Gunn, 2001; Wardman, 2001; Hensher, 2001; Hensher, 2004). The non-marginal value of
time furthermore gives a valuation for the ‘actual’ travel time, for example the value for 20
minutes of travel time. However in general, literature focuses on the marginal value of time.
One important reason being that it is easier to derive a marginal than a non-marginal value of
time (i.e. via stated choice experiments). In this paper the focus also lies on the marginal value
of time, because (the valuation of) travel time ‘changes’ can be seen as an important cost or
benefit component caused by a road pricing measure.
The average VOT estimated for the entire sample (on basis of table 8) amounts to 2.5
euro/hour. This value is low compared to other VOT’s found in literature (Gunn, 2001;
Wardman, 2001; Hensher, 2001). However, these other VOT’s were in most cases derived
from stated choice experiments, focusing on short-term choices (route choice, mode choice
etcetera), whereas the choice experiment used in this experiment aims at long-term (i.e.
location) choices. Thus travel time does not seem to be a very important factor in a location
decision. In combination with a high dislike for travel costs (amongst which are toll costs) the
resulting value of time is low. Thus, focusing on location choices, respondents seem to prefer
relatively low (direct) monetary trip costs, whereas the travel time itself is of less importance.
In conclusion, travel cost seems an important component in location decisions. First of all
respondents are more sensitive to travel costs than to housing costs. In the second place the
low VOT indicates that respondents value travel time less negatively than travel costs. Overall
this may lead to the conclusion that respondents in general prefer to pay somewhat higher
housing costs and accept longer travel times in order to avoid (high) travel costs.
5.2 Location preferences and explanatory variables
Additional to section 5.1, this section describes logit estimation results in which explanatory
variables, such as socio-economic, demographic, trip and house related characteristics have
also been taken into account. This analysis therefore gives a more differentiated insight into
the importance of the trip and location related variables for different types of respondents.
The model results used in the analysis in this section are based on logit estimation and are
presented in table 11. An explanation for the acronyms used in table 11 is given in table 10.
Two types of models have been estimated. The left part of table 11 shows the estimation
based on using a multinomial logit (MNL) model. Only coefficients that are significant with a
18