| Summary |
In view of the increasingly diversified choices of air transport services by travelers (for example, destinations, routes, and carriers), those considerations are critical to airline business strategies. Thus, understanding travelers' choice behavior is a very important issue.
This study uses aggregate panel data to construct a destination-route-carrier choice model based on the origin metropolitan area. This model keeps the total demand elastic; it can comprehensively consider the impacts of social-economic factors of the origin and destination city, as well as the influence of the service level of routes, carriers, OD airports, and connecting airports on air travel demand; it explores the substitution patterns among alternatives (such as the alternatives of carriers and destinations, the OD airports in the multi-airport metropolitan areas); it allows destinations with different attributes to have different substitution patterns, and substitution between aviation and non- aviation options is different by origin metro areas with different attributes; after improving the problem of fare endogeneity by the instrumental variable method, the estimated value of time and the fare elasticity become more reasonable.
In addition, the fare elasticities, which defined the market as a city-pair, in the literature would mostly fall between the fare elasticity of metropolitan area demand and the fare elasticity of city-pair demand estimated from our models. Therefore, when studying travelers’ behavior, ignoring the destination substitution may underestimate the pricesensitivity of travelers.
The model is also used to simulate airline strategies. The model can predict the changes in total demand in the origin metropolitan area, and the allocation changes in destinations, OD airports, carriers and routes. In addition, when multiple rival airlines make strategic adjustments at the same time, the effects of competition among airlines can be revealed. Airlines can choose better strategies and markets to increase their revenue based on the predicted results of this model. |
