Cross-nested Logit Modelling of the Combined Choice of Airport, Airline and Access-mode

Cross-nested Logit Modelling of the Combined Choice of Airport, Airline and Access-mode


S Hess, Imperial College London and RAND Europe; J Polak, Imperial College London, UK


This paper presents an analysis of the joint choices of airport, airline, and access-mode for air-passengers departing from the five main London airports.


The importance of sound analyses of air travel choice behaviour has grown significantly in importance in recent years, as the result of the increasingly precarious financial situation of many of the world?s leading airlines, increasingly complex regimes of regulatory, safety and environmental control, and growing problems of congestion both in land-side and air-side facilities. One area that is of particular importance is the analysis of the choices made by passengers departing from major multi-airport regions. The modelling of such choices is not only appealing from a research perspective, due to the complexity of the choice process, but is also of great practical policy importance in many large metropolitan areas. Indeed, the unprecedented increase in air-transport has led to major problems of congestion in a number of multi-airport regions, leading to urgent needs for capacity expansion, especially given the forecasts for continued strong growth. Given the long-term nature of any such expansion work, and the associated financial and environmental constraints, the generation of reliable forecasts of passenger levels (at the airports, as well as in the ground-level access-network) is a prerequisite to any planning process. The generation of such forecasts in turn leads to a requirement for an understanding of passenger behaviour, as yielded by studies of airport choice behaviour.

Consequently, it is of little surprise that this area of research has seen increased levels of activity in recent years. However, the majority of existing studies have been limited by the use of rather restricted model structures and over-aggregated data. Furthermore, studies of airport choice behaviour have generally failed to acknowledge the three-dimensional nature of the choice-set, in that passengers not only make a choice of departure airport, but additionally choose an airline and an access-mode. The joint analysis of these three choice-dimensions, and the interactions between them, can however lead to important gains in model performance and accuracy, as illustrated by Hess (2004).

This paper advances the current state of the art in the modelling of airport choice in two important respects. First, a significant methodological advance comes in the form of the use of cross-nesting model structures that allow for the joint representation of correlation along the three choice dimensions (airport, airline and access-mode), without requirements to use a multi-level nesting structure. And second, our data refer to the London area, which is arguably the most competitive multi-airport region in the world. The authors are not aware of any existing independent public-domain study that looks at the choice between the five main airports in this region. This area not only has higher levels of competition than the repeatedly used San Francisco Bay region, given that the four main airports are located at comparable distances (in terms of travel-time) from the city centre, but also has a much wider modal split for the access-journey. These factors make the area a very challenging candidate for a study of air-travel choice behaviour.

In common with most previous studies, the analysis shows that access-time is a prime determining factor in travellers? choices of departure airport, while flight frequency, access-cost and flight-time also play a consistent role. Other factors, such as air-fare, aircraft-type, waiting-time, walk-time, on-time performance, and allegiance to national carriers play a role only for some of the segments of the population. The very high implied values of access-time, which are consistent with studies in other regions, suggest that air-passengers, particularly those travelling on business, are very reluctant to accept increases in their access-journeys, and are highly captive to their local airport. As such, the attractiveness of outlying airports depends heavily on good access-connections, unless there are other incentives, such as low air-fares. This is reflected in the fact that only low-cost carriers find it possible to attract large numbers of passengers to outlying airports that are not served by convenient and fast ground-level services. It is conceivable that the sensitivity to access-time decreases with flight-time, such that moving long-haul services to outlying airports would seem wise; this however causes problems, as the associated (and necessary) short-haul feeder flights will also carry point-to-point passengers, which will again have a preference for more centrally-located airports.

In terms of model performance, all attempted nesting approaches lead to significant gains in model fit when compared to structures assuming strict independence between alternatives in the unobserved utility components along all three dimensions of choice. When accounting for correlation along just one dimension of choice, the best performance is obtained by the model using nesting by access-mode. However, the simultaneous treatment of correlation along all three dimensions has clear benefits, and the Cross-Nested Logit model clearly outperforms the three Nested Logit structures; this suggests that this model form can serve as a valuable tool in airport-choice modelling.

As an extension, model structures belonging to the family of Mixed GEV models are currently being estimated. These allow for a treatment of inter-alternative correlation in the same way as the Nested and Cross-Nested Logit models estimated on the data thus far, but additionally allow for random variations in behaviour across respondents within the same population segment, which can lead to great gains in model performance, as illustrated by Hess & Polak (2004). Finally, data from a recently conducted Stated Preference (SP) survey for air-travel choice behaviour were used in a parallel model fitting exercise; the results produced on these data were very similar to those produced with the Revealed Preference (RP) data, most notably in terms of the high values of access-time, lending some additional credibility to the results and conclusions of our study.


Hess, S. (2004), A model for the joint analysis of airport, airline, and access-mode choice for passengers departing from the San Francisco Bay area, paper presented at the European Transport Conference, Strasbourg, France.

Hess, S. and Polak, J.W. (2004), Mixed Logit modelling of airport choice in multi-airport regions, Journal of Air Transport Management, forthcoming.


Association for European Transport