Research recap: prepare for next week's Microfinance Impact & Innovation Conference

As we prepare for our upcoming Microfinance Impact and Innovation Conference that will take place on October 21-23 in New York City, we are looking back to the last time we drew together so many of our best minds in microfinance in one place. In October 2008, FAI and IPA co-hosted a microfinance conference at Yale University that focused on the first microfinance impact studies. Next week at the conference we’ll get a first look at the results of follow-up studies. To get you caught up on where things stood this time last year, below is a blog post from that conference by Timothy Ogden, editor-in-chief of Philanthropy Action.

We're live-blogging the Innovations for Poverty Action/Financial Access Initiative Microfinance Conference.

The first panel of the conference was on “Credit Product Design: Monitoring and Enforcement” – put more simply, how does the design of microfinance products affect repayment rates?

I’ll briefly summarize the findings of two of the studies that were presented.

1) Erica Field of Harvard presented on a study (conducted with Rohini Pande of the Kennedy School of Public Policy) she conducted in India with urban women borrowers in India. One of the accepted “rules” of microfinance is that rigid, frequent payments are necessary to ensure repayment. Typically borrowers must meet with their group and loan officer for repayments every week. These rigid schedules, though, may be bad for borrowers because they don’t allow the borrowers to adjust for a “shock” – a family sickness, bad weather, etc. – during any particular week that limits their earnings.

Field and Pande worked with an MFI to randomly assign some groups to a monthly repayment schedule. What they found was that there was no impact on repayment rates (in fact there was a higher default rate among weekly payment groups, though not a significant difference). This is very interesting for MFIs because the largest cost of delivering services is loan officers running these weekly meetings. If MFIs can move to monthly meetings rather than weekly ones, it would cut costs by nearly 75%, thereby allowing them to cut interest rates and enabling them to reach much more remote clients.

Interestingly though, Field and Pande also found that monthly payers worked more in the day before a payment was due than weekly payers – indicating that the borrowers weren’t able to save up money for repayment over the course of the month. The theoretical explanation is that either the borrowers themselves have a hard time holding on to cash in hand or that they cannot protect the cash in the household from others. This is consistent with the general theory of microfinance and would put an upper limit on how infrequently payment meetings could be held.

Field and Pande were also measuring social capital developed by participating in the groups and as a result found evidence that contravenes one of the underpinnings of the group lending model. While setting baselines for their social capital measures, they learned that the women in the groups did not know much about the other members of the group (e.g. names of husbands or children) even though group members recruit the other members of the group. Generally, it’s believed that the group lending model is important because forming groups serves as a way of screening out bad credit risks. Group participants won’t bring in members who won’t repay (since they’ll be liable to cover the default of other group members). Field and Pande’s data suggest that at least in this case there is actually little screening going on.

2) Craig McIntosh of Georgetown presented on a laboratory experiment with borrowers in Guatemala designed to understand how the implementation of a credit bureau affects borrower behavior. The reasons for doing the study in the first place came as a bit of a shock to me: competition in microfinance can be a bad thing.

One of the drivers of repayment in microfinance is that if you default on your loan, you can’t get another loan. But this only works when there is limited access to microfinance – e.g. there’s only one MFI in your village. When there are multiple MFIs a borrower can default on a loan and simply get another loan from another provider. I’ll quickly note that I’ve heard from people on the ground that “microfinance-kiting” has become common in urban Malawi – borrowers sequentially take out loans from a number of different MFIs using each new loan to repay a loan from a different lender.

The way to combat this behavior is to implement a credit bureau where MFIs can share data on their clients. But technical challenges aside, this is more problematic than it might seem. Since most borrowers are in a group, MFIs typically only keep repayment data on a group level (which for many reasons is appropriate since the group is the entity that is doing the borrowing). This group-level credit score though makes the risk to each individual borrower higher. If there is a deadbeat in your group you are not only shut out of credit from your current lender, you now are shut out from all the other MFIs too. This is a problem for MFIs because of a problem known as adverse selection. Essentially, because the risks are higher, the clients most likely to pay are less likely to take a loan – they’re unwilling to put their credit on the line when the stakes are even higher than normal.