Knowledge, congestion, and economics: Parameter uncertainty in Naor’s model

Abstract

This paper studies an extension of Naor’s model in which there is parameter uncertainty. In particular, the arrival rate is known, to customers and system managers, only through its distribution. For the observable case, the relationship between the optimal individual threshold and the thresholds for a social optimizer or revenue maximizer does not change from the classical model with a known arrival rate. However, in the unobservable case, it is shown that the decisions of the social optimizer and revenue maximizer no longer coincide. Furthermore, in the unobservable case with arrival rate uncertainty, the social optimizer induces a lower expected arrival rate than the revenue maximizer. This stands in contrast to the observable case, in which the social optimizer prefers a more congested system.