Alternative Methods for Cost Allocation in Stochastic Service Systems

In this paper we consider the question of allocating costs and setting prices for customers in a variety of queuing systems. We will focus on the problem of setting users fees within an organization for a shared resource, such as a computer center. We argue that both the Aumann-Shapley pricing rule, and a pricing rule based on the Shapley value of a finite game, are appropriate cost allocation methodologies. These approaches allow us to explicitly determine prices for customers in a queuing situation, who differ from one another in terms of their arrival rates, service rates, costs of lost work, and the number of simultaneous servers which they require.