Contamination Bias in Linear Regressions

Series

Econometrics Seminars and Workshop Series
Speaker(s)

Michal Kolesár (Princeton University, United States)
Field

Econometrics

Location

University of Amsterdam, Room E5.22 Amsterdam
Amsterdam
Date and time

March 10, 2023
12:30 - 13:30

Abstract

We study regressions with multiple treatments and a set of controls that is flexible enough to purge omitted variable bias. We show these regressions generally fail to estimate convex averages of heterogeneous treatment effects; instead, estimates of each treatment's effect are contaminated by non-convex averages of the effects of other treatments. We discuss three estimation approaches that avoid such contamination bias, including a new estimator of efficiently weighted average effects. We find minimal bias in a re-analysis of Project STAR, due to idiosyncratic effect heterogeneity. But sizeable contamination bias arises when effect heterogeneity becomes correlated with treatment propensity scores.

Joint work with Paul Goldsmith-Pinkham and Peter Hull

Link to article