Pareto-Improving Transition to Prioritized Service for Incentive- Compatible M / G / 1 Queues

Notably absent in the study of congested service facilities is the transition from a non-priority to a priority system. Building on the framework first presented in Mendelson (1985), we study this transition from the perspective of both the system manager and the system user. Using an M/G/1 priority queueing discipline with a fixed traffic assumption, we present three fundamental theorems about the transition: (i) private costs of individual users decrease after the transition, (ii) total delay costs decrease after the transition, and (iii) total revenue decreases after the transition. If the system traffic is not fixed, we prove that the transition cannot increase the private costs of all users. Because the analytical results do not ensure a good transition, we develop an optimization model that incorporates transition costs along with an efficient heuristic to solve the model. Simulation results demonstrate that (1) the initial post-transition solutions are typically Pareto-improving and (2) for non Pareto-improving solutions, the heuristic quickly generates high-quality solutions that cannot be easily improved by additional search.