On the Use of Discrete Choice Models for Airport Choice with Applications to the San Francisco Bay Area Airports

This paper describes an in-depth analysis of the combined choice of departure airport, airline and access-mode for passengers departing from the San Francisco Bay area. The analysis shows that several factors, most notably flight frequency and in-vehicle access-time, have a significant overall impact on the appeal of an airport, while factors such as fare and aircraft size have a significant effect only in some of the population subgroups. The analysis highlights the need to use separate models for resident and non-resident travelers, and to segment the population by journey purpose. The analysis also shows that important gains can be made through accounting for past experience at the different airports and through using a non-linear specification for the marginal returns of increases in flight frequency. In terms of model structure, the results suggest that the use of the Nested Logit model leads to significant improvements in model fit over the use of the Multinomial Logit model, although these improvements do not necessarily translate into significant advantages in prediction performance, which is already surprisingly good in the base models, showing the importance of using a detailed specification of utility.