Conceptual model of the shippers choice between sea, rail and road transport

The main objective of the project is to increase the understanding of which factors influence the choice between road, sea and rail transport, with a particular focus on uncertainty in demand and lead time for each mode and whether this is a barrier against achieving a modal shift. The academic ambition has been to develop a new methodology in the intersection between the freight modelling literature and the inventory theory literature and calibrate it for Norwegian conditions. As far as we know this is the first inventory theoretic model that (1) takes into account that there are economies of scale in transport costs and explicitly includes the vehicle size as a choice variable, and (2) is calibrated using basic data from Statistics Norway and the National Freight Model for Norway to calculate the full cost of uncertainty in demand and lead time for a population of firms, as opposed to a single or limited set of firms. A dataset has been established based on register data from the foreign trade statistics, in which the full set of Norwegian and foreign firms transporting goods between Norway, Netherlands and Sweden is included, including zip codes for the sender as well as the receiver. We have calculated transport times and distances with road transport for all zip code pairs as well as to/from the nearest port and rail terminal. From the foreign trade statistics, we are also able to calculate the commodity value, shipment size and annual demand for each firm. We have quantified the distributions of demand per time unit and transport times for (1) direct road transport, (2) access/egress by road, (3) rail transport and (4) transport by sea. These data, as well as cost parameters for various components of the logistics costs are used to calibrate the model to imports and exports of commodities between Oslo/Bergen/Kristiansand and Gothenburg, and between Oslo and Amsterdam/Rotterdam. We find that the additional logistics cost due to uncertainty in average is about 5-10 percent for direct road transport, 20-30 percent for transport chains between Norway and the Netherlands by sea and 40-45 % for rail transport between Norway and Sweden. In all cases, the additional cost due to uncertainty is approximately 3-4 times greater for sea/rail than it is for direct road transport. This supports the hypothesis that uncertainty is an important barrier to transferring goods from road to sea or rail. Moreover, it implies that analyses that are not taking uncertainty into account systematically will overpredict the potential for transfer. We find that the additional cost due to uncertainty is highly heterogeneous among firms. For transport chains by sea it varies between 0 and 80 percent, while for transport chains by rail it varies by 0 and 100 percent. Firms with high annual demand, high commodity value and longer lead times have a higher percentual cost increase due to uncertainty, while firms with low annual demand have a higher cost increase in absolute value (this difference is due to the fact that small firms in general also have more expensive transport solutions). The most important mechanism responsible for this cost increase is long lead time in combination with demand uncertainty. When the demand is uncertain there is a long time until a shipment will arrive, the firms will either have to pay the cost of a safety stock or bear the potential costs of stock-outs. This is one of the most important barriers for transferring goods from road to sea or rail; the additional cost due to uncertainty can only to a small extent be decreased by reducing lead time uncertainty since the demand, i.e. the rate at which the inventory depletes, still will be uncertain. It is therefore longer lead times, and not more uncertain lead times, that increases the additional cost due to uncertainty for sea and rail the most. The model predicts that firms with low demand uncertainty, low stock-out costs per unit, low commodity values and/or firms with the opportunity to keep a safety stock without a large increase in inventory costs will be least negatively affected by uncertainty in rail and sea transport chains. This is in accordance with what we observe in reality, and also gives some insights as to which market segments that should be the focus of policies if the objective is transferring goods from road to sea or rail. Moreover, conclusions from the project indicates that the most effective measures for achieving this would be measures that reduce the lead time or inventory costs.

1) Vi har med prosjektet bragt forskningsfeltene om godsmodellering og lagerstyring under usikkerhet nærmere sammen. Dette er nybrottsarbeid som vi anser som et akademisk bidrag. 2) Så vidt vi vet er vi de første til å kalibrere en modell med usikker etterspørsel og usikker ledetid med faktiske data til en hel populasjon av bedrifter. 3) Mer spesifikt har prosjektet økt kunnskap om sammenheng mellom transport og usikkerhet. 4) Metoden og resultater er sett i sammenheng med den nasjonale godsmodellen, som transportetatene bruker til sine strategiske godsanalyser. Vi har også beskrevet en metodikk som gjør det mulig å ta usikkerhet inn i godsmodellen. 5) Prosjektet har bidratt til å ta i bruk nye datakilder, ta i bruk gamle datakilder på nye måter, og for opplæring av nye transportforskere når det gjelder bearbeiding av store datasett. 6) Prosjektet er motivert ut fra miljøhensyn, og belyser avveiinger som er viktige for de totale CO2-utslippene som følge av godstransport.

The primary objective is to better understand the factors that influence the shippers' decisions to ship goods by sea or by road and how they interact to produce a certain outcome. This is to be achieved by constructing a generic model of the single shipper's choice situation, and to embed it in an equilibrium model with many shippers and with network externalities, in the sense that aggregate volumes by each mode determine the offered freight rate and the freight rates influence the decision of each single shipper. The model will take full account of uncertainty in demand for the goods and in lead times. One of its features is a very detailed representation of transport costs. Another is the upper and lower bounds on vehicle size for the distribution stages and line haul stages. For each of the transport modes, the shippers choose vehicle size, shipment size (shipments may be smaller than the minimal vehicle size) and reorder point. A choice model produces the choice probabilities. Aggregating over shippers, the expected total transport volumes for each mode that determine the freight rates are produced. The procedure is repeated for the new freight rate until convergence. The necessary data for the model are provided by utilising the data sources and data bases that are established in TØI's work with the National Freight Transport Model, by a review of literature and by the new commodity flow survey which is due in spring, 2016. Tests with the model will be done and are expected to provide new insight in how the mode split could be influenced, and the social efficiency of doing so. It is also thought that this model may point forward to a new generation of freight models where uncertainty and safety stocks play a larger part, and where the transport supply is not exogenously given. The key element of the model has already been constructed, see Minken and Samstad (2006) (in Norwegian with an English summary)