aryanpandey742/spatial-conflict-africa

# Spatial Analysis of Ethnic Partition and Conflict in Africa Testing whether pre-colonial ethnic partitioning across modern national borders, historical conflict exposure, and pre-colonial political instability predict contemporary conflict — using a ~2,500-cell geospatial panel of the African continent and spatially-corrected inference. **Stack:** Python (`GeoPandas`, `rasterio`, `rasterstats`, `scikit-image`, `SciPy`) for spatial data construction · Stata (`reghdfe`, `acreg`) for Conley spatially-corrected regressions ## TL;DR Colonial borders famously split many African ethnic groups across two or more modern states. I build a fine-grained spatial panel combining ethnic-boundary, conflict, terrain, climate, and historical data to test whether that partitioning — and the deeper history of conflict and political instability in a region — still predicts where conflict happens today. The core results hold up even after country fixed effects and Conley spatial correction: **partitioned ethnic groups, regions with more historical conflict exposure, and regions with a more unstable pre-colonial political history all see significantly more conflict today.** ## Research Question Does the colonial-era partitioning of ethnic homelands across modern borders, and the deeper historical exposure to conflict and political instability in a region, help explain the geographic distribution of conflict in Africa today? This combines themes from the ethnic partition literature (e.g., Michalopoulos & Papaioannou) with a historical-persistence approach to conflict. ## Data | Dataset | What it provides | |---|---| | Murdock ethnic boundary map (`borders_tribes.shp`) | Pre-colonial ethnic homeland polygons | | UCDP Georeferenced Event Dataset, v25.1 (`GEDEvent_v25_1.csv`) | Geolocated conflict events and civilian deaths, 1989–2024 | | Extended Ethnographic Atlas (`Extended_Atlas.xlsx`) | Ethnic-group centroids and pre-colonial political centralization (jurisdictional hierarchy, …