More Needed Research

Last week I attended the Microfinance Impact & Innovations Conference organized by the Financial Access Initiative (FAI) in New York. Never before, I daresay, has so much high-quality research (and so many high-quality researchers) on financial services for the poor been gathered in one place. The first two days were full of credible, largely randomized, studies, as well as rigorous thinking from industry consultants and managers.Within this field of research, the conference marked a huge step forward. The unprecedented efflorescence owes to the revolution in development economics that was launched barely ten years ago by Esther Duflo, Michael Kremer, and others. The movement skirts the big questions about economic development, instead attacking small questions with experimental rigor. The flowering owes also to the Gates Foundation, which through the FAI funded much of the research presented.The conference has been well-blogged. See this round-up and this and earlier posts from A View from the Cave. The best concise summaries of the new findings are on the FAI blog (day 1, day 2).An inevitable highlight for me was Esther Duflo's report on results from a microcredit impact study in Morocco---the first airing of these results except for a closed-door presentation at the Agence France de Developpement. (Write-ups by Forbes.com, Philanthropy Action, View from the Cave.) This is the first randomized test of microcredit in a rural setting, and in a setting that was "virgin" microcredit territory, to boot. (The previous two were done in Hyderabad and Manilla, which both had microcredit and other kinds of credit in relative abundance.) The study design was generally similar to J-PAL's Hyderabad study in that a microcreditor, Al Amana, agreed to randomize the order in which it rolled out its program into new territory.The results were pretty similar too. Take-up was far from universal: 16% of people who could borrow did so. After 24 months, indicators of poverty such as household spending and school enrollment had not budged. However, in continuity with the Hyderabad results, those who already had their own self-managed economic activities in operation---you know, farmers---invested more in those activities (bought more animals) and I think boosted sales and profits. They also diversified the types of animals they were raising. On the other hand, they worked less outside the home too, which may help explain the lack of net impact on income for the entire sample. Perhaps the loans appealed to the type of person who likes to work for herself. Esther emphasized that the analysis is ongoing. There is no paper yet.I also had a great conversation with Carlos Danel, co-CEO of Compartamos, about what constitutes transparency in disclosing interest rates to investors and clients. More on that later, I hope.Another highlight for me, and all present, was the lunch talk by Sendhil Mullainathan. He works at the intersection of economics and cognitive psychology, and the latter, he brilliantly showed, ]lends itself to great presentations. That's because the speaker can experiment on you, bringing the mind's peculiarities to life. Sendhil began with a puzzle: why do people sometimes treat predictable spending needs as emergencies, running to the moneylender to pay the teacher or midwife? Why do they do this even if they could cut costs by saving before instead of borrowing after? Perhaps, he suggested, they are so distracted by immediate needs that they cannot pay adequate attention to the future.The moment I remember most is when he flashed a series of images at us. Our task: focus hard and shout out the color of each image as quickly as we could. The third or fourth image looked like: Blue. You can guess what most people said. I know I got it wrong. Sendhil used this foible to argue that people have limited attention. You can't think about everything at once. Even when you have determined to focus your attention on a priority task---in our case, naming colors; in the case of the poor, saving for the midwife---your unconscious bucks up and throws you off kilter. You cannot completely control what you think about---more to the point, worry about. The function of financial services, he submitted, is not merely to help people assemble "usefully large lump sums," as Stuart Rutherford says, but to economize on the limited resource of attention. Once you bind yourself to making weekly payments on a loan or weekly contributions in a savings club (ROSCA), you've got a way to save. You've got one less thing to worry about.Intriguing. But I'm not sold yet. I do wonder if this economist is moving a tad too quickly to project the core metaphor of economics---the problem of allocating a limited, homogeneous resource---onto the human mind. I take the foible he demonstrated as a sign that information processing in the brain is massively parallel and largely automatic. Even as I was looking for color with one clump of neurons, another bunch of them was reading "Blue." That seems like a fuller description than that of the limited resource alone. But Sendhil is brilliant and has thought about this a thousand times more than I have, so maybe he could show me the error in my thinking. Maybe I just wasn't paying enough attention.Actually, for the conference as a whole, I didn't, at least not enough to absorb all the fresh findings. I found it hard to sit still and listen to a dozen research presentations in a row. Much as I believe in the value of this research, it did not fire me to play audience to it. Mostly, I was struck by the huge effort in each incremental step in the research enterprise takes: planning experiments, executing them, asking thousands of people hundreds of questions, entering and cleaning numbers, then crunching them. I don't have the constitution for it. But I'm glad the researchers do.Another highlight for many was the panel on the morning of the second day, "What Don't We Know That We Ought To?", which I was lucky enough to moderate. Featured were Christopher Dunford of Freedom from Hunger, Rich Rosenberg of CGAP, and Abhijit Banerjee (of MIT). They were predictably great. Laura Starita of Philanthropy Action has summarized the discussion. After the featured talent spoke, I abused my privilege as moderator to inject the questions that did fire me, which arise from the interactions of the three notions of success that form the core of my book (chapters 6--8):

• Will the in-depth, anthropological work represented by Portfolios of the Poor continue to stand as an independent source of insight into how financial services affect poor people? Or, to put it provocatively, could randomized studies like the ones at this conference some day take down Portfolios of the Poor the way they have the mythology about microcredit reducing poverty? Could we someday say, "Portfolios of the Poor was a nice collection of stories and hypotheses, but we tested the hypotheses rigorously and they didn't pan out"?
• Microfinance is singular within the worlds of philanthropy and foreign aid in the way it has fostered the growth of a big industry. These rigorous, randomized studies are relatively good at measuring direct impacts on clients. But that is not the same as judging the contribution of the microfinance industry to the development of nations. Imagine if randomized studies had been done on home mortgages in the U.S. in 2003--04. They might have shown home ownership, spending, etc., rising. But they would not have told us the whole story. A core question, it seems to me, is when is growth development? I don't know the answer to that, but it should also be an object of serious study, using quite different methods. (The crisis unfolding in Andhra Pradesh even as we spoke only dramatizes the need.)