This article discusses how the CMA has reached a ‘tipping point’ in ‘Big Tech’ merger enforcement, what the Competition Appeals Tribunal could do in Microsoft v Competition and Markets Authority to recentre and refocus the CMA’s work, and why this all matters to digital markets, and beyond.
War of the words
On 26 April 2023, the CMA shocked antitrust lawyers, journalists and the wider (mostly gaming) community by rejecting Microsoft’s remedies offer and prohibiting its acquisition of Activision Blizzard, on the basis that it would “alter the future of the fast-growing cloud gaming market, leading to reduced innovation and less choice for UK gamers”.
Microsoft condemned the decision as discouraging “technology innovation and investment in the United Kingdom”, while Activision said it would “stifle investment, competition, and job creation throughout the UK gaming industry”. Nvidia and Boosteroid bemoaned the “even deeper catalog of games” they would have received.
On 16 May 2023, the European Commission joined over thirty regulators around the world in approving the transaction, agreeing to a commitments package that it heralded as “a significant improvement for cloud gaming as compared to the current situation”. In turn, the CMA criticised the Commission for allowing “Microsoft to set the terms and conditions for this market for the next 10 years”, which was – in CMA chairman Marcus Bokkerink’s words – “turning a blind eye to anti-competitive mergers”.
“This trial has been a farce – I demand an appeal!”
On 24 May 2023, Microsoft appealed the prohibition decision, alleging fundamental errors in the CMA’s counterfactual, market definition, consideration of out-of-market constraints, market share calculation, the assessment of ability and incentive to foreclose, its dismissal of agreements with Nvidia, Boosteroid and Ubitus and its failure to have due regard to the need for comity with other authorities. Activision is also seeking to join as an intervenor.
A hearing date has now been provisionally set for 24 July 2023, with a judgment expected as early as September. If the appeal goes Microsoft and Activision’s way, a remittal decision could be made in time for Christmas 2023, keeping deal hopes very much alive.
So, who is right, what went wrong, and what can we expect (or hope for) from the appeal?
The Road to Prohibition was Paved with Good Intentions
The journey to the CMA’s prohibition of Microsoft / Activision – and the misuse of the counterfactual – began long before Bobby Kotick and Phil Spencer sat down to discuss possible strategic opportunities between the two businesses on 19 November 2021.
Little did they know then that – 18 months later – their predominantly vertical merger, viewed by most competition regulators for the last 50 years as presumptively pro-competitive, would be tripped up, not by a concern they would plan for – in console gaming – but by a strange unfamiliar, the ‘ecosystem’ theory of harm, a highly dubious counterfactual, and a questionable vertical ‘analysis’. And all this in a segment as amorphous as its namesake – the cloud – representing less than 1% of the total global gaming market. Nor could they have anticipated that no argument, no counsel, no remedies, no benefits, no plea for comity, and no pressure from every other international peer (except the US), would dissuade the CMA from its drive to prohibit.
“God help us, we’re in the hands of engineers economists”
In fact, the drumbeat for just this type of radical intervention has been growing steadily louder in the UK over the past 15 years, driven by a growing suspicion of large tech company acquisitions in digital markets. First, Google/Doubleclick was cleared by the European Commission in 2008, then Facebook/Instagram was cleared by the then Office of Fair Trading in 2012 and Facebook/Whatsapp was cleared by the Commission in 2014. During the same period, Microsoft completed a number of high-value acquisitions, first of Yahoo! Search Business (cleared in 2010) then of Skype (cleared in 2011) and finally LinkedIn (cleared in 2016). Verizon acquired the remainder of Yahoo in 2016 without opposition. By 2018, all the big names of tech were involved, with Apple acquiring Shazam in 2018, again cleared by the Commission. In fact, according to the CMA, over 400 acquisitions by the five largest digital firms from 2008 to 2018 went unchallenged.
All of these mergers took place in the digital sector, seen as critical to millions of consumers, and often involving ‘markets for attention’, network effects, multi-homing, and the buzzword of the 2010s, “Big data”. The CMA sensed that it was missing something and that the conventional analytical tools it had used to examine mergers might not be suited to the digital markets of the 21st century.
Enter Lear and Furman, with a new ‘shoot first’ approach to merger control
It was against this backdrop that two voices that would become critical to the future of UK merger control entered the debate – Lear and Furman. Both were tasked with examining the perceived enforcement gap and how to close it. Reporting in 2019, their recommendations would have significant ramifications for future merger control and, ultimately, lead to the prohibition of Microsoft/Activision.
Simply, Lear called for a more aggressive and risk-taking approach by competition authorities. Where there was uncertainty, Lear counselled for intervention. Where there was little evidence, Lear encouraged ‘joining the dots’. Where the facts and analysis collected did not meet the usual threshold for intervention, Lear advocated for that standard to be flexible. Similarly, Furman called for greater intervention in digital markets to tackle the perniciousness of ‘tipping points’, where ‘winners take all’ dynamics prevail, and welcomed more merger enforcement, believing that it would simply right the ship and reduce ‘false negatives’ (ie, the approval of mergers that should not have been allowed). Given the otherwise utterly permissive past approach of authorities to digital mergers, where – they asked – was the risk in greater enforcement?
“Even Nedry knew better than to mess with those raptor fences the counterfactual!”
Since then, however, a different kind of tipping point has been reached in the CMA’s approach to the counterfactual and one just as critical to the future of merger control.
Hitherto, the counterfactual, that humble baseline against which the CMA measures the potential harm arising from a merger, was a relatively benign tool, mostly adopting the ‘prevailing conditions of competition’ as the starting point for the case team’s in-depth analysis. It usually took market data and competitive interactions for the preceding 3-5 years to ascertain the extent of competition between the parties, relative to other players. It had occasionally been flexed but usually to take account of a weakening, rather than strengthening, competitive tension between the parties (most common in ‘failing firm’ scenarios). In the competitive assessment, possible future developments were considered (including, for example, the entry of new players) but merger control did not often stray into guessing what the future might entail, given the inherent uncertainty (and difficulties in substantiation) such an exercise would entail (absent a time machine, or the kind of pre-cogs envisaged by Philip K. Dick).
Now in 2019, Furman was calling for merger policy to be “more forward-looking”, to “take better account of technological developments”, and to boldly go where merger economists and lawyers had not gone before – to predict the future! And Lear was encouraging competition authorities to “accept more uncertainty in their counterfactual” embrace a “departure from the current practice” and “be somewhat imaginative with the definition of the counterfactual”. Even if that would risk failing the usual legal bar in terms of evidence and analysis.
However, while it was one thing to develop novel ‘innovation’ or ‘ecosystem’ or ‘dynamic’ theories of harm to tackle genuine issues, which external economists and lawyers could meaningfully engage with and challenge, it was quite another to lift the hood and start to tinker with the counterfactual itself. There was little in the way of guardrails as to the evidential and legal standard that should apply to any such departure. Yet the CMA, first in Meta/Giphy and now in Microsoft/Activision, uncoupled the counterfactual completely from prevailing conditions of competition, wielding it as a predictor of the future, and in doing so entirely decoupled itself from reality.
Lessons for the counterfactual from Meta V Competition and Markets Authority – fast forgotten?
The approach to the counterfactual, and novel theories of harm, ushered in as a result of Furman and Lear was, before Microsoft/Activision, most actively deployed in Meta/Giphy. Here, the CMA was concerned with Meta’s substantial market power in regard to display advertising and worried that the acquisition of Giphy – a small, unsuccessful and unprofitable player – could lead to a loss of dynamic competition and foreclosure of important input. The route to these findings required a very substantial change to the counterfactual, namely, that Giphy would have become much more important in future. Sound familiar?
On appeal in Meta v Competition and Markets Authority, the CAT, while otherwise endorsing the CMA’s approach and substantive findings (thoughts on that for another day), found that the CMA had not adequately explained how it had flexed the counterfactual and on what basis. It, therefore, encouraged the CMA on remittal (made on procedural grounds) to clarify and delineate its counterfactual findings between existing and future levels of competition.
First, in paragraph 68 of its judgment, the CAT proffered a calming counterweight to Lear and Furman on the counterfactual:
“Whilst defining markets in the context of an assessment of dynamic competition affords the competition authority an even greater margin of appreciation than would ordinarily be the case, it is incumbent upon the competition authority to be all the clearer in its analysis when considering the theory of harm and the counterfactual.”
Then, in paragraph 78, the CAT elaborated on its view that “clarity serves justiciability”; in particular, with regard to the counterfactual:
“it is important, when assessing risks to competition in merger cases, that a clear distinction be drawn between what is (the static case) and what might be (the potential case). We also consider – although this is very much a matter of labelling, and so of secondary importance – that a consideration of the counterfactual case ought to be confined to an assessment of static competition, and that an assessment of threats to potential (and, indeed, dynamic) competition be very separately demarcated.”
Thus, the Tribunal clearly signalled that, in scenarios where the CMA was interrogating novel concerns, it must be clear in its approach to the counterfactual and its theory of harm and clearly demarcate between findings based on the prevailing conditions and those based on potential future events happening. Where it chooses to adopt an alternative counterfactual based on a series of future events, it should clearly say it is doing so, and allow the parties (and the Tribunal) to determine whether it has met the appropriate evidentiary bar.
Yet, while the counterfactual section in the Final Report in Meta / Giphy ran to 36 pages, the same section in Microsoft/Activision runs to … 2 ½ pages. And while the CMA purports to adopt prevailing conditions of competition, it then goes on to say in that section how Activision will become much more important to cloud gaming in future (among a number of other assumptions). As we will see, this obfuscation is extremely problematic to the Parties’ rights of defence in the review and on appeal, given the fundamental importance of the counterfactual to this case. And it ignores the judicial steer from the CAT to clearly delineate assessments based on the here and now versus hypothetical futures.
Woah, rewind… Cloud gaming? Wasn’t this all about selling more Xboxes and less PlayStations?
In Microsoft/Activision, the CMA’s main concern since Spring 2022 had been a relatively conventional theory of harm – albeit one with its own problems – concerning whether Microsoft, post-transaction, could harm Sony by withholding Activision games (most notably, Call of Duty). The question, thus, was whether such a strategy could lead to a sufficient reduction in sales of PlayStation consoles to degrade Sony’s competitive position (through lost sales of its consoles, and related products and services) and push enough gamers to Xbox to make the strategy profitable for Microsoft.
Six weeks after its provisional finding upholding that theory of harm (and 28 weeks into Phase 2), the CMA U-turned, much to the disappointment of Sony, issuing an Addendum to its Provisional Findings that Microsoft would actually not have the incentive to withhold Activision games from Sony, a position confirmed in the Final Report. While now merely a footnote to the appeal, this concern’s demise serves as a stark example of what a conventional theory of harm looks like when articulated, and tested, and why the one that survived – foreclosure concerns in cloud gaming – is anything but.
I’m still confused – what is cloud gaming and what’s the concern?
In late March 2023, with only two months to go, and with remedies discussions well underway, the fall of the console gaming theory of harm focused all energies on cloud gaming. In a nutshell, the concern was that Microsoft would be enabled by the merger – through withholding Activision games from all other cloud providers – to leverage its ownership of Microsoft OS and Azure and its position in gaming and game publishing with Activision games in a way that would irreparably harm cloud gaming. For all the CMA’s concerns there remains a genuine and critical gating question – is cloud gaming even a separate market or “just another platform” on which to play games, alongside console, PC and mobile?
More fundamentally, while there is no doubt that Microsoft is strong in OS and the cloud and gaming, just as Google is strong in search, or Apple is strong in mobile phones, the question for the CMA is, as it has always been, not whether the acquirer is large, but whether the ‘merger effect’ will be one of a substantial lessening of competition. The burden remains on the CMA, regardless of the position of the acquirer. And there is precious little analysis as to how acquiring Activision will do this.
But that’s not the worst part.
The Counterfactual – and not the one you were expecting
Unlike console gaming, where Activision was recognised as an important (but by no means unique) games publisher, Activision is not present in (and maintains – most recently at the first case management conference at the CAT – that it will not enter) cloud gaming. Activision games, therefore, represent at present exactly 0% of all games played on the cloud.
Thus, had the CMA followed the important steer of the CMA in Meta/Giphy, it should have acknowledged that no concern could possibly arise on the prevailing conditions of competition.
Instead, the CMA sought to construct an ‘alternative’ counterfactual, whereby, absent the merger:
- Activision would – despite its stated position – enter cloud gaming in the near future;
- All of its games would be made available – including its premium “AAA” content like Call of Duty – on day and date release to all cloud gaming providers;
- Its games would become essential inputs to cloud gaming, ‘must-stocks’ that no other games publisher could match (or, in the CMA’s words, be a ‘sufficiently good alternative’);
- Network effects in cloud gaming would be so extreme that cloud gaming providers would require access to all games to be effective competitors (unlike, for example, online streaming services such as Disney, Netflix or Prime, which compete using differentiated content and incomplete catalogues); and
- Microsoft could (and would) withhold access to Activision titles from other cloud gaming providers to foreclose competition
I have discussed in past articles both the difficulties of compound probabilities in counterfactuals requiring multiple future events to occur and the critical role of the panel in ensuring that a counterfactual finding (especially one as fundamental to all elements of this case) meets the necessary standard of proof.
It should be evident from even a cursory glance just how tenuous the CMA’s counterfactual in this case is. Take one element out or even flex it a little (eg Activision holds back some games from the cloud or releases them later, Call of Duty isn’t as important to cloud gaming in the future as the CMA thinks it will be, other games are just as good, or better, cloud gaming providers don’t need to offer every game to compete) and the counterfactual, and the entire house of cards resting on it, collapses.
Destroying the World
As the Phase 2 deadline loomed large, and with no time to shore up its analysis, the CMA painted itself into a corner. On the one hand, it had constructed a purely hypothetical counterfactual, in a nascent and poorly defined market, on which it placed a questionable ‘ecosystem’ theory of harm. On the other, it remained chronically averse to behavioural remedies, that might offer a fair compromise.
The counterfactual had become the fulcrum on which the CMA’s entire case turned. But, by basing its competition analysis on an imagined future, the CMA could not support any of its concerns with real evidence or analysis and was forced to resort to further layers of hypothecation. Similarly, from a customer benefits and remedies perspective, by constructing a future where Activision offered up all of its games to the cloud, the CMA rendered it impossible for Microsoft to best it. Yet the CMA’s reservations with Microsoft’s remedies proposal were on the same dubious foundation as its competition concerns and at odds with the view of most other authorities, including the European Commission, that commitments would result in greater competition in cloud gaming.
In the final days, instead of recognising that the possibility and scope for any harm were unclear, and accepting a reasonable and proportionate remedy in mitigation, as the Commission had done, the CMA ostriched itself and blocked the transaction outright.
The Tribunal should remit this case…
Microsoft’s grounds of appeal cover every element of the CMA’s case and will provide many good reasons to overturn and remit the “largest case the Tribunal has ever considered”.
The most difficult aspect will be choosing which errors to highlight and major in. Microsoft may elect to dispense with expert reports (and so avoid a ‘battle of the experts’, which is difficult to win in Judicial Review cases) and focus on plain errors the Tribunal can readily see. They stand behind the shoulder of the Panel, comprised of lay members, after all. Microsoft may also elect to downplay the concerns with market definition (never an exact science) and focus on the failure to take account of relevant constraints (a more comfortable judicial review ground) whether in or outside the putative market.
The counterfactual remains the ‘weakest link’ in the case against Microsoft/Activision and simply cannot bear the reliance placed on it by the CMA’s substantive case, its rejection of benefits and its dismissal of proposed remedies. It will be fascinating to watch the critique of this unfold at the substantive hearing.
…with clear guidance
If the appeal is successful, it is critical that the remittal include clear guardrails, not only to ensure a legally sound outcome in this case but in future cases as well:
- First, the Tribunal should restore economically-sound analysis to the heart of the CMA’s work and hit the ‘reset button’ on unconventional and loosely defined concerns (whether labelled as ‘ecosystem’ or ‘innovation’ or ‘dynamic’). The conditions necessary for a theory of harm to hold must be clearly articulated, and objectively determinable. Speculative concerns are antithetical to the Parties’ rights of defence (both during and after the review), the ability of the panel to exercise effective oversight of the case team and, crucially, the Tribunal’s ability to exercise oversight on appeal. Theories, without evidence, must be rejected.
- Second, the Tribunal should urgently clarify the standard of proof applied by the CMA’s panel – not only in its final decision – but in each of the key elements that form that decision, including, critically, the counterfactual. If the CMA wishes to adopt an alternative counterfactual, it must present a clear evidential base for doing so. If the CMA cannot substantiate an alternative counterfactual on the balance of probabilities, especially one relied upon both to find harm and dismiss a solution, it must revert to the prevailing conditions of competition.
- Third, the Tribunal should disavow ‘in the round’ assessments that allow the CMA to lump a series of weak and speculative findings together. This approach undermines the rights of defence by obfuscating the actual decisions and findings made on each element of the CMA’s case. It makes it impossible to disentangle and scrutinise each element of the CMA’s decision. Such clarity, as called for before in Meta v Competition and Markets Authority, does indeed serve justiciability and is an essential quid-pro-quo for the due deference the Tribunal otherwise pays to the panel in judicial review.
- Fourth and finally, the Tribunal must challenge the CMA’s disdain for behavioural remedies – in guidance and practice – and direct it to consider the use of behavioural remedies as a legitimate and proportional tool; in particular, in borderline cases such as this one where the concern is speculative and uncertain. If the CMA is to become ever more expansive in its intervention, it must match this with a willingness to consider a wider category of remedies (as the European Commission has done).
We have reached a tipping point in the CMA’s approach, which, emboldened by Lear and Furman and a generalised concern with Big Tech, has shifted the presumption of harm against digital mergers from the get-go. So preoccupied with whether they could, the critical question of whether they should has fallen by the wayside.
Beyond the potential illegality of the decision in this case, degrading the standard of review in Big Tech cases -however well-intentioned- will inevitably lead to a weakening of approach in all cases. Where before evidence and careful weighing were centre stage if Microsoft/Activision is endorsed, the CMA may be tempted in the future to simply ‘predict’ a future ‘potential’ competition concern, and block mergers without the need for pesky data, documentation, or analysis. We as merger control lawyers should be very concerned with such a world.
In the final pages of its report, Lear recognised that pursuing a more radical approach would risk “falling short of the burden of proof […] required to satisfy to block a merger, thereby having their decisions successfully challenged in Court”.
Ultimately, the Tribunal now has a critical opportunity in Microsoft V Competition And Markets Authority to make clear that the CMA has indeed fallen short, correct the course and reset the focus on what is provable and probable, not possible. While Big Tech is a reason to refine merger control rules, it should never be a reason to disapply them altogether.
This should be an enduring lesson of Microsoft/Activision.
* Despite the author’s prior role as a Deputy Director of Mergers in the Competition and Markets Authority, the opinions of this post remain solely his own.