An Instrumental Variable Tree Approach for Detecting Heterogeneous Treatment Effects in Observational Studies

24 Pages Posted: 2 Oct 2017 Last revised: 25 Oct 2018

See all articles by Guihua Wang

Guihua Wang

University of Texas at Dallas - Naveen Jindal School of Management

Jun Li

University of Michigan, Stephen M. Ross School of Business

Wallace J. Hopp

University of Michigan, Stephen M. Ross School of Business

Date Written: May 15, 2018

Abstract

We develop a technique that incorporates the instrumental variable method into a causal tree to correct for potential endogeneity biases in heterogeneous treatment effect analysis using observational studies. The resulting instrumental variable tree approach partitions subjects into subgroups with similar treatment effects within subgroups and different treatment effects across subgroups. The estimated treatment effects are asymptotically consistent under very general assumptions. Using simulated data, we show that our approach has better coverage rates and smaller mean-squared errors than the conventional causal tree, and that a forest constructed using instrumental variable trees has better accuracy and interpretability than the generalized random forest.

Keywords: Heterogeneous treatment effect, big data analytics, machine learning, causal inference

Suggested Citation

Wang, Guihua and Li, Jun and Hopp, Wallace J., An Instrumental Variable Tree Approach for Detecting Heterogeneous Treatment Effects in Observational Studies (May 15, 2018). Ross School of Business Paper. Available at SSRN: https://ssrn.com/abstract=3045327 or http://dx.doi.org/10.2139/ssrn.3045327

Guihua Wang

University of Texas at Dallas - Naveen Jindal School of Management ( email )

P.O. Box 830688
Richardson, TX 75083-0688
United States

Jun Li (Contact Author)

University of Michigan, Stephen M. Ross School of Business ( email )

701 Tappan Street
Ann Arbor, MI MI 48109
United States

Wallace J. Hopp

University of Michigan, Stephen M. Ross School of Business ( email )

701 Tappan Street
Ann Arbor, MI MI 48109
United States

Register to save articles to
your library

Register

Paper statistics

Downloads
369
Abstract Views
1,292
rank
80,717
PlumX Metrics