Predictive Counterfactuals for Treatment Effect Heterogeneity in Event Studies with Staggered Adoption

Abstract

This paper introduces an approach for estimation of treatment effect heterogeneity in event studies with staggered adoption. Traditional impact evaluation methods can be near-term biased in those settings. The proposed approach attenuates biases and recovers heterogeneity more efficiently than traditional methods. It is shown that machine learning algorithms can be used to accurately predict counterfactuals, which can then be used to estimate a distribution of treatment effects. Simulations demonstrate how that approach can be accurate and efficient, even in the presence of dynamic treatment effects. The paper concludes with an application to a large energy efficiency program in the US.

Keywords: Causal Inference, Machine Learning, Event Studies, Energy Efficiency

JEL Classification: C18, C55, Q49