One-Sided Violence and Aid in Somalia: Exploring the Spatial Relationship Between AidAllocation and Violence Against Civilians

Sammanfattning

This thesis examines the spatial relationship between foreign aid allocation and one-sided violence in Somalia from 2000 to 2014. While aid is often understood as a tool for reducing suffering in conflict-affected settings, it is also delivered into contested environments where armed actors compete over territory, resources, legitimacy, and civilian cooperation. The study asks how aid allocation is associated with OSV across PRIO-GRID cell-years and whether this association differs between state and non-state perpetrators. Using a quantitative, observational, subnational research design, the analysis combines AidData's Somalia AIMS Geocoded Research Release with UCDP Georeferenced Event Dataset data on OSV and PRIO-GRID covariates. Linear probability models estimate whether previous-year aid exposure is associated with OSV incidence, while PPML count models assess whether aid is associated with the number of OSV events and fatalities. The findings show that aid and OSV are spatially and temporally connected, but they do not support a strong claim that aid increases civilian targeting. Aid-receiving cells are more likely to record at least one OSV event, especially non-state OSV, which partially supports the expectation that aid allocation is associated with local variation in civilian victimization. However, count models do not show that aid increases OSV intensity; in several specifications, the association with non-state OSV counts is negative. The state OSV relationship is weaker and less consistent, while spatial dependence is visible descriptively but not strongly supported statistically. The thesis contributes by studying aid and OSV locally, distinguishing perpetrator type, and avoiding causal overclaiming in Somalia’s conflict settings.