What Do We Know About the Impact of Microcredit? Review of an Important New Research Paper

Microcredit has arguably been through more ups and downs than virtually any modern poverty intervention. For decades up through the 1980s, the thought was that the poor would never repay loans; credit to the poor was widely viewed as throwing money down a deep, dark hole. Government subsidized credit programs often went bankrupt when politicians would forgive loans before elections. After the success of the Grameen Bank, the 2006 Nobel Peace Prize awarded to its founder Muhammad Yunus, and the replication of its high repayment rates across many parts of the globe, microcredit became the silver bullet for addressing global poverty. For years the development field lived in a microcredit lovefest. Credit provision to the world’s poorest would unleash the entrepreneurial energy of the world’s low-income entrepreneurs to bring millions out of poverty. By 2013, there were well over 200 million households accessing microloans.

But it was the entrepreneurial narrative that primarily fueled microcredit, not clear evidence of its impact. There was always the nagging feeling that perhaps the myriad anecdotes of microcredit’s successful entrepreneurs were providing a misleading picture of the average effect it actually realized on its borrowers. The bombshell that landed in 2015 was a series of six research papers in the American Economic Journal: Applied Economics reporting results of randomized controlled trials in diverse countries, results showing that microcredit seemed to lack the transformative effects on entrepreneurial outcomes that we thought it did. While there was some evidence that microcredit spurred entrepreneurial activity and increased business income (in two of the studies), these potential increases were offset by a loss in wage income. There were no effects on consumption. In other words, microcredit maybe caused people to become more entrepreneurial, but not to put more food on the table for their families.

There have been several critiques of these studies, including my own, which suggests that these studies measured impacts on mostly marginal borrowers, those late-takers of credit opportunities who are likely to realize lower impacts than the more proactive first-takers of credit likely to have more profitable projects to finance. But even more recent and compelling evidence for microcredit by Emily Breza and Cynthia Kinnan shows what happens when you take it away. One important newer paper found that when the Indian state of Andhra Pradesh temporarily shut down microcredit institutions, it resulted in large losses in household consumption, wages, and incomes: Microcredit seems to matter.

But an excellent new research paper by Mahesh Dahal and Nathan Fiala (forthcoming in April in World Development) offers perhaps the most substantive critique of the six microcredit randomized trials that have been so influential to shape the more pessimistic view on its impacts. As well it offers reasonable evidence that microcredit is likely to be doing at least partly what it is supposed to. The critique centers around the statistical power of the studies, which happened to show rather large “point estimates” of impact, but which were statistically insignificant due to the low take-up rate of microcredit in the treatment group relative to the control group. Essentially what Dahal and Fiala convincingly argue is that the design of these studies vastly lacked the sample size necessary to detect any reasonable-sized effects that microcredit could be expected to affect on borrowers. The authors then pool together the data from all of six of the studies and find in more precise estimates that as a result of access to microcredit, business profits increased between 28 and 40 percent, statistically significant at the 99% level of confidence.

Dahal and Fiala are careful to note that even in the combined meta-analysis of the data there are still issues with statistical power. But it is important to reflect on this average increase of 28 to 40 percent increase in business profits across these studies, the very studies that were initially invoked to represent the relative ineffectiveness of microcredit. This is a sizable impact that would compare relatively well with many other types of development interventions that are far more costly to deliver than microcredit. And while some of this increase is admittedly likely to be offset by reductions in wage income as borrowers become entrepreneurial, estimates such as this provide substantial evidence that, while not “transformative” on average, microcredit is doing what it is designed to do.

Moreover, we know from more recent work by Breza and Kinnan and their co-authors (recent Nobel laureates Esther Duflo and Abhijit Banerjee) that microcredit has wildly heterogeneous effects. While for a large group of borrowers, the effects are moderate at best and others are unproductive with their loans and may simply get farther into debt, there is a sub-group of borrowers that thrives with newfound access to credit. The authors show that among the “gung-ho entrepreneurs” in one of the six studies in India, microcredit doubled the revenues of businesses relative to control businesses.

So while we still have much to learn about microcredit impacts, a few issues have been somewhat clarified in the years since the set of 2015 studies. First, as the research stands, on average microcredit is very likely to exhibit reasonably sized impacts on microenterprise profits. Second, given the statistical problems noted by Dahal and Fiala, it is going to be difficult to obtain the sample sizes in the future needed to identify more of the nuanced effects of microcredit; we may have to be satisfied knowing that it is a fairly effective intervention that spurs entrepreneurship and increases business profits. Third, there is emerging but solid evidence that microcredit is “transformative” for perhaps the upper 10-15 percent of borrowers.

So to increase the average impact of microcredit, we need to become much better at targeting loans to the right borrowers. And this both sets a policy goal for practitioners as well as a research agenda for academics. Indeed in continuing to refine the effectiveness of microcredit, it may end up being true that “a little less microcredit is more.”

• ← Reflections on a Self-Experiment: 24 Hours of Homelessness in San Francisco
• 10 Ways Academics Can Help Development Practitioners →