Using Prices More Realistically as Decision Variables in Perishable-Asset Revenue Management Problems

This paper addresses a new slant to a problem which is general to many of the transportation industries—perishable asset revenue management. Traditional approaches have assumed that prices are fixed and solved for the optimal allocation quantities. Our approach recognizes that prices of the different classes affect demand and should therefore be included as decision variables to be optimized. We solve three different types of problems: (a) up to n price classes, distinct asset control mechanism, and no diversion, (b) up to 3 price classes, serial nested asset control mechanism, and no diversion, (c) up to 3 price classes, serial nested asset control mechanism, and diversion. Analytical results are provided in most cases and examples illustrate the results as well as the time required to solve these complex problems. Finally we look at the tradeoff involved between computational time and expected contribution when using heuristic decisions obtained from less realistic assumptions relative to the true optimal decisions. On average, the suboptimality ranged from 3.19% to 4.88% with a corresponding decrease in computing time required on the order of several minutes. Some trends are presented to help determine a priori which type of problems would tend to benefit most from the more accurate formulation. This should help managers decide when it is worth the extra computing time to come up with the true optimal solution.