Grahame Lynch, publisher of CommsDay has recently written an article about the cost of services on the National Broadband Network would be upwards of $200 per month to make an acceptable return on investment.

Engaging Concept Economics, Lynch - who in the past has opposed the investment in the NBN on Whirlpool Broadband Forums - and the analysis has made some interesting assumptions and in my opinion some key mistakes in their evaluation:

* The pricing of the services was based on using contention ratios at 50-1, which is significantly better than current broadband contention ratios used throughout the industry; in fact some four times the typical used for residential access networks.

* Concept claim their estimated prices were commensurate with international experience quoting Verizon's FiOS service which generates an ARPU of US$129 or A$185. Firstly they should have been using world parity pricing and not simple exchange rate calculations, which would put the price around the A$130-135.

* The Verizon ARPU of US$129 is a retail price which includes broadband, telephony and video as part of the make up services, not just broadband. Currently the average user in Australia is already paying about A$170-180 for their residential telecommunication services including broadband ($60), fixed line telephone rental ($29), telephone calls ($20) and payTV ($70).

* Concept does touch on the PayTV industry using the network, but fails to point out that this too would add to the income for the NBNco. The price of $215 which is suggested as the price of a broadband service to provide a suitable ROI, is purely scaremongering by those opposed to the construction of this network.

* What is not clear is if the claimed revenue of $215 required to substantiate the cost of the network is the wholesale income required by the NBNco or the retail prices offered to the public? If it is the retail price, then we are already paying prices around this mark.

* Concept suggested that 80% of metro and 20% of the regional back haul would be leased capacity; based on what price? With the roll out of more network, back haul pricing - which is expensive in Australia - is in my opinion likely to drop. Furthermore if the NBNco is building a network, then much of that network will provide its own back haul infrastructure or use existing assets of companies who come on board with the NBNco. So the 80% of back haul coming as least capacity - I think that highly unlikely.

* While on the back haul, Concept suggested the swapping of assets for equity was unlikely to be a cost saving to the venture. This is quite a wrong assumption, as it would be a direct benefit. The ROI calculations were based on a $43billion dollar investment by the NBNco; if the Government exchanges shareholding in the NBNco for assets of other carriers (such as offered by Optus) the actual investment required will be significantly less than the $43b, thus directly affecting the ROI calculations used by Concept.

* Concept also calculated the network to cost somewhere between $40-50b dollars; this based on what knowledge and experience? I doubt if any of the economists at Concept have every built a telecommunications network, let along a Fibre to the Premises network. It seems the range they quote is somewhat similar to the price quoted by Telstra during the NBN RFP process.

* Several telecommunication experts -including myself - have already commented that it will not take anywhere near the $43billion quoted by the government to build this network. The cost is expected to be more like $20-25b assuming the use of existing network infrastructure. So have Concept worked out the calculations themselves, or taken a few guesses on what others have publicly stated?

* At $43b the network cost is a surprisingly round figure of $5000 per home for the 8.5million homes in Australia. If we consider the 90% to be built with FTTH and the remaining with Wireless or Satellite, even at $3000 per home (which is twice the amount currently being achieved by Verizon) that is only $23b.

I think the most significant failing of the economic evaluation has been its reliance entirely on the residential space and does not consider the SME, Corporate, Education and Government business that could be carried over this network. Some $1b is spent by state governments on their telecommunication needs each year. A good percentage of that is private networks which would be well suited to the proposed FTTP technology.

There is also the CSIRO and aaRNET which spend millions of dollars each year on maintaining their fibre based infrastructure connecting some 55 locations across the country in addition to the 38 Universities. The NBN could help them reduce their operating costs while pouring money directly into the network.

There are also many other uses of this network outside of what we do today. Smart Metering, Security, Education and Health will all benefit and can pay for this network. If $215 was the correct price (which I don't for one moment suggest it is) a combination of services such as Internet, Telephone, PayTV, and those mentioned above would all be contributing to meeting and passing this price.

In summary it is my personal opinion this report should be discarded as irrelevant because it lacks objective evaluation, the technical knowledge of network construction and a detailed understanding of the other economic benefits which such a network can offer to the community.


At 25 April, 2009 23:59 Grahame Lynch said...

Unfortunately there are several errors of comprehension and analysis in this post which should be remarked upon.

CommsDay does not endorse the findings of Concept's findings, we merely asked for them and as anyone can see we asked a number of questions seeking substantiation. Your conflation is a bit like suggesting that Rupert Murdoch loves Kevin Rudd because a Newspoll grants the latter 68% popularity.

From what I do understand of the findings, Ergas used an improved contention ratio to account for the minimum speed requirements of the network and the likelihood it would be used for higher end applications. Are you seriously suggestion that ISPs currently employ contention ratios of 200-1? Here's one reference endorsing 50-1as a current norm...

You then go on to correctly point out that tripleplay costs around $170 already in Australia. Ergas never said his $215 end price was for Internet only, he simply said it was the revenue required to make the model work. Neither did CommsDay - which made the Verizon comparison, not Ergas as you incorrectly state - state that the Verizon price was anything but an ARPU for a blend of services. SO we agree, $170 is pretty normal for tripleplay and you need ARPUs at that level to provide a network that is definitely overprovisioned for tripleplay. SO what's the problem?

* I think you also misunderstand the point about pay TV providers, he is suggesting they won't be adding revenue to the NBN because they will prefer to stick with delivery media that is more cost effective or pervasive! You may disagree with that point but at least first understand it.

* It is very clear that Ergas is referring to retail and not wholesale revenue. It is made explicit in the first sentence. He has assumed that ISPs will only sign up for this if they can maintain their current margins. A fair assumption.

* Ergas assumes that with considerable competitive backhaul already in cities it makes sense to lease and not overbuild. Anyone who runs a business would concur with this impulse. He also ran his numbers through an assumption of 50% cheaper backhaul than current prices, which took around 5% off the total cost.

* I don't think you understand the point about asset swaps - it incurs a deferred cost because it has exactly the same ROI requirement as direct spend. It has a positive cashflow implication but makes not a difference to the total cost of the project and its ultimate ROI requirements. Economics 101 really...

* Concept's biggest sin on costing appears to have been to take the government's $43b at face value. You may disagree with that Stephen but you should take that up with the government not Concept or CommsDay. Perhaps you should ask the government why they might want to scare people off their policy with an incorrectly high number?

* The overall calculation has nothing to do with the "residential space", nor does it exclude "commercial demand." It is simply a calculation of how much revenue per line is required at specific take up rates for the thing to achieve the goal of profitability. It is an expression of a required ARPU, the A in which stands for average, not absolute.

* I suggest this analysis is irrelevent on the grounds of the last sentence which suggests that the benfefit of the network to the community isn't captured by the analysis.

Unless you can propose a way to capture those same benefits to the NBN bottom line (like say a Medicare levy or an airline petrol tax) or something, it is very relevant to suggest that for whatever goodness an NBN might have, the thing might either be too expensive to use under its current parameters or might go insolvent in spite of itself (highly likely if the government continues to encourage cheaper and cheaper copper prices via regulation and hurts the migration path to the NBN).

Finally, I completely take issue with your opening assertion that I oppose investment in the NBN. I have never said that, I have instead criticised specific aspects of the NBN RFP process and given the thing eventually collapsed, I would suggest my judgement wasn't too wrong there.

I think a FTTH network would be a great thing but I also demand a bit of policy rigour and economic analysis with it. The only reason that Telstra and the ACCC didnt agree on an original FTTN plan in 2006 (I recall) was over a disagreement in pricing of an average $12 a month. We're now talking about spending - THAT REQUIRE'S A CAPITAL RETURN AND IS NOT A SUBSIDY OR GRANT LIKE IN SINGAPORE OR MALAYSIA - of $5000 per household. I think we're entitled to ask questions about the economic rigour of this.

CommsDay has kicked off a debate about this. This is a good thing. It is certainly not irelevant, especially given the substantial comprehension errors displayed in this specific refutation of the Ergas results.

At 26 April, 2009 00:07 Grahame Lynch said...

Oh and my apologies for some of the typos, spelling mistakes and incorrect word usage in my preceding post. I wrote it in a moving vehicle and acccidentally posted it before revising it.


Post a Comment