We Need to Talk About Infrastructure Victoria | Samantha Dunn

We Need to Talk About Infrastructure Victoria

Infrastructure Victoria has a problem: it's relied too heavily on faulty data, and has made some questionable assumptions on costs.
Samantha Dunn
Wednesday, April 5, 2017


The aspirations driving the establishment of an independent arbiter of infrastructure needs were admirable. Infrastructure planning in Victoria has become too politicised. The old Labor and Liberal parties hastily come up with plans only to see them scrapped when they lose government. Major projects are supported or opposed not on their merits, but on whether political points can be scored. Years of planning and design work can be dismissed solely because it is perceived as being tainted by the prior government.

And hence the creation of an honest broker in the form of Infrastructure Victoria (IV). An independent statutory authority, it was created in 2015 to rise above the politics of the day, the pork-barrelling of local representatives and the lobbying of entrenched powers-that-be. Its central task: develop a comprehensive infrastructure plan that benefits the State as efficiently and fairly as possible.

In 2016 IV released a 30 year infrastructure strategy. With the help of engineers, economists and accountants, they analysed hundreds of different project options to narrow them down to the core priorities for the State to better function. They did some nominal consultation – with citizen juries derived from the city and the country – to get some input from the people that should know best; the people the infrastructure is supposed to support.

IV have put some thought into their process. In assessing 280 options, they considered:

  • Capital and running/maintenance costs of any infrastructure.
  • The contribution to identified high-level needs of that State.
  • The likely economic, social and environmental impacts of the option.
  • How each option may enable, complement or inhibit one another.
  • The level of community support for the option, drawing on ongoing public consultation and the reports of two citizen juries.
  • The resilience of the options under alternate future scenarios.
  • Any relevant interactions with current state planning strategies.

Furthermore, for a limited number of major transport projects, IV commissioned additional demand modelling and preliminary benefit cost analyses.

Seems like a comprehensive process? Perhaps. Until you start digging a bit deeper.

Where the IV 30-year Strategy Falls Apart

The key issue with IV’s methodology is that they did not have sufficient time or resources to do a comprehensive Cost-Benefit Analysis (CBA) of each of the 280 options. A Cost-Benefit Analysis is where the total benefits, measured in dollars and including broader benefits beyond the direct benefits to the users of the infrastructure, are compared against the costs of building and running the infrastructure.

IV made the decision to rely on existing CBAs, where they existed.

The only existing CBA analysis is usually that funded and conducted by Government agencies. The direction of those analyses, and what parts of them are released, have a perhaps unsurprising tendency to match the policy preferences of the Government of the day. 

This bias can be countered with alternative analyses. On some high profile projects there can be alternative assessments conducted by a Local Council or a think tank.

Hence IV’s rankings are being informed by data that was shaped to give an up-or-down influence on a project’s prospects.

Who killed Doncaster Rail?

A classic example of this is the Doncaster Rail project in my constituency. This project has been analysed in the past by PTV.However, PTV has been reticent to release the full assessment reports, which were paid for using taxpayers’ money, and only mere summaries have been released.

PTV made assumptions about Doncaster Rail that would lead to a CBA coming up negative. For example, the Doncaster Rail option considered had the rail line terminate at the existing Doncaster Park & Ride. The Doncaster Park & Ride is conveniently close to… nothing. It has been placed in its location solely due to its proximity to the freeway. It is too far to walk from the main activity centre – the Westfield Shoppingtown on Doncaster Road – or any other points of interest. The primary way to access it is by car, or the badly routed and rarely used feeder buses. 

The proponents of Doncaster Rail see it terminating at Doncaster Hill (i.e. the Shoppingtown). This is the main activity centre and has its own bus interchange, with major plans for expansion that will only heighten its importance to the local community. Train passengers from the city could disembark at the Shoppingtown, pick-up a few necessities or visit the bank, and then get a feeder bus from the existing bus interchange to their homes.

Any economic benefit analysis that has the train terminate at the Doncaster Park & Ride will clearly show little in the way of wider economic benefits. There are few opportunities for capital improvement or urban densification next to a 6 lane freeway. The passenger catchment is minimal at the Park & Ride. There are many opportunities for such benefits next to the Shoppingtown, as already shown by the medium density apartments in the vicinity of Doncaster Road, and the high foot traffic at the shopping centre. 

Why didn't they model a version of Doncaster Rail that terminated at Shoppingtown? Image via ycat.org.au

On the cost difference between road projects vs rail projects

As noted above, IV did procure some preliminary CBAs of major transport projects to determine their priority. Once you dig down into how these were calculated, some alarm bells start ringing.

The “site costs” for rail projects were much higher than road projects; a 25% loading on the construction cost versus 9%. Site costs for rail projects comprise bus replacement services and compensating the operator (Metro) for the fact the line is out of operation. IV have claimed that for some rail works (e.g. level crossing projects) they can spend as much on site costs as the constructions works. This claim has not been substantiated with evidence.

Road projects have analogous costs. When the Tullamarine Freeway was closed for widening over the recent summer holidays, motorists had to take major detours, which lead to delays, longer travel times, larger taxi fares, and flights missed. Yet these costs are paid for by private citizens, not the government. In any proper CBA, these costs would be entered as disbenefits in economic modelling of the project benefits. Did IV do this with the predicted private costs associated with major road projects? No, they didn’t.

Design costs and profit margins 

In IV’s modelling, design costs (what is paid to architects and engineers to design the project), and the profit margins (the difference between the contract value and project costs) were set higher for rail projects than for road projects. 

For rail projects design costs were set at a 15% loading on construction cost, and profit margins set as 8%. For road projects they were set as 9% and 6%, respectively. Like all assumptions in these CBAs, these assumptions were not footnoted or substantiated, so it took questions from my office to find out why.

For rail, it was claimed there are less engineering firms capable of doing rail projects in Australia, and therefore less competition and higher profits. Furthermore, it was claimed that more specialised experts need to be flown in from overseas to design rail projects, hence higher design costs. 

There’s one major flaw in such an assumption, and it is as follows: yes, there has been a lack of rail projects in Victoria prior to 2014. Compare that with the amount of design work done for roads, including the scrapped East-West Link, and clearly the rail engineering capacity in Victoria would be anaemic. Yet that is looking at 2014, not 2022 when 50 level crossings will have been removed (if the Andrews Government sticks to its promise), or 2026 when the $10 billion Melbourne metro project is scheduled to be completed.

My office asked Infrastructure Victoria whether the levels of competition in rail engineering, or the number of resident rail experts in Victoria, would increase in the 2020s. Their response was yes, it would. My office asked whether this was included in their analysis of profit margins. The response was, no, it wasn’t. Therefore IV’s estimates of the design costs in rail are likely inflated.

On the matter of profit in the road building sector: IV used an average of the profit margins on all road projects implemented by construction contractors working for AECOM, a major engineering firm. Incredibly, projects large and small were included, with IV telling my staff “any guy with a grader can build a road”. So the profit margin on $1 million in repairs to a dual carriageway country road in Gippsland was considered as indicative of the profit margin on the proposed $10 billion North East Link in Melbourne. This is clearly absurd.

A recent history of road projects in Australia have shown that they nearly always cost more to build than first estimated. And what is worse, is that the history of Public-Private Partnership road projects in Australia are littered with financial failures: think of the City Cross Tunnel in Sydney or the Clem7 Motorway in Brisbane.

“Government costs"

Government costs are the costs incurred within government departments and agencies in managing a project. Interagency coordination, data sharing, planning scheme amendments – they all take bums-on-seats and endless coordination meetings to make happen.

These costs were set much higher for rail projects than for road: a 10% loading on construction costs versus 6%. 

IV claim that this is because there are higher “interface costs” for rail projects: they are fiddlier to build so more parts of more agencies are involved. This is hard to believe: East-West Link, for example, was going to be an incredibly complex project involving demolition of houses, carving up parklands and turning Alexandra Parade into a trench for a couple of years. That kind of mess takes a lot of mopping up.

IV argue that the capacity of rail agencies and operators – VicTrack (which owns the track), PTV (which coordinates rail services), Metro Trains (which runs the trains), would have to be increased to on-board a new rail line like Rowville Rail or Doncaster Rail or an Airport Rail Link. In comparison, on-boarding a North East Link for VicRoads is relatively easy – somewhat plug-and-play.

Perhaps, but aren’t the rail authorities going to have get better at onboarding new rail projects anyway? Taking control of a major piece of infrastructure such as Melbourne Metro, with its 5 new stations and kilometres of underground tunnel will be a learning curve for PTV. When people practice things, they get more efficient at it.

10% loading on construction costs for rail projects compared to 6% for roads. Image: levelcrossings.vic.gov.au


In their cost estimates, IV claimed it cost up to $17,500 per square metre to build a rail bridge, and only $3,297 for a road bridge. That’s a pretty big difference, a factor of more than 5.

On questioning, IV said rail bridges were more expensive because they have to be both stronger (trains are heavier) and more “exact”, i.e. straighter, lower slopes, inability to curve around less solid earth. Another valid point was that rail bridges need barriers to prevent derailments, special gantries to allow emergency egress of drivers and passengers, and none of these applied to road bridges. These are all very valid points.

However, IV also claimed rail bridges more frequently needed to be built in-situ, that is, built on site, instead of fabricated off-site and bolted together on-site. It’s much cheaper to build a bridge deck off-site and truck it to the final location than build it in the constrained and exposed areas on-site.

You’ve probably found it hard to miss the State Government’s publicity around its elevated rail project, with fascinating animations of how it will be constructed on the Cranbourne-Pakenham line. What does that involve? Kilometres of prefabricated rail deck being hoisted into position using large gantries, not in-situ construction. Prefabricated rail bridges appear to be the order of the day – and engineering and construction companies in Victoria are getting better and better at doing it through practice.

Why does North East Link get such laudatory claims such as “city shaping”

In its evidence gathering, IV parroted an AECOM/PwC claim that North East Link would:

"[C]onnect the less developed north of Melbourne with the more developed south eastern areas with freeway standard connection. This could enable shaping of the city and improvement of community employment and economic outcomes.”

This seems to imply that IV think unemployed/underemployed people in the “less developed” outer northern suburbs of Melbourne should drive for at least an hour each way on the M80, North East Link, Eastern Freeway and EastLink to get to jobs in the southeast, paying a fistful of dollars for fuel and at least two tolls in either direction. This seems like a diabolical solution to a lack of jobs in the north, but that is how IV thinks: to twist Joe Hockey’s words: poor people do drive far, and they’ll do it a lot every working day if IV have their way.

Interestingly, the same claim about access to jobs and economic opportunity is not made about Rowville Rail. This rail project would allow anyone on any northern suburban rail line – as far as Cragieburn or with the extension, Mernda  – to get to the Monash Economic/Education Centre with a single transfer. IV do not allow language to creep into their report of Rowville Rail being “city shaping” or helping the hapless inhabitants of the “less developed north” get to work.

But a single transfer to get to Monash University isn't “city shaping”?

Distance & Time

My office has asked IV questions about the very strong likelihood that North East Link will increase congestion on roads such the M80, the Eastern Freeway, EastLink, Hoddle Street/Punt Road and Alexandra Parade. IV said modelling by their consultants, which they hadn’t questioned, had showed that Vehicle Kilometres Travelled (i.e. the sum of every car’s odometer) to increase, but Vehicle Hours Travelled (i.e. the sum of time everyone spends in their cars) to decrease. The trouble with this is it flies in the face of all respected academic research into the effect of new roads on traffic: new roads induce demand, which fills up the new lanes created, and the driving speeds decrease to what they were prior to the new road (or even slower). 

Initial modelling conducted by IV’s consultants bears this out: The Age has reported that train patronage on the South Morang and Hurstbridge Lines is expected to decrease by 25,000 passengers a day, which means tens of thousands of more cars on the streets.  North East Link won’t lead to a net decrease in peoples’ commuting times – at best they will stay the same, at worst the surrounding suburbs will neighbour a permanent traffic jam.

Noise & Amenity

IV gave North East Link a “moderately beneficial” rating for Noise & Amenity. That’s right: a stonking great big 6 lane toll road that will carve through remnant natural greenspace (Banyule Flats) will apparently improve noise and amenity.

This is daft. 

No rail project analysed by IV got such an accolade, even though rail lines take traffic off the roads and reduce the related noise and local air pollution.

6 lanes of "moderately beneficial" traffic. Image: RACV


There’s a saying in computer modelling: “garbage in = garbage out”. It doesn’t matter how finessed your methodology or sophisticated your approach or how good your computer code: if you are relying on bad data, then the outputs will be inherently flawed. This is where IV has tripped up. It has relied on past CBAs and modelling that were massaged to political ends. Specifically, current and past State Governments have a penchant for allowing TransUrban to build Toll Roads, and didn’t want to do the heavy lifting in expanding the public transport network. 

It’s a shame because the timeframe and the breadth of the IV 30-year strategy is impressive. The Strategy is up for review in 2019 – maybe at that juncture IV can amass some high quality data and do a better job of it.