We use data from an online marketplace of vacation rentals (Airbnb) collected in 19 major cities in North America and Europe to measure discrimination against ethnicminority hosts. This market has three interesting features: the existence of a detailed reviewing system, the high frequency of transactions and the panel dimension of the data. Using the fact that ratings provide potential guests with information about the quality of a listing, we build a credible measure of statistical discrimination, following a strategy a la ` Altonji and Pierret (2001). Hosts from a minority ethnic group charge 16% less than other hosts in the same cities. Controlling for a rich set of characteristics reduces the ethnic price gap to 3.2%. An additional review increases the daily price more for minority than for majority hosts. Estimating the parameters of a theoretical pricing model, we find that statistical discrimination accounts for most of the price differential: 2.5 percentage points. Joint with Morgane Laouenan.

Click here to read the paper.