Regulatory Comments

ICLE Comments on the EU Draft Merger Guidelines

Introduction and Overview

The International Center for Law & Economics (ICLE) is a nonprofit, nonpartisan research centre that promotes the use of law & economics methodologies to inform public-policy debates. We welcome the opportunity to respond to the Commission’s consultation on its draft Guidelines on the assessment of mergers under the EU Merger Regulation (the ‘Draft Guidelines’), which are intended to supersede the 2004 Horizontal Merger Guidelines and the 2008 Non-Horizontal Merger Guidelines. We respond to the consultation part-by-part below.

We share several of the draft’s animating instincts. We welcome its recognition that mergers ‘may enhance competitiveness and growth’ (para. 2), that scale can be procompetitive (paras. 11–18), and that innovation, investment, and dynamic rivalry deserve a central place in merger analysis. The draft also usefully moves toward symmetry between harms and benefits—articulating a ‘theory of benefit’ alongside the theory of harm (para. 25) and insisting that efficiencies be assessed with ‘an equivalent degree of likelihood over time’ as the harms they offset (paras. 297, 341). These are meaningful improvements that respond to long-standing critiques of the EU’s asymmetric ‘double standard’ in the treatment of merger harms and efficiencies.

Our central concern is one of legal certainty. Guidelines exist to make outcomes predictable, and predictable, consistent merger review has long been one of Europe’s chief competitive advantages—a reason capital, talent, and firms choose to locate and scale here. The draft, by contrast, tries to do too much at once: police market power, accelerate innovation, secure supply chains, strengthen European competitiveness, advance sustainability, safeguard resilience and defence readiness, protect media plurality, and align merger control with industrial policy.

As the list of relevant factors grows, the set of transactions whose outcomes firms can predict with confidence shrinks. A framework that is broad in ambition but thin in operational guidance becomes, in practice, a standing invitation to litigate rather than a guide to compliance. If the Commission is serious about promoting procompetitive mergers and advancing the Draghi agenda on European competitiveness (Draghi 2024), the Guidelines need fewer ambitions and clearer rules.

This tension is sharpest in the draft’s expansion of forward-looking, qualitative theories of harm, including loss of innovation competition, loss of investment and expansion competition, entrenchment of dominance, dynamic foreclosure, and portfolio effects. Many of these theories turn on contestable predictions about distant and uncertain future market conditions. The economics is genuinely unsettled: the relationship between market structure and innovation is non-monotonic and context-dependent (Gilbert 2006; Cohen and Levin 1989); vertical and conglomerate integration is on average neutral-to-procompetitive (Lafontaine and Slade 2007; Cooper et al. 2005); and the empirical record on ‘killer acquisitions’ and ‘data moats’ is thin and industry-specific (Cunningham, Ederer, and Ma 2021; Gautier and Lamesch 2022; Manne and Auer 2024).

Where the evidence is uncertain, error-cost analysis counsels caution. Presumptions and burden shifts that are administratively convenient but economically noisy risk condemning efficient deals and chilling the very investment and innovation the draft seeks to promote (Manne 2020).

In the comments that follow, we press three recurring themes. First, structural indicators should remain soft screens, not presumptions or burden-shifting devices. Second, the draft’s welcome commitment to symmetry must be carried through in practice: the evidentiary bar for speculative long-term harms should be no lower than the bar the parties must clear for the efficiencies that offset them. Third, the proliferation of non-competition objectives—resilience, sustainability, media plurality, democracy, and security—should be handled with restraint and, where they fall outside the European Union Merger Regulation’s (EUMR) legal mandate, kept out of the substantive assessment altogether. With those caveats, the draft can be sharpened into a document that delivers the certainty its own paragraph 5 promises.

Part I — Introduction and Guiding Principles (paragraphs 1–6)

The introductory paragraphs of the Draft Guidelines set the tone for the document and, in important respects, improve on their predecessors. Paragraph 2’s acknowledgment that mergers ‘may enhance competitiveness and growth’ by contributing to innovation, enabling entry, generating scale and scope economies, combining complementary capabilities, and improving allocative efficiency rightly recognises that the great majority of mergers are benign or procompetitive. That recognition is welcome in today’s competition-policy environment, where calls for more aggressive enforcement are common.

The Guidelines would be stronger still if they stated that most mergers do enhance competitiveness and growth. That would align with the Commission’s own practice of clearing most transactions unconditionally and swiftly, often under the simplified procedure. In any event, we welcome the recognition that consolidation is not presumptively suspect.

We caution, however, against the more programmatic language elsewhere in the introduction. Paragraph 5 states that the Guidelines ‘aim to increase legal certainty and predictability…thereby facilitating business and investment decisions within the internal market’. That is the right objective. But the draft’s expanded analytical canvas—discussed throughout these comments—works against that objective unless each new concept is accompanied by clear, administrable limits. The Commission should treat paragraph 5 as a binding design constraint: where a proposed concept cannot be operationalised in a way firms can predict, it should be narrowed or omitted.

Paragraph 6’s statement that the Guidelines bind the Commission, but that it ‘may depart from these Guidelines if necessary’, faithfully reflects the case law (e.g., CK Telecoms). But self-binding guidance has value only insofar as it promotes consistency. The more the text invites case-by-case, discretion-laden judgments—particularly through the ‘margin of discretion’ the draft repeatedly reserves to itself (e.g., paras. 20, 27, 300, 342)—the less paragraph 6 will constrain outcomes in practice. The Commission should be candid that discretion and predictability are in tension and should resolve that tension, wherever possible, in favour of greater clarity for parties.

Paragraph 3’s decision to consolidate horizontal and non-horizontal guidance into a single instrument is defensible. Modern markets, especially digital and innovation-driven markets, are increasingly hard to pigeonhole as purely vertical or horizontal. But consolidation should not blur the analytically important distinction between the two.

Horizontal mergers automatically eliminate a competitor. Harm from a non-horizontal merger, by contrast, depends on a subsequent, non-automatic, and often unprofitable strategic choice to foreclose, while the efficiencies of vertical integration—notably the elimination of double marginalisation—are frequently automatic (Cooper et al. 2005; Lafontaine and Slade 2007). A single document should preserve, not collapse, the distinction between these merger types in light of their different probabilities and mechanisms of harm.

The Commission should also keep the consumer-welfare standard, expressed through the ‘significant impediment to effective competition’ test (SIEC test), as the lodestar of the entire instrument. The introduction gestures toward a wider set of goals, but the legal question the Guidelines exist to operationalise is singular: whether a merger would significantly impede effective competition to the ultimate detriment of consumers. Anchoring every subsequent concept to that question gives the Guidelines their disciplining force.

An error-cost perspective reinforces the point. Merger review is a screening exercise conducted under uncertainty. Both false positives—deterring or blocking procompetitive deals—and false negatives—clearing harmful ones—are costly. But because the overwhelming majority of mergers are benign or procompetitive (para. 2), and because an unpredictable regime chills benign and harmful transactions alike, a framework that multiplies speculative theories of harm without commensurate limiting principles will, in expectation, destroy more value than it preserves (Manne 2020). The introduction should acknowledge this asymmetry and resolve doubt in favour of clear, administrable rules.

Relatedly, the Guidelines should recognise that acquisitions are a normal and largely healthy feature of a well-functioning market for corporate control. The ability to sell a firm—including to a larger incumbent—disciplines management, rewards founders and early investors who finance risky ventures, and reallocates assets to those who can deploy them most productively (Manne, Bowman, and Auer 2022). For start-ups in particular, acquisition is a primary route to exit and a central incentive to innovate and attract venture finance.

A.      Part I.A — The Role of EU Merger Control (paragraphs 7–18)

Second, Competitiveness and resilience are outcomes, not free-standing objectives

Paragraphs 7–10 reframe merger control as contributing to growth, investment, innovation, competitiveness, and resilience. We agree these are products of well-functioning markets, and consistent with the purpose of the EU Merger Regulation (EUMR). Productivity and competitiveness are mutually reinforcing (Syverson 2011).

Unlike productivity, competitiveness is not embedded in the EUMR and can easily be misread as protecting European-owned firms from foreign rivals, including by impeding their acquisition by, or partnership with, foreign undertakings. Yet foreign capital and know-how can drive growth.

Paragraph 8 rightly treats competitiveness and resilience as consequences of enforcement grounded in the significant impediment to effective competition (SIEC) test. But the Guidelines should describe them as byproducts of the EUMR’s single mandate, not as objectives to maximise. This is not merely semantic. Agencies charged with multiple, fuzzy, unranked missions face weaker incentives and weaker accountability. By contrast, a single welfare lodestar gives competition law coherence and predictability (Dewatripont et al. 1999; Wright and Ginsburg 2013).

We recommend that the Commission state clearly that its mandate is singular: to prevent a SIEC. Competitiveness and resilience are welcome consequences of discharging that mandate well. They are not a licence for protectionism or independent grounds to clear or block a merger.

Paragraph 9 frames resilience— including supply-chain security, critical infrastructure, defence readiness—as something merger control ‘strengthens’. Paragraph 10 likewise calls for ‘adequate weight’ to be given to scale, innovation, investment, and resilience as procompetitive factors. We support the impulse, but resilience is novel and underdefined.

Two problems arise. First, resilience cuts both ways. It may justify clearing a consolidating merger that builds redundant capacity. But the draft also treats increased import reliance as a resilience harm (para. 92), risking protectionism through the back door. Second, resilience often requires excess capacity, which sits in tension with static efficiency. The Guidelines offer no method to weigh the two.

Benefits from scale versus market power

The distinction between procompetitive scale and harmful market power (paras. 11, 18) is the correct organising principle. Paragraph 15’s catalogue of scale benefits also maps onto our consultation response. Scale is decisive where fixed costs are high, marginal costs are low, and network effects are strong. In those settings, complementary combinations can expand networks, intangibles, and access to finance without creating market power (Teece et al. 1997).

One phrase should be operationalised or removed. Paragraphs 11–12 and 15(b) commend mergers that reach ‘the necessary size to compete in global markets’. Scale is frequently procompetitive, but ‘necessary size’ is not a workable merger-control concept. The Commission should not define the scale a firm ‘needs’.

The phrase invites two opposite errors: licensing consolidation as a route to national champions, or implying the Commission can identify—and cap—the scale that is ‘enough’. Neither judgment is authorised by the EUMR. Scale is not a harm to be cured; the only basis to impede it is transaction-specific evidence of a SIEC. We recommend the Commission either operationalise ‘necessary size’ with criteria tied to the SIEC assessment or delete the phrase, leaving scale presumptively procompetitive and subject to challenge only where the evidence shows a SIEC.

We welcome paragraphs 16’s recognition that start-up acquisitions are ‘unlikely to give rise to competition concerns’ and paragraph 17 recognition of non-horizontal mergers’ stronger potential for integration. For paragraph 15 to have practical effect, the Commission should clarify how its listed benefits are evidenced and weighed.

Two further clarifications are needed. First, the Commission should keep the SIEC assessment separate from the investment-screening and foreign-subsidies regimes. Concerns about who owns a European firm, or about distortive foreign subsidies, fall under the Foreign Direct Investment Screening Regulation and the Foreign Subsidies Regulation. Importing those concerns into merger control would conflate distinct legal tests and erode predictability. Genuine resilience and security concerns belong in the Article 21(4) legitimate-interests procedure, not in an expansive competitive assessment. We return to this issue in Part III.

Second, the scale benefits listed in paragraph 15 should be cognisable like other efficiencies under the Part II theory-of-benefit framework. The relevant question is often not whether a standalone firm could replicate a capability, but whether it could do so at efficient scale, in time, and with the finance that EU firms often lack (Draghi 2024).

B.       Part I.B — Guiding Principles (paragraphs 19–51)

Paragraph 20 rightly recognises that competition is multidimensional and extends beyond price. But the Draft Guidelines’ open-ended list—especially references to ‘media and cultural diversity’, sustainability, and resilience—risks absorbing objectives outside the EU Merger Regulation’s (EUMR) mandate.

The draft’s reserved ‘margin of discretion’ to weigh incommensurable parameters also risks leaving outcomes to unstructured judgment. We recommend that the Commission confirm that non-price parameters matter only insofar as they are genuine parameters of competition in the relevant market. The Commission should also supply a structured framework—likelihood, magnitude, and timing—for weighing those parameters. We return to that issue in our comments on paragraphs 341-347.

We strongly welcome the symmetrical burdens in paragraphs 21-25. The Commission must ‘articulate and substantiate’ a theory of harm and bears the ultimate burden to show a significant impediment to effective competition (SIEC) (paras. 21, 23). Paragraph 25, in turn, adds a ‘theory of benefit’ that the parties must substantiate.

That symmetry will have integrity only if applied even-handedly. Efficiencies must be shown with a likelihood and timeframe ‘equivalent’ to the harm (paras. 27, 297, 341). That principle must run both ways. If the Commission builds a speculative innovation or entrenchment theory on developments years in the future, parties’ efficiency claims over a comparable horizon must be equally admissible. Otherwise, the ‘innovation paradox’ defeats the draft’s promised symmetry (Gurkaynak 2023).

We urge the Commission to state expressly that the evidentiary and temporal standards for theories of harm and theories of benefit are identical.

The evidence section faithfully restates the case law. We support footnote 61’s recognition that the absence of internal documents concerning a theory of harm does not prove that the theory is absent, and that post-announcement documents have limited exculpatory value. But the same scrutiny should apply to documents the Commission uses to build a theory of harm.

Two cautions are warranted. First, paragraph 29’s allowance for the Commission to rely on information ‘without having to verify in detail the credibility and reliability’ of all of it should be tempered by the ‘more likely than not’ standard. Probative value, not volume, must govern.

Second, paragraph 30’s treatment of third-party views should retain the caveat about commercial incentives. Competitors’ objections are not evidence of consumer harm (para. 21).

We welcome paragraph 32’s single ‘more likely than not’ standard for all mergers, regardless of the theory’s complexity (CK Telecoms). We also welcome the confirmation that a preliminary finding of harm is not a precondition for efficiency claims, and that early engagement—including during prenotification—is encouraged.

The Commission could usefully publish a short best-practices note on the evidence and methodologies it finds persuasive for a theory of benefit.

We broadly endorse the counterfactual section: premerger conditions should remain the usual benchmark (para. 38), with adjustments for sufficiently certain future changes, crises, and alternative mergers (paras. 39-42).

This matters especially for innovation. Where innovation is the primary parameter of competition, a dynamic counterfactual should be the rule. Research-and-development pipelines and technological trajectories cannot be captured by static price models.

Our principal caution is symmetry of certainty. The ‘sufficient degree of certainty’ required for counterfactual adjustments must apply equally when the Commission uses a dynamic counterfactual to build a theory of harm and when parties invoke it defensively. Under Tetra Laval, the more contingent and forward-looking the causal chain, the more cogent the evidence required—in both directions.

The failing-firm and failing-division criteria (paras. 45-51) restate the established three cumulative conditions and are unobjectionable. One refinement is warranted for innovation-intensive sectors: financial distress does not imply competitive irrelevance.

The Commission should assess whether a distressed firm retains the capabilities needed to develop and commercialise its innovation projects. It should also assess whether, absent the merger, those capabilities would be preserved or dissipated through exit or insolvency. This is especially salient in the European Union, where unharmonised insolvency regimes risk causing assets and know-how to be lost rather than reallocated.

The counterfactual should weigh the merger’s effects on market power against the consequences of losing the firm’s assets and innovation potential. That inquiry should not be confined to cases that meet the strict failing-firm criteria. Capability dissipation bears on the counterfactual even where the formal defence is unavailable.

Part II — Competitive Assessment, and Introduction to Market Power (paragraphs 52–59)

Paragraphs 52–54 establish the overall posture of the competitive assessment: a forward-looking, dynamic view of competition that considers not only short-term constraints, but also firms’ ‘capabilities and incentives to compete for future business’. That assessment is conducted, in principle, market by market, while remaining attentive to cross-market links, including complements, networks, shared technologies, bundling, multisided platforms, and ‘ecosystems’. We support the forward-looking orientation and the recognition that some industries are closely linked across markets. Two structural observations follow.

First, paragraph 54’s recognition that links across markets—including non-structural links such as distribution agreements, licensing, alliances, or asset sharing—may be relevant is analytically defensible. But it must be cabined. Taken too far, it would allow the Commission to aggregate the market shares of firms that are connected by contract but remain independent competitors, or to deny that a connected firm is a genuine competitive constraint.

This concern is more than hypothetical. The recent wave of AI partnerships—Microsoft/OpenAI, Amazon/Anthropic, and others—involves precisely such non-structural links. Yet the available regulatory record—including the Federal Trade Commission’s Section 6(b) staff report, the Competition and Markets Authority’s decisions on Microsoft/OpenAI, Microsoft/Mistral, Microsoft/Inflection, and Amazon/Anthropic, and the Commission’s own Competition Policy Brief—has identified only theoretical concerns and no concrete evidence of harm. Several authorities have also expressly declined to find a relevant merger situation (Auer and Zúñiga 2026).

These partnerships frequently enhance competition by supplying start-ups with capital, compute, and distribution while preserving their autonomy. We urge the Commission to make clear that the existence of a non-structural link does not justify share aggregation, or the dismissal of a firm as a competitor, absent concrete evidence that the link removes most of the competitive constraint the firms impose on each other.

Second, the draft’s general approach to market power in paragraphs 55–59 is sound in its core architecture and consistent with both the case law and modern industrial-organisation economics. Paragraph 55 correctly defines market power as the ability to maintain prices above—or quality, choice, capacity, output, investment, innovation, privacy, sustainability, or resilience below—competitive levels for a period of time.

Crucially, paragraph 57 states that market power ‘is assessed using a combination of factors, none of which is individually decisive’, and that structural indicators provide only ‘useful first indicators’. That is exactly right and should be preserved. It reflects decades of evidence that concentration is, at best, diagnostic and, at worst, a misleading predictor of competitive harm (Demsetz 1973; Berry, Gaynor, and Scott Morton 2019; Syverson 2019).

We also welcome paragraph 58’s recognition that, in ‘specific dynamic settings, a static assessment of market power may be less appropriate’, as well as the related willingness to adjust market-share and margin indicators for reasonably certain changes in the competitive environment. This is the correct instinct for fast-moving markets, where apparent dominance can erode quickly.

The key, again, is symmetry and discipline. Dynamic adjustments that may reveal hidden market power (para. 64(b)) must be applied with the same evidentiary rigour as dynamic adjustments that reveal hidden competitive constraints (paras. 86–100 on entry and expansion). The Commission should not adopt a posture in which dynamism is invoked to find power but discounted when it would constrain power.

A final word on the recurring reference to ‘ecosystems’ (paras. 54, 252–259). The term has intuitive appeal but no settled economic definition, and it risks becoming a label that licenses intervention without a concrete theory of harm (Colangelo 2026). Almost every successful firm sits within some web of complements, partners, and adjacent products. Describing that web as an ‘ecosystem’ does not, without more, establish either market power or a mechanism by which a merger would harm competition.

We therefore urge the Commission, wherever it invokes ecosystem or cross-market considerations, to identify the specific markets affected, the specific constraint allegedly removed, and the specific mechanism of harm. In other words, the same analytical discipline the draft rightly applies to single-market analysis should apply with at least equal force when a theory spans several markets. That analysis should also recognise that complex, multipronged theories of harm are, all else equal, less likely to occur.

A.      Part II.A.1 — Structural Indicators of Market Power (paragraphs 60–67)

We support the draft’s treatment of structural indicators as screens rather than presumptions, and we urge the Commission to hold that line. Paragraph 57 rightly frames market shares and the Herfindahl-Hirschman Index (HHI) thresholds in paragraph 65—below 1,000 as unconcentrated and above 2,000 as highly concentrated—as ‘useful first indicators’. Paragraph 64 likewise lists the many circumstances in which shares fail to capture market power. That framing should be retained without qualification.

EU merger control turns on a case-by-case significant impediment to effective competition (SIEC) assessment, in which shares and HHIs are first indications, not dispositive rules. The existing texts provide that thresholds ‘do not give rise to a legal presumption’. This architecture—soft screens, no burden shift, and proof of a SIEC remaining with the Commission on a more-likely-than-not standard, including below dominance—is a strength to preserve, not reverse.

The economic record does not support converting these screens into presumptions or making them stricter. Decades of industrial-organisation research find no stable, policy-reliable relationship between concentration and harm, and treat structural measures as poor standalone predictors of price effects (see, e.g., Demsetz 1973; Berry, Gaynor, and Scott Morton 2019; Syverson 2019). Concentration is frequently a product of efficiency rather than anticompetitive conduct. Effects are highly sensitive to market definition. And the ‘rising markups/growing concentration’ literature animating calls for stricter presumptions is itself contested.

Measured markups are sensitive to the accounting of intangibles and fixed costs. National concentration can rise even as local competition intensifies, making it a poor proxy for local conditions. The direction of causation also remains unresolved (Rinz 2022; Syverson 2019). Turning indicative screens into rebuttable presumptions would invert the EUMR’s architecture, increase Type I errors, and chill procompetitive deals—all to obtain an administrative convenience the economics do not justify.

We particularly welcome paragraph 64’s catalogue of the limits of share-based analysis. Shares may fail to capture the intensity of competition, as homogeneous products may compete vigorously even in concentrated markets. They may understate the constraint from an ‘important competitive force’ or a firm with ‘dynamic competitive potential’. Most importantly, ‘in nascent, fast-growing or short-innovation-cycle markets, market shares may provide a less reliable indicator’.

This last point is essential. Digital-market definitions often miss that users—and the advertisers seeking them—switch across a wide variety of services, not only functionally similar ones, often instantly and at no cost. The differentiation that forms the core of a challenger’s threat is too often treated as evidence that the firms do not compete. The January 2025 TikTok outage, after which usage migrated sharply to Instagram, YouTube, Facebook, and even Messenger, shows how broad and immediate cross-service substitution can be. In dynamic markets, share-based screens combined with narrow market definitions tend to overstate the market power of both merging parties and their rivals.

Two clarifications would improve predictability. First, the Commission should state expressly that shares and HHIs are screens, and that any SIEC theory—especially an ‘important competitive force’ theory or differentiated-products unilateral-effects theory—must rest on transaction-specific evidence of diversion, margins, and repositioning, not structural thresholds alone.

Second, the Commission should reaffirm or relax, not tighten, the existing market-share and HHI safe harbours, and should explain how other evidence—diversion ratios, margin or upward-pricing-pressure-type tools, capacity, and pivotality—is weighed once a screen is crossed. Clear, generous safe harbours are among the most valuable certainty-enhancing features a merger-control regime can offer. Because most transactions fall well below any plausible threshold of concern, bright-line screens spare benign deals the cost and delay of in-depth review.

The Commission should confirm that falling within a safe harbour creates a strong practical expectation of clearance, even though the ultimate test remains the SIEC standard. Narrowing the harbours, or qualifying them with open-ended exceptions, would transfer a large volume of benign transactions into uncertainty for no demonstrated benefit. The Commission should also guard against the descriptive bands in paragraph 62—‘low’, ‘moderate’, ‘material’, ‘high’, and ‘very high’—hardening into de facto thresholds.

Part II.A.2 — Other Indicators of Market Power (paragraphs 68–79)

Paragraphs 68–79 bring economically grounded indicators of market power—price sensitivity, margins, and barriers—to the foreground. The draft’s useful contribution is to make these indicators explicit, central, and assessed at the firm level, rather than scattered across market definition, effects, entry, and buyer power. We welcome that reorientation and comment on each indicator below.

Examining customers’ past propensity to switch and rivals’ propensity to expand—through churn, switching, elasticities, and natural experiments—is a sound, effects-based method. But low observed switching is ambiguous. It can reflect genuine lock-in and market power, or simply that customers have found a product that suits them better.

In digital markets, ‘inertia’ is not itself an indicator of harm. Lock-in can also intensify upfront ‘competition for the market’ (Klemperer 1987; Farrell and Klemperer 2007). The net effect of switching costs is therefore ambiguous and should be assessed on the evidence, not presumed.

The Commission should also distinguish switching from multihoming. When customers use several providers in parallel, ‘stickiness’ on one service overstates the constraint the customer faces. The relevant question is whether a customer can place incremental demand elsewhere in response to worse terms, not whether the customer terminates the service altogether. Cheap multihoming can discipline a high-share firm as effectively as switching.

We strongly welcome paragraph 70’s recognition that ‘high margins may be less likely to indicate market power if they are temporary… in markets characterized by a fast pace of innovation’. Schumpeterian rents reward risky investment. They are not necessarily symptoms of harm. Penalising them would blunt the very incentive the draft seeks to protect.

The Commission should apply that insight consistently to the industry-wide margins discussed in paragraph 73, which may reflect rewarded innovation as readily as barriers. Increases in markups do not systematically track the declines in business dynamism that entrenchment would predict (Albrecht and Decker 2026). A high margin is a residual whose source should be identified with evidence, not inferred from its level.

Paragraph 74’s recognition that firms may hold power despite low margins—through penetration pricing or multisided monetisation—is also welcome. In those cases, the relevant question is the net price across all sides, not the headline consumer price (Rochet and Tirole 2006). A low or zero price on one side is therefore not evidence of harm where another side cross-subsidises it. Likewise, a merger that internalises cross-side externalities may lower aggregate prices even while raising one fee.

The draft’s focus on demand-side switching and supply-side entry and expansion is the right one. Our principal caution concerns digital-specific barriers, especially data and network effects, which are too often treated as presumptive.

Data are largely non-rival, frequently replicable, and often obtainable through public sources, licensing, partnerships, or synthetic generation. Their value shows sharply diminishing returns, and freshness often matters more than depth (Manne and Auer 2024). The growth of generative-AI entrants despite incumbents’ vast data troves is hard to reconcile with strong data-entrenchment claims.

Network effects are likewise double-edged. They intensify ‘competition for the market’, can operate in reverse—as with MySpace—and rarely confer perpetual dominance absent exclusionary conduct (Evans and Schmalensee 2016).

Paragraph 79’s strategic-conduct barriers—limit pricing, excess capacity, intellectual-property enforcement, and loyalty rebates—should be applied cautiously. Competition on the merits is not a barrier. The Commission should require evidence that the conduct deters entry.

More broadly, the Commission should distinguish genuine impediments from features that are rewards or byproducts of competition on the merits. Intellectual property, economies of scale, and a strong brand can look like barriers, yet each is often a mechanism by which competition delivers benefits: patents reward innovation, scale lowers costs, and brands economise on search. Before treating a feature as a barrier, the Commission should require concrete evidence that it actually impedes timely and sufficient entry, not infer a barrier from an incumbent’s size or success.

The same caution applies to ‘tipping’ and ‘winner-take-most’ dynamics, which the draft treats as markers of entrenched power. Tipping often reflects the market rewarding a superior product. A tipped position is not durable where users multi-home, interoperability or data portability lowers switching costs, and adjacent innovation can redefine the product. The Commission should ask not whether a market has tipped, but whether the conditions to ‘untip’ it are present.

Part II.A.3 — Dynamic Competitive Potential (paragraphs 80–83)

The introduction of ‘dynamic competitive potential’ (paras. 80–83) is one of the draft’s most significant analytical innovations, and we support its underlying premise. In industries where innovation is an important parameter of competition, a static snapshot of market shares and margins can badly mischaracterise a firm’s competitive strength or weakness. A firm with a thin current share but powerful innovation capabilities may exert far more competitive influence than its static position suggests. Conversely, a firm with high current shares may be competitively fragile if its dynamic potential is exhausted. Capturing this is consistent with the economics of dynamic competition.

The same feature that makes this concept valuable, however, makes it hazardous to legal certainty if left undisciplined. The indicators listed in paragraphs 81–82—numbers and time to market of pipeline products, track record, research-and-development spend and headcount, patent citations, internal innovation targets, access to data or user traffic, ‘dynamic capabilities arising from a given business model’, complementarities, synergies between intangible assets, the ability to exploit network effects across products, and the breadth of an ‘ecosystem’—are so numerous and qualitative that many firms could be characterised as possessing, or lacking, dynamic competitive potential depending on which indicators the Commission emphasises.

Paragraph 81’s suggestion that ‘the high valuation of a target by the purchaser, especially compared to its turnover, may…provide an indication’ of competitive significance is particularly concerning. Acquisition price reflects many things: complementarity, option value, talent, the acquirer’s superior ability to redeploy assets, and ordinary competition in the market for corporate control (Manne, Bowman, and Auer 2022). Treating a high multiple as a marker of competitive significance risks penalising exactly the value-creating combinations the draft elsewhere applauds.

We therefore recommend three disciplines. First, the assessment of dynamic competitive potential must be symmetric. The same factors the Commission uses to attribute hidden strength to a merging party must also be used to credit the dynamic competitive potential of remaining rivals and entrants as countervailing constraints. The draft gestures toward this in paragraph 85; it should make the point explicit and apply it consistently.

Second, dynamic competitive potential, like all forward-looking elements, must be ‘predicted with a sufficient degree of certainty’ and substantiated with concrete, contemporaneous evidence. It should not be inferred from structural proxies such as research-and-development shares, which economic theory shows are unreliable predictors of innovation behaviour (Cohen and Levin 1989; Gilbert 2006).

Third, the Commission should clarify that high acquisition value is, at most, a prompt for further inquiry and never, on its own, evidence of a significant impediment to effective competition (SIEC). With these limits, dynamic competitive potential can sharpen the analysis. Without them, it becomes a licence to find market power wherever static analysis fails to.

Finally, we encourage the Commission to integrate the innovation life cycle into this assessment. Industries pass through stages of technological opportunity, and a firm’s dynamic competitive potential—and the competitive consequences of combining capabilities—depends heavily on where the market sits along that curve (Utterback and Abernathy 1975; Cohen 2010). All else equal, a merger of strong innovators in a maturing market with dwindling opportunity is more likely to dampen rivalry than a merger in a nascent, turbulent space where contestability and entry remain high. Building life-cycle reasoning into the dynamic-competitive-potential analysis would make it both more accurate and more predictable.

We would flag one further risk specific to this concept: double counting across the analysis. The same forward-looking narrative—‘this nascent technology will become competitively pivotal’—can be used simultaneously to inflate the merging parties’ competitive significance, discount the constraint from rivals on the view that their efforts will fail, and construct an innovation- or entrenchment-based theory of harm. Each step compounds the uncertainty of the last.

Consistent with Tetra Laval, the evidentiary burden should rise with the length and predictive uncertainty of the chain of inferences. The Commission should be required to show that the same dynamic assumptions are applied even-handedly to the parties and to their competitors. Where the evidence does not permit a confident prediction about how the technology or market will evolve, dynamic competitive potential should counsel caution about intervention, not support it.

Part II.A.4 — Countervailing Factors (paragraphs 84–110)

The countervailing-factors section—covering entry and expansion, out-of-market constraints, and buyer power—is well constructed. Paragraph 85’s instruction that countervailing factors be assessed ‘in line with…the criteria it applies to the assessment of harm’ reflects exactly the symmetry we have urged. We support the framework and offer refinements to preserve that symmetry in practice.

4.1 Dynamic entry or expansion of competitors (paras. 86–100)

The three cumulative criteria—likelihood, timeliness, and magnitude (para. 93)—are the established and correct test. We welcome paragraph 86’s recognition that entry may come from global firms outside the internal market. We also welcome paragraph 96’s flexibility on timing: a two-year default, but openness to longer horizons where market dynamics, the theory of harm, or entrants’ capabilities warrant.

That flexibility is essential. If the Commission assesses a loss of future competition over a multiyear horizon, it must equally credit entry and expansion that could constrain the merged firm over a comparable horizon. Long horizons for harm, paired with short horizons for offsetting constraints, would be indefensible.

Our main concern is the evidentiary asymmetry latent in paragraph 94, which requires ‘concrete plans to enter or expand at the relevant scale’ for entry to countervail a loss of existing competition, while elsewhere treating a target’s own potential entry as a competitive force on thinner evidence. Countervailing entry should, of course, be evidenced. But the standard should match the standard for treating a firm as a potential competitor whose loss is harmful. We return to this point in Part II.B.5.

Paragraph 100 already acknowledges that the perceived threat of entry can more readily countervail a loss of potential competition than a loss of existing competition. That is a sensible calibration. The Commission should make the parallel explicit: the evidentiary bar for crediting a constraint should mirror the bar for asserting the corresponding harm.

On imports (paras. 91–92), we caution against treating increased import reliance as itself a resilience harm. Imports are first a competitive constraint. Treating them as harm risks smuggling in protectionism and contradicts the draft’s own recognition (paras. 7–10) that openness drives competitiveness. Resilience concerns about import dependence, if entertained, should be confined to genuinely critical inputs and supported by evidence. They should not be used to discount foreign competition generally.

4.2 Out-of-market constraints (paras. 101–103)

We welcome paragraph 101’s recognition that the Commission considers all competitive constraints ‘irrespective of whether they arise from inside or outside the relevant market’. That is especially important given the Commission’s historically narrow market definition in digital markets, which tends to exclude differentiated and adjacent services that in fact constrain the parties.

Paragraph 102’s caveat—that such constraints are often limited because they are not immediate and effective—is an acceptable first approximation. But it should not justify disregarding real cross-market and cross-platform rivalry, especially in multisided settings where platforms that look like imperfect substitutes on the consumer side compete intensely for the same advertisers. The Commission should weigh out-of-market constraints on the evidence, not discount them categorically.

4.3 Countervailing buyer power (paras. 104–110)

The treatment of buyer power is orthodox and sound. Buyer power must exist and remain effective after the merger, since a merger of suppliers may itself reduce buyer power by removing a credible alternative (para. 105). We add only that buyer power and the corresponding monopsony analysis in Part II.B.2.4 should be applied with the same effects-based discipline as supplier-side power. Large, sophisticated customers able to sponsor entry or switch suppliers are a genuine and often underweighted constraint in industrial markets.

One structural point cuts across all three categories. In dynamic and digital markets, the entry, expansion, and repositioning that discipline incumbents are often financed by the very prospect of acquisition that the draft’s dynamic theories of harm would discourage. Venture capital flows to start-ups largely because acquisition offers a credible, lucrative exit. Chilling that exit reduces the financing available for the next generation of entrants and weakens the countervailing constraint of future entry (Manne, Bowman, and Auer 2022).

The Commission should therefore weigh, in assessing entry and expansion, the systemic effect of its own enforcement posture on the incentive to enter. Crediting entry as a countervailing factor while discouraging the acquisitions that finance entry would be internally inconsistent.

Part II.A.5 — Dominance and Other Types of Market Power (paragraphs 111–113)

The draft correctly treats dominance as a specific, higher degree of market power along a continuum (paras. 56, 111). It also states that a significant impediment to effective competition (SIEC) may arise below dominance (para. 113). This framing reflects the current SIEC standard as developed in the case law. The draft also rightly acknowledges that firms may legitimately acquire market power ‘through internal growth and competition on the merits’ (para. 56)—that is, through natural and efficient conduct. ICLE offers two observations to preserve certainty.

First, because several theories of harm are tethered to dominance—most notably entrenchment (paras. 252–259)—the Guidelines must define ‘dominance’ precisely and explain how it is established. Since United Brands and Hoffmann-La Roche, rigorous market definition and proof of market power within the defined market have been the foundations of dominance analysis.

Yet paragraph 54 invites assessment ‘across markets’ where they are linked as an ‘ecosystem’, and paragraphs 252–253 allow dominance—and, in particular, entrenchment—to be appraised ‘in the context of an ecosystem’. The risk is that dominance is established through narrative aggregation of positions across distinct markets, without defining a relevant market or demonstrating power within it. That would make dominance a conclusion derived from description rather than analysis.

Booking/eTraveli illustrates the danger. The ecosystem characterisation appeared only in a footnote, as Booking’s ‘wide range of services that cover multiple facets of the travel experience’, with no market-defined anchor (Colangelo 2026). The dominance threshold already embeds durability through very high shares sustained over time (para. 112). In the entrenchment scenario, which we address in Part II.B.7, the Commission should not be permitted to bypass that discipline through a loosely bounded ecosystem.

Second, the Commission should not treat below-dominance market power as a lower evidentiary hurdle. The SIEC standard—more likely than not, on a cogent and consistent body of evidence—applies whether or not the merged firm is dominant. The risk of dilution is most acute in the ‘important competitive force’ concept (paras. 138–141): a firm with a ‘relatively small—or even zero—market share’ may be labelled an important competitive force (para. 141).

That inquiry must rest on transaction-specific evidence of diversion, closeness, and margins—the tools the draft itself invokes (para. 138)—not on structural inference. The draft’s own limiter should also be applied strictly: no firm should be treated as an important competitive force where a sufficient number of comparable rivals remain (para. 141). Otherwise, the label can be turned against the very disruptive entrants whose competitive threat it is meant to capture.

Finally, ICLE welcomes the draft’s recognition that market power is a matter of degree and that the operative question is always whether the merger significantly impedes effective competition—not whether a label such as ‘market power’, ‘substantial market power’, or ‘dominance’ can be affixed. Some degree of market power is ubiquitous; only its amount matters (Landes and Posner 1980).

Identifying some degree of market power is therefore the beginning of the analysis, not its conclusion. The Commission must still show that the increment attributable to the merger significantly impedes effective competition. A finding of market power, without more, neither establishes a SIEC nor relieves the Commission of its burden of proof (Manne et al. 2024). Keeping that distinction sharp guards against the gradated framework sliding into a presumption that mergers involving large or successful firms are inherently problematic.

B.       Part II.B.1 — Anticompetitive Effects: Direct and Dynamic Effects (paragraphs 114–118)

Paragraph 114 introduces a new organising distinction between ‘direct’ effects—concerned chiefly with competition within current product markets—and ‘dynamic’ effects, which concern the merger’s impact on the merging firms’ and rivals’ ability and incentives to invest and innovate, as well as on future product-market competition. We support recognising that mergers can affect competition through more than near-term price effects. The draft is also right that the difference between direct and dynamic effects is ‘less a matter of kind and more a matter of degree’, since a single merger may exhibit both.

The introduction of a dynamic-effects category is, however, the hinge on which much of our concern about legal certainty turns. The dynamic theories of harm that follow—loss of investment and expansion competition, loss of innovation competition, loss of potential competition, dynamic foreclosure, and entrenchment—are inherently forward-looking, often qualitative, and frequently contingent on multiple uncertain future events. Three cross-cutting principles should govern this entire subsection and should be stated up front.

First, evidentiary rigour must be commensurate with uncertainty. The Court has held that the burden of proof rises with the uncertainty of the theory of harm (Tetra Laval). The more contingent or distant the alleged effect, the more demanding the evidence must be. Dynamic harms should therefore be substantiated with multi-sourced, triangulated, contemporaneous evidence—including internal documents, market studies, independent expert assessments, and, where available, ex post evidence from comparable cases—not with structural conjecture.

Second, harms and benefits must receive equal treatment. Dynamic effects cut both ways. The very features that can give rise to dynamic harm—the combination of capabilities, redeployment of research and development, and internalisation of complementarities—are also the principal sources of dynamic efficiency. The draft acknowledges this symmetry (paras. 297, 345). A merger should therefore not be charged with a speculative dynamic harm while its symmetric dynamic benefit is discounted for the same speculativeness.

Third, competition and innovation are non-monotonic. The economics literature is clear that the relationship between rivalry and innovation is non-monotonic and context-dependent: eliminating a rival does not always reduce innovation incentives and may, under some conditions, increase them (Aghion et al. 2005; Gilbert 2006; Denicolò and Franzoni 2010). The Guidelines should not embed an implicit assumption that fewer competitors mean less innovation. We develop these points in the subsections that follow.

We would add a fourth, practical principle: candour about the maturity of each theory. Some dynamic theories, such as loss of potential competition, have a longer pedigree and a reasonably developed evidentiary template. Others, including entrenchment of an ‘ecosystem’ position or dynamic conglomerate foreclosure, are comparatively novel and rest on contested economics. The Guidelines would improve predictability by acknowledging this gradient and by signalling that the more novel and contested the theory, the more exacting the required evidence and the narrower the circumstances in which the Commission will rely on it.

An error-cost lens makes the stakes plain. Because dynamic theories operate over long horizons and uncertain predictions, the cost of false positives—deterring procompetitive investment, research and development, and capability combinations—is especially high and especially difficult to detect after the fact (Manne 2020). That asymmetry counsels particular restraint in the dynamic domain.

One cross-cutting point deserves emphasis before turning to the specific theories: the dynamic counterfactual does much of the work in this entire chapter, and it must be constructed symmetrically. A dynamic theory of harm typically depends on a prediction about what the target or acquirer would have done absent the merger—launched a pipeline product, entered an adjacent market, or intensified a research-and-development race. The draft rightly requires (paras. 37–44) that such predictions be made with a ‘sufficient degree of certainty’.

That same standard must constrain the harm side. The Commission cannot treat a speculative future competitive contribution as a near certainty when building a theory of harm while dismissing the parties’ symmetric claims about future efficiencies as too uncertain to credit. Where the evidence does not support a confident prediction about the counterfactual, the appropriate inference is uncertainty about harm, not a presumption of it.

Part II.B.2 — Loss of Head-to-Head Competition (paragraphs 119–168)

The loss of head-to-head competition—unilateral effects from combining substitutes—is the draft’s most established theory of harm. The framework in paragraphs 119–168 is sound at its core. Our comments aim to keep the analysis anchored in transaction-specific evidence and to address the novel ‘specific market aspects’, especially labour monopsony.

We support the emphasis on closeness of competition—diversion ratios, margins, bidding, and switching evidence—as the analytical heart of this theory. This evidence, not structural presumptions, should govern. Consistent with Part II.A.1, high combined shares or Herfindahl-Hirschman Index increases may prompt closer inquiry, but they do not establish harm.

The question is whether the parties are close enough competitors that their combination would allow the merged firm profitably to worsen terms, accounting for repositioning and entry. Pricing-pressure tools—including upward-pricing pressure, gross upward-pricing pressure index, and merger simulation—are appropriate where the data support them. But they are models, not oracles, and their assumptions should be tested against the evidence.

A recurring problem is that closeness is assessed against an artificially narrow market, inflating diversion between the parties and understating diversion outside it. This is especially true where users multi-home and switch cheaply. Diversion ratios should therefore be estimated against the full set of realistic alternatives, including out-of-market and cross-platform options, not a narrow set that builds the conclusion into the premise.

Internal documents are probative, but they should be weighed against the quantitative record and scrutinised equally in both directions (paras. 26–31). Casual references to ‘competitors’ do not establish substitution.

The ‘important competitive force’ concept usefully captures firms whose influence exceeds their share, including mavericks and disruptive entrants. But its under-definition invites expansive and unpredictable application. The Guidelines should specify what the concept means and how it is evidenced. Relevant evidence may include a documented role in driving price or innovation responses, or a track record of disruption. The concept should not substitute for transaction-specific evidence of the constraint imposed. Otherwise, it risks manufacturing concern about firms with modest shares.

The treatment of specific market aspects is a welcome attempt to tailor the analysis. On multisided platforms, harm and efficiency must be measured at the platform level, accounting for cross-side externalities and net prices. Differentiation on one side does not preclude intense competition on another (Rochet and Tirole 2003).

On labour markets, we support confining the analysis, as the draft does, to ‘solely how the merger impacts market power on labor markets’, excluding effects ‘unrelated to the loss of competition resulting from the merger’. A labour-market theory should be entertained only where there is a clear, measurable loss of competition in a properly defined labour market, supported by evidence of employer concentration, limited mobility, and a plausible monopsony mechanism.

The Commission should separate genuine monopsony from restructuring, offshoring, or headcount reductions that flow from efficiency gains, which fall outside the EUMR. A labour effect is typically downstream of a product-market harm, since reduced output is what cuts derived labour demand. The Commission should require the same effects-based showing as for any unilateral-effects theory and should not treat job losses as such as a competition concern.

Paragraphs 167–168 would treat ‘non-structural links’ between a merging party and a competitor—including distribution, licensing, alliances, intellectual-property sharing, and network sharing—as reducing competition. Where those links ‘remove most of the competitive constraint’, the draft would aggregate the firms’ shares. This should be confined to genuinely exceptional cases tied to evidence approaching common control.

Such links are pervasive, ordinary, and usually procompetitive. Their mere existence does not make two undertakings a single competitive unit. Aggregating shares on that basis would penalise efficiency-enhancing cooperation and understate the constraint imposed by connected but independent rivals.

Consistent with error-cost logic (Manne 2020) and the Commission’s burden to substantiate harm on cogent, consistent evidence (paras. 23, 26), the Guidelines should state three principles. First, non-structural links should bear on the analysis only exceptionally. Second, share aggregation should require evidence of common control or its functional equivalent. Third, absent such evidence, a connected firm remains a full competitive constraint to count, not discount.

Part II.B.3 — Loss of Investment and Expansion Competition (paragraphs 169–174)

This new theory of harm posits that a merger may lessen competition by reducing the merging firms’ incentives or ability to invest in, or expand with, existing tangible assets. Paragraph 171 would assess that risk by reference to the parties’ market power over existing assets, their dynamic competitive potential, the degree of dynamic competitive interaction, closeness of competition, the number and capability of remaining competitors with investment capabilities, and entry or expansion by rivals.

We recognise the intuition. Capacity, investment, and expansion are genuine parameters of competition. But the theory requires careful limits to avoid condemning ordinary, efficiency-driven rationalisation of capacity and investment.

Two cautions are paramount. First, investment and expansion decisions are precisely where merger efficiencies are most likely to arise. A combined firm may invest in shared infrastructure, data centres, or capacity that neither party could justify alone, internalising returns across complementary assets and a larger addressable base. A reduction in duplicative investment is frequently an efficiency, not a harm.

The Guidelines must therefore distinguish genuine, merger-specific suppression of competitive investment from the elimination of wasteful duplication. They should require evidence that post-merger investment or capacity would fall below the competitive level, not merely that two independent investment programmes become one.

Second, because this theory rests heavily on ‘dynamic competitive potential’ and ‘dynamic competitive interaction’, it inherits all the certainty concerns raised in Part II.A.3. The factors in paragraph 171 are qualitative and elastic. Without discipline, almost any merger between two firms that invest could be framed as a loss of investment competition.

The Commission should require three showings: first, transaction-specific evidence that the parties’ investment programmes are genuine substitutes that discipline one another; second, credible evidence, predicted with sufficient certainty, that the merged firm would reduce investment or expansion below the counterfactual; and third, an honest accounting of offsetting investment efficiencies, assessed on the same evidentiary terms. Without those limits, the theory risks penalising the very scale-enabled investment the draft elsewhere encourages (paras. 11–15).

The risk of overapplication is especially acute because capacity, investment, and expansion decisions are continuous responses to demand, technology, and the cost of capital. Two independent firms will often make duplicative or partially overlapping investments. Combining them and rationalising those investments is, in the ordinary case, exactly the kind of efficiency that justifies a merger.

The theory of harm therefore asks the Commission to identify the rare case in which the merged firm would withhold investment or capacity the market would otherwise have received, and to do so through a prediction about counterfactual investment behaviour years into the future. That is a demanding evidentiary task, and the draft should say so plainly. In particular, the Commission should not treat a post-merger reduction in aggregate capital expenditure as evidence of harm. Reduced duplication and the reallocation of capital to higher-value uses are precisely how scale efficiencies manifest.

We also note that the boundary between this theory and the loss-of-innovation theory in Part II.B.4 is blurry, since investment in new capacity and investment in research and development can shade into one another. The Commission should avoid charging the same conduct twice under two labels. Whichever framework applies, offsetting efficiencies—including lower costs, the ability to fund investment that neither party could finance alone, and the elimination of double marginalisation in any vertical dimension—should be credited on the same evidentiary terms as the alleged harm (paras. 297, 341).

Part II.B.4 — Loss of Innovation Competition, and the Innovation Shield (paragraphs 175–192)

The dedicated treatment of innovation competition is among the draft’s most welcome features. Innovation and price operate through distinct mechanisms, and marginal-pricing logic is not designed to capture innovation effects. A separate, properly bounded framework for innovation harm is therefore the right step. The challenge is to capture genuine innovation harms without chilling the many mergers that combine capabilities and advance innovation.

Capabilities, contestability, and the non-monotonicity of innovation

Our central substantive point is that innovation analysis must move beyond structural proxies. Paragraphs 175–191 rest on the ‘innovation space’ construct, the ‘important innovative force’ label (paras. 189–191), and whether a ‘sufficient number’ of comparable innovators remain (para. 188). These are coarse proxies prone to overreach. Because the relationship between competition and innovation is non-monotonic (Aghion et al. 2005), with theory and evidence pointing in both directions (Cohen and Levin 1989; Gilbert 2006), sound innovation analysis must be fact intensive and case specific (Bourreau, Jullien, and Lefouili 2024).

More fundamentally, the draft conflates two concepts. The resource-based view treats firms as bundles of tangible and intangible resources (Barney 1991). Capabilities are different: they are the organisational ability to deploy those resources to a coordinated end (Helfat and Peteraf 2003). Footnote 265 collapses the distinction by defining ‘innovation capabilities’ to include ‘innovation resources’—describing what firms own, not what they can do (Ünekba? 2026).

The distinction matters. Resources can often be rebuilt or bought, so acquiring them need not durably harm competition. Capabilities, by contrast, are organisationally embedded, path dependent, and imperfectly imitable (Barney 1991; Teece, Pisano, and Shuen 1997).

Innovation harm also depends on contestability—whether the market remains open—not only on incentives. A merger that brings overlapping capabilities under unitary control, ‘thereby thwarting future variety in new product development’ (Sidak and Teece 2009), is more likely to harm innovation than one that combines complementary capabilities. Such complementary combinations often expand the frontier and may be a capital-starved project’s only route to market.

The Guidelines should therefore treat complementary-capability combinations as presumptively procompetitive and account for the innovation life cycle. Rivalry is more likely to be dampened in maturing markets than in nascent ones (Utterback and Abernathy 1975).

General innovation competition and the innovation shield

The ‘loss of general innovation competition’ theory, which assesses rivalry industrywide, risks stretching potential competition into speculative pathways that are easy to assert and hard to rebut. If retained, it should require cogent evidence that the parties are genuinely substitutable in innovation and that no comparable capability remains.

The ‘innovation shield’ (para. 192) is a welcome safe harbour for start-up acquisitions, consistent with emerging evidence that such deals are rarely anticompetitive (Ederer, Seibel, and Simcoe 2025; Manne, Bowman, and Auer 2022; Auer and Zúñiga 2026). We support it, with refinements.

First, the shield must be genuine: ‘does not find a SIEC’ should mean what it says. Second, its thresholds and the ‘three independent comparable rivals’ test must be precise enough for parties to self-assess.

Third—and most important—the shield withholds protection from ‘gatekeepers’ and the largest firms by status, routing their deals to the stricter entrenchment standard in Part II.B.7. That asymmetry imports Digital Markets Act-style, size-based regulatory presumptions into merger control. The concern is not that gatekeepers receive less protection. It is that case-specific merger analysis is distorted when a deal’s treatment turns on regulatory status rather than evidence of harm. Applied outside the Digital Markets Act’s institutional and legal context, such presumptions risk condemning the cross-market entry by which large firms discipline one another and deterring procompetitive scale-ups.

The framework must also be symmetric with the dynamic-efficiency analysis in Part II.C. The same capability combinations that may reduce innovation competition are frequently the largest source of dynamic efficiencies, including faster development, broader variety, and financing for stalled projects. A loss-of-innovation theory premised on developments years in the future must be matched by innovation efficiencies assessed over the same horizon and under the same standard. Crediting speculative harm readily while treating benefits grudgingly would reproduce the ‘innovation paradox’ (Gürkaynak 2023) and chill the very innovation the draft seeks to protect.

Part II.B.5 — Loss of Potential Competition (paragraphs 193–207)

ICLE supports recognising potential-competition harm where it is real: where actual competitors impose no effective constraint (para. 195), the target is uniquely placed to enter, and no other potential entrant remains (para. 194). Eliminating a likely, timely, capability-backed entrant in those conditions can harm competition, and the framework rightly permits intervention.

The draft distinguishes two limbs: an ‘actual constraint’, where the incumbent is already reacting to the threat (para. 196), and a ‘future constraint’, where the entrant has the ability and incentive to enter in the foreseeable future (para. 200). These track the established perceived and actual potential-competition doctrines, and both can be genuine.

Two disciplines should be explicit. First, the actual-constraint limb must not rest on the incumbent’s subjective perception alone. The draft helpfully conditions that limb on objective evidence of feasible entry and disapplies it where the entrant is ‘objectively unable to enter’ (para. 197). That objective anchor should be a necessary condition, not one factor among many.

Second, paragraph 202’s statement that ‘specific entry plans’ are ‘not necessary’, together with its relaxed posture for fast-moving markets, risks converting hypothetical entry into presumed entry. Because potential-competition harm is inherently probabilistic, its strength must scale with the likelihood and timeliness of entry, predicted ‘with a reasonable degree of certainty’ (para. 206) and to the demanding standard the Court requires for prohibitions.

Crucially, not all acquisitions of potential competitors are harmful. Many are procompetitive. Where the target cannot enter effectively, where the merger combines complementary capabilities, or where it gives a capital-starved firm the resources to scale, the deal is likely to enhance competition and innovation. A vibrant acquisition market is itself an input to venture financing and entrepreneurship: the prospect of acquisition is often the principal route to liquidity and a key reason early-stage innovation is financeable (Manne, Bowman, and Auer 2022; Draghi 2024).

Nor should the ‘killer acquisition’ concern, addressed in the Draft Guidelines under loss of innovation (fn. 258), be generalised into a presumption. Its empirical base is industry specific. Even in pharmaceuticals, only 5–7 per cent of acquisitions plausibly fall into a ‘killer acquisitions’ category (Cunningham, Ederer, and Ma 2021). The evidence does not transfer to digital markets, where innovation is faster and often complementary rather than substitutive. In the Commission’s own information-and-communications-technology merger cases, no transaction was followed by the disappearance of the target’s products or by reduced entry or innovation (Ivaldi, Petit, and Ünekba? 2025). And a study of 1,200 technology acquisitions—co-authored by an originator of the killer-acquisition empirical work—finds that post-acquisition patenting tends to rise, not fall (Ederer, Seibel, and Simcoe 2025). Distinguishing a genuine ‘kill’ from ordinary integration is also methodologically fraught, so a digital killer-acquisition presumption would invite false positives (Manne 2020).

The right posture is an ability-incentive-effect inquiry applied to the counterfactual. Would the merged firm profitably shelve or degrade the target’s offering? Would doing so raise long-run profits net of efficiencies? Would consumers be worse off than in the realistic counterfactual—which, for many start-ups, is struggle to scale or outright failure, not independent success? Where elimination of nascent competition is feasible and profitable, the framework already permits intervention. Where procompetitive explanations dominate, blocking the deal sacrifices efficiencies and raises the risk of costly false positives (Manne 2020). The Guidelines should state that potential-competition harm is one explanation among several and often not the most probable.

Finally, the ‘reverse killer acquisition’ variant—that the acquirer would itself have entered but now will not—is doubly speculative. It requires the Commission to predict both foregone independent entry and the shelving of the combined capability, in tension with the recognition that combining complementary capabilities is usually procompetitive. The theory should require concrete, contemporaneous evidence of an abandoned entry plan that the merged firm has the ability and incentive to shelve, weighed against integration efficiencies.

By symmetry with Part II.A.4 on countervailing factors, the evidentiary bar for treating a firm as a lost potential competitor should equal the bar for crediting its prospective entry as a countervailing constraint.

Part II.B.6 — Foreclosure (paragraphs 208–251)

We strongly support the draft’s retention of the ability-incentive-effect framework for foreclosure (paras. 208–251). It is the correct organising test and reflects both the case law and the modern empirical record. Unlike horizontal mergers, non-horizontal mergers do not automatically eliminate a competitor. Foreclosure harm depends on a subsequent strategic choice that must be feasible and profitable. By contrast, the efficiencies of vertical integration—especially the elimination of double marginalisation (EDM)—are frequently automatic and often arise where foreclosure risk is alleged to be highest (Cooper et al. 2005; Lafontaine and Slade 2007; Crawford et al. 2018).

6.1–6.2 Ability and incentive to foreclose (paras. 219–242)

We welcome the draft’s insistence that both ability and incentive be established, with incentive analysed using profitability-based tools. The Commission should make explicit that ‘foreclosure shares’ or vertical-concentration measures cannot establish a theory of harm absent an incentive story grounded in profitability. Converting such screens into presumptions would contradict the framework the draft endorses.

EDM and other verifiable, merger-specific efficiencies should be evaluated within the same arithmetic as the foreclosure analysis, not at a separate stage where they are easily discounted.

Two refinements would help. First, input and customer foreclosure should remain analytically distinct. Input foreclosure requires upstream power sufficient for input degradation materially to raise rivals’ costs. Customer foreclosure requires that the downstream affiliate account for such a large share of demand that its withdrawal denies upstream rivals efficient scale. A merger may raise one concern without raising the other.

Second, the incentive inquiry must account for recoupment across both levels. Foreclosure is profitable only if downstream gains exceed upstream losses, net of diverted sales and customers’ ability to sponsor entry or switch. Most vertical mergers are cleared (Lafontaine and Slade 2007).

6.3 Effect on competition, and ‘dynamic’ foreclosure (paras. 243–251)

The draft extends the analysis to dynamic effects, including foreclosure of innovation inputs—such as data, intellectual property, or platforms—on which rivals’ future innovation depends. We agree this is conceptually possible. Placing a critical, irreplaceable input under unitary control and restricting access can impair rivals’ innovation; the abandoned Nvidia/Arm deal is a frequent illustration. But the analysis must turn on how critical and irreplaceable the input is, and whether rivals can use alternatives, build their own, acquire substitutes, or innovate around it.

The mere addition of an element to a portfolio—whether ‘financial strength’, as in GE/Honeywell, or breadth, as in Booking/eTraveli—should not suffice (Teece 1986).

We caution against paragraph 251’s treatment of ‘diagonal’ mergers as presumptively unlikely to generate efficiencies. Many such combinations—including the AI partnerships now reshaping the industry—supply complementary capital, compute, and distribution that let targets scale, with no concrete evidence of foreclosure to date (Auer and Zúñiga 2026). A category-level presumption is inconsistent with the evidence and with the draft’s own recognition that mergers without head-to-head overlap have stronger integration potential (para. 17).

Those AI partnerships test the framework. Cloud providers supplying compute to AI labs involve the input dependencies a dynamic-foreclosure theory targets. Yet the record shows theoretical concerns but no concrete foreclosure. On balance, these arrangements have expanded competition by financing capital-starved entrants. The test must be applied to the facts: whether the input is truly scarce and irreplaceable, whether withholding would be profitable, and whether rivals have alternatives. Harm should not be inferred from structure. A firm that has just paid to strengthen a partner ordinarily has an incentive to see it succeed, not foreclose it.

Dynamic theories also carry a temporal hazard. They depend on a chain of predictions: that an input will remain scarce, that no substitute will emerge, that the merged firm will withhold it, and that rivals cannot adapt. Each link compounds the uncertainty. Interoperability and data-portability obligations, open standards, and ordinary entry of substitutes often dissolve the scarcity on which the theory rests. Where the Commission cannot predict with confidence that an input will remain a durable bottleneck, the theory should not carry the case.

Finally, where a foreclosure concern is made out, access and nondiscrimination commitments can often address it at far lower cost than prohibition while preserving integration efficiencies. Consistent with proportionality, the Commission should prefer the least restrictive effective remedy and should not treat prohibition as the default response to a contingent foreclosure theory.

Part II.B.7 — Entrenchment of a Dominant Position (paragraphs 252–259)

Entrenchment is one of the draft’s most novel and legally precarious theories of harm. Under paragraph 252, entrenchment occurs when a merged firm gains control of assets that ‘structurally create or reinforce existing barriers to entry and expansion’, reducing contestability and deterring future entry, expansion, or innovation, whether within a ‘core market’ or across an interconnected ‘ecosystem’.

We do not dispute that a merger could, in principle, reinforce dominance by acquiring a genuinely critical, scarce, and non-replicable asset. But as drafted, the theory rests on concepts—‘ecosystem’, ‘core market’, ‘strategic’ assets, network effects, and customer inertia—that are too ill defined to deliver predictable outcomes. It risks condemning the ordinary accumulation of complementary capabilities that is competition on the merits.

The pivotal weakness is the ‘ecosystem’ construct (paras. 252–253). In competition law, ‘ecosystem’ largely operates as a descriptive label for dynamics already captured by established categories, such as aftermarkets, multisided platforms, and portfolio effects, rather than as a freestanding analytical framework (Colangelo 2026). EU practice bears this out: the term has been used loosely (Google Android), relegated to a footnote (Booking/eTraveli), and avoided by the Court (Android Auto). In practice, it has not influenced market definition, conduct assessment, or merger review.

Allowing dominance to be established across a loosely defined ‘ecosystem’ of ‘closely related markets’ would expand the theory beyond its proper bounds and undermine the work the dominance threshold is meant to do. We return to that concern in Part II.A.5.

Entrenchment should also be anchored in durability. Entrenched market power is power that is durable and hard to dislodge, shown by long-term evidence such as persistently stable shares or enduring barriers. It is distinct from merely substantial or current market power (Manne et al. 2024).

Equally important, network effects, scale economies, and customer inertia—the very ‘market dynamics’ paragraph 254 treats as making entrenchment more likely—are double-edged and frequently procompetitive. Network effects intensify competition for the market and rarely confer perpetual dominance absent exclusionary conduct. Markets that once looked entrenched have repeatedly been disrupted: MySpace by Facebook, Symbian and BlackBerry by iOS and Android, and incumbents by generative-AI entrants (Liebowitz and Margolis 1999; Evans and Schmalensee 2016). Treating these features as presumptive markers of entrenchment risks penalising successful firms. Acquiring complementary assets to improve an integrated offering is a normal mechanism of competition, not a structural harm.

If retained, the theory should be confined to four conditions. First, there must be a rigorously established, durable, pre-existing dominant position in a properly defined core market, not an aggregated ‘ecosystem’. Second, the acquired asset must be genuinely unique, scarce, or irreplaceable and important to competing in the core market, and rivals must be unable to replicate or work around it within a commercially relevant horizon. Third, the Commission must show a merger-specific mechanism by which contestability would actually fall—through higher barriers, deterred entry, or reduced innovation—causally tied to the transaction and predicted with the heightened certainty the Court requires for forward-looking harm (Tetra Laval). Fourth, the Commission must credit, symmetrically, the integration efficiencies that the same combination produces.

The draft’s own limiters—paragraph 255’s ‘no plausible connection’ and paragraph 258’s rival counterstrategies—point in this direction but are too loosely drawn. Without these limits, entrenchment becomes a roving commission to block complementary acquisitions by large firms, with substantial error costs and little predictability.

A further concern is that the theory risks importing into ex ante merger control the conduct-based concerns properly addressed, if at all, under Article 102 and the Digital Markets Act—without any conduct having occurred and without those regimes’ safeguards. Reinforcing a dominant position is not unlawful in itself. The law prohibits abusive conduct that exploits or extends dominance, and the Digital Markets Act already addresses the concern that the largest platforms may entrench themselves through conduct.

Layering a broad, structurally framed entrenchment theory on top raises a real risk of double regulation and of penalising firms for size and success rather than identifiable harm. The Commission should state expressly that entrenchment requires a merger-specific mechanism of harm, not merely the observation that a strong firm has become stronger.

Part II.B.8 — Coordination (paragraphs 260–281)

The coordinated-effects framework in paragraphs 260–281—reaching terms of coordination, monitoring and deterring deviation, and resisting disruption—faithfully tracks the Airtours conditions and remains the right organising structure. We support its retention. Our comments concern the evidentiary burden and the calibration of the analysis to modern market features.

First, the Guidelines should be explicit about what evidence shows that a merger makes coordination ‘more likely than not’, consistent with CK Telecoms. It is not enough to recite that a market is concentrated, transparent, or symmetric. The Commission should specify which market facts move the needle: transparency sufficient to monitor deviations, symmetry of incentives and capacities, the elimination of a maverick, and the absence of effective outsider or buyer disruption. It should also identify what documentary or empirical evidence is required to establish each, particularly in differentiated-products settings or where algorithmic pricing is alleged to facilitate coordination. The broader empirical record is relevant context: most vertical and many horizontal combinations do not produce coordination, and the theory should not be applied on structural inference alone.

Second, the elimination-of-a-maverick strand connects to the ‘important competitive force’ concept and should be held to the same evidentiary discipline. We return to that concept in Part II.B.2.3. The Commission should require concrete evidence that the target in fact played a disruptive, coordination-defeating role, not merely that it had a modest share or a different business model. Differentiation is frequently the form competition takes, not evidence of its absence.

Third, the draft should retain a realistic account of the factors that disrupt coordination, including maverick rivals, lumpy or infrequent orders, heterogeneous costs, demand volatility, and the difficulty of sustaining tacit collusion in fast-moving or innovative markets. In dynamic settings, unstable market positions and the pace of innovation often make durable coordination implausible. The Commission should weigh those features symmetrically, rather than focusing only on conditions conducive to coordination.

We would add a specific note on algorithmic pricing, which the draft and contemporary enforcement increasingly invoke as a coordination facilitator. The economics are unsettled and frequently misunderstood. Pricing algorithms can intensify competition by speeding price discovery and lowering search and menu costs. The conditions under which they sustain genuinely collusive outcomes—as opposed to rapid competitive responses—are narrow and contested.

The Commission should not treat the use of common pricing software, or the technical possibility of algorithmic monitoring, as evidence of coordination risk. It should instead require concrete evidence that the merger makes a sustainable collusive equilibrium materially more likely. Coordinated-effects prohibitions have historically been, and should remain, demanding and comparatively rare, reflecting both the stringency of the Airtours conditions and the practical difficulty of sustaining tacit collusion.

Part II.B.9 — Other Anticompetitive Effects (paragraphs 282–290)

9.1 Access to commercially sensitive information (paras. 282–286)

The concern that a merger may give the merged entity access to rivals’ commercially sensitive information—particularly in vertically integrated settings—is legitimate but narrow. It has also historically been addressed effectively through behavioural commitments, such as information firewalls and ring-fencing.

The Guidelines should make clear that this theory requires evidence of three things: the information is genuinely competitively sensitive; the merged firm would have both the ability and incentive to exploit it to rivals’ disadvantage; and doing so would harm competition net of efficiencies. Where concerns are credible, targeted commitments should generally be preferred to prohibition. The mere possession of information by a vertically integrated firm, which is common and usually benign, is not itself a harm.

This theory overlaps substantially with both vertical foreclosure and the established rules on information exchange under Article 101. The Commission should not convert that overlap into a standing structural presumption. Access to a rival’s data becomes a competition concern only where the information is competitively sensitive, is not otherwise available through public sources, licensing, or the merged firm’s own operations, and could be deployed to the rival’s disadvantage in a way that harms competition net of efficiencies.

The Commission’s own practice illustrates the danger of stopping short of that inquiry. In General Electric/Alstom, the Commission observed that the merged entity could aggregate customer information on demand patterns relevant to continued innovation. But the mere capacity to aggregate information, which is common to any integrated supplier, is not itself a harm. The Commission must show that the information confers a durable advantage that rivals cannot replicate, and that the merged firm has both the ability and incentive to exploit it to foreclose—not merely that it would hold data its constituent businesses already possessed.

Where a credible concern is identified, it will often be amenable to a targeted behavioural remedy. Information firewalls, ring-fencing, and non-discrimination commitments have a long and effective track record in vertically integrated mergers precisely because the concern is narrow and the conduct it targets is severable from the integration’s efficiencies. Consistent with proportionality, prohibition should be reserved for the rare case in which no such commitment can neutralise the concern.

An overbroad information theory carries its own error cost. The routine information flows that accompany vertical integration—better demand forecasting, coordinated production planning, and the elimination of double marginalisation—are ordinarily procompetitive. A presumption against them would deter the very integration the draft elsewhere recognises as beneficial.

9.2 Portfolio effects (paras. 287–290)

Portfolio effects have a long and cautionary history in EU merger control, from Guinness/Grand Metropolitan through the GE/Honeywell debate. The economic learning since then counsels scepticism. Bundling and a broad product range are frequently procompetitive: they can lower transaction costs and prices and improve the user experience. A theory that a firm will leverage a portfolio to foreclose rivals must overcome the same ability-incentive-effect hurdles as any foreclosure theory.

We are particularly concerned that ‘portfolio effects’ not become a relabelled ‘ecosystem’ theory. We return to that concern in Part II.B.7. Much of what the draft describes as ecosystem entrenchment is, in substance, the portfolio-effects analysis of the past. Importing that analysis without its hard-won limits would be a step backward.

The Commission should confine portfolio-effects theories to cases with concrete evidence that the combined portfolio confers the ability and incentive to foreclose, and that foreclosure would harm consumers net of the efficiencies that broad portfolios typically generate.

Taken together, both limbs of this subsection should be governed by the same discipline that applies to foreclosure generally: a concrete mechanism, ability and incentive grounded in profitability, an effect on competition net of efficiencies, and a strong preference for the least restrictive effective remedy where a concern is made out.

Neither access to information nor breadth of portfolio is, in itself, a harm. Each is a common and usually procompetitive feature of integrated firms. The risk we flag throughout these comments—that loosely framed, structurally inferred theories will sweep in benign conduct—is most acute for these ‘other’ effects precisely because they are open-ended. The Commission should anchor them firmly in the effects-based framework it endorses elsewhere.

C.      Part II.C — Benefits from Mergers (Efficiencies) (paragraphs 291–301)

The efficiencies chapter is, in our view, the draft’s most important improvement. It recognises a ‘theory of benefit’ (para. 25), distinguishes direct and dynamic efficiencies in parallel with direct and dynamic harms (para. 294), and, above all, insists that efficiencies be assessed with ‘an equivalent degree of likelihood over time’ as the harms they offset (para. 297).

Those changes respond directly to the long-standing critique that EU merger control has applied a ‘double standard’: demanding rigorous, quantified, near-term proof of benefits while constructing harms more freely and over longer horizons. We commend this reorientation and urge the Commission to carry it through consistently. The non-exhaustive list in paragraph 298—scale and scope economies, combinations of complementary assets and capabilities, procurement synergies, access to critical inputs, new, improved, or more affordable products, access to finance, better allocation of research-and-development resources, and enhanced ability and incentive to innovate—is sound and economically literate.

The single most important thing the Commission can do here is to mean what paragraph 297 says. Symmetry is not merely a matter of fairness. It is a matter of analytical coherence. If a dynamic harm premised on the discontinuation of a research-and-development project years in the future is cognisable, then a dynamic efficiency over the same horizon must be cognisable on the same terms. Conversely, parties should not be required to quantify and prove benefits to a standard the Commission does not apply to its own theories of harm.

We ask the Commission to add an explicit, operative statement that the standard of proof, the quality of evidence required, and the temporal horizon are identical for theories of harm and theories of benefit. The Commission should also commit, as paragraph 36 suggests, to early and good-faith engagement on efficiencies during the review.

This reform is consequential because the efficiency defence has, to date, been largely theoretical in EU practice. Although the 2004 framework nominally permitted efficiencies to be weighed, no merger has been cleared primarily on the strength of an efficiency defence. Practitioners have widely regarded the defence as something close to a dead letter, partly because of the demanding and asymmetric evidentiary burden, and partly because raising efficiencies was perceived as conceding the existence of harm (Padilla 2019).

The draft’s structural changes—an affirmative theory of benefit, parity of evidentiary and temporal standards, and encouragement of early engagement—are precisely what is needed to make the defence operative rather than ornamental. We therefore urge the Commission not only to adopt these provisions but also to signal, in its decisional practice, that a well-substantiated efficiency case can and will carry the day. A defence that exists on paper but never succeeds in practice does nothing for the competitiveness the draft seeks to promote.

Part II.C.1 — Assessment of Direct Efficiencies (paragraphs 302–323)

The familiar trio of verifiability, merger specificity, and benefit to consumers (paras. 303, 309, 314) remains the right structure for direct efficiencies, and we support it. Our comments aim to ensure those criteria are applied in a way that does not, in practice, render the defence illusory, as has too often been the case.

Verifiability (para. 1.2)

Quantification is appropriately the gold standard. Cost data showing expected reductions in marginal or variable cost, or the achievement of minimum efficient scale, are persuasive forms of evidence. But the Commission should define minimum standards of proof by reference to the totality of the evidence and should accept that not all genuine efficiencies are precisely quantifiable ex ante. This is especially true of efficiencies arising from asset or capability combinations, which may be demonstrated through internal documents, premerger plans, independent expert studies, and past examples of realised efficiencies.

The same evidentiary discipline the Commission applies when it builds a theory of harm from internal documents and market evidence should suffice for efficiencies. A best-practices note on substantiating efficiencies, paralleling the notice on the submission of economic evidence, would materially help both parties and the Commission.

Merger specificity (para. 1.3)

The merger-specificity test asks whether efficiencies could be achieved by less restrictive means, such as licensing or a joint venture. That test has intuitive appeal, but it should be applied with caution. From a transaction-cost perspective, so-called ‘less restrictive alternatives’ are often not less restrictive in practice. Contracts and joint ventures are costly to negotiate, implement, and monitor, and they create lock-in, hold-up, and rent-extraction risks that may lead firms to forgo the arrangement entirely—so the efficiency never materialises (Williamson 1985; Klein 1996).

Management research likewise treats partnerships and mergers as distinct organisational modes, not functional substitutes (Hagedoorn and Sadowski 1999). We welcome the draft’s recognition that a merger’s greater profitability than an agreement does not, by itself, defeat merger specificity, and that alignment of interests may be ‘very difficult’ (para. 312). The Commission should resist assuming that integration efficiencies can necessarily be replicated by contract.

Benefit to consumers (para. 1.4)

The requirement that benefits accrue to ‘substantially the same consumers’ who would otherwise be harmed (para. 314) is, as drafted, quite narrow. The Commission should clarify how it will handle two recurring trade-offs.

The first is the intermediate-versus-final-consumer trade-off, especially in multisided markets where a merger may raise costs for business users while lowering prices or improving service for end users. Consistent with the Court’s emphasis on protecting consumers, such trade-offs are best resolved in favour of final consumers rather than edging toward a ‘trading-partner welfare’ standard.

The second is the out-of-market trade-off. Past practice has linked the consideration of out-of-market efficiencies to ‘considerable commonality’ between consumer groups (Mohan 2014). The Commission could usefully clarify when groups are sufficiently common to permit balancing.

The decisive practical question for pass-on is the degree of residual competitive pressure. Efficiencies are more likely to reach consumers where rivalry remains. In fast-moving markets, entry and potential competition often supply that pressure and explain why the efficiency-seeking transaction was undertaken in the first place.

One practical point cuts across all three criteria: they should be applied as a genuine, winnable defence, not as a series of hurdles each of which can be used to reject the claim. In particular, the Commission should avoid the pattern, common under the prior framework, of acknowledging an efficiency in principle while finding it insufficiently verifiable, insufficiently merger specific, or insufficiently likely to be passed on, with the cumulative effect that no efficiency ever quite qualifies.

Symmetry requires that each criterion be applied with the same realism, and the same tolerance for reasonable inference, that the Commission applies when assembling a theory of harm from comparable evidence. The burden on the parties is one of production and substantiation. Through the accretion of stringent subtests, it should not become a practical impossibility.

Part II.C.2 — Assessment of Dynamic Efficiencies (paragraphs 324–338)

The express recognition of dynamic efficiencies—improved or new products, better distribution or production, and enhanced ability and incentive to invest and innovate—is especially valuable. Dynamic efficiencies are often the largest welfare gains a merger can produce, and they are systematically undervalued by frameworks built for static price effects.

We support the draft’s approach, including its acknowledgement that agreements such as joint ventures may, in some circumstances, not be realistic alternatives to a merger where dynamic efficiencies are concerned (para. 333). That is an important and correct concession to the transaction-cost and management literatures.

The principal risk is that the verification standard for dynamic efficiencies becomes more demanding than the standard the Commission applies to its own dynamic theories of harm. Paragraph 326 asks parties to ‘explain and substantiate with a sufficient degree of likelihood’ and ‘as concretely as possible’ the nature of the investment or innovation. If applied asymmetrically, that standard would reproduce the ‘innovation paradox’ and defeat the symmetry of paragraph 297.

Innovation is, by definition, uncertain and forward-looking. That uncertainty afflicts innovation theories of harm just as much as innovation efficiencies. It would therefore be incoherent to demand a higher evidentiary threshold for innovation efficiencies than for innovation harms (Padilla 2019).

The appropriate evidence is the same on both sides: the parties’ assets and capabilities, including research-and-development budgets, staff, facilities, and strategic plans; whether their research programmes overlap, enabling productivity gains, or are complementary, enabling synergies; business and integration plans; product-development forecasts; and independent expert studies. That evidence should be assessed in light of the market’s innovation cycle.

On timing, we support the draft’s preservation of an open-ended ‘timely’ standard rather than a fixed limit, and we urge the Commission to apply it with explicit symmetry. If the Commission accepts a theory of harm premised on the loss of a pipeline project several years from market, efficiencies expected over a comparable horizon must also be admissible.

The Commission has itself recognised divergent horizons in innovation cases, from one to two years in Novartis/GSK Oncology to nearly a decade in Dow/DuPont. The same case-by-case, life-cycle-sensitive calibration should govern efficiencies.

We also encourage the Commission to build ex post evaluation into its practice (Komninos and Petit 2021). Periodic reviews of past mergers would generate empirical evidence on when and how dynamic efficiencies materialise, strengthening future ex ante assessment and improving transparency.

To make dynamic-efficiency analysis concrete and predictable, the Commission should align it with the innovation framework in Part II.B.4 and with the innovation shield. Where a merger combines complementary research-and-development capabilities, supplies a capital-constrained innovator with the resources to bring a project to market, or internalises spillovers between adjacent research programmes, those should be cognisable dynamic efficiencies.

The evidence that establishes such efficiencies is the same evidence the Commission would examine to assess innovation harm: pipelines, capabilities, integration plans, and the market’s innovation cycle.

The Commission should also make clear that the dynamic efficiencies of vertical and conglomerate integration—improved coordination of complementary investments, faster iteration, and the elimination of double marginalisation that frees resources for further investment—are presumptively credible. That presumption would be consistent with the empirical record showing that such integration is, on balance, procompetitive (Lafontaine and Slade 2007; Crawford et al. 2018).

Part II.C.3 — Balancing Benefit and Harm (paragraphs 339–357)

The balancing chapter is where the draft’s symmetry must become operational. We welcome much of its structure: the recognition that efficiencies need only ‘offset’ the harm (para. 339); the appropriate focus, in symmetric cases, on whether dynamic efficiencies offset dynamic harms (para. 345); and the willingness to use quantification tools, net-present-value analysis, and willingness-to-pay methods where the parameters are comparable (paras. 343, 347).

The draft’s use of risk-regulation logic—weighing both the likelihood and magnitude of effects across time (para. 341)—is a sensible way to compare near-term and distant effects without categorically discounting the latter.

We have two concerns. The first is the ‘margin of discretion’ the Commission reserves for balancing ‘incommensurable’ price and non-price parameters (para. 342). This is where the draft’s ambitions and its certainty objective most directly collide. How is the Commission to weigh, for example, a resilience benefit that may require excess capacity and thus static inefficiency against an efficiency harm? How should it weigh a sustainability benefit against a price increase? The draft supplies a vocabulary but not a method.

We do not suggest these trade-offs are easy. We suggest that unstructured discretion to resolve them is the opposite of the predictability paragraph 5 promises. At a minimum, the Commission should commit to a structured, transparent framework—likelihood, magnitude, timing, and the parameter affected—and should explain its weighting in each decision, so that the exercise is reviewable and learnable rather than ad hoc.

The second concern is the latent thumb on the scale in paragraphs 345–346. Those paragraphs suggest that where a merger produces both substantial harm and substantial benefit, the analysis should lean toward intervention because ‘competition is an important long-term driver of efficiency and innovation’ and lower residual competition reduces the incentive to maintain efficiencies. There is a kernel of truth here: pass-on depends on residual competitive pressure. But that point should not harden into a presumption against close cases.

Symmetry requires that genuine, well-evidenced substantial benefits be credited even when harms are also substantial. Otherwise, the ‘offset’ standard in paragraph 339 quietly becomes an asymmetric ‘clearly outweigh’ standard imposed on the parties alone. The cleaner and more defensible approach is to enlarge the Commission’s temporal and analytical horizon symmetrically—evaluating harms and benefits over the same timeframe and calibrating the weight of each strictly to the evidence—rather than discounting long-term benefits while entertaining long-term harms.

Finally, on balancing across different consumer groups or markets, we encourage the Commission to adopt a transparent and consistent rule. Where harmed and benefited consumers are largely the same group, balancing is straightforward. Where they differ—harm to one customer group or market, benefit to another—the draft should make explicit when out-of-market or cross-group benefits may be counted, building on the ‘considerable commonality’ approach of past practice (Mohan 2014).

Two principles should guide that analysis. First, in multisided markets, the relevant unit is the whole platform. A benefit to one side funded by a charge on another is therefore not properly characterised as a cross-market trade-off. It is an integral feature of a single product.

Second, where a genuine cross-group trade-off exists, the Commission should resolve it by reference to total consumer welfare in the affected markets, transparently and symmetrically. It should not privilege the harmed group in a way it would not privilege a harmed group when the roles are reversed on the benefit side. Consistency here is, once again, the route to predictability.

Part III — Measures to Protect Legitimate Interests (paragraphs 358–362)

We welcome the draft’s defence of the ‘one-stop shop’ and the Commission’s exclusive competence over EU-dimension mergers (paras. 358–359). Both are important contributors to legal certainty, reduced administrative burden, and integration of the internal market. We also welcome the placement of the burden on member states to prove that any Article 21(4) measure pursues a genuine legitimate interest and complies with the general principles of EU law, including proportionality and nondiscrimination (paras. 360–362).

As a matter of principle, the Commission should read Article 21(4) narrowly and conservatively, and should continue to see through pretextual invocations of ‘public policy’ used to build national champions or shield domestic firms—as it did, for example, in resisting Italy’s attempt to block Vivendi/Mediaset. The value of Article 21(4) lies precisely in its exceptional character. Allowing it to expand would fragment the internal market the one-stop shop is designed to protect.

Codifying the Article 21(4) framework in the Guidelines usefully promotes predictability, provided the substantive bar remains high. The clearer the Commission is about what does—and does not—qualify as a legitimate interest, and about the evidentiary and proportionality showings member states must make, the less room there will be for Article 21(4) to be used as an industrial-policy or protectionist instrument.

The placement of legitimate-interest review in a narrow, tightly bounded carve-out reflects a sound constitutional logic that the Guidelines should make explicit. The EU Merger Regulation (EUMR) confers on the Commission a single, competition-based competence. Article 21(4) is the limited mechanism by which member states may protect a narrow set of non-competition interests, subject to EU oversight.

That structure exists to prevent the substantive competition assessment from becoming a vehicle for industrial policy, protectionism, or contested social objectives. That is precisely why the non-competition considerations that appear elsewhere in the draft—resilience, sustainability, and media and cultural diversity—are best handled, where legally cognisable at all, through this channel rather than absorbed into the significant impediment to effective competition (SIEC) analysis.

Confining such interests to Article 21(4), and policing that channel strictly, is the cleanest way to reconcile the draft’s broader ambitions with the certainty and neutrality of merger control.

A.      Part III.A — Substantive Assessment of Legitimate Interests (paragraphs 363–389)

Public security (paras. 368–371)

We support the draft’s strict, case-law-grounded reading of public security. Public security must be interpreted narrowly, cannot be determined unilaterally by member states without EU control, may be invoked only against a ‘genuine and sufficiently serious threat to a fundamental interest of society’, and must not be misapplied to serve economic ends.

Paragraph 370 is particularly important: member states or their nationals are, prima facie, not a threat to another member state’s public security. Intra-EU mergers should therefore raise public-security concerns only exceptionally. The Commission should hold that line firmly.

The principal risk is that public-security and foreign-investment rationales will be used to block or redesign deals for protectionist reasons, as in aspects of GE/Alstom. The Guidelines should clarify when such interventions are, and are not, consistent with Article 21(4).

Media plurality and ‘democracy’ (paras. 372–374)

This is where we most strongly urge restraint. The draft’s framing is correct: the Commission assesses media-sector mergers ‘according to the same principles as mergers in other sectors’, taking diversity into account only where it is a relevant parameter of competition, while media-plurality review proper proceeds under the European Media Freedom Act and national law (paras. 373–374). We support that framing but caution against drift beyond it.

The EU Merger Regulation (EUMR) prevents abuses of market power. It is not a tool for regulating speech, editorial diversity, or viewpoint pluralism, and ‘democracy’ has never been a value it protects. No merger has been blocked for ‘harming democracy’, nor would the Treaties permit it. ‘Harm to democracy’ is contested, politically charged, and analytically unstable. The link between mergers, lobbying, and democracy is unsettled and rests on contestable modelling (Valletti and Broso 2024). Lobbying is, in any event, a recognised feature of democratic participation in the EU (Articles 10–11 TEU; the transparency register). Importing such an ill-defined goal would exceed the Commission’s mandate, require amendment of the EUMR and arguably the Treaties, and damage the predictability and neutrality of merger review.

Media plurality proper raises distinct difficulties that confirm it belongs with media-specific instruments. Measuring viewpoint diversity is far harder than assessing product choice. It requires defining, gauging, ranking, and prioritising opinions—normative judgments the Commission is neither mandated nor equipped to make. Nor do all media mergers reduce plurality. Diversification can be efficiency driven, and ‘more’ media is not always better given clutter, disinformation, and low-quality reporting.

Where the Commission does consider diversity, it should do so only as a genuine competition parameter, such as product variety and consumer choice. Editorial and viewpoint-pluralism concerns should be left to the European Media Freedom Act, national media laws, and—where a member state acts—the narrow Article 21(4) mechanism, approached with scepticism toward pretextual claims.

For Article 21(4) media-plurality claims, the Commission should apply clear filters. It should require demonstrable risks to diversity, not slogans about identity or sovereignty; market-wide impact, not firm-specific interest; and proof that citizens’ access to diverse information would meaningfully decline despite abundant channels. It should also insist on consistency with EU values and competition principles, including no shielding of inefficient or state-favoured firms, and should require transparency and proportionality. These filters would respect legitimate member-state interests while preventing Article 21(4) from creeping beyond its exceptional scope.

Other public interests, proportionality, and non-discrimination (paras. 366–367, 375–389)

On ‘other public interests’ and general principles, including proportionality, nondiscrimination, and compatibility with other EU law, we support the draft’s insistence that member-state measures must pursue a real legitimate interest, must not constitute arbitrary discrimination or a disguised restriction on free movement of capital or freedom of establishment, and must be suitable and proportionate.

The Commission should resist broadening the recognised-interest categories. It should treat increased market power vis-à-vis public authorities, or other diffuse ‘public interest’ concerns, as outside its mandate where they are untethered to a competitive effect on consumers in a relevant market. Clear, law-based boundaries protect both the integrity of the EUMR framework and the legitimate prerogatives of member states.

B.       Part III.B — Procedural Framework (paragraphs 390–399)

The procedural framework for Article 21 review—notification and standstill, the range of possible Commission decisions, and the interaction with other EU-law proceedings (paras. 390–399)—is a useful codification. We support clarifying the procedure by which member-state measures affecting EU-dimension mergers are assessed.

Predictable procedure is itself a contributor to legal certainty. The Commission should be explicit about timelines, the evidentiary submissions it expects from member states, and the consequences of failing to substantiate a claimed legitimate interest, including the Commission’s power, reflected in paragraph 362, to find an infringement of EU law.

We offer two observations. First, the procedural rules should reinforce the substantive discipline of Part III.A. A member state invoking a legitimate interest should bear a clear, front-loaded burden to identify the specific risk and provide specific evidence, so that pretextual or protectionist measures can be filtered early.

Second, the Guidelines should clarify the relationship between parallel national review—for example, a media-plurality review under national law implementing the European Media Freedom Act, or foreign-investment screening—and the Commission’s EUMR assessment. Firms should understand which proceedings apply, in what sequence, and with what standstill consequences. Coordination and sequencing clarity would reduce duplication, avoid staggered review, and protect the one-stop shop the draft rightly defends.

In closing, we reiterate our overarching message. The Draft Guidelines contain genuine and welcome advances: symmetry between harm and benefit, a dedicated and more sophisticated treatment of innovation and dynamic competition, the innovation shield, and a candid recognition that most mergers are procompetitive. Those advances will deliver their promised benefits only if the instrument resists the temptation to do everything at once.

We respectfully encourage the Commission, in its next draft, to pair its ambitions with clear, administrable rules: soft structural screens rather than presumptions; symmetric evidentiary and temporal standards for harms and benefits; tightly bounded and well-defined dynamic theories of harm; and a disciplined, narrow treatment of non-competition objectives. That is the surest way to honour the draft’s own promise, in paragraph 5, to increase legal certainty and predictability—and, in doing so, to advance the competitiveness of the internal market.

Selected References

Acquisti, A., C. Taylor, and L. Wagman. 2016. ‘The Economics of Privacy’. Journal of Economic Literature 54 (2): 442–492.

Aghion, P., N. Bloom, R. Blundell, R. Griffith, and P. Howitt. 2005. ‘Competition and Innovation: An Inverted-U Relationship’. Quarterly Journal of Economics 120 (2): 701–728.

Albrecht, B. C., and R. A. Decker. 2026. ‘Markups and Business Dynamism Across Industries’. International Journal of Industrial Organization, in press.

Auer, D., and G. Manne. 2024. Is Data Really a Barrier to Entry? Rethinking Competition Regulation in Generative AI. Mercatus Center.

Auer, D., and M. Zúñiga. 2026. ‘AI Partnerships and Competition: Damned if You Buy, Damned if You Don’t’. International Center for Law & Economics.

Barney, J. B. 1991. ‘Firm Resources and Sustained Competitive Advantage’. Journal of Management 17 (1): 99–120.

Berry, S. T., M. Gaynor, and F. Scott Morton. 2019. ‘Do Increasing Markups Matter? Lessons from Empirical Industrial Organization’. Journal of Economic Perspectives 33 (3): 44–68.

Bourreau, M., B. Jullien, and Y. Lefouili. 2024. ‘Mergers and Innovation’. Working paper/survey.

?irjevskis, A. 2019. ‘The Role of Dynamic Capabilities as Drivers of Business Model Innovation in Mergers and Acquisitions of Technology-Advanced Firms’. Journal of Open Innovation 5 (1): 12.

Cohen, W. M. 2010. ‘Fifty Years of Empirical Studies of Innovative Activity and Performance’. In Handbook of the Economics of Innovation, vol. 1. North-Holland.

Cohen, W. M., and R. C. Levin. 1989. ‘Empirical Studies of Innovation and Market Structure’. In Handbook of Industrial Organization, vol. 2. Elsevier.

Colangelo, G. 2026. ‘Do Ecosystems Exist in EU Competition Law?’ European Law Review 51 (3): 305.

Cooper, J. C., L. M. Froeb, D. P. O’Brien, and M. G. Vita. 2005. ‘Vertical Antitrust Policy as a Problem of Inference’. International Journal of Industrial Organization 23 (7–8): 639–664.

Crawford, G. S., R. S. Lee, M. D. Whinston, and A. Yurukoglu. 2018. ‘The Welfare Effects of Vertical Integration in Multichannel Television Markets’. Econometrica 86 (3): 891–946.

Cunningham, C., F. Ederer, and S. Ma. 2021. ‘Killer Acquisitions’. Journal of Political Economy 129 (3): 649–702.

Demsetz, H. 1973. ‘Industry Structure, Market Rivalry, and Public Policy’. Journal of Law and Economics 16 (1): 1–9.

Denicolò, V., and L. A. Franzoni. 2010. ‘On the Winner-Take-All Principle in Innovation Races’. Journal of the European Economic Association 8 (5): 1133–1158.

Dewatripont, M., B. Jewitt, and J. Tirole. 1999. ‘The Economics of Career Concerns, Part II: Application to Missions and Accountability of Government Agencies’. Review of Economic Studies 66 (1): 199–217.

Draghi, M. 2024. The Future of European Competitiveness: Report to the President of the European Commission. European Commission.

Ederer, F., R. Seibel, and T. Simcoe. 2025. ‘Digital (Killer?) Acquisitions’. Working paper.

Evans, D. S., and R. Schmalensee. 2016. Matchmakers: The New Economics of Multisided Platforms. Harvard Business Review Press.

Farrell, J., and P. Klemperer. 2007. ‘Coordination and Lock-In: Competition with Switching Costs and Network Effects’. In Handbook of Industrial Organization, vol. 3. Elsevier.

Gautier, A., and J. Lamesch. 2022. ‘Mergers and Acquisitions in the Digital Sector: More than Just Killer Acquisitions?’ Information Economics and Policy 59: 100949.

Gilbert, R. 2006. ‘Looking for Mr. Schumpeter: Where Are We in the Competition–Innovation Debate?’ Innovation Policy and the Economy 6: 159–215.

Gürkaynak, G. 2023. ‘The “Innovation Paradox” in Merger Control’. Commentary.

Hagedoorn, J., and B. Sadowski. 1999. ‘The Transition from Strategic Technology Alliances to Mergers and Acquisitions’. Journal of Management Studies 36 (1): 87–107.

Hagiu, A., and J. Wright. 2020. ‘Data-Enabled Learning, Network Effects, and Competitive Advantage’. Working paper.

Helfat, C. E., and M. A. Peteraf. 2003. ‘The Dynamic Resource-Based View: Capability Lifecycles’. Strategic Management Journal 24 (10): 997–1010.

Holmström, B., and P. Milgrom. 1991. ‘Multitask Principal–Agent Analyses: Incentive Contracts, Asset Ownership, and Job Design’. Journal of Law, Economics, & Organization 7 (Special Issue): 24–52.

Ivaldi, M., N. Petit, and S. Ünekba?. 2025. ‘Killer Acquisitions: Evidence from European Merger Cases’. Antitrust Law Journal 86 (3): 647–695.

Klein, B. 1996. ‘Why Hold-Ups Occur: The Self-Enforcing Range of Contractual Relationships’. Economic Inquiry 34 (3): 444–463.

Klemperer, P. 1987. ‘The Competitiveness of Markets with Switching Costs’. RAND Journal of Economics 18 (1): 138–150.

Lafontaine, F., and M. E. Slade. 2007. ‘Vertical Integration and Firm Boundaries: The Evidence’. Journal of Economic Literature 45 (3): 629–685.

Liebowitz, S. J., and S. E. Margolis. 1999. Winners, Losers & Microsoft: Competition and Antitrust in High Technology. Independent Institute.

Manne, G. A. 2020. ‘Error Costs in Digital Markets’. In The Global Antitrust Institute Report on the Digital Economy.

Manne, G. A., and D. Auer. 2024. From Data Myths to Data Reality. International Center for Law & Economics.

Manne, G. A., D. Auer, L. Radic, and M. Zúñiga. 2024. Comments of the International Center for Law & Economics on the CMA’s Draft Guidance for the UK’s Digital Markets Competition Regime. International Center for Law & Economics.

Manne, G. A., S. Bowman, and D. Auer. 2022. ‘Technology Mergers and the Market for Corporate Control’. Missouri Law Review 86 (4).

Mohan, S. 2014. ‘Out-of-Market Efficiencies in EU Merger Control’. Commentary.

Padilla, J. 2019. ‘The Treatment of Efficiencies and Harms in Merger Control’. Commentary.

Rinz, K. 2022. ‘Labor Market Concentration, Earnings and Inequality’. Journal of Human Resources 57 (S): 251–283.

Rochet, J.-C., and J. Tirole. 2003. ‘Platform Competition in Two-Sided Markets’. Journal of the European Economic Association 1 (4): 990–1029.

Rochet, J.-C., and J. Tirole. 2006. ‘Two-Sided Markets: A Progress Report’. RAND Journal of Economics 37 (3): 645–667.

Sidak, J. G., and D. J. Teece. 2009. ‘Dynamic Competition in Antitrust Law’. Journal of Competition Law & Economics 5 (4): 581–631.

Syverson, C. 2011. ‘What Determines Productivity?’ Journal of Economic Literature 49 (2): 326–365.

Syverson, C. 2019. ‘Macroeconomics and Market Power: Context, Implications, and Open Questions’. Journal of Economic Perspectives 33 (3): 23–43.

Teece, D. J. 1986. ‘Profiting from Technological Innovation’. Research Policy 15 (6): 285–305.

Teece, D. J., G. Pisano, and A. Shuen. 1997. ‘Dynamic Capabilities and Strategic Management’. Strategic Management Journal 18 (7): 509–533.

Ünekba?, S. 2026. ‘The EU’s Draft Merger Guidelines and the “Innovation Shield”: Never Judge a Shield by Its Cover?’ Kluwer Competition Law Blog, May 8.

Utterback, J. M., and W. J. Abernathy. 1975. ‘A Dynamic Model of Process and Product Innovation’. Omega 3 (6): 639–656.

Valickova, P., T. Havranek, and R. Horvath. 2015. ‘Financial Development and Economic Growth: A Meta-Analysis’. Journal of Economic Surveys 29 (3): 506–526.

Valletti, T., and M. Broso. 2024. ‘Mergers, Lobbying, and Elections: Is There a “Curse of Bigness”?’ Working paper.

Weyl, E. G. 2010. ‘A Price Theory of Multi-Sided Platforms’. American Economic Review 100 (4): 1642–1672.

Williamson, O. E. 1985. The Economic Institutions of Capitalism. Free Press.

Wright, J. D., and D. H. Ginsburg. 2013. ‘The Goals of Antitrust: Welfare Trumps Choice’. Fordham Law Review 81 (5): 2405–2423.