Microfinance on the Margin
New blogs and articles in the public online press have brought increasing attention to the results of the six randomized trials on microfinance presented in the January 2015 issue of the American Economic Journal: Applied Economics. Make no mistake: The impact of microfinance–in its credit form–was oversold from the beginning, its rate of implementation far outpacing positive evidence of impact for nearly two decades. What allowed this to occur was the strong impact of narrative, which as I discussed in a World Bank Impact blog, is often far more powerful than hard statistical evidence in motivating people to action. Everyone was infatuated with the narrative of microfinance: Those on the left loved its story of trickle-up financing, its focus on women and artisans, and direct empowerment of the poor. Those on the right loved how it promoted grass-roots capitalism, fostered a culture of entrepreneurship, and all without large subsidies. How powerful is narrative? The power of narrative moved microfinance from a few million borrowers worldwide in the early-1990s to over 200 million borrowers just two decades later. Microfinance emerged as the biggest grassroots movement in the history of micro-level development interventions. Everyone loved microfinance…until recently, when we found out it didn’t work as well as we thought.
While it is true that microfinance is not the silver bullet that its advocates had maintained, there is good reason to believe microfinance has more impact than is revealed by the “big-six” RCTs–or at least how their results have been interpreted in public discourse. This is the subject of a forthcoming comment on these studies in the Journal of Development Effectiveness. The crux of the comment relates to a statement that the researchers carefully make in the introduction to the studies:
“Another key caveat is that these studies have nothing to say about impacts on inframarginal borrowers. It may well be the case that impacts are substantially different on the borrowers and/or communities already being served before the lenders in these studies began experimenting on the margin.” (Banerjee, Karlan, and Zinman, p.3)
The issue is that there is good reason to believe that experiments evaluating microfinance on the external margin, may substantially understate the average impact of microfinance across the general population, and specifically on the impact microfinance has had on inframarginal borrowers, those who already had loans when the experiment took place. The model that lays out this idea formally, built around the Rubin causal model, tries to capture the temporal selection dynamics into microfinance, which is highly endogenous with respect to underlying productivity. The intuition is that if every borrower has an underlying level of productivity and an idiosyncratic shock in every period that affects productivity (the latter determined by opportunities, changes in motivation, etc.), then a borrower only takes a loan when her productivity is greater than the microfinance interest rate, and this is more likely to be true sooner for borrowers with this higher underlying productivity. As a result we should expect early takers of microfinance to be more productive (and exhibit larger impacts) than the kinds of borrowers who take up microfinance loans only on the external margin (maybe only when induced by an experiment). The analogy I make in the paper is with a new iPhone model: The iPhone junkies who camp out overnight in line to buy the new iPhone model on its release day are likely to realize a substantially higher level of consumer surplus than people who only get it months or years later through a special promotion, or when the phone is marketed later in marginal sales regions.
To make a long story short, where these studies do a great job is in telling us that more microfinance in saturated areas is unlikely to be very helpful or effective. This is important. But what they do not tell us is the average impact of microfinance among those who have taken microfinance loans. And this is also important. And it is important because while we might want to dial back microfinance in saturated areas, we may very likely want to keep rolling it out and testing it in yet unserved areas.
Hints of the potentially greater impact in less-served areas are actually found in these studies themselves. Three of the studies (carried out in Mexico, Bosnia, and Mongolia) were implemented in fairly saturated microfinance contexts. The other three (India, Morocco, and Ethiopia) were implemented in less-saturated contexts as seen in the above figure from the paper. But what is interesting is that even though insignificant impacts were found on self-employment income across the board, all of the high point estimates are found in these non-saturated countries. And indeed there is a pretty strong relationship between having high point-estimates of impact in these studies and having your experiment being run in an area that is not saturated with microfinance at baseline. The figure plots point estimates of impacts on net income from self-employment divided by the control mean against the existing level of microfinance lending in each country study overlaid by a quadratic fit. (See figure below.)
Although the authors of these studies go to extreme care to qualify what one can infer from the set of experimental results, that these results only measure the impact of microfinance on the external margin, this same level of nuance is often less manifest in public commentary, especially by some non-academics who want to keep things simple. The disappointment that microfinance hasn’t produced the results that were initially claimed may provoke a sense of disillusionment that finds itself eager to muster the microfinance lynch mob. But while microfinance as an intervention on the margin may show impacts limited to moderate improvements in consumption smoothing and growth in entrepreneurship, there is still room for further research on its impact on highly productive and eager entrepreneurs in developing countries. These are the “infra-marginal” borrowers, those high-productivity borrowers with high business aspirations for whom the opportunity to borrow at reasonable interest rates makes them jump at the microfinance opportunity as soon as its available. The result of this new research may be that less microfinance is more, and that we just have to keep microfinance in its niche.
Follow AcrossTwoWorlds on Twitter @BruceWydick