The story of toll roads operator Brisconnections whacked into administration is a warning about bad infrastructure projects that no one wants, inappropriate funding models and wildly optimistic traffic forecasts. It’s the same pattern played out with other big infrastructure investments – think Sydney’s Lane Cove and Cross City tunnels, which initially collapsed into receivership, and Brisbane’s Clem Jones tunnel which went into administration. According to a new global study, this might be par for the course.
The study, by Danish academic Bent Flyvbjerg from Oxford University’s Said Business School, found that traffic forecasts for infrastructure projects around the world are usually way off target. We’re not talking about 5 or 10 per cent out either – they can be wrong by more than 40 per cent, sometimes even way more.
Read his paper and you end up feeling that these forecasts might as well have been put together with a crystal ball, dart board and Ouija board. And the problem is investors, who wouldn’t take a punt on where the All Ords would be 10 years from now, thought the forecasts were more or less accurate and got burnt. The risks were downplayed by decision makers and if the study is right, what happened to Brisconnections was perfectly normal, and totally predictable.
That means two things:
Firstly, certain kinds of infrastructure are not as safe as many investors think. It’s a high-risk, high-return proposition from the conception to the operation state.
Secondly, governments and investors will have to rethink the kind of infrastructure we need, and re-evaluate the funding models. If we want infrastructure like roads and rail, we will have to develop a type of funding more robust to withstand forecasts that are totally off beam.
The study, Survival of the unfittest: why the worst infrastructure gets built – and what we can do about it , covered 208 projects in 14 countries on five continents. It found a standard deviation for traffic forecasts of 44.3 per cent. Put another way, it means that in about 68 per cent of cases, assuming a normal distribution on the bell curve, traffic forecasts will be out by 44.3 per cent. The remaining 32 per cent could be out by even more.
Now that variation could go up or down, the traffic could be more than forecast or way less. It’s just that in Australia so far, it’s been a case of the latter. Either way, the forecasts are wrong, it’s no precise science.
The Brisconnections debacle fits that model contained in the study. Airport Link's pre-GFC traffic forecasts had 136,000 vehicles per day using the road one month after opening, rising to 186,000 one year later. But only 81,000 used it in the first month, roughly in line with Flyvbjerg’s figures, and this slipped to 47,000. Lane Cove, Cross City and Clem Jones had a similar problem, the forecasts were way off target.
It doesn’t matter how sophisticated the modelling might be, it’s difficult to pick precise demand per day on a single piece of infrastructure five to 10 years’ time. Whether you’re predicting traffic demand or where the stock market will go, forecasting is a perilous business. As Flyvbjerg shows, consultants are paid millions of dollars to make predictions and without fail, they get it wrong.
In the Australian Financial Review last week, Queensland’s former Labor Treasurer Andrew Fraser wrote that Airport Link was a "zinger” of a deal and that the people of Queensland will forever have $5 billion worth of infrastructure. What he neglected to say was that it’s worth nothing close to that now. He also didn’t mention that the Bligh government only put in $270 million. It’s not like it had that much skin in the game to do its job and evaluate the traffic forecasts more critically.
Fraser said the forecasts were wrong because they were pre-GFC. But the Flyvbjerg study shows that’s a lousy excuse and sloppy rhetoric. The inaccuracy, Flyvbjerg says, has been constant over a 30-year period and there has been no sign of improvement. The GFC is just a red herring.
Flyvbjerg looks at a swag of infrastructure projects that went horribly wrong: the Eurotunnel which had produced a return on investment of -14.5 per cent and the Danish Great Belt rail tunnel, the second longest underwater rail tunnel in Europe, which turned out to be non-viable even before it opened. Then there was Boston’s Big Dig, LA’s subway, San Francisco’s Bay Bridge, Denver’s new International Airport, and the New Woodrow Wilson Bridge in Washington, DC. In Britain, there was the London Tube public–private partnership, the West Coast Main Line upgrade, the Railtrack fiscal collapse, the Scottish parliament building and the Humber Bridge.
Now the only reason you build roads, rails and bridges is to create a stronger economy. It’s difficult to identify major economic success directly attributed to these projects.
So how did we get this "survival of the unfittest”?
"They survived because benefit–cost ratios presented to investors and legislators were hugely inflated, deliberately or not,’’ Flyvbjerg writes.
Robert Bianchi, a senior lecturer at Griffith Business School and expert in PPPs, says governments and investors need to be more rigorous in their evaluation of infrastructure. Good infrastructure, he says, has to provide economic benefit and has to be, unlike Airport Link, heavily in demand. If it’s not, don’t build.
"If you knew 10 years ago the real demand for these projects was only a third of the demand you had expected, they would not have been built. With the costs involved, they become financially unviable,’’ Bianchi says.
"Governments have a fiduciary duty to every long-term investor in this country. Instead, what they’re doing is saying the company that has the largest cash flow gets the deal but that may not be any good."
He says that building anything underpinned by such variable forecasts needs to be financed in such a way that it’s going to be robust to that variation. Instead of 80 per cent debt and 20 per cent equity, he suggests it should be around the other way.
Greater equity would make investors more aware of the risks they’re running. Still, he concedes that’s problematic if there was trouble and the share price tanked. Nonetheless, the business would still be operating and the bond holders would be less likely to end up on the hook.
If we want infrastructure driving economic growth, we need to change the model.
In the end, it would come down to a more rigorous evaluation of the costs and benefits of infrastructure we need. Forecasts, with all their variation, and business cases should be made subject to independent review and stakeholders, including super funds, should be allowed to evaluate them. Greater transparency would help deal with unpredictability of forecasts as detailed in the Flyvbjerg study. All these steps would produce a better result than the dart-throwing we’ve seen so far.