Crime under-reporting in Bogota: a spatial panel model with fixed effects

We examine spatial spillovers in violent crime and its under-reporting in Bogota, Colombia, using a cuadrante (quadrant) level data. To model spatial spillovers, we use a spatial panel model with fixed effects, and to address under-reporting, we use the stochastic frontier approach as a tool. The novel statistical approach is combined with a database of police-reported crimes in Bogota to examine how influential surrounding areas with high criminal offenses are on crime (under)reporting. The results suggest that spatial correlations are highly significant and that under-reporting is mainly related to interactions with other localities, which have important public policy implications.