Understanding the Value of Travel Time.: Advanced modelling techniques applied to the National German Value of Travel Time and Travel Time Reliability Study

Abstract

Project appraisal is an essential part of policy making, in transportation and elsewhere. To this end, different valuation methods are used. One of the most important evaluation tool is the Cost-Benefit-Analysis. Often, travel time savings account for the largest share of the gains in Cost-Benefit-Analyses. Therefore, they are a central element of the analyses, making its accurate determination of great importance for transport policy appraisal and investment decisions. The time changes are evaluated using the willingness to pay of travellers to save time resulting in the most important number in transport economics: the value of travel time. The \emph{German Value of Time and Value of Reliability Study} was the first official national study estimating values of travel time for Germany. For this purpose a large nationwide data set was collected. It covered six travel modes and five trip purposes, included several transport related attributes and different kinds of stated preference experiments, as well as two time horizons in the choice experiments. This great complexity and detail and the large sample size allows the investigation of various aspects and perspectives of time valuation. This thesis makes use of this rich data to compare a variety of approaches for travel time valuation, comparing different state-of-the-art model formulations, and covering more controversial topics, such as the use of values of time from long-term decisions. The values of time are derived using various model formulations and splits of the data. The impact of the method on the accuracy and trustworthiness of the estimates is evaluated, and used to make recommendations for future studies. The results show various interesting aspects. First, business as a travel purpose was found to be very special, and should thus ideally be modeled separately. Also shopping trips show different characteristics than the other purposes. Second, the VTT follows systematic patterns with changing agglomeration size, indicating that a differentiation on this level might be beneficial. Differentiating by local income, on the opposite, did not exhibit a recognisable pattern. Using a proven method to estimate the standard error of the estimated VTT, it was shown that for the formulation with the best model fit, a latent class model, accuracy in the VTT estimates was actually worse, illustrating how important error estimates are in the process of selecting the best model. A formulation of the VTT based on long term decisions did not bring any satisfying results. In particular, it seems that the other attributes that were part of the choice were valued much more strongly than changes in travel times. This thesis presents an in-depth and comprehensive evaluation of travel time with data from a national VTT study. The conclusions drawn from the analyses can be used as guidance for future transport project evaluation and policy making.