Abstract:
This paper investigates the relationship between socio-economic inequalities of various dimensions and crime at the local level in South Africa. It uses a novel panel dataset of police precincts that combines crime records from the South African Police Service with socio-economic data from two population censuses and household surveys. The identification exploits variation of inequality and crime across time and space, controlling for socio-economic and demographic characteristics of police precincts, province-specific time trends, police cluster-fixed effects, and spatial correlations. I find robust evidence for a significant, positive, and linear relationship between income inequality within precincts and local rates of violent crime, and an inverted u-shaped relationship with property crime. Education and housing inequality and racial heterogeneity are also positively correlated with crime. While inter-racial inequality contributes more to property crime, intra-racial inequality contributes more to violent crime. Richer precincts display higher property crime rates as compared to their poorer neighbors.
