全球智库信息集成服务系统

• Models with multiplicative error structures fit the SC data better and give more reasonable VTTs.
• We do not observe significant reference dependence in the stated choices.
• Travellers with higher (personal) incomes and those making longer journeys have higher VTTs.
• Respondents react more negatively to tolls than to other driving costs, but not significantly so.
• The design of the experiments — specifically the position of time and cost in the stated choice alternatives — impacts the resulting VTTs.
• For commute and other travel, the VTTs are influenced by the socio-economic characteristics of travellers (we do not observe significant variation in VTTs for business travellers due to socio-economic characteristics, except whether they are reimbursed for travel).
• Significant random heterogeneity in free flow and congested VTTs for commute and other travel is observed, in addition to other measured covariates (we were not able to identify random variation terms for business travel, perhaps because of the relatively small sample size for this segment).
• In contrast to other recent studies, the resulting VTTs were not impacted by the survey recruitment methodology, specifically whether respondents were recruited from an online panel or intercepted while making an in-scope journey.
• We do not observe difference in the formulation of VTTs for drivers and passengers, although they will differ in practice because of differences in their socio-economic and trip characteristics.
• There is little evidence of tunnel phobia. In fact, the tolled tunnel is viewed positively relative to the untolled routes, over and above the toll and travel time savings.

This study quantifies the willingness of car and light van travellers to pay to use a proposed new tolled tunnel to travel across the harbour in central Copenhagen. The crucial issue in predicting whether drivers will choose to pay the toll is to understand the 'value of time' (VTT) of each driver; that is, the value that each is willing to pay to save a minute of his or her time (and that of the passengers, if any) and whether they have a bias against tolls. To fulfil the research purpose, a stated choice (SC) experiment was undertaken with 3,688 travellers in cars and vans travelling in the relevant corridor. Discrete choice models (DCM) were then developed to quantify travellers' VTTs. We found that DCMs with multiplicative error structures gave a better fit to the data and more reasonable estimates for VTT and travellers with higher incomes and those who made longer journeys placed higher value on travel time savings. We observe significant random heterogeneity in the value of free flow and congested travel times for commute and other travel, in addition to systematic variation by income, trip length and other measured socio-economic covariates. In contrast to other work, we find little evidence of reference dependence in the survey responses or that the survey recruitment methodology impacted the resulting VTTs. We find that respondents reacted more negatively to tolls than other driving costs, but not significantly so.

• Chapter One

  Introduction

• Chapter Two

  Design of the survey and stated choice experiments

• Chapter Three

  Data collection

• Chapter Four

  Characteristics of the sample data

• Chapter Five

  The stated choice models

• Chapter Six

  Summary and recommendations

• Appendix A

  Survey Questionnaire

• Appendix B

  MNL model development results