Overbooking Analysis for Fare Class Dependent Oversale Costs

This paper explores the revenue management problem in the presence of no-shows. The service provider must pay the oversale cost if the shows at the service delivery time exceed the available capacity. Unlike the traditional overbooking model, we consider class dependent oversale costs. Assuming that denying higher fare classes (i.e., higher oversale cost) is more expensive, service priority is given to higher fare customers over lower fare customers. We formulate dynamic programming to solve the overbooking problem with fare class dependent oversale costs. However, dynamic programming suffers from dimensionality. To address this problem, we propose the linear programming (LP)-based heuristic. We also propose a simple heuristic using the aggregated oversale cost to evaluate the performance of the LP-based heuristic. Our numerical study shows that the LP-based heuristic performs well, especially when the spread of the oversale costs is large.