Game-theoretic models for competitive pricing and network design problems

Private roads competition is one of the important issues under a Build-Operate-Transfer (BOT) scheme. When two or more competing firms operate multiple toll roads, their profits are interrelated due to demand inter-dependence and each firm must consider what its competitors' choices are likely to be. The main task of this dissertation is to develop game-theoretic approaches to the study of the road network that involves multiple toll roads operated by competitive private firms. The strategic interactions and market equilibria among the private firms are analyzed in determining their supply or price or both over the network. Furthermore, exact bounds are established for the efficiency loss of private road competition in some simplified or special cases.; First, the effects of oligopolistic equilibria are examined on a network of parallel roads. The inefficiency is bounded by considering general demand and cost functions. As a supplement, both toll and capacity competition among private asymmetric roads is also considered; next, the case of fixed OD demand with continuously distributed value-of-time is investigated. The efficiency loss is discussed under different kinds of regimes and the construction capacity set by a monopolist is considered as well as the toll charge. A rate-of-return (ROR) regulation which may have attractive advantages is suggested for the regulatory authority; for further extension of previous work, the problems are examined in a context of general traffic networks. The study shows that generally private pricing and competition can be both profitable and welfare-improving.; Investigating the subgame perfect equilibrium in an oligopolistic freight market is another task of this dissertation. A partially non-cooperative game among shippers, carriers and infrastructure companies (IC) is modeled by a three-stage procedure. The study shows that the equilibrium flows can also maximize total system profits if the IC and the carrier both use vertically efficient nonlinear pricing schedules.