🤖 AI Summary
This paper addresses the problem of predicting the distribution of individual treatment effects (ITE), moving beyond the conventional focus on average treatment effects (ATE). We propose a covariate-adjusted counterfactual prediction framework that integrates quantile inference with finite-sample distribution theory, yielding the first statistically valid distributional inference method for ITE under weak assumptions—applicable to real-world settings such as randomized controlled trials (RCTs). Our theoretical contribution reveals a critical insight: even when ATE is statistically insignificant, ITE may exhibit strong heterogeneity—some individuals experience substantial benefits while others suffer significant harm. Empirical analysis across five microcredit RCTs demonstrates this phenomenon: the 10th percentile of income change is −12.5%, whereas the 90th percentile is +13.6%, confirming the presence of substantial, directionally opposing heterogeneous treatment effects.
📝 Abstract
Important questions for impact evaluation require knowledge not only of average effects, but of the distribution of treatment effects. What proportion of people are harmed? Does a policy help many by a little? Or a few by a lot? The inability to observe individual counterfactuals makes these empirical questions challenging. I propose an approach to inference on points of the distribution of treatment effects by incorporating predicted counterfactuals through covariate adjustment. I show that finite-sample inference is valid under weak assumptions, for example, when data come from a Randomized Controlled Trial (RCT), and that large-sample inference is asymptotically exact under suitable conditions. Finally, I revisit five RCTs in microcredit where average effects are not statistically significant and find evidence of both positive and negative treatment effects in household income. On average across studies, at least 13.6% of households benefited, and 12.5% were negatively affected.