Sorting based on observable features such as gender or race has a widespread negative perception. Peer-to-peer markets, fostered by the boom of online applications, are no exception to this trend. Policymakers around the globe are urging platforms to reduce the information that users share through their profile pictures and names, as a way to prevent discrimination. Given the role that observable features play building users’ trust, these measures may hinder their participation in asymmetric fashion, affecting especially specific population segments. To analyze the implications of profile information on female participation and ethnic sorting, I focus on BlaBlaCar, the world´s leading car-sharing platform for non-professional drivers and passengers. I construct a novel data set that contains detailed information of all the users on both market sides and all transactions in the routes that connect eight of the largest cities in France, between October 2020 and March 2021. Using a structural framework that accommodates the main strategic decisions of both market sides, this paper shows that women prefer to travel with other women and that there exists a substantial degree of ethnic-based homophily. I develop a unique identification strategy, based on the capacity of the driver to reject passenger requests, to disentangle the preferences of drivers and passengers. This paper also provides evidence that alternative designs limiting the sorting abilities of users do not necessarily benefit ethnic minorities and that women from the ethnic majority tend to be the population segment whose participation and welfare reduces the most when anonymous marketplaces are imposed. Alternative designs are showed to influence the pricing decisions of the different population segments in various ways, depending on the side of the market whose sorting ability is affected and the proportion of each population segment amongst supplying and demanding agents.
Supplementary notes can be added here, including code and math.