Microcredit has been expected to deliver an escape route out of poverty for large numbers of people in developing countries. However, experts are less hopeful now that numerous experiments in low- and middle-income countries have failed to find significant effects of microcredit on household incomes, living standards or asset ownership. The reasons are not well understood.
One possibility is that current practices do not select the right borrowers i.e. they need to possess special skills or complementary assets to realise large returns. To test this hypothesis, we designed a novel approach called TRAIL (trader–agent intermediated loans), where a selection of borrowers were delegated to agents chosen randomly from prominent local traders or informal lenders. The idea was to harness their experience of lending within the local community. Appointed agents were incentivised by commissions that depended on the loan repayment of the borrowers they recommended. Borrowers were individually liable for their loans. The loan duration matched the growing season of the local cash crop (mainly potatoes), allowing the use of the loan for agricultural production. Access to future credit was conditional on repayment. There were no group meetings, nor any savings requirements.
We conducted a field experiment in two districts in West Bengal, India to evaluate TRAIL against group-based lending (GBL), the traditional method of delivering microcredit. In the GBL treatment, village residents were invited to form groups with five members and then apply for joint-liability loans. Twenty-four villages were randomly assigned to each treatment. The TRAIL and GBL arms differed only in borrower selection procedures and loan liability; all other loan terms were identical. Only those village residents who owned less than 0.6 hectares of cultivable land were eligible to participate in either scheme. Loans were offered to 10 randomly chosen households from those recommended (in TRAIL villages) or those who formed groups (in GBL villages). The schemes were implemented over 8 quarterly loan cycles from 2010 to 2013.
We found that TRAIL loans increased the production of potatoes and farm incomes by 27–37% without any decline in income from any other source. Consistent with other recent microcredit experiments, the outcomes for the GBL borrowers did not noticeably change. Detailed analyses of our data show that the borrowers selected by the TRAIL agent were significantly more productive than those in the self-selected GBL groups. TRAIL loans had higher take-up rates and the scheme cost less to administer. We found no evidence that TRAIL agents siphoned off the benefits that accrued to borrowers. Finally, landless borrowers saw the largest income increases, so that TRAIL benefitted the least wealthy households the most.
These results suggest the need for further experimentation with the TRAIL scheme. Doing so could shed light on the validity of these results in other contexts, as well as the effects of varying scale and commission rates to ensure financial sustainability.
By Pushkar Maitra, Sandip Mitra, Dilip Mookherjee, Alberto Motta and Sujata Visaria