Series
Blogs
Up nextFormer executive Vice-President Vestager, 27 June 2024: “The main problems are reduced contestability, dynamism and innovation. These problems arise across sectors and markets, digital or traditional. So we are called upon to develop novel theories of harm that take into account these market dynamics in a wide range of sectors.”
In a probe into the EU's merger control landscape, Rhino has unearthed a shift in the European Commission's approach to assessing mergers. Over the past decade, our analysis of non-simplified procedure decisions reveals a transformation in the predominant theory of harm, moving from predominantly focusing on horizontal issues to other types of concerns (Figure 1).
In 2014-2016, the EC predominantly applied a horizontal theory of harm in approximately 45 decisions each year. Fast forward to 2020-2023, this number has almost halved to approximately 25 per year. Looking at the same data in percentages, the EC predominantly applied a horizontal theory of harm in almost 50% of cases from 2015 to 2017, and in less than 40% of cases from 2020 to 2023 (Figure 2).
Step 1 - We determine the intensity of each of the horizontal, vertical and conglomerate assessment in each merger decision. This intensity rating, which ranges from very low to very high, is based on text recognition software applied to a database of EU merger control decisions. The software attributes a numerical value to each decision based on key words that are representative for a horizontal, vertical, or conglomerate assessment. Whether a decision’s intensity is qualified as high or low, depends on how the value compares to the average values for the same type of decision in the dataset. For example, a phase I remedies decision’s horizontal intensity assessment will be determined in relation to the average values for all phase I remedies cases in the dataset.
Step 2 - Cases are categorized based on the most intense scrutiny they have undergone. For instance, a decision with a 'very high' horizontal assessment intensity with 'average' vertical and 'low' conglomerate assessments is labelled 'Horizontal'. But if all assessment types share equal intensity, the decision is tagged as 'Horizontal / Vertical / Conglomerate'. This approach helps us understand the depth of the EC’s concerns in different areas.
What is going on?
The data doesn’t show why there is increased scrutiny of vertical and conglomerate aspects of cases in the EC’s decisions. However, the data suggests a trend that the EC's concerns are evolving, or perhaps the nature of mergers themselves are changing, with companies possibly steering clear of transactions that might raise the more obvious/traditional horizontal issues and focus more on their product pipeline or on their portfolio growth.
As we have described before, part of the explanation is that there is an increased focus on the complicated cases. In addition, there is more willingness to explore theories of harm beyond unilateral horizontal effects. In the case of ecosystems, the EC will even undertake a ‘multi-directional’ examination of the possible interactions between all products that fit within an ecosystem. The EC has the backing from the General Court for this expansion, which ruled in the Android case that “the relevant markets that make up that ecosystem may overlap or be connected to each other on the basis of their horizontal or vertical complementarity”.
However, the trend is not limited to digital markets (Table 1). Looking at the unconditional approval decisions in 2023 with at least an average vertical assessment and at least an average conglomerate assessment, only one (M.10855) can be considered a digital market transaction, with the others belonging to more traditional markets.
| Case | Parties | Sector | Vertical | Conglomerate |
| M.10988 | BROOKFIELD / CAMECO / WESTINGHOUSE |
Nuclear power plant equipment | Average | High |
| M.10855 | CAPGEMINI / ORANGE / JV | Cloud services | Average | High |
| M.10841 | FIRMENICH INTERNATIONAL / KONINKLIJKE DSM |
Chemicals | High | High |
| M.10821 | YOKOHAMA RUBBER CO / TRELLEBORG WHEEL SYSTEMS HOLDING |
Tyres | Average | High |
| M.10815 | DEUTSCHE TELEKOM / ORANGE / TELEFONICA / VODAFONE / JV |
Digital identification services for targeted advertising |
Very high | Very high |
| M.11177 | PFIZER / SEAGEN | Therapies to treat cancer | Average | High |
In the United Kingdom, the CMA has been keeping an eye on non-horizontal theories of harm for some time, prohibiting its first vertical merger in 2016, in the case of ICE/Trayport, which involved the acquisition by ICE (the largest operator of exchanges and clearing houses in the trading of wholesale European utilities) of the Trayport software which was necessary for competing trading platforms to effectively offer their services.
It officially amended its substantive guidance (the so-called Merger Assessment Guidelines) in 2021 to remove the reference that ‘it is a well-established principle that most [non-horizontal mergers] are benign and do not raise competition concerns:’ (Merger Assessment Guidelines, OFT124; paragraph 5:6:1), reflecting growing concerns that non-horizontal theories of harm could result in the foreclosure of current or future rivals.
In the years that followed ICE/Trayport, the CMA’s merger enforcement was heavily influenced by a desire – shared at the time by many other authorities globally – to remedy perceived historical underenforcement, especially in digital markets (as highlighted in the Lear and Furman reports). Perhaps unsurprisingly therefore, many of the non-horizontal mergers giving rise to detailed investigations and/or remedies relate to this sector – including Facebook/Meta, Microsoft/Activision and LSE/Quantile. In particular, similarly to the EC, the CMA has been canvassing ‘ecosystem’-type theories of harm in cases such as Meta/Giphy and Microsoft/Activision.
Notwithstanding the above, conducting a review of the statistics on mergers involving non-horizontal theories of harm investigated by the CMA, it is not in our view possible to discern any major trends regarding the cases considering non-horizontal theories of harm. While the number of cases involving non-horizontal theories of harm has fluctuated year on year, and there have been a number of high profile non-horizontal cases throughout this period, there is no clear upward trend in the last six or seven years.
| CMA Phase 2 Non-Horizontal cases 1 January 2022 - 15 February 2025 |
Theory of Harm | |
| 2025 | Boparan / ForFarmers | Close rivals; vertical |
| Lindab / HAS-Vent | Close rivals; vertical | |
| 2024 | AlphaTheta / Serato | Close rivals |
| Adobe / Figma | Potential competition | |
| UnitedHealth Group / EMIS | Close rivals; conglomerate | |
| Broadcom / VMware | Vertical foreclosure | |
| 2023 | Microsoft / Activision | Vertical foreclosure |
| London Stock Exchange / Quantile | Vertical foreclosure | |
| Sony Music Entertainment / AWAL | Close rivals / potential | |
What the data does indicate is that non-horizontal theories of harm represent a significant portion of cases giving rise to remedies or prohibitions compared to horizontal mergers. In our view, this is consistent with the CMA’s concerns regarding foreclosure risks in non-horizontal theories of harm and the growing number of complex/ecosystem-type mergers. This is also doubtless driven by the fact the CMA’s historical opposition to behavioural remedies, which would typically be the “fix” in a vertical or conglomerate merger.
Our data, however, is inherently backwards looking. Recent developments in the UK, crystallised most clearly in last week’s Draft Strategic Steer to the CMA from the UK Government and the CMA’s response suggest that going forwards, the CMA’s appetite for pursuing more novel or speculative theories of harm may be at least slightly curtailed.
Our data, however, is inherently backwards looking. Recent developments in the UK, crystallised most clearly in last week’s Draft Strategic Steer to the CMA from the UK Government and the CMA’s response suggest that going forwards, the CMA’s appetite for pursuing more novel or speculative theories of harm may be at least slightly curtailed.