Towards a Better Appraisal of Transportation Policies : Some Evidence on the Willingness to Pay in France
M Jacquot, CETE de l'Est, T Anselme, CETE Normandie-Centre, FR
How much are drivers ready to pay for a shorter trip ? To answer this, an empirical analysis was conducted on route choices made by road users in France. And the result is : on long trips, they accommodate with very expensive tolls!
Two important issues for transportation planning are concerned with the value of time : firstly the effect of transportation costs on the agents behaviour and secondly the economic appraisal of transportation policies. Until now, little evidence was gathered in France on the first issue : the only available case-study dates back from 1996 and the whole framework for traffic assignment from the early 1960s. Furthermore the french guidelines dealing with the economic appraisal of motorway projects state that willingness to pay and richness generated by a gain of travel time should be the same, without giving any empirical evidence. It led to the use of standard values for the willingness to pay which might be not realistic.
To give a new insight on these issues, we focused on developing a whole calibration procedure for a traffic assignment model for trips with personal vehicles. A lot of data is indeed collected in France on the drivers route choice : Origin/Destination surveys are regularly conducted on many roads and motorways. Three different databases could be collected : a first one containing revealed preferences for trips in whole France for the period 1995 ? 2000, a second one with revealed preferences for trips in the region Normandie and the period 2007-2010 and a last one with stated preferences on the use of the different roads crossing the Vosges mountains (survey conducted in 2010).
Different assignment models were calibrated and compared. Firstly we investigated the current recommended formulation of the assignment : a deterministic model among efficient alternatives (according to a dual criteria of cost and travel time) with a lognormal distribution of the willingness to pay. While the calibration of not-measurable costs such as comfort or changes in the formulation of the generalised cost (operating costs, depreciation costs, speed by road types,...) didn't lead to significant improvements of the model, the model was very sensitive to the distribution of the willingness to pay, under the condition that its median value was a linear function of the trip distance. The distribution obtained is very different than the one commonly used : the parameter sigma is on the order of magnitude of 2 to 3 while it is commonly admitted that it should be near the value of the income distribution ranging from 0,5 to 0,7 ; the median values are very low for short trips (under 1 ?/h) and very high for trips above 400 km (above 300 ?/h).
Two other issues were analysed with this deterministic assignment model. Trip purposes, each with a special distribution of the willingness to pay, were introduced. While consistent distributions were obtained (for instance, business trips have the highest median for short trips), the improvement of the model was significant but small, due to the strong correlation between the trip purpose and the trip distance. The tests showed also that the model obtained for the dataset covering whole France could be to some extent transferred to other situations : once applied on the dataset for the region Normandie, it produced a reasonable fit of the data. Finally discrete choice models were tested in place of the deterministic model, with very different results among the databases, showing how the selection of a model formulation is specific to each particular case.
As a conclusion, this study can be seen as a comprehensive analysis of the calibration of an assignment model from a practical point of view. Some empirical values concerning the willingness to pay were obtained, which give a new insight on the effect of transportation costs on agents behaviour. This research aims thus to be an important contribution towards a better traffic forecasting on toll transportation projects.
Association for European Transport