Systemic Discrimination: Theory and Measurement

Economics often defines and measures discrimination as disparities arising from the direct effect of group identity. We develop new tools to model and measure systemic discrimination, which captures how discriminatory decisions in other domains---past, future, or contemporaneous---contribute to disparities in a given decision. We show that systemic discrimination can be driven by disparate signaling technologies or differential opportunities for skill development. We then propose a new measure based on a decomposition of total discrimination into direct and systemic components, and show how it can be used to estimate systemic discrimination in both experimental and observational data. We illustrate these new tools in three applications, including a novel Iterated Audit experimental paradigm with real hiring managers. The applications also identify behavioral frictions that blunt the impact of individual-level interventions and perpetuate systemic discrimination, suggesting the need for systems-based policy responses to systemic discrimination.