Update of the NEAC Modal-split Model
E van der Leest, M Duijnisveld, P Hilferink, NEA, NL
This paper presents the research for a new modal-split model for NEAC, a European Transport Forecast Model. Model assumptions, results of the model development and a comparison on methodology and results of the new and previous model are discussed.
The NEAC model and information system consists of models and methods for constructing databases describing the current freight transport flows and for developing forecasts of future freight transport flows. NEAC describes all intra- and inter-regional freight transport in Western and Eastern Europe by transport mode and by commodity type. Results are expressed in total weight of the goods. In NEAC, a classical four step modelling approach is used. This approach consists of the following steps:
? Trade and transport generation;
? Regional distribution;
? Mode choice;
? Route choice/assignment.
The third step, the model for mode choice, is the subject of this paper. Examples of applications of NEAC are transport flow analyses, corridor analyses, infrastructure analyses, market potential analyses and policy impact analyses. For example, in the TEN-STAC project, the Trans European Network was revised with the help of NEAC. More information about NEAC can be found on the NEAC website (www.nea.nl/neac).
In 2005, NEA carried out a research project to update the modal-split model. This paper presents the assumptions and results, highlighting the differences between the previous and the new version. The goal of the research project was to develop a model that could replace the outdated modal-split model. As well as the calibration of updated parameters the aim was to also incorporate the transport mode short sea in the model. To achieve this NEA has developed and calibrated a completely new model.
Research for the previous version of the model led to the conclusion that an aggregate analysis of the modal split on the basis of multinomial logit models renders no satisfactory results. The previous NEAC modal-split model therefore contained a segmentation of the transport markets. Within each segment of the market the development of the modal split was dependent on the transport time and the cost of transport of each mode in relation to other modes (cross elasticities). Segments were identified based on commodity groups, distance and total weight transported between regions. In general, the previous model worked in such a way that an increase in relative tariffs of one mode led to an increase in the proportion of the other modes (the magnitude depends on the elasticities). The previous model was based on the theory that the main determinants of mode choice are cost and time of transport, which in their turn are dependent on the state of the infrastructure and policy measures which have an effect on the transport market.
In the research project to update the modal-split model it was concluded that the previous version of the model was unsuitable for the more detailed regionalisation used in the NEAC database and the added transport mode short sea. The NEAC dataset used for the calibration of the new version of the model includes transport chains between 268 regions in 37 European countries specified for 11 commodity groups: NSTR 1 digit with crude oil separate; and 4 modes of transport: road, rail, inland waterways and short sea. It was concluded from the review of literature and applications of relevant models that the best approach for the new modal-split model would be the use of logit models combined with an adjusted segmentation of transport markets. This approach is also based on the theory that the main determinants of mode choice are cost and time of transport, which are dependent on the state of the infrastructure and policy measures which have an effect on the transport market.
The origin ? destination ? commodity group combinations in the dataset were divided into market segments using the chi-squared automatic interaction detection technique. The modal-split is the criterion for the segmentation and the available transport modes, the commodity group, the geographic area, the total annual transport volume and the distance between origin and destination are the predictors. The more homogeneous datasets of the market segments improve the explanatory power of the modal-split model.
The selected configuration of the model is a multinomial logit model and the parameters of the model have been calibrated separately for every segment. The explanatory variables used in the model are the alternative specific constant, the relative total cost per ton, the relative total time, the distance, the total annual transport volume, border resistance dummies for road and rail transport, domestic and intercontinental transport dummies and a main port region dummy. The dependent variable in the model is the market share of each available transport mode. In this paper an analysis of the effects of the new NEAC modal-split model on forecasts of main European transport flows is presented. The modal-split model for freight transport within the TRANS-TOOLS research project is based on the new NEAC model. The goal of TRANS-TOOLS, which is co-funded by the European Commission (DG-TREN) under the 6th Framework Programme for Research and Development, is to produce European transport network models to overcome the shortcomings of current European transport network models.
Association for European Transport