Spatio-temporal autocorrelations of homicides in municipalities of Mexico from 2011–2019

In this work, spatial autocorrelations of homicides (intentional, unintentional and total) per municipality of Mexico were calculated monthly for the period 2011–2019. Spatial autocorrelation quantifies how much an element is similar to its neighboring elements based on a defined criterion of spatial proximity. To measure autocorrelations of municipalities in terms of their homicide rates per 1000 inhabitants, we employed the local Moran index, an indicator widely used in spatial statistics. For every municipality, the local Moran index was calculated for every month in the period 2011–2019. We identified in each resulting map those municipalities with recurrent in time high-high (AA) local Moran index autocorrelations, that is, municipalities with high homicide rates surrounded by municipalities also with high rates. Municipalities that appear more frequently during the 2011–2019 period, with autocorrelations of intentional and total homicides AA almost in all cases, form clusters among themselves. Intentional and total homicides have a spatial pattern that is much more agglomerated in clusters compared to unintentional homicides. In general, clusters of intentional and total homicides consist of more municipalities than clusters of unintentional homicides. Chihuahua, Nuevo León and Tamaulipas are the three states with the highest number of AA homicide autocorrelations, followed by Durango and Sinaloa. Most of the AA autocorrelation clusters are located in the Golden Triangle region (Chihuahua, Durango and Sinaloa) and in the Nuevo León and Tamaulipas area on the border with the United States of America. It was determined that the homicide phenomenon in Mexico is not randomly distributed throughout the territory.