Sunday, September 21, 2014

Ergas: Using "Economics" to Lie to the Australian public

How do you turn a profitable high-growth, high gross-margin business at the centre of the current rapid transformation of the global economy into an overall negative investment? If you're Henry Ergas, you concoct a deeply flawed Economic Cost Benefit Analysis.

Henry Ergas calculated in 2009 that retail prices for NBN subscribers would be $170/mth, with $133 in the metro areas and $380/mth in non-metro areas. The costs & pricing were modelled for him by an experienced ex-Telstra employee who became the General Manager, Pricing, of NBN Co from Nov-2009 to June-2014, while actual retail prices today are one-third his 5 years old prediction.
Thus, for the most likely estimate of $170 per month, unit costs in metropolitan areas are of $133 per month, while those in non-metropolitan areas are just under $380.
Ergas quotes his own study, now proven to be wildly incorrect, in the NBN Cost Benefit Analysis, as "fact". That's the start of his falsehoods and misrepresentations.
Ergas is calculating a Cost Benefit Ratio for an entity that doesn't exist and converting Profits into disbenefits.
He's counts only charges for NBN Co incurs and excludes as "transfer" the down-stream costs incurred by the other four Industries involved: Telecomms Retailers, backhaul/transmission providers, backbone/international connectivity and hardware/software vendors.

He then calculates "revenue" as retail income, conflating five separate, independent Industry sectors into one, while not accounting for their different cost structures and that they all make significant Accounting Profits. Ergas' composite entity, "The NBN", is far larger than the government owned "NBN Co" he uses, and it is majority private sector and profitable during his entire 25 years.

The whole CBA and it's subsidiary forecast, "Domestic bandwidth requirements in Australia, A forecast for the period 2013-2023", cited as 'Demand Forecast', is a thinly veiled attempt to justify Abbott's 2013 policy launch statement "25Mbps is 'more than enough' for Australia". Turnbull quickly had to 'clarify', or retract, that statement [current link to statement].

The problem isn't that Communications Chambers (UK) "magically" found the median household demand would be within the Abbott 25Mbps limit, while ignoring all Industry Standard Quality of Service metrics inventing a new Service Level Metric of "degraded minutes", but that Ergas tried to pull a swiftie: the bandwidth demand projection is only for a fraction of the Economic Forecast period of the CBA, to 2023 not to 2040.

"Median" usage is not financially interesting in Internet sales: it accounts for only 6%-7% of download traffic and 1%-2% of upload. All profits arise from the top 10%-20% of subscribers, the rest get a free ride or are subsidised. This is well documented in the Sandvine reports referenced in the reports.

Even the concocted "degraded minutes" metric shows a hyperbolic collapse (think 'goes vertical') in service quality if demand slightly exceeds available capacity. The Demand Forecast could've, and should've, been run out to 2040, a trivial spreadsheet exercise. But that would shown a massive problem with the VDSL2/FTTN network, especially if they'd plotted "degraded minutes". With the unsupported growth projections, by ~2030 the FTTN network becomes unviable, as video services are unwatchable during peak periods - the impact of "degraded minutes" is a complete loss of watchability, as anyone who's ever experienced a Youtube etc "buffering" event knows.

Ergas would like us to believe he's calculating the economic profit, not accounting profit in the CBA.
They are related: Opportunity Costs are deducted from accounting profits to reveal economic profits, answering the question: could I be doing better things with my money, time and other scarce resources?

Rather predictably, Ergas finds, in present value, that all government funded NBN rollouts will incur an economic cost (vs the wholly private rollout being zero cost) and the Coalition plan is the least costly option. This despite the 2012 NBN Co Business Plan predicting a profit of $54 billion and total direct returns to the taxpayer of $85 billion. Table 1, pg 10.
No further rollout: -$24 billion
MTM scenario: -$6.1 billion [on a $29.5 billion Govt. capital]
FTTP scenario: -$22.2 billion [on $30.4 billion Govt. capital]
How does Ergas' FTTP scenario flip $85 billion in direct government revenues, excluding NBN Co taxes and the additional taxes on profits & GST from all retail sales and related businesses, around 2-4 times that amount? [$85 billion = 7.1% IRR on $30.4 billion over 25 years is $54 billion, plus repayment of capital]

Ergas is directly stating that there's a direct $100+ billion Opportunity Cost embedded in the FTTP project, or around $70 billion more than an FTTN/MtM rollout. This is fantastical.

If you add the additional taxes on wages, profits and GST paid by all the businesses using or related to NBN Co services, there's probably $200 billion not being counted over the 25 years of the CBA.

So, where's this incredible $100 billion in Opportunity Costs that Ergas found? pg 10 informs us:
This is because it is more costly and slower to deliver, which delays the realisation of benefits. This means that its costs are higher and its benefits lower than the unsubsidised rollout scenario. The net costs of this scenario also include the net costs of delivering higher speeds to rural and remote areas via fixed wireless and satellite. [from the MTH, $6 billion cost: This largely reflects the net costs of delivering higher speeds to rural and remote areas via fixed wireless and satellite.]
and pg 11 offers an alternative explanation:
The FTTP scenario provides higher speeds within the fixed line area at higher cost and with a slower rollout. This reduces net benefits by a further $16.1 billion ... The FTTP scenario only makes the community $2 billion better off, in net terms, than under the no further rollout scenario.
The point about Business Profits are that they crystallise the consumers "Willingness to Pay" (WTP), related to "Price Elasticity of Demand".

Businesses, government and private, can only make a Profit if Revenue exceeds Expenses, if customers are willing to pay more for the product/service than it costs the vendor to produce.

If the customers are willing to pay more based on a feature that doesn't cost the vendor much, we can have "tiered pricing", leading to a reduction of "consumer surplus" - or increased profits based directly on value-in-use to the customer, not cost-of-production for the vendor.

Ergas does note "Consumer Surplus" in the CBA, where it had been conspicuously lacking in his 2009 paper.

The point about profits still applies: a vendor can only charge as much as the customer is willing to pay for the features. Any business, monopoly or competitive must obey this iron-clad law.

That the 2012 NB Co Business Plan was able to demonstrate profitability proves they were meeting the customer WTP for the product. The Opportunity Costs of delay to consumers are ridiculous and have no basis in fact. We only have the NBN Co project because the Coalition Government failed to act in 2001, when it fully controlled Telstra, to rollout a broadband network. In 2005, Sol Trujillo personally pitched John Howard to build an FTTN, and was flat-out rejected. The refusal to force structural separation of Telstra before the final sale was bizarre and intentionally created an uncontrollable monopoly player. The start date for costs to the community of delay should be 2001.

Which leaves three related aspects of the CBA's Big Lie:

  • Consumers have NO "Willingness to Pay" to speed, only price for a feature set.
  • The basic consumer trade-off embodied in tiered pricing is well known, but never considered or evaluated by Ergas: consumers trade Time for Dollars. The key value-in-use of higher bandwidth plans is based on how each consumer values their time, yet this is ignored.
  • The long-term "Elasticity" of Telecomms Services is embedded in the 2012 Business Plan, but is ignored by Ergas et al. The pricing of 12/1 and 100/40 services ($24 & $38/mth) would reduce to the same $11.75, while 1000/400 would reduce from $150/mth to ~$40.

Here's a proof of all these propositions: "Google Fiber" in the USA.
They are running symmetric, ethernet based, services to consumers at just three price points (note, no download/upload volume charging):

  • $120/month for 1Gbps and 150 TV channels on a 1 year contract,
  • $70/month for 1Gbps on a 1yr contract, and
  • $0/month + $300 once-off fee for guaranteed 7 years access at 5Mbps.

Google aren't just the smartest programmers on the planet, they are some of the best business brains and highly value economists in their work. They will be making direct profits from these services as well as driving additional business to their other lines of business.

Does Ergas really want us to believe that "Nobody wants more than 25Mbps" or "They aren't willing to pay for more than 100Mbps"?. The truth, as shown by Google, is that people have a preferred price-point and its close to the ARPU now in Australia. Google are demonstrating that the additional costs added to Retail Pricing are due to inefficiencies and excessive profits in the private sector, not the economics of the public sector wholesaler, NBN Co.

The WTP curves at the heart of Ergas' economic modelling aren't just irrelevant, they are wrong.
He's confused cause and effect, turning "speed" from a dependent variable into the independent, or controlling variable.

We know very, very well from the early days of ADSL1 when access rates were capped at 1.5Mbps and plans were tiered from 64kbps, that consumer surplus can be exploited. But with ADSL2/2+, no speed tiers were offered.

Why? Because "speed" is a hygiene factor, not a feature demanded & valued  by the consumer.
Already by 2002, customers were wanting more than the 3-5Mbps average service they were supplied, and willing to pay up to price-point for it. There is more than a decade's worth of pent-up demand in one of the most noted "Rapid Adopter" nations known.

There's another misleading false premise about Funding built into the Ergas modelling, it has two sides:

  • The Opportunity Costs for the Govt. related to Funding are zero. The money used was never going to be invested elsewhere, there is no "second best alternative".
  • The money invested was borrowed at 2.5%, not the 8% "discount rate" used by Ergas in the CBA. At best that's double-dipping, in the NPV and "opportunity cost".
The government bond rate is at historic lows, there has never been a better time to borrow and invest in Government Infrastructure projects that return a direct return.

How is that Economic affect missed by Ergas? It's very, very large, around 6%/yr on $30 billion over 25 years.

This also raises the real cost to the taxpayer of the NBN Co investment: I suggest that because it was always designed to make a profit and pay-back the borrowed money, the actual cost is the additional interest paid on the (max) $30.4 billion borrowed. At 3%, this is under $1 billion/year at peak, budgeted to be paid off in full around 2028. The real cost that should be using, based on the investment funding model, should be under $10 billion for FTTP. Even on the unbelievable Ergas model, FTTP is preferable and profitable, both in the Economic and Accounting senses.

There's also the $35 billion already realised in direct and quantifiable benefits to the community in increased equity in Publicly Listed corporations in the Telecomms sector, notably Telstra, Optus, TPG and iiNet.

Previously I've written that Ergas' modelling fails a basic sanity test: if the FTTP rollout by NBN Co cannot clear it's costs of $30.4 billion, then over the 25 year analysis, Ergas is claiming less than 0.05% impact on Productivity in Australia, on a GDP of $53T - $60 trillion. The Productivity Commission wrote a report from the early 2000's concluding that half the productivity growth of businesses is due to Information Technology: the single best investment any business could make was in its computing and networking.

I also have a very strong objection to referring to the NBN Co investment as "The Largest Infrastructure Project even in Australia", sometimes qualified with "Start Up'.

My answer is "So What?". The Public Service and Federal Government routinely manage, allocate and oversight amount ten times this.

Just Defence over the CBA period, not in any way an "investment", but only an expenditure, will at spending levels of 1.7% - 2.0% of GDP consume around $1,000 billion of taxpayer dollars.

Is the NBN Co investment "expensive"? Not compared to anything else
Is the NBN Co project necessary? Exceedingly so, it's been delayed by around 15 years!
Is a FTTP by NBN Co justified? Yes, it makes a profit, it is not an expense item, and is the only option capable of meeting currently known future needs.

This is the most base of the lies and omissions underpinning the CBA:
Ergas' modelling completely dismisses Risks, especially of "Black Swan" events, also named by Rumsfeld as "Unknown Unknowns". If Ergas' modelling were for a military campaign, it would be rejected out-of-hand.
The primary lesson learned of Technology and the Internet over the last 25 years, and wilfully ignored by the CBA, is:
  • Until it arrives, nobody can forecast, or even guess at, The Next Big Thing powered by the Internet, and
  • We are still being surprised with new Internet products and applications: the biggest Internet revolutions are ahead of us, not behind as assumed in the CBA.
    • In just 7 years from 2007, the smartphone went from an undesirable business device powered by Microsoft, to over 90% market penetration in the USA. Coupled with commodity back-ends, this is transformative technology with near zero barriers to entry. 


  1. Amazing deconstruction, Steve. Would be very interesting to see how Mr Turnbull would respond to it.

  2. Brandon, Thanks for your kind comment.

    Turnbull has assiduously ignored me.
    His staffer, Stephen Ellis did once swear at me in an email, which Renai Le May actually turned into a story.

    Short answer: no :(

    All my best