How to Evaluate Microlending Efforts

Recently the Boston Globe ran an article discussing program evaluation of microlending programs. The article described some popular microlending programs, which typically offer small loans to folks in third-world countries to start their own businesses. More information on one popular microlending program, from which its founder won the Nobel Peace Prize, can be found here.

I'm interested in examining this article from the perspective of a program evaluator. As the article stated:

Karlan and his co-author, Jonathan Zinman, an associate economics professor at Dartmouth, looked at a bank in the Philippines that offered microloans. They created their controlled experiment by altering the algorithm the bank used to evaluate creditworthiness so that some borderline applicants were randomly denied loans while other otherwise identical applicants had loans approved. The researchers then followed up with the borrowers and nonborrowers to see what difference the loan had made.

The answer was not much. Neither household income nor spending rose for those who got microloans. And borrowers who did put the money into their businesses - instead of using it, as many did, for household expenses - actually shrank rather than grew their businesses. Karlan and Zinman suggest that this might be because the business owners were taking advantage of the loan to fire unproductive workers to whom they owed financial favors, and those firings seemed to explain the very small gains in profit Karlan and Zinman found. In addition, the gains accrued only to male entrepreneurs, not the women usually targeted by microcredit programs.

The second study, co-authored by Abhijit Banerjee and Esther Duflo, economics professors at MIT, along with Rachel Glennerster, executive director of the Poverty Action Lab, and an MIT economics doctoral student named Cynthia Kinnan, found a slightly larger impact, though a selective one. Working with a microcredit bank in India that was looking to expand in the city of Hyderabad, the researchers did find some small positive effects. Borrowers who already had a business did see some increase in profit. Households without businesses that the researchers judged more predisposed to start one were found to cut back on spending, suggesting they were saving to augment their loan for a capital business expense like a pushcart or a sewing machine. The researchers also found small but encouraging shifts in household spending across the board, with less money spent on “temptation goods” like alcohol, tobacco, and gambling.

Still, overall household spending - a key indicator of financial well-being - stayed about the same. And the researchers found no effect on children’s health or education levels, and the women in the borrower homes were no more likely to play a role in household decisions than those in the control group.

Is this how you would have evaluated this program? What other methods could have been used? Do these indicators do a good job of addressing the research questions? If not done already, microfinance efforts are certainly ripe for some qualitative research to gain an understanding on what impact, if any, these programs have had on participants. I suppose there might be a large difference in how poverty is defined by evaluators and participants!