Abstract:We aim to evaluate the causal impact of an intervention when time series data on a single treated unit and multiple untreated units are available, in preand posttreatment periods. In their seminal work, Abadie & Gardeazabal (2003) and Abadie et al. (2010) proposed a synthetic control (SC) method as an approach to relax the parallel trend assumption on which difference-in-differences methods typically rely upon. The term “synthetic control” refers to a weighted average of control units built to match the treated unit’s pre-treatment outcome trajectory, such that the SC’s post-treatment outcome predicts the treated unit’s unobserved potential outcome under no treatment. The treatment effect is then estimated as the difference in post-treatment outcomes between the treated unit and the SC. A common practice to estimate the weights is to regress the pre-treatment outcome process of the treated unit on that of control units using ordinary or weighted (constrained) least squares. However, it has been established that these estimators can fail to be consistent under standard time series asymptotic regimes. Furthermore, inferences with synthetic controls are typically performed by placebo test which lacks formal justification. In this paper, we introduce a proximal causal inference framework for the synthetic control approach and formalize identification and inference for both the synthetic control weights and the treatment effect on the treated unit. We further extend the traditional linear interactive fixed effects model to accommodate more general nonlinear models allowing for binary and count outcomes which are currently under-studied in the synthetic control literature. We illustrate our proposed methods with simulation studies and an application to the evaluation of the 1990 German Reunification. K eywords: consistency; inference; panel data; proximal causal inference; synthetic control. 1 ar X iv :2 10 8. 13 93 5v 1 [ st at .M E ] 3 1 A ug 2 02 1

A Way to Synthetic Triple Difference

Decomposing Triple-Differences Regression under Staggered Adoption

Synthesized difference in differences

Generalized Synthetic Control Method: Causal Inference with Interactive Fixed Effects Models

Differentially Private Synthetic Control

Non-linear Triple Changes Estimator for Targeted Policies

An Effective Treatment Approach to Difference-in-Differences with General Treatment Patterns

Difference-in-Differences Meets Tree-based Methods: Heterogeneous Treatment Effects Estimation with Unmeasured Confounding.

Synthetic controls with staggered adoption

On Proximal Causal Inference With Synthetic Controls

Difference-in-Differences for Policy Evaluation

The Augmented Synthetic Control Method

What's Trending in Difference-in-Differences? A Synthesis of the Recent Econometrics Literature

Synthetic Controls for Experimental Design

Sequential Synthetic Difference in Differences

Continuous difference-in-differences with double/debiased machine learning

Parallel Trends and Dynamic Choices

A Bayesian Alternative to Synthetic Control for Comparative Case Studies

Constructing Synthetic Treatment Groups without the Mean Exchangeability Assumption

Eliminating Systematic Bias from Difference-in-Differences Design: A Permutational Detrending Strategy

A Semiparametric Instrumented Difference-in-Differences Approach to Policy Learning