![Sample our Geography journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5937/TOC-Subject-Sample-2021-Geography.jpg"})
Abstract
A primary motivation of this paper is to draw together, in one source, information on the nature, extent and performance of Australia's evolving toll road network which is currently spread across many disparate published and unpublished sources. This paper provides key information (e.g. length, toll rates, year opened, operator(s) and payment alternatives) on all of the fully interoperable toll roads in Australia that are present in Sydney (e.g. the M2, M4, M5, etc.), Melbourne (CityLink and EastLink) and Brisbane (the Gateway Bridge, the Logan Motorway and the Gateway Extension). Where available, we compare and discuss actual traffic levels and forecasts, revealing the sizeable gap or ‘error’ in forecasts, especially during the first year of operation. Ordinary least squares regression and panel random effects regression models are developed to identify potential sources of explanation of differences in error forecasts between the Australian toll roads at various points post the opening date. A separate analysis of a sample of toll roads in the USA was also undertaken that supports the main findings from the Australian toll road network. Key influences on errors in forecasts are the capacity of a toll road, the elapsed time that the toll road has been in place, the specific period of time in which a tolled road is introduced into the network (which influences the complexity of route options including multiple tolled routes and hence toll saturation), the length of the tolled route, the presence of cash payment and the charging regime (i.e. fixed vs. distance‐based or variable user tolls).
Acknowledgements
The extensive comments of Demi Chung and advice from Leon Drew of Thiess and three referees are appreciated. Robert Bain also advised us during various stages of the research. This paper was prepared in response to a continual stream of requests to provide advice on Australia's toll roads.
Notes
1. Toll roads as a concept however are not new. For example, the world's first turnpike was approved in 1706, a section of the London‐Coventry‐Chester road in the UK (Willumsen, Citation2005); while in Australia, the first tollway (i.e. a bridge over South Creek at Windsor) was built in 1802 (Ernst & Young, Citation2008).
2. The policy motivation also includes value for money, off‐balance sheet commitments and greater control on costs.
3. The media in Australia has a field day every time a toll road is opened which fails to meets its forecast traffic levels within a few months. The media typically fails to (or does not want to) know that ramp‐up takes time—typically two years, and that any statements of failure are premature until ramp‐up is finalized.
4. Goods and Services Tax (GST) is a broad‐based tax of 10% on most goods, services and other items sold or consumed in Australia.
5. A tag is an electronic device that can be purchased from a tag provider and attached to the inside of a vehicle's windscreen. All tags work on all motorways in Australia. Different tag products are available from each of the motorway operators, all with different set‐up and running costs. When travelling through a toll collection point, sensors detect the tag and the toll is automatically charged to the toll account. There are no additional costs for using most tag products.
6. A pass is an arrangement between a driver and a motorway to use that motorway for an agreed period of time. A driver provides the number plate details to the motorway operator, and they will charge the toll and any applicable set‐up and administration fees to the driver's credit card or bank account. There may be additional charges for using a pass.
7. Sourced from NRMA http://www.mynrma.com.au.
8. This is a similar contribution to Port Botany (evaluated on a 2007 individual year net economic impact basis), the major port in the Sydney metropolitan area.
9. Complicated formulas (all of the formulae are tied to inflation, with one also tied to the change in average weekly earnings) have been inserted into the contracts the NSW Government has signed with tollway operators to gradually raise tolls until the projects, and their finance and operation, are paid off. They also have minimum rates should the consumer price index fall below a certain point.
10. Conventionally, toll escalations are indexed to CPI. Some road concessions apply other escalation formula (e.g. ED) due to other reasons such as political reason or pressure from the private consortia.
11. Statewide Roads operates and maintains two sections of the Motorway, a 2.5 km section between Silverwater Road and James Ruse Drive at Auburn and a 10 km section between May's Hill and Prospect. The RTA operates the remainder of the M4.
12. One of the NSW Auditor General reports that the locations of these plazas are on the sections that are not owned/built by the private sector. The reason was to allow the operator to capture more traffic in order to ensure the return to private capital is met.
13. The M5 is a missing link built between the two existing freeways, where the traffic pattern was well established before the opening of M5—it may be one of the reasons that the M5 has a good toll income that the operators do not wish to disturb.
14. For the payment matrix, please refer to http://www.westlinkm7.com.au/cmsAdmin/uploads/WL_toll_advert_Jun09_001.pdf
15. The maps are sourced from Queensland Motorways (www.qldmotorways.com.au).
16. Car: includes cars towing a trailer or caravan; LCV: any cab chassis 1.5–4.5 tonnes GVM, two axles; HCV: includes rigid trucks with three or more axles or over 4.5 tonnes GVM; buses with 13 or more seats including driver and articulated trucks.
17. For all charging rates, please go to http://www.eastlink.com.au/page.aspx?code=TOLLS#classes.
18. It is suitable for those who don't want a tag, including motorcyclists. The non‐tag accounts only work on EastLink and CityLink, and there is an image‐processing fee payable in addition to the toll for each trip made (23 cents per trip on EastLink, 75 cents per trip on CityLink). This fee is waived for motorcycles. The Breeze non‐tag account is a pre‐paid account, and will be available to be opened with a single opening payment of $40.
19. Toll prices are given in http://www.qldmotorways.com.au/www/index.aspx?ItemID=118.
20. Trip generation is usually predicted (as a linear extrapolation) from linear regression models in which traffic counts are a function of demographic and geographical data, which may be old and/or incomplete.
21. Forecasts of land‐use development are based on land‐use plans. However, the actual implementation in land use may be different.
22. In Næss et al. (Citation2006), all sampled roads in America are toll roads.
23. The ramp‐up duration is the period between the start of tolling and steady‐state operations.
24. Toll road forecasts may be rebased, given a poor opening‐year performance.
25. The M7 is a distance‐based toll road. For trips of less than 20 km, the toll is calculated on a cent per km rate (i.e. uncapped), and for trips of more than 20 km the toll is capped at 20 km (i.e. capped).
26. The sources of traffic numbers are given in Appendix B.
27. We were only able to evaluate Year 1 forecasts of these five toll roads in Australia, given the non‐availability of such data for the other toll roads.
28. The majority of toll road performance data (actual and predicted traffic) are obtained from Australian Securities Exchange release, Road and Traffic Authority, the Audit Office of New South Wales, a number of reports from consulting and toll road companies (e.g. Ernst & Young, Sydney Roads Group, Transurban, etc.), etc. Unpublished data are sourced from newspapers (e.g. Sydney Morning Herald, The Age and Toll Road News) and one unpublished paper from the School of Architecture, the University of Sydney. See Appendix B for all sources.
29. Different combinations of explanatory variables were tested, and the selected final model is the one with not only a good model fit statistically (e.g. R 2), but also a plausible explanation of underlying economic or behavioural processes.
30. The crucial issue in the toll road context is the fact that governments select concession operators on the basis of their confidence in the traffic and hence revenue forecasts, among other factors. For non‐tolled roads there is no selection process and indeed one might be critical of the rigour of assessment of forecast for roads provided and operated by the public sector. Bain (Citation2009a) found that “forecasts for toll‐free roads do not display the strong systematic tendency towards optimism bias” (p. 477).
31. Recent toll roads have also only been in operation for two or three years.
32. For a commuter working in the Central Business District, this amounts to $130 per week or $6240 per annum (based on a 48‐week working year).
33. The newly established toll roads (after 2000) are part of Sydney's Orbital Network such as the M7 and the LCT (see Figure in the paper). Another recently build tollway is the CCT which links Darling Harbour to Rushcutters Bay, avoiding 16 sets of traffic lights westbound and 18 eastbound.
34. Another type of regression models for panel data is the fixed effects model, which has toll road‐specific constants. This type of models requires within group variation in all explanatory variables for the fixed effects estimator, which is not appropriate for our sample (e.g. the length of a toll road is constant over time).
35. We have limited the investigation of the toll elasticity to the 28 observations where the toll is fixed. We do not have an average toll per trip from the roads with a distance‐based or variable user charge regime.
36. The car fleet is the main user of toll roads, for example, 93% of vehicles using the M5 are cars. Hence, car toll prices are used.
37. In the OLS model, the implied elasticity is −0.156, and is statistically significant at the 99% level of confidence. Since the sample size is less than 30 (28 and 12 in Table and Table , respectively), we used the Student's t‐distribution with n – k – 1 degrees of freedom (k is the number of regressors) to test critical values, instead of the normal distribution. The critical value (α = 5%) for n = 28 and k = 5 is 2.07, and 2.31 for n = 12 and k = 3.
38. Linton Besser, Transport Reporter for the Sydney Morning Herald, writes on 11 March 2009 that fluctuations between per kilometre prices of Sydney's major toll roads will get worse after the F3 extension and the M4 East are built. He suggests that this could be a huge barrier to implementing a workable congestion charging scheme in Sydney.
39. The global financial crisis has slowed the process of consortium evaluation and bidding down substantially, as of March 2009, with a number of these proposed new toll roads under a PPP partnership currently on hold.
40. This may have contributed to the erroneous forecast. In focus groups (see Chung et al., Citationforthcoming) one banker commented that the forecast errors were mainly due to the difficulty in estimating short trips, while long trips were indeed performing better than the forecast.
