Boese, VanessaEberhardt, Markus2021-02-162021-02-162021-02http://hdl.handle.net/2077/67727We motivate and empirically analyse the idea that democratic regime change is not a dis-crete event but a two-stage process: in the first stage, autocracies enter into an ‘episode’ of political liberalization which can last for years or even decades; in the second stage, the ultimate outcome of the episode manifests itself and a nation undergoes regime change or not. Failure to account for this chronology risks biased estimates of the economic e˙ects of democratic regime change since this ignores the relevance of the counterfactual group in which liberalisation did not culminate in a democratic transition. Using novel Varieties of Democracy (V-Dem) data on Episodes of Regime Transformation (ERT) for a large sample of countries from 1950 to 2014 we study this phenomenon in a repeated-treatment di˙erence-in-di˙erence framework which accounts for non-parallel pre-treatment trends and selection into treatment. Our findings suggest that a single event approach signifi-cantly underestimates the economic benefits from lasting democratic regime change.engDemocracy, Growth, Political Development, Difference-in-Difference, Inter- active Fixed EffectsText