The Developing Problem of Mobiles

News that regulators in Europe have approved the merger of Germany's third and fourth mobile operators have been followed by the suggestion that "clearance of the E-Plus deal is tipped to spark consolidation throughout EC markets."

Here in Australia we should not be surprised by this development. After all we had five full network operators at one stage, which reduced to four when One.Tel imploded. That was further reduced to three when Vodafone and "3" (Hutchison) merged to form VHA.

In Australia there was no attempt to pursue the additional remedies or undertakings that are occurring in Germany. The merged operator is required to make one third of airtime available to three Mobile Virtual network Operators, and to relinquish excess spectrum to either a new entrant or to the three MVNOs.

At least one reason for not acting on spectrum here is that the merger took place before the Digital Dividend auction. As it transpuires that auction only interested a half-hearted bidder in TPG - and VHA did not bid for 700 MHz. That means there is a fully available unsold lot.

Secondly two of the entrants started their lives as "MVNOs" - though more accurately resellers - in One.Tel and Hutchison (pre Orange).

The term MVNO itself has a very loose collection of meanings. When initialy coined it referred to a genuine virtual operator who ran all the components of a network other than the base stations. That specifically included an HLR and the ability to enter directly into interconnect contracts to earn the rents from fixed to mobile termination.

In practice it now refers to an operator that buys airtime but provides no mobile functionality. That includes doing its own billing, rating and value added services, but little else. The user of an MVNO service sees no product differentiation other than price and brand. Typically operators embrace MVNOs as a way to run "off-brand" strategies to target the "value" end of the market without having to erode the margins they earn from the base that is prepared to pay a brand premium.

There are two policy questions. The first is exactly how many operators is a viable number given the economics of network operation. The second is what is the number of operators required to provide effective competition.

At the bottom of this post is a very short history of the Australian cellular mobile market. It traces the progress of the number of operators from one, to three, to five and then back to three. In the current configuration, the third placed operator is still struggling in the market.

A key differentiation is that the backward compatibility of the standards built around GSM/3GPP

Opinions differ dramatically on what the end point will be. The current agreed facts are that Telstra has significantly wider coverage, and despite being relatively spectrum constrained (compared to customer numbers) has mostly better capacity within coverage areas. It is also generally agreed that the traffic volumes don't exist to justify two operators in the more marginal areas.

Vodafone is currently making overt calls for mandated roaming on Telstra's network.  The consequence of mandating roaming would be for Telstra to make no more investment in extending its coverage. The $100 million currently up for grabs over 4 years is much smaller than Telstra's directly funded network growth.

The problem for mobile markets is that a dominant player in coverage gets a disproportionate share of customers and gets to charge a price premium. (For students of Industrial Organisation it is worth noting that industry EBITDA margin is distributed in proportion to square of market share - which would be the expectation of theory for a Cournot oligopoly).

That operator gets to reinvest some of that extra profit in extending the reach of the network and consolidating its lead. Before he departed Optus, Kevin Russell was on the record as saying that the strategy for Optus had to be matching Telstra on coverage and outperforming on customer service. The problem is that the first of these is probably now unachievable, and their resorting to adds (some of which the courts found misleading) to claim that near enough is good enough is probably a concession on that point.

The question is whether there are any strategies that stop the market completely tipping to monopoly.

The short answer is probably, but I struggle to see them, and more importantly they probably require co-ordinated action by VHA and Optus and they have a poor track record.  One such scenario would see VHA retreat to a metro only strategy and pay a premium for roaming onto the Optus network. That creates VHA as a differentiated player with price plans to match - but gets a higher contribution to Optus for the little roaming.

The answer to the first question is that the viable number of operators may only be one, but there may be strategies relying on differentiation for two or three.

The question of how many are needed for a "competitive market" depends a whole lot on the definition of competition. Unfortunately the current definition used by the courts in Australia identifies "vigorous rivalry" as sufficient, and explicitly allows for what economists call "strategic interaction" in which the decision of one player depends on the decisions, or possible decisions, of others.  That's how the Cournot oligopoly mark-up comes about.

Also unfortunately oligopolistic markets are hard to analyse in theory. Straightforward optimisation models are limited and game theory based models are open to too many variables. In a 1973 paper 'A simple model of imperfect competition, where 4 are few and 6 are many' Selten determined that with less than 4 participants strategic co-ordination would occur, while it wouldn't with six. As the abstract says

"The theory presented in this paper investigates the connection between the number of competitors and the tendency to cooperate within the context of a symmetric Cournot model with linear cost and demand, supplemented by specific institutional assumptions about the possibilities of cooperation. Cooperative forms of behavior are modelled as moves in a non-cooperative game. The proposition that few suppliers will maximize their joint profits whereas many suppliers are likely to behave non-cooperatively does not appear as an assumption but as a conclusion of the theory.provided."

A 2004 paper using experimental economics concluded the numbers were smaller than Selten estimated. The abstract notes:

"In this paper we investigate how the competitiveness of Cournot markets varies with the number of firms in an industry. We review previous Cournot experiments in the literature. Additionally, we conduct a new series of experiments studying oligopolies with two, three, four, and five firms in a unified frame. With two firms we find some collusion. Three-firm oligopolies tend to produce outputs at the Nash level. Markets with four or five firms are never collusive and typically settle at or above the Cournot outcome. Some of those markets are actually quite competitive with outputs close to the Walrasian outcome."

These theoretical constructs would support the practical claims - such as these made in Canada - that the current market structures aren't competitive.

The policy question is what to do about the conjunction that mobile markets are trending to monopoly, or to very close to monopoly, while effective competition will require four viable operators or more.

One way would be to pursue the strategies we have pursued for the last three decades of regulating mandated access to the infrastructure. This has been recently called for - effectively - in Australia by a fixed operator wanting to resell the Telstra 4G mobile product. Telstra has predictably resisted the call.

(As an explanation Macquarie has actually made the complaint on the basis that Telstra's refusal to supply the monopoly mobile product is having the effect of reducing competition in the market for fixed services to business clients which Telstra sells in bundles. The complaint relies on Part XIB of the Competition and Consumer Act that relies only on an effects test. What Macquarie wants is the right to sell the services, which Telstra could agree to do, or the ACCC could mandate under Part XIC.)

The case rests on the assertion that Telstra's position as the only supplier to have chosen to build base stations in specific markets constitutes a "monopoly." Ignoring for the moment the fact the claim can be made on an effects test, is Telstra's purpose in making 3G available for resale but not 4G to advantage itself in the fixed market (or even mobile)?

A reasonable argument can be made for declining to resell on purely technical grounds. With a mobile network the core objective is to provide consistency of service - in coverage and capacity. Allowing another provider air time resale of a new network has the potential to disconnect the network planning function from sales activity resulting in poor network performance.

The second question is what it is that is being sought as the remedy. Simple airtime resale still results in an uncompetitive position. The real benefit of fixed and mobile bundling comes from service integration, not just pricing plans. To achieve that the person seeking access needs a deeper relationship than airtime sale - it actually needs the ability that a true MVNO brings.

And this is the longer run policy challenge in mobiles, especially in Australia. Creating an open access fixed line market levels the playing field (though even on that field Telstra and iiNet are dramatically winning), but that levelling dissipates if the mobile network operators can provide service integration not available to others.

Hopefully the ACCC and Government policy makers will look at this latest claim in the context o the longer term evolution of mobile markets and the access requirements for integrated services.

A very short history of the Australian Cellular Mobile Market

In Australia the policy history is that, in common with all other markets, mobiles started as a single monopoly, usually but not always operated by the incumbent fixed line operator. This was the standard configuration for first generation - basically analog - mobiles. (There is another fascinating tale to be told about the global approaches to the technology choice and standards through the four generations, but that will be avoided here).

With the advent of 2G and the related issue of the feeing up of additional spectrum, most markets moved to a multi-operator model. Typically this was to licence two operators per market. In australia the extensive report by AUSTEL can be effectively paraphrased as "Well one or two is too few, and infinite or unlimited number is too many, so let's have three." And so the licences that were won by Optus (it was bundled with the fixed licence) and Vodafone (as a sideline, the licence originally went to a group that included a sister company of what became AAPT).

So Australia became one of the earliest countries to go beyond two mobile licences. These 1991 licences were mandated for 2G GSM and in the designated 900 MHz band. Telstra was required to make its AMPS product available for resale to entrant carriers at a discount, and to no others, and carriers were required to achieve a 80% population coverage by 1997. The licences prohibited new AMPS networks being built and required switch off of AMPS by 1 January 2000.

I am not aware of any detailed commentary on the importance of AMPS resale, other than that I believe Vodafone never used it and that Optus in subsequent proceedings has referred to it as an essential condition.

As it transpires the coverage obligation was superfluous. Early on Telstra and Optus started to compete on coverage - but this was promoting regional coverage to purchasers based in the cities.

When the 850MHz spectrum used for AMPS was becoming available, the international community had also designated 1800MHz for mobile expansion. Both lots of spectrum were auctioned together with no technology choice mandate. This saw three new players acquire spectrum AAPT, Hutchison and One.Tel.

AAPT never developed a network (another story), Hutchison built an 850 CDMA network in Sydney and Melbourne, and One.Tel built a 1800 GSM network in the capital cities. Both entered into roaming agreements with Telstra to use Telstra's infrastructure outside their footprints.

By the time of the 2 GHz auction (spectrum designated internationally for 3G) took place there were only four networks, One.Tel having collapsed (once again, another story, of which their failings in mobile were only a part). At the time Hutchison wondered whether it needed to bid since it had extensive unused 1800 MHz holdings. They initiated a process to have 1800 also designated as 3G so that equipment could be made for it...this eventuated, but only long after the auction was complete.

There followed a short phase of pair-wise co-operation between Telstra-Hutchison and Optus-Vodafone. The former was fractured when Telstra utilised the decision by AT&T to build a 3G network at 850 MHz (in typical US fashion the market size of the client was sufficient to get the technology variant delivered).  Hutchison and Telecom New Zealand as owners of 850 MHz spectrum in Australia and New Zealand were unable to reach agreement to extend the co-operative build to this network. Eventually Hutchison merged with Vodafone, but the post merger execution resulted in network failures that sparked a customer exodus that continues today. Interestingly through this period Optus's market share has stayed relatively static, and Telstra has gained more from the structure.