No abstract

Randomized experiments, as developed by Esther Duflo and Abhijit Banerjee at the Abdul Latif Jameel Poverty Action Lab (J-PAL), offer a novel, evidence-based approach to fighting poverty. This approach is original in that it imports the methodology of clinical trials for application in development economics. This paper examines the analogy between J-PAL's field experiments in development economics and randomized controlled trials in medicine. Randomized controlled trials (RCTs) and randomized field experiments (RFEs) are commonly treated as identical, but such treatment neglects some of the major distinguishing features that make each experiment specifically apt for use in its respective field. The central claim of this paper is that the analogy between medicine and development economics is incomplete because the central dimensions of RCTs are not simply different, but altogether lacking in J-PAL's approach. This weakens both the political and the theoretical power of such experiments in development economics.

While the history and methodology of laboratory experiments in economics have been extensively studied by philosophers, those of field experiments have not attracted much attention until recently. What is the historical context in which field experiments have been advocated? And what are the methodological rationales for conducting experiments in the field as opposed to in the lab? This article addresses these questions by combining historical and methodological perspectives. In terms of history, we show that the movement toward field experiments in economics has two distinct roots. One is the general orientation of medical and social sciences to evidence-based policy evaluation, which gave rise to randomized field experiments in economics (e.g., behavioral public policy, poverty alleviation policy). The other is an awareness of several methodological limitations of lab experiments in economics, which required practitioners to get out of the lab and into the field. In these senses, the movement is a consequence of influences from both outside and inside economics: the general evidence-based trend in policy science and an internal methodological development of experimental economics. In terms of methodology, we show that these two roots resulted in two somewhat different notions of “external validity” as methodological rationales of field experiment. Finally, we suggest that analysis of experiments as exhibits highlights a methodological strategy in which both strands complement each other.

No abstract