Car Users’ Trade-Offs Between Time, Trip Length, Cost and Road Pricing in Behavioural Models

Abstract

Stated Preference (SP) experiments are the obvious choice in order to forecast travellers’ responses to alternatives that do not exist today. SP experiments can in addition change the variables in a controlled factorial design, whilst Revealed Preference surveys (RP) have to rely on measured observations. Preferences concerning correlated explanatory variables such as travel time and travel length can therefore often be estimated more easily by SP experiments than by RP surveys. However, respondents may not act as they claim in the interview and the design may affect the results.

The chapter illustrates, based on an SP experiment compared with the AKTA data (Chapter 6), how the definition of variables can influence the results obtained. Value of travel time, VoT, (both free flow and congestion travel time), choice of time-of-day of travel and route choice are considered both without and with road pricing. It is shown that the design of the experiment seriously affects the result especially with respect to VoT. It is then shown how the income variable and distributed coefficients (taste variation) can improve the model fit and its behavioural accuracy. The model structure obtained corresponded well to the best models from the field experiment (RP) in AKTA, although the size of coefficients differed somewhat.

It appears that (marginal) cost is a problematic variable compared to trip distance as the respondents had serious difficulties in estimating the cost per kilometre. Road pricing was considered a bit worse than a pure marginal driving cost in the road pricing situation, i.e. car users preferred paying for fuel than paying road pricing. But the total cost after road pricing was introduced was weighted less than before road pricing compared to travel time — or in other words travelling time was weighted higher. This can be explained by the fact that travellers have restrictions and inertia in their possibilities of changing behaviour as well as the fact that the car-users who still use the car are the ones with the higher willingness to pay and the travellers who change behaviour have the lower willingness to pay.

Mai-Britt Herslund is thanked for the work on SP design and the interviews, Christian Würtz on work collecting and processing the GPS data, Paolo Menegazzo on work regarding the route choice model estimator and data processing and Majken Vildrik Sørensen is thanked for work on the SP and model estimation, including work reported in Nielsen and Sørensen (2004) upon which section 18.4.3 is based. The AKTA project was primarily financed by the EU’s 5th Framework Programme and the municipality of Copenhagen, with co-funding of the research part by the Technical University of Denmark (DTU). Principle contractor of AKTA was the City of Copenhagen, from which Poul Sulkjær is thanked. DTU, the Danish Road Directorate and the Ministry of Transport were assistant contractors. The AKTA SP was funded by the Danish Transport Council and the remaining work on this was allowed to be finalised by self-funding from the DTU and the Danish Transport Research Institutes (DTF)