On the observational implications of taste-based discrimination in racial profiling

This paper contributes to a growing literature that attempts to determine whether disparities in police stops and searches of potential criminals of different races stem from taste-based discrimination. The key challenge in making this evaluation is that police officers have more information than the econometrician and thus racial disparities in police behavior may result from these unobservable factors rather than discrimination. We develop a general equilibrium model of police and potential criminal behavior that encompasses key models in the literature. We highlight the assumptions needed for existing methods of detecting racial discrimination to hold. In particular, we show that when there are increasing costs to search, existing tests for discrimination can give incorrect results. Given the potential importance of these costs, we then propose some alternate methods for detecting racial bias in police behavior.