Self-selection and the Value of Travel Time
S Mabit, M Fosgerau, Danish Technical University, DTU Transport, DK
The current paper formulates a model where mode choice is introduced as part of the value of travel time estimation to separate mode and user type effects.
We are often concerned with estimation of the value of travel time (VTT) as it plays an important role in demand models and appraisal. In most models, it is essential to recognise that the VTT is heterogeneous in the population. This may be across several dimensions, e.g., mode and income. As mentioned by Wardman (2004) it is difficult to separate such mode and user type effects. In many empirical studies the combined effect of the two effects is that individuals in car are seen to have higher VTT than for example in bus. The question is then if this happens because the mode choice of car raises VTT or because individuals with higher VTT choose car? If the latter is true then we have a self-selection problem and if this is not incorporated into the modelling then the VTT estimates may suffer from selection bias.
Self-selection and selection bias is recognised as an important concern in transport modelling. Still little work has been done to correct for this. Some exceptions are Dubin and McFadden (1984), de Jong (1990), and Train and Wilson (2007), and some effects of self-selection are discussed in Daly and Carrasco (2006). Only a few studies investigate the effect of self-selection on VTT estimation these are Eklof and Karlsson (1999), Mabit and Fosgerau (2006), and Fosgerau et al. (2007).
The contribution of the current paper is that we formulate a model where mode choice is introduced as part of the VTT estimation to separate mode and user type effects. In the paper we estimate VTT taking the model choice into account through a selection model that reflects the way people become part of the sample. The model consists of two sub models. A mixed logit (ML) model used to estimate VTT as is standard practice and a selection model to estimate the way individuals self select into the sample. In this respect the model follows the selection model introduced by Heckman (1979). Here the model type is generalised to allow for correlation between the selection equation and the coefficients in the ML model.
The present investigation differs from Mabit and Fosgerau (2006) in two ways. First, we test how to generalise the selection model using random probit. Second, we investigate a different sample to see if the results are more general or merely sample specific. We apply the model to a recent stated preference panel data set.
Our re-examination of the model serves to strengthen the general knowledge about how self-selection affects transport modelling and in particular it develops a model to separate user type effects from mode effects in the estimation of VTT.
Daly, A. and Carrasco, J. (2006), The influence of trip length on marginal time and money values, presented to Conference of International Association for Travel Behaviour Research, Kyoto.
Dubin, J. A. and McFadden, D. (1984), ?An Econometric Analysis of Residential Electric
Appliance Holdings and Consumption?, Econometrica, Vol. 52, pp. 345-362.
Eklof, J. and Karlsson, S. (1999), Testing and correcting for sample selection bias in discrete choice contingent valuation studies. SEE/EFI Working papers Series in Economics and Finance, No. 171.
Fosgerau, M., Hjort, K. and Lyk-Jensen, S. V. (2007), Between mode differences in the value of travel time: self-selection or strategic behaviour??, presented at European Transport Conference, Leiden.
Heckman, J. J. (1979), ?Sample selection bias as a specification error?, Econometrica 47 (1), 153?162.
Jong, G. C. de (1990), ?An indirect utility model of car ownership and private car use?, European Economic Review, 34, pp. 971-985.
Mabit, S. L. and Fosgerau, M. (2006), Controlling for sample selection in the estimation of the value of travel time, presented to Conference of International Association for Travel Behaviour Research, Kyoto.
Train, K. and Wilson, W. (2007), ?Estimation on stated-preference experiments constructed from revealed-preference choices?, Transportation Research B, forthcoming.
Wardman, M. (2004), ?Public transport values of time?, Transport policy 11, 363?377.
Association for European Transport