An NBN retrospective - Part 5 ? X
Home GitHub Contact

Notes from 2022

This is Part 5, where we discuss in brief the economics of the NBN, and why it simply doesn't stack up in its current form.

Part 1 focuses on the cultural deficiencies at NBNCo.

Part 2 focuses on the failings of FTTN.

Part 3 deals with the HFC network and ways in which it can be salvaged

Part 4 explains why the LTE network was poorly thought out from the start.

Part 6 hammers home the fact that all of this is a result of shameless and disgusting short sighted political manoeuvring by the Liberal Party to protect the interests of Rupert Murdoch and other corporate friends such as Telstra.

An NBN retrospective - Part 5.

One of the major hurdles to the NBN being a profitable enterprise is low Average Revenue Per User (ARPU). The cause for this is multi-faceted, and I make no pretences that I am an economist so this section will be short, however there seem to be two major causes for NBNCo’s money troubles:

The cost of purchasing bandwidth on the NBN is very high compared to other wholesale networks. NBNCo’s pricing structure involves two access charges, billable to RSPs: the Access Virtual Circuit (AVC) and Connectivity Virtual Circuit (CVC).

The AVC is a per-connection charge, essentially the equivalent of the traditional line rental. AVC is charged based on the link speed between the end user and the RSP’s network and can be considered the “base cost” of providing an NBN service. For example, a 50Mbps downstream/20Mbps upstream (50/20) AVC provided over FTTP or HFC costs an RSP $34 per month. This is not the only cost to the RSP, however. Per the Wholesale Broadband Agreement (WBA), the AVC charge only covers the speed of the connection between the end user and the handoff to the RSP’s network. Using the plumbing analogy, the AVC charge determines only the size of the pipe between the end user and the POI. The RSP must also pay for the data sent across that connection.

CVC is a very controversial and divisive topic. CVC is a charge billed to RSPs based on how much data they transfer over the NBN, and currently costs $17.50 per Mbps per POI. A few years ago, most RSPs were engaged in a pricing race to the bottom in order to try and capture new customers during their transition to the NBN. This led to many severely under provisioning CVC at the POI, causing excruciating peak hour slowdowns even on FTTP. Again using the plumbing analogy, CVC determines the size of the pipe between the POI and the RSP’s own network. In order to keep costs down, RSPs were under sizing the pipe for the amount of flow, and as such the pipe was continually clogged.

The ACCC eventually intervened, forcing RSPs to advertise the “typical evening speed” for each service, which is a decent indicator of whether or not the RSP has purchased enough CVC from NBNCo. This of course led to RSPs actually purchasing enough CVC, as none wanted to be seen selling 100/40 AVCs which could only realistically hit 25/5 during peak hour. This stopped the race to the bottom price-wise, and now most RSPs offer very similar prices for equivalent services.

Part of this under provisioning scheme remains today, however. One may have noticed that a 50/20 AVC would, naturally, require 70Mbps of CVC to use to its full potential and as such should cost $30+(70*$17.50). But the typical 50/20 service costs around $70 a month, not $1,259 a month. This is because RSPs use statistics to minimise the amount of CVC they must purchase at each POI. The key assumption is that not everyone is going to be saturating their connection at the same time 24/7, and so RSPs can provision enough CVC to satisfy average peak demand across a given POI, rather than trying to provision enough bandwidth to satisfy constant 100% utilisation.

The average amount of CVC provisioned per AVC across the NBN is around 2.3Mbps, up from around 1.8Mbps during the worst of the pricing war. If we assume a provisioning of 2Mbps CVC for a typical 50/20 service, then the NBN cost-price of such a service is $691. NBNCo are attempting to bring this figure up to 2.5Mbps, however this problem is indicative of a wider issue with NBNCo’s pricing scheme, which is acting as a huge barrier to uptake of higher speed plans.

The cost of bandwidth is fixed. That is, it costs NBNCo the same whether you download 1GB or 1TB. Thus, CVC is a manufactured cost. CVC was conceived in order to help fund the rollout by minimising the impact on the public treasury and is essentially a bandwidth tax. It also served a political purpose; the Liberal Party would attack the NBN for being an unprofitable venture and so CVC was used to ensure that the network would be profit-making early in its life to pre-emptively quash such an angle from its detractors.

Under the original FTTP plan, the CVC charge was to be gradually lowered and then removed as NBNCo’s costs fell. The key to making this work was that FTTP’s maintenance costs are extremely low, and beyond the initial rollout, overheads would be somewhat fixed outside of upgrades and emergencies. However, in late 2013, plans changed. With the change to FTTN/HFC/FTTB/anything-that-isn’t-FTTP, NBNCo could no longer rely on the assumption of low and fixed overheads. FTTN and HFC are extremely expensive. Telstra’s old copper network alone costs NBNCo around $1bn a year simply to keep running, let alone fault fixing and upgrades.

The reality is that the change in rollout plans has precluded the elimination of the CVC charge, as was originally planned and has locked out much of the rollout footprint from being able to attain even the currently available high speed plans. Broadband prices will continue to remain artificially high simply to cover the inordinate expense of maintaining this failed, dilapidated, stillborn network while also trying to generate the 7% ROI expected of it.

Read the epic conclusion in Part 6.