Harvard Business School study finds Airbnb hosts accepted guests with stereotypically black names 42% of the time versus 50% for white names
Study Finds Racial Discrimination by Airbnb Hosts — Renters with names that seemed African American had a harder time booking reservations …
Context & Ripple Effects
The Harvard Business School study sent matched booking requests to Airbnb hosts and found guests with stereotypically African American names accepted at a rate of 42% versus 50% for white-sounding names — an eight-point gap on identical requests. It landed as peer-reviewed evidence behind what Fusion had already been reporting: Airbnb publicly struggling against racial discrimination on its own user-to-user marketplace.
The finding put a platform whose product IS host discretion under direct scrutiny, and the parallel was immediate — months earlier, an MIT co-authored study found [[a:876921|Uber drivers in Boston canceled rides for men with black-sounding names more than twice as often]]. Within roughly sixteen months of publication, Airbnb moved from denial to signing a California agreement letting regulators audit hosts for discriminatory behavior.
First-order effects
- Airbnb faces immediate pressure to redesign its booking flow — the study's design (identical profiles, name-only difference) makes host-side discretion itself the identified defect, not fraud or pricing.
Second-order effects
- Regulators gain a template: the California audit-rights deal shows states can extract monitoring concessions from marketplaces once a controlled experiment documents discrimination rates, and other platforms with discretionary sellers face the same playbook.
Third-order effects
- If the pattern holds, peer-to-peer marketplaces drift toward instant-book and identity-verification models that strip per-request host veto power — trading the 'live like a local' intimacy that defined home-sharing for standardized acceptance rules borrowed from hotels and ride-hailing.
The trend: Platform marketplaces are being pushed from trust-based host discretion toward audited, algorithmic matching as field experiments quantify discrimination their own interfaces enable.