Working Paper: NBER ID: w24814
Authors: Dmitry Arkhangelsky; Guido Imbens
Abstract: We develop a new approach for estimating average treatment effects in the observational studies with unobserved cluster-level heterogeneity. The previous approach relied heavily on linear fixed effect specifications that severely limit the heterogeneity between clusters. These methods imply that linearly adjusting for differences between clusters in average covariate values addresses all concerns with cross-cluster comparisons. Instead, we consider an exponential family structure on the within-cluster distribution of covariates and treatments that implies that a low-dimensional sufficient statistic can summarize the empirical distribution, where this sufficient statistic may include functions of the data beyond average covariate values. Then we use modern causal inference methods to construct flexible and robust estimators.
Keywords: average treatment effects; observational studies; unobserved heterogeneity; fixed effect models; propensity score
JEL Codes: C1; C21; C23; C31
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Traditional fixed effect regression models (C23) | biased estimates of causal effects (C51) |
| Proposed alternative method (C59) | better capture true causal relationships between treatment and outcomes (C32) |
| Modeling conditional assignment probabilities (C25) | relax functional form assumptions on potential outcome distributions (C51) |
| Proposed estimator (C51) | more nuanced understanding of treatment effects across clusters (C32) |