The spatial pattern of crime in Minas Gerais: an exploratory analysis

Abstract

This paper is aimed at examining spatial pattern of crime in the state otMinas Gerais, Brazil. In methodological terms, this paper uses exploratory spatial data analysis (ESDA) to study the distribution of crime rates in more than 750 municipalities of this state for 1995. The findings reveal that crime rates are distributed non-randomly, suggesting positive spatial autocorrelation. Moreover, spatial heterogeneity represented by both the positive local spatial association (spatial regimes of HH and LL municipalities) and the negative local spatial association (spatial regimes of HL and LH municipalities) is identified. Only one spatial outlier ofHH type is detected, but it does not have any influential impact on the level ofspatial autocorrelation, measured by the Moran s 1. Furthermore, a simple spatial regimes model with inertial effects, but without any causal covariates, has an explanatory power of about 70% of the variation in the crime data. In this regard, the inertial effect is a very relevant aspect ofthe crime in Minas Gerais. The pattern ofspatial distribution revealed through ESDA provides an empirical and solid foundation for the further econometric specification ofmultivariate models.

Keywords

• exploratory spatial data analysis (ESDA)
• crime rates
• spatial regimes
• inertial effects
• spatial autocorrelation