Theory and empirics are ambiguous on the effect of democracy on growth. Cross-country studies find that democracy has no significant impact on growth. In contrast, within-country studies find a strong positive effect of transition to democracy. We reconcile this inconsistency by showing that the positive effect of political transition is a result of swift regime change and not democratization. We identify and examine 90 successful, failed, and gradual transitions that have occurred over the last half century. This new classification permits us to compare successful episodes of democratization with unsuccessful ones -- as opposed to with the counterfactual of no transition. We find that both successful and failed transitions boost long-run growth by about one percentage point, but gradual change is quite costly in economic terms. The results imply that the growth dividend from political transition is a result of regime change and not democratization, and also offer new evidence on the importance of the speed of transition for economic growth. The results are robust to a number of alternative specifications, to stricter and more lenient definitions of democratic transition, and to including reverse transitions.
