Platforms and Prejudice: The Cycle of Racial Disparities on a Labor Platform

Abstract

We examine a multistage evaluation process to understand the conditions under which racial minorities are at risk for disparate work outcomes using data from a platform-based skilled labor market for small business entrepreneurs. Analyzing various evaluative outcomes including job cancellations, ratings, and job allocation for nearly 100,000 jobs, we find that workers perceived as racial minorities (relative to white) receive significantly more cancellations, as well as more one-star ratings and fewer five-star ratings, even among top performers. Factors that reduce uncertainty, including worker quality, and both worker and customer experience, help attenuate the disparity in cancellations but not in ratings. Further, we demonstrate the cyclical nature of bias entrenchment and the interplay of bias with formal practices—ratings and cancellations are mutually reinforcing, and feed back into priority algorithmic work assignments. Our research illustrates the critical role that uncertainty plays in magnifying bias against racial minorities in labor market contexts and that the mechanism driving bias is a function of the evaluation stage. We contribute to the sociological literature on bias in labor market processes, status and inequality in organizations, and work precarity.