Drought and Conflict Dynamics in Africa: A Disaggregated Analysis of Short- and Long Term Effects

Abstract

This thesis examines the relationship between drought and conflict in Africa using subnational panel data for 51 countries from 1996 to 2024. It combines drought data from the Standardised Precipitation Evapotranspiration Index (SPEI) with conflict event data from the Armed Conflict Location and Event Data Project (ACLED) and estimate a disaggregated model on different types of conflict at the ADM1 province level. The results indicate that drought does not have an immediate effect on overall con flict incidence, but that prolonged exposure, particularly over two to four years, increases conflict risk. The effects vary by conflict type. Lower-intensity forms of social unrest, particularly protests and to a lesser extent riots, respond quickly to drought conditions, while organised armed violence, such as battles, shows no imme diate effect and may even decline in the short run, suggesting temporary constraints on mobilisation, logistics, and resources. The heterogeneity analysis indicates that the effects are concentrated in regions dependent on rainfed agriculture, whereas local economic development, proxied by night-time light intensity, plays a weaker role. These findings are consistent with a threat multiplier interpretation, whereby drought amplifies existing socio-economic vulnerabilities rather than acting as a direct trigger of violence. Overall, the results suggest that drought primarily shapes the timing and form of conflict rather than uniformly increasing conflict incidence, highlighting the import ance of distinguishing between different types of violence in the climate–conflict analysis