Report: 'Have all the time savings been achieved?'
The working paper by Dr John Odgers ('Have all the time savings been achieved?') is published by GAMUT to encourage an informed and vigorous debate about the fundamental issues surrounding infrastructure construction. The paper raises important issues. Responses are published below.
Philip Hay (received 04/09/2009)
This paper is of no great academic merit and of no value to the much needed debate on public transport expenditure until it addresses the cost of not investing in the freeway system, or cost alternatives. A retrospective analysis should consider the cost of transport in a Melbourne with the road system of 1995 with traffic flows of 2009, and some other realistic scenarios as well (to answer the question : what could we have done better?)
The use of gross average metrics in a study of this type is disappointing. Speed and trip time frequency distributions should be considered. Freeways alter distance travelled as well as speed. Reductions in the variability of trip times have a societal and economic benefit.
Essentially, the study suggests that after 14 years of population and vehicle usage increase, travel time have remained steady, which many would consider a good result.
Matt Mushalik (29/09/2009)
(1) You have to distinguish between 2 types of time savings:
1.1 Private motorists
1.2 Commercial traffic
The time savings from 1.1 are virtual. They will not translate into cash
dollars in the economy. By contrast time savings of commercial traffic will indeed turn into
additional turnover.
(2) Peak oil (2005-2008) will change the whole transport world.
Please visit my web site:http://www.crudeoilpeak.com. Time for private motorists will not matter, so there will be huge economic losses after peak oil.
(3) I did a slide show at MAV in the treasury theatre:
http://www.mav.asn.au/CA256C320013CB4B/Lookup/Smart_ConferenceAug09/$file/Smart_Conf_Aug_09.pdf> If you want to have a look at my slide show, please let me know
Professor Dr Max Lay AM, FTSE (received 06/10/2009)
This response to the suggestion on the Gamut web-site has also been prompted by requests from Nick Low and Bill Russell for my comments.
1. I disagree with the view that this is an important paper. In my view it misses key marks, as I will try to illustrate below.
1a. However, my basic thesis is that what should happen in transport and what has usually happened with Melbourne's road system is that the system is developed within a strategic and policy framework, with individual components then being tested against economic and social criteria. In many cases, these criteria are best seen as filters eliminating bad projects and as prioritising tools for good projects. It is wrong to see community transport investment as primarily an economic activity.
2. It has been common in the past for many authors to note drops in travel times over time, particularly in papers about increasing congestion. Often such authors have then used the observation to argue for more investment in transport (particularly in roads, I will admit). [I have found most of those papers to be very debatable, for reasons not relevant to this discussion.] The Odgers paper certainly draws different and relatively unique conclusions from such previous observations.
3. New transport investments do induce new traffic. Usually, models account for this - the main exception was the extraordinary omission of induced traffic from British models in the second half of the last century.
4. Transport is a derived demand and that demand will increase as economies increase and as populations grow. Over time we need more schools, more hospitals and more transport, or the existing facilities become over-used. Often market mechanisms respond to this increase in demand, but transport infrastructure is largely supplied by governments who do not respond to markets with nearly the same speed. 4a. In my book Melbourne Miles in 1992 I made a study of previous Melbourne traffic planning forecasts. I showed there how important population changes were and how most of these forecasts had been very wrong. The examples in Melbourne Miles also demonstrate to major impacts of land use changes. 4b. In his paper Odgers makes the claim: 'Indeed, based on the evidence presented and analysed in this paper, one could be led to the conclusion that investments in Melbourne’s urban road network have resulted in more time being used by Melbourne’s motorists rather than less time'. It is as if nothing else happened in Melbourne during this time, other than roads were built. It is as if the population did not expand and the city did not become larger and more prosperous. Just as well Odgers does not apply the same logic to the falling performance on Melbourne's rail system. Indeed, I find Odgers' Figure 3 to be very supportive of the merits of recent road investments and Figure 4 to in part reflect changes to public transport usage.
5. Governments do respond to electoral pressures and in the last century these largely related to road provision.
6. In the very first sentence of his paper Odgers says: The foremost economic benefit postulated and claimed for all road network investments is the value of travel time saved. It would have been useful if Odgers had given some references for this unsupported claim. In the 4th (2009) edition of my Handbook of Road Technology, I discuss the benefits of road projects (Chapter 5) and note, for instance, that vehicle operating cost reductions can be of at least the same order as travel time savings. In addition, enhancements to land values provide an important alternative measure.
7. In this context it is interesting that the first road scheme announced in Victoria after I read the Odgers paper a month or so ago was the Western Link, which Minister Pallas justified in his press statements as fulfilling a "strategic" need. No hint of travel time reductions.
8. It should also be noted that travel times are at a minimum in free flow conditions. The maximum traffic throughput is at speeds well below free-flow. Thus aggregate travel time savings do not have a one-to-one link with travel speeds. In addition, the travel time savings used in benefit-cost analyses are the differences in time on two routes – specifically, the route used before the project was implemented and the route used after implementation. A uniform drop in travel times would not alter the difference. Odgers gives no indication that he has explored this difference and to do so would require a detailed probing of the models used, as the City Link roads do not have a one-to-one correspondence with the pre-existing roads
8a. Odgers points out that the average speeds on Melbourne freeways have been about 80 km/h (with a speed limit of 100 km/h to produce a very skewed distribution). He fails to mention that 80 km/h is close to the speed at which a road delivers its maximum traffic throughput (my Handbook, Chap 17).
8b. In an ideal urban road system, no one would be travelling at over 80 km/h. The concerns would be with the management of traffic at the ends of the links. Many - including me - have argued that most urban roads should be designed to operate at 80 km/h.
8c. I cannot imagine any design concept for an urban road system based on travel speeds. The concepts are, rather, based on providing accessibility in both a land-use sense and in the "strategic" sense referred to by Minister Pallas (point 7). Travel time and vehicle operating cost savings are useful measures, but they have to be seen in the larger strategic context. Of course, I do admit that they do creep into press releases from time-to-time as - for the layman - they are readily understood measures.
9. In addition, much of the benefit of a road (particularly from truck traffic) is delivered by operations outside the peak times of personal travel.
10. In his second sentence, Odgers says
'This paper’s aim is to empirically test whether the very substantial economic resources that have been consumed over the last two or so decades in the construction and use of major road network additions in Melbourne have helped to achieve the travel time savings which formed the main foundation of their economic justification'.
If this view were to be tested, it cannot be done by a time-series analysis for the reasons noted in Comment #4 above, unless one removes the confounding effects of all the other events that have occurred over the time period, such as population and area increases. The alternative test would surely be to observe what happens when one of the investments is removed from the system for a day (e.g. with one of the occasional closures of Burnley Tunnel on City Link during which the city almost comes to a halt.) It would be instructive to see Odgers' NPV calculations for the time differences if Burnley tunnel were to be removed.
10a. Odgers also says:
'Hence major road infrastructure initiatives and the consequent economic investments have not yet delivered a net economic benefit to either Melbourne’s motorists or the Victorian community'.
This statement also does not pass the "if we did not have it" test. Nor does it reflect the land value increases associated with these investments, as most graphically illustrated by the Western Ring Road.
11. I also debate Odgers' further attempts to continue to make the above point. For example, take the statement:
'… the results from this study suggest that the core of travel times savings benefits, which is an increase in average travel speeds…'
However, as a simple test recall that time = speed times distance. Thus, the author's claim only holds if distance is constant. But if a new route is shorter, then there can still be a time savings, even if speeds are lower. Odgers partly recognises this in the paper, but can find no distance data to quantify his view.
12. Odgers uses City Link as his prime example. However City Link was not advocated for its travel time savings, but for its strategic intent. In 2002 the author and Daley published a definitive review of City Link in the international journal Transport Policy. It lists the five key purposes of City Link. All are strategic and none involve reduced travel times. Similarly, pages 9 to 17 of a City Link review publication (Tagaza 2002) extensively covers the lead-up to City Link, including statements by Ministers and review panels before the project was approved. This fulsome review makes not one mention of travel time savings! Where is the substance of Odgers’ rather Quixotic windmill?
I can also recount my personal experience as a bureaucrat initially presenting the City Link concept to the State minister of the time. The proposition was about the need to clear the south bank of the Yarra of traffic (e.g. queues of trucks on Linlithgow Ave). That promise has been well delivered.
13. There is nothing in either Table 8 or Figure 7 to even vaguely support the following "clear" statement about Eastlink by Odgers. In fact, EastLink's travel time savings have recently been independently assessed. The lack of travel time savings that were projected and used to justify the economic benefits that would result from both City Link and more recently East Link is clearly evident in both table 8 and in figure 7.
14. Odgers says: 'Figure 5 presents the reported data for the years 1996-97 to 2006-07 for the urban road network monitored by Vic Roads. Based on these data the projected increase in flow of traffic does not appear to have been achieved'. However, the figure does not support the claim - assuming that the unlabelled vertical axis does refer to percent variability - instead it suggests to me the system is maintaining its level of service, particularly as many travelers rate variability highly as their estimate of trip time must be average time plus likely negative variability.
15. I should also add that I am disturbed by the accuracy that Odgers' ascribes to most of his data. For instance, there is much evidence (Hensher in press, Flyvberg 2003, Melbourne Miles) that traffic forecasts are typically +/- 25%. Similarly, flow measurements are scaled-up from smaller counts. And yet Odgers gives data up to 5 significant figures and average travel speeds over the network, for instance, to fractions of a km/h.
16. Odgers' final paragraph is: 'In closing, perhaps the German word Schlimmbesserung — meaning an improvement that makes things worse — is an apt descriptor for the massive program of new road construction that has marked Melbourne’s ‘solution’ to its transport challenges over the last several decades'.
17. My closing critique is that in my view, and having worked in and observed traffic in many cities, Melbourne's traffic moves reasonably well and has benefited by the steady development of its road network. I cannot imagine how the city could operate without at least its current modest network and, although I support major investment in public transport, it will have little impact for a decade and only minor impact when completed. Many travelers have no option but to use the road system, and little prospect of this changing. As Cox said earlier this year at the National Infrastructure Conference (Cox 2009): So do we have the balance right in the number and length of freeways in our capital cities? The major factor in this discussion is that, because freeway travel is only 70% of the cost of travel on the arterial road network, then cities with a higher percentage of freeway travel will have lower business travel costs than other cities – and will therefore be a more competitive city in the global economy. In my opinion, this paper is of no great academic merit and would be of no value to the much needed debate on public transport expenditure until it addresses the cost of not investing in the freeway system, or cost alternatives.. A retrospective analysis should consider the cost of transport in a Melbourne with the road system of 1995 with traffic flows of 2009, and some other realistic scenarios as well (to answer the question : what could we have done better?)
The use of gross average metrics in a study of this type is disappointing. Speed and trip time frequency distributions should be considered. Freeways alter distance travelled as well as speed. Reductions in the variability of trip times have a societal and economic benefit.
Essentially, the study suggests that with after 14 years of population and vehicle usage increase, travel time have remained steady, which many would consider a good result.
References:
Cox, J. 2009. DO WE HAVE THE BALANCE OF FREEWAYS RIGHT IN AUSTRALIAN CITIES, National infrastructure Conference, April
Daley. K. F. & Lay, M. G. 2002. The Melbourne City Link Project. Transport Policy, 9:261-7.
Lay, M. G. 2009. Handbook of Road Technology. 4th Edn. London: Spon (Taylor & Francis)
Tagaza, E. 2002. "Journey and arrival: the story of the Melbourne City Link." North Melbourne: IEAust
Please contact: Danielle Hearn: dhearn@unimelb.edu.au