Abstract
Abstract
This paper examines whether democratisation has a robust average effect on aggregate territorial carbon emissions. Using a country-year dataset covering 183 countries over 1960–2010, we compare countries that experience democratic transitions or reversals with those that remain non-democratic throughout the period. The empirical strategy combines heterogeneous difference-in-differences with mean-group estimation and constructs common proxies from never-democracies to account for non-parallel emissions trends, global shocks, and country-specific responses. The preferred static specification produces a positive estimate of 0.069 log points, while the corresponding dynamic estimates are much smaller and mostly statistically insignificant. Adding income and trade, which may operate as post-transition channels, reduces the static estimate to 0.007. A cross-validated interactive fixed-effects counterfactual estimator that conditions on income and trade also produces an imprecise average treatment effect, with no statistically significant event-time effect during the first 20 post-transition years. The point estimates differ by pre-treatment state capacity in the direction predicted by the theoretical framework, but the between-group difference is imprecisely estimated. Historical estimates using V-Dem over 1789–2015 are similarly sensitive to specification, while country-specific estimates display substantial heterogeneity. Overall, democratisation has no robust average effect on aggregate emissions across specifications, countries, and time horizons.