Cars & traffic

Dec 13, 2012

Who got the facts on traffic forecasts wrong?

The Fairfax press claims a consulting firm hired to work on Melbourne's planned East-West Link got the traffic forecasts for Brisbane's failing Airport Link toll road wrong. Is that right and is it a bad thing?

Alan Davies — Editor of The Urbanist

Alan Davies

Editor of The Urbanist

Forecasts for vehicle use of Brisbane's Airport Link. Source: Veith Lister Consulting (VLC)

In an article titled ‘Optimism bias’ on toll road plan published last week, The Age claimed a Qld company “that overestimated the traffic on Brisbane’s financially failed airport toll road has been hired to predict the traffic on Melbourne’s proposed $10 billion east-west toll road.”

Like most exercises that require gazing into the future, forecasting traffic is a hard business at the best of times. There are many variables at play and ample opportunity for the real world to depart from the assumptions made before the road in question opens for public use.

So it’s unrealistic to expect any forecast to be spot-on. However the projections relied upon by investors in Brisbane’s Airport Link toll road were a long way from spot-on.

The Product Disclosure Statement prepared by the proponents of Airport Link, BrisConnections, forecast 163,269 vehicles would use the new tunnel on an average day in the first three months after opening. That forecast assumed the full toll would apply but in fact no toll applied during this period.

As it turned out, actual usage during the three month toll free period averaged just 76,935 vehicles per day, less than half the forecast. So whoever did the forecasts got the numbers spectacularly wrong.

Neither Melbourne nor any other Australian city needs a problem like that. However it wasn’t the “Qld Company” referred to by The Age – Veitch Lister Consulting (VLC) – that got the forecast for Airport Link wrong.

In fact, The Age got it wrong. VLC had nothing to do with the forecasts that investors in Airport Link relied on. The Age does note in para 4 that “VLC was not employed” by BrisConnections, but that’s to cover its arse – the implication of the headline and lead paras is that VLC is culpable for Airport Link’s debacle (read The Age’s story here).

What actually happened is VLC prepared and published its own forecast for Airport Link as an entirely independent exercise, at its own cost. The company says it was an “in-house research project”.

I expect VLC was engaging in some aggressive marketing. The company probably expected the forecasts used by the proponent would turn out to be nonsense, making its own look good in comparison.

And they do. VLC’s forecasts are much closer to actual use than the proponent’s (see exhibit).

It’s still early days, but over the initial toll-free three month period, VLC says its forecast of average daily traffic, computed on the same basis as the proponents (except VLC assumed no toll), was between 75,460 and 91,600.

That’s between -2% and +19% of actual patronage. Pretty good for any forecasting exercise and very good compared to the proponents +213%, especially considering the proponent’s forecast assumed full tolling (over the entire 15 months!).

So VLC didn’t do the hopelessly optimistic forecasts that appear to be the undoing of Airport Link. More importantly from the point of view of Melburnians, the ones they did do have proven so far to be very good!

The Age got its story from Greens State Leader, Greg Barber, who the paper says:

called into question the Victorian government’s assertion that the east-west link would carry up to 100,000 vehicles a day.

Based on the estimates in the 2008 Eddington study, Investing in Transport, the preliminary Benefit to Cost ratio for the East-West Link is in the order of 0.5 i.e. it’s negative.

Now that it’s effectively committed to the project, the Victorian Government will of course be looking for ways to get that figure into positive territory. So Mr Barber’s concern is understandable.

Pointing the finger at VLC however is wrong. In fact, Mr Barber should be comforted, as I am, that the Government is hiring a firm that makes a public virtue of the accuracy of its forecasts.

I like the idea of a transport modelling consultant whose competitive advantage lies in the superiority of its forecasting technology. Some might also be comforted to learn that VLC did the transport modelling underlying the negative BC ratio estimated in the Eddington Study.

The best-known cases of enormous gaps between forecast and actual traffic involve private proponents seeking to attract capital. However, in the case of the East-West Link, VLC’s role is as an adviser to the Government.

As an aside, I don’t know where Greg Barber got the figure of 100,000 vpd on the East-West Link from, or even what it means.

(Visited 12 times, 1 visits today)


Leave a comment

Your email address will not be published. Required fields are marked *

21 thoughts on “Who got the facts on traffic forecasts wrong?

  1. Dudley Horscroft

    Not quite sure where this discussion is going. But apart from the issue of who is responsible for bad traffic forecasting (await results of court case(s), perhaps?), there is the question of how well the effects of a toll on traffic are built into the equations. This comes from “Streets Blog” in the USA:

    “Clark Williams-Derry of Seattle’s Sightline Institute once told me that the highway lobby often places more value on the time people waste in traffic than the drivers themselves do:

    “If you actually make people pay out of pocket to get a faster trip, people aren’t willing to spend that much. We have two bridges across Lake Washington. One has to be replaced. They put a toll on it to help pay for the project, and immediately traffic plummeted. People are driving around to get to other side, driving on the other road, sitting through congestion, timing their trips differently in order to avoid tolls. People really, if they have to pay out of pocket for a quicker trip, they don’t place much value on it.

    “So we’re expecting taxpayers to foot the bill for things the drivers themselves will not pay for.

    “An even more stunning example of this comes from another Seattle road project, the Alaskan Way Viaduct, projected to cost about $4 billion. “The total amount they think they can raise from toll-payers for this project is $200 million,” Williams-Derry said. “So we’re expecting taxpayers to pay 20 times as much as drivers themselves are going to pay.”

    “It’s what Yglesias calls the “push the costs onto other people” plan. The few who benefit from bad urban policy want the cost to be borne by the many who suffer from it.”

    Consider, for example the “Waste Connex” plan for Sydney. According to the “Open Road” it has a $10 billion price tag. Assume that when built it carries 400 000 cars per day (guesswork based on number of lanes compared to Sydney Harbour Bridge). Then those drivers should pay a toll such that the annual net revenue equals $500M per year (based on a 5% interest rate). This works out at $3.47. This is required for ‘NET’ revenue to cover interest on the capital costs to eternity. Consider that there is maintenance, policing (apparently all tunnels over about 200 m have to have a fully manned operations room to oversee traffic conditions and institute emergency responses) (so we were told when we pressed for a tunnel for the Sexton Hill By-Pass) and this leaves nothing for depreciation, etc, and the minimum toll could easily be $7.00. This figure I have seen as a figure that the ‘toll’ should be capped at. But at that toll will the traffic still be 400 000 cars per day? Perhaps 200 000? Then the toll should be more like $14.00 per day.

    It is worthwhile looking at the recently released Report by GHD for the Department of Infrastructure and Transport on over optimistic traffic forecasts- can’t access the URL while this cite is open but try googling “GHD | Revised Final report: Causes of over-optimistic patronage forecasts for selected recent toll road projects” which is the footer for the Report. Very Interesting!

  2. Austin M

    The black and white argument about traffic/transit demand is a bit old hat. With people basically labelling things based on if they support or oppose a project.

    Induced demand – Build a freeway/transit in the middle of nowhere connecting nowhere to nowhere and see how much demand is induced. My tip is none because these are not really trip generators in themselves. It is mostly the change in land use that dictates the inducement of demand. Building for induced demand is a statement about building unnecessarily ahead of the game and I think we are far removed from that position most of the time. Some dictatorships etc. on the other hand…

    Latent demand – That is when you build a road/transit and it is quickly filled because there was a pre-existing demand for its use. IMO this is where we often are and where we should be positioned with most infrastructure we build (its where the bang for buck is).

    Impending demand – What ties them both together the induced demand that’s coming (usually from land use change) that will pile on top of the existing latent demand.

    The fact is that the roads/transit we often build are politically popular with constituents (otherwise we wouldn’t build anything because it would be a waste of money for votes). Currently new roads are seen as failures by detractors as they either don’t have enough traffic (a waste) or have heaps of traffic and have induced demand away from more sustainable alternates and encouraged even more traffic through car dependance.

    I see a time coming soon where transit is also seen by detractors (especially inner city capacity upgrades) are seen the same way as they either don’t have enough use (a waste) or have heaps of use and have induced more people into an already overcrowded space. The argument being we should have spent a smaller sum of money to encourage the use of latent capacity to instead live local and work local thus being much more sustainable. The fact still remains though that “fixing” congestion on any mode buys votes and most people accept it improves a cities functionality over time.

  3. Alan Davies

    Another case where the facts (not revealed until later in the story of course) don’t support the sensationalism fabricated by The Age.

    By George, Warnie’s fired up

  4. Dylan Nicholson

    As I understand it the most efficient speed for modern engines is a steady 90-100 km/h. Certainly a lot more so than constantly accelerating from 0 to 60. But as you say, the longer distances thus traveled rule out many of the benefits.
    And sure, induced traffic isn’t seriously disputed by anyone – but you’re hardly going to make the argument that we should destroy all roads/railways/airports etc. to curb pollution-creating travel in order to reduce the pressure we put on the environment. The solution is actually relatively simple: ensure that everyone pays the full cost of travel, including as many externalities as can be reasonably estimated. Relatively simple but politically difficult enough that it will take us a long time to reach that point (especially considering it will almost certainly require more effective income distribution in order to ensure the most disadvantaged can still afford to travel).

  5. Tom the first and best


    Induced traffic is an internationally recognised phenomenon.

    Car are allowed on residential streets and other local streets and that is how they get to the freeways and tollways and thus they do increase traffic on local roads.

    I take it that you are not denying that higher speeds (as do occur on free/tollways) use more fuel.

  6. Dylan Nicholson

    Tom, I have a fair bit of respect for the work that has gone to into some of arguments on the ptua site, and have read many of them myself, but those ones are definitely not terribly convincing (cherry picking of examples being the most obvious flaw). If I had my way, cars would *only* be allowed on dedicated highways/freeways, and not allowed on to residential streets except in special circumstances 🙂

  7. Tom the first and best


    Freewanys do not create economic growth. Places with less freeways have more growth.

    Freeways do not reduce pollution, they increase it. Freeways cause traffic that would not otherwise have been there and this increase traffic on stop start roads. Freeways increase speeds which increase fuel use.

  8. Dylan Nicholson

    Tom, I’m no fan of freeways but they do help enable us to generate levels of economic wealth that can be then used to reverse or minimize the environmental damage we do as a species. And generally traffic moving on freeways will cause less pollution than traffic moving through connector roads with traffic lights etc. You aren’t going to stop people building freeways – the best we can hope for is that the externalities are fully priced in and that they’re seen only as part of a multi-mode transport network.

    Alan, no, sorry, I don’t think much of VLC’s track record at all! “Not as bad as some” isn’t good enough.

  9. Alan Davies

    Dylan Nicholson #11:

    readers should be rightly concerned about using forecasts from a company like VLC to determine whether such a major use of public funds is justifiable

    They should be very concerned about whether the EW Link is justifiable. They should therefore insist proper forecasts are done to inform the decision. But shouldn’t they be glad a company with VLC’s track record is doing the transport modelling rather than someone else with a more, er…pliable…track record?

  10. Tom the first and best


    Freeways do not have environmental benefits. There mean higher speeds, which use more fuel. They mean traffic that would not have otherwise happened (induced demand) that again uses more fuel. They mean massive construction related emissions and destruction of habitat. Freeways should be demolished not built.

  11. Dylan Nicholson

    I read it, I just wouldn’t use the word ‘wrong’. Even if the article was written in a way that could be a little misleading (and I don’t see much reason to assume it was intentional), readers should be rightly concerned about using forecasts from a company like VLC to determine whether such a major use of public funds is justifiable. The fact that even with a forecast that is most likely overly optimistic the numbers don’t really seem to add up is to me quite enough reason to look for better ways to spend the money – there’s no question there is a real congestion issue between the end of the Eastern Freeway and Flemington Rd, but the idea that one great long road-only tunnel is going to solve it seems so obviously contradicted by facts that I can sadly only assume it’s being pushed ahead for entirely the wrong sorts of reasons.

  12. Alan Davies

    Dylan Nicholson #9:

    What you and Greg aren’t getting is that the 120,000 VLC figure and the 163,269 BrisConnections figure aren’t estimated on the same basis i.e. they’re not comparable. They differ on day of week, ramp up, and application of toll. However when they are looked at on a comparable basis, VLC say their estimate is very close to actual use.

    The Age “got it wrong” for implying VLC was the consultant who did the numbers for Airport Link – see headline and first 3 paras of The Age’s story.


    A QUEENSLAND company that overestimated the traffic on Brisbane’s financially failed airport toll road has been hired to predict the traffic on Melbourne’s proposed $10 billion east-west toll road.

    Queensland-based Veitch Lister Consulting predicted in July that Brisbane’s Airport Link M7 would attract 120,000 vehicles a day toll-free when opened, taper off to 85,000 vehicles within six months and eventually sink to 60,000 vehicles a day.

    But the road, which opened four months ago, carried just 53,000 vehicles a day last month, less than two-thirds Veitch Lister’s forecast.

    Yes, that’s followed by these weasal words:

    Veitch Lister was not employed by the operator, BrisConnections, to do the modelling but did it independently. Its forecasts eclipsed BrisConnections’ own, which predicted that the road would gradually ramp up to 160,000 vehicles a day.

    I’ve already told you in my article about para 4 so you knew beforehand that VLC didn’t do the official forecasts, but if you weren’t alerted you might not pick up on it so easily. And given what para 4 says, why would the paper slant paras 1-3 the way it did?

    Note that the consultants who actually did the forecasts for BrisConnections aren’t mentioned anywhere in The Age’s article, so what implication are readers going to take about VLC?

    I suppose Greg’s “anti Green” criticism derives from articles like my recent less-than-enthusiastic one on the Greens HSR proposal. Politicans tend to see those who disagree with them as the enemy.

  13. Dylan Nicholson

    Alan, I still don’t see how the Age “got it wrong” – the 120,000 figure was forecast by VLC for the toll-free period, as per the link Greg supplied. And Greg – while I do agree I’m not sure what Alan has an issue with in this particular case, but I don’t see any need to attack him as being ideologically anti-Green (if he was, I wouldn’t bother reading/posting here). He’s a good person to have on your side, even if he won’t (and shouldn’t!) always agree with you.

  14. Greg Barber

    I’ll give ’em credit for one thing though – at least they predict traffic falling as tolls rise over the first year and a half. Brisconnect seems to have reversed the concept of price elasticity.

  15. Greg Barber

    Sure they do. And the table, (rather than the chart which you reproduced), shows exactly how they did it, after the road opened. The chart, by starting three months out, buries this former prediction. The really sweet part was Mike Veitch saying “It’s not rocket science”. No Mike, it’s much harder.

    The Age didn’t ‘misquote’ me on traffic numbers because they didn’t quote me on traffic numbers. You just read it that way. You did that because you wanted to, because on this blog, even when the Greens are right, they are still wrong.

  16. Alan Davies

    Greg Barber #5:

    No Greg, VLC’s update doesn’t refer to a new set of figures that conveniently revises its forecast after the event, as you suggest. Rather, the update compares the two forecasts on a like-for-like basis, most especially taking account of ramp-up and weekday vs every day.

    Sorry to hear The Age misquoted you on the 100,000 vpd figure. You should take that up with the paper. I certainly didn’t accept the figure at face value (see my last para!).

    As it happens I largely agree with the sentiments in your last three paras, but they aren’t the issues I discussed in this particular article.

  17. Greg Barber

    Veitch Lister predicted 120,000 vehicles per average weekday by the end of the three month toll free period (August to October 2012).

    Reality never exceeded 85,000.

    Within two weeks of tolls it fell to 59,127. Brisconnect’s most recent November report has it at that same level.

    Veitch Lister’s prediction for the Nov 2013 to April 2014 period is as follows:

    “VLC’s Zenith travel model suggests that towards the end of this low toll period, the road could carry in the order of 85,000 vehicles on a normal weekday”

    I don’t like their chances. Traffic volumes tend to settle down pretty quickly on these things and no amount of tweaking tolls or even changing surrounding road access will make a dramatic difference. If that’s what happens, then Veitch Lister will have performed about as well as the EastLink model, which now sees about two thirds of the predicted traffic.

    Alan, your Veitch Lister chart comes from the page headed “Zenith Airport Link Forecasts – Update”. That’s the NEW forecast they made AFTER the road opened.

    PS I’ve never said anything about an inner Melbourne east-west tunnel carrying 100,000 vpd. Funny how you’ve worked so hard to junk every claim in the Age article, but you take that one at face value. I have no idea where a number like that would come from but I wont quit until the government releases their forecasts.

    I do know that about 135,000 cars come off Melbourne’s Eastern freeway every day, but only a small proportion of those are going where the tunnel is supposed to go. See analysis here:

    And an even smaller proportion will be willing to pay a toll of a couple of dollars – BrisConnect certainly proved that.

    Now why don’t you turn your creative energies to the other statements in that Age article. Such as the RACV’s view that an official government forecast should be used for the tender bids. That way, we load all the traffic risk onto the public and the dills who invest in these things sue our arse off when Veitch Lister get it wrong. The private operator only wears the construction risk and the minimal operating costs.

  18. hk

    The economic survival of Melbourne, Victoria and dare I say Australia is dependent on an effective and sustainable freight transport network and operating system. Sure there needs to be a balance between the economic, environmental and equity factors in any predictive modelling of how much freight needs to be transported through the urban community.
    All the existing E-W transport flows need to be managed better. Maybe, as a first step and a better solution is to manage freight flows over the next twenty years by grade separating the Princes Street intersections at Lygon and Nicholson Streets with a lowering of the existing road surface.
    PT management would also benefit by grade separating traffic flows by lowering one of the roads at a few more intersections. There are several projects in Melbourne and the world where underground car parks and access roads are built over to generate sufficient revenue to justify private sector investment return to manage vehicular traffic and parking.

  19. Krammer56

    The real difficulty is actually the very sensible proposition behind a PPP project – the developer wins the right to build something and charge for it to make a profit, but it has to bear the risk that the revenue (e.g. tolls) might not be up to scratch.
    The major problem is the competitive environment that requires “winning” the rights. This means all bidders have to under-estimate the costs and over-estimate the revenue (volumes x toll) to have any chance of winning – hence the optimism bias.
    Where real risk-reward situations exist, PPPs can be a very good way to get infrastructure. This approach has generated some very good road projects (if you like road projects) at no (little) cost to the taxpayer. On a few of them, the developer has taken a real bath – most relevant are the two Sydney and two Brisbane stand-alone road tunnels (not a good omen for the EW Tunnel). On others, the high-risk construction elements have been completed at no cost to the taxpayer despite major cost overruns due to technical difficulties (e.g. City Link Tunnels).
    However the path Victoria seems to be heading down – a pseudo-PPP approach where most of the risk is retained by the State but the private sector makes what are effectively guaranteed profits – is not a sensible approach. The Peninsula Link is one example – construction risks were negligible (except maybe for the drought breaking) and the traffic risk is largely borne by the State as the Government, not the drivers, pay the tolls.
    There seems to be a softening up approach under way on the EW link in recognition that the PPP industry won’t fall for another tunnel financial failure. The RACV suggested the solution to over-estimates of traffic volumes was for the State to provide the forecasts – effectively guaranteeing the traffic levels and hence toll revenue. Even Tony (fiscally responsible) Abbott’s $1.5B (promised for a project with no plans, cost estimate, business case or environmental approvals) will only be a drop in the ocean.
    The EW Tunnel’s cost is just too high to be covered by tolls and will require major taxpayer $$ to get across the line. With the cost estimate rising, the chances of the benefit cost ratio being anywhere near 1.0 are negligible. This is especially the case if it is a toll road, as the economic and environmental benefits used to justify a new road are reduced if it is tolled because less traffic uses the new, freer-flowing road.

  20. Alan Davies

    Dudley Horscroft #1:

    Maybe it’s the inevitable outcome of Fairfax out-sourcing sub editing? The reporter files a fair story but some faceless sub puts a sensational and silly spin on it?

    Not sure about all those “anti-light rail people in authority”. Lots of new LRT in the US. Now even the O’Farrell governments building light rail from Circular Quay via George St to Randwick and Kingsford.

  21. Dudley Horscroft

    “In fact, The Age got it wrong.” What did you expect, it’s what the Fairfax Press does.

    But this is a general problem with roads and road related projects, you start out with the figure that will give you the return on the investment – the capital cost of the project – and guess a tariff for the vehicles, and then guess how many you will need to provide this return. Not a really logical way of doing it. With some busways in the USA, where the return is a ‘virtual’ one, ie, we assume that a toll would have been paid, and then assume a demand level, the results have been ludicrous. With light rail, perhaps because there are so many anti-light rail people in authority, the forecast demands have often been substantially exceeded. Of course, sometimes the ‘money-bags’ has been particularly cheese paring, and has cut essentials, such as the length of line (Buffalo), or the number of trams (Croydon) or demolished a housing estate on which the line was relying for patronage (Sheffield). Then people ask why the forecasts were not made.

    Suggest, do not rely on a promoter’s forecasts. Make your own judgement, and take all these forecasts with a pinch of salt. And if a politician or an ex-politician makes the pitch, hold on to your cheque book.

    I wonder if we shall ever see the full cost details for “Waste Connex”?

Share this article with a friend

Just fill out the fields below and we'll send your friend a link to this article along with a message from you.

Your details

Your friend's details