A Nobel Prize for “Randomistas”
By Asanka Wijesinghe
On 02 November 2019
The hype generated since the announcement of the Nobel prize in Economics has not yet died down. The “randomistas”-the researchers who widely use randomized control trials (RCTs) for causal identification- were exultant hearing that Abhijith Benerjee, Esther Duflo, and Michael Kremer were awarded the 2019 Nobel Memorial Prize in Economic Sciences. The trio were awarded the Nobel for “their experimental approach to alleviating global poverty”. Esther Duflo’s achievement is remarkable as she is the youngest person to ever win the Nobel in economics. She was a John Bates Clark Medal winner and a MacArthur “genius grant” winner. Now she is the second woman to win the economics Nobel.
The Nobel Prize in economics (yes, it is a real Nobel irrespective of the tiny group of dissenters) in this year is significant due to the radical empirical approach advocated by the laureates. It is undeniable that Benerjee, Duflo, and Kremer transformed the development economics research by pioneering a rigorous approach to empirical research. Robert Solow once remarked that the most important tool for research was one economics professor and one research assistant. Benerjee, Duflo, and Kremer changed this conventional research production function. In their empirical approach, large survey teams, coordinators, and management skills are employed. They work closely with policy makers. Beyond that, they set up Abdul Latif Jameel Poverty Action Lab (JPAL) which has inspired many development researchers. When the news broke out, twitter feed was abuzz with stories from researchers about how they interacted with the laureates. Everybody had something to relate with them, showing their impact and their approachability.
The tool kit used to analyse poverty was broadened over the time. The 2015 laureate Angus Deaton integrated the measurement of well-being of the poor into the analysis of poverty. A well-articulated microeconomic theory on incentives and information was also developed, and the knowledge on behavioral constraints was broadened. Benerjee, Duflo, and Kremer built upon this foundation to enrich the field with rigorous causal identification method. The RCTs are now used in impact evaluations worldwide. Academically, RCT based studies are mushrooming in economic journals. In top five economic journals, ten out of 271 articles were RCT studies in 2015 while it was zero in 1990s. What is a RCT? Why is it useful to enhance our knowledge on poverty and poverty eradication? RCT helps us to solve the puzzle that why cannot we achieve even the low-hanging fruits in our fight against poverty.
Worldwide, nearly 20 millions of children are not immunized each year despite of its efficacy to save lives. Is this due to the high opportunity cost associated with the loss of daily wage resulting from visits to vaccination centers? Duflo and Benerjee designed two interventions in Udaipur to answer this question. In intervention A, “monthly vaccination camps” were set up in randomly chosen villages and in intervention B extra incentive of one kilo of lentils were given in randomly chosen villages. The results were compared against a group of comparison villages or a treatment group. They found that regular health camps increase immunization from 6% to 18% while providing lentils increases it up to 39%. The random allocation ensures that any difference between the treatment and the control groups . must be due to the treatments (intervention A and B). The clear identification strategy has made RCT a giant compared to the other empirical approaches that are used to make causal inferences, elevating the status of the RCTs to the “Gold Standard” of evidence .
RCTs are not devoid of criticism. Once, Lant Prichette remarked that “if experiments were the hallmark of science, alchemists would win Nobel prizes”. Severe criticisms are leveled on the randomistas’ claim that their method should be the “gold standard”. The randomistas react with equal severity claiming that alternatives to RCTs are “opinions, prejudices, anecdotes, and weak data”. Though experimental design is useful in generating exogenous variation of the variables of interest, researchers stress the need to acknowledge the unacknowledged costs including violations of basic principles of research ethics, faux exogeneity that compromises even the internal validity of estimates, and the failure to utilize prior knowledge built using other tools. The 2015 laureate Angus Deaton remarks that “randomization does not equalize everything …it does not automatically deliver a precise estimate of the average treatment effect”. The demand from the constructive critics is not an outright denouncing of RCTs but using it in the cumulative program, combined with the other methods which are similarly imperfect, and move beyond investigating “what works” to “why things work”.
While we acknowledge the strength of the RCTs, we should continue to debate on how to improve our knowledge on poverty. The Nobel Prize 2019 catalyzes this debate. Every incremental contribution to our knowledge on poverty will gradually construct the world we want to see one day. As Duflo noted once, “there is not going to be “le grand soir”-one day, the big revolution- and the whole world is suddenly not corrupt. But maybe you create a small little virtuous group here and something else there. All these things are incremental”. I believe that this year’s Nobel is for these invaluable incremental changes.
Asanka Wijesinghe
PhD student at the Ohio State University
The interpretations and conclusions expressed in this blog are those of the individual author and do not necessarily reflect the views or policies of CEPA.
