Multiple signals in the multidimensional limit pricing game: Theory and evidence from the United States airline industry

In this dissertation, we study the limit-pricing game with multiple signals under multidimensional uncertainty. The potential entrant faces an asymmetric information problem on both the cost and demand sides. By revealing quantity and price information, the incumbent can signal market characteristics to the potential entrant and influence his entry decision. The directions of distortion in strategic variables (price and quantity) in such a signaling game are actually ambiguous. Another interesting finding is that the equilibrium outcome of the incumbent firm may not always be a price and quantity pair on the demand curve. Use of the method of "forward induction reasoning" pins down the reasonable (or "intuitive") equilibrium. We also confirm the fundamental properties of past studies: the players (incumbent and potential entrant) can benefit from more revealed information in a separating equilibrium.; The model is extended to the case with oligopolistic incumbents and with continuous types. Corresponding to the signaling limit-pricing game with distortion in the strategic variable, we find that in equilibrium the incumbents' strategies, price and quantity pairs, deviate from myopically optimal levels to influence the entry decision. By modifying the entrant's beliefs, asymmetric strategies may exist due to interactions among the incumbents. We also find that the incumbent sacrifices less profit to implement an entry-deterring strategy if the degree of current market competition or the barrier to entry is increasing. However, the effect of the change in the potential market competition is unclear.; An implication of limit and predatory pricing is that entry and exit should be associated with deviations of the incumbents' prices and sales from their short-run optimal path. We conduct an empirical test on the theory by using data from the U.S. airline industry. The results of this paper support the extended limit-pricing hypothesis, but fail to support the predatory pricing hypothesis. Since the probability of entry (and exit) cannot be observed, but the occurrence of entry (and exit) can be observed, logit and probit models with fixed effects (and random effects) have been employed.