Large weather and conflict effects on internal displacement in Somalia with little evidence of feedback onto conflict

Abstract

Extreme weather and conflict may drive forced displacement. However, their individual contribution to displacement is not fully understood due to challenges around isolating individual channels of causality. Here, we use novel disaggregated data on internal displacement in all of Somalia's subregions from 2016 to 2018 broken down by reported reason of displacement and combine it with weather and conflict data. This allows us to isolate the effects of extreme weather and conflict on forced displacement, as well as the effects of displacement on conflict itself. We find large non-linear effects of weather on displacement where an increase in temperature anomalies from 1 °C to 2 °C (to approx. 1.5 standard deviations, SD) leads to a tenfold increase in displaced people, and a reduction in precipitation from 50 mm to 0 mm (approx. 1.5SD) leads to around a fourfold increase in displacement. We find significant effects of conflict events on displacement (which are masked when the data is aggregated) with a 1.5 standard deviation increase in conflict events increasing displacement 50-fold. We further show that displacement itself has little detectable effect on the occurrence of conflict events.