State-dependent local projections– the dynamic effects of regime transitions

This article proposes using state-dependent local projections (SDLP) to forecast the dynamically asymmetrical effects of political regimes or policy transitions. By conditioning on the ex ante status, the SDLP method estimates first-differenced local projections that depend on the previous regime type, allowing for asymmetric regime transitions and reversals. Monte Carlo simulation results demonstrate that the SDLP estimator involves a smaller bias than existing local projections and performs well with small samples. We apply the proposed SDLP method to examine the effects of democratization and autocratization on economic growth. Our empirical results suggest that the accumulated impacts of democratization on economic growth are positive and increase over time, while the negative effects of autocratization are more substantial and increase over time. These findings suggest that democracy promotes economic growth, whereas reversals to autocracy have an even greater detrimental impact on growth.