Abstract
The Value of Travel Time Savings (VTTS) is the monetary value attached to reductions in travel time. VTTS is an important willingness to pay (WTP) estimator which plays a crucial role in the evaluation of transport projects and pricing policies; it is often the case that travel time savings account for the biggest part of total monetized benefits in a transport project. Approximately four decades after the initial efforts to estimate VTTS, the indicator itself and its extensions remain in the centre of transport economists’ interest.
One of these extensions is the large variations that characterize VTTS estimates across a plethora of factors. The scope of the thesis is to answer whether the observed cross transport mode variations in the estimates are mainly due to one broad family of factors, namely the user type effects (socioeconomic variables such as income, gender etc.), or due to a second family of factors that consist the mode effects (attributes of the mode such as speed, safety and comfort). A third theoretical possibility is that the differences stem from the strategic behavior of the respondents (the agents escape the context of the experiment and perceive incentives to not reveal their true WTP) in the choice experiments (CE).
The origin of VTTS variations is a rather critical question, especially when it comes to complicated transport projects which involve simultaneous changes in the market shares and in WTP of the travelers in a given transport mode. What is investigated is rather the relative power of these effects than the dominance of one of them; therefore, we admit an ex-ante coexistence of at least user type and mode effects rather than checking for a polar case. Our focus is on the relative power of the two effects which is responsible for the change in total monetized benefits.
The empirical results of the study can be grouped in two parts. In the first one, re-estimation of VTTS with mixed logit from Norwegian SP data takes place. The estimations involve experimentation with various mixing distributions. The result is a variety of estimates which enrich the empirical findings of the Norwegian Value of Time (VoT) study in 1997. In addition, models which control for income and gender are developed, allowing for socioeconomic segmentation of VTTS.
The second part, which answers the central question, adopts the model with Normal mixing distribution and exploits the experimental design of the study in order to separate the effects from each other. Then, practical LR tests are employed to check for the significance of the observed differences in VTTS. The concluding remarks pose future challenges which arise from the complex nature of the issue.