Monday, 26 November 2012

NBN: The Economics of FTTH vs FTTN

Do you know a Bargain when you see one??
Is either NBN Policy proposal offering a bargain or the opposite, a "lemon"?

First, what do we mean by "bargain"?

My "New Oxford American Dictionary" defines bargain as:
a thing bought or offered for sale more cheaply than is usual or expected : the secondhand table was a real bargain | [as adj. ] household and electrical goods at bargain prices.
an advantageous purchase, especially one acquired at less than the usual cost: The sale offered bargains galore.
Collins World English Dictionary, via
a. something bought or offered at a low price: a bargain at an auctionb. ( as modifier ): a bargain price
something acquired by or as if by bargaining; especially : an advantageous purchase <at that price the car is a bargain>
The dictionary examples seem to point to price: The cheapest things are bargains.

But look again, the wording is often "advantageous", not "lower than normal price". Bargains imply both "value for money" and "notably lower than normally expected price". Purchase price is only one of many factors influencing value.

If you've bought any second-hand cars, you know that what looks like "a great bargain" on the day might turn out to be "a real lemon"... Nasty problems, whether existing or latent ('broken' vs 'will break soon'), may cost several times the notional price of the car to repair. A $2,000 "classic" car that winds up needing extensive rust-repair or (hidden) accident damage fixed may turn into a $15,000 vehicle, still with a worn engine, suspension, steering and transmission!

Just because the sale price is low, doesn't make an item "a bargain", unless you can afford to discard it, for free, when it proves unserviceable.

The other type of value-decision is constant price, but different quality.

Which is a better deal for $15,000, a 2 year-old Hyundai (ie. 'cheap' car) or a 10 year-old BMW or Mercedes (ie. 'premium' car) with a higher build-quality and inclusions? It depends...

If you can work on the cars yourself, have access to specialist tools and test gear and can source parts at reasonable prices, then the older car may well suit your needs better.

If you're at the mercy of others, especially the official Dealers who aim to extract high premiums over the entire life of the product, the newer car may well suit you better.
If you're not going to drive the car far each year, have had it checked for problems by a good, local mechanic who specialises in that brand and who will maintain it for you, then you will, as far as these things are guaranteed, get a bargain by buying the older, 'premium quality' car.

What are we buying, as voters and taxpayers?

How, as consumers, do we "buy" at the next election an NBN that will be at least "good value" and at best "a bargain"?

Firstly, just what is being sold?
What is the up-front price, what's included and not included, what hidden costs might there be and what are the future costs that are reasonably predictable?

At the moment, this completely disregards: are the two options even roughly equivalent. Is one a beat-up old "paddock car" and the other shiny and new?

Update 28-Nov-2012: I've corrected the Fibre component to $26.8B (was $26.4B) and total project to $37.4B (was $43B).

The GPON FTTH solution, for 93% premises, starting to be rolled out is not $37.4B: the fixed-line fibre is $26.8B. The rest is for Fixed Wireless, Satellite and other system/project costs.

An FTTN solution, without copper-line leasing or purchase costs, for 98% of premises was estimated by Telstra to cost around $15B. Paul Fletcher, in Wired Brown Land, suggests a cost of $9-10B for ~90% of premises. The last 1.1M premises were estimated to cost $4-5B, more than four times each than the first 4M premises.

To be generous, let's compare a $10B FTTN with $26B FTTH, both to roughly 90% of premises.

The elephant in the room, the unknown cost, is: What will Telstra charge to use its copper?
Whilst the current agreement pays Telstra $11B in NPV, for access to ducts, pipes, pits and exchanges, it does not pay for access to copper: payments only accrue when copper lines are disconnected.

But here's a surprise: the biggest single component of replacement cost of Telstra's Customer Access Network is the "ducts and pipes". Together with copper cables, well over 50% of replacement costs. One estimate from the ACCC is over $40B (for a competitor) to build a replacement copper C.A.N.

Update 28-Nov-12: With my limited knowledge, I hadn't understood just how deep the Telstra pipe-and-duct network went. It extends 100mm conduit well into the local neighbourhood, with 25mm conduit on "service drops" to premises for the lead-in. A large part of the $11B NPV given to Telstra is for access to that network. NBN Co will be upgrading parts of the network and installing its own pits and probably many new service drops (old, congested, collapsed and to wrong location on-site). Because paying to reuse the Telstra pipe-duct network resulted in a saving, I'm guessing a net cost of $10B, with $2-3B in upgrades. It must be noted that the current SSU contract doesn't buy anything from Telstra: it allows access and allows Telstra to decommission and remove its copper after a declared date in each region (I didn't read who owns the pipes-ducts after then). There is a deadline that NBN Co must meet to declare a service, or Telstra gets to keep its copper in service. IANAL, but that sure looks to me like the Coalition can't rely on the existing SSU and automatically acquiring the pipe-duct network. To be an "apples-and-apples" comparison, they still need to provide an irrevocable right-to-use the $10B pipe-duct network and to upgrade it to the minimum usable standard, as NBN Co will.

The GPON FTTH project is two projects in one:
  • For around $12B in construction costs $10B in Telstra payments and their own upgrade costs , lay build and upgrade 100,000km of pipe (conduit) and associated connection cabinets.
  • For $14B $14-16B, lay a GPON FTTH inside the new pipe/conduit network.
If Telstra has known since the early 1990's that it would've been needing to move residential premises to fibre in the mid-term, why did it not start to lay the conduit then or since. As a 20-year project, or 5,000km/year, it would've allowed a refresh of copper cabling where needed (lowering maintenance costs), been well within their technical and capital means (under $500M/year), and given them a $10+B asset to lease or sell now. They were able to garner the majority of their $11B NPC from this source.

Which prices should we compare for FTTH and FTTN?
  • $40B for a copper CAN to $26.8B for GPON FTTH?
  • $10B FTTN to $14B for the GPON component of the FTTN + $2-3B for upgrades?
Neither of these support the Coalition assertion, "FTTN will be 3-4 times cheaper to build than an FTTH".

The first bargain NBN Co is delivering is the $12B $10B conduit-to-the-premises, included in the price. If ever new cables need to be run (upgrades or maintenance), the majority of the construction costs are already covered.

The FTTN does not include this new and very important feature. If this was a car, the salesman would be now saying: "Oh, did you want tyres and rims on your new car?" This new "Structured Cabling" approach became standard in computer network nearly 20 years ago. It's not a new idea, nor outrageous to expect.

For a proper comparison, an FTTN has to include the same options: conduits and cabinets to 75% of premises served.

The apples and apples comparison for FTTH to FTTN is: $26.8B vs $20B ($10B+$10B)
Not that much of a difference.

What do each solution cost in running costs, repairs and maintenance?

All the FTTN proponents acknowledge the ageing copper network has rising maintenance costs, and these are higher than for a) fibre optic in general and b) a brand new fibre network.

By how much?  Nobody has been willing to put a figure on that.

We can get a hint from Telstra itself, from their ASX released Briefing Paper, Aug-2005, "The Digital Compact and National Broadband Plan". Page 18 cites: "14.3M fault calls, 14+% of all lines have fault".

This is without considering the running and maintenance of those 70-350,000 little grey boxes spread across the countryside for vandals to smash, weather to impact and heat to break the electronics in... The cost isn't in electronics, though they were estimated at over 50% of the node costs in 2009, it's the millions of kilometres that technicians will have to drive, reducing their productive hours to below 50% of their hours.

That's the second bargain NBN Co will deliver: much lower operational and maintenance costs.
This is a "gift that keeps on giving": maintenance and electronics upgrades are always going to many-fold cheaper than an FTTN. It designed to be efficient.

What are the major project costs?

In the ACCC Report to the NBN Expert Panel in 2009, the cost modelling says that for ULLS (Unbundled Local Links), the cost of capital was 75% of Telstra's costs. The Cost of Capital is the overwhelming cost in every NBN Solution.

The Federal Government, with its AAA credit-rating and "risk-free" status, borrows money much cheaper than anyone. The ACCC analysis added a 6-7% "risk margin", on top of the "risk-free" rate, to the modelling of a privately built FTTN.

Funding NBN Co with "risk-free" debt effectively halves the dominating cost in the project.
A 50% discount: sounds like a great bargain to me!

What the ACCC didn't discuss was IRR (Internal Rate of Return) and Profit Margin.
As a wholly-owned Government Business, NBN Co is expected to return 350 basis points (3.5%) above the reference rate: this is barely above household mortgage rates and much, much lower than any business loan for this project.

The Internal Rate of Return and Profit Margins demanded of NBN Co by the Government are extraordinarily low. They are many times (4?) lower than Telstra or a private consortium would charge.
A 75% discount on the profit margin:  sounds like an outstanding bargain to me!

How do we know what Margins Telstra likes to charge (apart from the headline Profit Margin of 42+% of turnover)? From the Aug-2005 ASX Release. Page 13 gives EBITDA Margins for various services.
They range from 88% for Domestic Long Distance (STD calls) to 55% for Local Calls and Basic Access and 40-42% for Mobiles and Data Services.

Because of the very modest financial target set by the Government, we know that NBN Co does not have to charge the extraordinarily high margins Telstra has been used to. As consumers, we get to enjoy lower charges and better services.

Lifetime and Upgrade: Do I get a 15,000km or 150,000km warranty?

The new equipment installed for both FTTN and FTTH solutions will have a 30-year economic design life. Whichever we choose, we are locked into the decision for a very long time. Most people think of a 25-30 year mortgage as "a lifetime"... This is the same sort of commitment.

FTTN proponents acknowledge the capability gap between ADSL/VDSL and GPON Fibre.

Their usual response is "nobody will ever want or need more than DSL can deliver" or "Fibre is sooo expensive, who'd want to buy it?"

The speed difference, out of the box, is 80:1000Mbps at best, but for guaranteed rates 12-25:1000Mbps. That's the difference between walking and a 320kph Ferrari.

For twice the upfront price, one-tenth the running costs, 40 (forty) times the speed and better quality?
Is that the bargain of the century or what??
A Ferrari that uses less fuel than a moped, for just twice as much.

As I've noted above, the monthly rental for subscribers is based on many more factors than he headline/up-front cost of the project. Twice as much up-front does not translate to "Twice the rental".

While the new nodes and electronics installed for an FTTN will have a 30-year economic life, does the ageing fault-ridden copper network come with a warranty of any sort?

That's the weakness of the FTTN: the copper is old and reaching the end of its service life in the majority of premises, it doesn't come with a warranty. [no references found]

You get what you pay for: new GPON Fibre will have a 30-year service life, will support 100-times speed improvement as it is,  (just add new electronics either end) and, if you need to replace or augment it, being in conduits - that's really quick, cheap and easy.

Just 15 years ago most Australians on the Internet used 14.4k-28.8bps over the same phone lines they are now getting 4Mbps on. Those same phone lines will be upgraded to 12-25 Mbps by an FTTN.

Is that a 1,000-fold or 6-fold improvement? Which ever way you count it, that's really impressive improvement.  How much more improvement can we expect from copper DSL services?

There's the rub... FTTN advocates will say "People have 1Gbps or 10Gbps over copper in the laboratory"... Kinda, sorta, but not really.

You can get 100Mbps over new Cat-5 ethernet cable: for 100m, sometimes 500m.
You can get 1Gbps over new Cat-6 ethernet cables: for 10-30m.
You can get 10Gbps over purpose built cables, either Thunderbolt (active parts) or SFP+: for 2-3m.

Because as frequencies increase, the signals travel in a smaller band just on the surface of a conductor, they are unduly affected by imperfections, corrosion and the breakdown of joints. In the real-world with old cables designed for 3Khz voice, not Mhz DSL, you will never see 1Gbps on copper. The speed that GPON Fibre starts at.

At best, over the next 10 years we'll see a 2-4 fold increase in DSL speed, and then that will probably be it.

Pay 50% more and get a lifetime warranty on a new product: sounds like a really good deal to me. That's the difference with fibre.

In another 10 years will you be wanting 12Mbps or 40Mbps, 100Mbps or more?
Chances are you'll want a lot more bandwidth than 12Mbps and a DSL FTTN just won't take you there then, or for the following 20 years.

In 10-15 years, will the Coalition bow to public pressure for Better Broadband and adopt a GPON FTTH? That replacement will cost only marginally less than the current implementation: the DSL FTTN will cost additional money to remove. It isn't just "building one to throw away", it's also "paying almost the say to remove it as put it in".

Building a DSL FTTN to throw-away: that's not a bargain and disposing of 10,000's of nodes won't be cheap, quick or easy.


GPON Fibre is a bargain compared to an DSL FTTN solution, even a hybrid FTTN solution.

The construction/install cost does not reflect what you'll pay per-month.
The basic price, for 12Mbps, will be higher than current DSL services, but if you're getting 2-4 times the speed as the majority of DSL subscribers will, that's still good value.

  • GPON FTTH is delivering is a $12B conduit-to-the-premises, included in the price, allowing cheap, fast, easy upgrades, repairs and additions.
    • FTTN does not provide the conduit, leaving us in the nineteenth Century.
  • GPON FTTH will deliver: much lower operational and maintenance costs, probably 2-5 times lower.
  • Using government "risk-free" funding and a comparatively low Return (3.5%), more than halves the financing costs, which normally account for 75% of internal costs.
    • Any FTTN that is built via a Public/Private partnership or solely by Private, will have much higher financing costs and many-fold higher Gross Margins to provide shareholder returns.
  • The guaranteed bandwidth of GPON FTTH starts at forty times faster than DSL and will only improve. 100Gbps ethernet fibre-optic is currently commercial available, not a laboratory pipedream.
    • The problem isn't that DSL FTTN can or cannot match speed now, but that it will never scale to meet future demand. It is at the end of its lifecycle.
  • The copper network is old and near the end of its service life: it's warranty period has expired, while GPON Fibre, which has lower maintenance costs, will be new and easily last the next 30 years: Fibre comes with a "lifetime warranty".
  • An FTTN built now, will cost us twice: Once to install and again, nearly as much, to remove.
    • What's the sense in building something intending to throw it away. It doesn't make sense.
DSL FTTN might seem cheaper up-front, but is like a clapped out old car on its last legs: it will cost you dearly to run and maintain, let you down when you need it most and will one day die on the side of some lonely road.

But there's a "gotcha" nobody in the Coalition will address: "What will Telstra charge?"
The SSU is about an FTTH not FTTN, Telstra will be obliged to pursue its shareholder interests and charge $8-$12B for its copper network, or lease it for (much) more. Trying to finesse this charge by partnering with Telstra shifts costs to the financing side: much more expensive.

Bargains represent good value, not just "cheapest to buy".
The GPON FTTN is a great bargain, a DSL FTTN is a lemon that for the next 30 years you'll regret buying.

No comments:

Post a Comment

Note: only a member of this blog may post a comment.