NBN on steriods

Posted by

The AECOM-Grimshaw-KPMG-SKM high-speed rail study is an impressive collation of effort but it doesn’t really give you any answers. It’s bit like solving a rubik’s cube, great work but what have you really achieved.

In particular, the report gives no estimate of the net present value of the HSR project either in a commercial or a wider economic sense.

The report does say:

Separate forecasts for the impacts on commuting in the Newcastle to Sydney corridor were prepared. In these forecasts, commuter fares are lower, representing a potential subsidy.

What about subsidies for other routes? All other proposals for high speed rail have required hefty government support. It would be surprising if this proposal was any different.

The thing is they have actually done the work to calculate this. They have calculated the costs, the fares, the demand and even some external environmental and health-related (less car accidents) benefits.

I have picked out the data that I can from the report to make a conservative estimate of a required subsidy of $4.3 billion per year! We could build a Melbourne to Brisbane inland freight rail line for that ANNUAL cost.

The subsidy works out at $80 per trip. We could basically afford to give everyone of the estimated 54 million trips a year a free flight from Sydney to Melbourne at that rate.

I wonder why they don’t give these figures in the report. Do you think the headlines this morning might have been a little bit different if they had?

My approach

Using the information in the report you can get a quick estimate of what the cost in dollar terms will be. I have to make some assumptions but I have always been generous to the case for a high speed train when doing so.

I have made no effort to back out the cost of wider benefits but note these would need to add up to $4.3 billion per year to make the case for the project to proceed.

Capital costs

The media today widely report the total cost at $108 billion but this is an upper estimate. The P50, least-cost estimate (50% chance of cost not being exceeded) comes out at $75.7 billion (p. D21). On top of this we need to add client planning and procurement costs which the report estimates at 10 to 15 per cent (p. ii).

Adding 10 per cent gives a total captial cost of $83.3 billion.

Amortising that over the estimated 50 year life of the network (and using their 7 per cent cost of capital) gives an annual capital cost of $6 billion.

Operating costs 

Operating costs are harder to estimate because of the limited data the report provides.

In 2036, the report estimates 43 rollingstock are needed at the captial cost of $40 million each (p. D26).

These are assumed to be leased over 30 years at an 8 per cent interest rate. This gives an annual cost of $3.6 million per train set or $152.8 million per year in total for the 43.

Commercial ticketing is assumed to cost 5 per cent of revenues, which from the below analysis, comes out at $131.1 million per year (p. D26).

Infrastructure maintenance and asset replacement costs are estimated to cost $400,000 per track kilometre per year. Google maps says that Melbourne to Brisbane via Syndey is about 1800 km so that is $720 million per year (pp. D24-D25).

I have not been able to estimate the costs of rolling stock maintenance, labour costs, electricity costs and administration costs due to a lack of data. Their assumptions are provided though (p. D26) and if anyone can find an estimate for train kilometres or train hours I should be able to work out a number for most of these.

Total estimated annual operating costs therefore come to $1 billion.


The report models that just under 54 million trips will be taken on the high speed rail in 2036. This represents about 30 per cent of trips taken in this corridor every year.

The report assumes that fares will be comparable to the costs of air travel. It assumes a “business” fare and a “non-business” fare. Demand is assumed to be 50/50 between these classes on routes between the capital cities so I have just taken an average of the fares to calculate the average fare on any route.

A table is provided for trips between centres. To simplify matters I have assumed that anyone in the “intermediate” category actually pays the same fare as the longer route between cities. That is, someone getting on at Albury to go to Sydney pays the Melbourne to Sydney fare.

I have estimated all fares not directly given in the report at 20c per kilometre for business travel and 10c per kilometre for non-business travel.

All of these assumptions mean there is an upward bias in my revenue estimates.

I estimate total annual revenue of $2.6 billion or just under $50 per trip.

Net annual costs

Capital costs = $6 billion

Operating costs = $1 billion

Total costs = $7 billion

Trips per year = 53.9 million

Total cost per trip = $130

Average revenue per trip = $50

Average subsidy per trip = $80

Total annual subsidy = $4.3 billion      

UPDATE: I have just found that the report does say this:

There are few examples of a HSR service fully recovering capital costs except in the very long term. Operations and maintenance costs are usually self-funding, but infrastructure costs are unlikely to be fully recovered without a significant government contribution. (p. 13)

Which is exactly what is shown here. The fact that the report doesn’t fully highlight the true costs of the network reveal that it is not a dispassionate analysis of the project.

Andrew Bolt reckons that Labor’s support for the study is another sop to the Greens.


  1. Interesting reading your more detailed analysis of the figures given my own more cursory post at http://www.stubbornmule.net/2011/08/train-in-vain/. We both come to the same conclusion though that in the first decade or two the government subsidy would be in excess of $6bn/year. But as I point out independent estimates of passenger numbers on the Eurostar are more like 15-20m passengers by 2036. The Eurostar between London and Paris (and onto Brussels) is a good model but has a much greater potential passenger base. It takes about as long to travel by train from central London to central Paris via St Pancras and Gare du Nord as the plane flight via Heathrow to Charles de Gaulle given both airport trips include at least an hour extra for the journey from central London and Paris. 2 hours in the case of Heathrow on a bad day which is most days. It takes 20-30minutes to get from central Sydney and Melbourne to their respective airports on a good day so I am guessing a lot of business travellers would still choose to fly. So if Eurostar is going to get a maximum of 20m by 2036, up from 10m today, then Melb-Syd-Bris isn’t going to get more than 5m, max, in my opinion.
    I also question the cost of capital of 7% which will only be true if it is government guaranteed. In which case it is just a PPP. A structured bond which capitalised the unpaid interest would rapidly increase in notional. It’s possible and has been done before (Perth Airport bonds come to mind) but not at that time scale or size. A tax-free capitalising bond might make it more palatable but then the lost tax would also run into the billions per year so the opportunity cost would still be enormous. Any investment banks involved would expect a huge premium for structuring and placing with investors.

  2. I agree with you on both counts.

    Their patronage seemed optimistic and those stats you have provided confirm it. Thanks.

    On the cost of capital, I just used their figure, which is Infrastructure Australia’s recommendation. They also use figures of 4% and 10% for sensitivity.

    But it’s not actually clear what they use the cost of capital for because they present no estimates of NPV or financing costs.

    The bottom line is that even their own rosy figures this would be a monumental waste of money.

  3. not sure how you got $50 per trip, airfares are a lot more that that, currently melb-syd, ecconomy one way, Virgin is $185, qantas $175 and jetstar $130

Comments are closed.