There has been a lot of speculation and debate recently about the possible underenforcement of mergers in dynamic markets. Certain past cases such as Facebook/Instagram and Google/Doubleclick are often cited as examples where unease about the true transaction rationale and market impact has grown since their respective merger clearances. At the same time, we have recently conducted a high number of in-depth Phase 2 investigations, taking strong action to block or break up certain mergers, with numerous mergers being abandoned after we have raised competition concerns.
I’d like to begin my address against this backdrop by discussing whether the CMA is becoming increasingly interventionist in merger control generally (and dynamic markets in particular). I will comment on the extent to which we are investigating mergers differently now. I also want to touch on whether any evolution in approach requires substantive or jurisdictional updates to legislation or our guidance.
1. Is the CMA becoming increasingly interventionist in merger control generally (and in dynamic markets in particular)?
It is important to start with some historical context to UK merger control. For decades, the UK reviewed mergers against a broad public interest test which required authorities to take into account ‘all matters which appear to them in the particular circumstances to be relevant’. This made it difficult for firms to predict merger control outcomes, or even the grounds on which decisions would be taken, particularly as the Director General of Fair Trading (DGFT) and the Monopolies and Merger Commission (MMC) played an advisory role and Ministers made the final decision. This changed with the introduction of the Enterprise Act in 2002, which placed decision-making in the hands of independent authorities, with a consumer-focused test, and decisions based on rigorous economic analysis. This has been positive for investment and innovation in the UK economy. The CMA’s merger control work is estimated to have delivered direct benefits to consumers of, on average, £183 million annually 2016-2019 (CMA110: CMA Impact Assessment Report 2018/19). This figure does not of course take into account the wider benefits to investment that come from the existence of the independent regime.
I believe there are four factors relevant to considering whether the CMA is now more interventionist: two relate to merger control generally, and two are particularly relevant to dynamic markets.
A. Horizontal mergers in concentrated markets
First, the number of mergers we block in a given year inevitably depends on the mergers that are announced that year, and recently we have been seeing horizontal mergers in markets that are already very concentrated (including in markets where we already suspect that there might be poor market outcomes) that have ultimately required intervention.
As Valletti and Zenger have warned, mergers are more likely to raise competition concerns where merger parties’ pricing power is already large to begin with, and in a current environment of high and increasing margins, we should be vigilant when assessing mergers in concentrated markets (see footnote 1). In concentrated markets, each player – however small – may represent an important competitive constraint that helps prevent price rises and promote quality, service and innovation. Removing such a competitor, especially where limited constraints remain, is likely to be problematic.
We saw this in, for example, Sainsbury’s/Asda, Tobii/Smartbox and Ecolab/Holchem. After in-depth investigations, the CMA’s independent Phase 2 inquiry groups decided that the only effective way to address competition concerns would be to block (or reverse when already completed) these mergers. Similarly, Illumina abandoned its acquisition of PacBio after serious competition concerns had been highlighted by the CMA’s Phase 2 inquiry group.
There is also obviously an increased risk of collusion (or at least weak competition) in highly-concentrated markets.
B. Richer evidence sources
Second, we are now using a wider and more sophisticated range of evidence as part of our analysis, partly driven by the increasingly dynamic nature of the markets that are the subject of our merger reviews. I’d like to take this opportunity to run through some of the key evidence sources we tap into.
i. Internal documents
Internal documents will often set out the merger parties’ commercial strategies and provide insights into how the markets (and the merger parties’ products) may develop in future. They are particularly helpful in cases involving dynamic markets. We are especially interested in discussions of future competitive threats and how the merger parties intend to respond to them.
We sometimes see differences between the picture portrayed to investors and Boards in certain documents, and that submitted to competition authorities. Thorough investigation of these documents is therefore needed to understand these differences, and the background to these documents and the purpose for which they were produced form part of this investigation.
When reviewing the merger parties’ internal documents, the evidential weight placed on such a document may depend on its context. We will consider the author of the document, the intended audience, the timeframe and the purpose for which it was created (See for example, PayPaliZettle and Illumina/PacBio). An understanding of the merger parties’ decision-making processes and hierarchy can also be helpful background. This is also consistent with how we seek to understand the context of internal documents in Competition Act cases.
Internal documents from third parties (most commonly competitors and potential competitors) can also be useful to forward-looking assessments. Such information requests will typically be targeted at the third parties’ commercial strategies and future plans. This gives us a better understanding of the likely competitive constraints on the merger parties in future. Context will again be important to determine how much weight to place on third-party internal documents.
ii. Deal valuation materials
Another tool at the CMA’s disposal is to review the documents discussing the target’s valuation. By this, I mean scrutinising the price of the company acquired and the models used by the acquirer to value the target.
This analysis can provide helpful insight into the acquirer’s expectations for the future success of the target, as well as whether the valuation includes a premium to account for the potential earnings growth of the target, including in the acquirer’s hands. The price ultimately paid by the acquirer may not match the acquirer’s valuation of the target, and may depend on the acquirer’s perception of the level of competition to buy the target.
We recently analysed deal valuations in PayPal/iZettle and Illumina/PacBio. We looked at estimated synergies, taken from internal documents and deal models, as well as equity analysts’ comments at the time of the transactions and calls with advisors to better understand the valuation and sales process:
In PayPal/iZettle, PayPal’s valuation of the iZettle business was much higher than a proposed IPO valuation. Our work on the valuation and estimated synergies ultimately led us to conclude that the price PayPal paid for iZettle was justified by commercial valuation and synergies rather than reflecting an anti-competitive premium.
In Illumina/PacBio, our work on the valuation model was helpful to understand the high purchase price, the forecasted areas of future growth, as well as in determining projected future levels of R&D in the years after the merger and comparing these to what they would have been absent the merger.
Both cases show that deal valuation models and board presentations can be strong sources of evidence for the merger parties’ motivations and the deal rationale. This is particularly useful for dynamic mergers where the future is less certain, and therefore the merger parties’ plans for the merger are most illuminating.
iii. Other third-party evidence sources
In addition to internal documents and deal valuation materials, we may also look at other third-party evidence when conducting a forward-looking assessment. For example, third-party forecasts and analysts’ reports.
We are also increasingly considering whether to make use of our investigatory powers to hold witness interviews, similar to those used regularly in the US. For example, we recently interviewed Amazon’s senior management as part of our Phase 1 investigation in Amazon/Deliveroo.
In addition to allocating more resources to consider certain theories of harm in a given merger case, we are building a better understanding of dynamic markets through other functions and internal initiatives. For example, knowledge gained through our current market study on online platforms and digital advertising will no doubt prove useful in future mergers relating to the digital advertising ecosystem, and our DaTA team are increasingly helping us to get to grips with how firms use data and algorithms in their business models, and how we can use data and documents more effectively in our day-to-day work.
iv. Limitations of historical and static evidence in dynamic markets
Typically, we consider ‘static’ or recent evidence such as market shares, win/loss data and switching data as useful for many merger assessments as it is informative of competition in the near future.
But in dynamic markets, this is not necessarily the case. As firms in dynamic markets are typically continuously evolving and releasing new products or services, historical or static data may not accurately reflect the changing market position. For example, market shares may not accurately reflect the market power of a recent entrant or of a player who has recently released a disruptive new product. Indeed, merger parties sometimes encourage us to take a dynamic perspective when assessing the potential competitive constraint from other actual or potential competitors, or the impact of market developments that reduce the merger parties’ historic strength.
We expect these first two factors to lead to more competition concerns being identified and more merger interventions in general. Beyond that, there are two factors which are particularly relevant to dynamic markets.
C. The need to consider uncertain future market outcomes in dynamic markets
In dynamic markets where products, processes and/or business models evolve rapidly, we must consider uncertain future market outcomes. In doing so, we should learn from previous mergers that have contributed to poor outcomes for consumers – in terms of higher prices as well as reduced innovation, quality or service – and where wrong decisions may have been taken. Our approach is, and has been, evolving to help us make forward-looking assessments.
We need to make decisions under greater levels of uncertainty than before. Some say that the correct response to this increased uncertainty about future competitive outcomes is to not intervene, however, it would be incorrect in terms of standard economic analysis. We still have a balance of probabilities test and this means that we should still intervene if we think, based on the available evidence, that the merger is more likely than not going to result in a substantial lessening of competition (SLC). This test remains applicable however uncertain the future is.
Our evolving approach should, as always, learn from previous mergers and consider what could have been done differently. Market participants in several sectors often accept that various historic mergers have contributed to poor market outcomes (see footnote 1). Facebook/Instagram, Facebook/WhatsApp and Google/DoubleClick, PriceWaterhouse/Coopers & Lybrand and Boeing/McDonnell Douglas are seen by some as examples of merger control gone wrong. We now know that:
Since Facebook bought Instagram, Instagram has become a thriving, monetised social network. As Lear found in its review of digital merger decisions, the merger gave Facebook a competitive advantage, arguably allowing it to offset declines in its original platform;
Facebook gave misleading information in the Facebook/WhatsApp merger review about whether it could match Facebook and WhatsApp user accounts;
After buying Doubleclick, Google now has a particularly strong position in advertising intermediation (with an estimated share of supply of over 90% as a publisher ad server in the UK), and there are legitimate concerns about potential conflicts of interest. (CMA’s Interim report in the online platforms and digital advertising market study);
Following various mergers in the 1990s, the statutory audit market is now made up of the Big 4 firms only, making it highly-concentrated with a lack of choice and resilience (Final Report: Statutory Audit Services Market Study); and
Instant messages and internal Boeing documents recently released to congressional investigators show employees discussing efforts to manipulate US and international safety regulators, revealing the lengths that some companies will go to evade scrutiny from regulators and giving a sense of a market that is not working well.
We need to make the most of our merger control tools when our assessment is necessarily forward-looking and apply a degree of scepticism and robustly test submissions by merger parties (and indeed, third parties) about future market developments and transaction rationales.
False negatives (allowing a merger that should have been blocked) can be particularly costly for consumers in the digital sphere. Many digital markets exhibit characteristics that lead to markets tipping, such as two-sidedness, network effects, and large economies of scale.
In markets that are prone to tipping, competition is likely to be for the market rather than in the market. While competition for the market is valuable for consumers as it ought to lead to the best firm winning the market, a merger that allows an incumbent to avoid a new round of competition for the market is likely to have a high cost for consumers, in that they do not benefit from the competitive phase and the ‘wrong’ firm may end up winning the market. In concentrated digital markets, it becomes even more important to protect competition for the market.
We are very mindful of the importance of protecting potential competition and dynamic competition, as reflected in recent reports and commentary. Valletti and Zenger have pointed out that if dominant incumbents spend billions on targets with no earnings because they are considered to pose a risk of future competition, these incumbents clearly take potential competitors seriously and, therefore, so should competition authorities (see footnote 4).
We are advancing our approach through casework. For example:
In PayPal/iZettle, we assessed a dynamic counterfactual in a fast-moving and dynamic payment services industry;
In Experian/Clearscore, we considered whether the merger would lead to a reduction in the rate of innovation (such as product improvements), a non-price theory of harm, which is something we are likely to see increasingly in future due to online services being offered for free (or for zero-price);
In Thermo Fisher/Roper, we explored potential non-price effects of the merger (which was both horizontal and vertical in nature) on quality and future innovation; and
In Illumina/PacBio, we examined whether the merger would reduce levels of innovation in a dynamic and rapidly developing sector, drawing on evidence such as internal documents and valuation models.
The evolution of our approach is also supported by the increasing evidence of poor outcomes for consumers in numerous markets. We are aware of poor market outcomes from our own experience of such cases, from our engagement with consumer organisations (who are particularly well-placed to offer intelligence due to the scale of their interactions with consumers), with Parliament and with market participants both during merger investigations and through external consultations.
We are not alone in believing that our approach to merger assessment, and that of many other competition authorities, needed to evolve. We believe that the carefully researched academic evidence from Philippon, Jason Furman and his colleagues, the Stigler Centre report, Jonathan Baker, Valletti and Zenger, and numerous others, supports the view that there have been poor outcomes for consumers as a result of mergers in a number of dynamic, and digital, markets.
D. Challenges with remedies in dynamic markets
Finally, in dynamic markets, it can be particularly difficult to identify and design effective behavioural remedies thereby making structural remedies or prohibition more likely solutions to competition problems. We have also learnt from the remedies we’ve accepted when allowing certain previous mergers that we need to be more sceptical about whether anything besides structural remedies can comprehensively address the competition problems we identify. These factors may make our decisions appear more ‘interventionist’ than in the past.
Behavioural remedies are, by their very nature, difficult and resource-intensive to implement, monitor and review. These difficulties and costs do not just fall on the CMA – they are felt by merger parties as well as their customers and competitors who may wish to understand and monitor remedies, with consumers bearing the ultimate cost of any ineffective remedies. Given the level of uncertainty inherent in dynamic markets, it can be particularly difficult to foresee how any behavioural remedies offered will work in changing market environments. It is also more difficult to predict the level of monitoring and enforcement that will be required to ensure that the proposed remedy is as effective as possible.
A number of recent cases have highlighted the challenge of accepting behavioural remedies in dynamic markets. Thermo Fisher/Roper, ICE/Trayport and Experian/Clearscore all considered possible behavioural remedies that did not proceed due to concerns they would not be effective in remedying the SLC identified. This was at least in part due to the evolving nature of the markets in question and the difficulties in catering for all possible market developments and eventualities.
Our regular evaluations of the impact and delivery of past remedies consistently find structural remedies to be generally superior to behavioural remedies in terms of their effectiveness, risk profile and durability, supporting the use of behavioural remedies in very limited circumstances only. Our most recent report found that the long-term impacts of three historic hybrid and ‘quasi-structural’ remedies we have recently reviewed remained uncertain with ongoing risks to consumer outcomes.
For these reasons, amongst others, we prefer structural remedies over behavioural ones, and it is often the case that prohibition is the only appropriate remedy in these markets.
Ultimately it is consumers who bear the risk of ineffective remedies. We should therefore not shy away from prohibition if it is the only effective remedy.
Interim measures also play a fundamental role in enabling us to put in place effective remedies in completed mergers that are possible in our voluntary regime. In dynamic markets, where it is harder to predict how the market will develop, and where such developments can happen very quickly, it will be even more vital that the merger parties are held separate and that we preserve our ability to remedy any SLC found.
2. Will this result in any changes to the jurisdictional or substantive tests?
We still believe our merger control regime is largely fit-for-purpose. But we are continuing to think about whether there is a case for legislative changes to jurisdictional and/or substantive tests as we review more mergers and gain more experience of digital and dynamic markets.
Both the Furman report, and the reform programme set out in Lord Tyrie’s letter to BEIS outlined several proposals for how the CMA’s review of mergers could be improved and how its tools could be strengthened to ensure more effective results for consumers in the UK. Brexit and the increased workload for the CMA also plays a role in this discussion. We will need to review more multi-jurisdictional mergers, and as has been mooted, a mandatory, suspensory regime for these types of mergers (while retaining the numerous benefits of a voluntary regime) would likely help ensure that merger parties engage proactively with the CMA and that we are best able to reach key decisions on substance and remedies in parallel with other jurisdictions. We are continuing to evaluate how such reforms might be implemented in practice.
As mentioned in our Online Platforms and Digital Advertising Market Study Interim Report, we have also been exploring whether there is a case for introducing a parallel regime for acquisitions by companies designated as having ‘strategic market status’ (SMS companies), building upon the regulatory framework envisaged by the Furman report. We are considering whether such a regime is justified based on concerns about potential historic underenforcement and our understanding – through expert reports and the CMA’s case and policy work – of particular features of digital markets that increase the risks of consumer harm arising from acquisitions by particularly powerful companies. SMS companies may use acquisitions to eliminate potential competition or to leverage their position into adjacent markets in an anti-competitive manner. Small, nascent or potential competitors to particularly powerful companies (in both their own and adjacent markets) could be important sources of competition worth protecting.
We would need to think carefully about how to design and apply an SMS-specific mergers regime to achieve the right balance between (i) ensuring proper scrutiny of transactions that could give rise to harm and (ii) avoiding undue uncertainty and burdens on businesses and the CMA as well as other unintended consequences. This would inform the approach to the regime’s jurisdictional aspects (i.e. what transactions by SMS companies should trigger review) and substantive aspects (i.e. whether the existing substantive SLC test, or the burden and/or standard of proof remain appropriate for such transactions).
3. Review of the Merger Assessment Guidelines
Given the evolution of our thinking in relation to dynamic mergers that I have outlined today, and to reflect a changing economy that relies ever more heavily on digital markets and use of the internet, it is also appropriate that we update our Merger Assessment Guidelines to reflect some of our learnings from over the past 10 years.
Towards the end of last year, we had a fruitful ‘call