Estimating values of travel time savings for toll roads: Avoiding a common error

Traditionally, the empirical valuation of travel time savings (VTTS) is obtained from a linear utility specification in a discrete choice model, which implicitly assumes a risk-neutral attitude. This paper draws on recent contributions by the authors that accommodate the attitude towards risk within a non-linear utility specification as a preferred framework within which to value travel time savings. The interest in the non-linear form is motivated by the evidence in Hensher et al. (2011) that mean estimates of VTTS in a proposed toll road context are significantly lower when account is taken of risk attitude. The percentage reduction in the estimate of mean VTTS is approximately (coincidentally) equal to the actual percentage error in traffic forecasts associated with the new tollroad two years after opening. If we could show that this evidence of a lower mean estimate under the non-linear treatment is found in other data settings, then we gain confidence in suggesting that the linear-utility assumption to valuing travel time savings might be a potential contributor to over-predicted tollroad traffic forecasts. The non-linear model is applied herein to two other tollroad choice data sets and we find that sampled car commuters tend to be risk taking when decision making is subject to risk (due to the presence of variability in travel times). The model produces lower mean VTTS estimates than the traditional (linear) model, providing additional evidence of a systematic over-prediction of VTTS under the linear assumption. This paper suggests that future empirical studies on valuing time savings (and variability) should address the attitude towards risk.

► Identifies a limitation of the traditional framework for valuing travel time savings.
► We propose an improved modelling framework in which risk attitude is included.
► Evidence from three stated choice data sets shows systematically lower mean VTTS estimates from the new model.
► A first study that investigates the potential contributor to tollroad demand forecast errors of allowing for risk attitude.