Since I will be working on a microfinance project in the summer, I have been reading and reviewing related articles. I just read this new AER paper “Credit Market Consequences of Improved Personal Identification: Field Experimental Evidence from Malawi” by Xavier Gene, Jessica Goldberg, and Dean Yang. Here’s the abstract:
We implemented a randomized field experiment in Malawi examining borrower responses to being fingerprinted when applying for loans. This intervention improved the lender’s ability to implement dynamic repayment incentives, allowing it to withhold future loans from past defaulters while rewarding good borrowers with better loan terms. As predicted by a simple model, fingerprinting led to substantially higher repayment rates for borrowers with the highest ex ante default risk, but had no effect for the rest of the borrowers. We provide unique evidence that this improvement in repayment rates is accompanied by behaviors consistent with less adverse selection and lower moral hazard.
Asymmetric information is one major imperfection of the credit market. Lenders do not know whether individuals borrowing money are risky or not, and to mitigate this problem, they impose collateral requirements. Once the money is borrowed, there is also the potential of moral hazard because the bank doesn’t supervise the borrower, and the borrower might choose a riskier project since he already has the money in his hand.
Microfinance institutions are famous for using social capital instead of tangible physical collaterals to reduce default rate. Village life is not anonymous, and one’s reputation gets ruined if he doesn’t pay back the loan. Group lending also enhances the mutual supervision between each group and monitors the riskiness of the projects that individuals choose. But due to rapid growth and increased competition, an increasing number of clients are “overindebted”. Like traditional lenders, microfinance institutes are also turning to credit bureaus (which requires the existence of identification methods).
The authors use a randomized control trial, so they are able to conclude causal impact of fingerprinting on the repayment and investment choices of borrowers. A few points are worth noting in this paper:
First, they predicted a “credit score” for each of the participants (in control and treatment group). They limit the sample to control group and regress repayment rates on club-specific and farmer-specific variables. The predictions are fitted values using this regression model. Here is their justification:
Conceptually, the resulting index will be purged of any bias introduced by the effects of fingerprinting on repayment because it is constructed using coefficients from a regression predict-ing repayment for only the control (nonfingerprinted)farmers.
I don’t find this convincing, though. The observations in the treatment group will be used twice — in the regression and then in prediction.
Second, they coded interaction terms of treatment and predicted likelihood of repayment. Incorporating this term enables them to examine the treatment effects for borrowers with different ex ante risk levels, which is a major interest of the paper.
Third, they record land devoted to different kinds of crops and inputs into paprika as a measure of moral hazard. Because by devoting more land to grow paprika, the farmer is reducing her risk of not being able to pay back the loan.
They also measured the cost-effectiveness of fingerprinting, which is a common thing to do in development program evaluation. The total benefit per individual fingerprinted is calculated to be MK 490.50 (US$3.38) and total cost per individual fingerprinted MK209.20. Benefit-cost ratio is 2.34.