The value of travel time changes: theoretical and empirical issues

The value of travel time is a key input for the evaluation and comparison of transport projects. Travel time savings often constitute a great part of the benefits of a project, and therefore the value assigned to them is crucial for cost-benefit analyses. Less often, there may be some transport projects which lead to travel time increases (e.g. repair or maintenance works). In general, it is essential to determine the valuation of travel time changes (VTTC). The VTTC is defined as the marginal rate of substitution between travel time and travel cost for an individual. Hence, the VTTC is likely to differ across people and even for the same person under different circumstances. Therefore, for policy-making it is recommendable to have a set of VTTC based on factors that could influence time valuation: e.g. income or journey length. Stated choice (SC) experiments constitute the most popular method to estimate the set of VTTC of a population. These experiments offer travellers hypothetical choice scenarios to observe their VTTC. Therefore, the choice context may also play on role on valuation: e.g. size and sign of the time changes offered. The choices are analysed using discrete choice models to estimate the VTTC. The aim of this thesis is to increase our understanding of a population’s underlying set of VTTC. To achieve this target, we first explore a series of key sources of variation of the VTTC and relate them within the framework of microeconomic theory. Potential confounding between sources is investigated. Secondly, this thesis identifies, relates and compares two popular modelling approaches to estimate the VTTC: Random Utility and Random Valuation. Finally, our research analyses the role of the design variables used in the SC experiment on the estimation of the set of VTTC. The empirical work has been carried out using datasets from the last national VTTC studies in the UK and Denmark. The results provide valuable insights, from which would highlight the following: i) the Random Valuation approach proves to be superior to the traditional Random Utility approach, and in general gives a systematically lower VTTC, ii) confounding is apparent between some sources of variation and the design variables influence most model estimates that determine the set of VTTC, iii) journey length effects do not exist in the data explored, as opposed to what earlier works report. The findings of this thesis have important implications for appraisal.