Ideas to Action:

Independent research for global prosperity


David Roodman's Microfinance Open Book Blog


*Yet Another Randomized Trial of Microcredit

The latest randomized study of the impact of microcredit has popped up on the web. Snarky blog post title notwithstanding, I very much welcome having yet another randomized test of microcredit---by my count, the fifth---because only after we test in a variety of forms and circumstances can we generalize with (cautious) confidence. We have been fortunate in the diversity so far: group and individual microcredit, rural and urban locales in India, the Philippines, Morocco, Mongolia, and now Bosnia and Herzegovina.

The cooperating lender in this newest study was EKI, one of a clutch of microlenders created and financed by outsiders after the explosion of Yugoslavia. I believe it is the first non-profit studied, an important distinction given all the debate about the role of the profit motive in microcredit. And, somewhat bizarrely, the study brings diversity of another kind to the literature: where the India and Morocco trials took place in overheating markets, this one occurred as economic crisis hit and a microcredit bubble popped. In December 2008, as the experiment began, EKI had a "portfolio at risk" (loan amounts outstanding owed by those at least 30 days behind on repayment) of just 1.63%. Within a year, the PAR shot to 10.83%.

This study, like the others, tracks impacts over an arguably short period, about a year. The write-up is by a lot of the people who ran the experiment in Mongolia. But where that one randomized at the village level---whole villages either got offered microcredit or didn't---this one experiments at the individual level using the method of Karlan and Zinman, in which some borderline loan applicants are randomly "unrejected." It thus shares other traits of the Karlan and Zinman Philippines study. It looks only at individual microcredit. As the authors take pains to emphasize, the "unrejection" design makes it a study of the impacts on marginal rather than typical clients, ones who would not normally have been offered microcredit. And, as Milford Bateman, who happens to know Bosnia and Herzegovina well, might hasten to note, it does not look at the displacement effect: non-borrowers outside the sample might be hurt by stronger competition from borrowers in it.

The average subject household had 3.2 people and a combined income of 18,175 marks ($11,123)/year, working out to about $10/person/day.

The results of the experiment accord with those from other locales, but in a way that gives food for thought. There is no sign of improvement, relative to the control group, on household spending, a key poverty indicator. But there are lots of signs of microcredit stimulating microenterprise. Microcredit caused 6 percentage-point increases in households reporting self-employment, and those reporting owning a business. Or more like: amidst the economic crisis, households with access to EKI microcredit were 6 percentage points less likely to close their businesses. Less-educated households tended to go for agricultural activities, such as buying livestock or fertilizer. Better-educated (more urban??) households tended invest in service businesses, perhaps such as petty retail.

Meanwhile, there are signs, as in India and the Philippines (and elsewhere I think), that households, once prodded by the credit into investing in a business activity, marshaled other resources too. Those offered loans cut back on alcohol and cigarettes. Those with less education also spent less on food. Those who probably had more savings, as indicated by having higher education or an existing business, withdrew savings as they took the loans. Finally, in low-education households, which were more engaged in farming, the loans led family members in the 16--19 age range to work 35 hours more per week than in the control group, and to attend school 20 percentage points less often.

The researchers found interesting patterns when they split their sample by gender. The increase in business starts (or reduction in shutdowns) occurred mostly when the borrowers were female. The marshaling of resources along with credit---tapping savings; spending less on food, cigarettes, and alcohol; employing teenagers---occurred on the male side. This suggests that it is the men who are more apt to invest in substantial businesses.

For many, I think the most thought-provoking results from this study relate to the reduction in spending on food and the increase in teenage labor. I hesitate to say "child labor" since the definition of adulthood is fuzzy, context-specific, and generally lower in poor, agricultural societies. It is entirely possible based on what I know about Bosnia and Herzegovina (almost nothing) that these low-education, rural families in the midst of an historic economic crisis made wise choices under their circumstances, taking advantage of the credit, tightening their belts, and pulling their teenagers out of school to work on the farm. In Sri Lanka, David McKenzie and coauthors found that it was indeed the male microentrepreneurs---the ones whose activities here are associated with teenage labor---that earned much higher profit. Perhaps after the study concluded, the households that made these sacrifices ended up better off than similar families in the control group.

But the study nevertheless highlights the tension between microcredit as an intervention and our understanding of how poverty is usually reduced. Countries do not get rich by encouraging poor farmers to farm a bit more. They do, however, get rich through processes that involve lots more investment in education. I am not in a position to gainsay these families' choices to pull their late-teens out of school, this study makes me more confident that an intervention that drives an economy toward household-level production doesn't push it toward long-term prosperity.

Of course, one can say the same about projects to bring solar lamps and other good things to the poor. A more realistic goal, and worthy if achieved at reasonable cost, is to make EKI and its peers into self-sustaining institutions that do more than just credit, bringing a broader and safer range of financial services to deserving people.

Related Topics:


CGD blog posts reflect the views of the authors, drawing on prior research and experience in their areas of expertise. CGD is a nonpartisan, independent organization and does not take institutional positions.