Joint with Andrew T. Little.
We analyze a formal model of policing in which an officer decides how to allocate his time interacting with members of two social groups, both while policing and in his non-work social life. Our major contribution is to relax the standard assumption of rational choice models that officers should form “correct” beliefs about who commits more crimes. Instead we start with the intuitive premise that the officer is a limited Bayesian and does not fully distinguish among his experiences of crime at work and in his social life. We show that the officer’s policing decisions end up being influenced by how he spends his time “off the clock” even though we make the stark assumption that his social life should have no actual impact on the facts on the ground while he’s at work (and vice versa). This occurs because he endogenously forms exaggerated beliefs about the relative prevalence of crime among members of the group with whom he spends less time off the clock. As a result, he overpolices that group. Moreover, his overpolicing of one group causes him to spend less time socializing with members of that group, reinforcing his discriminatory policing.
Selected Presentations: APSA Conference (September 2020, Virtual)