The Value of Travel Time: Random Utility versus Random Valuation

This paper identifies, relates and compares two popular modelling approaches to estimate the value of travel time changes. The first (random utility) assumes that the random component of the model relates to the difference between the utilities of travel options; the second (random valuation) assumes that it relates to the difference between the value of travel time and the suggested valuation threshold. The existence of the random valuation approach is related to the nature of a particular type of stated choice data: binary choices where only two attributes are traded off. A majority of national studies on the value of travel time conducted in Europe use this kind of data. This paper gives details on the theoretical relationship between the two approaches and compares them empirically at several levels of model sophistication. This ensures fairness in the comparison and allows us to disentangle the real differences between the approaches and among levels of sophistication, in terms of value of travel time and model fit. The models are estimated on two datasets corresponding to national studies in the UK and Denmark, obtained through stated choice designs built on the same basis. The results show a consistent superiority of the random valuation approach and a systematic gap in the value of travel time between approaches regardless the level of model sophistication. A surprisingly similar pattern across models is found in both countries. This raises questions about the validity of results using the random utility approach and about the role of the stated choice design.