The Use of Randomized Evaluations in Microfinance
As the microfinance industry continues to expand, and innovations in lending methods and financial product offerings continue to emerge, the need for sophisticated program evaluation tools has become increasingly apparent. To this end, researchers have started utilizing randomized control trials – a methodology developed for use in the medical field – to better examine what is and what isn’t working in the fight against global poverty. The evaluations seek to create real world laboratories by selecting certain groups of people – the treatment group – to receive a particular financial product, and maintaining a second group – the control group – with which to compare results. (Note: Randomized controlled trials are one method under the impact evaluation umbrella, which describes efforts to evaluate program interventions and their impact.)
One of the reasons that randomized trials are so popular is because many of the traditional methods of impact evaluation have been criticized for being subject to self-selection bias. That is, the evaluations were arguably unable to isolate whether a given result was due to the impact of the financial instrument being studied, or something inherent in those individuals who successfully became recipients of those instruments. With randomized control trials, however, such bias is far more controlled because the groups being compared are randomly assigned. Thus, for example, instead of asking how recipients of microloans differ from the rest of the population after the fact, they manufacture two similar groups, provide one of them with access to microloans (but not the other), and then observe what happens.
Several microfinance programs and products have been evaluated for their impact, in both beneficial and adverse terms, caused by such programs. Researchers have evaluated microcredit, savings and microinsurance among other initiatives. Under the umbrella of Innovation for Poverty Action (IPA), leading researchers in development economics from Harvard, Yale, MIT, and LSE have been measuring microfinance program impacts in 40 countries. One study, for example, focused on savings by the poor. Their research has shown that while the poor participate in informal savings groups, they underutilize formal savings (banks) and tend to spend savings compulsively. The solution centered on developing a ‘commitment’ savings product that help savers set voluntary savings amount and timelines while removing savings from the household and providing security.
Dean Karlan, Professor of Economics at Yale University, has used randomized trial methodology to examine the impact of microcredit. He concluded that “the canonical case for microcredit– that access increases profits, business scale, and household consumption– is not supported on average.” Notwithstanding, in his research in the Philippines he found what he considers to be a valuable “social component” of microfinance, noting “(t)hat microloans increase ability to cope with risk, strengthen community ties, and increase access to informal credit. Thus, microcredit here may work, but through channels different from those often hypothesized by its proponents.”
Microinsurance is another area that has been examined by impact evaluations researchers, including Jonathan Morduch, Professor of Public Policy and Economics at NYU. A recent study notes that “impact evaluations can test the effectiveness of two different insurance products or test the effect of specific elements of the products, such as different marketing techniques, pricing structures or distribution channels”. Such evaluations, it is argued, can enable appropriate operations and access to clients which would enable increased sustainability and social impact.
Another adherent of randomized trials, David Roodman, Research Fellow at the Center for Global Development, notes that the use of randomized controlled trialsoffers “more credible data than microfinance has had before” in terms of evaluating its impact. His recent review of “Randomized Test of Microcredit in Mongolia” provides an interesting take on the ‘method of randomization’ used in the studies that entailed group-based loans versus individual microloans.
But such trials are not without their shortcomings. Policymakers such as Abhijit Banerjee of the Abdul Latif Jameel Poverty Action Laboratory (J-PAL) have remarked that they aren’t interested in whether a certain financial instrument works in a few villages, they want to know if it works nationwide. In this respect, randomized trials might not be as useful.
That the value of randomized trial research is creating a new body of knowledge for microfinance adherents is contained in a new paper “Latest Findings from Randomized Evaluations in Microfinance”. Research shows that delivery of financial services products for the poor, while facing limitations, is a viable approach as long as its design is appropriate to meet their needs. We will continue following this exciting research and any insights it reveals.