Estimating Individual Causal Effects

Abstract

The following papers in progress are the chapters of my dissertation. My dissertation argues that the focus on estimating average treatment effects in most empirical work is both misleading because average effects rarely ever apply to any individual in particular and hides treatment effect heterogeneity that has important implications for both theoretical and practical purposes. I develop a framework for estimating individual causal effects using matching methods and a Bayesian model. I then test the performance of my model via simulation, demonstrate the use of the framework in various empirical applications, and discuss how to adapt and extend the framework in various ways.

Publication
Harvard University
Date
Links
  1. A Framework for Estimating Individual Causal Effects (ICEs)
  2. A Simulation Study (web appendix)
  3. Estimating ICEs in Two Applications
  4. Concluding Remarks and Extensions