Price Competition and Platform Design in Two-Sided Networks

Abstract

The owners of large online platforms, like Airbnb or Amazon Marketplace, are able to affect the probability that a given buyer observes a given seller on their platform. Buyers and sellers therefore interact on a bipartite graph with links between them weighted by the probability of observation. Platform owners are then able to design the network by choosing the observation probabilities. A seller's price is decreasing in their Bonacich centrality. In order to maximize profits, the platform owner connects an entrant to buyers according to a measure of their own centrality in the graph. While a network where buyers observe and trade with all sellers with probability one maximizes consumer surplus, such a network does not necessarily maximize the platform's profit. The platform owner faces a trade-off between increasing observability and reducing competition, which explains why buyers tend not to be able to observe all sellers on online platforms. If sellers are identical, increasing seller prominence for any consumer segment increases competition, and profit is maximized when it is randomized across all sellers.