Working Paper: CEPR ID: DP13659
Authors: Markus Eberhardt
Abstract: I revisit the causal relationship between democracy and growth as recently studied in Acemoglu, Naidu, Restrepo, and Robinson (2019, ANRR). I demonstrate the sensitivity of their results to sample selection by dropping a small number of observations in a non-random fashion and use these findings to motivate a generalisation of their empirical approach. My own analysis relaxes the assumption of (i) a common democracy-growth relationship, and of (ii) the absence of strong cross-section correlation. Adopting novel methods for policy evaluation I find a robust positive long-run effect of democracy albeit with only around half the magnitude of that found in ANRR.
Keywords: democracy; growth; political development; spillovers; difference-in-difference estimator; interactive fixed effects
JEL Codes: O10; P16
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| sample selection (C34) | (democracy -> economic growth) (O00) |
| omitted observations (C20) | (democracy -> economic growth) (O00) |
| democracy (D72) | per capita GDP (E20) |
| democracy (D72) | economic growth (O49) |