An Analysis of the Factors Determining Crime in England and Wales: A Quantile Regression Approach

We examine how socio-economic and police enforcement variables affect property and violent crimes at different points of the crime distribution in England and Wales over the period 1992-2007. By using data from 43 police force areas, we examine how the effect of real earnings, unemployment, crime detection rate, income inequality and proportion of young people varies across high and low crime areas. Six crime categories are examined - burglarly, theft and handling, fraud and forgery, violence against the person, robbery, and sexual assault. Using a quantile regression model, we find that there are statistically significant differences in the impact of explanatory variables on various crime rates for low and high crime areas. For example, not only does unemployment increase crime but it does so more in high crime areas. Higher detection rates reduce crime rates and the effect is stronger in low crime areas. There are also differences in distributional impact on crime rates for real earnings, income inequality and proportion of young people. Thus, our work points to the need to look beyond the usual mean effects of policing and socio-economic factors on crime and consider their impact on the entire distribution of crime rates. This will enable us to tailor policies that are particularly effective at different points in the crime distribution. Further, given the differential impact of earnings and unemployment across high and low crime areas this provides insight into why paradoxically recessions may have no impact on crime or even lower it.