Australia’s Broadband Experience – A Case Study in how Government Broadband Policy can go horribly wrong

I was given the opportunity at Broadband World Forum Asia (in Hong Kong on 11-12 April) to present some of my views on Australia’s Broadband Policy disaster.

It seemed to provoke lots of good questions and interaction from the audience.

Telecom Times published the piece below – the full article is available here.

The presentation is available below.


  • David Havyatt

    Well I cant follow that presentation just from some slides.

    Absolutely nothing wrong with the monopoly broadband argument, it still is a natural monopoly just as the copper loop was and is. Yes in markets that are almost exclusively high rise – just like Australia’s capital city CBD’s – you can get fibre competition because the cost structure changes.

    There was massive execution error with NBN Co – the three main ones being (1) publishing a Corporate Plan with targets before the agreements with Telstra were operationalised, the 2010 plan caused a world of pain (2) not pursuing the ‘build drop’ approach from day 1 (3) signing two year rather than four year construction contracts.

    All of that created the political risk that was realised with the Coalition government deciding on the technology pivot.

    • Thanks for the comments David. Appreciate your thoughts.

      I won’t comment on the execution issues at NBN Co – water under the bridge now.

      The real problem I have is with the natural monopoly argument. What service are you referring to? In the old days we had local fixed telephony which was a natural monopoly just like electricity, gas and water supply. But it was blown away by technology improvements – mobile, VoIP, broadband etc.

      Are you arguing fixed broadband is a natural monopoly? Mobile, HFC and DSL networks all offer similar speeds now with incremental upgrade paths to higher speeds with investment in deeper fibre networks (more cell sites, node splitting and FTTdP respectively). FTTP is likely to be a natural monopoly for speeds of 1Gbps and above in Greenfields situations. In other Brownfields situations deeper fibre backhaul for mobile, HFC and DSL networks will mean a low cost jump to FTTP so that eventually 1Gbps and above can be competitive in residential areas (just as it is today in CBD areas for business).

      This is not only in dense cities with apartments (eg. Singapore, Hong Kong, Tokyo). We are seeing this trend in all metropolitan areas in developed and developing countries. Rural and regional areas are likely to remain natural monopolies (or require significant subsidies) for some time to come. This is where government investment policies should be focussed for social and economic reasons. In metropolitan areas government should look to foster competing private investment infrastructure investment in deeper fibre networks.

      Maybe it is too late to break up NBN Co – but eliminating the anti-competition regulations would be a good first step to ensuring incremental private fibre investment is encouraged rather than distorted. NBN Co would then need to respond with more fibre investment of its own. Surely this is a good thing.

      • David Havyatt

        ‘Natural monopoly’ is a technical term that identifies the circumstances in which all the desired output can be delivered at lower cost by one provider rather than by more than one provider…a condition known as cost sub-additivity. That applies to any fixed line network. It doesn’t matter whether the service is substitutable or not (i.e. f only half the population has a fixed line connection and the other half decide to use mobile, the fixed line connections are still provided at lower cost by one provider rather than two.)

        It doesn’t apply to mobiles because typically the density of base stations required to provide service is such that it isn’t really cheaper to have one network – except in the case of rural and remote areas. That is an interesting policy question now before the ACCC – and the ultimate question is exactly where is the line you draw between the two? (My answer is that the existing model works fine, Telstra is disciplined on how much it can charge by the markets where it faces competition, so it doesn’t actually get to monopoly price even in the areas where they have a monopoly. Their service costs more because they choose to serve more high cost areas).

        That everyone else in the world has a policy of promoting inefficient entry is no reason to do it in Australia…we are the ones who proved the case over natural monopoly of a service with HFC duplication.

        The logic of the NBN was premised on an FTTP deployment to all premises. The logic that it was cheaper to do that than the two step process was and is unassailable.

        Meanwhile I really want someone to do more analysis of the Akamai data – those low download outcomes are not a consequence of the last mile speeds. I wouldn’t be at all surprised if the major cause is the way transit works in Australia (flip side of people don’t know how to do peering properly). I asked Akamai by email whether they could break down that speed data by ISP (they have the originating IP address so should be possible) – but got no reply. Maybe you know someone who will.

        • David, you seem to be describing FTTP as the service rather than the technology used to deliver the service. If half use FTTP for their broadband and half use mobile then you seem to be saying they are each separate monopolies. But if customers can choose between them for their broadband needs then surely that can’t be a monopoly.

          On your definition if I want to travel from my home to the CBD then each of the available infrastructures (road, train, tram etc) are what economists would call monopolies. But to me they are alternative choices I can make to achieve the same objective.

          Broadband is now the common objective that can be serviced by mobile, FTTx and HFC networks (all networks that were originally built for different objectives). With more fibre investment in each of these speeds will increase and the networks will converge towards FTTP. Each network owned by different companies will actually compete unless the regulator allows monopolies to form through takeovers. I am sure many would argue that this should be allowed to happen but I for one would argue strongly against such an outcome as it would just mean higher prices and less innovation.

          Akamai’s reports are useful but the cause of speed differences is multi-faceted as you know. Akamai have reported the Average Peak Connection Speeds for some time (but curiously not in their latest report). This is the average of the highest connection speed calculated for each unique IP address and as a result a much better reference point for access network speeds as in a contention-less connection the access speed will be the limiting factor (although WiFi is a bigger factor in FTTP networks). Australia’s result has steadily been dropping relevant to others on this measure for some time.
          The Average Connection Speed measure is probably a better for assessing the overall end-user experience but the causes of differences are difficult to pin-point.

        • David Havyatt

          Gary, you asked by tweet why if broadband is a natural monopoly the provider needs legislative protection from competition.

          This confounds two separate issues and overlaps with a third.

          A natural monopoly that is unregulated can protect itself from inefficient overbuild if it is able to practice differential pricing. Assume that there are only two geographies, a low cost area and a high cost area. All providers face the same marginal costs in the two areas, but the monopolist still has scale advantage and so should always win on price. More dramatically, the high cost area is usually less attractive to competitors (your concession that maybe a natural monopoly exists in regional and remote areas). As a consequence the monopolist if needs be can add additional margin to the unattractive areas to subsidise its services in the low cost region.

          That all changes when you institute uniform pricing – as we did for NBN Co. Now the low cost areas are particularly attractive for the competitor because the natural monopolist is forced to price above cost. This is why in Oz we’ve just had an extensive inquiry about levy TPG and other fibre operators need to pay for the provision of satellite and wireless, but it misses the point that TPG is still “cherry picking” low cost to sere premises within the fixed footprint.

          These issues don’t emerge in mobile because the connection point isn’t geographically located. Telstra can charge a higher average price because all users get benefit from the wider geographic footprint. But you see the idiotic arguments from Vodafone arguing for mandated roaming because consumers in regional areas supposedly deserve to pay the low prices that Vodafone charges – ignoring the fact that VFs average price is low only because they choose not to serve high cost areas.

          The third area of concern is the difference between infrastructure competition and service competition. The quality of a broadband service depends on a lot more than the access speed, it depends on the contention ratios inside the service providers network and the decisions the service provider makes about where it connects to ‘the Internet’ including whether it makes direct connection to large content sources or accesses them through transit (or peering) arrangements. The problem with all models of access network infrastructure competition that we have seen to date is that it constrains the development of this retail market. Absent ULL declaration there would have been two broadband ISPs in Australia. Consumers in markets served by some greenfields operators do not have a wide choice of retailers, as is also true of TPG. “Open Access” alone doesn’t result in retailers offering.

          The NBN model chosen in Australia was the right choice. The company made a number of execution errors, and the model has been destroyed by the MTM which in turn was based on a totally flawed strategic review.

          • Thanks again for the comments David.

            I get your points about differential pricing. Here in Hong Kong we are seeing extreme variations in pricing between competitive and monopoly fixed broadband areas. Hong Kong still has approximately 20% of households under monopoly because costs to islands and villages are too high relative to market size.

            The incumbent has raised prices considerably in these areas to compensate for lower prices where the competitor (ie my company HKBN) has taken away significant market share in competitive areas. By the way we have over 50% residential market share in the the 80% of households we have coverage – so the monopoly has been well and truly broken in 80% of Hong Kong.

            Uniform pricing is important politically in Australia but I never understood why it was not a cap rather than the absolute price. NBN Co would then have been able to compete by lowering prices in cities to match competitors. Sure this would have put pressure on returns given it could not raise prices elsewhere like has happened in Hong Kong.

            But as mentioned in earlier comments there are always geographic areas which are either un-economic or monopolies for telco services. These must involve government intervention (USO style) and price regulation to ensure services at socially acceptable outcomes. Even in Hong Kong there is still a fixed telephony USO – without it there would be no fixed telephone service to outlying islands and villages. The incumbent uses this to provide ADSL services still where so fixed broadband is more or less universal as well.

            But Uniform pricing or a price cap policy just adds to making more areas un-economic or monopolies. In Hong Kong we are still building to some villages because we are attracted by the higher prices in those locations than in the more denser areas. If price caps were applied in Hong Kong then we would not build to those areas.

            The same principles apply across all countries in my view.

            There are locations which are competitive and private operators should be encouraged to invest if they believe returns on the investment are acceptable. The locations which are un-economic or natural monopolies need government support and regulation to achieve socially acceptable outcomes.

            I believe the same issues do apply in the mobile market. Each operator’s prices are geographically uniform because the service is mobile as you say. This eliminates the need for price uniformity regulation. There are areas where Telstra is the only provider (monopoly areas) and areas not serviced (un-economic). Government has decided to support build in the un-economic areas through the Blackspot program. If roaming is declared then investment by Telstra will be less than if unregulated as it’s returns in monopoly areas will be reduced, in effect creating more areas which are un-economic and (probably) necessitating more subsidies through the blackspot program – hence the idiocy in roaming regulation which we clearly agree on.

            But mobile highlights exactly my point. There is natural uniform pricing and no need to enforce a natural monopoly in specific areas through regulation. Where the natural monopoly geographically exists in regional and rural areas Telstra is able to defend its monopoly without regulation supporting it and still be profitably in the cities where there is competition. The competitors compete where they can and the government subsidises extra build out to satisfy social and political needs. I can’t see why this model is also not applicable to fixed broadband with a uniform price cap policy – albeit the price cap will mean more monopoly and un-economic areas than without the price cap.

            The focus on retail competition in Australia has been a failure. It might work in a market where technology developments run slowly and markets are relatively stable (eg. electricity, water etc). Retail innovation can then be proportionally more valuable – eg. focus on marketing, customer service, bundling etc. But in telecommunications technology has provided too many disruptions, innovations and value increasing opportunities and network investment has been squashed by the focus on retail over infrastructure competition. Australia’s slide in international broadband rankings is the result. Sure there are less retail options for customers but this is less important than the benefits that come from the investment in new technologies. Again mobile shows the way. Retail competition has been limited but the gain has been significant increases in technology investment providing far superior services over time. If you ask a mobile customer today if they would rather have a choice from many GSM retail providers all operating on a common mobile network or a 4G service from 3 different operators I am pretty sure I know what they would prefer. Even mobile users in monopoly areas have benefited from this investment as it more cost effective to make these national networks than isolate just to competitive areas.

            The fixed and mobile networks are in any case converging. Densification of cell sites and 5G technologies are going to make it a viable substitute in more segments of the market. Unless a crazy government decides to apply the same fixed broadband regulatory regime to mobiles we are likely to see mobile effectively out invest fixed broadband in Australia. This is unfortunate as it will come at a higher cost than if both markets were on a level playing field and synergies could be used across networks.

            The original NBN 1.0 model was a good choice but not necessarily the right choice. An investment in an national FTTP network would have been a huge national asset that enabled Australia to leap-frog into the future and essentially recover the lack of investment in fixed telco infrastructure of the 2000s. Mobile networks would have also benefited from this fibre investment for the ongoing densification of cell sites. Now that we don’t have a commitment for a national FTTP network from either side of politics we must rely on private investment incentives to complete the job. This would have been the best model in any case back in the early 2000s after the pay TV wars had left Telstra with a HFC and copper monopoly. But Telstra’s share price during the privatisation process was deemed a priority rather than a pro-competitive infrastructure investment climate.