Airport choices and resulting catchments in the US Midwest

Air travelers originating from the catchment areas of airports - typically defined by administrative or geographic boundaries - will sometimes "leak" across these boundaries to farther, often larger airports that offer more attractive services and options. To investigate the attractiveness of different airports based on their proximity and geographic position relative to one another, we assess the airport choices of air passengers across a large region of the US Midwest that spans multiple administrative boundaries and airports. We combine a dataset of air tickets purchased by these passengers with other publicly available datasets, and apply mixed logit airport choice models based on air service attributes and airport access distance. We take multiple subsets of the total air tickets dataset using Halton sampling in order to achieve model estimation convergence. Model estimates are compared between data subsets, and consistently lead to the following key results. First, changes in air service attributes on short- and medium-length routes will have a more uniform influence on air passengers' airport choices, compared with long routes. Second, comparable proportional changes in air service attributes lead to greater changes in the catchment areas and market shares of small and medium airports, compared with large hubs. Within these service characteristics, airfare is found to impact passengers' airport preferences more than flight frequency. Finally, the catchment areas and market shares of small- and medium-size airports are influenced by the proximity and geographic location of neighboring airports; this differs from the catchments of large airports, which strongly extend into multiple jurisdictional boundaries in all directions despite the presence of surrounding airports. The study results provide further insight into the influence of air service attributes on longdistance airport choices over large geographic regions served by a diverse set of airports, and the resulting implications on multimodal transportation planning within and across such regions.