Frontiers in Operations: Fair Funnels: Bias, Performance, and Interventions in Multistage Hiring Processes
Abstract
Problem definition: Many organizations face evidence of discrimination in their hiring and career advancement outcomes: underrepresentation of a group that cannot be explained by quality differences. Representation is generally the result of a funneling process through multiple stages. Therefore, policymakers seeking to intervene (typically by introducing regulations or incentives) must decide which stage to target. Moreover, stakeholders care about both representation and about the quality of hires. How does the multistage funnel structure interact with discriminatory judgment behaviors to impact hire representation and quality? What is the best way to design regulations and incentives within funnel structures to improve representation and performance, and can they improve both? Methodology/results: We develop a stylized mathematical model that captures taste-based and statistical discrimination in a multistage funnel. We show that three distinct discriminatory judgment behaviors work through the funnel in three distinct ways to cause underrepresentation. We then endogenize the funnel’s threshold decisions to study the impact and effective design of regulations and incentives. Whereas common wisdom would suggest targeting regulations and incentives at the stage where underrepresentation appears most severe, our analysis suggests a simple recommendation is surprisingly robust irrespective of where the underrepresentation manifests: regulate the top, subsidize the bottom. We identify several distinct asymmetric dynamics that support this prescription: asymmetry in regulation workarounds, asymmetry in regulation win-win opportunity, and asymmetry in subsidy propagation. Managerial implications: Our results help organizations better understand how discriminatory judgment behaviors affect representation and quality at various stages in the funnel. They help guide policymakers on where to target interventions, and their likely consequences on representation and performance outcomes.
History: This paper has been accepted in the Manufacturing & Service Operations Management Frontiers in Operations Initiative.
Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2025.0053.

