Saturday, February 20, 2016

The MTM-NBN rip-off: saving $250 Million costs the taxpayer $125 Billion.

Related links. This was originally a comment on a previous post.

There's a point  related to hidden OpEx costs which makes a nonsense of Cost-Per-Premises of "nbn ltd":
the bulk of maintenance costs will be spent on the Copper Pair network. [50% of all? FTTN 50% higher?]
Where is that in the CPP costings?
That, for me, makes these company documents not 'business' but 'political', they are misleading figures.


PROFIT is the difference between Revenue and Expenses, business focus should be on 'the bottom line', not any one line item at the exclusion of all others.

Expenses include Depreciation on the Capital Expenditure and Interest on Loans.
Equity, as in Shares and Government 'equity injections' to 'nbn ltd' only appears on the "Balance Sheet" (Statement of Financial Position) not on the "Profit & Loss" (Statement of Financial Performance).

Equity is given in return for Dividends.
Which aren't a Liability & later Expense like Interest, but a voluntary payment decided by the Board, paid out of Retained Earnings [an Asset not P&L account] but sometimes funded from borrowings (not sustainable).

What's the real difference to the taxpayer between the 2013 NBN plans of the Coalition and ALP?

Look as hard as you like in the Federal Budget and you won't find any line-item for NBN Co or "nbn ltd".

The taxpayer is NOT liable for the 'Peak Funding' cost of building the NBN, this is a furphy ("Cost") put around by Turnbull and never successfully challenged by the ALP & Greens.

Only if the company is sold or goes into liquidation, just like AUSSAT before it in 1992, will the losses be crystallised and flow into the Federal Budget and affect the Budget Deficit / Surplus.

The Budget Fiscal Impact of the NBN, the thing that matters to taxpayers and is tracked by the PBO (Parliamentary Budget Office), is solely the Financing Costs of the Equity given to NBN Co/nbn ltd, until it is repaid, if ever.

At the then current 10-yr Govt Bond Rate (F2) of 4.68%, Sept-2013,
and now, 2.80% (18-Feb-2016)

  • on the ALP max Govt Equity, $30.4 billion
  • vs the LNP max Govt Equity, $29.5 billion.

In 2013: $42.12 million per year @ 4.68%.
In 2016: $25.20 million per year @ 2.80%.

Assuming equity return in ten years, prior to the 2013 election, the FULL difference in Fiscal Budget Impact between the ALP and LNP plans was $420 million dollars. Million with an 'M', not Billion with a 'B'.

Right now, the original Conroy / Quigley / Young "93% Fibre" plan would, in total, cost $250 Million more than the rather dysfunction MTM-NBN.
If you adjust for CPI at the RBA's 2.5% 'target rate', a 176% change over 23 years, it's a $142.8 Million difference in 2018 dollars.

The Astute Reader will have noticed "until Equity is repaid" - this isn't how ordinary businesses are funded. Shareholders don't get repaid their $1 for BHP or ANZ shares, they solely get dividends, or sometimes their shares are bought back, at (near) market price, by the company.

Given this peculiarity, the primary taxpayer question about Interest on Equity is:
"For how many years?"

We know from the 2013 NBN Business Plan of Mike Quigley, that the full draw-down of Govt Equity (remember it's NOT a loan) occurred around 2017 or 2018 and was fully repaid by 2028, a non-standard accounting approach adopted by the Board in lieu of better directions, then a portion of Free Cash Flow would be returned to the owner (Fed Govt as representatives of the taxpayer) as Dividends.

In 2013, the NBN Co Business Plan had an Internal Rate of Return, (IRR) for the 30 years from 2010 to 2040 of 7.1% on a total Capital (Equity) of $30.4 billion.
That is the sum of all dividends divided by the (equity times years), not a single-year value.

To simplify the equation, it's 7.1% of $30.4 billion for 23 years (2017 to 2040), for DIVIDENDS.
A tad shy of $50 billion, in nominal dollars, not CPI adjusted nor Net Present Value (NPV) with a Discount Rate.

Where repayment of the Equity is included (not standard business practice), that's repayments of $80.4 billion, PLUS the value of the business as a Going Concern: one with no debt, very strong cash flows and low maintenance and network upgrade costs.

A reasonable Price to Earning Ratio of 12:1 or 14:1 would see another $50 billion (not CPI adjusted) realised for the taxpayer, based on Earnings of ~$4 billion / year in 2040.

By 2040, you'd expect 100Gbps available to consumers at a modest premium, should they wish it.
Today's $10,000 100Gbps GBIC's will, via Moore's Law, be commodity items worth ~$100 in 2040.
We've 60 years of data on Moore's Law for Silicon:

  • a 100-fold reduction in component costs over 2 decades is low-ball, it's 3.25 years per halving in cost.

More realistically, the price reduction will be a 1000-fold (think smartphones vs 1990 super-computers), a halving of cost every 2 years, but might show up as 10Gbps GBIC's being so cheap, they are included on CPU chips, effectively zero-cost.

In 2016, the LNP finally released it's IRR, a paltry 2.7% - 3.5%, still based on $29.5 billion equity.

What we don't know is if that includes, like the 2013 Mike Quigley / Harrison Young Business Plan, a full repayment of the Equity, nor do we know when the Free Cash Flow of "nbn ltd" becomes unencumbered by Debt.
From the consistently lower Revenues & higher Expenses of the MTM-NBN, that's very unlikely.

Ignoring their higher Rate of Spend, using a 23-year return period to 2040, the LNP's IRR on $29.5 billion represents $18 - $24 billion return to the taxpayer.

On top of this, the long-term maintenance and upgrade costs of the two Hybrid Fibre networks (Coax and Copper Pair / xDSL), are much higher: in fact, full replacement.

Even to get to the unproven 1Gbps of "G.fast" will require massive Capital works, including extensive civil works, exactly the part of the costs that Turnbull notes "is not subject to Moore's Law". Digging holes, placing active devices and running/terminating cables only increase in cost.

Those billions being sunk into tens of thousands of Nodes is meant to be 100% discarded.

The Turnbull MTM plan has only ever included a single in-place technology upgrade to the Nodes, the addition of 'Vectoring'. That planned cost doesn't appear to me to be included in the 2016 business plan and its low IRR.

VDSL2 Vectoring does very complex signal processing, which tells us three things about the cost of the single board:
  • it's got to be high-power to achieve the necessary performance over even 192 lines.
  • Top-of-line commodity CPU's and GPU's are required & expensive to buy and run.
  • The motherboards, being a high-performance, low-volume, specialised design will be very expensive and may have a short service life or, like the 486 CPU, have serious faults requiring replacement.
I'd expect that upgrade to cost 25%-50% of the current Node rollout budget, solely on the cost of the technology and that every single Node has to be reconfigured.

G.fast must completely replace all the VDSL2 equipment and network, throwing away almost all of the current investment.
They must use customer-powered micro-nodes - burying complex, expensive equipment in suburban pits and expecting them to work reliably must be one of the most 'optimistic' Technical, Business and Political decisions I've heard.

Any upgrade to the 34% unlucky subscribers to be connected to the single-shared medium of Coax are going to face the same sort of costs. What the proponents of HFC never say is that they have to deploy "Nodes" as well, they don't call them 'Nodes'.

Just like FTTN, FTTx, FTTB and FTTdP, they need to convert Digital-over-Fibre to Copper...

  • The only way to reduce shared-medium congestion is to reduce the number of clients per segment: code for "we'll have to double or quadruple the number of node-equivalients we've deployed, plus rollout more Fibre to connect to them".

That's on-top of deploying ever more expensive transmission technologies. DOCCIS 3.1 does a "massive" 1Gbps... Wow, that's good, until you look at real-world current Fibre.
So what's the upgrade path? How do they get to the 10Gbps already embedded in the NG-PON2 standard [with four colours, its 40Gbps to subs per fibre.]

At what point does the already two-decades old Telstra Coax become too degraded to use, or needs replacing with a higher-standard Coax to handle higher bandwidths?
Duplicating the $800M write-off already of the Optus Coax network.

Do we believe that the old Telstra HFC network will last unchanged until 2040?
At which time that buried cable will have been in service around 50 years!!!

The Telstra decision to bury the Coax included a very expensive upgrade cycle, one that the MTM-NBN must now bear, having assumed control and effective ownership from Telstra.

The Turnbull / Morrow / Switkowski Business Plan doesn't include predictable & foreseeable network upgrade costs out to the lifetime (2040) of the project, at least not that we can see.

Compared to the conservative $130 Billion returned to the taxpayer (not CPI adjusted) from the 93% Fibre plan of 2013, even the high-ball $24 billion return of the MTM-NBN looks very sad, especially as they'll either have to fund a full replacement with Fibre or a complete rebuild of their Hybrid Fibre networks (HFC, FTTN), costing at least as much (CPI-adjusted) as the current rollouts, due to the ZERO CapEx Reuse inherent in the technology choices, not the 50% Turnbull claimed in his 2013 Election "Broadband Plan". Another lie?

There's the choice between the Plans:
  • 93% Fibre costs $250M more, will last 50-years, can be infinitely upgraded in-place very cheaply and will return the Taxpayer $125 BILLION more, versus
  • MTM-NBN will last maybe 10-15 years, requires full replacement with no savings and returns, with NO current data on those future costs.
$250 Million more now buys us $125 BILLION in future savings.
Seems like a "no-brainer" decision to me.


Source for 10-year Govt Bond Rates. RBA's F2 series, daily for 2016 and monthly for 2013.
RBA data: http://www.rba.gov.au/statistics/tables/