On terror, drugs and racial profiling

We show that for racial profiling (defined as policy rules that employ statistical discrimination based on racial attributes) to be efficient in fighting ordinary crime, it needs to focus on the racial composition of marginal offenders. Efficiency thus may counter-intuitively call for targeting the group with the lower offending rates. In the context of terror, however, it has to be based primarily on differences in offending rates across racial population groups (group-wise averages). We demonstrate that, assuming correlation between race and crime, racial profiling would nearly always be efficient. Finally, we discuss equity considerations and suggest that if awarding compensation is perceived to be a viable policy option, it should be paid on an ex ante basis. (C) 2008 Elsevier Inc. All rights reserved.