Entitled “Making economics relevant again,” a recent article in the New York Times described the path-breaking work of Esther Duflo, Abhijit Banerjee and their colleagues at MIT’s Jameel Poverty Action Lab, who use evaluations based on randomized trials to determine what works and what doesn’t in various social programs.  Much of their work has been on South Asia, as has some of the World Bank’s impact evaluations in education and health. 

In his blog post on the New York Times article, Dani Rodrik asks how these impact evaluations can be generalized.  This is precisely what the World Bank’s Development Impact Evaluation (DIME) initiative seeks to do by undertaking several evaluations of similar interventions, such as conditional cash transfers, and then aggregating them into a “meta evaluation” that provides guidance on the circumstances under which the intervention will and will not work. 

Neither the Times article nor Dani’s post mentioned an important contribution of impact evaluations:  to build political support for pro-poor programs.  If a rigorous impact evaluation shows that the program actually works, then it makes the case for scaling it up that much stronger, as the experience with Mexico’s Progresa  shows.  Of course the flip-side of this is that people who have a stake in a particular program are reluctant to subject it to rigorous impact evaluation, or try to discredit or ignore the results of an evaluation that shows the program did not benefit the poor.  As the title of Lant Pritchett’s paper on the subject puts it, sometimes “it pays to be ignorant.”