Association for Public Policy Analysis & Management

A common critique of home-sharing platforms is that they enable hosts to impose costs on their neighbors. We consider four potential public policy responses that differ in whether the decision right to host is allocated to: (1) individual tenants,
(2) building owners, (3) cities, and (4) a social planner. We find that (2) and (4) are equivalent, with (3) leading to too little hosting, and (1) to too much hosting. The efficiency of (2) depends on building owners being indifferent between allowing and banning home-sharing in their buildings. To assess this ``no policy arbitrage'' prediction, we constructed a dataset of NYC rental apartments listings. Although we do not observe building home-sharing policies, there are several ``policy'' attributes captured in the data, such as whether buildings allow subletting. Consistent with our prediction, we find that policy choices have no detectable effect on rental prices. Despite the attractiveness of the equilibrium of policy (2), tenants must ``sort'' across buildings, potentially at substantial cost. We explore this sorting with an agent-based model, and show how individual preferences and moving costs affect the equilibrium.