return with their catch. They are able to lend money to the poor because they have found a solution to the four problems faced by any lender. How do they do this? The comparative advantage of money lenders is that they live in the community where they have access to information about potential borrowers. They know who is who, who does what, what reputation each person has regarding hard work and honesty,
4
10/19/10
de Janvry and Sadoulet
and who is connected with whom. This allows them to defeat the problems of AS and MH in selection, monitoring, insurance (partially), and enforcement. They can give insurance to their clients, but only for idiosyncratic shocks: if everyone in the community is affected by a drought or a plague (i.e., by a covariate shock), he has difficulty providing insurance, except on a limited basis based on his accumulated reserves. A broader insurance scheme going beyond the community would be needed for this. Since knowledge of his clientele is limited to the community where he lives, this is a service he cannot offer. The insurance problem is thus incompletely resolved. This is the so-called dilemma of the agrarian community: local information is quite perfect (the universal pastime of gossiping takes care of this), but local information is not sufficient for some transactions such as insuring covariate risks. Beyond local information, money lenders also have access to local social capital in that they can ostracize someone who does not repay, diminishing the reputation a person has, with costs for him or her in terms of future transactions with other community members such as landlords, employers, and merchants. They can also take some forms of collateral that a bank could not take such as animals, small durable goods (as pawnbrokers do), and use of the clients land for a season (land pawning which is frequent in the Philippines). They can also play repeated games with their clientele that wants to keep access to future loans if there is limited competition among money lenders. So, local money lending seems to work pretty well in meeting the financial needs of the poor. What are the drawbacks? Why do money lenders have such poor reputation and are typically called usurers? The main problem is the high cost of credit. Interest rates are typically 1% per day in the Dominican Republic and 400 to 500% per year in South Africa. This may be due to monopoly power, but interest rates are high even when there is competition among moneylenders. This is for several reasons. One is that money lenders need to keep high liquidity positions to be able to immediately give emergency loans to their clients. This is what they are here for. A child is sick in the middle of the night and a taxi is needed to take him to the hospital. A quick visit to wake up the moneylender enables parents to have the cash needed for the trip. The dilemma of the agrarian community also plays a role: if all businesses go bad at the same time, high reserves need to be kept to weather out the period where all clients need postpone repaying their loans. This very high cost of loans clearly limits their use by households to lump-sum payments such as quickly coping with shocks, short run liquidity needs to make payments (for example for food at the local store or to an MFI to make interest payments on a loan), high return operations such as buying and selling animals and goods, and very small amounts to make transactions. Hence, money lenders provide extremely useful services to the poor, which is the reason why they exist, but their services are extraordinarily expensive, calling on alternative institutional innovations. IV. Informal microfinance institutions: ROSCAS Can poor people with no access to formal loans and in need of lump sums of cash to meet particular expenditures do better than using loans from moneylenders? If these expenditures are not for risk coping, one option is the Rotating Saving and Credit Associations (ROSCAS). These are spontaneous associations that generally exist without
5
10/19/10
de Janvry and Sadoulet
any formal legal constitution. They serve both to promote savings and to gain access to lump sums of cash. They are part of the extraordinary institutional innovativeness that one encounters at the grassroots level. We find them from Haiti (where they are called tontines), to African villages, and among undocumented immigrants in East Oakland. These associations are able to solve the problems of selection and enforcement. Because they perform irrespective of the use members make of the cash lump sums, they are not concerned with monitoring of project implementation and of limited liability. They also fulfill a very important disciplinary function in helping individuals acquire the discipline of regularly saving small amounts of money from their incomes. As such, they provide a disciplinary device that helps defeat tendencies to procrastinate in saving money. Here is how they work. A group of people who trust each others constitute a ROSCA. They are typically people who live in the same neighborhood, migrated from the same village, or work together for instance in a meat packing plant in East Oakland. The association has N members who meet at regular intervals, typically monthly. At each meeting they all make an equal deposit d. One member takes home the lump sum Nd contributed at one meeting. Different ROSCAS have different rules of attribution: there can be a preassigned turn often decided by an initial random draw, or there can be bidding on turns with the highest bidder coming first. Note that this is not meant to be effective for emergencies: you only get the lump sum when the group meets and when it is your turn. It is however effective for lumpy expenditures such as a business investment, the purchase of a durable good, or a lifecycle expenditure known ahead of time such as a dowry. Consider the following example where 10 members meet monthly and each puts $10 on the table. The first winner gets $100 immediately, interest free. He will repay to the others at zero interest through his contributions at the nine subsequent meetings. The 10th winner gets $100 having made payments for nine months. Hence, he could have saved alone just the same. And he has lost earning interest on his money during the nine months where he made his $100 contributions. So why did he do it? Advantages of membership to a ROSCA are the following: If the ROSCA is a repeated game, being last, which is a random draw, is only for the first round. Subsequent rounds come with the same regularity for all. Loss of interest for those with higher turns is zero if money could not be deposited in interest bearing accounts or used for business. The ROSCA is largely able to defeat the selection and enforcement problems as members self-select and know each others well. One of the most important functions of ROSCA membership is that it provides a disciplinary device, i.e., a nudge to save via the routine of contributions at regularly scheduled meetings (Thaler and Sunstein, 2008). It helps defeat the well known human tendency of procrastinating with such activities as savings that can forever be postponed to tomorrow, and thus never happen.
6
10/19/10
de Janvry and Sadoulet
ROSCA membership often has other benefits such as information sharing among group members, organizing other business deals, allowing women to be away from home, and organizing social functions. Disadvantages include the following: Rigidity in the timing and amount of access to a lump sum of money. This rigidity makes the lump sum rarely effective for insurance. Risk of moral hazard since a member who had an early turn can decide to stop coming to the subsequent meetings, effectively defaulting on others. This is particularly tempting if the amounts of money contributed at each meeting are very large. Thus, ROSCAS are not infallible.
V. Microfinance with group lending The heart of the MFI revolution consists in replacing the use of assets as collateral, that the poor do not have, by social collateral, that the poor can provide (Armandariz de Aghion and Morduch, 2005). This is the idea of using solidarity groups where members are jointly liable for the loans taken by individual members. The approach was pioneered by the Grameen Bank. Today, it is estimated that the MFI movement has some 2,500 institutions that serve some 70 million clients in more than 100 countries, with outstanding loans reaching some $25 billion. Information on the MFI movement is provided by the Microcredit Summit website at http://www.microcreditsummit.org/enews/. MFIs were initially mainly non-profit organizations, motivated by giving access to capital to the poor. Because rates of repayment tend to be high, interest rates are high, and the poor are hundreds of millions of potential clients, MFIs are increasingly commercial banks that seek fortune at the bottom of the pyramid, with profit-making as the motive (Prahalad, 2006). Increasing competition among MFIs for good borrowers has created a rapid mission drift away from serving the poorest to finding viable clients among the poor (where poor is here defined as households without assets that can serve as collateral, i.e., people who are asset poor), usually the least poor who may be the best potential entrepreneurs. Group lending has been effective in solving the four lender problems, including at least partially the insurance problem. Here is how it works. The MFI asks potential borrowers to self-select into groups of 5 to 30 members. Members typically choose to associate with others they know well, helping defeat the adverse selection problem which the MFI could not solve directly as it is based on information that members have about each others, not the institution. Loans are individual, but all group members are jointly liable for the repayment of all loans taken by group members. This means that each member is responsible for repaying the loans of those who may default. And the whole group loses access to future loans if all loans are not repaid. In this fashion, the group serves as social collateral, acting as a substitute
7
10/19/10
de Janvry and Sadoulet
for the collateralizable assets that poor members do not own (Ghatak and Guinnane, 1999). The loans made by the MFI are small and inferior to the amounts each member would like to borrow. Loans are gradually increased to reward good repayment behavior. This creates dynamic incentives that induce borrowers to repay their current loans in hopes of getting larger loans in the future. In so doing, borrowers accumulate reputation that could be made public through a credit bureau. Another feature of group lending by MFIs it that frequent installments (partial repayments of the loan) are required, typically on a weekly or by-weekly basis. While this as a huge transactions cost for the borrower, it may in fact be desired by both the MFI and the client. On the side of the MFI, advantages of frequent installments are that it allows officers to better monitor their clients and that it helps them screen borrowers who have a steady income flow in addition to the expected return from the investment project funded by the loan. On the side of the client, frequent installments provide a disciplinary device that avoids procrastination (hyperbolic discounting whereby you will start being concerned with the future, but starting tomorrow) in saving, and shelters the money earned from pressures from kin and neighbors to share the available cash, which can be irresistible when there are strong solidarity norms. In that sense, taking an expensive loan to cover an expenditure such as house repairs when the money could have been saved appears irrational. Rationality for this decision is that frequent installments serve as a commitment device for saving more than the borrower could have done by herself, at a very high cost. In that sense, the client effectively borrows to save! How does group lending help solve the lender problems? Self-selection by group members helps solve the AS problem. Because members are jointly liable, they have a strong incentive to select other members with a good expected performance, and to expel non-performing members. They also have a strong incentive to monitor and help each others projects using local information and direct assistance, in what is referred to as peer monitoring. For this, more homogeneous groups in terms of what people do (they may for instance all be in the restaurant business) are more effective for monitoring. The group can enforce payment on its members based on social capital (ostracism in the community), inter-linkages among members (based on interlinked transactions that will be explained later), seizure of collateral (e.g., personal belongings), and threats of being expelled from the group. Analyzing group lending by FINCA in Peru, Karlan (2007) finds that monitoring and enforcement activities by group members do improve group lending outcomes, and that social connections (i.e., groups that are more similar culturally and more geographically concentrated) facilitate the monitoring and enforcement of group members on each others. Finally, the group can mutually insure against idiosyncratic risks (but not against global shocks): successful members can advance repayment of the loan of a member with a true failure due, for example, to illness. For insurance purposes, large heterogeneous groups are more effective as they diversity risks (better to have occupations heterogeneous that do not suffer from covariate shocks). How large and homogenous a group should be thus responds to a trade-off: small and homogenous is better for monitoring to avoid free-riding (MH); large and
8
10/19/10
de Janvry and Sadoulet
heterogeneous is better for mutual insurance. The Grameen Bank requires groups to not have less than 5 members. Members usually prefer smaller over larger groups, suggesting that the MH problem is of more concern to them than the insurance benefit. Finally, the MFI can enforce repayment by using dynamic incentives (gradually larger loans based on a good repayment performance) and collective sanctions (the whole group loses access to future loans if one loan is not repaid). So what are some of the potential disadvantages of the group lending approach? One is that credit remains expensive. The MFI will typically charge a 7-10% interest for management over the 15-20% interest rate charged by the bank from which the capital is coming, thus reaching interest rates in the range of 22 to 30%. Those are of course much lower than the rates charged by moneylenders (400-500%). Yet, investment opportunities should be quite good to perform with loans that expensive. For that reason, MFI loans have been more successful in commerce and in rapid turnover micro-enterprises than in agriculture with longer cycles, lower profitability, and higher risks. Another disadvantage for the lender is that the whole group will default if those who would not have defaulted individually cannot repay for defaulters, even though they could have paid for themselves. However, experience shows that repayment rates are high when MFIs have good credit officers working closely with their clients. Data in Table 1 show that repayment rates within 30 days of due date are 98.1% for the Grameen Bank, 97.1% for Banco Sol, and 98.9% for Compartamos. Group lending can also be interpreted as a learning stage toward obtaining individual loans. As members need larger loans, they can become increasingly different from each others, making joint liability and mutual insurance difficult. The groups reputation of good repayment performance can be registered in a credit bureau, and allow members to qualify for more flexible and larger individual loans. In that sense, group lending can be seen as a step in a credit ladder leading to individual banking. Some other issues in group lending worth mentioning are the following: (1) NGOs engaged in lending activities sometimes have difficulty in enforcing repayment as they are seen by clients as soft-hearted charitable institutions. Lessons are that it is better to clearly separate charitable activities from lending operations. NGOs engaged in welfare activities should set up separate branches for their microfinance operations. (2) Kiva (http://www.kiva.org), Prosper (http://www.prosper.com), and other internet micro-lending institutions allow face-to-face relations between dispersed lenders and dispersed borrowers, under the guarantee of the managing MFI. This approach to MFI lending has seen an extraordinarily rapid growth with successful repayment rates. (3) Group lending is sustainable and replicable only if linked to financial markets to access capital, as opposed to being dependent on donations for loanable funds. Donations can be used to cover the start-up costs and learning phase of new MFIs until they become competitive, as done by Accin International. Unitus (http://www.unitus.com/) channels internet donations specifically to assist MFI startups. (4) Rising competition among MFIs reduces the power of dynamic incentives to repay loans as alternative sources of loans become available to individual borrowers (McIntosh
9
10/19/10
de Janvry and Sadoulet
and Wydick, 2005). Competition can thus increase the rate of default. This is when introduction of a credit bureau that provides competing lenders with information on the past repayment performance of potential clients and their total levels of indebtedness becomes truly important to lenders. Introduction of a credit bureau in Guatemala allowing MFI lenders to share positive information (i.e., information about both past defaults and levels of indebtedness) about their clients has been shown to help reduce adverse selection and moral hazard behavior, significantly improving the repayment performance for participating institutions (de Janvry, McIntosh, and Sadoulet, 2010). (5) International donors supporting non-profit MFIs for their role in helping the poor gain access to capital are increasingly concerned with mission drift away from poorer potential clients. This is especially the case as commercial banks that go into microfinance (lending without collateral) take the better clients away from the non-profit pro-poor lenders. The issue of achieving profitability (which is necessary to have a sustainable operation with a high rate of loan repayment) while at the same time serving the relatively poorer potential borrowers remains an area of active research and experimentation by academics and the MFIs themselves. To conclude, group lending is the core innovation of microfinance lending. The key features of group lending are: Self-selection by group members Individual loans, with joint liability providing social collateral Dynamic incentives Frequent installments and self-control Mutual insurance Peer monitoring Social capital and inter-linkages among members Collective sanctions VI. Village banks A major lesson derived from MFI operations is that the poor need access to financial services that go beyond loans to include savings, insurance, and financial transfers (Collins, Morduch, Rutherford, and Ruthven, 2009). Poor people need in particular safe and profitable savings instruments for lump-sum payments, self-insurance (risk coping), and life-cycle expenditures (dowries, retirement). Institutional innovations for this are village banks promoted in particular by FINCA (a Washington-based organization), credit cooperatives, local savings-and-loans associations, credit unions, and self-help groups in India also called Accumulating Saving and Credit Associations (ASCA). While the main challenge of lending to poor people was to find a substitute to formal collateral, the main challenge in offering savings services to the poor is guaranteeing security of deposits in an institutional context that is weakly regulated at best and highly decentralized. The worst savings service that could be offered to poor people (as a substitute for animals, grain stocks, and jewelry as savings instruments) is savings accounts from which money will be stolen, an altogether not uncommon event.
10
10/19/10
de Janvry and Sadoulet
A FINCA-type village bank typically has some 200-400 members and formal bylaws (Figure 2). It mobilizes savings from members (internal account) and also receives loans and grants (external account). This equity can in turn be lent to members under the form of individual or group loans. Loans are relatively safe as they benefit from proximity lending: they are made based on the basis of extensive personal information among members. The village bank can thus manage the problems of AS and MH based on local information. It can insure borrowers against idiosyncratic shocks based on local information and social capital, offering limited liability on loans (the option to reschedule loans) if there is genuine failure to repay.
Figure 2. Village bank
Village banking is, however, not free from difficulties. The approach assumes that members can have sufficient management capacity to self-manage the village bank. This is a big assumption. The risk of loss of savings through mismanagement and theft is a stark reality. For that reason, it is important that local MFI that engage into savings mobilization be part of a legal regulatory framework. But the risk is that regulation be too rigid and demanding (e.g., in Mexico), making the mobilization of savings by village banks difficult. As the group gets larger, which is necessary to cover fixed management costs, AS and MH may occur as shared information about members becomes more tenuous. Yet, village banks are very important in providing access to more complete financial services to millions of poor people, often with the strong support of formal banks as in India. VII. MFIs with individual loans: Proximity lending MFIs have also developed methodologies that allow them to make individual loans, still without collateral. These loans are generally given by profit-oriented MFIs which are in the business of microfinance to capture a market niche which commercial banks are unable to penetrate. They are opening a huge market for small loans to the micro-enterprise sector, which can be very profitable as long as interest rates are high, repayment rates are also high, and there is still little competition. How do they obtain high repayment rates without collateral? Three instruments are used for this. One is dynamic incentives: make small loans, inferior to the size demanded by clients, with a steep increase in loan size over time to reward good performance, and loss of access to credit otherwise. This works better if the MFI has more monopoly power. The second is to require co-signers on the loan who serve as personal collateral. The problem with this is that if there is default on the loan, it will typically be difficult to enforce payment on the co-signer. And third, is proximity lending as the MFI makes intensive use of well informed credit agents for the
11 10/19/10
de Janvry and Sadoulet
selection and monitoring of clients. These agents are typically themselves from the community which gives them access to information about potential borrowers, somewhat similar to moneylenders (Fuentes, 1996). Agents are given incentive contracts whereby they benefit from high repayment rates through bonuses. With motivated and skilled credit agents, success (for example as achieved by Bank Rakyat Indonesia) shows that individual lending without collateral can also work with clients that have acquired the discipline of making payments, sometimes acquired through a previous phase of group lending. Credit bureaus are important for information sharing about clients among lenders when competition rises and dynamic incentives are weakened. VIII. Local sources of credit based on inter-linkages in value chains When there are market failures, inter-linkages between two transactions often allow each transaction to perform better than it would alone. This is the theory of interlinked transactions (Bardhan, 2003). The two transactions have greater value together than the sum of the value of each transaction separately because each transaction improves the value of the other. Inter-linkages can help solve the problems of AS, MH, and insurance in credit transactions. Consider the following examples: A local merchant gives credit to a farmer and will buy his products at harvest time under favorable conditions. A local landlord gives credit to a worker and also regularly rents land to him as a tenant when there is a lot of competition among potential tenants to get access to land. A local moneylender gives credit and also provides insurance to his client when hit by a shock with no other insurance option for the client. In all cases, the inter-linkage of credit with a second transaction is used to put pressure on the recipient of the loan to repay: the borrower would be cut-off from the benefits of the other transaction if he does not repay the loan, creating an incentive to repay. In a sense, the other transaction serves as collateral for the credit transaction. This form of lending happens increasingly within value chains. A value chain is the sequence of transactions that run from accessing inputs, to the production, processing, marketing, and consumption of particular commodities. Transactions in these value chains are commonly linked. A provider of inputs will provide credit to a farmer and get paid through a share of the harvest. A supermarket will finance the production of its providers, and impose strict quality and timing requirements on product delivery. Value chains increasingly serve as instruments where financial transactions occur along with transactions of inputs and products. IX. Can the poor be insured? The promise of index-based weather insurance An important dimension of financial services is the provision of insurance. Lack of access to insurance reduces the ability of the poor (who are particularly risk averse as exposure to an uninsured shock can have devastating consequences on survival) to take on risks that would be rewarded by higher expected incomes. As we saw before in the
12
10/19/10
de Janvry and Sadoulet
course, exposure to risk requires them to engage in costly (1) risk management strategies (e.g., non-adoption of more productive but more risky technologies, diversifying production away from comparative advantage, accumulating liquid assets such as animals as precautionary savings even though they have a lower return than fixed investments such as pumps for irrigation (see Rosenzweig and Wolpin, 1993)); and (2) risk coping strategies (e.g., taking children out of school to make them work while knowing that once away from school they are unlikely to ever return to school, consuming next years seeds that will compromise ability to plant in the next season, selling productive assets such as land that will be hard to re-accumulate after the shock is gone, and compromising long term child development with short run reductions in nutrition expenditures). Exposure to risk also reduces the ability to take loans as projects may fail and collateral that may have been pledged may be lost (Boucher, Carter, and Guirkinger, 2008). Lack of insurance thus acts as a major constraint on the development of financial markets. Yet, traditional forms of insurance (that indemnify the insured based on verifiable individual losses) have largely failed because of the well known problems of adverse selection and moral hazard. This insurance coverage only works for undoubtedly verifiable events such as loss of life or pregnancy, but it has failed to insure business activities and especially farming. The result is that the poor are largely unable to reduce the risks to which they are exposed, perpetuating poverty as they must manage risk on their own, and creating a source of new poor as they are exposed to shocks. An interesting institutional innovation that is being tried in several countries such as Mongolia, India (Basix), and Mexico (Agroasemex), is index-based weather insurance (Skees, 2009). In this case, indemnity payments for weather shocks are not based on an assessment of individual losses, but on a single, easy-to-observe indicator such as local rainfall or average crop yields across farmers in a particular region. This insurance is thus only partial: it protects individuals against covariate risks (average weather in the area where the weather station is located) and not against idiosyncratic risks (plot-specific weather events). Basis risk (the remaining uninsured residual risk) can thus be high and needs to be covered through other mechanisms such as mutual insurance (which is precisely what local households can insure for as these risks are not covariate) or social safety nets (such as the National Rural Employment Guarantee Scheme in India). Under index-based insurance schemes, insurance contracts are issued to farmers for a fee and indemnity payments are determined by the level of the single selected index (rainfall as done in India, some index related to predicted weather-induced livestock mortality as done in Mongolia). The approach thus overcomes the classical insurance problems of AS and MH, as well as of costly assessment of individual losses. Bundling access to credit with index-based weather insurance is a promising way to reduce some of the covariate risks, and thus allow farmers to make greater use of credit and engage in more remunerative (and more risky) activities. Interesting pilot programs are in place for livestock in Mongolia, crops in India and Mexico, and groundnuts in Malawi (United Nations, 2007).
13
10/19/10
de Janvry and Sadoulet
Major issues still to be addressed with index-based weather insurance schemes are the following: (1) Help farmers understand the concept of insurance. It is a priori not trivial for a farmer to understand that, if a shock did not happen in a particular year, the insurance company should not simply return the money. For that reason, most insurance programs will start with heavy subsidies until the understanding of insurance principles has been acquired. (2) There is a well known tendency in the psychology of behavior to under-insure (cockeyed optimism). And there is a tendency to engage in narrow bracketing whereby some risks are highly insured while other risks are not, instead of insuring the whole portfolio of sources of income (Rabin, 1998). (3) Reduce basis risk by having more geographically fine-grained observations on climatic events. One promising new option is to use satellite-based observations of weather or biomass to estimate rainfall at the pixel level. (4) Build trust in the insurance agency that it will indeed make payments when a disaster covered by the insurance contract occurs. This requires giving clients formal or informal recourse in case the insurer does not deliver on promise (think of hurricane Katrina in 2005 in New Orleans with a large share of insurance claims still not settled). (5) Provide the opportunity for local insurance agents (who have the benefit of access to local information and local social capital) to re-insure at a broader scale in order to diversify their portfolios beyond local covariate risks. One such company that allows insurers to re-insure is Swiss Re headquartered in Zurich, Switzerland. X. Conclusions on MFI: How useful are they for poverty reduction? MFIs are institutional innovations that help provide access to financial services for the poor. While initial emphasis was principally on access to loans (following the lead of the Grameen Bank), services have been gradually broadened to include savings, insurance, and money transfers. The much heralded microfinance revolution helped reduce credit market failures for millions of poor people, principally women who constitute 97% of the Grameen Banks clients and 98% of those of Compartamos (Table 1). Can we say that this has helped reduce poverty? Surprisingly very few solid evaluations of this desirable impact are yet available. And solid evaluations (that avoid the selection bias in seeking to obtain a loan that biases benefits upwards as it attributes superior entrepreneurship to the loan) tend to show more modest results than claimed by the advocates of microfinance (Banerjee, Duflo, Glennerster, and Kinnan, 2010). In spite of this, preliminary conclusions are the following: (1) Successful use of loans by the poor requires them to have access to good projects. This necessitates a favorable investment climate: a reliable policy context, good public goods and institutions for investment, and help from NGOs to prepare projects. Hence, the development of MFIs is complementary to these determinants of success, not a substitute. MFIs without profitable investment opportunities accessible to the poor will not reduce poverty. (2) Loans will tend to be most useful for the most entrepreneurial among the poor, not the poorest of the poor. In that sense, the MFI approach is also a complement to social assistance targeting the poorest, not a substitute.
14
10/19/10
de Janvry and Sadoulet
(3) Yet, there have been remarkable success stories in taking people out of poverty through microfinance, particularly in creating income opportunities for women. (4) MFIs are thus not the magic bullet for poverty reduction, but they offer important partial solutions that deserve improvement and expansion. (5) Further institutional innovations are needed to expand financial services both in content (savings, insurance, transfers) and in coverage (avoiding mission drift away from the poorer people as commercial providers enter the field and competition rises). In that sense, it is an on-going revolution with much creative thinking and entrepreneurship yet to happen. It is indeed an exciting and promising field for development economists to work on. References Akerlof, George. 1990. The Market for Lemons: Quality Uncertainty and Market Mechanisms. The Quarterly Journal of Economics 84 (August): 488-500. Armandariz de Aghion, Beatriz, and Jonathan Morduch. 2005. The Economics of Microfinance. Cambridge: The MIT Press. Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and Cynthia Kinnan. 2010. The miracle of microfinance? Evidence from a randomized evaluation. MIT: Department of Economics. Bardhan, Pranab. 2003. Poverty, Agrarian Structure, and Political Economy in India : Selected Essays. New Delhi: Oxford University Press. Boucher, Stephen, Michael Carter, Catherine Guirkinger. 2008. Risk Rationing and Wealth Effects in Credit Markets: Theory and Implications for Agricultural Development. American Journal of Agricultural Economics 90(2): 409-423. Collins, Daryl, Jonathan Morduch, Stuart Rutherford, and Orlanda Ruthven. 2009. Portfolios of the Poor: How the Worlds Poor Live on $2 a Day. Princeton University Press. de Janvry, Alain, Craig McIntosh, and Elisabeth Sadoulet. 2010. The Supply- and Demand-Side Impacts of Credit Market Information. Journal of Development Economics 93(2): 173-188. de Soto, Hernando. 2000. The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else. Basic Books Fuentes, Gabriel. 1996. The Use of Village Agents in Rural Credit Delivery. Journal of Development Studies 33: 188-209. Ghatak, Maitreesh, and Timothy Guinnane. 1999. The Economics of Lending with Joint Liability: A Review of Theory and Practice. Journal of Development Economics 60: 195-228. Karlan, Dean. 2007. Social Connections and Group Banking. The Economic Journal 117(February): F52F84 McIntosh, Craig, and Bruce Wydick. 2005. Competition and Microfinance. Journal of Development Economics78: 271-98. Perloff, Jeffrey. 2008. Microeconomics. Addison-Wesley. Prahalad, C.K. 2006. The Fortune at the Bottom of the Pyramid: Eradicating Poverty Through Profits. Wharton School Publishing. Rabin, Matthew. 1998. Psychology and Economics. Journal of Economic Literature
15
10/19/10
de Janvry and Sadoulet
36(March): 11-46. Rosenzweig, Mark, and Kenneth Wolpin. 1993. Credit market constraints, consumption smoothing, and the accumulation of durable production assets in low-income countries: Investment in bullocks in India. Journal of Political Economy 101(2): 223244. Rutherford, Stuart. 2000. The Poor and Their Money. Delhi: Oxford University Press. Sengupta, Rajdeep, and Craig Aubuchon. 2008. The Microfinance Revolution: An Overview. Federal Reserve Bank of St. Louis Review (January-February): 9-30. Skees, Jerry. 2009. Innovations in Index Insurance for the Poor in Lower Income Countries. Agricultural and Resource Economics Review. Forthcoming. Thaler, Richard, and Cass Sunstein. 2008. Nudge: Improving Decisions About Health, Wealth, and Happiness. Yale University Press. United Nations. 2007. Developing Index-Based Insurance for Agriculture in Developing Countries. Sustainable Development Innovation Briefs, Department of Economic and Social Affairs. Williamson, Oliver. 1985. The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. New York: The Free Press. World Bank. 2005. Equity and Development. World Development Report 2006. Washington D.C.
16
10/19/10
de Janvry and Sadoulet
Financial institutions 1. Formal lender (Formal collateral) 2. Money lender (Local information & social capital) 3. ROSCAS (Local information & social capital) 4. MFI with group lending (Self-selection, joint liability, dynamic incentives, mutual insurance, collective sanction) 5. Village banks (Savings, proximity lending) 6. MFI with individual loans (Reputation, dynamic incentives, co-signataries) 7. Interlinked transactions (Value chains, interlinkages) 8. Credit bureau (Information sharing among lenders) 9. Index-based insurance (Weather-based covariate risks)
Selection AS in participant selection +
Monitoring MH in project implementation +
Insurance MH limited liability No Some limited liability Not covariate "Dilemma of the agrarian community" No
Enforcement MH in repayment +
Remaining problems Wealth-constrained market No insurance for borrower Very high interest rates
+
+
+ Not useful for insurance Small lump sums of money No interest on money handed out Risk of group default No insurance for covariate shocks Graduation to private loans? Risk of loss of savings (regulation) Weak management capacity Risk of double-dipping Imperfect information for credit officers Specific transactions only
+
+
+
+
+ Better small groups +
Not covariate Better large groups Not covariate
+
+
+
Partial
Partial
Partial
Partial
+
+
+
+ Accuracy of information No information on new clients Only insures weather-based covariate risks Leaves large "basis risk" uninsured Need build trust in provider
+
NA
NA
+
+
+
+ Re-insurance
NA
+ sign indicates that the institution has the ability to solve the corresponding problem. NA indicates not applicable.
Table 2. Financial institutions and their ability to overcome capital market failures for the poor
17
10/19/10