Drought impacts on armed conflict primarily explained by pre-existing conflict risk

Abstract Although extensive research has examined climate-conflict relationships, systematic analysis of how climate impacts heterogeneously affect armed conflict dynamics remains limited. Without understanding these heterogeneous effects, our ability to understand and respond to climate-related armed conflict risk remains incomplete. This paper systematically investigates the heterogeneous causal effects of droughts on both conflict incidence (extensive margin) and conflict intensity (intensive margin) across 8 world regions, 4 temporal and spatial resolutions, and 3 conflict types, reporting only patterns that hold consistently across specifications. Using a causal forest and a calibration test for heterogeneity reveals some evidence that droughts heterogeneously affect conflict incidence. Zero inflated negative binomial models that control for grid cell and yearly effects and include drought interactions with socioeconomic, political, and demographic variables show significant evidence that droughts heterogeneously affect conflict intensity. When examining individual interaction terms, baseline conflict risk-estimated using a random forest with the same socioeconomic, political, and demographic predictors plus conflict history-emerges as the only significant source of heterogeneity that is consistent across specifications. This suggests that other variables primarily influence drought–conflict relationships through their contribution to overall conflict risk, rather than through independent causal channels. The results demonstrate that even when droughts do affect conflict intensity heterogeneously, their overall role in armed conflict dynamics remains limited.