A Discrete Choice Study to Assess the Impact of Reliability on Passenger Rail Demand

A Discrete Choice Study to Assess the Impact of Reliability on Passenger Rail Demand


R Batley, M Wardman, J Shires, J N Ibáñez, J Dargay, ITS, University of Leeds, UK; G Whelan, MVA Consultancy, UK


We report a study of the impact of punctuality and reliability on UK passenger rail demand. Dynamic econometric methods are employed to yield evidence on market elasticities, and discrete choice methods to yield evidence on monetary valuations.


Almost all evidence relating to the behavioural impacts of rail travel time variability, and indeed travel time variability on other modes, has been drawn from Stated Preference (SP) analysis. A problem with the SP method in general is non-commitment bias, whereby individuals are not obliged to respond in practice in a manner consistent with how they respond to hypothetical scenarios. Hence there are several instances in rail research where what seem to be exaggerated values of factors such as late time and crowding have been estimated. Matters are made worse in the context of travel time variability because presentation of the concept to individuals is far from trivial. Furthermore, the usual approach (in the Passenger Demand Forecasting Handbook, at least) has been to estimate valuations of reliability, and from these to infer demand impacts. This method of forecasting demand has attracted some criticism, particularly as it may inflate the Generalised Journey Time (GJT) elasticity.

This paper reports a study of the impact of punctuality (i.e. where a train runs to time) and reliability (i.e. whether a train runs at all) on passenger rail demand, funded by the UK Department for Transport. Acknowledging the above issues, this study aspires to place greater emphasis on actual (i.e. revealed and reported) behaviour, and to obtain direct estimates of the effects of changes in travel time variability on rail demand.

The study develops two parallel research themes, motivated by an interest in revealing distinct but complementary insights, as follows:
- aggregate: dynamic econometric methods are employed to yield evidence on the market elasticity properties of changes in reliability, and the characteristics of any inherent dynamics
- disaggregate: discrete choice methods are employed to yield evidence on behavioural responses to reliability at the individual level, and associated monetary valuations

Aggregate analysis
Econometric analysis of rail ticket sales data has made a major contribution to understanding rail demand, and there is no reason in principle why this method cannot also yield insights into the demand impacts of train punctuality and reliability. Naturally, any such analysis might yield particular insights when applied to routes and periods that have been subject to travel time variability. Acknowledging this, we convened a meeting with industry representatives for purposes of identifying potential study contexts. Guided by this consultation, we have assembled a sample of 240 O-D pairs (480 single flows) on a 13-period basis from 1996 until the present, and this will form the basis for our econometric analysis. This analysis is supported by data from various sources; principally LENNON ticket sales data, generalised journey times from MOIRA and performance data directly from train operating companies (TOCs); and will exploit panel data methods.

Disaggregate analysis
Whilst the econometric analysis is couched in terms of some notion of the aggregate, disaggregate analysis permits more detailed consideration of the agents that comprise the market. In the present case, disaggregate analysis offers a means of revealing travellers? attitudes to risk and monetary valuations of punctuality and reliability. It was decided at the outset of the project that disaggregate analysis might usefully combine three elements: retrospective questioning, revealed preference (RP), and stated preference (SP). With a view to implementing these three elements, the industry consultation sought to identify potential study areas which exhibited either, or both of, two identifiable characteristics, thus:
- where passengers have experience of changing levels of punctuality and reliability (i.e. retrospective questioning)
- where passengers have a choice between services with different levels of punctuality and reliability (i.e. RP)
On this basis we developed a passenger questionnaire appealing to both characteristics. In order to mitigate the risks of relying entirely on revealed and reported preferences, a final component of the questionnaire was SP. As a means of presenting the concepts of punctuality and reliability in the SP, we employed a development of Hollander?s (2006) presentation, which itself might be seen as a simplification of the so-called ?clock face? presentation adopted by Bates et al. (2001).

In respect of the econometric analysis, we are presently finalising the collation of data and will shortly commence modelling. In respect of the disaggregate analysis, we have pilot-tested our passenger survey, and the field application will take place during February. The project is due to be completed by the end of March, when we will report analysis of the impact of changes in punctuality and reliability on rail demand, as well as associated monetary valuations.


Association for European Transport