Résumé
We explore the relationship between drought and civil war. We show that the link between rainfall, temperature and civil war found in the literature may be driven by aggregate shocks (such as global climate) that were not accounted for. A standard differences-in-differences specification relying only on within country variation reveals a much weaker and insignificant link between weather variables and civil war. To increase statistical power, we propose a country-specific measure of drought that describes social exposure to water stress in a more efficient way than rainfall and temperature. We use the Palmer Drought Severity Index (PDSI) which combines rainfall, temperature (and soil texture) and takes important aspects that were missing from previous studies into account: nonlinearities - the effect of contemporaneous rainfall and temperatures depends on the climate history -, interaction effects - e.g. low rainfall is more important in hot years -, and threshold effects due to the limited capacity of the soil - e.g. rainfall water will run off when the soil layers are full. We continue to find a weak positive link between drought and civil war.