A Replication of "Comparative Politics and the Synthetic Control Method" by Abadie et al. (2015)

Abadie, Diamond, and Hainmueller's (2015) analysis of the costs of German reunification is a flagship study for the "synthetic control method" (SCM). Yet three issues can be identified. First, Abadie et al. remove countries that performed poorly in the 1990s from their donor pool, potentially biasing their results to amplify the negative impact of reunification on West Germany. Second, Abadie et al. use nominal GDP per capita to measure growth but report using real GDP per capita in 2002 USD. When the reported series is used, their results become more fragile: they remain statistically significant at p ﹤ 0.10, but the results of other robustness tests that Abadie et al. include are weaker. Third, even those fragile results reflect an arbitrary choice of reference year, given the instability that results from the spatial Gerschenkron effect: when Abadie et al.'s study is replicated using real GDP per capita series that are re-referenced using every year from 1960 to 2003, only 8 out of 44 synthetic controls generate treatment effects with p ﹤ 0.10.