Wednesday, February 10, 2016

Things I don't understand about NBN Economics: how does $1,500/premises become $4,500?

I can’t figure out the basic cost equation of the NBN.

If it’s costing ~$1,500/premise to install new fibre, how does the full cost become $4,500?
What does the extra $3,000 buy us?

The POI’s (Points of Interconnect) and transit network, the network operations centre, the provisioning & billing systems and what else? But these are common costs across all networks, it cannot be them.

FTTP costs: 2017 nbn co Corporate Plan (pg 53) Browfields: $4,411 Greenfields: $2,608. [Aug-2016]
Why does it cost $1,803/premise more for 'Brownfields' connections? It's 50% more than the pre-2013 costs. Where did these new costs come from?

There is a difference between ‘passing’ and connecting premises, so what proportion of the $1,500 is 'passing' and 'connecting'? That's hard to find and doesn't account for a three-fold increase.

Edit: Population Sources added at end, 27-Aug-2016.
Edit: Fibre cost-estimate Sources added at end, 20-Feb-2016.

1. It’s got to be much cheaper to run new fibre than copper. But how much cheaper?

In 2004, Telstra (Ted Pretty) declared that greenfield Fibre Customer Access cost 1.5 times as much as Copper, but within 18-24 months would cost the same.

So what’s happened in the intervening 10years?
What’s the current ratio? New Fibre has to be now significantly cheaper to install than Copper.
On the usual Moore’s Law and ‘learning curve’ improvements, you’d have to guess Fibre is “cheaper”, but how much?

It can’t be the factor of 10 you’d expect with electronics / Moore’s Law, because it costs money to dig holes and bury conduit - that's common to both.
Fibre cable, per meter and per circuit, is a fraction of the cost of copper to buy, install and join…

This is cost-per-premises, not cost per Mbps, which is already 10:1 in favour of Fibre / GPON, with another factor of 10 coming in the next few years.

Note: This is ’new' not the incremental cost of adding equipment to an existing Copper Network.

2. "Google Fiber" (& Facebook, Amazon, …) will likely disrupt the Aussie Telco market, sooner than people realise.

Google are reputed to have costs ~ $1,500/premise, if they target parts of the Australian market, they’ll kill the competition.
If Vertigan et al got real "Infrastructure Competition", it might scare the pants off them.

It cost $20-$25M for Telstra to build the South Brisbane fibre rollout in 2013 to pass 18,000 premises and connect 13,000.
Without benefits of scale, it cost Telstra $1,300-$1,550/premise for FTTP in a built-up Inner Urban area, presumably the most expensive Real Estate around and high Brownfields cost.

That’s broadly consistent with the $1,250 being reported by Quiqley’s NBN, allowing for economies of scale and 'Learning' effects.

Does that include the Customer Premise Equipment, as it does with Google?
"Google Fiber" charge $300 for a lot of equipment: a WiFi access point, a file server and a PVR/Video server. Nothing up-front for the fibre connection.

3. Broadband isn’t “massive capital”, it’s very modest per-premise and many times lower than the per-household consumer electronics spend.

The per-premises cost of installation of Fibre ($1,500) is a relatively small cost compared to householder CapEx and maintenance costs. Not every household will opt for a fixed line connection, lowering the network costs. We could expect 50%-75% connection rates.
  • the average household of 2-3 people will have 2-4 smartphones @ $250-$1000 each and replace them 2-3 years. The cost may be buried over a 2year Mobile Phone carrier contract.
    • households are spending $500/year just on smartphones (guesstimate)
  • households also invest significantly in electronics that require Internet back-end connections
    • WiFi router/firewall and repeaters
    • tablets (>1/person), replaced 3-4 years
    • laptops (1-2/house), replaced 4-5 years
    • gaming consoles(>1/house), replaced & updated 3-5 years
    • large-screen TV (4K has arrived at consumer prices) and sound system. replaced 4-5 years
    • PVR, replaced 4-5 years
Households are spending well over $1,500/year ($12B total) on electronics, gaming and entertainment systems, by these estimates.

They pay extra for access plans, mobile and fixed, plus more again for content (NetFlix, Foxtel, ...)

My guess is around $25B/year ($2,750/premises) is being spent on new 'Internet Accessing' consumer electronics - much larger than the once-off CapEx of installing a Fibre connection.

If household were given the option of paying off the $1500 connection over 5 years, then paying maintenance of $150/year ($12/mth), wouldn’t they jump at that?

You’d then have ISP’s and content providers charging for the upstream part.
From Google Fiber, we know this can be done for under $70/mth for Internet only (uncapped volume).

This two-part fee was the charging model adopted by TransACT in the ACT region and proved quite unpopular with consumers, who will pay a premium for the simplicity of a single bill.

For a modified model to work, ISP/RSP's would have to pay the network owner a direct access fee on behalf of each single-bill client.

4. Despite being a “Wide Brown Land” , we’re the most urbanised nation on Earth.

We already have a lot of long-distance fibre connecting Cities and town.
Anyone who claims Australia has “geographical” challenges and “low population density” is being disingenuous or deliberately ignorant.

The ABS population studies show that we cling to coast and overwhelmingly huddle together in a very few cities.
  • the Big 5 capitals, have ~60% of the population with the ABS noting "Two in Three Australians live in a Capital City" [the target region for Sol's 2005 "NBN" proposals.]
  • "Major Cities" (>1M) account for 70% of Australians
  • urban settings get us up to 89% (45 cities with >30,000ppl).
  • just over 2% of 22M Australians live in 'Remote' (318,969) and 'Very Remote' (206,051) areas
We have three very different Broadband networks to be built, upgraded and maintained:
  • servicing the compact Urban markets, overwhelmingly the Capital Cities & mainly the Big 3.
  • servicing ‘the country’,
    • i.e. Regional towns and small towns & villages, already with telephones and grid power
  • servicing ‘the outback’.
    • Isolated Rural Homesteads and remote small settlements. [ideal for Satellite]
There’s a secondary issue: universal mobile (3G/4G) coverage.
We’ve a large highway & road network that we need to cover with mobile network.

The notion that we can have 3 competing carriers both adequately cover all areas outside the major populations and do it economically without sharing any facilities is ludicrous in the extreme.
Where there is almost no 'usual' traffic and coverage areas are measured in hundreds of square kilometres, "infrastructure competition" isn't just economically inefficient, it is demonstrably the worst possible model.

Any dominant player (e.g. Telstra) will be able to outbuild everyone else and purloin all subscribers who need access 'in the bush', not just living there, but visiting and driving through. It doesn't take long for them to take all the revenue and drive the other players out of the game due to high costs of duplicating infrastructure.

The killer in this game for the non-dominant mobile carriers is upgrading all remote base stations to new standards. From 3G to 4G and the LTE's speeds & frequencies coming soon.

In the dense, urban areas, there's high enough population density generating enough traffic to support multiple players overbuilding each other's mobile networks. They will also generate enough revenue to be able to upgrade their networks competitively.

There will also be, if anyone wanted to calculate it, a mobile phone carrier equivalent to "Goyder's Line" in South Australia: the places where average rainfall just stops being able to support cropping.

There are four regions to be identified in Australia:
  • the majority, where there is insufficient traffic & revenue to support any fixed mobile network
  • the very large area where there's enough traffic/revenue to support a single, but not two, mobile network
  • the modest area where there's enough demand to support two, but not more, mobile networks
  • the Cities that can support many mobile networks, but possibly not in all suburban areas.
Should we consider the "Wide Brown Land" one homogenous market, or actually segment it geographically by demand & revenue, then attempt to define economically sustainable long-run Policy solutions? I think the later.

In reality, Australia is for most services a very compact, very accessible market, beating the USA, UK and Europe on most measures.

Where the bulk of our population lives & works, Cities & large Regional centres, there is room for multiple operators and even "Infrastructure Competition", given a strong regulator.

Australia has the distinction of being the only market where a well-funded, large-scale Cable TV deployment failed. Directly due to the Consumer Regulator allowing insane overbuilding.


Summary

So why is rolling out a new connection to every urban premises, with nearly infinite upgradability and 50+year life, such a Big Deal?
It's really very cheap at around 50% of the yearly per-household expenditure on Consumer 'internet enabled' Electronics.
Anyone who focuses on the total project cost, without relating it back to per-household expenditure, is trying to deceive or confuse people.

How did the per-premises cost get bulked up by a factor of three?
Someone is having us on...


Population Sources: 27-Aug-2016

Wikipedia, source ABS, 45 cities with pop. > 30,000.
https://en.wikipedia.org/wiki/List_of_cities_in_Australia_by_population

Infrastructure Australia: https://infrastructure.gov.au/infrastructure/pab/soac/files/2015_SoAC_Chapter_2.pdf
In 2013, more than 18 million Australians lived in the 20 major cities.
Most of Australia’s recent population growth has occurred in the cities,
with the largest 20 all enjoying population growth over the last decade.
Australia’s major cities are now home to almost 80 per cent of the Australian population, the result of a process of urbanisation that has been occurring for more than a century.
As Figure 2.2a shows, in 2011 almost half of the population lived in the 3 most populous capitals:
Sydney, Melbourne and Brisbane.
Approximately 15 per cent of the population lived in cities and towns of between 30,000 and 85,000 with the remainder in rural towns and remote settlements.
The latter have experienced uninterrupted decline in the number of people living there since 1911,
while the share of the population living in cities and towns of between 30,000 and 85,000 has remained steady since 1991 after a long and slight increase.

Bureau of Stats, 4102.0 - Australian Social Trends, 2008 [pop data, 2006 Census]
http://www.abs.gov.au/AUSSTATS/abs@.nsf/Lookup/4102.0Chapter3002008
Between 1996 and 2006, Australia's population grew by 2.4 million people.
In 2006, Australia's population reached 20.7 million people.
More than two-thirds of people lived in Major Cities (68%) and
the remainder (32%) were in Regional and Remote areas.
The proportion of the population living in each of the Remoteness Areas (broad geographical areas sharing common characteristics of remoteness) varied considerably across the states and territories.
For most of the large states, including New South Wales, Victoria, South Australia and Western Australia, people were concentrated in the Major Cities.
Queensland had a relatively high proportion of its population in the Inner and Outer Regional areas (37%) compared with the other large states.

Fibre cost-estimate Sources 20-Feb-2016:

Project Finished: http://exchange.telstra.com.au/2013/01/24/south-brisbane-exchange-migrations-completed/
[deadlink: https://web.archive.org/web/20130103174129/http://www.fibretosouthbrisbane.com.au/index.html]
Progress: http://exchange.telstra.com.au/2012/05/10/5000th-customer-migration-achieved-in-the-fibre-to-south-brisbane-project/
Announcement: http://exchange.telstra.com.au/2011/06/01/full-speed-ahead-for-south-brisbane-fibre-connections/
News Ltd on Queensland Children’s Hospital project: http://www.news.com.au/national/telstra-deal-adds-millions/story-e6frfkp9-1226076707082
  • $50 million paid to Telstra in compensation, includes real-estate for Exchange site.
  • $28 million interest-free loan to Telstra, for construction costs for 18,000 premises passed.
  • I presumed $20-$25 million on network and the remainder for Exchange fit-out & equipment.
Minister stands by phone exchange call: 2011 Courier Mail story.

Brisbane Times article, as tabled in Qld Parliament.
"as a matter of compensation the state is obliged to meet the cost of replacement of the old copper exchange" (quote from Minister)
This includes a $28 million loan that will be repaid to the Queensland Government.
The loan pays for the replacement cabling to local residences.
Qld Audit report on LCCH project in South Brisbane
  • relocation of the Telstra exchange—required because of changes mandated by the Coordinator-General for road realignment which required removal of Telstra's telephone exchange. This added $73.579 million to the project budget, which included a $28 million interest free loan to Telstra that Telstra is required to reimburse to the Queensland Government on or before July 2018.