Semiparametric Bayesian Difference-in-Differences

TL;DR

This paper introduces two novel Bayesian methods for difference-in-differences analysis, providing semiparametric inference with proven frequentist validity and demonstrating strong finite-sample performance.

Contribution

The paper develops two new Bayesian approaches for DiD with theoretical guarantees and extends the framework to staggered and repeated cross-sectional designs.

Findings

Bayesian methods outperform frequentist counterparts in simulations.

Proven semiparametric Bernstein-von Mises theorems for the proposed methods.

Effective in empirical applications with strong finite-sample results.

Abstract

This paper studies semiparametric Bayesian inference for the average treatment effect on the treated (ATT) within the difference-in-differences (DiD) research design. We propose two new Bayesian methods with frequentist validity. The first one places a standard Gaussian process prior on the conditional mean function of the control group. The second method is a double robust Bayesian procedure that adjusts the prior distribution of the conditional mean function and subsequently corrects the posterior distribution of the resulting ATT. We prove new semiparametric Bernstein-von Mises (BvM) theorems for both proposals. Monte Carlo simulations and an empirical application demonstrate that the proposed Bayesian DiD methods exhibit strong finite-sample performance compared to existing frequentist methods. We also present extensions of the canonical DiD approach, incorporating both the staggered design and the repeated cross-sectional design.