Essays in Transport Economics.

Abstract

The PhD thesis consists of four self-contained chapters in the area of Transport Economics. The main aim of the thesis is not to produce a single message which is supported by all four chapters. Rather, each chapter is written to make a contribution of its own. The thesis covers a wide range of issues such as modelling behavioural reactions to travel time variability, the measurement of the cost of travel time variability, the labour market implication of changes in commute costs, and the application of discrete choice models to investigate variations in willingness to pay for travel information systems across individuals and the implication of model assumptions on the estimated distribution. Chapter 1 is titled: “Testing the slope model of scheduling preferences with stated preference data”, and is a joint work with Katrine Hjorth and Jeppe Rich. This study used a stated preference data to challenge the theoretical equivalence of two methods for measuring the value of travel time variability: the slope model of the scheduling approach (Fosgerau & Engelson, 2011) against its reduced form model. The analysis is based on data from two choice experiments that are identical except one has a fixed departure time while the other allows respondents to choose their optimal departure time. According to the scheduling model, the two experiments yield the same result if travellers can freely choose departure time to maximise utility, and if the distribution of travel times is independent of departure times. It turns out that the empirical results in this paper do not support the theoretical equivalence of the two models as the implied value of travel time variability under the reduced form model is an order of magnitude larger. This finding is robust and is in line with a recent Swedish study by Börjesson et al. (2012). Because of data better suited for the analysis, we ruled out some potential explanations lined up by past research for the observed discrepancy between the two models. Although the similarity of results across studies could suggest the presence of a more fundamental problem in estimating the valuation of travel time variability based on data from hypothetical experiments, it is recommended to test the equivalence of the models based on real life data before we can rule out hypothetical bias as a potential explanation for the discrepancy. (A paper based on this chapter was presented at the 3rd Symposium for the European Association for Research in Transportation, Leeds, UK, 10-12 September, 2014.) Chapter 2 is titled “Valuation of travel time variability with endogenous scheduling iv of a meeting time”, and is a joint work with Mogens Fosgerau. The chapter involves a theoretical model to examine the choice of an optimal meeting time in a situation where individuals can freely choose meeting times. It extends the model of Fosgerau et al. (2014) by introducing a notion of a designated meeting time and a penalty that may be imposed when one arrives later than the meeting time. Such a meeting time can be obtained as an agreement outcome in a bargaining process over potential meeting times. The model considers two individuals who choose departure and meeting times in the presence of uncertain travel times for a trip towards a joint meeting. An important feature of the model is the physical property that a meeting starts only when both individuals arrive at the destination. The study shows the existence of a unique optimal meeting time and a unique Nash equilibrium in departure times. It finds that an increase in the variance of the difference between individual travel times is costly for both individuals. It also find that an increase in travel time variance of one person is costly for both. Compared to Fosgerau et al. (2014), the introduction of a lateness penalty allows an additional mechanism through which a change in travel time variance of one individual affects the pay-off of both individuals. (Previous versions of this paper were presented at the 2nd Symposium of the European Association for Research in Transportation, Stockholm, 4-6 Sept, 2013; and at the ITEA’s Annual Conference and Summer School on Transportation Economics, Toulouse, 2–6 June, 2014.) This paper is related the scheduling model in chapter 1: both models consider scheduling choices in the presence of travel time variability. They differ in two important respects: First, whereas the model in this chapter allows individuals to choose a meeting time, the slope model assumes a fixed arrival time. Moreover, while the slope model takes scheduling choices merely as a personal matter, the model in this chapter allows strategic interaction in scheduling choice. As a result, the slope model does not capture the effect of improved variability of travel times for one person on another. Chapter 3 is titled:“Advanced methods make a difference: A case of the distribution of willingness to pay for advanced traveller information systems”. This study is concerned with the use of discrete choice models to estimate the distribution of willingness to pay for advanced traveller information systems and the implication of certain model assumptions on the estimated distribution of willingness to pay. The study uses a flexible estimation v method based on data from a stated choice experiment designed to measure the willingness to pay for several types of information that an advanced traveller information system can provide. Different models were estimates that vary in terms of restrictions embodied. While simpler and relatively more advanced models yield nicely dispersed distribution for willingness to pay, this distribution ceased to exist when some restrictions are set free. The less restrictive model fitted the data better, and in this model, which combines the latent class and mixed logit models, it turns out that the data do not reveal any dispersion in the willingness to pay for advanced traveller information systems. Results indicate that a significant share of individuals is unwilling to pay for advanced traveller information systems and that willingness to pay is tightly distributed among those who are willing to pay a positive amount. Findings in this study illustrate the importance of model specification testing, and that results regarding the estimated distribution of willingness to pay can be highly dependent on restrictions built into the model. (A paper based on this chapter is under review at Transportation Research Part C: Emerging Technologies, and was presented at the 94th Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., 11-14 January 2015.) Chapter 4 is titled “The effect of a firm’s relocation distance on worker turnover”, and is a joint work with Ismir Mulalic, Jos van Ommeren and Ninette Pilegaard. Using a matched worker-firm data from Denmark for the years 2000-2007, this study examines whether and how much a firm’s relocation distance is related to worker turnover. Firm relocation alters the pattern of commutes to workers: some workers benefit from shortened commutes while other face lengthened commutes. The costs of residential mobility and long distance commuting could induce those whose face lengthened commutes to move jobs. Those who faced longer commutes incur higher commuting costs; and they can minimise these costs by moving residence to shorten commutes or by moving to a nearby job. When the costs of residential mobility and long distance commuting are higher, job mobility becomes a more attractive proposition. The analysis finds a positive and significant but moderate effect of relocation distance on worker turnover. This effect is robust to the inclusion of firm level characteristics and year and municipality fixed effects. Results in this chapter establish that, on average, a 10 km increase in relocation distance leads to a 2–4 percent increase in the annual rate vi of worker turnover at the firm level over a period of three years, including the year of relocation. The estimated effect is stronger in the first year after relocation and pales away after the third year as workers more or less fully adjust to the relocation. It is not surprising that we obtained a smaller effect since, first most firms relocated locally. Second, the high rate of job mobility in Denmark means that workers expect to be mobile in the labour market; hence, it may matter less when their firm relocates. Moreover, it is possible that workers knew about the relocation decision and left the firm in the years and months before the relocation. The study also examines whether the distance of relocation captures the effect of changes at the firm because of the relocation. Results indicate that, after controlling for relocation distance, firm relocation has no significant effect on worker turnover.