A very interesting microfinance experiment is in the new issue of the American Economic Review, one of the premier journals in the field (Published, but gated, version here. Ungated version here). The paper is by FAI Affiliate Xavi Giné, Jessica Goldberg (see her recommended reading on savings here), and Dean Yang. It's not often that microfinance makes the pages of AER; it's a testament to the work that Xavi, Jessica and Dean did to set up this experiment and their careful analysis of the data.
In brief, the experiment tested the effects of fingerprinting borrowers from a microcredit program in rural Malawi. I had the opportunity to interview Xavi and Dean (separately) for my upcoming book on economic field experiments and we talked about this work. I’ll let them explain the project and its implications in their own words.
First, here’s Xavi. The context is a discussion of how sophisticated, or unsophisticated, poor households are in their choices and use of financial services:
Xavi: I would agree with that. Individuals may not know whether they are facing the best product available but once they borrow, they do respond to incentives. Let me give you another example: the work I did with Jessica Goldberg and Dean Yang in Malawi. So in Malawi there’s no unique identification system but the turnover of loan officers is high, especially in rural areas. This means that institutional memory about borrowers is lost when a credit officer moves to another job. What’s happening in this situation is that a lot of folks try to borrow and never repay because they know there is little penalty for defaulting: they can give another name or somehow convince the next loan officer that they’ve never borrowed before. The response from lenders is to limit the supply of credit. So banks are playing a one-shot game, where every season there’s a tabula rasa. They can’t offer dynamic contracts because there is no way of rewarding the good borrowers or penalizing the defaulters.
Enter fingerprinting. What the lender can now do is to require fingerprinting as part of the loan application so it can create credit histories for its borrowers. Credit officers can now tell their clients, “You can default, but there is no way you’ll be able to fool us into giving you another loan in the future.” All of a sudden that creates the possibility for the bank to use dynamic incentives, provide better terms for faithful repayers and shut out the defaulters.
What actually happened there—going back to your sophistication story—is that the individuals that were more likely to default ex anteactually increased repayment dramatically. In fact, it’s quite interesting. First off, they borrowed less, making sure that they could repay whatever they took out. There was also less diversion of the loan (these are agricultural loans to buy inputs for a crop; the loans are funded by an institution that is buying the harvest) into other spending. So what these people did is they put more of the loan and more effort into farming and they get better yields and they repay at a higher rate.
Here’s my back and forth with Dean on the issue of paternalism—how much should MFIs try to control what borrowers do with a loan?
Tim: The results [in the Malawi fingerprinting study] line up very nicely with theory in terms of affecting adverse selection and moral hazard. [But] I’ve long thought one of the better aspects of microcredit is that it’s less paternalistic than a lot of other charity approaches. It lets people make decisions on their own about how they will spend the money. It’s not forcing them to do one thing or another. On the flip side, sometimes paternalism helps. And that’s not just in terms of helping people do what we as outsiders think they should do, but in terms of helping them achieve what they say they want to achieve but find difficult to accomplish.
Overall the history of development is littered with a lot of programs designed to control the behavior of the poor, based on the belief, explicit or implicit, that they aren’t able to control themselves. [Conversely, there is the] Development as Freedom concept that escaping poverty is about people gaining more and more control of their lives. [But this work suggests that sometimes] helping them gain more control of their own lives in the long-term might require exercising some control over them in the short-term.
Dean: I think that’s a first order question when thinking about our fingerprinting study in Malawi. What ended up happening in practice... was that the fingerprinted farmers in the highest risk category of borrowers responded to the fingerprinting by allocating more land to paprika and using more inputs on their paprika plots. And this was good news to the lender because that’s what the loans expressly were for. [The] concern of the lender was that the loan be used for paprika cultivation and not something else. This was important to the lender for two reasons. One, because the whole reason why the loan was given was because paprika-growing was thought to be a profitable and income-increasing activity for these farmers. Secondly, because the paprika was sold through a controlled marketing channel. Basically the only practical option for the famers was to sell to this one paprika buyer, and the company had an agreement with the lender that it would extract repayment of the loan from the farmers’ crop proceeds before turning over the remainder of the proceeds to the farmers. So growing paprika not only was thought to be income-raising for the farmers but also dramatically increased the lender’s likelihood of getting repaid because the paprika buyer was garnishing the farmers’ proceeds to repay the loans.
The very interesting possibility, which we unfortunately didn’t get strong enough statistical results on, is the possibility that compelling farmers in this highest risk category to grow paprika could have actually raised their income. Unfortunately our sample sizes weren’t large enough and therefore the statistical significance levels aren’t what we would want them to be, but there’s verysuggestive evidence in the paper that income actually rose substantially for farmers in this highest risk category. So when they were compelled to grow paprika instead of using the loan for something else, they actually experience an increase in income. I very much want to do a future study that increases the sample size so we can test to see whether or not in fact there were increases in income due to fingerprinting.
If that result holds, it really does connect back to your question. If by cracking the whip on borrowers we find they actually raise their incomes that would bring the question to the fore very, very prominently. I don’t think this intervention would cross the line into an ethically questionable area. In fact, I think it’s still quite far from an ethnically questionable area. The context is really private lending transactions; lenders providing loans for a specific commercial purpose. I don’t think it’s questionable that a lender can, if it chooses to do so, seek to require that the loan be used for the purpose that the loan was intended for. Particularly if empirically we find that strong control over the use of the loans ends up being beneficial for farmers, by raising their incomes.
But I think it raises other very interesting and important questions. If the results hold in a larger sample you have to ask why the farmers weren’t taking these actions on their own. Certainly that would call for more work explaining why farmers weren’t taking advantage of these opportunities. Are the farmers present-biased? Is there some other factor preventing them from taking these actions without compulsion from the lender?