Civil conflicts in Africa: Climate, economic shocks, nighttime lights and spill-over effects

This study examines the relationship between weather conditions, economic shocks and civil conflicts in Africa. While most studies rely on country-level data sets, this study exploits a panel data set of African first-order administrative units covering 1992-2010. Since sub-national gross domestic product for Africa is either unavailable or of poor quality, nighttime light data from satellites is exploited to predict economic growth at the sub-national level. In addition to IV/GMM estimation, the Lasso estimator is employed in order to generate optimal instruments for economic growth from rainfall and temperature variables. It is demonstrated that the Lasso estimator successfully addresses the challenges arising from non-linearities, heterogeneity across climate regions and weak identification. Furthermore, spatial econometric methods account for conflict spill-overs via political, geographical and ethnic ties. Estimation results provide no evidence that economic growth shocks have a significant causal impact on violence, but prices of capital-intensive commodities seem to be associated with civil conflicts.