Menu lock

Transport - general

Feb 16, 2012

Belated Valentine's Day message: Debunking another spurious correlation........ Dilbert, by Scott Adams

In my last post, I looked at the enormous differences in the cost of subways between some countries and cities. The “take home” message is to be careful extrapolating overseas experience to Australia because there is immense variability across projects.

A closely related issue though, is the remarkable cost escalation of what on the surface are the same projects. The cost of the California High Speed Rail (HSR) project has just doubled to $65 billion. In Australia, the capital cost of Sydney-Melbourne HSR went from an estimated $25 billion in 2008 to circa $60 billion in the Federal Government’s new study, released in late 2011. Some argue these estimates are so exorbitant they must be a deliberate ploy to sabotage new public transport projects.

Fortunately, Dr David Levinson at the University of Minnesota started an exercise to collect all the reasons why transport projects in the US have become so expensive. He cites the $175,000 average cost for traffic signalling equipment at intersections in the US. “It sure seems like we should be able to build this cheaper”, he says. “Think about it, $175K for 12 lightbulbs on a timer. What’s going wrong?”

He advances eight possible explanations or hypotheses (this is my summary – his full explanation is here):

  • Higher standards. “Our obsession with safety, features, environmental protection, and quality drive up the cost……”
  • Accountability. “Public works agencies are spending Other People’s Money, and so are less motivated to get value for the dollar than an individual consumer on their own……..”
  • Lack of competition. “There is no or eBay for public works. I cannot go on Amazon and buy a transit bus or an interchange……..”
  • Limited economies of scale. “When everything is bespoke, there is no opportunity for standardization and economies of scale…..”
  • Inadequate scoping. “We have investments that don’t match actual demands……We have big buses serving few passengers. We have overgrown highways. We have a fear of building too small and having congestion or crowding so we build too big.”
  • Asymmetries in benefits and costs. “Benefits are concentrated, costs are diffuse. As a result, the known beneficiaries lobby hard for projects, but not just to build it, but to build it in a way that is expensive…..”
  • Remoteness. “Remote actors cannot have precise information about local conditions and in the absence of a free market in transportation (there is generally one buyer, who is generally a government agency), prices are not clear. As a result these remote actors misallocate because they are misinformed.……”
  • Deficiencies in analysis. “No one actually does B/C analysis……”.

His comments on higher standards are especially interesting:

Engineering design is often 20% of project costs. If only we would tolerate a few more deaths, a bus without AC, pollution, and frequent breakdowns, our initial costs would be lower. But when do reasonable investments become gold plating? Does the firetruck really need to do a 360 degree turn on the cul-de-sac, or can it back out?

Dr Lisa Schweizer from the University of Southern California, Los Angeles, provides four supplementary hypotheses (this is also a summary – full description here):

  • The highest demand areas for maintenance and new stock occur in places that are expensive. “I wonder how much of the costs of, say, intersections have to do with land costs….the Northeast Corridor and California links of the proposed high speed rail make the most sense in terms of service and users, but they are also the most expensive to build”
  • Project creep. “Standards have risen……noise walls hither and thither, etc. It’s hard for me to say that these costs aren’t necessary because the politics of getting something built pretty much requires the outlay”
  • Envy is a much bigger problem in public works than in personal life. “Jurisdiction X got a light rail link. I pay taxes for those things, why does Jurisdiction X get it when my neighborhood/district doesn’t?”
  • Benefit cost is only as good as the integrity of the data and the analysts. “With nonuser benefits and nonmarket benefits thrown in, the b/c ratio is a politically constructed number”

Based on comments from readers and input from others, Dr Levinson increases his list to a whopping 38 explanations (although I think there’s a bit of double counting). Here’re some of them:

  • Starchitecture. The original design for the WTC PATH rail station in New York was $3.3 billion, subsequently scaled back to $2.2 billion. Even so, says Stephen Smith, that’s more than Paris spent on its entire new 9km Metro Line 14
  • The need to build while keeping current operations going. According to Dr Levinson “This reduces construction space, reducing time, increases set-up/break-down costs, and otherwise adds to total costs. Construction is much faster (and thus cheaper) if rebuilding could be done on a closed facility. See the London Underground as the classic example of the high cost of doing construction only at night and weekends, but keeping the line in operation”

These two are from reader David King:

  • “Public-private partnerships trade additional up front costs for faster construction. See this story
  • “Open government/costs of democracy. The planning process is required by law to bring in as many stakeholders as possible. This has (potentially) led to transportation investment being sought and justified for non-transportation concerns. Transportation investment is now used for social, moral and economic goals that are not directly related to mobility”.

And there are more from others:

  • Baumol’s Disease: “It involves a rise of salaries in jobs that have experienced no increase of labor productivity in response to rising salaries in other jobs which did experience such labor productivity growth”
  • Experience. One readers says: “I think the reason has to do with experience and competence. We have no high speed rail lines, so any high speed rail line built in the US will be a “one off”. Same applies to subways and light rail. If you don’t do something regularly, you never develop an expertise that will reduce costs because you are constantly reinventing the wheel. The US has become a “stupid” country. We are paying the price for being stupid.”

In Australia, transport projects have to compete for resources with the booming mining industry. As implied in the comments section of my last post on this topic (Why do subways cost more here than elsewhere?), another possible explanation is the small size of the industry in Australia. This gives scope for collusion although it’s probably more likely to manifest as lack of competition. There’s also the possibility that public sector clients are unknowingly co-opted by major contractors i.e. they unwittingly take on the same world view as the contractors and suppliers.

Update 27 Feb 2012: David Levinson has updated his original list, see here. And again – now up to 43 reasons.

We recommend

From around the web

Powered by Taboola


Leave a comment

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

10 thoughts on “Why is infrastructure so bloody expensive?

  1. Thteribl

    Cheap, quick or good …. Choose any two !

  2. Dudley Horscroft

    I suspect that both the Bellistener and Socrates figures are far too low. Even single tram track is Christchurch is costing $5M per km. No utilities to shift on the Redcliffe route, but the ground appears – from the NearMap view – to be largely swamp – so double it for poor soil conditions and the need to embank much of it. That brings you up to $126M for single track. Quite a few bridges, probably at least a dozen – or level crossings, so add $120M for them.

    With stations over 2 km apart on average, existing trains should be able to average 50 km/h taking about 15 minutes from Petrie to Kippa-Ring. Use higher power (all three coaches powered) and this could perhaps be cut for one three car train to make a 30 minute interval service, or two trains on a 15 minute interval service, so for 6 coaches say $18 to $20M. Signalling? With single track and a loop in the middle, signalling does not need to cost $1M per km, as it is only at the loop that there need to be signals to regulate entry and exit into the single track sections (level crossings need standard traffic lights for road and rail traffic). Say $1M for this. Add $2M for overhead and the total in round figures is $270M.

    Add stations – 6 at $1M each – and that is pushing the price up greatly! Train storage – a siding exists at Petrie Station for this, and it would in any case be reasonable to store one set at Petrie on the line and the other set at Kippa-Ring in the station. Maintenance – QR already has suitable maintenance facilities. Let’s call the hardware costs a round $300M.

    Engineeering design allow 10% of hard costs, rather more than your 4%, gives an extra $30M – all done in house by QR. Legal, and all other paperwork costs, say another $20M, total now $350M. Then that great big black hole – provision for unforseen costs, usually around 20%. So add another $70M, total now $420M.

    We have now got a railway costing $420M that can provide a 15 minute interval service to connect with Brisbane trains at Petrie. Who gets the other $700M or $1200M? Your guess is as good as mine, but it does not need to be there.

    I opt for single track on the basis that it is at the end of the line and passenger demand has, in the past, been presumed to be too low to justify a rail line. So ST is more appropriate than double track. And if ST is used the construction cost is near but not quite, halved. With comparatively low demand, three car trains should suffice, so platforms need be no more than three cars long. If demand becomes high, then stations can be lengthened to accommodate 6 car trains running through to Brisbane. And if it is still higher, then the line can be doubled so you can run a 10 or 5 or 2 minute interval service. But is this likely? No, so don’t prepare for a demand that is hardly likely to eventuate. Remember the Airport line is largely single track, and the line to Robina was opened as single track and doubled later.

  3. Socrates


    Your rail cost figures are well out of date. Rail track in an urban setting (fenced) costs $2 million per track km, more over soft ground like around Redcliffe. Add another million per km for signalling, and $2million for power supply. Stations are $10 to $15million each thanks to DDA legislation. Then $15m per train for rolling stock, bridges, and train storage and maintenance. Maybe the cost exaggeration problem is going on at Redcliffe but I doubt you could do it for under a billion. The Seaford extension in Adelaide is costing $250 million for 5km.

  4. Socrates

    Thanks for the fascinating thread. I should declare a bias here because I am an engineer working in transport planning. I would add the following comments:

    Engineering costs – there is a definitional problem here that hides the real issues. Actual engineering design rarely costs more than 4% on road or rail projects (slightly more on small jobs, but never more than 10%). Yet overheads do total 20%. Why? Bureaucracy. Most government agencies charge around 8% of project costs as their project admin. Other departmental costs are obviously being hidden in this. Further, because most govt. agencies have lost their skill base, they now hire separate third party verifiers (+6%). The rest goes in planning, environmental and consultation processes.

    Managing perception – agree strongly with the problem of massaging figures to appear “clever” for delivering an over costed project on budget.

    PPPs are very costly, with no evidence they reduce costs or risks, which can never entirely be evaded by govt. for essential services. There is an enormous cost in legal and finance fees. I have reviewed PPP proposals where the financier wanted 6% just to arrange (not supply) the finance.

    Obviously more standardisation would save money. Here I do find the role of some of my fellow engineers embarrassing, with an absurd duplication of standards based on no more than parochialism and, I suspect, ego. Everyone knows we have different rail track gauges. Did you know we also have different codes for rail signalling in each state? And different brake systems, suspension, and even power voltage for electric trains. Why? A narrow minded industry protects its turf in each state. Why do we even have state standards for what is safe? The laws of physics do not vary from Sydney to Melbourne.

    Finally I agree with the comment on land costs. The more developers dominate planning agencies, the less land is put aside for future corridors. The corridor for Melbourne Eastlink was put aside in the 1950s. The land corridor for the m7 was valued at $300+ million. Alternatively, if there is no corridor, tunnelling doubles or even triples project cost.

    I disagree with the comment on quality. Building something well rarely costs much more, but may double the asset life. The USA has a whole generation of cheap, poorly maintained bridges that will cost billions to repair or replace, or more will fall into rivers.

  5. Bellistner


    so a project that should cost a billion (say), is costed at two billion.

    should say

    so a project that should cost a billion (say), is costed at two billion, confident that even if the project blows out by half a billion to $1.5bn, it’ll still be $0.5bn ‘under budget’, which would cue the smiling faces at a ribbon-cutting ceremony and self-congratulatory press releases and sound bites

  6. Bellistner

    Part of the high cost is, I think fear by Governments of being critised by the media, so a project that should cost a billion (say), is costed at two billion. Having found out what the budget is, the contractors then bid at or around the two billion mark. Then there’s a blowout somewhere along the line and the additional cost is applied to the two billion rather than taken out of the profit from the realistic one billion cost (and the blowout is agin costed at twice what it should really cost).

    An example is rail. All three levels of Government are contributing to the Redcliffe rail link (a branch promised for a hundred years, finally getting built). The cost has increased for the ‘original’ 2010 budget of $1.16bn to somewhere in the region of $1.6bn, while the actual cost of preparing the formation, buying and laying the track, sleepers and jewelery, and comissioning the signalling should cost well under half a million a kilometre (for a total cost of some $12.6m). Even adding in the cost of six stations and a few bridges, you’re still a magnitude out from the ‘final’ cost.

    Perhaps the ideal situation is to have a private entity do the whole shebang, with an on-the-quiet govt-backed loan, and a handover to the government after 50 years or so (similar to Airtrain, except that they have a 35yr handover and it was privately financed). Without the overt inclusion of tape-cutting government ministers, costs are lower because the private enterprise has a limited amount of capital it can raise.

    However, the ‘ideal situation’ rarely come about, for the simple reason that while operations on infrastructure are profitable, the actual infrastructure itself rarely is.

  7. Russ

    When the Southern Cross development was taking place the train from Sydney became a bus from Albury. This was noted on the ticket by referring to “the $700 million redevelopment of Southern Cross station”. Which raises the question of why the cost of a project merited a mention in relation to the inconvenience?

    Actually the Hollow Men answered that.

    The major difference in forward budgeting between France/Spain and Australia/Britain/USA is that the former has a long-term plan for certain infrastructure with the intention of building up a network over 50+ years. the latter have dozens of individual projects that undergo C/B analysis some of which are built, but where the knowledge/expertise/equipment from one project doesn’t get carried over onto the next. When the former run into problems it affects the speed of fulfillment of the overarching plan, which annoys comunities waiting on their projects. When the latter run into problems it blows a whole in the budget, but the primary effect is to call into question the management ability of the government.

    And hence, over-budgeting projects has two political upsides: a government can brag about the money being spent ($10 billion transport plan, back-ended to when you’ll no longer be in government and contingent on CBA) and it avoids looking like bad managers because even though the projects are costly, the less of them there are, the easier it is to meet expectations.

    It is interesting on this comparison that the recent HSR study assumes a 5 year construction period with 5 year ramp-up to full value, 20 years of appraisal and a 30 year residual value. Whereas, real-life big transport projects have 30-50 year construction time frames and at least for the right-of-way which is always a significant component cost no discernable end-date for residual value.

  8. michael matusik


    A great topic and one that I have been tracking up here in Queensland…especially in relation to infrastructure charges levied on new urban development…which is very high and rising. Many of the reasons why are covered in your summary/analysis.

    I also have been counting the number of workers on road and transport repair sites in recent years. Two years ago the average was five per pot hole filling operation…today it averages (in Queensland) 15, yes 15…and most are stop-go people or they seem to be watching not doing – a bakers dozen has been the average count over the last twelve months.

    See it is easy to have a job target – i.e. Anna Blighs 100,000 new jobs – if you employ most of them and they do non-productive work – these costs must be recouped (hidden) somewhere and the cost of new public infrastructure is one such place.

    We too have become a stupid country…thank our lucky stars that we have stuff in the ground that someone wants…and even then we hamper production via strikes, whilst our competitors – Africa, Russia, South America – watch, learn and start to take our market share.

    Again a good post…hopefully it will get a showing in today’s crikey email.



  9. Andrew Herington

    A huge problem in Australia is scope creep – once the idea is approved – a rail extension to X, instead of just being a rail extension it grows in size and complexity and the lack of public service accountability thinks nothing of staying within budget. Massive contingencies are attached to projects and rather than being kept for emergencies they are dished out to do various add ons in response to public consultation and difficulties in the planning and environmental process. The tendency for engineers to “gold plate” is perfectly matched to politicians desire to build the biggest and best and projects rapidly escalate in cost. A single suburban station in Melbourne is being built at a cost of $85 million – because it has the capacities of a major interchange when all that is needed is a platform for people to get on and off the train. A culture shift is needed to get back to affordable basics and make the most of existing infrastructure

  10. Austin M

    I think balancing the size of the contract is important aswell… on our very large PPPs and other contracts we are effectively paying for the fact only one or two players can afford to tender and the companies that are tendering are management companies that then subcontract out the bulk of the actual work to other contactors (and usually most of the risk aswell). Yes there are potential for economies of scale but in reality I think we just get another overhead added to an already costly project.

    Eatlink is a prime example of what people refer to as a good ppp (in atleast the construction timing and end product). Yet I often wonder if it had gone out as individual packages/sections would it have been done at a lower cost. The bulk of the work on Eastlink was done by multiple contractors on individual sections anyway (with the exception of some bulk purchases of things like bridge beams etc.) so it remains to be seen as to what the advantage of paying an overarching consortium and their financier large sums of money in management was really adding in corresponding value to the project.