The experimenters strike back in microcredit research
Tuesday, June 10, 2014
“GIVE a man a fish”, said Bono, an Irish singer turned philanthropist, “he’ll eat for a day. Give a woman microcredit, she, her husband, her children and her extended family will eat for a lifetime.” In the print edition of April 19th, we reviewed a paper by the World Bank that supports the notion that microfinance reduces poverty. The authors of the study, Shahidur Khandker and Hussain Samad, analysed observational data from households in Bangladesh spanning over two decades. They disputed the pessimistic results reported from many randomised controlled trials (RCTs) and found, on the contrary, that microcredit increases personal expenditure, household assets, labour supply and schooling of children. Now the experimenters are hitting back.
A new experimental study by Bruno Crépon, Florencia Devoto, Esther Duflo and William Pariente investigates the impact of a microcredit programme in rural Morocco. The authors randomly assigned 162 villages to either the treatment group (given access to microcredit) or the control group (not given access to microcredit). The authors, all of whom are affiliated with the research organisation J-PAL (to give it its full name, the Abdul Latif Jameel Poverty Action Lab), found that microcredit is not an instrument “that fuels an exit from poverty…at least in the medium run.” On average, borrowers invested more in their own businesses, for example by buying cattle. However, since takers of microloans spent less time on wage labour without increasing the time they devoted to self-employment, the introduction of microcredit appears to have decreased the number of hours spent on work. The positive effect on investment, the authors reckon, was consequently offset by a reduction in income from wages. Overall, the J-PAL researchers conclude, access to microcredit did not lead to gains in income, consumption or education.
Reconciling the seemingly conflicting results from the two papers is not straightforward. While Messrs Khandker and Samad study long-run effects, Mr Crépon and his co-authors do not, which is one of the criticisms that the former raise of randomised evaluations of microloans in general. It is possible that microcredit can raise income and reduce poverty, but only in the long run (which one might expect from the effect of the programme on investment). The two studies also take place on two different continents. An alternative hypothesis is therefore that what seems to work in Bangladesh does not work in Morocco. This issue is what economists usually term “external validity” and it refers to the extent a study’s results are generalisable outside the context in which it took place. Hopes of finding a cure to poverty that will work everywhere have not been realised. Access to microcredit may not be an exception.