Association for Public Policy Analysis & Management

Police officers routinely stop and search African Americans and Hispanics more frequently than others. Determining to what extent these discrepancies occur due to racial bias is an ethical and legal imperative. A common approach is to test if the rate of searches that uncover crime, or the `hit-rate', is lower among African Americans and Hispanics compared to whites. However, several economists have cautioned against using hit-rate tests because they rely on the assumption that police aim to maximize arrests, and are not valid if police aim to minimize crime instead. I derive an empirical test to identify police objectives from a model of optimal police spending with racial profiling. A social planner chooses the level of police spending in order to minimize social costs of crime and policing. Police search citizens in order to either minimize crime or maximize arrests. The model yields testable implications for how police expenditures and racial discrepancies in arrest rates vary with racial composition. Data from U.S. cities are consistent with a model of arrest maximization, and inconsistent with a model of crime minimization. This supports the validity of the hit-rate test for police bias.