Assessing the Influence of Distance‐based Charges on Freight Transporters

Abstract

Congestion charging has become a high‐agenda theme in many urban environments, with a growing recognition of a need to commit more effort to establishing the impact that various charging regimes might have in reducing traffic congestion and, as a corollary, in raising revenue that can be used to re‐invest to improve transport infrastructure in general and public transport services in particular. In addition to the political commitment, a major challenge being faced is behavioural—a need to understand more fully the role that specific charging regimes might play. The paper investigates the potential influence of variable user charges in the freight distribution chain. A choice‐modelling framework is presented that identifies potential responses from the freight transport firm to distance‐based charging within the context of the wider spectrum of costs and benefits delivered in terms of travel time savings and increased trip time reliability.

Acknowledgements

Support for this research was provided by the Australian Research Council Discovery Program (Grant No. DP0208269) on Freight Transport and the Environment. Discussions with John Rose and Stephane Hess are appreciated. Two referees are thanked for their comments.

Notes

1. In 2015 it is estimated that congestion costs in the largest Australian cities will be as follows: Brisbane, A$9.3 billion, A$4600/head; Sydney, A$8.8 billion, A$2000/head; and Melbourne, A$8.0 billion, A$2100/head (Gargett and Cosgrove, 1999). These estimates include costs of delays and higher fuel consumption, but exclude costs of extra accidents and higher vehicle emissions linked to congestion.

2. Results from Sweden’s experiment as of May 2006 show that car traffic to and from the inner city has fallen by 25% since the scheme was introduced. Public transport patronage has increased by 8% since last year, which translates into a daily increase of 50 000 passengers.

3. The ‘Eurovignette’ directive lays down common rules on how European Union states may charge heavy goods vehicles for using the road network. The main objective is to ensure that road usage better reflects its true impact on society and the environment at large by introducing a ‘user pays’ and a ‘polluter pays’ principle. It also aims to shift freight away from roads onto other modes of transport such as rail and waterways.

4. Since this study was completed, the Western Motorway in Sydney (M7) has opened. It has a distance‐based charge with a cap of close to A$13 for trucks over its 45 km of new infrastructure.

5. The summary of trip details that appears when clicking on ‘Trip Details’ includes: the name of the client or freight firm involved, the type of truck used, the primary contents of the truck, the amount paid for the delivery of goods, the kilometres travelled, the last location of loading before delivery, the total number of locations at which the truck delivered goods, the allowable lead time, the time from the request of delivery to the departure of the truck, and (in the case of questionnaires given to sampled clients) the value of the cargo. This last element is omitted from questionnaires given to representatives of freight firms as they are not prompted for this information.

6. As the tasks are likely to involve some unfamiliar terms, respondents are given definitions of the terms ‘attribute’ and ‘alternative’, and informed that a showcard is available for any unfamiliar terms in the choice sets. Respondents were also informed that any details relating either to the trip or to the relationship between the two firms that are not shown in the choice sets can be found by clicking on the buttons labelled ‘Trip Details’ and ‘Relationship Details’, respectively.

7. As set out in Greene et al. (2006), the random parameters specification can accommodate correlation amongst the alternatives. Since β _q can contain alternative specific constants which may be correlated, this specification can induce correlation across alternatives. It follows that the model does not impose the IIA assumption. Restrictions can be imposed at numerous points in the model to produce a wide variety of specifications.

8. Preliminary analysis revealed that the degree of heterogeneity in reference trips was sufficiently high that some outliers obscured the inferential power of the data. After careful consideration, the following observations were removed from the final sample: (1) trips based on a fuel efficiency of over 101 litres/100 km (or approximately twice the average fuel consumption for the larger trucks in the sample); (2) trips based on a probability of on‐time arrival less than 33%; (3) round trips (or tours) of less than 50 km; and (4) round trips of more than 600 km. The trips eliminated, based on low fuel efficiency, may have obscured the results due to significantly prohibitive values for fuel cost and variable charges, reflecting reference trips that are too atypical to be pooled with other trips. An alternative source of obscuring effects via low fuel efficiency may be that the implied values of fuel efficiency were inaccurate, and hence either made the trade‐offs implausible to respondents or reflect an inability of the respondent to offer meaningful information on which to base the alternatives. The trips eliminated, based on the low probability of on‐time arrival, are likely to have obscured the results because the trips involved travel quality significantly worse than the remainder of the sample, making the pooling of these trips into the sample problematic. Similarly, extremely short or long trips may have involved trade‐offs that are significantly different to the trade‐offs made by respondents in the sample at large.

9. The spread is the standard deviation .

10. For a triangular with mean = spread, the density starts at zero, rises linearly to the mean, and then declines to zero again at twice the mean. It is peaked, as would be expected. It is bounded below at zero, bounded above at a reasonable value that is estimated, and is symmetrical such that the mean is easy to interpret.

11. Road transit time often only represents a small proportion of the total order lead time; and many other non‐transport factors can exert as much if not greater influence on reliability.

12. The assumption of a single respondent is just for convenience and comparative reasons and does not have any further implications. Any other sample size could have been used, but it is common in the literature to base it on a single respondent.

13. The summary of relationship details that appears when clicking on ‘Relationship Details’ includes: the length of the relationship between the two organizations, their contractual arrangement, the organizations that have input into the routing and scheduling of the trip, and (in the case of respondents representing freight firms) the proportion of business represented by the relationship with the client. This last element is omitted from questionnaires involving sampled clients as they may not know this information in the marketplace.