The Nature, Detection, and Avoidance of Harmful Discrimination in Criminal Justice

Abstract

We provide a selective survey of the literature on discrimination by criminal justice agents, and argue for a taxonomy of harms that differs from conventional approaches. Discrimination can be self-defeating if it reduces welfare among targets of discrimination while serving no legitimate purpose for the discriminating party. Even if a legitimate purpose is served, discrimination can be deliberative or demeaning, resulting in welfare losses that need to be accounted for. Deliberative and demeaning discrimination can also be self-defeating, through general equilibrium effects on witness cooperation, clearance rates, and preemptive and retaliatory violence. We consider how harmful discrimination can be detected and avoided, as well as changes in our understanding of discrimination resulting from the proliferation of predictive algorithms.

Keywords: Discrimination, Policing, Criminal Justice, Algorithms

JEL Classification: J15, J18, K40