2021/11/21 by Oscar Engelbrektson, Engelbrektson, Oscar
Economics, Econometrics and Finance · Mathematics · Medicine · #Advanced Causal Inference Techniques #Econometrics (econ.EM) #Economic Policies and Impacts #FOS: Computer and information sciences #FOS: Economics and business #Hormonal Regulation and Hypertension #Methodology (stat.ME) #Statistical Methods and Inference
paper · pdf · doi:10.48550/arxiv.2111.10784
openalex publication_date 2021/11/21 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
This paper extends the literature on the theoretical properties of synthetic\ncontrols to the case of non-linear generative models, showing that the\nsynthetic control estimator is generally biased in such settings. I derive a\nlower bound for the bias, showing that the only component of it that is\naffected by the choice of synthetic control is the weighted sum of pairwise\ndifferences between the treated unit and the untreated units in the synthetic\ncontrol. To address this bias, I propose a novel synthetic control estimator\nthat allows for a constant difference of the synthetic control to the treated\nunit in the pre-treatment period, and that penalizes the pairwise\ndiscrepancies. Allowing for a constant offset makes the model more flexible,\nthus creating a larger set of potential synthetic controls, and the\npenalization term allows for the selection of the potential solution that will\nminimize bias. I study the properties of this estimator and propose a\ndata-driven process for parameterizing the penalization term.\n