Analysis of price competition with yield management in the United States airline industry

Ticket prices in the U.S. airline industry are volatile, sometimes during price wars, in which the airlines alternately reduce their prices, and very frequently in up-and-down price shifts. The only explanation offered in the literature as to why prices are volatile while market conditions do not change is due to core theory. However, that explanation only accounts for price wars. This study proposes an explanation of both kinds of price fluctuations.; A game-theoretical model is developed that integrates key elements of price competition between major and low-fare-low-cost airlines, and of yield management; two fields that have previously been studied only in isolation to each other. In the model, a major airline competes with a low-fare-low-cost airline and offers two fare classes, unrestricted and restricted. In maximizing its revenue, it simultaneously decides its ticket prices and the number allocated to the restricted class. The low-fare-low-cost airline sets the ticket price of its sole fare class. Customers are characterized by parameters from a continuum of possible values, which in combinations lead to many possible preference patterns. This is a richer depiction of customer behavior than in previous studies.; Cycles of price fluctuations can occur if there are discontinuities of the marginal revenues of fare classes. These discontinuities are formed by a shift of demand rate of a class when a different class becomes full. If there are no multiple fare classes offered for the same service, i.e., there is no yield management, then this shift can not occur.; In this study, we search for optimal responses and pure strategy Nash equilibria of the game. A set of sensitivity analyses validates the model by showing conformance with common sense reasoning. The estimation of the number of seats sold is simplified by treating customer arrivals as a continuous flow instead of discretely. It is proved here that the continuous flow model is asymptotically nearly equivalent to the discrete customer model. This provides a theoretical justification of the continuous flow approximation commonly used in yield management studies.