The paper describes a model for forecasting freight transport on the Great Belt Fixed Link currently being built to connect East and West Denmark. The model is estimated on both revealed and stated preference data (RP and SP) using an approach later proposed in a paper by J Swait et al. The results provide justification for the approach. The SP method has been applied in a novel way to successfully estimate the size of the discretionary discounts given by ferry lines which by nature are confidential. Validation of SP estimates is given by the RP data. The model integrates route and mode choices in a nested multinomial logit model with three choice levels: mode, route choice for outgoing, loaded trucks and route choice outgoing, loaded trucks and route choice for returning empty trucks. Modes are different types of trucks, rail and combined transport, routes are the existing ferry lines and the new fixed link. Ferry fare structures are different for loaded and empty trucks so mode choice should be affected by the route choice accessibility for empty trucks. The model is formulated such that this dependence is achieved, while the empty and loaded route choices are otherwise independent. The formulation is interpreted to show that the combined empty and loaded route choice accessibility for a given mode expresses the cost per tonne of freight of sending a loaded truck with a fixed proportion returning empty. Furthermore, the results show that the combined empty and loaded route choice accessibility is very significant in explaining mode choice. Estimation is done with data from 1989 and the paper provides validation of the results with a backcast using 1993 data. Some examples of elasticities derived from the model are given and compared with elasticities from other sources.
Abstract