Article ID: | iaor2001829 |
Country: | United States |
Volume: | 33 |
Issue: | 2 |
Start Page Number: | 182 |
End Page Number: | 191 |
Publication Date: | May 1999 |
Journal: | Transportation Science |
Authors: | Chatwin Richard E. |
Keywords: | programming: dynamic, markov processes, yield management |
We analyze a model of airline overbooking in which customer cancellations and no-shows are explicitly considered. We model the reservations process as a continuous-time birth-and-death process with rewards representing the fares received and refunds paid and a terminal-value function representing the bumping penalty. The airline controls the reservation acceptance (birth) rate by declining reservation requests. Assuming that the fares and refunds are piecewise-constant functions of the time to flight, we demonstrate that a piecewise-constant booking-limit policy is optimal, i.e., at all times, the airline accepts reservation requests up to a booking limit if the current number of reservations is less than that booking limit, and declines reservation requests otherwise. When the fare is constant over time or falls toward flight-time, the optimal booking limit falls toward flight-time.