We study possible worker-to-employer discrimination manifested via social preferences in an online labor market. Specifically, we ask, do workers exhibit positive social preferences for an out-race employer relative to an otherwise-identical, own-race one? We run a well-powered, model-based experiment wherein we recruit 6,000 workers from Amazon's M-Turk platform for a real-effort task and randomly (and unobtrusively) reveal to them the racial identity of their non-fictitious employer. Strikingly, we find strong evidence of race-based altruism – white workers, even when they do not benefit personally, work relatively harder to generate more income for black employers. Self-declared white Republicans and Independents exhibit significantly more altruism relative to Democrats. Notably, the altruism does not seem to be driven by race-specific beliefs about the income status of the employers. Our results suggest the possibility that pro-social behavior of whites toward blacks, atypical in traditional labor markets, may emerge in the gig economy where associative (dis)taste is naturally muted due to limited social contact.
discrimination worker-to-employer social preferences taste-based discrimination Gig Economy mechanical turk Structural Behavioral Economics