Decomposing the Congestion Effect and the Inference Effect of Competition: A Field Experiment

Abstract

Are firms more or less likely to enter a market if they observe that competitors have entered? This most basic question has received contradictory empirical answers. The normative recommendation to firms that make entry decisions is therefore ambiguous. We reconcile this controversy by introducing demand uncertainty as a moderator of how entrants respond to existing competition. We distinguish between two effects of competition on entry decisions: a negative “congestion effect,” where competition dissipates profit when demand is fixed and is known, and a positive “inference effect,” where firms infer high demand from a large number of competitors. To tease apart these two effects empirically, we use field experiment data from a website that brings together buyers and sellers of used goods. Before each potential seller made a posting request, the website randomized whether to disclose the number of buyers and/or sellers, and the exact number to disclose. We find evidence for a positive inference effect: When the number of buyers is not disclosed, the overall effect of the number of sellers on entry is neutral; when the number of buyer is disclosed, however, a larger number of sellers lowers the entry propensity due to the congestion effect. We discuss how our results should affect the information disclosure strategies of two-sided platforms.