Spectrum is all around us. These invisible waves are essential for devices that communicate wirelessly. Whether using a mobile phone to make a phone call, send a text message or browse the web, the device will use spectrum to connect to a local mast and carry out your connection command.
Only a finite amount of spectrum is available. All other aspects of running a mobile network can be bought – masts, wires, electronics, land, contracts, experienced management, branding, and arguably even customers. But, new frequencies cannot be plucked out of thin air. Its allocation in the UK is carefully controlled by Ofcom which authorises and manages the use of spectrum, taking into account the ever-growing demand for it.
At the moment, the four mobile network operators in the UK have fairly similar levels of available capacity. That is not an accident. Ofcom has long believed that completely symmetrical levels could reduce competition by making coordinated effects become more likely; and significantly asymmetrical levels could reduce competition because it would be hopeless to try to compete with the market leader.
As the CMA proceeds with its in-depth phase 2 investigation of the Vodafone/Three merger (discussed in a previous blog post), it will undoubtedly be mindful of any risks facing competition if the balance of allocated spectrum is disrupted. Ofcom has gone to great lengths to maintain that balance, and there are no significant spectrum auctions in the pipeline.
Ofcom policy
The decision on whether to allow the Vodafone/Three merger will obviously be taken by the CMA rather than Ofcom. However, Ofcom’s stance will be hugely important.
In December 2022, Ofcom published a paper on its future approach to mobile markets and spectrum. It said that “any future mergers would need to be assessed on a case-by-case basis, rather than on a presumptive view of the appropriate number of competitors” and any decision on whether to allow a merger “will turn on the effectiveness of competition that can be expected in the market after the merger”. This was widely interpreted as a softening of Ofcom’s stance towards consolidation in the industry. That may be the case, but it always seemed to me like a neutral statement of the legal framework.
Ofcom’s detailed stance towards spectrum holdings seems more likely to be indicative of the outcome in the Vodafone/Three case. A regulator’s policy stance on an issue might change, but only if the underlying rationale also changes. That would require a change in the facts and/or the economic theory. The finite nature of spectrum leads to some unique features for policymakers trying to stimulate competition between operators.
Asymmetric spectrum holdings
When Ofcom has awarded spectrum licences in the past it has been careful to avoid creating any significant asymmetries because of the risks posed to competition. In an award made in March 2020, it said that “competition may be weaker either if one (or more) [network operator] has a very high share of spectrum, or one (or more) [network operator] has a very low share”. Ofcom has taken the view that, if one network operator is left with a high share of spectrum, it may end up with an unmatchable competitive advantage in some customer segments even if the other MNOs remain credible market players. It could be able to offer superior services that its rivals will simply be unable to replicate. While there may be short term benefits that consumers may realise, such an advantage could also reduce competition and lead to higher prices or lower quality services overall. Ofcom gave the example of one operator having a 40% share of overall spectrum and the other three operators having an average of 20% each. It said that this level of asymmetry would give rise to detrimental effects.
Another concern that Ofcom has previously voiced is spectrum hoarding. If a network operator already holds a large share of spectrum, it will likely be less incentivised to make good use of any additional spectrum that it acquires in an auction. Comparatively, a network operator with less capacity will likely want to use its spectrum more quickly and likely would have been a more aggressive competitor had it won the spectrum.
Ofcom has also raised the concern that a network operator holding excess spectrum capacity may distort the market by threatening rivals with aggressive price cuts. This, in turn, may put rivals off from competing aggressively alongside them. The network operator will then become the only viable option for smaller wireless communications services seeking spectrum and this may lead to a softening of competition at the wholesale level and in turn hinder competition at the retail level.
With excess spectrum comes the ability for a network operator to launch new services without impacting their spectrum allocated for other services they already offer. This, as Ofcom have said, will make it harder for rivals without excess spectrum to keep up, as they may have to re-purpose some of their existing spectrum in order to launch new services.
Considering the risks that Ofcom believes arise where one operator has more than twice the average of its rivals, it is not surprising that it chose to cap BT-EE’s spectrum holdings in 2018 (when BT-EE’s share of ‘immediately usable’ spectrum was reduced from 42% to 39%) and again in 2020 (when BT-EE was restricted to acquiring a maximum of 120 MHz of the 200 MHz to be awarded). If the Vodafone-Three merger were to go ahead, it would leave the merged party with around 54% of all useable low band spectrum, which would be over four times the amount of spectrum held by the closest rival, BT-EE (13%). Even if Ofcom has been relatively tolerant of spectrum asymmetries in individual bands in the most recent auction, this level of asymmetry in low band spectrum appears high. This is particularly stark when combined with sites which together create the capacity needed to serve customers.
The CMA would need to make itself comfortable that these levels of asymmetry will not enable the merging parties to wield capacity strategically post-merger, softening competition and reducing investment.
Sites asymmetries
In March 2020, Ofcom said that, “[i]t remains our judgment that competition concerns about asymmetry in relation to capacity and average speeds may generally arise when one MNO has around 37% of overall spectrum“. It therefore seems that Ofcom’s concerns about asymmetric spectrum holdings are ultimately a concern about asymmetric overall capacity.
Both spectrum and physical infrastructure contribute to an operator’s overall capacity, i.e. consumers’ ability to download and upload data using their mobile devices. As stated above, spectrum is the commodity that cannot be increased. Additional infrastructure sites can always be built, at least in theory. However, this is too simplistic as it is not so easy to build sites in a short time frame. An asymmetry in sites would therefore exacerbate the spectrum issues.
As I mentioned in my previous post, the four operators currently have around 18,000 to 20,000 sites each. A post-merger Vodafone-Three would have 36,000 sites (although it seems to intend to reduce that to around 26,000 sites).
The merger would also leave the parties with a foot in both network sharing agreement camps. This would disrupt EE’s network sharing agreement with Three, and O2’s with Vodafone, particularly because, post-merger, Vodafone-Three would not need to depend on these agreements as much. This could leave rivals in a less attractive position and could further hamper their ability to grow their number of sites.
The parties will not only have a much larger estate of sites than any of their rivals, but their sites and spectrum will combine to provide the merging parties with over 60% of total market capacity, which is much larger than the other two operators and almost double what a level playing field would require in the three player market.
If the concerns that Ofcom has previously raised with capacity are applicable by analogy here, then the number of sites will also be a big issue in the CMA’s phase 2 analysis.
Remedies
To overcome some of these asymmetry issues, the parties may end up offering to (or being forced to) divest some of their spectrum or sites to make way for a new fourth entrant. To some extent, this is the obvious thing to do, and it has been tried multiple times across Europe. Most recently, the European Commission cleared the Orange-MásMóvil merger in Spain on the condition that spectrum was sold to a new entrant—the Romanian mobile virtual network operator Digi—in the hope that it would help foster competition post-merger. Digi seems well placed to enter the market. It is the largest and fastest-growing virtual network operator in Spain with a relatively large, fixed broadband network that it was continuing to build.
In the UK’s case, the CMA would need to be confident that a new fourth operator would be forthcoming, and would have a good chance of succeeding. There are some well-established virtual network and fixed broadband operators in the UK, but it is difficult to see how a fourth player that is less experienced in the UK market, less well-resourced and less asset rich than its competitors would re-create the level of competition we see in markets today.
If the CMA finds that the merger is expected to result in a substantial lessening of competition, the remedies seem difficult.
Conclusion
The UK sits quite low in the international league table for 5G rollout. Two of the four network operators are failing to meet their cost of capital. Something needs to change. But, that does not necessarily mean that this merger can easily be cleared by the CMA or that a viable remedy is easy to find. It may be that the economic theory behind Ofcom’s spectrum policy yields a different result when applied to the specific facts of this merger, but Ofcom might struggle to rectify the competitive balance if the merger ends up having an adverse impact on the sector. There are no significant spectrum auctions in the pipeline, and it seems unlikely that Ofcom could order the divestment of sites.
The next step will be for the CMA to publish the full text of its Phase 1 decision and the list of issues that the inquiry group is intending to address in Phase 2. The statutory deadline for a final decision is currently 18 September 2024, although we should expect the eight-week extension to be taken in a case like this. The final decision will therefore be in mid-November (and any aggrieved parties can get their appeals filed before Christmas!).
