With evidence growing that traditional microfinance may not be offering the expected poverty alleviation benefits, it may be time to consider alternative approaches to microcredit, suggest Professors Pushkar Maitra, Sandip Mitra, Dilip Mookherjee, Alberto Motta and Sujata Visaria. In this article, they describe the results of their experiment with Agent Intermediated Lending, an approach that seeks to make microcredit work for the agricultural poor.
When the 2006 Nobel peace prize was awarded to Mohammad Yunus and the Grameen Bank, the Nobel Committee’s citation said: “Lasting peace cannot be achieved unless large population groups find ways in which to break out of poverty. Microcredit is one such means.”
There are many hopes pinned on microcredit, and not surprisingly, the number of households with access to microcredit has grown rapidly in the past decade. However, evidence is increasingly emerging that microcredit may not be delivering all the outcomes that have been hoped for. In what came as a surprise to many, a recent high-profile study conducted by the Jameel Poverty Action Lab at the Massachusetts Institute of Technology found that when Spandana, a microfinance institution (MFI), began to provide microcredit services in Hyderabad slums, take-up was low. Even among those who did avail of microcredit, the credit infusion did not change household consumption levels or assets (Banerjee et al 2013). Contagious defaults are another serious problem that can undermine the viability of MFIs, as witnessed in the 2008 microfinance crisis in India (Banerjee and Duflo, 2011; Gine et al, 2011).
In the traditional model of microcredit, borrowers self-organise into groups and each member is jointly liable for the loans of all group members. Loan disbursal is sequential, and typically, repayment begins one or two weeks after disbursal, with weekly repayment instalments over the period of a year. If any member defaults, the entire group is cut off from all future lending. From the MFI’s financial viewpoint, microcredit has been very successful; even though borrowers do not post collateral, repayment rates are typically very high — more than 95 percent. It is generally believed that it is the design of the microcredit product that contributes to this high repayment rate: safe borrowers form groups with other safe borrowers (selection), monitor each other’s loan usage (peer monitoring), impose sanctions on those who default, and repay on behalf of defaulting group members (joint liability). There is also some evidence that high frequency repayment schedules develop a culture of repayment among borrowers, and ensure that they repay the loans before they allocate the cash to other competing uses (Field et al, 2013).
Recent evidence, however, suggests that these same features may also contribute to some of the problems with microcredit. The rigid, high-frequency repayment schedules and joint liability can dampen risk-taking (Fischer, 2013). Assuming, for instance, that A and B are members of the same group, A might worry that the riskier is B’s project, the higher the chance that B will default on her loan and cut A off from future loans as a side-effect. But if riskier projects are also the ones with higher returns, this can dampen the effects that microcredit has on income generation and subsequently on poverty. Low tolerance for risk taking and strict repayment rules on the part of MFIs also significantly restrict the set of project choices available to borrowers. Most importantly, this prevents the use of microcredit in financing agricultural working capital, despite the fact that the majority of the poor in developing countries work in the agricultural sector.
Strict repayment rules can also create tremendous pressure on borrowers to repay even in dire circumstances, as the spate of suicides by microcredit borrowers in Andhra Pradesh in 2010 revealed. The joint liability feature can also backfire, as seen in the 2009 Kolar crisis in Karnataka: when a religious organisation issued a directive forbidding its followers to transact with microcredit institutions, it led to contagious defaults across two districts of the state (Gine et al, 2011). Over and above these factors are the high costs of attending group meetings and attaining the savings targets imposed by MFIs. In our conversations with microfinance clients in West Bengal, all of these were reported as factors that limited benefits from microfinance.
Low tolerance for risk taking and strict repayment rules on the part of MFIs also significantly restrict the set of project choices available to borrowers. Most importantly, this prevents the use of microcredit in financing agricultural working capital, despite the fact that the majority of the poor in developing countries work in the agricultural sector.
An Alternative Microcredit Model
Our project is an attempt to design an alternative to traditional microcredit that will solve some of these problems. In our approach, borrowers are liable only for their own loans (individual liability), and therefore, do not face constraints on project choice from group members. But this also means that they cannot free ride on successful group members. There is also no incentive for contagious default. As with most microcredit, there is no collateral requirement so that poor households can also access the loans. In fact, to ensure that the very poor do get the loans, we impose a requirement that all borrowers own less than 1.5 acres of cultivable land. Borrowers also do not have to attend group meetings, monitor other borrowers or be monitored by MFI officials, or meet any savings targets. Most importantly, our scheme was designed to facilitate the financing of agricultural activities; the loan cycles for our product match agricultural cycles, and repayment is only due at the end of four months. Loans are provided after the harvest to allow farmers, if they so choose, to store crops to sell later, when prices are typically higher.
The key question that naturally arises is: how are borrowers selected? As we stated earlier, it is generally accepted that group liability solves a critical information problem: borrowers have better information about each other than an external MFI does, and the joint liability contract incentivises safe borrowers to form groups with other safe borrowers, which contributes to high repayment rates. In our approach, we have removed joint liability, but have replaced it with an alternative mechanism to select safe borrowers. We call our approach Agent Intermediated Lending (AIL). Agriculturalists typically interact with a range of local entities: traders who buy their output, shopkeepers who sell them inputs, moneylenders who give them credit, and so on. These intermediaries have information about a farmer’s ability and willingness to repay, which can be used to the MFI’s advantage. In our approach, these intermediaries are appointed as agents of the MFI and are asked to recommend borrowers for AIL loans. They are incentivised to choose safe borrowers and to ensure that repayment is timely, through a commission based on repayments made by the borrower.
To examine how such a scheme might work in the real world, we teamed up with Shree Shanchari (SS), a Kolkata-based MFI, to conduct a field experiment in two districts of the potato-growing belt of West Bengal. In 24 randomly selected villages, SS appointed local traders and lenders as agents. We call this the trader-agent-intermediated lending (TRAIL) scheme. One TRAIL agent was appointed per village. The agent’s role was limited to recommending a set of borrowers from the village. To a random subset of these recommended households, SS advanced starting loans of INR 2,000 (approximately US$40 at the prevailing exchange rate) for four months at an annual interest rate of 18% (the average interest rate charged by MFIs in these villages was 24% per annum). The repayment due four months later was INR 2,120.
In our approach, we have removed joint liability, but have replaced it with an alternative mechanism to select safe borrowers. We call our approach Agent Intermediated Lending (AIL). Agriculturalists typically interact with a range of local entities: traders who buy their output, shopkeepers who sell them inputs, moneylenders who give them credit, and so on.
Upon full repayment, the borrower became eligible to borrow INR 2,660 (133 percent of 2,000) in the next cycle (covering the next four months), and loan sizes in subsequent cycles became progressively larger. The first loan cycle was in October-November 2010, coinciding with the planting season for potatoes, the major cash crop in this region. To facilitate credit access for post-harvest storage, borrowers were allowed to repay the loan in the form of potato bonds rather than cash. In this case, the repayment was calculated at the prevailing price of potato bonds. While the stated purpose of the loans was agricultural, households were not required to report or prove to us the intended or actual use of the loan.
In another 24 villages, SS introduced its own, traditional group-based lending (GBL) approach, with the modification that these loans were also given for four-month periods. For all loans, repayment was due in a single lump sum at the end of four months, which is the typical length of the potato-growing cycle.
Our team of investigators conducted repeated household surveys of a sample of 50 households in each village. The sample was designed to include households that received loans, households that were recommended or formed GBL groups but were not randomly selected to receive a loan, and households that were not recommended by the trader-agent and did not form GBL groups. Through these surveys, we collected information on household demographics, assets, landholding, cultivation, land use, agricultural input use, sale and storage of agricultural output, credit received and given, incomes, and economic relationships within the village. These data were matched with loan record information from SS.
A Viable Model?
At the end of two years, the take-up and repayment rates of TRAIL loans were significantly higher than that of GBL loans. So in terms of conventional metrics, TRAIL loans have a significantly better performance. The analysis in the paper shows that the superior performance of the TRAIL scheme can be explained by its selection of safer borrowers. There is no evidence of collusion between agents and borrowers, say, in the form of risky borrowers “bribing” agents in return for recommendations. Instead, we find that agents are likely to recommend safer borrowers from within their own clientele, i.e., the group of households to whom they have been lending in the past. The TRAIL scheme was also more cost-effective, with administrative costs at about 10-20 percent of the costs of the GBL scheme. However, the GBL scheme was more successful at targeting landless borrowers. TRAIL borrowers were more likely to own intermediate landholdings (approximately 0.45 acres).
The TRAIL scheme was more cost-effective, had higher repayment rates, and imposed lower costs on borrowers through reduced monitoring, liability and MFI controls. The GBL scheme was better at targeting the poorest households. One implication of our results is that MFIs could consider a multi-pronged approach where GBL and AIL schemes are offered at the same time.
These results suggest that the AIL approach could provide a useful alternative to traditional microcredit. The TRAIL scheme was more cost-effective, had higher repayment rates, and imposed lower costs on borrowers through reduced monitoring, liability and MFI controls. The GBL scheme was better at targeting the poorest households. One implication of our results is that MFIs could consider a multi- pronged approach where GBL and AIL schemes are offered at the same time. It is possible that the poorest households would self-select into GBL contracts, while creditworthy small and marginal landowners would be recommended for AIL loans. Note, however, that in our experiment, the AIL and GBL schemes were offered in different villages. Therefore, we cannot speak to the possible interactions between the two schemes if they were introduced in the same villages simultaneously. Such interactions would have to be analysed before a concrete recommendation could be made.
Positive Impacts on Agriculture
In more recent and preliminary results, we are also finding that the TRAIL scheme had significant impacts on the agricultural activities of borrowers. TRAIL borrowers were more likely than GBL borrowers to expand potato cultivation and produce higher outputs after receiving the loans. The marginal rate of return to these loans could be higher than 100 percent in a year when potato prices were high. A closer look at the borrowers shows that the bulk of these effects are coming from households with 0.5 to 1 acre of land. It is these households who gain the most from the loans. When we decompose the results in GBL villages by household land size, we find the same pattern. In GBL households with 0.5 to 1 acre of land, we see significant positive effects on potato cultivation and output, but in households with smaller land size, we do not; it is just that in GBL villages, it was primarily the poorer households who were selected into the microcredit scheme; therefore, the aggregate impact of the GBL scheme is not significant. These findings, although preliminary, seem to suggest that extending the duration of a loan to four months makes it possible for microcredit to finance agriculture and may help to improve output and incomes for households with the appropriate landholding size. However, the group liability feature means that it is the poorer households who predominantly select into these loans, and they have less to gain. Why this is so is an important question, and one that needs further investigation.
However, the picture that begins to emerge is that microcredit products should be designed differently depending on the goal in mind. For example, whereas GBL loans are better at targeting the extremely poor, they are less likely to have impacts on agricultural output, and therefore, on income. GBL loans serve an important purpose: as our research shows, they target the very poor, and research by other scholars suggests they help the poor smooth consumption (Islam and Maitra, 2012, Morduch, 1998). However, the evidence increasingly suggests that they play a limited role in alleviating poverty. Instead, the modified contractual terms in variants such as the AIL approach could be more effective at targeting those who stand to gain from micro-loans, and allow them to invest the funds in agriculture, earn high returns and possibly escape poverty.
Funding was provided by the Australian Agency for International Development through an Australian Development Research Award, the International Growth Centre at the London School of Economics, the Hong Kong Research Grant Council’s General Research Fund and the United States Agency for International Development.
Banerjee, Abhijit V, Esther Duflo, Rachel Glennerster and Cynthia Kinnan (2013). “The Miracle of Microfinance? Evidence from a Randomised Evaluation”, Working Paper, Massachusetts Institute of Technology.
Banerjee, Abhijit V, and Esther Duflo (2011). Poor Economics: A Radical Rethinking of the Way to Fight Global Poverty (New Delhi: Random House India).
Field, Erica, Rohini Pande, John Papp and Natalia Rigol (2013).
Does the Classic Micro finance Model Discourage Entrepreneurship among the Poor? 103(6): 2196-2226.
Fischer, Greg (2013). “Contract Structure, Risk-sharing and Investment Choice”, Econometrica, 81(3): 883-939.
Gine, Xavier, Karuna Krishnaswamy and Alejandro Ponce (2011).
“Strategic Default in Joint Liability Groups: Evidence from a Natural Experiment in India”, Working Paper.
Islam, Asadul and Pushkar Maitra (2012). “Health Shocks and Consumption Smoothing in Rural Households: Does Microcredit have a Role to Play?” Journal of Development Economics, 97(2): 232- 243.
Maitra, Pushkar, Sandip Mitra, Dilip Mookherjee, Alberto Motta and Sujata Visaria (2013). Agent-Intermediated Lending: A New Approach to Microfinance”, BREAD Working paper 375. Available at: http://ipl.econ.duke.edu/bread/abstract.php?paper=375
Morduch, Jonathan (1998). “Does Microfinance Really Help the Poor? New Evidence from Flagship Programs in Bangladesh,” Working Paper, New York University.