Spatial analysis of homicidies on Northeast in 2010 and 2019

Abstract

We first looked for spatial correlation of lethal violence in Brazil’s Northeastern municipalities in 2010 and 2019 by conducting an exploratory spatial data analysis (ESDA), which showed clusters of homicides in both years, specifically 106 high-high clusters in 2010 and 84 in 2019, and 18 low-low clusters in 2019. Clusters of high homicide rates were more common in the states of Bahia (27 and 16 in 2010 and 2019, respectively), Piaui (21 and 29) and Paraiba (20 and 10); Paraiba had also the greatest number of low homicide clusters in 2019 (7). We then tested spatial econometrics models such as SAR (Spatial autoregressive), SAC (Spatial Autocorrelation) and SEM (Spatial Errors Model), including sociodemographic variables relevant in determining violence indexes. Those estimations showed there is a positive relation between wealth inequality, population density, urbanization, employment and homicide rates, whereas earnings and the proportion of young men are negatively associated.