Spatial dynamics of short-term crime trends in Mexico city

We analyse short-term spatial crime dynamics in Mexico City across census blocks nested within neighbourhoods. Using generalised additive spatial models, we examine sigma convergence and beta convergence as bounded-period indicators of a narrowing of spatial disparities in crime, rather than as evidence of long-run structural equilibrium. Results show strong sigma convergence for traditional frequency-based indicators, but weaker convergence for harm-based measures. Beta convergence is robust only for crime harm index per victim (CHIV), whereas the remaining indicators display threshold effects, partial convergence and divergence in high-crime places. Compositional and contextual factors significantly shape these dynamics, and the results should be interpreted in light of the COVID-19 shock and the structural instability detected around 2020/2021. Findings highlight the value of multi-indicator, multi-scale, harm-aware and spatially explicit analyses for identifying where spatial disparities in crime are narrowing, where they are not and what kinds of interventions may be required.