Friday, November 27, 2009

Do we need 4 NBNs?

In the telco trade press there was some commentary following a report by Southern Cross Equities. I haven't read the full research but I gather the conclusion is that the NBN as a wholesale only access network means competitive advantage will come from core network, or brand. The research concluded therefore that Telstra, iiNet and Austar (Austar?) are well placed to benefit from the NBN.

The report highlights one of the current imponderables in the NBN discussion, which is exactly where the boundary points between the NBN as an access network and vrious core networks will exist. More importantly, should NBN Co be a participant in the national backhaul market as well as the access market.

It is hard to conclude that NBN Co should be allowed to be vertically integrated in backhaul and access and to at the same time compete in the national backhaul market, for exactly the same reasons that current policy rejects the access player participating in retail services.

But leaving "backhaul" to a competitive market also is a nightmare because of the differences between routes. As I've previously noted distance and traffic density are really important. In fact, the biggest issue for getting "equivalent prices" acrosss Australia isn't the differential cost of access but the differential of backhaul. It is also nonsense to suggest that "competitive supply" is a good outcome when the issue is lack of demand to meet the cost of ione fibre.

This suggests that a national strategy of one backhaul network still makes sense, priced at the same cost per megabit no matter where it is delivered. But that one "backhaul" network probably shouldn't be part of the NBN access network for the reasons above. This ultimately is the structure of the proposal that was put to the Government for NBN 1.0 by Axia.

So we need two NBNs at least - the national NBN in fibre access and the national NBN in backhaul. But there are still two more we need.

The first is to realise that exactly the same economic factors that mean we need one fibre access network are starting to come to play in mobiles. The two factors are that higher speed networks require wider allocations of contiguous spectrum. The ongoing fragmentation of spectrum for competitive carriers just doesn't suit the technology. The second is the existing inefficiency - the fact that a user can experience congestion i an area covered by a base station that is underutilised. The latter factior has seen its expression in the desire regularly expressed by regional consumers for roaming, as again noted by the RTR.

This issue now meets the challenge of using "advanced wireless" for the NBN. A couple of facts are relevant. The first is that the regional needs are best met using lower frequency spectrum for propagation. The second is that the areas that need to be served by the NBN wireless are not viable as stand alone wireless networks, they are better as additional sites on a single network.

This suggests that the govrnment coul take an NBN approach to future spectrum allocation. In particular when the 700 MHz (digital dividend) spectrum is allocated it should be allocated as one national lot that includes a must build condition as well as a must wholesale condition. To make the lot attractive it could include an allocation of the 2.5 GHz as well. Indeed it seems likely that the best outcome in 2.5 will be to seek only two licencees not three or more. The existing 1.8 GHz and 900 MHz allocations could be refarmed to provide a third operator with an LTE capability.

This requires the Government to think harder about the policy issues in spectrum.

But the same inefficient use of spectrum issues are now emerging in television broadcasting. In the analogue days there was a logic between the association between a 7 MHz frequency slot and a "channel" of broadcasting, but with digital the 7 MHz slot ca be used for different combinations of channels depending on coding. But each licencee using a 7MHz slot can have spare data capacity that is "going to waste". There has been insufficient discussion about why we are operating digital television through a separate multiplex for each licencee rather than combining all signals into one multiplex. The efficiencies are greater than in just frequency utlisation, but in frequency utilisation they are large.

So is there a logical policy for four NBNs - fibre access, backhaul, mobile and digital television? These are efficiency questions that all ultimately pit a static productive efficiency against a dynamic allocative efficiency argument.*

I'm not sure I know the answer. The outline here is not meant to be advocacy for these positions as such, but they do seem to be the kinds of discussions we should be having.

* Note the usual policy discourse defines three kinds of efficiency, allocative, productive and dynamic. I dispute that there are three - instead there are really four - there are the classic productve and allocative efficiency both of which are "static" fficiencies described by whether what happens in a specific point of time is efficient. The other are dynamic productive and allocative efficiency - the latter being the one most commonly thought of as "dynamic" efficiency which is choices over time about when to invest. At the CPRF Ben Frens gave a paper in which he referred to a concet of "governance" efficiency which he referred to as spectrum sharing and flexibility. I think this is the concept of dynamic productive effociency.

No comments: