Synthetic Control Method with Pre-Trends: the Case of Augusto Pinochet’s Chile

This paper studies the sensitivity of the synthetic control method estimates to the existence of “pre-trends,” i.e. situations when the synthetic control is unable to replicate the trend of the outcome variable of the treated unit in the pre-intervention period. Using Pinochet’s Chile as an example, we show that the paths of the synthetic control
and treated unit diverge before the “intervention” begins. We further confirm this by backdating the intervention. As a result, the synthetic control method estimates for Chile are highly sensitive to the choice of the donor pool and the pre-treatment period. Our results indicate that in settings where pre-trends may be present, the synthetic control method should be used with caution.