There is a dearth of research on crime in South Africa, which is particularly problematic in this country given the extraordinary high crime rates reported here. Common correlates of crime, such as unemployment, poverty, and inequality, are also at extreme levels in South Africa – making the investigation of the determinants of crime even more pertinent in this context. We combine published crime statistics with demographic data from the 2011 South African Census Community Profiles to investigate which socio-economic factors attract crime at a police precinct level. In particular, we investigate whether, and to what extent, precinct-level unemployment and income and intra-precinct inequality are related to reported crime rates within a particular precinct. The expectation was that resource-acquisition driven crimes (i.e. property and robbery crimes) would be attracted by high levels of income and inequality in a precinct, and low levels of unemployment. Further, we hypothesised that at some high level of both income and inequality, crime levels would decrease due to individuals beginning to take protective measures against crime. Through a combination of nonparametric and parametric analyses, including an IV regression design, we found support for this protection hypothesis in the case of property crime but not in the case of robbery crime. No socio-economic factors were significantly related to robbery crime in our analyses. We also investigated violent crime, which, due to the interpersonal and psychological nature of such crime, we hypothesised to vary positively with inequality and unemployment and vary negatively with income.
Although we did find positive relationships between violent crime and income, we found that at high levels of precinct-level income, violent crime decreased. We did not detect any relationship between inequality and violent crime, nor between unemployment and any crime type. The interpretation of these findings is that, where significance was found, certain socio-economic factors attract certain types of crime. This is not to say that insignificant findings signal no impact of the indicator on crime, since individual-level driving factors cannot be investigated using precinct-level data.