Debate: War and Post-Communist Democratization: Dealing with Old Comrade Endogeneity

Horowitz’s article claims to present strong empirical support for the argument that involvement in war negatively affects democracy. It is argued here that the results based upon regressions for countries of the post-communist area are flawed. As Horowitz denies the existence of a substantial reciprocal effect of democracy on war, he does not address the arising endogeneity problem at all. This results in an overestimation of the negative consequences of war on democracy. Building on the theories of democratic peace, democratization-related conflict and the outbreak of civil wars, it is asserted that countries which are ex ante less democratic in the transition period have a higher probability of war involvement. A reanalysis tackles the occurring endogeneity problem by using change in democracy as the dependent variable. The results provide evidence that the original models suffer from endogeneity bias. The effect of war on the difference in levels of democracy is not significant, and the model performs much worse in predicting the change in democracy rather than the level. One conclusion is that further research on the consequences of war on democracy has to deal both theoretically and methodologically with the reversed causality problem.