A model was developed for the use of the French Ministry of Transportation, and financed by its main Directions (DRAST, DTT), with the objective of taking into account exogenous variables for predicting national transport activity one-to-two years ahead. The paper deals with a model for short-term total demand for land transport, relating passenger and freight traffic to their main determinants, representative of transport supply (prices, networks length) and demand (production/consumption). A modal approach that treats road and rail transport separately was followed by a bimodal road/rail approach. The model was constructed on a quarterly basis, over the period 1980-1998. In the modal approach, ARIMAX modelling with exogenous variables was used, and then vectorial auto-regressive modelling with exogenous variables (VARX) in the bimodal approach. The elasticity values of the modal traffic indicators with regard to their main determinants are discussed, and compared with other values obtained with annual models constructed on the same data, or obtained with other methods. The model's empirical performance is discussed. Although statistically superior, the bimodal form is not easy to interpret, the supply and demand effects being hard to separate. For the covering abstract see ITRD E126595.
Abstract