Understanding the Structural Drivers of Violence in Colombia: A Hierarchical Bayesian Approach

This research project explores the relationship between structural development indicators and security outcomes across Colombian municipalities using a hierarchical Bayesian model. By analyzing data from the Modern Cities Index (ICM) for the period 2010–2022, the model accounts for the nested structure of municipalities within departments and allows both baseline levels of violence and the effects of key predictors to vary by region. While global trends already show heterogeneous results for each predictor, department-level estimates further display detailed regional heterogeneity, underscoring the need for context-sensitive analysis. The findings highlight the utility of Bayesian methods for policy design and suggest that effective violence reduction strategies must be tailored to the specific institutional and historical conditions of each region.