Democracy, growth, heterogeneity, and robustness

I motivate and empirically investigate differential long-run growth effects of democratisation across countries. While the existing literature recognises the potential for such heterogeneity, empirical implementations to date unanimously assume a common democracy-growth nexus across countries. Adopting novel methods for causal inference in policy evaluation I relax this assumption to confirm that in the long-run democracy has a positive average effect on per capita income of around 10%, adopting a range of alternative definitions for regime change in the form of binary indicators. Guided by existing hypotheses, additional analysis probes the patterns of the heterogeneous 'democratic dividend' across countries. A second common feature of this literature as well as cross-country growth empirics more generally is the absence of concerns for sample selection or influential observations. I carry out two rule-based robustness exercises to demonstrate that my empirical findings are highly robust to substantial changes to the sample.