[LINK] Comparing the NBN to Sweden for CBA

[Nick Ross]

>From our own Chirgs :)

I've got a follow up coming to this. People often say that Japan and
Korea's usage of their networks shows that there's little demand for NBN
speeds, but Sweden is first in the world according to Cisco...

http://www.abc.net.au/technology/articles/2012/05/01/3493131.htm

On one side of the NBN debate, the Opposition continues to call for a
cost-benefit analysis. Those who resist the calls say the indirect benefits
are too hard to quantify. Wouldn't it be nice if the two sides could be
brought together, at least on this point?

It's just feasible that such a study could be carried out - and it might
come down in favour of the NBN, if Sweden's experience is anything to go
by. A report from that country provides both a useful case study, and helps
indicate how such an analysis might be designed.

While only a preliminary study, conducted for Sweden's ICT research body
Acreo, this document<http://www.broadband-europe.eu/Lists/Competences/Prestudy%20socio-economic%20return%20of%20FTTH.pdf>,
"Socio-economic return of FTTH investment in Sweden, a prestudy" suggests
that the economic benefits of fibre-to-the-home pay for the network in just
3.5 years.

Not ten, twenty, thirty, or off-in-the-never-never: three-and-a-half years.

On the fact of it, though, the study is *bad* news for NBN-boosters,
because Sweden could get its FTTH for AU$1500 less per household than we
are. I will get to that detail later. First, I'd like to see what we can
learn from Sweden in terms of economic benefits.

*Sweden: a natural experiment*

So how could Sweden's experience help put some light into the NBN debate,
and maybe take some heat out?

What the authors, Marco Forzati and Crister Mattsson, have attempted is to
factor in what happens to other sectors of the economy once FTTH is
installed; and Sweden is a natural experiment for this, because it can
compare municipalities with different rates of fibre penetration, from zero
to about 60 per cent.

Because it's a preliminary study, the authors have been conservative, and
limited their work to a small number of quantifiable outcomes. On the cost
side, therefore, they have looked only at the cost of deploying the civil
infrastructure of the network- the fibres, which Forzati & Mattsson call
the "passive infrastructure". On the benefit side, they have included only
a handful of factors: impact on GDP, reducing government costs, better
value telecommunications to end users, and employment effects.

There are, the authors note, many other ways in which better networks might
generate returns, and they suggest further study to look at how other
factors may be quantified. For example, they note the likely positive
benefits in e-health - but since these aren't quantified, they're not
included in the Forzati & Mattsson study.

*Investment is an input to GDP*

I know it sounds obvious, but the NBN isn't just a black hole into which
money disappears from our economy. It's an investment with both direct (the
jobs and investment in the network) and indirect impacts on GDP.

Forzati & Mattsson's study uses work from Germany ("The Impact of
Broadband on Jobs and the German Economy", Katz et al, 2009) which finds
that investment in broadband returns 93 per cent of its cost as increased
GDP. And that's before what Forzati & Mattsson call the "uncaptured"
benefits of the network are accounted for.

Those indirect benefits include some of the things we've heard about in
Australia - e-health, education, new services and so on. I'd like to focus
on factors that are under-represented in the Australian debate, but which
Forzati & Mattsson have quantified.

*Cheaper government*

We have already seen providers on the NBN competing in their service
offerings. That competition can be expected to continue - even the US Trade
Representative welcomes the "level playing field" of the NBN - and it's
easy to forget that households won't be the only winners.

You don't have to get far outside Australia's major cities before telecoms
costs escalate - and in some places, a high capacity (say, 100 Mb/s)
business-grade service is simply unobtainable today.

Even in urban Stockholm, the rollout of fibre shaved 30 per cent off the
city's telecommunications costs. Across Sweden, the study finds, ubiquitous
fibre would save local governments more than AU$70 million each year.

Australia could realize even greater savings: we have a larger number of
local governments (Sweden has 290, Australia has more than 600 councils),
more of which face the problem of remote communications - and the NBN would
also offer savings at the state and federal government tiers.

Going by Sweden's average of about AU$240,000 saved per municipality per
year, Australia's local government sector alone could save nearly AU$150
million annually when the NBN is rolled out.

*Employment*

Sweden also offers a natural experiment from which the authors have been
able to estimate the impact of FTTH on employment, by comparing employment
trends between municipalities with and without fibre. Some of this is, of
course, the direct impact of employing people to install fibre - but in the
longer term, that impact washes out of the system.

Their finding is that the long-term impact on employment is worth 0.2 per
cent increase in employment, per 10 per cent of fibre penetration.

Australia and Sweden are different economies - our unemployment rate is
lower, for a start, leaving less room for improvement - but even a
contribution of 0.5 per cent to employment would go a long way to getting a
payoff from the NBN.

*Real Estate*

The one part of the study that doesn't translate easily to Australia, I
suspect, is their work on real estate values - because as I will discuss
below, the Swedish housing market is so different to Australia's.

Even so, it's interesting to note that they believe the presence of fibre
in a home is worth about AU$6 per month to the occupants (in terms of extra
rental value) - which quickly adds up when you're talking about millions of
households.

*Is the NBN too expensive?*

The study's big downside for the Australian experience is that it suggests
we're paying way too much for the NBN - an average $3360 per connected
premises (based on $37 billion government input and a target of 11 million
households), compared to Sweden's average of AU$1800 per household.

Let's look at per-premises costs in more detail:

*Sweden, house:* average connection cost 18,000 Krona, or AU$2560 (passive
infrastructure only)
*Sweden, apartment: *average connection cost 10,000 Krona, or AU$1425
(passive infrastructure only)
*Sweden average connection (all premises types):* 12,700 Krona, or AU$1800
(passive infrastructure only)
*Australia NBN, all premises types:* average connection cost: AU$3360
(active and passive infrastructure)

There are key issues to consider when discussing network cost:

   - Australia's investment includes both active and passive network
   components, whereas Forzati & Mattsson only account for the cost of the
   passive infrastructure;
   - Australia and Sweden have vastly different population patterns.

The Swedish study only includes the "passive" costs - so a comparison is
misleading. To accurately compare Australia and Sweden, we would need to
have one key datum from NBN Co: when the network is completed, how much
will be invested in active equipment - the network switches, the boxes
being installed in peoples' homes, the points of interconnect (POIs), data
centres, other exchange equipment, and the software that provides retailers
with access to NBN Co's system?

And the homes we live in have a huge influence over the cost of the
network. The Swedish data tells us that connecting a house is 80 per cent
more expensive than connecting an apartment.

A big reason that Sweden can get a cheaper network than Australia is that
more of their dwellings (52 per cent) are apartments than houses (42 per
cent). In Australia, housing stock is overwhelmingly in freestanding or
semi-detached houses, which make up more than 85 per cent of the dwellings
reported in the 2006 Census.

If Sweden reflected Australia's housing patterns, the cost of a national
FTTH rollout would be about 33 per cent greater than Forzati & Mattsson
estimate.

And if Australia's NBN costs include AU$11 billion in active network
equipment, our per-*house* investment is pretty much in line with the cost
of connecting a house in Sweden. We're not only at the mercy of covering a
geography that's 17 times as big as Sweden: our love of freestanding
houses, is loading up the cost of the NBN as well.

Even so: the price premium of the NBN compared to the Forzati & Mattsson
study doesn't mean it can't break even; it merely delays the payoff - and
not by much. If Sweden bore the same costs as the NBN (with all other
factors unchanged), the payoff would be delayed by one year, to
four-and-a-half years.

So perhaps there's no reason for the NBN's supporters to fear a
cost-benefit analysis.

*Richard Chirgwin is from The Register*