| This thesis models the equilibrium network structures of competing hub-and-spoke airlines that choose which routes to serve to maximize their profits subject to their competitors' choices of routes. The model simulates airline traffic patterns and traffic volumes for air travel between twenty-two major US cities. Considering only the 22 cities with airports, there are 464 origin/destination city-pair combinations and 10,684 origin/hub/destination triples. Deciding whether service should be provided on each of these triples involves on the order of 210684 permutations. An integer-programming algorithm---simulated annealing---determines a Nash equilibrium in origin-hub-destination routes served by profit-maximizing airlines. Airlines take demand and prices as given by functions of city size, distance, and other characteristics of the origin and destination cities. A modal-choice model assigns passengers to routes through alternative hubs. Airlines experience economies of scope from joint production of trips between different city pairs using aircraft that exhibit economies of scale. As airlines add origin-destination pairs to their networks, the costs of additional circuity of routes through their hub eventually overwhelms the economies of scope---limiting the size of their networks. Parameters of the fare functions and modal-choice model are estimated from the FAA's ten-percent sample of flight coupons. I compare results from multiple runs of the algorithm for consistency and also with the actual passenger-routing data, and investigate the model's applicability to policies that affect network configurations---such as, congestion pricing, airport expansion, mergers, and price regulation. |
