The Augmented Synthetic Control Method

The synthetic control method (SCM) is a popular approach for estimating the impact of treatment on single unit in panel data settings. “synthetic control” weighted average units that balances treated unit’s pretreatment outcomes and other covariates as closely possible. A critical feature original proposal to use SCM only when fit excellent. We propose Augmented an extension settings where such infeasible. Analogous bias correction inexact matching, augmented uses outcome model estimate due imperfect then de-biases estimate. Our main proposal, which ridge regression model, directly controls while minimizing extrapolation from convex hull. This estimator can also be expressed solution modified problem allows negative weights some donor units. bound estimation error this under different data-generating processes, including linear factor show how regularization helps avoid over-fitting noise. demonstrate gains with extensive simulation studies apply framework 2012 Kansas tax cuts economic growth. implement proposed new augsynth R package.