Primary Submission Category: Synthetic Control Method
Identification and Estimation of Casual Effects with Synthetic Controls in the Presence of Interference
Authors: Wang Miao,
Presenting Author: Wang Miao*
Synthetic control methods are increasingly popular for evaluating the causal effect of a treatment
taking place on only one unit while no perfect control units are available.
Synthetic control methods leverages a set of imperfect control units that are not affected by the treatment to predict the potential outcome of the treated unit had the treatment did not occur, and then estimates the treatment effect.
While treatment effects on the control units may also be present in many applications,
most current synthetic control methods do not admit such interference effects.
In this paper, we propose a novel synthetic control method that allows for interference.
We establish the identification, estimation, and inference for both the treatment effect and interference effects.
Our approach requires the number of interfered units to not exceed half of the total number of units, which is plausible in certain situations including several well studied examples.
However, we do not require exact prior knowledge on the interference structure.
Robust regression is implemented to produce a consistent estimator and a reasonable inferential procedure.
We illustrate with simulations and an application to evaluating the effect of relocating the US embassy on the number of conflicts in the Middle East,
with comparisons to competing methods.