Microsoft / Activision - The Counterfactual (and probabilities)
There will be many challenging questions for the UK's Competition Appeals Tribunal to answer in the event that it is called to opine on the CMA's decision to prohibit Microsoft's acquisition of Activision.
One that I wanted to explore in this post is in relation to the counterfactual underpinning the CMA’s finding of an SLC in cloud gaming.?
While every counterfactual is to some extent forward-looking, this is particularly true here, since Activision games were not widely available to cloud gamers at the time of the CMA’s decision. The CMA therefore had to take a view on a number of factors, including (i) when and to what extent Activision games would be available, (ii) whether they would be available to some or all providers, (iii) how important those games would be compared to other games available to cloud gamers, (iv) which business models would proliferate in the cloud gaming sector and (v) how other gaming providers and publishers might respond should Activision games be made available only on Microsoft services.
Putting the SLC analysis to one side, one reasonable question to pose at the outset is the appropriate probability threshold that should apply to a counterfactual comprised of multiple future events occurring.
In principle, the CMA (and therefore the CAT) applies a simple 'balance of probabilities' test (ie the CMA must provide sufficient evidence and analysis to show that the counterfactual is more than 50% likely to occur). Certain mergers, such as those involving horizontal competitors in concentrated markets, cross the threshold relatively easily (since the CMA can rely on a settled counterfactual, usually involving existing high combined shares of supply, limited known competitors or prospects for entry, internal documents pointing to existing competitive interaction, and so on).
However, in a scenario such as Microsoft / Activision, where the counterfactual and SLC finding relate to a merger occurring in a nascent market with an unclear counterfactual, the probability threshold gets more interesting (and, arguably, much harder for the CMA).
This is particularly true when the counterfactual the CMA relies upon is made up of a series of assumptions, namely that, absent the merger:
On the above basis, the CMA found that Microsoft, were it to withhold Activision games from its rivals (which the CMA found it would have the incentive to do) could hinder those rivals from scaling quickly enough to the point that competition is harmed.
What I find intriguing is the probability threshold that should apply to each of the above ‘findings’ as regards the counterfactual, all of which the CMA would acknowledge are, to an extent, speculative.?
For example, the timing of release, and precise scope, of Activision content that would be available is uncertain, as is the total number of providers that would have access to some or all content. Further, the assertion that there would not be ‘sufficiently good’ (in the CMA's words) alternatives to Activision content is necessarily speculative, since it is opining on the state of the market in the next 3-5 years, and all based on historic data in relation, predominantly, to console and PC games, not cloud gaming. There is also significant uncertainty as to the particular business model that cloud gaming providers will use (eg BYOG, B2P, etc.) and this is acknowledged by the CMA. Finally, assumptions regarding the importance of network effects, while the market remains nascent, cannot be made with certainty.
Why is this important??Well, in a case where the counterfactual relies on the probability of one event occurring, that event must have an even or greater likelihood of occurring for the legal probability threshold to be met.?However, each additional independent event that must occur impacts the probability of that counterfactual occurring.?To find the probability of two independent events occurring, you multiply the probability of the first event by the probability of the second event). If two events each have an even chance of occurring, the probability of both occurring is therefore only 25%.? If three events each have an even chance of occurring, the probability of all three occurring falls to 12.5%, and so on.? In fact, if a counterfactual is based on three independent events occurring, each must have a probability of 80% or greater of occurring for there to be a better than even chance of all three occurring.
Which is a long-winded way of saying that a simple ‘more likely than not’ approach is arguably not appropriate in a case where the CMA has established a counterfactual that relies on a series of events occurring in the future. Instead, the CAT would be right to closely scrutinise the evidential and analytical basis of each finding in its own right, in determining whether the outcome the CMA proposes is 'the most likely'.
Tofa? A?k
1 年Kap?daki Dü?man https://g.co/kgs/HjXseL Can you make the avatar of the enemy movie Vasili zaytsev at the door, call of Duty Mobile, and make sure that the Tarkan and Kurt character from the Turks will play the game more people Ronan Scanlan Call of Duty Endowment Activision
Economics Director at Ofcom
1 年It is for this reason that the Furman review called for “a more economic approach to assessing mergers [which] would be to weigh up both the likelihood and the magnitude of the impact of the merger. This would mean mergers being blocked when they are expected to do more harm than good. The Panel called this a ‘balance of harms’ approach” [Para 3.88] “The balance of harms test would have similarities with the government’s recognised approach for making regulatory decisions, which draws on the principles of cost-benefit analysis. This can combine qualitative and quantitative analysis and judgements, with various techniques for addressing the challenges of uncertainty. This approach is frequently used for significant and complex government decisions, for example for public health proposals, environmental protection, or major infrastructure investment” [footnote 18]. The CMA did not seek to change the test, but it appears that they took up the spirit of the recommendation of the Furman review.?
Economics Director at Ofcom
1 年You will know that the CMA does not seek to assign probabilities to different counterfactuals. Rather, it seeks to take a view in the round. I don’t expect the CAT to re-open whether the CMA assigned the correct probability of the conditions you identified, or whether this probability was greater than 50%. The CMA’s ability to form a judgment on this seems unlikely to be illegal or irrational or to have involved procedural impropriety. Nonetheless, you have a point about how one should deal with competing counterfactuals, all of which are uncertain. The CMA often seeks to focus on the competitive effects analysis, rather than agonising over the counterfactual, but this does not really change the nature of the challenge.
Economics Director at Ofcom
1 年I think that the CMA has been pretty clear that uncertainty, even a lot of it, cannot be used as a ‘get out of jail free card’. Whereas earlier OFT/CC/CMA decisions may have given the parties the benefit of the doubt, the CMA does not intend to do this going forwards. It is taking this approach in light of the evidence, it perceives, of underenforcement, which includes the lack of review/challenge to huge numbers of tech mergers; the KPMG report evaluating entry and expansion in UK mergers; the E.CA report on ex-post evaluation of vertical mergers; and the Lear ex-post assessment of merger decisions in digital markets. Your analysis suggests that it is only under a number of cumulative speculative conditions in which an SLC may be found and the stand-alone probabilities of each would need to be multiplied. This assumes that each condition is entirely independent of the others, which may not actually be the case. It may also be that the CMA considered that some of these conditions were highly likely, even if others had a lower likelihood of occurring. You are also implying that all other outcomes, other than the conditions you outline, lead to an absence of an SLC, which may not be the case.
Chair, Antitrust Practice Preiskel& Co LLP at Preiskel & Co LLP
1 年Isn’t there a bigger problem about Cloud gaming needing to operate at low latency for gamers to be able to play the game without the interactions being delayed ? In a country which is now below 10% cover age and well below the OECD average for Fibre to the Home, are the basic communications connections good enough for cloud gaming to be a significant business in the U.K. countries like Sweden and Spain have over 60% FTTH connections and the there is likely to be a cloud gaming business industry in many other countries but is it likely, on a balance of probabilities, that the U.K. position will change enough over the next 5 years for Activision Blizzard to be a realistic competitor in a U.K. counter factual ?