The opportunity cost of intrastate violence and the out-of-sample validity of commodity price shocks

Abstract

Recent work in economics and political science has identified a positive correlation between commodity prices and violence—which are theoretically linked through an opportunity cost mechanism. The findings have important implications for the academic understanding of violence and for policy makers working to anticipate and prevent it. Despite the importance of the findings, they are based on in-sample validation and aggressively aggregated data. There is a growing understanding, both within this literature and in the broader social science literature, that out-of-sample validation should be considered essential before any policy recommendations are made based on academic research. I find that while coffee price shocks are still correlated with intrastate violence in a higher resolution test (district-month), they fail to be useful out-of-sample. These results should prompt a reevaluation of scholarly work based on the relationship between commodity prices and violence and of any policies that have been crafted based on it.