A Disaggregate Freight Transport Model of Transport Chain and Shipment Size Choice

A Disaggregate Freight Transport Model of Transport Chain and Shipment Size Choice


E Windisch, Université Paris-Est, FR; G de Jong, Significance, NL; R van Nes, S Hoogendoorn, Delft University of Technology, NL


A freight transport model that simulates transport chain and shipment size choices in a simultaneous way is established. The model set-up accounts for a logistics perspective. MNL and NL models are estimated; influencing attributes are investigated.


"The objective of this study is to establish a disaggregate freight transport model, based on the unique and vast data source of the Swedish Commodity Flow Survey of the year 2004/05, which models the choices of transport chain and shipment size in a simultaneous way, hereby incorporating a logistics perspective. The proposed model set-up aims to overcome the lack of freight transport models taking account for current trends in logistics and to reflect real-life decision making as appropriate as possible.

Large effort was put in the presumable right interpretation of stated responses recorded in the Commodity Flow Survey, and in defining detailed cost functions for an appropriate estimation of cost attributes of choice alternatives. Multinomial logit models and nested logit models are then estimated significantly. Statistically better results are obtained by the NL models, indicating a very high correlation of shipment size alternatives and therefore a considerably higher substitution across shipment size alternatives than across transport chain alternatives in the choice process.

Several characteristics of sending units, of the shipments and of choice alternatives are identified to have significant influence on decision making. Most influential are the value density of a shipment, the commodity type and the cargo type of a shipment. Cost attributes of choice alternatives have mostly minor influence on decision making. However, results also show that a significant contribution to the utilities of choice alternatives remains unexplained. This unexplained part stems from unobserved attributes that point to the general preference of smaller shipment sizes. In current logistics development, new strategies like continuous replenishment, vendor managed inventory or just-in-sequence are applied and make frequent shipments, with therefore small shipment sizes, necessary. Such forms of agreements between senders and receivers remained unobserved in the data set, and are therefore probably reflected in relatively high alternative constants.

Trial applications of the model support the model?s applicability for policy simulations. Calculating cost elasticities by simulating cost increases and decreases shows that the latitude in meaningful policy measures is indeed extremely limited. Neither cost increases for lorry transport, nor cost decreases for any other transport chain, can provoke a remarkable shift in decision making that is targeted at a reduction on road freight transport.
Results of this study show that possibilities for influencing decisions in freight transport, specifically regarding the choice of mode usage, seem to be very limited.

The outcomes suggest that only changes in infrastructure (e.g. allowing sending units more direct access to rail facilities or waterways) could provoke an often longed for decline in road freight transport."


Association for European Transport