Working Paper: NBER ID: w25532
Authors: Dmitry Arkhangelsky; Susan Athey; David A. Hirshberg; Guido W. Imbens; Stefan Wager
Abstract: We present a new estimator for causal effects with panel data that builds on insights behind the widely used difference in differences and synthetic control methods. Relative to these methods we find, both theoretically and empirically, that this "synthetic difference in differences" estimator has desirable robustness properties, and that it performs well in settings where the conventional estimators are commonly used in practice. We study the asymptotic behavior of the estimator when the systematic part of the outcome model includes latent unit factors interacted with latent time factors, and we present conditions for consistency and asymptotic normality.
Keywords: No keywords provided
JEL Codes: C01
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| synthetic difference in differences (SDID) (C22) | credible estimates of causal effects (C51) |
| synthetic difference in differences (SDID) (C22) | performance compared to DID (C52) |
| synthetic difference in differences (SDID) (C22) | performance compared to SC (D29) |
| inclusion of time and unit weights (Y20) | robustness of SDID (C52) |
| DID assumptions are suspect (D80) | SDID provides better estimates (C13) |
| synthetic difference in differences (SDID) (C22) | lower root mean squared error compared to DID (C52) |
| synthetic difference in differences (SDID) (C22) | lower root mean squared error compared to SC (C52) |