High-speed rail: an expensive hobby

By David Walker (Google profile)

First posted at Club Troppo, but expanded here

I like trains. For a while when I was a kid, I spent Saturdays clambering around Adelaide's Mile End Railway Museum and most of my pocket money buying items for an elaborate train set. Which may explain how I found myself reading KPMG's Phase 2 report in the federal government's High Speed Rail Study, looking at a high-speed rail line from my home town of Melbourne to Sydney and on to Brisbane.

Sadly, the Phase 2 report essentially restates, with arithmetic, everything that most analysts keep concluding about a Melbourne-Sydney-Brisbane High Speed Rail (MSB HSR) project. It would be, like my old train set, a bit of a money pit. It would be very expensive ($114 billion in 2012 dollars), would never earn back its capital costs, would take decades to build, and would not be very green. My train-loving inner child would get a huge buzz from seeing shiny locomotives surging through the countryside, but its economics don't stack up.

It's useful to have this point repeated. There's a stream of public commentary that says, essentially, that anyone who opposes high-speed rail is just lacking in vision, because, well, to quote a cringeworthy lead paragraph from The Age's transport reporter, "High speed rail would be the best". The low point of this vein of commentary was probably Gordon Weiss's evidence-free paean to HSR in the Global Mail in 2012, memorable mostly for a bizarre string of approving references to totalitarian governments: "Benito Mussolini not only made trains run on time, he made them run fast ... Unfortunately, Mussolini's equally manic support for Hitler interrupted HSR technology ... the Central Committee changed [China's rail system] with Mussolini-like bravura."

When that's the quality of policy debate, a little realism about costs and benefits should be welcome.

Costs may be underestimated

HSR supporters object to several aspects of the study. In particular, they argue that the $114 billion estimate is way too high. Hey, another study says we could do the job for a mere $70 billion!

This is an unlikely criticism to anyone who understands the recent trends in major developed-country infrastructure projects. The clear trend is for projects to come in way over their original cost estimates.

Why's that? The best current analysis of the nature of budget overruns on large projects comes from Danish economic geographer Bent Flyvbjerg, co-author of the book Megaprojects and Risk: An Anatomy of Ambition. Flyvbjerg's numbers show that big cost overruns are the norm. He speculates that they happen because the all the pressure on the promoters is to lie about the numbers. As Megaprojects puts it:

"[The Aalborg study] and other studies of cost development in major transport infrastructure projects reveal the same overall pattern: cost overruns above 40 per cent are common, especially for rail projects, and overruns above 80 per cent are not uncommon … Cost overrun today is the same order of magnitude as it was ten, thirty or seventy years ago. If techniques and skills for estimating costs and avoiding cost overrun in transport infrastructure projects have improved over time, this does not show in the data. No learning seems to take place in this important and highly costly sector of public and private decision making. This seems strange and invites speculation that the persistent existence over time and space and project type of significant and widespread cost overrun is a sign that equilibrium has been reached: strong incentives and weak disincentives for cost underestimation and and thus for cost overrun may have taught promoter what there is to learn, namely that cost underestimation and overrun pay off. If this is the case, cost overrun must be expected and it must be expected to be intentional.”

So while the MSB HSR may come in at under $100 billion, that is not the way to bet. (Flyvbjerg and his co-authors also found that rail project proposals systematically overestimated passenger numbers. Flyvbjerg sets out his views succingtly in the freely available 2009 paper Survival of the Fittest: why the worst infrastructure gets built—and what we can do about it.) 

Developed-world infrastructure is expensive

And why so are big infrastructure projects so expensive in developed-world nations, even before the cost overruns? We don't know for sure, but the University of Minnesota’s David Levinson has compiled a list of 39 hypotheses about developed-world infrastructure build costs. The leading candidates (not that I would see them all as important) include:

• Commitment to tunnelling for road and rail rather than tearing up big chunks of cities, My first guess is that this is the biggest single factor.
• High build standards generally (safety, social, environmental, architectural).
• One-off nature of the projects. (China, building 20,000 kilometres of HSR lines, will get some economies of scale, as has Europe to a lesser extent. Australia won't.)
• High labour costs.
• High land costs.
• Process costs to get the power to acquire and use land. (The Chinese government can happily tell 1.3 million people to move when it wants to build a dam; Canberra rightly cannot .)
• Insufficient commitment to detailed benefit/cost analysis, which in turn leads to a get-it-done-at-any-cost mentality.
• Lack of competition among construction services providers.
• Poor government oversight.
• Political pressure to add extra features (tunnels to protect creeks and forests, extra stops at smallish towns etc)

I would add that the Sydney-Brisbane HSR line faces particularly unfriendly natural terrain, compared to most of China and anywhere else.

By far the best public analysis of the high-speed rail issue appears to be Alan Davies’ work at The Urbanist, and he does not appear very surprised by KMPG’s estimate.

So I think it’s hard to argue those numbers are on their face ridiculous.

Environmental benefits are thin

Just as bad is the tendency to ignore the report's conclusions on the MSB HSR's environmental impacts. "I’d think HSR would have considerable ... environmental benefits which don’t seem to be mentioned in the report" suggested one Club Troppo commenter. Actually, the environmental benefits are in chapter three of the report, which has a sub-section titled "System-wide environmental impacts during operation", and in Appendix 5G, which goes through the modelling of those impacts. The problem for HSR boosters is the environmental story that the report tells.

The report claims, for instance, that an MSB HSR will cause a net increase in emissions. Why?

Partly because building a new rail system uses a heap of cement and steel, which are high-emission materials.

And partly because when you build new pieces onto an existing traffic network, you encourage activity on other parts of the network. Rather than competing with car travel, as some HSR supporters assume. high-speed rail is complementary to it. (Most modes of transport are complemetary too each other, mobility begetting more mobility.)

But mostly because right now there's a lot of unmet demand for flights out of Sydney Airport. When we build our rail line, we won't get less traffic through Sydney Airport; we'll just get new traffic replacing old.

This is a variant of the point often made by critics of new freeways: they don't reduce congestion, because they pull new users onto the roads. (Note by the way that we do get benefits in these situations; those benefits just don't include reduced congestion.) Because this point is so often made by anti-freeway, pro-rail commentators, I suspect it will be a little hard for HSR fans to deny.

You can construct scenarios where HSR reduces emissions.

• One is the scenario where Sydney gets a second airport, so that HSR is no longer filling unmet Sydney airport demand. But of course an extra Sydney airport makes the economic case for HSR weaker. And even in this scenario, HSR remains an extraordinarily expensive way to reduce CO2 emissions.
• In another scenario, we produce all our electricity from renewables. The Phase 2 study ddin't look at this, but should have; it would make the HSR look at least somewhat greener.

But the report argues reasonably convincingly that HSR is not a great tool for emission reduction. Yes, when it's up and running the MSB HSR should create less greenhouse gas emissions than air and road travel (though this depends on the trains being pretty full). But HSR ends up producing a lot of net new transport activity, which means emissions go up rather than down. The report estimates "a net increase in overall transport emissions of 32 million tonnes of CO2–e over the evaluation period up to 2085".

Summary: HSR still doesn't stack up

High-speed rail has rarely been economically compelling anywhere. But like many forms of infrastructure, it has most appeal when you are starting with a relatively clean state and high population densities, like Japan after World War II. In Australia, with its built-up cities and long inter-city distances, HSR makes less sense.

For Australia, high-speed rail looks a little like a very expensive hobby. We have far better ways to spend $100 billion.

Addendum: The World Bank on high-speed rail

In 2010 the World Bank - an organisation historically very friendly to huge infrastructure projects - published a paper on high-speed rail and economic development. Its conclusion neatly summarises what I believe to be the best view of high-speed rail, and is worth repeating:

[H]igh-speed rail is now a tried and tested technology that delivers real transport benefits and can dominate market share against road and airline transport over the medium distances that many inter-city travelers confront. However, the demographic and economic circumstances that could support the viability of high-speed rail are, in global terms, limited. The number of passenger transport corridors of the requisite length, that are already capacity constrained, and where there is sufficiently dense potential demand by people of adequate purchasing power, is limited; some may be in countries where the implementation capacity may be lacking.

The combination of supportive features that exist on the eastern plains of China such as the very high population density, rapidly growing disposable incomes, and the prevalence of many large cities in reasonable proximity to one another (creating not just one city-pair but a string of such pairs) are not found in most developing countries. Nor could all countries assemble the focused, collective capacitybuilding effort, and the economies of scale that result, that arise when a government can commit the country, politically and economically, to a decades-long program over a vast land area. Even in China, the sustainability of railway debt arising from the program as it proceeds will need to be closely monitored and payback periods will not be short, as they cannot be for such “lumpy” and long-lived assets. But a combination of those factors that create favorable conditions of both demand and supply, comes together in China in a way that is distinctly favorable to delivering a successful high-speed rail system.