ASC Workshop: Synthetic Control Time Series Experiments: The Case-Study Approach to Causal Inference

Session Submission Type: Workshop

Abstract/Description

The Synthetic Control Method is an increasingly popular approach to quasi-experimental causal inference and policy evaluation. The method involves the construction of a control time
series which optimally mimics the characteristics of the treated series up to the point of the
intervention as a weighted combination of less-than-ideal, but uncontaminated “donor pool”
units. Because the synthetic control is constructed from a set of uncontaminated controls, the post intervention synthetic series is intended to approximate the treated series "had the
intervention never occurred". After briefly situating the method within causal inference and
quasi-experimental literatures, the workshop will walk participants through real-world
applications of the method, including replications of peer-reviewed synthetic control studies.
The replications and illustrations will familiarize participants with the process and
implementation of the synthetic control routine, from data cleaning and setup through post-estimation procedures. Examples will also highlight common pitfalls and researcher checks that are essential to valid inference. The examples presented in this workshop are derived from a book project on synthetic control designs currently out for review.

All necessary data, .ado, and .do files will be provided by the instructor prior to the workshop.

*Participants will need to bring laptops with Stata installed to follow along with the examples presented.

Sub Unit

• Area XVII. Workshops / Quantitative Methods

Instructor

• Bradley Bartos, University of Arizona, School of Government and Public Policy