The Income Elasticity of the Value of Travel Time is Not One Number
M Börjesson, WSP Analysis & Strategy, SE; M Fosgerau, Technical University of Denmark, DK; S Algers, Royal Institute of Technology, SE
In 1994 a VoT study was undertaken in Sweden. In 2007 a new VoT study was undertaken. One objective was to study the income effect based on data from 2007 as well as from 1994. This paper describes the methods, the data collection and the results.
Value of time (VoT) is often considered as the most important benefit when assessing transport infrastructure projects. Stated Choice methods have now become something of a standard tool to estimate the monetary VoT. Knowledge on the relationship between demand and income growth as well as the relationship between VoT and income growth is therefore essential. In the literature, one important reference is the meta analysis by Shires and de Jong (2006), using a large number of cross sectional studies to estimate the dependence between VoT and income (among other factors). Another important reference is Fosgerau(2005), emphasizing the need to consider not only the change in VoT per se, but also the fact that income impacts covariates that influence the VoT as well. For forecasting as well as for estimation of VoT, random utility models have traditionally been used. From these, the VoT can be derived as the relation between time and cost parameters. The VoT may increase as an effect of lower marginal utility of money as well as an increased marginal utility of time. For assessment, this does not matter, but for fore¬casting, this may lead to different effects. So, to be able to make a proper project appraisal, we need to know not only how the VoT changes, but also to what extent this change reflects changes in marginal utilities of money and time.
In 1994, a VoT study was undertaken in Sweden by the national transport planning authorities (Algers et al 95) to facilitate improved VoT for project appraisal. In this study, a relatively weak relationship between income and VoT was found. The result was not convincing enough to make the authorities depart from their practice to assume a constant future VoT (a decision partly based on other considerations). In forecasting however, the results motivated a reduction of the marginal utility of money proportional to half the income growth.
In 2007, a new VoT study was commissioned by the Swedish transport planning authorities. One objective of this study was to study the income effect based also on time series data, i.e. based on data from the 2007 study as well as the 1994 study. This paper describes the methods used for this purpose, the data collection and the results.
We use the multinomial logit model to estimate cost and time parameters from Stated Choice data sets for 1994 and 2007, including covariates related to trip and trip maker attributes. We test for differences attributable to the survey year.
We also use a mixed logit model to estimate the VoT directly, based on the econometric model used by Fosgerau (2005). This model is based on the fact that the State Choice exercise uses only cost and time attributes, and thus can be formulated as the traveller accepting or declining a travel time saving bid. In this setting we similarly test the dependence of the VoT with respect to covariates and survey year, also recognising that there is an unsystematic variation in the VoT.
We finally develop an income elasticity also accounting for the dependence between income and travel distance.
We use the 1994 Swedish value of time study data for car trips, and the replication of this survey in 2007. The 2007 replication was made using exactly the same questionnaire as in 1994, sampling car travellers in the same way and in the same places, and using the same experimental design, with the sole change of increasing the cost levels by 40 percent, corresponding to real income growth and inflation. The sample size of 2007 (500 observations) was designed to make it possible to reject a unit income elasticity. The 1994 data contains about 700 observations.
The project is ongoing, but some preliminary analysis indicates that the main effect is related to the cost parameter. Final results will be available during the spring of 2009.
Algers, S, Lindqvist-Dillén, J, Widlert, S, (1995). The National Swedish Value of Time Study, paper presented at the PTRC European Transportation Forum in Warwick 1995. Proceedings of seminar F, P393 ISBN 0-86050-283-X.
Fosgerau, M. (2005) Unit income elasticity of the value of travel time savings.
a. NECTAR Conference, Las Palmas G.C. June 2-4, 2005.
Shire, J, and de Jong, G (2006) An international meta-analysis of values of time. Annex A to HEATCO, Deliverable 5, Proposal for Harmonised Guidelines, EU 6th framework program.
Association for European Transport