Pricing, routing, and incentive compatibility in multiserver queues

This paper studies pricing and incentive issues in the assignment of customers to servers in a system that suffers congestion effects. When customers have private information about their waiting costs, a system administrator who wishes to maximize steady-state net benefits per unit of time (i.e. total benefits from service minus total waiting costs) may do so using a pricing and routing scheme that is incentive compatible; that is, no customer has any incentive to reveal his private information untruthfully. When the system administrator wants to maximize toll revenue, the optimal scheme involves higher tolls, and hence lower congestion, than is socially optimal.