Indicate development sites, Melbourne (source: SGS)
Indicative development sites, Melbourne (source: SGS)

Infrastructure Victoria’s new 30-year draft strategy is very good but, as I noted last week, that doesn’t mean it’s free of shortcomings (see Is Infrastructure Victoria’s 30-year strategy any good?).

A key flaw is the claim that the cost of providing physical infrastructure (i.e. new suburban roads, sewer lines, power substations) and social infrastructure (i.e. schools, recreation reserves, public transport) for new development on the urban fringe is “two to four times more” than the cost of development in established areas:

The established areas of Melbourne to the east and the south are well served by infrastructure and offer the opportunity to further intensify housing. Accommodating residential growth through this kind of infill development also has the potential to save infrastructure costs, with greenfield urban expansion estimated to cost between two to four times more than infill, depending on the capacity of existing infrastructure to support additional people.

This is a very important proposition because it underpins one of Infrastructure Victoria’s major recommendations: that densities in established areas should be increased (see Can the politics be taken out of infrastructure planning?). The claimed difference in cost might be true but equally it might not; the fact is neither Infrastructure Victoria nor anyone else really knows. Infrastructure Australia peddled the same trope when it released its Australian infrastructure plan back in February (see Are infrastructure costs a lot higher in the outer suburbs?).

Infrastructure Victoria commissioned consultants SGS Economics and Planning to look at relative development costs at various urban locations but the brief only asked for a literature review (see Comparative costs of urban development: a literature review). While SGS concluded “from the costs that could be compared within the texts” that greenfield urban development “cost approximately 2-4 times more than infill”, it also set down some caveats:

Most authors rely on their own reviews of existing literature about infrastructure costs; very few produce their own original costings. (We) caution against unqualified promulgation of infrastructure cost ‘benchmarks’ for different development settings; costs are heavily dependent on area-specific factors.

SGS say only two of the studies reviewed produced their own costings with direct input from civil engineers; one by SGS itself and one by Trubka et al. The key finding though is that “much of the reviewed literature relies on costs presented in Trubka et al (2009)”; those costs “were intended by the authors to be replicated and tested by others”.

It’s never a good idea to rely on one or two studies to make policy, especially when it shapes how billions of dollars might be spent. But it gets worse; I previously looked at the Trubka, Newman and Bilsborough report back in 2010 and 2011 and wasn’t impressed. I pointed out in this article, Are infrastructure costs higher on the fringe?, that the underlying data is out of date and the methodology isn’t transparent.

As I said then, Trubka et al didn’t in fact calculate costs themselves either. They sourced their estimates from a 2001 report, Future Perth, prepared by the WA Planning Commission to assess infrastructure costs in Perth. Future Perth didn’t calculate its estimates from first principles either; rather, it surveyed 22 earlier studies, some dating from as far back as 1972 and some relating to costs in the USA and Canada. Future Perth is a working paper and hasn’t been published – hence the rigour of its methodology and the quality of the 22 studies it drew from hasn’t been tested.

As I noted when discussing Infrastructure Australia’s report, this area of policy and research seems to be a black hole. The sort of reliable data on contemporary infrastructure costs that’s required to inform policy-making in 2016 and beyond just doesn’t exist. We can’t be confident we have any idea what the difference in the cost of providing urban infrastructure is in different locations.

It might seem intuitively obvious that the cost of providing physical and social infrastructure in established areas should be only a quarter of the cost on the fringe; after all, it seems much of the infrastructure already exists. But in the inner 10 km ring that Trubka et al looked at, most of the “spare” capacity is long gone (see Is unused infrastructure capacity in the inner suburbs all used up?). Retrofitting expansions or new works is difficult and expensive. Established areas are more intensively developed, values are higher, sites are smaller, access is harder, opposition from neighbouring land uses is more intense, and the cost of avoiding disruption of other activities is higher.

I’m in favour of action to increase density in established suburbs but policy needs to be made on the basis of reliable evidence, not intuition, wishful thinking or motivated reasoning. I don’t know for certain if it costs more or less to provide infrastructure in the established suburbs, but neither does anyone else including Infrastructure Victoria. It’s shocking that city managers continue to make policy without evidence; both government and academia have dropped the ball here (see Is academia researching the urban issues that matter?).