The Unassuming Economist

Source: Ideas for India

In a recent column for the Ideas for India blog, development economist Jean Dreze writes about the perils of experimental policymaking. While data-based policy design is quite the rage now with randomized control trials (RCTs) being used to gather evidence on “what works” and then scaling up whatever does, Dreze writes how a more comprehensive approach to policymaking would be one where insights from data are interspersed with a sound mix of understanding the issues, value judgments, and deliberation on inclusivity.

He makes a case for this argument by discussing an experiment conducted in the Indian state of Bihar during 2012-13 to study a new financial management system in the Mahatma Gandhi National Rural Employment Guarantee Act (Banerjee, Duflo, Imbert, Mathew, and Pande (2020). From delivering counter-productive results in the form of reducing, not enhancing, the baseline expenditure on this scheme to delays in workers’ payments and the lack of significant differences between outputs of the treatment and control groups, this article draws attention to the challenges associated with hasty rollouts of interventions and subsequent conclusions based on them without negating the learnings. It concludes with some best practices for engaging with governments, conducting experiments at scale, and ensuring that the “do no harm” principle of RCTs stays put.