Journal of European Competition Law & Practice, 2024, 1–22

https://doi.org/10.1093/jeclap/lpae033
Article

How to Fix a Failing Art. 102 TFEU: Substantive

Interpretation, Evidentiary Requirements, and the
Commission’s Future Guidelines on Exclusionary

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

Abuses
Heike Schweitzer, Simon de Ridder
Professor Dr. Heike Schweitzer, LL.M. (Yale) holds a Chair for Private Law and Competition Law and Economics at Humboldt-Universität zu Berlin. Simon de Ridder
is doctoral researcher at the Chair. For their comments on an earlier draft of this paper, we are grateful to Elias Deutscher, Andreas Engert, Lennart Enwaldt,
Johannes Holzwarth, Philipp Hornung, Pablo Ibáñez Colomo, Thorsten Käseberg, Robert Welker, and Wouter Wils. For valuable support in conducting the
empirical analysis, we thank Maximilian Wolters. Of course, all opinions and remaining mistakes are ours.

Key Points

• The duration of adversarial Art. 102 proceedings before the Commission has significantly increased in recent years, such that
an effective deterrence and protection competition is no longer ensured.
• The case law on Art. 102 TFEU is partially unclear and incoherent; future guidelines should offer a conceptually coherent
framework and address the dysfunctionalities of public enforcement.
• In support of the endeavour, we propose an interpretation of Art. 102 TFEU that we consider to be in line with the case law
and with a ‘workable effects-based approach’.
• A speedier enforcement presupposes a clarification of the law of evidence, in particular with a view to proving potential
anticompetitive effects; in this vein, we identify settings in which potential anticompetitive effects may be inferred, or in
which a ‘quick look’ may suffice.

I. The ‘Article 102 package’—aims of future exclusionary conduct shall consolidate the Union Courts’ case law
guidelines on exclusionary abuses on Art. 102 TFEU, including the 32 judgments on exclusionary con-
duct that have been passed since 2008; provide companies with
On 27 March 2023, the Commission has published its ‘Article 102
greater legal certainty; and enable a consistent enforcement by
package’: in an ‘Amending Communication’,1 it has modified its
the Commission, national competition authorities, and national
so-called ‘2008 Priorities Guidance’2 and officially distanced itself
Courts.
from some of the core principles that previously characterised the
In fact, the project is even more ambitious. A Policy Brief4
‘more economic approach’ to the enforcement of Art. 102 TFEU.
that DG Competition has published alongside the ‘Article 102
No less importantly, the Commission has announced its inten-
package’ reveals the broader agenda: the initiative strives to take
tion to adopt guidelines on exclusionary abuses in the course
a new look at exclusionary abuses in the light of 15 years of
of 2025. A ‘Call for Evidence’3 has initiated the consultation
experience gained in the enforcement of Art. 102 TFEU based
which shall lead up to the publication of a first draft in mid-
on a ‘Priorities Guidance’ that has proven to be—at least par-
2024. According to the ‘Call for Evidence’, future guidelines on
tially—dysfunctional. In its future guidelines, the Commission
aims to develop ‘A dynamic and workable effects-based approach
1 Communication from the Commission: Amendments to the Communica-
tion from the Commission—Guidance on the Commission’s enforcement pri-
to abuse of dominance’, i.e. an interpretation of Art. 102 TFEU that
orities in applying Article 82 of the EC Treaty to abusive exclusionary conduct allows for a speedier enforcement with a realistic expenditure of
by dominant undertakings [2023] OJ C116/1 (‘Amending Communication’).
2
resources. The outstanding importance of an effective enforce-
Communication from the Commission: Guidance on the Commission’s
enforcement priorities in applying Article 82 of the EC Treaty to abusive exclu- ment ‘in economically difficult times and in view of the increasing
sionary conduct by dominant undertakings [2009] OJ C 45/7 (‘2008 Priorities
Guidance’).
3 EU Commission, ‘EU Competition Law—guidelines on exclusionary
abuses by dominant undertakings: Call for evidence’, <https://ec.europa.eu/
info/law/better-regulation/have-your-say/initiatives/13796-EU-competition- 4 Linsey McCallum and others, ‘A dynamic and workable effects-based
law-guidelines-on-exclusionary-abuses-by-dominant-undertakings_ approach to abuse of dominance’, European Commission, Competition policy
en > accessed 31 March 2024. brief No 1/2023 (‘Policy Brief’).

Received: May 1, 2024. Accepted: May 3, 2024

© The Author(s) 2024. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which
permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
2 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

market concentration in various sectors’ is also emphasised in the adversarial proceedings and all types of decisions in cooperative
‘Amending Communication’. proceedings (i.e. commitment decisions or prohibition decisions
Indeed, rethinking Art. 102 TFEU in the light of the principle of following a cooperative proceeding).
effectiveness is urgently needed. In part II of this paper, we show As illustrated by Fig. 1, the overall level of enforcement of Art.
how the duration of Art. 102 proceedings before the Commission 102 TFEU remains relatively constant between 1996 and 2019. In
has significantly increased over time, such that Art. 102 enforce- this period, the annual number of decisions roughly f luctuates
ment ultimately risks to become ineffective in protecting undis- around 2.5. At a closer look, two trends are noticeable: first,
torted competition. In part III, we brief ly summarise important after the entry into force, in May 2004, of Regulation 1/2003,6
cornerstones of the substantive interpretation of Art. 102 TFEU. the number of decisions in adversarial proceedings has dropped
Where the case law seems unclear or incoherent, we propose a significantly, while the number of decisions in cooperative pro-

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

reading that we consider to be in line with a ‘workable effects- ceedings has risen. This ref lects a change in the Commission’s
based approach’. In part IV, we focus on the law of evidence: if enforcement practice in immediate reaction to the introduction
the current handling of the effects analysis is (partially) responsi- of commitment decisions in Art. 9 of Regulation 1/20037 as a for-
ble for the prolongation of Art. 102 proceedings, settings should malised cooperative tool and thus an alternative to the ‘normal’
be explored in which potential anticompetitive effects may be enforcement procedure under Art. 7 of Regulation 1/2003.8
inferred. Part V brief ly concludes. Secondly, the most recent period from 2020 to 2023 is partic-
ularly interesting. After an enforcement high from 2008 to 2019
(between 2.5 and 3 decisions per year compared with between
II. Dysfunctionalities in enforcing Article 2 and 2.5 decisions per year from 1996 to 2007), the numbers
102 TFEU have dropped significantly: from 2020 to 2023, the Commission
In its Policy Brief, DG Comp rightly highlights the need for a only adopted one decision per year, on average. Further, in this
handling of Art. 102 TFEU that allows for an effective, manageable period, the Commission did not adopt a single prohibition decision
enforcement. This suggests a brief analysis of whether, and if so but commitment decisions only. This comes after a period from
how and why, an effective enforcement is currently failing. In 2008 to 2019 in which the number of decisions in adversarial
the EU, Art. 102 TFEU has been enforced persistently over the proceedings had risen.
years. Nonetheless, the development of the duration of Art. 102
proceedings raises concerns. Furthermore, public enforcement 2. The duration of proceedings
of Art. 102 TFEU at the EU level is currently at a low, but it is The duration of Art. 102 proceedings is an important parameter
too early to tell whether this is an actual trend or just a short- for assessing the public enforcement’s success. Where enforce-
term anomaly (A.). Regarding the duration of Art. 102 proceedings, ment proceedings become excessively long, abusive conduct may
various factors have contributed to this development, including, not be ceased in a timely fashion, but continue to unleash its
importantly, particularities and complexities of the enforcement exclusionary potential until the competitive structure is harmed
procedure. In our analysis, we inquire whether and how the inter- for good. As a result, and apart from purely allocational ineffi-
pretation of Art. 102 TFEU and the evolution of the law on evidence ciencies, Art. 102 TFEU may lose its deterrent effect—and thereby
may have contributed to prolonging abuse proceedings, and how an important part of its preventive effectiveness.9 Public enforce-
future guidelines may help in promoting a more ‘workable effects- ment notwithstanding, dominant positions may be reinforced or
based approach’ (B.). even expand, and competing innovation paths may be perma-
nently altered.
A. Current shortcomings in numbers
We examine existing dysfunctionalities in the Commission’s 2.1. The development of the duration of proceedings over
enforcement of Art. 102 TFEU by empirically analysing the time
Commission’s decisions relating to proceedings under Art. 102 Fig. 2 shows the median duration of the Commission’s Art. 102
TFEU (or the previous Treaty provisions on abuse of dominance) proceedings sorted by the same eight periods as above.10 We
since 1992. The dataset we use consists of a list of the published measure the duration of the proceeding in months.11 The end
decisions relating to a proceeding under Art. 102 TFEU between of the proceeding is the date of the decision. The starting point
1992 and 20235 and features self-collected data that were taken of the proceeding is the earliest (formal or informal) procedural
from these decisions. We are recording for each decision: the
name of the decision, the year the decision was issued, the 6 Council Regulation (EC) No 1/2003 of 6 December 2002 on the implemen-
duration of the proceeding in months, the length of the decision tation of the rules of competition laid down in Articles 81 and 82 of the Treaty
[2003] OJ L1/1.
in pages, the type of the decision (e.g. commitment decision), 7 For an instructive overview of the commitment decision practice see
and what triggered the proceeding (e.g. a complaint lodged by a Niamh Dunne, ‘Commitment Decisions in EU Competition Law’ (2014) 10
competitor). Journal of Competition Law & Economics 399, 406. The main categories of
commitment decisions in Art. 102 cases are decisions in the energy sector, in
technological markets and in financial markets.
1. The overall enforcement activity 8 Regulation 17/62 (First Regulation implementing Articles 85 and 86 of the
Treaty [1962] OJ 13/204) did not provide for a comparable legal framework for
Since 1992, our dataset includes 67 Commission decisions relating cooperative enforcement.
to a proceeding under Art. 102 TFEU. For 32 years, that amounts 9 On the importance of considering deterrence effects when choosing
optimal legal standards see, for example, Yannis Katsoulacos and David Ulph,
to a mean of 2.1 decisions per year. The following figure shows ‘On Optimal Legal Standards for Competition Policy: A General Welfare-
the mean number of decisions issued by the Commission per Based Analysis’ (2009) 57 Journal of Industrial Economics 410; Jan Broulík,
year, sorted by eight periods of four years each. It also shows the ‘Preventing Anticompetitive Conduct Directly and Indirectly: Accuracy Versus
Predictability’ (2019) 64 Antitrust Bulletin 115, 118.
difference in the type of decision between prohibition decisions in 10 Given the sometimes small numbers of decisions in these respective
periods, the findings have to be interpreted carefully and any conclusion can
only be tentative.
5 We exclude negative clearance decisions due to their special legal con- 11 For reasons of practicality, we count both the month in which the starting
tent, which hinders an effective comparison with other decisions. point occurs and the month in which the ending point (i.e. decision) occurs.
 Journal of European Competition Law & Practice, 2024 | 3

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

Fig. 1. Mean number of Commission decisions per year differentiated between adversarial and cooperative proceedings.

Fig. 2. Median duration of proceedings regarding all Commission decisions, in months.

step mentioned in the decision. This may be a formal step such the first half of the period under review, the median duration of
as the Commission’s decision to initiate proceedings. However, it proceeding was 50 months; in the second half, it was 46 months.
may also be the filing of a complaint with the Commission or an In the periods from 2008 to 2011 and 2020 to 2023, with a
inspection at the office of an undertaking under investigation. In particularly high share in cooperative enforcement, the duration
defining the starting point of the proceeding this way, we want to of proceedings was below average. However, given that coopera-
ensure that our measurement of the duration does not depend tive proceedings have very much dominated Art. 102 enforcement
on the Commission’s discretion to formally open proceedings in a since 2005, the fact that the duration of proceedings across all
given case.12 decisions only decreases by four months after 2004 is remarkable.
Fig. 2 does not suggest any particular trend regarding the dura- Taking a closer look at the median duration of adversarial
tion of proceedings, but rather a remarkable degree of stability: in proceedings versus cooperative proceedings in the period between
2004 and 201913 (Fig. 3), two findings strike the eye:
First, and hardly surprising, the median duration of adversarial
12 This methodology is certainly not flawless. For example, it can lead proceedings is longer than the median duration of cooperative
to certain distortions in comparing ex officio investigations with proceedings
initiated by a complaint: only for the latter will we be able to observe the actual proceedings—at first sight, commitment decisions seem to deliver
point in time at which the Commission first gained relevant information about on the promise. The size of the margin varies, however. While in
the alleged infringement. By contrast, any informal procedural step by the
Commission depends on the Commission’s discretion, as well. Consequently, two periods, the margin is considerable (36 months (2008–2011)
ex officio proceedings may still seem shorter in our dataset than they actually
are. However, with a view to illustrating the problem of excessively long
proceedings, we consider that our methodology of measuring their duration 13 This is the only period in which this differentiation can be made in a
is ‘good enough’. useful way, as can be seen in Figure 1.
4 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

Fig. 3. Median duration of proceeding differentiating between adversarial and cooperative proceedings (between 2004 and 2019), in months.

and 53 months (2016–2019), respectively), it seems rather small until 2007,17 while ten have been issued since 2008.18 The two
in the other two periods (five months (2004–2007) and one month last decisions in adversarial proceedings in our period under
(2012–2015), respectively).14 review, Google Search (AdSense) and Qualcomm (predation), are also
Secondly, while the duration of cooperative proceedings has those with the longest proceedings of all Commission decisions
decreased from 2004–2007 onwards,15 adversarial proceedings are reviewed, with 111 months (or 9.3 years) and 122 months (or
getting longer: from 2004 to 2007, their median duration was 10.2 years), respectively.
61 months; from 2016 to 2019, it was 88 months.16 The period
from 2012 to 2015, where the median duration of adversarial B. Possible reasons and the role of guidelines
proceedings was rather short, appears to be just an exceptional The increase in the duration of adversarial proceedings and of the
outlier that may be attributed to the particularities of a very small increasing number of excessively long proceedings calls for a reac-
number of cases: between 2012 and 2015, our dataset includes tion: where Art. 102 proceedings last six, eight, or even ten years,
only four decisions in adversarial proceedings. the Commission’s decision will typically be unable to remedy the
resulting harm to the competitive structure, and the enforcement
2.2. The occurrence and distribution of exceptionally loses any deterrent effect. In highly dynamic sectors—like the
long proceedings digital sector—the dominant undertakings may, by their conduct,
The development of the median duration of Art. 102 proceedings have created long-lasting barriers to market entry, and may have
over time is not the only factor to be analysed. Additionally, a look inf luenced for good the direction of investment and innovation.
at the spread in the duration of proceedings and the distribution Even if one were to presume—which is not at all clear—that
of particularly long proceedings over time is called for. While longer proceedings lead to more accurate outcomes and thereby
there may always be outliers, a recurrence of excessively long reduce error costs in individual cases, the enforcement regime’s
proceedings may indicate problems in Art. 102 enforcement in ‘systemic’ error costs increase as decisions that come too late to
and of itself. For example, looking at our dataset, 16 decisions effectively remedy an abuse are ultimately analogous to ‘false
have followed proceedings that were longer than 72 months (or negatives’. Also, if we assume a positive correlation between
six years). Out of these 16 decisions, only six have been issued long proceedings and administrative costs, excessively long pro-
ceedings may result in a lower level of overall competition law
enforcement.

17 Van den Bergh Foods Limited (Cases IV/34.073, IV/34.395 and IV/35.436)
Commission Decision of 11 March 1998; Virgin/British Airways (Case IV/D-
14 Absent a more in-depth analysis of the reasons for this small margin, 2/34.780) Commission Decision of 14 July 1999; Deutsche Post AG (Case
no policy conclusions can be drawn from this finding, at least not across all COMP/35.141) Commission Decision of 20 March 2001; DSD (Case COMP
decisions. Of course, Figure 3 appears to confirm that cooperative enforcement D3/34493) Commission Decision of 20 April 2001; AstraZeneca (Case COM-
does lead to shorter proceedings than adversarial enforcement. This finding P/A.37.507/F3) Commission Decision of 15 June 2005 and Coca-Cola (Case COM-
is, however, subject to the above-stated incapability of our methodology to P/A.39.116/B2) Commission Decision of 22 June 2005.
measure the duration of a proceeding reliably given that, before the formal ini- 18 Intel (Case COMP/C-3/37.990) Commission Decision of 13 May 2009;
tiation of proceedings, it lies within the discretion of the Commission whether RAMBUS (Case COMP/38.636) Commission Decision of 9 December 2009; Rio
it will mention, in a later decision, any prior informal steps, see above n 12. Tinto Alcan (Case COMP/39.230) Commission Decision of 20 December 2012;
Many commitment decisions are silent about these informal procedural steps. Slovak Telekom (Case AT.39523) Commission Decision of 15 October 2014; ARA
15 Cooperative proceedings lasted for a median 56 months in 2004 to 2007 Foreclosure (Case AT.39759) Commission Decision of 20 September 2016; Google
and from then on were considerably shorter (33 months (2008–2011), 47 months Search (Shopping) (Case AT.39740) Commission Decision of 27 June 2017; Baltic
(2012–2015), and 35 months (2016–2019), respectively). Rail (Case AT.39813) Commission Decision of 2 October 2017; BEH Gas (Case
16 A selection bias may result from the differentiation between cooperative AT.39849) Commission Decision of 17 December 2018; Google Search (AdSense)
and adversarial proceedings, if the more complex cases were typically be dealt (Case AT.40411) Commission Decision of 20 March 2019, and Qualcomm (preda-
with in an adversarial setting. Nonetheless, the trend remains problematic. tion) (Case AT.39711) Commission Decision of 18 July 2019.
 Journal of European Competition Law & Practice, 2024 | 5

1. Reasons for exceptionally long proceedings assessment compared with the Commission’s decisions in the
An adequate policy reaction to the trend towards longer Art. 102 1990s.27 While a quantitative effects analysis will not necessarily
enforcement proceedings presupposes a sound understanding prolong proceedings,28 it may become a complex and time-
of the causes. Obviously, these causes can be manifold—both consuming exercise at times. The Commission’s self-commitment
across all cases and in each individual case. First, complexities to conduct an as-efficient-competitor (AEC) test even in rebate
of procedural law and/or practice may play a role,19 just like and exclusive dealing cases29 —for as long as it lasted—was an
procedural strategies—both by the Commission and by the under- additional factor that contributed to prolongations. But even the
taking under investigation.20 Regulation 1/2003 is currently being qualitative effects analysis has become much more elaborate
evaluated.21 A future reform may try to address some of the and complex,30 not least in light of the Union Courts’ request for
procedural causes. a concrete and context-specific analysis.

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

Secondly, the type of abuse cases may have changed over time.
2. The role and limits of guidelines in addressing the
For some time, Art. 102 TFEU cases primarily addressed strategies
by former state monopolies to defend their monopolistic position
inefficiencies?
in liberalised markets. More recently, an increasing number of With its 2023 Amending Communication, the Commission has
Art. 102 TFEU cases deals with the conduct of firms that have partially withdrawn the 2008 Priorities Guidance’s costly policy
gained their dominant position based on superior performance choices, in particular the need to show direct consumer harm
and innovative strategies. In these cases, Art. 102 enforcement and the need to conduct an AEC test when analysing exclusiv-
must carefully consider the effects that competition law interven- ity rebates. However, over the last years the Union Courts have
tion may have on undertakings’ incentives to invest and innovate, developed their own variety of an ‘effects-based approach’. Hence,
which may complicate the distinction between pro-competitive one may question whether and to what extent there remains
and anticompetitive conduct. In this regard, it does not come as room for future guidelines to contribute to a speedier, and thereby
a surprise that out of the ten Art. 102 proceedings the duration more effective enforcement of Art. 102 TFEU. While Commission
of which stretched over six years and more, two proceedings guidelines can redefine competition policy, they cannot change
concerned Google,22 and three proceedings concerned technology competition law.31
and Chip markets.23 Nonetheless, another three proceedings dealt At the same time, one should not underestimate the inf luence
with old style natural monopolies.24 guidelines may have on the evolution of competition law.32 In
Thirdly—and this is where our focus lies from now on—the competition law, the Union Courts rule on the legality of Com-
increase in the duration of Art. 102 TFEU proceedings from mission decisions in individual cases (Art. 263 TFEU), or they rule
2004 to 2007 coincided with the Commission’s endeavour to on the interpretation of the competition rules in the context of
develop a ‘more economic approach’ to Art. 102 TFEU which
ultimately resulted in the publication of the 2008 Priorities
choice if they had also been offered the product of their favourite OEM and/or
Guidance. With the 2008 Priorities Guidance, the Commission retailer with x86 CPUs from Intel’s competitors.’
committed to engage in a detailed, case- and context-specific 27 In the 1990s, the room in the Commission decisions dedicated to the
effects analysis of likely foreclosure in essentially all cases.25 effects analysis was quite limited. Usually, the Commission showed the anti-
competitive effect purely qualitatively, cf. e.g. British Midland v. Aer Lingus
Although the Commission quickly gave up the attempt to (Case IV/33.544) Commission Decision of 26 February 1992, recitals 26 ff;
offer direct proof of consumer harm and simply deduced it Warner-Lambert/Gillette and Others and BIC/Gillette and Others (Cases IV/33.440
and IV/33.486) Commission Decision of 10 November 1992, recitals 25 ff; Irish
from a finding of an otherwise anticompetitive foreclosure Sugar plc (Cases IV/34.621 and 35.059/F-3) Commission Decision of 14 May
instead,26 the Commission’s effects analysis shifted towards 1997, recitals 114, 125, 135; Amministrazione Autonoma dei Monopoli di Stato (Case
IV/36.010-F3) Commission Decision of 17 June 1998, recitals 35, 37; Ilmailu-
a significantly more extensive and often more quantitative laitos/Luftfartsverket (Case IV/35.767) Commission Decision of 10 February 1999,
recitals 42 ff.
28 For example, in the Telekomunikacja Polska case, the Commission based
19 See, for example, Johannes Laitenberger and James Kröger, ‘Towards an its quantitative analysis on data that it had received from the dominant
“even more efficient” approach?’ [2023] Zeitschrift für Europäisches Privatrecht undertaking and the Polish telecommunications authority (UKE). The analysis
621, 628 ff. of the conduct’s effects on the competitive structure apparently remained
20 Procedural (law) reasons for long proceedings may include priority set- manageable—Cf. Telekomunikacja Polska (Case COMP/39.525) Commission Deci-
ting by the Commission (e.g. in Qualcomm (predation) (n 18)), high procedural sion of 22 June 2011, recitals 820 ff.
standards (frequently related to protecting the defendant’s rights to be heard 29 See on this complex rendition of the AEC test below, III. B. 4. 1.
and to access to the file) or overlong and ultimately fruitless commitment nego- 30 For example, in Baltic Rail, the Commission’s effects analysis spans 30
tiations (e.g. in Google Search (Shopping) (n 18), and Google Search (AdSense) (n 18)).
pages (almost a third of the entire length of the decision)—see Baltic Rail (n 18),
Procedural instruments that might shorten proceedings—interim measures in
pp 65–94. See on the effects analysis in this case also below, IV. C. 2. 2.
particular—are rarely used; for their particular potential, see Massimiliano
31 For the Union Courts’ conclusive competence to interpret EU law see
Kadar, ‘The Use of Interim Measures and Commitments in the European
Commission’s Broadcom Case’ (2021) 12 Journal of European Competition Law Art. 19 TEU, 267 TFEU. Furthermore, for the non-binding nature of Commis-
& Practice 443, 451. sion guidelines see Case C-376/20 P Commission v CK Telecoms UK Investments,
21 EU:C:2023:561, para 125. See also the European Parliament’s Report on institu-
See on this, Massimiliano Kadar, ‘Evaluating 20 Years of Regulation
tional and legal implications of the use of ‘soft law’ instruments (<https://www.
1/2003: Are EU Antitrust Procedures “Fit for the Digital Age”?’ (2024) 85 Antitrust
europarl.europa.eu/doceo/document/A-6-2007-0259_EN.html> accessed 31
Law Journal 577; Laitenberger and Kröger (n 19); and on the experience with
March 2024), cautioning against implicit legislation by means of, inter alia,
Regulation 1/2003, Wouter PJ Wils, ‘Regulation 1/2003: An Assessment After
interpretative communications and an extrapolation of the case law of the
Twenty Years’ (2023) 46 World Competition 3. The experience under national
Court of Justice ‘into unchartered territory’.
procedural law suggests that the enforcement of Art. 102 TFEU might be
32 Which degree of discretion the Commission possesses in drafting guide-
substantially facilitated if a special procedure were introduced for enforcing
the abuse prohibition without fines. lines remains controversial. In favour of a broad degree of discretion—even of
22 the admissibility of deviations from the case law: Pınar Akman, ‘The Euro-
Google Search (Shopping) (n 18); Google Search (AdSense) (n 18).
23
pean Commission’s Guidance on Art. 102 TFEU: From Inferno to Paradiso?’
Intel (n 18); RAMBUS (n 18); Qualcomm (predation) (n 18). (2010) 73 Modern Law Review 605, 626 f.; Rohner (n 26) 274. For a more
24 Slovak Telekom (n 18); Baltic Rail (n 18); BEH Gas (n 18). narrow construction: Håkon A Cosma and Richard Whish, ‘Soft Law in the
25 2008 Priorities Guidance, para 20. Field of EU Competition Policy’ (2003) 14 European Business Law Review
26 Tristan Rohner, Art. 102 AEUV und die Rolle der Ökonomie (Nomos 2023) 25, 52; Liza Lovdahl Gormsen, A Principled Approach to Abuse of Dominance in
220. For example, in Intel, after 151 pages of AEC test, the Commission finds European Competition Law (CUP 2010) 158; Liza Lovdahl Gormsen, ‘Why the
a reduction of consumer choice on not more than six pages—see Commission European Commission’s Enforcement Priorities on Article 82 EC should be
Decision in Intel (n 18), pp 459–464. The argument is summarised in recital 1603: Withdrawn’ (2010) 31 European Competition Law Review 45, 50; Oana Andreea
Intel’s conduct ‘excluded, limited, or delayed AMD x86 CPUs in the market. Ştefan, ‘European Competition Soft Law in European Courts: A Matter of Hard
As such, Intel’s exclusionary practices had a direct and immediate negative Principles?’ (2008) 14 European Law Journal 753, 764; see also Opinion of AG
impact on those customers who would have had a wider price and quality Kokott in Case C-95/04 P British Airways v Commission, EU:C:2006:133, para 28.
6 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

referral proceedings (Art. 267 TFEU). Even in referral proceedings, suggested likely anticompetitive foreclosure with ‘an adverse
the questions raised by national Courts must be accompanied by impact on consumer welfare’ to be identified based on ‘qualitative
a summary of the context in which the question arises. Whatever and, where possible and appropriate, quantitative evidence’.40
the type of proceeding, the Courts’ judgments must therefore At first sight, the ECJ’s determination in Servizio Elettrico
be contextualised. This is also true for the seemingly broad and Nazionale that the ‘ultimate objective’ of Art. 102 TFEU is the ‘well-
general statements on the interpretation of Art. 101 or Art. 102 being of consumers—both intermediate and final consumers’41
TFEU that judgments often entail: In any given case, they are seems to follow that logic. However and importantly, the sentence
meant to provide reasoning for the concrete decision or the is embedded in a broader explanation of the objectives and the
answer to a question referred. By implication, they, too, must be structure of Art. 102 TFEU: Art. 102 TFEU is characterised as one
contextualised. of a series of rules ‘intended to prevent competition from being

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

Against this background, guidelines are a useful instrument to distorted contrary to the public interest and to the detriment of
propose a coherent and comprehensive interpretation of Art. 102 individual undertakings and consumers, and thus to contribute to
TFEU, to identify the generalisable substantive and evidentiary economic welfare in the European Union’. Moreover, the ECJ refers
principles, to separate them from more context-specific findings to the established case law that Art. 102 TFEU ‘seeks to sanction
and to fill conceptual gaps. The effort to systematise and concep- not only practices likely to cause direct harm to consumers but
tualise the Courts’ jurisprudence in this manner, and also to adapt also those which cause them harm indirectly by undermining
the interpretation of Art. 102 TFEU to changing market realities,33 an effective structure of competition’.42 From this passage,
may be considered a core part of the Commission’s responsibility it follows that harm to consumer welfare—uncontroversially
for the implementation and orientation of competition policy.34 broadly construed43 —is an objective, but not a precondition for
In the following text, we try to identify fundamental substantive finding an abuse. The ECJ seems to have rephrased what Richard
(III.) and evidentiary principles (IV.) of the law on abuse. Posner has observed for US antitrust law quite some time ago:
‘Efficiency is the ultimate goal of antitrust, but competition a
mediate goal that will often be close enough to the ultimate goal
III. Abuse of dominance—conceptual to allow the Courts to look no further’.44 It is only at the stage
cornerstones of the substantive law on of the ‘efficiency defence’ that the dominant undertaking may
abuse invoke an absence of consumer harm;45 and in fact, it is (only)
A. The goals of Article 102 TFEU the efficiency defence that requires a formal acknowledgment
Obviously, future guidelines on exclusionary abuses will have to of consumer welfare within the system of EU competition law’s
start with a restatement of the goals of Art. 102 TFEU. In doing goals.
so, they will very likely not dive into old battles. The drafting of It follows that, Servizio Elettrico Nazionale notwithstanding, con-
the 2008 Priorities Guidance had been accompanied by an intense duct may be considered to be prima facie abusive even where proof
discussion about the objectives of European competition law in of consumer harm may be difficult: Servizio Elettrico Nazionale
general and Art. 102 TFEU in particular.35 Proponents of a ‘more was hardly intended to overrule the case law on a dominant
economic approach’ aimed to settle on the consumer welfare goal undertaking’s refusals to supply wholesalers engaging in paral-
as a clear and measurable reference point for EU competition lel exports, for example.46 Conduct that clashes with the EU’s
law.36 According to this approach’s opponents, EU competition overarching market integration goal47 will arguably continue to
law was to hold on to the broader goal of protecting open markets be outlawed, with particularly high demands on an efficiency
and the competitive process37 and to be guided by the overarching defence in view of the normative weight of the EU’s internal
objective of the EU to establish and protect a well-functioning market policy.
internal market with undistorted competition.38 It is a separate question whether EU competition law addi-
The 2008 Priorities Guidance did not address the dispute on a tionally pursues broader political and societal aims. According to
theoretical level, but it let the Commission’s enforcement agenda some,48 this is what the Commission’s Amending Communication
revolve around a welfare maximisation metric: (only) those cases suggests when it states that the competition rules can contribute
were to be prioritised in which ‘cogent and convincing evidence’39 to ‘achieving objectives that go beyond consumer welfare, such as

33 40 2008 Priorities Guidance, para 19.

See Massimiliano Kadar, Johannes Holzwarth and Virgilio Pereira, ‘Abuse
41 Servizio Elettrico Nazionale (n 38), para 46. Arguably, the term ‘well-being
of dominance under Article 102 TFEU: a survey on 2023’ (2024) 15 Journal
of European Competition Law & Practice (published online 6 May 2024), p 2, of consumers’ is used synonymously with the economic concept of consumer
pointing to increasingly concentrated sectors and the role of digital markets for welfare in this passage. This is also suggested by the fact that the aim of
the EU economy, and suggesting that these changes have made false negatives protecting ‘consumer well-being’ is to justify the ‘efficiency defence’.
in the enforcement of Art. 102 TFEU ‘both more likely and more costly’. 42 Servizio Elettrico Nazionale (n 38), para 44 (and 47) with further references
34 For this, see Case C-234/89 Delimitis, EU:C:1991:91, para 44. to prior case law. See also: Fernando Castillo de la Torre, ‘Is the Effects-Based
35 For a summary see Anne C Witt, The More Economic Approach to EU Approach Too Cumbersome?’, in Adina Claici, Assimakis Komninos and Denis
Antitrust Law (Hart 2016) ch 4 with further references to the debate. See also Waelbroeck (eds) The Transformation of EU Competition Law (Wolters Kluwer 2023)
the contributions in Daniel Zimmer (ed) The Goals of Competition Law (Edward 146.
43 The concept of consumer welfare does not refer to, and must not be
Elgar 2012).
36 See, inter alia, Okeoghene Odudu, ‘The wider Concerns of Competition measured in terms of, price only, but also encompasses choice, quality, and
law’ (2010) 30 Oxford Journal of Legal Studies 559, 605–612; Pınar Akman, The innovation—see Servizio Elettrico Nazionale (n 38), para 46; C-307/18 Generics (UK)
Concept of Abuse in EU Competition Law: Law and Economic Approaches (Hart 2012) and Others, EU:C:2020:52, para 165.
44 Richard A Posner, Antitrust Law (2nd edn, University of Chicago Press
49–106.
37 See, inter alia, Adrian Künzler, Effizienz oder Wettbewerbsfreiheit? (Mohr 2001) 29.
45 Against this background, the broader wording in the Amending Com-
Siebeck 2008) 323 ff. See also: Heike Schweitzer, ‘Efficiency, political freedom
and the freedom to compete’, in Daniel Zimmer (ed) The Goals of Competition munication, para 1 appears appropriate: the competition rules contribute to ‘a
Law (Edward Elgar 2012) 169 ff. level playing field where markets serve consumers’.
38 46 See Joined Cases C-468/06 to C-478/06 Sot. Lélos kai Sia and Others,
See Art. 3 para 1 lit b TFEU and Protocol No 27 on the internal market
and competition, according to which the internal market includes a system EU:C:2008:504.
ensuring that competition is not distorted. See also, recently, Case C-377/20 47 Even Servizio Elettrico Nazionale (n 38) explicitly acknowledges the rele-
Servizio Elettrico Nazionale and Others, EU:C:2022:379, paras 42–43. vance of this goal in paras 42–43.
39 2008 Priorities Guidance, para 20. 48 See e.g. Van Bael & Bellis on Competition Law, March 2023, p 6.
 Journal of European Competition Law & Practice, 2024 | 7

plurality in a democratic society’.49 The Policy Brief further refers competition authority may still, in a first step, rely on a relatively
to the Union Courts’ case law—in particular to Google Android50 — abstract analytical template to provisionally infer an abuse. But if,
for its claim that ‘ensuring consumer choice is a means to ulti- based on supporting evidence, the dominant undertaking submits
mately guarantee plurality in a democratic society’. In addition, that its conduct was not capable of restricting competition, the
Commissioner Vestager’s frequent references to goals such as competition authority must ‘examine whether, in the particular
‘fairness and level-playing field, market integration, preserving circumstances, the conduct in question was indeed capable of
competitive process, consumer welfare, efficiency and innovation, doing so’.61 It is this requirement to show potential anticompet-
and ultimately plurality and democracy’ are quoted.51 itive effects in each single case and with a substantial degree of
As Massimiliano Kadar and Johannes Holzwarth have recently context-specificity (at least where they are contested during the
clarified,52 these references should not be read as an attempt administrative proceeding with a sufficient degree of substanti-

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

to expand the objectives of Art. 102 TFEU. Rather, they amount ation—on this see below, IV. B. 2.) which is at the heart of the
to a recognition of interdependencies in a unitary legal order: Union Courts’ version of an effects-based approach: the judicial
an effective enforcement of competition law can—for example authorisation to presume anticompetitive effects based on certain
and of course—function as a protective barrier against excessive types of conduct alone has been withdrawn.
concentration and the redistributive effects associated with it. To From an enforcement perspective, it is difficult to overstate
this day, the European debate has not joined the Neo-Brandeisian the relevance of this shift. Nonetheless, the case law is open to
chorus across the Atlantic that aims for a formal expansion of different interpretations. On the one hand, it might imply that,
goals,53 nor should it. Even more clearly, it cannot be the task of at least in adversarial settings, an abuse analysis will always—or
future guidelines on exclusionary abuses to redefine the objec- almost always62 —entail a full-f ledged economic analysis of any
tives of competition law. and all potentially relevant facts,63 i.e. of all facts that may poten-
tially have some impact on the ‘competitive structure’. In that
B. The concept of abuse—which criteria? case, the main function of the law is to provide for a procedure
1. An ‘effects-based approach’— a concept in need of within which both competition authorities and the undertaking
specification under investigation can produce and present economic evidence,
According to a broad consensus, any approach to exclusionary and that ensures access to the file and the right to be heard. If—
abuses will need to be ‘effects-based’ of some sort. The requisite and only if—a competition authority has duly considered all facts
effects relate to a given practice’s capability to restrict competi- and evidence submitted, it may, in the final assessment of the
tion54 —not to consumer harm. The ECJ’s jurisprudence has been effects, benefit from a limited margin of appreciation. Almost by
consistent in this respect from the beginning.55 What has changed necessity, such an approach will tend to prolong proceedings in
over the last 15 years or so are the requirements for showing such all settings where dominant undertakings may benefit from such
capability. In its earlier jurisprudence, the ECJ was satisfied with prolongations. Large amounts of complex evidence will then be
a rather abstract, rule-based approach—at least with regard to produced that enforcers will feel compelled to engage with in
some types of conduct: the use of loyalty rebates or exclusive deal- order not to risk losing the case in Court. Based on this sort of
ing arrangements by a dominant undertaking sufficed for pre- an effects-based approach, competition authorities will be able
suming an abuse. In defence, the dominant undertaking could rely to handle less and less cases. Economic complexity and the cost
on an objective justification and, from British Airways56 onwards, of enforcement are maximised. Legal certainty—a major building
on an efficiency defence. In its more recent jurisprudence the block of an effective protection against distortions of competi-
ECJ has significantly tightened the evidentiary requirements: a tion—is reduced.
capability to restrict competition must be shown in concreto, i.e. in A different reading would suggest that the case law on exclu-
the circumstances of the case at issue,57 based on an analysis of sionary abuses has not outright abandoned a reliance on simpli-
the market(s) in question and an understanding of the functioning fying rules that implicate an inference of anticompetitive effects,
of competition on that or those market(s).58 It must not merely but that these rules have become more differentiated and sen-
be hypothetical,59 and it cannot be deduced from an ‘analytical sitive to context within a structured proceeding in which the
template’ based on the type of conduct alone.60 With a view burden to produce evidence and the burden of proof may shift
to loyalty rebates and exclusive dealing, the Commission or a to and fro.64 An important challenge in ensuring an effective
enforcement of Art. 102 TFEU would then lie in redefining and
specifying the preconditions under which anticompetitive effects
49 Amending Communication, para 1.
50
can be initially inferred or even presumed,65 what kind of evidence
Case T-604/18 Google Android, EU:T:2022:541, para 1028.
51 For references see fn 3 of the Policy Brief. must be produced to counter the inference or presumption, and
52 Massimiliano Kadar and Johannes Holzwarth, ‘Effects-based approach?
Effects-based approach! The European Commission’s “Article 102 Package”’
[2023] Neue Zeitschrift für Kartellrecht 333, 334. Similarly: Kadar, Holzwarth 61 Servizio Elettrico Nazionale (n 38), para 50; Case C-413/14 P Intel v Commis-
and Pereira (n 33), p 3.
53
sion, EU:C:2017:632, paras 138 and 140.
See e.g. Lina M Khan, ‘Amazon’s Antitrust Paradox’ (2017) 126 Yale Law 62 Most commentators accept that there are some ‘by object’ abuses (in
Journal 710, 737 ff; Lina Khan, ‘The New Brandeis Movement: America’s Anti-
analogy to Art. 101(1) TFEU) which allow the Commission or competition
monopoly Debate’ (2018) 9 Journal of European Competition Law & Practice 131;
authority to infer effects; but they argue that this ‘box’ should be narrowly
Tim Wu, ‘After Consumer Welfare, Now What?’ [2018] CPI Antitrust Chronicle,
construed—see, for example, Pablo Ibáñez Colomo, ‘Competition on the merits’
April 2018.
54
(2024) 61 Common Market Law Review 387, 410–414.
Servizio Elettrico Nazionale (n 38), para 50; Case C-680/20 Unilever Italia 63 Justin Lindeboom speaks of an ‘all things considered’-approach, see ‘For-
Mkt. Operations, EU:C:2023:33, paras 41–42; Case C-333/21 European Superleague
malism in Competition Law’ (2022) 18 Journal of Competition Law & Economics
Company, EU:C:2023:1011, para 130.
55
832, 858 ff.
Case 85/76 Hoffmann-La Roche v Commission, EU:C:1979:36, para 91. 64
56
Generally on this point: Arndt Christiansen and Wolfgang Kerber, ‘Com-
Case C-95/04 P British Airways v Commission, EU:C:2007:166, para 86. petition policy with optimally differentiated rules instead of per se rules vs. rule
57 Servizio Elettrico Nazionale (n 38), para 50; Unilever Italia (n 54), paras 41–42; of reason’ (2006) 2 Journal of Competition Law & Economics 215. See also: Pablo
European Superleague (n 54), para 130. Ibáñez Colomo, ‘Structured Legal Tests, Effective Judicial Review and Missing
58 European Superleague (n 54), para 130. the Trees for the Forest’ (2022) 13 Journal of European Competition Law &
59 Unilever Italia (n 54), para 42. Practice 461–462.
60 European Superleague (n 54), para 130. 65 On the terminology see below, IV. B. 1.
8 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

what additional evidence may then revive the case. There is both criteria, there are settings where the ‘no competition on the
no basis for a general argument that inferences and presump- merits’ criterion gains its own weight.
tions should be construed restrictively, such as to minimise ‘false When the two criteria are met, there is a prima facie case for
positives’ in any given case. Where the beneficial settings are an exclusionary abuse. Nonetheless, the dominant firm can still
difficult to separate from the harmful ones and harmful effects come forward with an objective justification or efficiency defence.
are frequent, a legal requirement of economic precision case
by case can come at the cost of systemic underenforcement.66 3. Conduct capable of hindering competition—the
Broader inferences may lower overall error costs. Consequently, ‘effects’-criterion in Art. 102
the challenge for an effects analysis within a legal proceeding An abuse of dominance presupposes that a dominant undertaking
is not to burden enforcement institutions with the task to fully engages in conduct that is capable of hindering competition.

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

replicate economic complexity. Rather, account must be taken While the Union Courts’ case law has f leshed out the concept in a
of the way legal rules function: if their aim is to minimise error series of judgments, a relevant degree of fuzziness persists. Part of
cost at the systemic level, the calculus must include their effect this fuzziness may be inevitable. Nonetheless, some clarification
on norm addressees and enforcement institutions.67 This is the is called for.
interpretation we endorse.
3.1. Adverse effects on an effective competition structure
2. Anti-competitive effects—preconditions for establishing In Hoffmann-La Roche, the concept of abuse was defined as ‘an
an abuse objective concept relating to the behaviour of an undertaking in
The concept of exclusionary abuses refers to conduct that is a dominant position which is such as to inf luence the structure
capable of producing anticompetitive effects.68 The 2008 Priorities of the market’ and which ‘through recourse to methods differ-
Guidance has equated anticompetitive effects with a conduct’s ent from those which condition normal competition’ has the
capability of foreclosing AECs. The Union Courts’ case law sug- effect of ‘hindering the maintenance of the degree of competition
gests a broader approach: in order to qualify as an abuse, the existing in the market or the growth of that competition’74 —a
relevant conduct must, first, have the (actual or potential) effect formula much repeated since then. More recently, the ECJ also
of ‘hindering the maintenance of the degree of competition still refers to practices ‘capable of adversely affecting, by way of
existing in the market or the growth of that competition’. However, resources other than those which govern normal competition,
a lessening of competition, namely a foreclosure of less efficient an effective competition structure’.75 By implication, this effects-
competitors, may result from the very process of competition that criterion does not relate to consumer harm (see above, III. B.
competition law aims to protect—an old and fundamental insight 1.). Rather, what matters are the conduct’s potential effects on
that has recently been dubbed the ‘AEC principle’ (see below, III. B. barriers to market entry or expansion,76 e.g. by hampering actual
4.). A second criterion is therefore needed to distinguish between or potential competitors’ effective access to resources, data, or
pro-competitive and anticompetitive practices. Since Hoffmann-La customers, by removing or restricting the buyers’ freedom as
Roche, the Union Courts require that adverse effects on competi- regards choice of sources of supply,77 or by otherwise creating
tion must result from ‘recourse to methods different from those competitive disadvantages for actual or potential rivals, thereby
which condition normal competition in products or services’.69 shielding the dominant undertaking from competition, reducing
Unfortunately, this second criterion is sometimes difficult to han- competitors’ incentives to innovate78 or enabling the dominant
dle. At a relatively abstract level, almost any type of ‘normal’ com- undertaking to expand into adjacent markets irrespective of the
mercial conduct can be turned into an instrument of exclusion by superior attractiveness of its offer. As these examples illustrate,
a dominant undertaking, depending on the circumstances and the the concept of ‘adverse effects on the competition structure’ is
precise form the conduct takes. This is true, for example, for price much broader than the concept of (full) foreclosure or marginal-
competition, refusals to deal, or product design. More recently, isation of competition.79 The potential of a practice to cause
the Union Courts tend to ask whether the relevant practice lies adverse effects on the competition structure will reinforce the
within or outside the scope of ‘competition on the merits’.70 But risk implicit in the position of dominance itself, namely the risk
that concept, while arguably of increasing importance,71 is in need that the dominant undertaking will negatively inf luence, ‘to its
of explanation itself. own advantage and to the detriment of consumers, the various
We will discuss these two criteria (restrictive effects and no parameters of competition, such as price, production, innovation,
competition on the merits) separately below. While it is true variety, or quality of goods or services’.80 Consequently, potential
that some tests (in particular the AEC test) let the two criteria
collapse72 (see below, III. B. 4. 1.) and that, ultimately, ‘anticom-
in Case C-377/20 Servizio Elettrico Nazionale and Others, EU:C:2021:998, para 48;
petitive effects’73 must be established—a term that conf lates Damien J Neven, ‘The As-Efficient Competitor Test and Principle. What Role
in the Proposed Guidelines?’ (2023) 14 Journal of European Competition Law
& Practice 565, 569: ‘ . . . the absence of normal competition or competition on
66 For a similar argument, albeit in a different legal context, see Justice the merits are just different paraphrases for the presence of anti-competitive
Breyer, Dissenting Opinion, Leegin Creative Leather Products, Inc. v. PSK, Inc., effects’.
551 U.S. 877 (2007). 74 Hoffmann-La Roche (n 55), para 91. See also Servizio Elettrico Nazionale (n
67 Justice Breyer, Dissenting Opinion, Leegin Creative Leather Products, Inc. v. 38), para 44 with further references.
PSK, Inc., 551 U.S. 877 (2007). 75 Servizio Elettrico Nazionale (n 38), para 68; Unilever Italia (n 54), para 36. See
68 Servizio Elettrico Nazionale (n 38), para 50. also the Amending Communication, Annex para 1. For a critical view regarding
69 Hoffmann-La Roche (n 55), para 91. See also Case C-209/10 Post Danmark, the criterion of negative effects ‘on the competitive structure’ see Neven (n 73)
EU:C:2012:172, para 24; Servizio Elettrico Nazionale (n 38), para 44; European 566 and 568–569: The term would be ill-defined and the concept could be used
Superleague (n 54), para 125. as ‘a shortcut to avoid the analysis of anticompetitive effects’.
70 76 European Superleague (n 54), para 131.
See below, III. B. 4.
71 77 Post Danmark (n 69), para 26; referring to Case C-280/08 P Deutsche Telekom
See Kadar, Holzwarth and Pereira (n 33), p 3.
72 Emphasised, for example, by Ibáñez Colomo (n 62) 415. v Commission, EU:C:2010:603, para 175. See also: Case T-604/18 Google Android (n
73 The term makes it clear that this precondition requires both the finding 50), para 281.
78 See, for example, Case T-604/18 Google Android (n 50), para 294.
of restrictive effects and a determination that these do not result from pro-
79 For this see also Policy Brief, p 4.
competitive conduct. Consequently, some argue that the ‘no competition on the
merits’ criterion should not be assessed separately—see Opinion of AG Rantos 80 Case T-604/18 Google Android (n 50), para 281.
 Journal of European Competition Law & Practice, 2024 | 9

adverse effects on the competitive structure come with a risk of likely than its absence’,92 the ECJ merely repeated that the anti-
consumer harm. But they must not be confused with consumer competitive effect must be likely,93 thus declining to establish a
harm: as such, the issue of consumer harm will only be addressed clear probability threshold. Under the substantive law on Art. 102
in the context of the efficiency defence. TFEU, the ECJ holds on to a vague ‘capability standard’,94 thereby
Nor must the concept of ‘competitive structure’ be confused delegating genuine probability assessments to the standard of
with ‘market structure’.81 An analysis of the (actual or potential) proof.
effects of a dominant undertaking’s practice on the ‘competitive Simultaneously, the ECJ is determined in its requirement of
structure’ will focus on effects on the ‘structural’ preconditions concreteness: The alleged anticompetitive effects must not be
that determine how competition functions in the relevant mar- ‘purely hypothetical’,95 i.e. potential effects must not be analysed
kets. Some refer to effects on the ‘competitive process’ in this in the abstract. The determination of effects ‘may entail the use of

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

regard,82 i.e. to the process that results from this structure. By different analytical templates depending on the type of conduct’;
focusing on the ‘competitive structure’, the Union Courts empha- but even then, it must be made ‘in the light of all the relevant
sise the need to focus not on short-term effects, but to adopt a factual circumstances’, including ‘the conduct itself, the mar-
dynamic perspective. ket(s) in question [and] the functioning of competition on that or
those market(s).’96 Ultimately, the standard of ‘concretely likely’
3.2. Actual or potential effects adverse effects on competition requires competition authorities
The finding of an abuse does not presuppose proof of actual anti- and Courts to identify a context-specific97 theory of how precisely
competitive effects:83 anticompetitive conduct can be prohibited the relevant conduct is capable of triggering adverse effects on
already before harm to competition has occurred. EU competi- competition and verify whether the market conditions on which
tion rules are meant to react to the risk for competition that this theory is based are met or at least likely to arise.98 According
certain conduct raises.84 This raises the question of what level of to Elias Deutscher, a ‘realistic prospect’ of anticompetitive effects
risk—or which probability of anticompetitive effects—will trigger is required.99
the abuse prohibition. Interestingly—and contrary to the law on The relatively vague ‘capability’ standard comes with the dis-
merger control85 —the Union Courts have been unwilling to estab- advantage of uncertainty. Simultaneously, it makes the prohibi-
lish a clear substantive standard of probability, however. Instead, tion of Art. 102 TFEU sensitive to the potential gravity or magni-
the case law refers to ‘potential’,86 ‘probable’,87 or ‘likely’88 effects, tude of harm to competition that is at issue:100 Along that line,
to the capability of the conduct to cause anticompetitive effects,89 practices that may plausibly lead to a long-time, full foreclosure
or to conduct that ‘tends to’90 cause adverse effects seemingly of competition may qualify as abusive even where they are merely
interchangeably.91 In Post Danmark II, where the referring court capable of, but not ‘more likely than not’ to produce such effects—
explicitly asked how likely the anticompetitive effect of a rebate in particular if these practices cannot be shown to come with
scheme must be in order to constitute an abuse, and where AG significant pro-competitive benefits. While this does not amount
Kokott, in her Opinion, argued that, ‘on the basis of an overall to an attempt to quantify the expected harm (as well as the
assessment of all the relevant circumstances of the individual expected pro-competitive benefits),101 it opens the possibility for
case, the presence of an exclusionary effect [must appear] more
92 Opinion of AG Kokott in Case C-23/14 Post Danmark II, EU:C:2015:343,
para 82. AG Kokott further explains this standard in paras 83–84: A higher
81 Erroneously suggesting that commitment to maintaining an ‘effective bar than ‘more likely than not’ (e.g. a requirement that exclusionary effects
competition structure’ is analogous to protecting a given ‘market structure’: must be ‘very likely’ or ‘particularly likely’) would be inappropriate, due to the
Pınar Akman, ‘A Critical Inquiry into “Abuse” in EU Competition Law’, (2024) 44 special responsibility of the dominant undertaking not to undermine effec-
Oxford Journal of Legal Studies (published online 7 March 2024), p 26. tive and undistorted competition in the internal market. ‘That responsibility
82 See e.g. Raffaele Di Giovanni Bezzi, ‘A Tale of Two Cities: Effects Analysis entails some obligation to exercise restraint on the market. The dominant
in Article 102 TFEU Between Competition Process and Market Outcome’ (2023) undertaking must therefore refrain from all commercial practices which are
14 Journal of Competition Law & Practice 83; Renato Nazzini, The Foundations likely to produce an exclusionary effect’. A ‘more likely than not’ standard is
of European Union Competition Law (OUP 2011) 14 ff. In its decisions, the Com- also advocated in the literature—see, for example, Pablo Ibáñez Colomo, ‘The
mission sometimes explicitly uses the term ‘competitive process’ in its effects (Second) Modernisation of Article 102 TFEU: Reconciling Effective Enforcement,
analysis, see e.g. Google Android (Case AT.40099) Commission Decision of 18 July Legal Certainty and Meaningful Judicial Review’ (2023) 14 Journal of European
2018, recitals 863, 971, 1142. See also 2008 Priorities Guidance, para 6: ‘The Competition Law & Practice 608, 618.
emphasis of the Commission’s enforcement activity in relation to exclusionary 93 Post Danmark II (n 83), para 67.
conduct is on safeguarding the competitive process in the internal market.’ 94 See Elias Deutscher, ‘Causation and counterfactual analysis in abuse of
Simultaneously, the 2008 Priorities Guidance distinguished between ‘protecting dominance cases—lessons from the General Court’s Qualcomm ruling’ (2023) 19
an effective competitive process’ and ‘protecting competitors’. European Competition Journal 481, 509 ff. From the case law see e.g. Unilever
83 See Case C-52/09 TeliaSonera Sverige, EU:C:2011:83, para 64; Case T-336/07 Italia (n 54), para 41 (‘that conduct had [ . . . ] the ability to restrict competition
Telefónica and Telefónica de España v Commission, EU:T:2012:172, para 90; Case C- on the merits’).
457/10 P AstraZeneca v Commission, EU:C:2012:770, para 112; Case C-23/14 Post 95 Post Danmark II (n 83), para 65; the ECJ referred to para 80 of AG Kokott’s
Danmark II, EU:C:2015:651. Also: Castillo de la Torre (n 42) 152. Opinion (n 92) in this regard. For the same phrasing see: Servizio Elettrico
84 See Unilever Italia (n 54), para 41. In terms of the German legal tradition, Nazionale (n 38), para 70; with a reference to Post Danmark II (n 83), para 65;
Art. 102 TFEU qualifies as a ‘Gefährdungsdelikt’—see also Opinion of AG Kokott Unilever Italia (n 54), para 42 (with reference to the presumption of innocence,
in Case C-8/08 T-Mobile Netherlands and Others, EU:C:2009:110, para 47. which suggests that the ECJ, in this passage, deals with the standard of proof).
85 See Case C-413/06 P Bertelsmann and Sony Corporation of America v Impala, 96 European Superleague (n 54), para 130. For the need to consider the
EU:C:2008:392, para 52: in order to prohibit a merger, a significant impediment market context when determining potential effects see also: Policy Brief, p 3
to competition must be more likely than not to occur. See also: CK Telecoms (n (no ‘simplistic or formalistic standard’), referring to Generics (UK) (n 43), para
31), para 84. 154; Telefónica (n 83), para 175; Deutsche Telekom (n 77), para 175; TeliaSonera
86 Case T-612/17 Google and Alphabet v Commission (Google Shopping), (n 83), para 28; Case T-235/18 Qualcomm v Commission (Qualcomm—exclusivity
EU:T:2021:763, para 438; TeliaSonera (n 83), para 77; Post Danmark II (n 83), para payments), EU:T:2022:358, paras 396–398; Case C-165/19 P Slovak Telekom (n 89),
66. para 42.
87 Post Danmark II (n 83), para 74. 97 Unilever Italia (n 54), para 44.
88 Post Danmark (n 69), para 44; Post Danmark II (n 83), para 67. 98 See Servizio Elettrico Nazionale (n 38), para 70: If a competition authority
89 Case T-201/04 Microsoft v Commission, EU:T:2007:289, para 867; Case C- assumes a causal link between a certain line of conduct and (potential) restric-
165/19 P Slovak Telekom v Commission, EU:C:2021:239, para 109; Case T-814/17 tions of competition, the specific market conditions on which the assumption
Lietuvos geležinkeliai v Commission, EU:T:2020:545, para 80; Telefónica (n 83), para is based must be met or at least likely to arise.
268; Post Danmark II (n 83), paras 31, 68. 99 Deutscher (n 94) 510.
90 Lietuvos geležinkeliai (n 89), para 80; Telefónica (n 83), para 268. 100 See also: Deutscher (n 94) 515.
91 Policy Brief, p 3 (with fn 22): ‘Despite the varied terminology, the applica- 101 In favour of an ‘expected value’ test in the context of digital mergers:
ble legal standard endorsed by the Union Courts and applied by the Commis- Massimo Motta and Martin Peitz, ‘Big Tech Mergers’, Discussion Paper Series
sion must be understood as being one and the same’. CRC TR 224, May 2020, pp 35 ff.
10 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

some degree of balancing of likely harm and likely benefits, possi- Art. 102 enforcement ‘unduly burdensome or even impossible’.110
bly including the question whether any pro-competitive benefits For this position, the Policy Brief relied on findings by the GC
could be realised by other, more pro-competitive means. Simulta- in Google Shopping and in Google Android: in Google Shopping, the
neously, the requirement to analyse the likely effects in light of GC highlighted that in some settings, trying to determine the
all relevant facts of the case underlines that Art. 102 TFEU shall dominant undertaking’s conduct absent the alleged abuse can
not capture abstract, but concrete risks to competition.102 This become highly speculative.111 Consequently, it was satisfied with
does not exclude the use of inferences based on economically a showing of a correlation between the relevant conduct and a
informed experience and/or sound economic theory. But such ‘modification of competition’ on the relevant markets and ‘addi-
inferences must remain subject to the challenge that, given the tional information, which may include . . . the assessments of
circumstances of the case, no concrete risk to competition exists. market participants’.112 In Google Android, the GC confirmed that,

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

In light of this case law, Art. 102 TFEU’s reference to ‘potential where a restriction of competition and its anticompetitive effects
effects’ should therefore be read to clarify two things: a dominant were otherwise shown, the Commission was not required to carry
firm’s conduct may qualify as abusive irrespective of whether out a counterfactual analysis.113
actual effects have already materialised. But the risk of anticom- By contrast, the GC quashed the Commission’s Qualcomm deci-
petitive effects must be plausible and concrete in light of the facts sion, both on procedural and substantive grounds. According to
of the case—without there being a quantified level of probability. the GC, the Commission had not established that the competitive
The requirement of concreteness marks the Union Courts’ situation would have been better without Qualcomm offering
departure from Hoffmann-La Roche103 and its shift towards an exclusivity payments to Apple conditional upon Apple’s com-
‘effects-based approach’: In Hoffmann-La Roche, the ECJ was willing mitment to purchase its entire requirement of LTE chipsets for
to accept, for certain types of conduct, analytical templates that iPhones from Qualcomm as compared with the situation with
inferred potential anticompetitive effects from certain forms of these exclusivity payments, because Apple had no technical alter-
conduct alone, at least in typical settings.104 According to the ECJ’s native to Qualcomm’s LTE chipsets.114
more recent jurisprudence, ‘the abusive nature of a practice does Given this tension in the GC’s more recent case law, a debate
not depend on the form it takes or took, but presupposes that has ensued whether a counterfactual analysis is indeed required
that practice is or was capable of restricting competition and, in Art. 102 cases,115 and if so, in what form, or whether the
more specifically, of producing, on implementation, the alleged causality between conduct and anticompetitive effects can also
exclusionary effects’, where this condition ‘must be assessed be established otherwise. In a careful enquiry, Elias Deutscher has
having regard to all the relevant facts’.105 set out why requiring a counterfactual analysis in the strict sense
may at times be an overly demanding standard that is bound
3.3. Causality of conduct for adverse effects to lead to type II errors when applying Art. 102 TFEU:116 inter
In order to find an exclusionary abuse, some sort of causal link alia, it will not take account of potential concurrent causes of
between the dominant undertaking’s conduct and the (actual or anticompetitive foreclosure, i.e. of settings, where high barriers
potential) anticompetitive effects must be established: the effects to entry and high switching costs may themselves be a relevant
must be attributable to the dominant undertaking’s conduct.106 and possibly sufficient cause for the absence of competition, but
Typically, causality is proven by reference to a ‘counterfactual’: where additional anticompetitive strategies may further decrease
a competition authority will compare the likely state of com- contestability and establish an additional layer of protection for
petition with the allegedly abusive conduct with an alternative an entrenched position of dominance; and—if wrongly concep-
scenario without that conduct. In the 2008 Priorities Guidance, the tualised, as arguably happened in Qualcomm—a counterfactual
Commission committed to identifying a suitable ‘counterfactual’ analysis can run counter to the ‘capability standard’ of anticom-
in each Art. 102 case.107 Obviously, all that can be required is a petitive effects, i.e. to the fact that all that is required is that the
plausible causality,108 i.e. a showing that the conduct was capable conduct was ‘capable of’ producing actual or potential effects.
of producing or likely to produce actual or potential effects.109 This implies that the Commission need not settle on the most
In their recent Policy Brief, DG Comp officials have questioned likely counterfactual but may simultaneously consider ‘but for’
the commitment to establish a clear counterfactual, at least scenarios with a probability of less than 50 per cent—provided
where the market realities as they might have evolved without the they are realistic.117 For example, it may consider that, absent
allegedly abusive practice become speculative: in such a setting, the allegedly abusive conduct, market entry would have occurred,
requiring the Commission to establish a ‘nexus of full causality’ even if not under a ‘more likely than not’ standard, due to high
between conduct and anticompetitive effect would risk to render
110 Policy Brief, p 4.
111 Case T-612/17 Google Shopping (n 86), para 377 speaks of an ‘arbitrary or
even impossible exercise’. See also: Castillo de la Torre (n 42) 168 with further
102 In the German legal tradition, Art. 102 TFEU would therefore qualify as a references.
‘konkretes Gefährdungsdelikt’ as opposed to an ‘abstraktes Gefährdungsdelikt’. 112 Case T-612/17 Google Shopping (n 86), paras 376 ff, para 382. It would then
103 Hoffmann-La Roche (n 55), para 91. be for the dominant undertaking to come forward with relevant information
104 In its subsequent decision-making practice, the Commission did verify that might cast doubt on the existence of a causal link (para 382).
whether an assumption of anticompetitive effects was justified in the circum- 113 See Case T-604/18 Google Android (n 50), para 893.
stances of a given case, i.e. whether the setting was atypical and therefore 114 Case T-235/18 Qualcomm (n 96), paras 400–415. For an in-depth analysis
required a closer effects analysis—see Case T-201/04 Microsoft (n 89), para 868. see Deutscher (n 94).
105 Servizio Elettrico Nazionale (n 38), para 72; Generics (UK) (n 43), para 154, 115 For a defence of the counterfactual analysis see Neven (n 73) 566 and
and Case C-165/19 P Slovak Telekom (n 89), para 42, as references. 571–572, arguing that the counterfactual is ‘one of the most fundamental
106 Case T-612/17 Google Shopping (n 86), para 441. disciplines in the assessment of effects’ and should not be relaxed. Similarly
107 2008 Priorities Guidance, para 21. Daniel Higer, ‘Die jüngste Initiative der Kommission zu Art. 102 AEUV: Abkehr
108 See, inter alia, Deutsche Telekom (n 77), para 257; and Telefónica (n 83), para von einer bloßen Prioritätenmitteilung hin zu Leitlinien’ [2023] Neue Zeitschrift
394–402, esp. 402. für Kartellrecht 385, 389. For the need to conduct a counterfactual analysis of
109 Servizio Elettrico Nazionale (n 38), para 98: The competition authority had some sort see also: Opinion of AG Kokott in Case C-48/22 P Google Shopping,
to show, supported by evidence, that the procedure to collect its customers’ EU:C:2024:14, paras 169 ff.
116 Deutscher (n 94) 484 ff.
consent for the transfer of data favoured a transfer of the data to its own
subsidiaries. 117 Deutscher (n 94) 511–512.
 Journal of European Competition Law & Practice, 2024 | 11

barriers to entry. By contrast, proponents of a counterfactual Consequently, ‘competition on the merits’ must not be read as
analysis will typically require the Commission to make the most referring to a list of practices that dominant undertakings must
likely ‘but for’ scenario its reference point—which will frequently not use. Instead, it stands for the general idea that, first, dominant
be the status quo ante. Often enough, this will be a situation of undertakings shall compete on account of their skills, abilities,
entrenched dominance where entry cannot be expected with a and superior performance,125 striving to offer consumers the best
51 per cent probability. The analysis will then tend to neglect value for money or most attractive product or deal instead of
the adverse impact of the dominant undertaking’s conduct on competing by obstructing their competitors’ effective access to
dynamic competition and contestability: implicitly, the ‘capability markets, resources, data, or customers or by otherwise creating
standard’ of anticompetitive effects would turn into a ‘likely competitive disadvantages for actual or potential rivals; and, sec-
actual effects’ standard.118 ondly, that they must consider the effects of their conduct, due to

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

All this cannot justify abandoning the need to establish a their ‘special responsibility’ not to allow their conduct ‘to impair
causal link on the basis of a ‘but for’-test, i.e. the need for some genuine undistorted competition on the common market’.126
sort of counterfactual analysis.119 However, in the context of a In some settings, these principles alone will suffice to iden-
potential effects analysis, particular caution is required: a show- tify abusive conduct. As the ECJ has put it in Servizio Elettrico
ing that the relevant conduct was plausibly capable to contribute Nazionale: ‘conduct which has the effect of broadening consumer
to anticompetitive effects in the specific context of the case must choice by putting new goods on the market or by increasing
suffice; and/or that it contributed to making market entry or an the quantity or quality of the goods already on offer must, inter
expansion of competitors more difficult.120 Along these lines, it alia, be considered to come within the scope of competition on
may be enough for the Commission to show a positive correlation the merits’.127 By contrast, practices which cannot be plausibly
between the allegedly anticompetitive conduct and anticompet- explained as performance-based competition, but only by the
itive effects that is not easily explained by other factors, or to dominant undertaking’s interest to eliminate competitors—i.e.
show that, absent the conduct, a realistic prospect of market entry practices that, in the context of a given case, neither contribute to
would have existed.121 productive efficiencies nor to enhanced consumer attraction, but
tend to foreclose competitors—will constitute an abuse under Art.
4. No ‘competition on the merits’ 102 TFEU. The inquiry into the underlying object of a dominant
Establishing adverse effects on the competitive structure does not firm’s strategy is frequently referred to as the ‘no economic
suffice for establishing an abuse (see above, III. B. 2.). In addition, sense’-test—a test that does not define competition that is not
any actual or potential exclusionary effects must not result from on the merits128 but helps to identify clear cases of so-called
‘normal’ competition, i.e. from ‘competition on the merits’122 or ‘naked exclusion’. In such cases, competition authorities may
‘performance-based’ competition.123 infer anticompetitive effects (see below, IV. C. 2.). The two criteria
Despite Hoffmann-La Roche’s reference to ‘methods different for establishing an abuse then collapse.
from . . . normal competition’, ‘normal’ competition and practices Frequently, a dominant undertaking’s potentially exclusion-
that fall outside the scope of ‘competition on the merits’ cannot ary practices or strategies will come with some potential effi-
be identified based on their form alone: the concept of abuse does ciencies. In those cases, additional tests and criteria for distin-
not (only) refer to practices that are inherently improper,124 or that guishing between pro-competitive and anticompetitive conduct
are always improper when applied by a dominant undertaking. will be needed:129 the concept of ‘competition on the merits’
Rather, depending on the circumstances of a case, almost any merely serves as a general umbrella for these tests which may
type of otherwise normal commercial conduct can be turned into look different in different settings. In the Union Courts’ more
an instrument of anticompetitive exclusion by a dominant under- recent jurisprudence, the so-called ‘AEC principle’ has gained
taking, including price competition, product design, or innovation. prominence (4.1.). But given a dominant undertaking’s special
responsibility not to undermine effective and undistorted com-
petition in the internal market, even the AEC principle comes
118 Deutscher (n 94) 508; referring to Case T-612/17 Google Shopping (n 86),
para 377.
with limitations (4.2.). In other words, ‘competition on the merits’
119 See also the Opinion of AG Kokott in Case C-48/22 P Google Shopping, is a concept with different facets, some of them more, some
EU:C:2024:14, paras 169 ff. of them less normatively loaded. Practices that do qualify as
120 For a similar line of argument, against the need for a counterfactual
analysis in the strict sense, see Castillo de la Torre (n 42) 170–171. ‘competition on the merits’ exclude a finding of abuse. On the
121 Deutscher (n 94) 509. other hand, practices that cannot be explained as a legitimate part
122 For that wording see Post Danmark (n 69), para 22; Case C-413/14 P Intel (n of performance-based competition and result in the foreclosure of
61), para 134; Servizio Elettrico Nazionale (n 38), para 45. See also: 2008 Priorities
Guidance, para 6. competitors, or where the pro-competitive justification is clearly
123 For that wording see Post Danmark (n 69), para 24 (‘normal competition out of proportion to the anticompetitive effects, will typically
on the basis of the performance of commercial operators’). See also Liza Lov-
dahl Gormsen, A Principled Approach to Abuse of Dominance in European Competition
Law (CUP 2010) 45 (fn 155), who links the idea of ‘competition on the basis of 125
performance’ to the German concept of ‘Leistungswettbewerb’ (see below n For references of the Union Courts to the concept of ‘performance-
124), as in the German version of the ECJ’s judgment in Hoffmann-La Roche (n based competition’ see Post Danmark (n 69), para 24 (‘competition on the
55), para 91. basis of performance’); Case C-62/86 AKZO v Commission, EU:C:1991:286, para
124 70 (‘competition on the basis of quality’); Case C-202/07 P France Télécom v
Pablo Ibáñez Colomo suggests that the Union Courts’ case law in the
Commission, EU:C:2009:214, para 106 (‘competition on the basis of quality’).
1970s and 1980s was informed by the idea that lawful and unlawful conduct 126
could be told apart in the abstract—see Ibáñez Colomo (n 62) 389. It is Case C-322/81 Michelin v Commission, EU:C:1983:313, para 57.
127 Servizio Elettrico Nazionale (n 38), para 85.
unclear whether the ECJ’s early case law was indeed based on this assumption.
128 For the fact that the category of conduct that is not ‘on the merits’
See, for example, Case 6/72 Europemballage and Continental Can v Commission,
EU:C:1973:22, para 27, emphasising that an abuse may exist ‘regardless of is broader than the category of ‘naked exclusion’ as identified by the ‘no
the means and procedure by which it is achieved’, rather focusing on effects economic sense’ test see: Case T-612/17 Google Shopping (n 86), para 188.
(para 26). In Germany, where the concept of ‘competition on the merits’ was 129 This has also been pointed out by AG Rantos in his Opinion in Servizio
discussed early on (since the late 1920s, as ‘Leistungswettbewerb’), it was Elettrico Nazionale (n 73), para 55, pointing out the ‘abstract’ nature of the
uncontroversial that this concept did not allow for an abstract distinction concept of ‘competition on the merits’ and that it ‘cannot be defined in such a
between lawful and unlawful conduct—see, for example, Burkhardt Röper, ‘Zur way as to make it possible to determine in advance whether or not particular
Verwirklichung des Leistungswettbewerbs’ in Hans-Jürgen Seraphim (ed) Zur conduct comes within the scope of such competition’. See also Neven (n 73) 570
Grundlegung wirtschaftspolitischer Konzeptionen (Duncker & Humblot 1960) 261 ff. and 572: The concept of ‘competition on the merits’ as such is not operational.
12 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

qualify as an abuse. What is more, a lower evidentiary threshold information is known to the norm addressee.141 Consequently,
for proving anticompetitive effects may apply to practices that are the AEC test has become, for these types of cases, the standard
clearly no part of performance-based competition (more on this, analytical template to identify an exclusionary abuse.142 One of
see below, IV. C.). its advantages is that it lets the two legs of the abuse analy-
sis collapse: the identification of ‘anticompetitive foreclosure’ is
4.1. The AEC principle and the AEC test based on one single criterion. Prices below some measure of the
Competition ‘may, by definition, lead to the departure from the dominant undertaking’s own cost do not qualify as ‘competition
market or the marginalisation of competitors that are less effi- on the merits’. Simultaneously, they are capable of adversely
cient and so less attractive to consumers from the point of view affecting the competition structure.143
of, among other things, price, choice, quality, or innovation’.130 Attempts to adapt the AEC test to other types of conduct

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

Consequently, a practice’s exclusionary effects alone cannot con- come with significant complications, however.144 First, the AEC
stitute an abuse. In Servizio Elettrico Nazionale, the ECJ has recently test quickly loses its objectivity when applied to non-price-related
restated this well-established tenet—lately referred to as the ‘as- practices, given that a hypothetical AEC would need to be compa-
efficient competitor’-principle (AEC principle): it is ‘in no way rable with the dominant firm not only in terms of cost structure,
the purpose of Art. 102 TFEU to prevent an undertaking from but also in terms of capacity to innovate, quality, and other
acquiring, on its own merits—on account of its skills and abilities respects.145 Secondly, one of the great advantages of the AEC test
in particular—a dominant position on a market, or to ensure that in price-based cases—the ability of the dominant undertaking
competitors less efficient than an undertaking in such a position to conduct the test itself—is no longer present when the test is
should remain on the market.’131 This insight applies to both applied to non-price-based conduct. Rather, additional informa-
price-based and non-price-based practices.132 tion will be needed that lies outside the dominant undertaking’s
The AEC test—or rather the different varieties of AEC tests sphere. In the case of conditional rebates, for example, the ‘con-
that have been developed for different types of conduct133 —is an testable share’ must be determined, i.e. the share of a customer’s
attempt to make the AEC principle operational by taking cost as purchase requirements that can realistically be switched to a
the metric for efficiency.134 The 2008 Priorities Guidance has com- new competitor in a given time.146 Consequently, an AEC test
mitted to regularly conducting an AEC test in cases of price-based becomes much more complex, time-consuming, uncertain, and
conduct.135 In predatory pricing136 and margin squeeze cases,137 controversial. Given the GC’s unwillingness to grant the Com-
an AEC test, i.e. a cost–price-comparison based on the cost struc- mission a margin of appreciation, there is a significant risk of
ture of the dominant firm, typically is the most straightforward losing a case in Court.147 The complex version of the AEC test
way to distinguish between efficiency-based competition on the proposed for conditional rebates in the 2008 Priorities Guidance
merits and anticompetitive exclusion. According to a consistent and the difficulties the Commission has experienced in putting it
line of cases, a foreclosure of AECs qualifies as sufficient condition into practice testify to the limited practicability of the AEC test in
for an abuse,138 absent an objective justification or efficiency these cases.148 The Commission’s Intel decision is illustrative: the
defence.139 The implementation of the AEC test will normally be AEC test spans 151 pages149 or 29.2 per cent of the total decision
manageable for the Commission.140 Simultaneously, all relevant length; respectively, 68.3 per cent of the pages that deal with the
finding of an abuse. It has been estimated that conducting the AEC
130 Servizio Elettrico Nazionale (n 38), para 73. test alone took up two years.150 Furthermore, with the growing
131 Servizio Elettrico Nazionale (n 38), para 73. According to Adriano Barbera, complexity, the risk of error increases,151 and with regard to any
Nicolás Fajardo Acosta and Timo Klein, ‘The Role of the AEC Principle and Test
in a Dynamic and Workable Effects-Based Approach to Abuse of Dominance’
(2023) 14 Journal of European Competition Law & Practice 582, 583, the AEC
principle thereby recognises the importance of ‘appropriability’ in dynamic
competition.
132 Servizio Elettrico Nazionale (n 38), para 79. See, however, Neven (n 73)
141 Cf. for predation: AKZO (n 125), para 74; for margin squeeze: TeliaSonera
571, who doubts that the AEC principle ‘offers useful guidance for non-pricing
practices’. (n 83), para 41. This advantage is also emphasised in the literature, see, for
133 For the fact that there is not one singular AEC test, but that the way example, Barbera, Fajardo Acosta and Klein (n 131) 587.
142 AKZO (n 125), para 72; Case T-271/03 Deutsche Telekom v Commission,
that the AEC principle is translated into a testable mathematical proposition
will depend on the type of conduct, see Barbera, Fajardo Acosta and Klein (n EU:T:2008:101, para 194.
143 Post Danmark (n 69), para 25; Servizio Elettrico Nazionale (n 38), para 72.
131) 587.
134 Barbera, Fajardo Acosta and Klein (n 131) 587. See also: Ibáñez Colomo (n 62) 415.
135 144 On the functioning of the AEC test in these cases, see Nazzini (n 82)
Cf. 2008 Priorities Guidance, para 23 in section C ‘price-based exclusion-
ary conduct’. For the fact that the AEC test is primarily applicable to price- 235–242; and Barbera, Fajardo Acosta and Klein (n 131) 588–590.
145 See Unilever Italia (n 54), para 39.
based practices see Servizio Elettrico Nazionale (n 38), para 80. Very explicit on
this point: Opinion of AG Kokott in Google Shopping (n 119), para 197: ‘the scope 146 For an illustration of the complexity of the calculation see Commission
of application of the as-efficient-competitor test should not be extended to Decision in Intel (n 18), recitals 1010–1011. Cf. on that, Robert Lauer, ‘The Intel
practices which bear no relation to price competition’. See, however, Barbera, saga: what went wrong with the Commission’s AEC test (in the General Court’s
Fajardo Acosta and Klein (n 131) 585, for the proposition that, in principle, AEC view)?’ (2024) 20 European Competition Journal 45, 58 ff.
tests could be developed also for non-price conduct. 147 For critical comments on the failure of the GC to recognise ‘the margin
136 For an explanation of how the AEC test works in predation cases see of error that is inherent in the implementation of the test’ and to grant
Barbera, Fajardo Acosta and Klein (n 131) 587. the Commission a ‘confidence interval around the point of estimate of the
137 For an explanation of how the AEC test works in margin squeeze cases difference between cost and effective price’ see, inter alia, Neven (n 73) 566 and
see Barbera, Fajardo Acosta and Klein (n 131) 588. 575. In particular, the Court’s view that there is ‘a single correct number’ for the
138 Servizio Elettrico Nazionale (n 38), para 76; Unilever Italia (n 54), para 39. contestable share would misunderstand ‘the nature of the empirical exercise
139 However, according to the 2008 Priorities Guidance, para 27, a finding that the implementation of the AEC test involves’.
148 Cf. for the critique Fiona M Scott Morton and Zachary Abrahamson, ‘A
that a price charged by the dominant undertaking has the potential to foreclose
equally efficient competitors should be considered just one piece of evidence, Unifying Analytical Framework for Loyalty Rebates’ (2017) 81 Antitrust Law
to be integrated ‘in the general assessment of foreclosure . . . , taking into Journal 777, 796 ff; Lauer (n 146) 56 ff.
149 Commission Decision in Intel (n 18), pp 302–453.
account other relevant quantitative and/or qualitative evidence’. See also
Neven (n 73) 570. Similarly Barbera, Fajardo Acosta and Klein (n 131) 591–592. 150 Cf. Wouter PJ Wils, ‘The Judgment of the EU General Court in Intel and
140 For the manageability of the AEC test in a margin squeeze case see, for the So-Called “More Economic Approach” to Abuse of Dominance’ (2014) 37
example, Commission Decision in Slovak Telekom (n 18), pp 228–262: the AEC test World Competition 405, 431.
spans only 34 pages or 8.4 per cent of the total decision length, even though the 151 Castillo de la Torre (n 42) 191–192: ‘If the EU Courts are too demanding
dominant undertaking did not fully comply with the Commission’s requests for on how the AEC test is conducted, they may be providing arguments not to use
information. it’.
 Journal of European Competition Law & Practice, 2024 | 13

error or uncertainty, the GC grants the defendant the benefit of to the Union Courts’ case law163 —a foreclosure of AECs is a
the doubt.152 sufficient condition for finding an abuse,164 competition may
Against this background, the 2008 Priorities Guidance’s propo- also be harmed by the exclusion of less efficient competitors.165
sition to make the AEC test a central and preferred analytical As recently reiterated by the ECJ in European Superleague,166 con-
methodology even beyond purely price-based conduct153 has long duct may be categorised as an abuse ‘not only where it has
been abandoned.154 After Intel and Google Android, the Commission the actual or potential effect of restricting competition on the
has refrained from conducting AEC tests outside the realm of merits by excluding equally efficient competing undertakings
predation and margin squeeze cases.155 In doing so, it can rely on from the market(s) concerned, but also where it has been proven
a consistent jurisprudence that acknowledges the Commission’s to have the actual or potential effect—or even the object—of
discretion in choosing the most suitable methodology for identi- impeding potentially competing undertakings at an earlier stage

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

fying anticompetitive conduct.156 Simultaneously, the more cau- . . . from even entering that or those market(s) and, in so doing,
tious assessment of what the AEC test can achieve is confirmed preventing the growth of competition therein to the detriment of
by a literature that increasingly acknowledges its limits.157 Impor- consumers, by limiting production, product, or alternative service
tantly, the AEC test does not inquire into the pro-competitive development or innovation’.167 Consequently, where the structure
or anticompetitive incentives that may have informed the dom- of the market—in particular high barriers to entry—makes it
inant undertaking’s conduct, and does not provide any theory of particularly difficult for an entrant to achieve a comparable level
harm:158 in the context of some theories of harm (like ‘compe- of efficiency,168 including due to strong positive network effects
tition softening’), the AEC principle and test are not informative or economies of scale, where, as a consequence, the presence
and may lead to type II errors.159 In the context of other theories of the dominant undertaking weakens competition such that
of harm, they may lead to type I errors.160 Consequently, as as-efficient competitors cannot emerge, or where the dominant
emphasised by Barbera, Fajardo Acosta, and Klein, ‘there is no position results from legal privileges that the undertaking enjoyed
one-size-fits-all answer to the question of whether and to what in the past, in particular from monopoly rights,169 the AEC prin-
extent the AEC principle and tests remain appropriate and useful ciple will not limit the application of Art. 102 TFEU:170 EU com-
in discerning anticompetitive conduct.’161 petition law recognises that even less efficient competitors can
exert competitive pressure such that ‘the market structure and
4.2. Limits of the AEC principle (and of AEC tests) the choices available to consumers do not deteriorate further’171
Beyond the AEC test, the AEC principle serves as an important and that, contrary to the position adopted in the 2008 Priorities
reminder of the aim and function of Art. 102 TFEU. But like Guidance, the AEC principle therefore does not serve as a ‘soft
the AEC test, it is not conclusive:162 while—at least according safe harbour’.172 Where barriers to entry are high and competitive
pressure from as-efficient competitors is structurally weak, the
152 negative effects on productive efficiency as they may result from
See Case T-604/18 Google Android (n 50), paras 643–644; Case T-286/09
RENV Intel Corporation v Commission, EU:T:2022:19, paras 244 ff. For a critical the entry or continued presence of less efficient competitors are
perspective see Castillo de la Torre (n 42) 164–165. Similarly Laitenberger and considered less important than their potentially positive impact
Kröger (n 19) 625: ‘in complex cases, the judicial review puts a very heavy
burden on enforcement’. on allocative and dynamic efficiency.173 As the notion of ‘competi-
153 See 2008 Priorities Guidance, paras 23 and 27: If the data suggested that tion on the merits’ encompasses the possibility of such balancing,
an equally efficient competitor could compete effectively with the dominant
undertaking’s pricing conduct, the Commission would not intervene. it is broader than the AEC principle and the AEC tests that help
154 The Amending Communication merely confirms a prior shift in the operationalising its meaning.
Commission’s practice, see Amending Communication, Annex para 3. See also:
Policy Brief, pp 6–7. For a different view on the appropriateness of an AEC test
in case of non-price conduct see Barbera, Fajardo Acosta, and Klein (n 131) 585:
‘there is no reason, in principle, why [an AEC test] cannot also be developed for whether the conduct of an undertaking in a dominant position is in keeping
non-price conduct, such as self-preferencing and bundling’. with the means of competition on the merits’. For the opposite position see
155 Akman (n 81), p 22, suggesting that the AEC principle is conclusive.
Recent publications personally linked to DG Comp suggest that the 163
Commission is planning on continuing to refrain from conducting AEC tests See, for example, TeliaSonera (n 83), para 43.
164 As recently observed by Damien Neven, a main function of the AEC
in these cases, see especially Kadar and Holzwarth (n 52) 333, pointing to the
high consumption of resources, the problems in accurately calculating the principle is that it marks ‘a red line’—see Neven (n 73) 565 and throughout.
contestable share and the general ‘administrability’. However, Neven points out that the ‘red line’ principle may lead to type I errors
156 According to long-standing case-law, the AEC test is just one method- (at pp 576, 577–578) and suggests that this concern should be reflected in future
ological option—see Post Danmark II (n 83), paras 56–61; Servizio Elettrico guidelines.
165 See 2008 Priorities Guidance, para 24; Policy Brief, fn 44; Opinion of
Nazionale (n 38), para 81; Unilever Italia (n 54), paras 57–58. Even if the Com-
mission does not conduct an AEC test itself, it is obliged to engage with an AEC AG Kokott in Google Shopping (n 119), para 194. Also Neven (n 73) 570; Ger-
test presented by the dominant undertaking as part of its defence. From the main Gaudin and Despoina Mantzari, ‘Google Shopping and the As-Efficient-
Commission’s perspective, this may, however, be more manageable and less Competitor Test: Taking Stock and Looking Ahead’ (2022) 13 Journal of Euro-
risky (further on this: see below, IV. B. 2.). pean Competition Law & Practice 125, 126–127; Pablo Ibáñez Colomo, ‘Anti-
157 Whether these limits justify dispensing with an AEC test beyond the competitive Effects in EU Competition Law’ (2020) 17 Journal of Competition
established categories of predatory pricing and margin squeeze, or rather calls Law & Economics 309, 339.
166 European Superleague (n 54), para 131.
for an AEC principle and AEC tests that are framed not as a generally applicable
167 European Superleague (n 54), para 131, with reference to Generics (UK) (n
principle, but as contingent on a specific theory of harm, is currently being
debated. In favour of a framing of the AEC principle and tests as contingent on 43), paras 154–157. Inter alia, the wording in European Superleague encompasses
varying theories of harm: Neven (n 73) 580. cases in which the dominant undertaking denies rivals a minimum scale—on
158 This has recently been pointed out by commentators, see Barbera, this, see: Benjamin Klein, ‘Exclusive Dealing as Competition for Distribution
Fajardo Acosta and Klein (n 131) 587 and 591; Neven (n 73) 574. “On the Merits”’ (2003) 12 George Mason Law Review 119.
159 168 Post Danmark II (n 83), para 59.
Neven (n 73) 579.
160 169 Emphasised by Ibáñez Colomo (n 62) 407.
Neven (n 73) 577–578.
161 170 The same may be true where efficiencies result from the scope of
Barbera, Fajardo Acosta and Klein (n 131) 592.
162 For the fact that the concept of ‘competition on the merits’ is broader products that the dominant undertaking offers, but hypothetical competitors
than the AEC test see Unilever Italia (n 54), para 39: an abuse is established either do not and cannot offer the same range of products—see Neven (n 73) 570–571.
171 Opinion of AG Kokott in Google Shopping (n 119), para 195.
‘where the conduct complained of produced exclusionary effects in respect of
172 See the Amending Communication, rephrasing para 23 of the 2008
competitors that were as-efficient as the perpetrator of that conduct in terms of
cost structure, capacity to innovate, quality’; or ‘where that conduct was based Priorities Guidance; Neven (n 73) 566.
on the use of means other than those which come within the scope of “normal” 173 Barbera, Fajardo Acosta, and Klein (n 131) 592 and 593, discussing
competition, that is to say, competition on the merits’ (emphasis added). See when—and when not—the goal of promoting contestability should prevail over
also Opinion of AG Kokott in Google Shopping (n 119), para 196: ‘The ACE test the principle that artificially keeping less efficient firms in the market will not
is not generally applicable, let alone an essential prerequisite for determining be in the interest of efficiency and consumers.
14 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

IV. Effects analysis—standard of proof, theoretical level,179 it is sufficiently f lexible to accommodate that
assessment of evidence, and analytical standard in its practical implementation.180
shortcuts Theoretically, a f lexible, yet relatively moderate substantive
threshold of anticompetitive risk must therefore be proven under
The brief sketch of the basic legal structure of Art. 102 TFEU shows a rather demanding standard of proof. In practice, the Union
a considerable degree of consolidation in some areas of the law; Courts tend to conf late the substantive threshold and the stan-
significant uncertainties persist in others. Most importantly, the dard of proof, however—in particular when they assume that,
principles that are to guide the effects analysis remain a puzzle. in establishing ‘likely’ or ‘probable’ anticompetitive effects, the
While there is room for future guidelines to clarify the law and norm addressee will benefit from a presumption of innocence.
to push it into an appropriate direction, the substantive law on In fact, the presumption of innocence applies to the standard of
abuse will always remain a ‘standard’, not a ‘rule’.174

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

proof only, but presupposes that the substantive legal threshold
There is reason therefore for future guidelines to venture for ‘potential anticompetitive effects’ under Art. 102 TFEU is pre-
beyond the incoherencies and uncertainties of substantive defined.181 If the Union Courts apply the presumption of inno-
law. Arguably, the uncertainties surrounding the evidentiary cence to the determination of potential anticompetitive effects
standards likewise contribute to exceptionally long Art. 102 nonetheless, this can only mean that a demanding standard of
proceedings. In the following text, we propose to clarify first proof will apply with regard to the facts that substantiate the exis-
the standard of proof (A.) and subsequently the conceptual tence of a relevant and sufficiently concrete risk to undistorted
framework for the assessment of evidence (B.). Our focus will be competition.182
on analytical shortcuts for determining potential anticompetitive The causal hypothesis meant to establish an infringement of
effects (C.). Our main claim is that a major, if not the most Art. 102 TFEU is not treated as a fact to be established to that
important, contribution of future guidelines on exclusionary same standard of proof. It is rather treated as a specification of the
abuses to a more effective approach to Art. 102 TFEU may be to substantive norm for which the Commission possesses some mar-
further clarify and develop these analytical shortcuts. Ultimately, gin of appraisal. Consequently, a ‘manifest error of assessment’
the evidentiary framework for the effects analysis should meet standard applies. This is how Unilever Italia must be read when it
a number of requirements: in line with sufficiently reliable finds that the demonstration of effects must ‘be based in tangible
economic experience and knowledge, the analysis must establish evidence which establishes, beyond mere hypothesis, that the
a concrete risk of anticompetitive foreclosure; at the same time, practice in question is actually capable of producing such effects,
it must remain manageable such as to enable an enforcement since the existence of doubt in that regard must benefit the
that considers the enforcing institutions’ constraints of time undertaking which engages in such a practice’:183 the requirement
and resources and the need to enforce Art. 102 TFEU in a timely of ‘tangible evidence’ applies to the facts—but the evidence must
fashion. prove no more than that the relevant conduct is capable of produc-
ing anticompetitive effects ‘beyond mere hypothesis’.184
A. The standard of proof
As a matter of substantive law, the finding of ‘potential anticom- B. Assessment of the evidence
petitive effects’ presupposes that a concrete and real, not merely When it comes to the assessment of the evidence, the principle
hypothetical, risk of anticompetitive effects is established. The of unfettered evaluation of evidence applies: generally, the Union
Union Courts have rejected a specification of the level of risk Courts are free in assessing the probative value they assign each
in probabilistic terms (see above, III. B. 3. 2.). In principle, this single piece of evidence (or a set of evidence) and, ultimately,
substantive threshold is to be distinguished from a specification
of the standard of proof, i.e. the degree of probability with which
the presence of relevant preconditions of an abuse must be proven 179 Kalintiri (n 175) 81–85.
180 Kalintiri (n 175) 85; similarly Fernando Castillo de la Torre and Eric
as a matter of procedural law:175 According to Art. 6 ECHR, pro-
Gippini Fournier, Evidence, Proof and Judicial Review in EU Competition Law (Edward
ceedings which allow for the imposition of fines176 and which Elgar 2017), para 2.029: ‘generally judges see no essential difference’. In a few
therefore qualify as quasi-criminal177 are governed by a ‘beyond cases, the ‘beyond a reasonable doubt’ standard has been explicitly referred
to—see Opinion of Mr Vesterdorf acting as AG in Case T-1/89 Rhône-Poulenc v
a reasonable doubt’ standard.178 The Union Courts frequently Commission, EU:T:1991:38, p II-954; similarly Joined Cases 29/83 and 30/83 CRAM
refer to a ‘full conviction’ standard instead. While it may not be and Rheinzink/Commission, EU:C:1984:130, para 16.
181 See, for example, Unilever Italia (n 54), para 42 (with further references to
synonymous with a ‘beyond a reasonable doubt’ standard on a case law). In other cases, it remains unclear whether the Courts aim to specify
the level of risk required by Art. 102 TFEU or rather the standard of proof—
Critical in this regard: Castillo de la Torre (n 42) 153–154.
182 For the standard of proof regarding such facts see Opinion of AG
Kokott in Case C-8/08 T-Mobile Netherlands and Others, EU:C:2009:110, fn 60: ‘the
standard of proof determines the requirements which must be satisfied for
174 For the general distinction see: Louis Kaplow, ‘Rules Versus Standards: facts to be regarded as proven.’
An Economic Analysis’ (1992) 42 Duke Law Journal 557. 183 Unilever Italia (n 54), para 42; citing Case 27/76 United Brands v Commission,
175 See Andriani Kalintiri, Evidence Standards in EU Competition Enforcement EU:C:1978:22, para 265, and Joined Cases C-89/85, C-104/85, C-114/85, C-116/85,
(Hart 2019) 72–73; 79–80. If both the substantive threshold and the standard C-117/85 and C-125/85 to C-129/85 Ahlström Osakeyhtiö and Others v Commission,
of proof were interpreted in probabilistic terms, this could theoretically lead EU:C:1993:120, para 126.
to a finding that relatively low probabilities of anticompetitive effects would 184 If read in this light, Unilever Italia does not interfere with the principle
suffice for the finding of an abuse. For a different perspective see Deutscher (n that it is for EU law to define the substantive requirements of an abuse as
94) 504 ff. well as the allocation of the burden of proof, but for national procedural law
176 For this power of the Commission in proceedings under Art. 7 of to define the standard of proof (see recital 5 of Regulation 1/2003) and to
Regulation 1/2003 see Art. 23 of Regulation 1/2003. decide on the rules on evidence (see also Case C-211/22 Super Bock Bebidas,
177 For the general criteria for qualifying proceedings as ‘criminal’ within EU:C:2023:529, para 55, with a view to Art. 101 TFEU). What is more, given that,
the meaning of Art. 6 ECHR see Engel and Others v The Netherlands, App no at the national level, Art. 102 TFEU may be enforced in a purely administrative
5100/71 (ECtHR 8 June 1976), cf. Kalintiri (n 175) 25 ff. See for competition procedure, with no threat of fines (See, for example, § 54(1) GWB (Gesetz gegen
authorities’ proceedings Menarini Diagnostics SRL v Italy, App no 43509/08 (ECtHR Wettbewerbsbeschränkungen; German Act against restraints of competition)),
27 September 2011). a ‘beyond reasonable doubt’-standard cannot be generalised. National rules on
178 Minelli v Switzerland (1983) Series A no 4, para 37; Poncelet v Belgium App standard of proof and on evidence must, however, be in line with the general
no 44418/07 (ECtHR 30 March 2010), para 50; Garycki v Poland App no 14348/02 principles of equivalence and effectiveness—cf. Case C-453/99 Courage and
(ECtHR, 6 February 2007), para 68. Crehan, EU:C:2001:465, para 29.
 Journal of European Competition Law & Practice, 2024 | 15

the body of evidence as a whole, with possible interconnections to be proven and a fact B—in our case: potential anticompetitive
and corroborations between the different pieces of evidence.185 effects—which is inferred. Where such a link is known to exist,
The law on evidence does not stipulate a general prioritisation factual inferences may justify an abbreviated assessment of the
of quantitative over qualitative evidence. From an evidentiary evidence for a Court to reach a full conviction that a certain type
perspective, both come with equal weight. What ultimately mat- of conduct will likely lead to anticompetitive effects. A defendant
ters is that, based on the totality of the evidence, the Court is may try to raise doubts about the strength of the empirical link
fully convinced that the relevant conduct is capable of producing between the relevant set of facts and the fact inferred, either
anticompetitive effects. generally or in the context of the specific, potentially atypical,
Beyond these basic principles, a conceptual framework for case; or may resort to challenge the plausibility of the inferred
the assessment of evidence—and in particular circumstantial fact directly—for example by conducting an AEC test to show

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

evidence on potential anticompetitive effects—is notably absent. that, despite the factual inference, anticompetitive effects are
Nor have the Union Courts developed a clear terminology. This unlikely. In such a case, the Commission will need to engage with
has led to a significant degree of uncertainty—an uncertainty that the evidence brought forward. But provided that the existence
induces the Commission to engage in an ever more expansive of an empirical link and/or its applicability in the given setting
effects analysis in order to reduce the risk of later losing the case is not fundamentally called into doubt, the initial inference will
in Court. continue to carry (some) evidentiary weight.
Although factual inferences may reduce the scope of the fac-
1. Inferences, presumptions, and other analytical tual inquiry that is required to come to a full conviction of fact
shortcuts—preliminary remarks on terminology B, the law of evidence sometimes goes further and recognises a
In the majority of cases, the Commission will focus its inquiry lowering of the evidentiary threshold. This is where ‘real’ pre-
on potential effects on the competitive structure. As potential sumptions (or ‘legal presumptions’) come into play: they direct
effects cannot be proven directly, some sort of indirect evidence the Court to conclude, from the presence of fact A (or a set of facts
will be required: proof of anticompetitive effects will be based A1, A2, and A3), that fact B (e.g. potential anticompetitive effects)
on indicia from which it can be inferred that anticompetitive is likewise proven, irrespective of the strength of the empirical link
effects are sufficiently likely. The selection of indicia will derive between fact A and fact B. This is justified because presumptions
from a causal hypothesis—frequently referred to as the ‘theory of do not (or not primarily) follow an empirical rationale. Rather, they
harm’—which must be plausible and sufficiently reliable. Absent are based on normative grounds: generally speaking, they shall
an established legal presumption that would direct the Court ensure the well-functioning of the law and law enforcement as
to infer anticompetitive effects on a normative basis, the causal an administrative system. In some cases, legal presumptions shall
hypothesis, together with the proof of the facts on which it is elicit information from the party against which the presumption
based, must lead to the full conviction of the Court that anti- works, because that party has better access to the relevant infor-
competitive effects are sufficiently likely (see above, IV. A.). The mation.187 Frequently, they are meant to lower the administrative
defendant can question the general plausibility and/or reliability cost of proving a given fact, such as to ensure the manageability
of the theory of harm; it may argue that it does not fit the special of proof, reduce overall error cost, and increase legal certainty. In
setting of the case, or it may try to show that the facts do not that case, legal presumptions—like factual inferences—will have
support the causal hypothesis. Finally, it may try to disprove some empirical grounding, but the empirical link between fact A
anticompetitive effects, e.g. based on an AEC test. The Courts and fact B to be presumed may be weaker than what would be
will ultimately decide whether they are convinced by the causal required in order to recognise a factual inference. In any case,
hypothesis and by the facts that are presented to substantiate it. it will be of the essence that fact A is readily observable by the
Sometimes, an abbreviated analysis may suffice to infer poten- norm addressee and the law enforcement authority at relatively
tial effects—namely if, based on the economically informed expe- low cost.
rience of the Court, the combination of a small, selected number In some cases, legal presumptions are rebuttable, in others
of factors (e.g. a certain type of conduct, combined with a certain irrebuttable. Either way, the undertaking under investigation may
degree of dominance and a certain market coverage of a practice) challenge the presence of fact A, i.e. of the preconditions of the
leads to a high probability of harm to competition. We refer to presumption. In case of a rebuttable presumption, it may also
such inferences as ‘factual inferences’.186 The legitimacy of a challenge fact B, but the burden of proof is reversed. Only if it fully
factual inference depends on the existence of a strong empirical proves the absence of fact B will the burden of proof shift back to
link between a pre-selected fact A (or a set of facts A1, A2, and A3) the Commission. However, before the burden of proof is reversed,
thought must be given to whether the circumstances to be proven
are within the sphere of the party bearing the burden of proof, i.e.
185 For an attempt to sketch the theoretical framework for the assessment
of evidence see Opinion of Mr Vesterdorf acting as AG in Rhône-Poulenc v Com-
whether they are readily observable to that party. Typically, this
mission (n 180), p II-954. The approach has been adapted by the Union Courts— will not be the case for the non-existence of potential anticom-
see cf. Joined Cases T-25/95, T-26/95, T-30/95, T-31/95, T-32/95, T-34/95, T-35/95,
petitive effects. Consequently, the burden of proof for potential
T-36/95, T-37/95, T-38/95, T-39/95, T-42/95, T-43/95, T-44/95, T-45/95, T-46/95, T-
48/95, T-50/95, T-51/95, T-52/95, T-53/95, T-54/95, T-55/95, T-56/95, T-57/95, T- anticompetitive effects will normally not be reversed. In some
58/95, T-59/95, T-60/95, T-61/95, T-62/95, T-63/95, T-64/95, T-65/95, T-68/95, T- instances, such effects may, however, be irrebuttably presumed.
69/95, T-70/95, T-71/95, T-87/95, T-88/95, T-103/95 and T-104/95 Cimenteries CBR
and Others v Commission, EU:T:2000:77, para 1838. For the holistic approach to
the assessment of evidence see Castillo de la Torre and Gippini Fournier (n 180) 2. Indirect proof of potential effects
78–86; see e.g. Case T-321/05 AstraZeneca v Commission, EU:T:2010:266, para 477
with further references to the case law. The above categorisation allows us to describe more precisely how
186 Sometimes, these inferences are also referred to as ‘factual the preconditions for proving potential anticompetitive effects
presumptions’. To clearly distinguish factual inferences from legal
presumptions, we exclusively use the term ‘factual inferences’ in the following
text. Contrary to ‘factual inferences’, legal presumptions do not have a purely
empirical, but (also, sometimes primarily) a normative basis (e.g. the goal to
elicit information from the defendant or to react to the cost and/or difficulties 187 For this purpose of presumptions see Cyril Ritter, ‘Presumptions in EU
of access to evidence). competition law’ (2018) 6 Journal of Antitrust Enforcement 189, 206.
16 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

have shifted: in the past, the ECJ had been ready to accept irre- explanation and must be considered in the overall assessment
buttable legal presumptions of anticompetitive effects for some of the case. Simultaneously, where errors in the AEC test are
types of conduct (in particular exclusionary rebates and exclu- exposed,194 or where the AEC test was not applicable (see above,
sive dealing/exclusivity payments) based on the type of conduct III. B. 4. 1.),195 the Commission may continue to fully rely on the
alone and had thereby lowered the evidentiary threshold from factual inference.
proof of a concrete likelihood to an abstract risk of anticompet- With regard to other types of conduct, the Union Courts have
itive effects. The more recent jurisprudence has rejected these recognised factual inferences already some time ago. Predatory
‘normative’ presumptions, however. Arguably, the former pre- pricing and margin squeeze cases may be considered to fall into
sumptions have been turned into factual inferences—which may, this category, where an abuse will generally be inferred from
however, depending on the type of conduct, need to be based on an AEC test that shows that as-efficient competitors could not

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

a more complex set of facts (e.g. the degree of dominance or the compete.
duration and/or market coverage of the practice). Whether or not factual inferences apply, recent case law is
A comparison between Hoffmann-La Roche and the ECJ’s Intel explicit that ultimately, the full conviction of the Court of the
decision of 2017 illustrates this shift: based on a relevant prob- concrete potential of anticompetitive effects is required.
ability—i.e. a general risk—that loyalty rebates and exclusive The Commission must verify whether the market conditions
dealing may anticompetitively foreclose competition, Hoffmann- on which its theory of harm is based are met or at least likely
La Roche established an irrebuttable legal presumption that the to arise in the specific context of the case.196 In Unilever Italia,
adoption of such practices by a dominant undertaking constitutes the ECJ, with a view to the showing of anticompetitive effects of
an abuse.188 The dominant undertaking could plead for an objec- exclusivity clauses, demanded an analysis of ‘factors specific to
tive justification or an efficiency defence. But it could not rebut the circumstances of the case, such as the extent of that conduct
the possibility of potential anticompetitive effects. The critique of on the market, capacity constraints on suppliers of raw materials,
this presumption as ‘formalistic’ and ‘disconnected from modern or the fact that the undertaking in a dominant position is, at least,
economics’189 overlooked its normative basis: the presumption for part of the demand, an inevitable partner’.197 Generally, the ECJ
facilitated enforcement, allowed for a high degree of legal cer- asks for ‘specific, tangible points of analysis and evidence’.198
tainty, and was arguably based on implicit assumptions on what Whenever the dominant undertaking submits, supported by
level of risk EU competition law should accept regarding adverse evidence, that anticompetitive effects are unlikely or implausible
effects on competition. A comprehensive economic assessment in the circumstances of a given case, the Commission has to
would therefore need to be based on a comparative analysis of hear the submission and review the evidence. This is part of
error costs.190 the Commission’s evidentiary burden—it must fully convince the
In its Intel judgment of 2017, the ECJ has reversed course, Court of the likely anticompetitive effects. But it also follows from
however. While claiming to ‘clarify’ the law on loyalty rebates general procedural principles—in particular the right to be heard
and exclusive dealing,191 it has turned the irrebuttable presump- of the undertaking under investigation. On the procedural side,
tion into a different type of inference—but contrary to the GC’s the Union Courts have emphasised the Commission’s obligation
reading192 not into a rebuttable presumption—but into a fac- to examine ‘carefully and impartially all the relevant aspects of
tual inference of (concrete) potential anticompetitive effects:193 the Individual case’.199 So far, the requirement of the ‘relevance’200
the burden of proof is not reversed, such that the dominant
undertaking would need to proof the implausibility of potential 194 One may assume that verifying an AEC test presented by the dominant
anticompetitive effects—frequently an impossible task. Rather, undertaking is much less time-consuming and resource-intensive than con-
ducting an AEC test of its own.
where the dominant undertaking casts doubt on the potential for 195 See Opinion of AG Kokott in Google Shopping (n 119), para 196: ‘in so far
anticompetitive effects by presenting substantiated evidence— as [the AEC test] is not applicable, neither the Commission nor the General
e.g. an AEC test aiming to show that as-efficient competitors Court can be compelled to consider arguments by the dominant undertaking
in connection with its use’.
are not foreclosed—the Commission has to revisit whether it 196 See Servizio Elettrico Nazionale (n 38), para 70: If a competition authority
can prove potential anticompetitive effects to the full conviction assumes a causal link between a certain line of conduct and (potential) restric-
tions of competition, the specific market conditions on which the assumption
of the Court. Where the factual inference is not refuted, it will is based must be met or at least likely to arise.
continue to carry evidentiary weight. If the dominant undertaking 197 Unilever Italia (n 54), para 44: When assessing whether an undertaking’s
conduct is capable of restricting effective competition, a competition authority
has shown special circumstances, the factual inference may be may rely on the guidance from economic sciences, confirmed by empirical
weakened and need to be refined. An AEC test—if conducted or behavioural studies. However, factors specific to the circumstances of the
in a reliable and error-free manner—does not, as such, disprove case must be taken into account, ‘such as the extent of that conduct on the
market, capacity constraints on suppliers of raw materials, or the fact that
the factual inference, but may constitute a factor in need of the undertaking in a dominant position is, at least, for part of the demand,
an inevitable partner’. In Unilever Italia, the ECJ has required such analysis
irrespective of the defendant coming forward with substantiated evidence
suggesting the absence of anticompetitive effects. It is unclear whether the
188 Hoffmann-La Roche (n 55), paras 89–90. ECJ thereby strives to establish a qualified standard for factual inferences of
189 See, for example, Akman (n 81), p 6, with further references to the wide- anticompetitive effects.
spread criticism of an overly ‘formalistic’ jurisprudence on Art. 102 TFEU in the 198 European Superleague (n 54), para 130.
early years. 199 Servizio Elettrico Nazionale (n 38), para 52. See also Post Danmark II (n 83),
190 See, to that respect, also Wils (n 150) 428, arguing that the degree of para 68. Critically Castillo de la Torre (n 42) 174 ff.
legal certainty and the corresponding allocation of risk should be considered 200 See, for example, Servizio Elettrico Nazionale (n 38), para 72: Whether a
in interpreting Art. 102 TFEU. practice is or was capable of restricting competition ‘must be assessed having
191 Case C-413/14 P Intel (n 61), para 138: ‘that case-law must be further regard to all the relevant facts (see, to that effect, judgments of 30 January
clarified in the case where the undertaking concerned submits, during the 2020, Generics (UK) and Others, C-307/18, EU:C:2020:52, paragraph 154, and of 25
administrative procedure, on the basis of supporting evidence, that its conduct March 2021, Slovak Telekom v Commission, C-165/19 P, EU:C:2021:239, paragraph
was not capable of restricting competition and, in particular, of producing the 42).’ See also: Unilever Italia (n 54), para 40: ‘it is for the competition authorities
alleged foreclosure effects’. to demonstrate the abusive nature of conduct in the light of all the relevant
192 See Case T-286/09 RENV Intel (n 152), paras 124, 129. factual circumstances surrounding the conduct in question (judgments of 19
193 Case C-413/14 P Intel (n 61), paras 138–140. See also Castillo de la Torre (n April 2012, Tomra Systems and Others v Commission, C-549/10 P, EU:C:2012:221,
42) 180–181; Raffaele Di Giovanni Bezzi, ‘Anticompetitive Effects and Allocation paragraph 18, and of 12 May 2022, Servizio Elettrico Nazionale and Others, C-377/20,
of the Burden of Proof in Article 102 Cases: Lessons from the Google Shopping EU:C:2022:379, paragraph 72), which includes those highlighted by the evidence
Case’ (2022) 13 Journal of European Competition Law & Practice 112, 122–124. adduced in defence by the undertaking in a dominant position.’
 Journal of European Competition Law & Practice, 2024 | 17

of the dominant undertaking’s submission has been interpreted examination of actual effects. This may be true in atypical and
generously. In order to keep the evidentiary proceedings manage- novel abuse cases: the potential of some practices to adversely
able, a rigorous relevance check would be called for: only if the affect competition has not yet been sufficiently explored, and
evidence submitted by the dominant undertaking is capable of relevant theories of harm may be ambiguous in their prediction.
undermining the full conviction regarding the likelihood of effects For example, potential effects may depend on typical consumer
should the Commission be required to engage in further inquiry behaviour. In these cases, actual effects may provide the most reli-
and analysis.201 able indication. Atypical bundling/tying is a case in point: when
Regarding the relevance of an AEC test presented by the analysing Microsoft’s bundling of the Windows operating system
dominant undertaking,202 the Union Courts’ recent jurisprudence with the Windows Media Player, the Commission considered that,
suggests that it will always meet the initial threshold of relevance: contrary to the classical tying case, PC users could and did to a

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

a showing that as-efficient competitors retain the ability to certain extent download third-party media players through the
compete may cast doubt on the likelihood of anticompetitive Internet free of charge. Therefore, the Commission could not
effects.203 Nonetheless, even if conducted in a methodologically assume without further analysis that a tying of the Media Player
reliable and error-free manner, such an AEC test is not necessarily with the operating system would, by its very nature, be liable to
determinative: For example, the Commission may demonstrate foreclose competition.207 This approach was confirmed in Google
that, given the setting of the case, an abuse already follows from Android:208 as it was easy for mobile users to obtain general
the foreclosure or marginalisation of less efficient competitors. search or browser apps competing with those that were subject
The AEC test would then be irrelevant for the outcome of the case. to Google’s tying policy, the Commission established that that
policy had the actual effect of making it ‘harder for competing
3. Actual effects search services to gain search queries and the revenues and data
An infringement of Art. 102 TFEU does not presuppose proof of needed to improve their services’, that they increased ‘barriers
actual anticompetitive effects (see above, III. B. 3. 2.).204 However, to entry by shielding Google from competition from other search
direct proof of actual effects may sometimes be the best evidence services’, and that they reduced ‘incentives for the innovations
available, in particular where the allegedly abusive conduct has that competitors marketing specialised search services . . . wished
persisted over a longer period of time and where the exclusionary to offer’.209
effects are obvious.
Frequently, the causes of any actual effects will be contro- C. Towards novel analytical shortcuts
versial and the attempt to offer additional evidence to resolve In the search for a ‘more workable approach’ to Art. 102 TFEU, a
such controversy will tend to prolong proceedings, however. The clarification of the evidentiary rules and requirements in future
Commission will then rather opt for showing potential effects on guidelines on exclusionary abuses holds great potential. While the
the basis of indirect evidence. Union Courts’ recent jurisprudence has not shown much open-
It is a separate question whether the absence of actual effects ness or sympathy for attempts to re-establish conduct-based legal
may qualify as indirect proof that a certain line of conduct was presumptions with a normative grounding,210 there seems to be
incapable of producing anticompetitive effects. As a general mat- ample room for developing factual inferences tailored to different
ter, the ECJ has dismissed such allegations.205 Yet, sometimes, types of abuse (1.). Simultaneously, the Union Courts’ dislike of
and if backed up by additional evidence, the absence of actual conduct-based presumptions need not extend to presumptions
anticompetitive effects may indicate that anticompetitive effects with a different peg: in particular, presumptions based on the
were unlikely in the circumstances of the case.206 If not delimited ‘object’ of a given type of conduct or on the identification of an
by clear and transparent criteria, the latter jurisprudence has exclusionary strategy have recently attracted increasing interest,
the potential to significantly prolong proceedings: as a practical partly under the term of ‘naked exclusions’. The consequence of
matter, the principle that the Commission is not required to finding an abuse ‘by object’ may be that no effects analysis is
inquire into actual effects would be undermined. For that reason, required (2.). In other settings, an effects analysis of some kind will
a dominant undertaking that offers proof of a lack of actual be indispensable to establish an infringement of Art. 102 TFEU,
anticompetitive effects should be required to also refute or at but an abbreviated analysis may suffice. This may be the case,
least weaken the theoretical link between the relevant conduct in particular, where the anticompetitive effects clearly outweigh
and potential anticompetitive effects before the Commission’s any pro-competitive justification (3.). Finally, where the relevant
obligation to further investigate is triggered. conduct does not fall within the scope of ‘competition on the
Finally, there are exceptional cases where—contrary to the rule merits’ for other reasons, the evidentiary threshold for the proof
that no actual effects need to be shown—potential effects are of potential anticompetitive effects may legitimately be lowered
difficult to prove to the full conviction of the court without a close to a showing of ‘plausible adverse effects’ (4.). While the analytical
shortcuts discussed under 1. and 2. can be categorised as factual
201 For a critical perspective regarding the ‘all circumstances considered’-
inferences and legal presumptions, respectively, the shortcuts
jurisprudence see Castillo de la Torre (n 42) 175–176. developed under 3. and 4. do not qualify as inferences of either
202 See, for example, the undertaking’s presentation of an AEC test in kind: in these settings, anticompetitive effects are not inferred,
Qualcomm (Exclusivity payments) (Case AT.40220) Commission Decision of 24
January 2018, recital 487. but their proof is otherwise facilitated.
203 As a matter of policy, a different approach would also seem feasible: To
ensure the administrability of abuse proceedings, one may want to require the
defendant to first show the inadequacy of the Commission’s methodological
choice, or that the outcome of the Commission’s analysis was inconclusive,
before obliging the Commission to consider a different type of test. This 207 Microsoft (Case COMP/C-3/37.792) Commission Decision of 24 March
approach would protect the Commission’s margin of judgment in choosing an 2004, recital 841. See also Case T-201/04 Microsoft (n 89), para 977.
adequate methodology to establish anticompetitive effects in a given case. 208 Case T-604/18 Google Android (n 50), para 295.
204 TeliaSonera (n 83), para 64; Post Danmark II (n 83), para 66; Servizio Elettrico 209 Case T-604/18 Google Android (n 50), para 294.
Nazionale (n 38), para 54, and Unilever Italia (n 54), para 41. 210 See, however, Laitenberger and Kröger (n 19) 632, pointing towards
205 Servizio Elettrico Nazionale (n 38), para 54; Unilever Italia (n 54), para 41. suggestions for a well-dosed . . . ‘form based’ or—as some might prefer to call
206 Servizio Elettrico Nazionale (n 38), para 56. it—‘more normative approach’.
18 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

1. Towards a novel generation of factual inferences whether that categorisation may usefully be transplanted to Art.
Legal presumptions as defined above lower the evidentiary 102 TFEU219 —the concept of ‘restrictions by object’ is not a model
threshold based not only on an empirical link between fact A and of clarity even in the context of Art. 101 TFEU: in the past, one
fact B, but also, and importantly, on some normative goal they might have argued that it established an irrebuttable presump-
are meant to promote.211 Factual inferences, by contrast, have a tion of illegality for certain types of typically harmful agreements
purely empirical basis. The evidentiary threshold—in our case: under Art. 101(1) TFEU—although a presumption subject to a
full conviction of the court of potential anticompetitive effects— possible exemption under Art. 101(3) TFEU. In some more recent
remains unchanged. But based on a factual inference, i.e. the judgments, it seems to serve as a justification for a quick-look
economically informed experience of the court that the presence effects analysis with the aim to determine whether, in a concrete
of certain facts will typically lead to anticompetitive effects, setting, a typically harmful agreement has, or has not, plausible

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

the requisite scope of the factual inquiry may be narrowed.212 pro-competitive effects.220 Whenever plausible and sufficiently
According to Pablo Ibáñez Colomo, these inferences213 should no significant pro-competitive effects are established, a full effects
longer be based on the form of the conduct alone, but include analysis will be required.221
some context-specific indicia214 which would need to be further The basic analytical framework underlying the latter
tailored to the different types of abuse. Given their empirical approach—the finding that a typically harmful practice for which
basis, factual inferences would need to be based on sound no plausible pro-competitive effects can be demonstrated may be
economic insights which would, however, need to be translated presumed to be capable of producing anticompetitive effects—
into manageable legal tests.215 Ideally, the Commission’s future is not limited to Art. 101 TFEU. It has also been applied in Art.
guidelines on exclusionary abuses would propose an initial set 102 cases.222 The rationale is partially empirical, considering
of tests. But arguably, the development of manageable factual that conduct with an anticompetitive object comes with a high
inferences would be a continuous joint endeavour of competition probability of potential anticompetitive effects. However, it is
lawyers and economists. backed up by a normative component, given that it is hard to see
why such conduct should be protected. Because of its normative
2. ‘Naked exclusion’—no effects analysis in case of an foundation, this shortcut qualifies, in our categorisation, as a legal
anticompetitive object presumption.
2.1. ‘Naked exclusion’ and ‘abuses by object’
Beyond factual inferences, there may be room for the develop-
2.2. How to identify naked exclusions
ment of new kinds of presumptions. In particular, the idea of a Where the conduct of a dominant undertaking qualifies as a
presumption based on the absence of a pro-competitive rationale ‘naked exclusion’, the proof of an abuse is significantly facilitated.
is gaining prominence. As no effects analysis is required, the concept has the potential
A first indication can be found in the 2008 Priorities Guidance: to contribute to a more effective enforcement of Art. 102 TFEU—
‘If it appears that the conduct can only raise obstacles to com- provided that cases of ‘naked exclusion’ can be identified with
petition and that it creates no efficiencies, its anticompetitive ease and a sufficient degree of legal certainty.
effects may be inferred’.216 According to this wording, the legal Some examples of ‘naked exclusion’ are mentioned in the 2008
treatment of the so-called ‘naked exclusions’ would amount to Priorities Guidance: no effects analysis shall be required where a
a quick-look effects analysis: where plausible pro-competitive dominant undertaking ‘prevents its customers from testing the
effects are absent (i.e. no ‘competition on the merits’) and a products of competitors or provides financial incentives to its
general potential for anticompetitive effects is obvious, ‘it is not customers on condition that they do not test such products, or
necessary for the Commission to carry out a detailed assessment pays a distributor or a customer to delay the introduction of a
before concluding that the conduct in question is likely to result competitor’s product’.223 A part of the Intel case is illustrative
in consumer harm’.217 ‘Naked exclusions’ might then amount to of the latter: according to non-contested findings of the Com-
just another sub-category of factual inferences. mission, Intel had paid three computer manufacturers (HP, Acer,
By contrast, parts of the literature have referred to these set- Lenovo) for halting or delaying the launch of certain products con-
tings as ‘abuses by object’, likening them to ‘restrictions by object’ taining competitors’ x86 CPUs, thereby limiting the sales channels
in Art. 101 TFEU.218 However—and irrespective of the debate available for these products.224 The Commission characterised
this conduct as a ‘naked restriction’ and concluded that, as the

211 Often, the normative goal is accompanied by an empirical link in cases 219 See, on the one hand, Case T-612/17 Google Shopping (n 86), paras 435 ff,
of legal presumptions, too. This link might, however, not be strong enough to explicitly negating the existence of ‘abuses by object’. See, however, European
justify a factual inference, see above IV. B. 1. Superleague (n 54), para 131, where the possibility to prove an anticompetitive
212 Efforts in this direction also appear at Kadar and Holzwarth (n 52) 333: object is mentioned, albeit without clarifying the legal preconditions and
‘Setting up clear-cut principles of enforcement based on experience and sound implications of such a finding.
economics is not incompatible with an effects-based approach’. 220 Such a view is suggested by Super Bock (n 184), paras 34–37.
213 Pablo Ibáñez Colomo speaks of ‘proxies’. Within the terminological cate- 221 See Super Bock (n 184), para 36.
gorisation proposed above, these ‘proxies’ would qualify as ‘factual inferences’. 222 See, for example, Case T-203/01 Michelin v Commission (Michelin
214 Like, for example, the degree of dominance, market coverage and dura- II), EU:T:2003:250, para 241; Case T-340/03 France Télécom v Commission,
tion of the practice etc. See Ibáñez Colomo (n 92) 619–620. EU:T:2007:22, para 195: ‘showing an anticompetitive object and an anticompet-
215 José Luís Cruz Vilaça, ‘The intensity of judicial review in complex itive effect may, in some cases, be one and the same thing. If it is shown that
economic matters—recent competition law judgments of the Court of Justice of the object pursued by the conduct of an undertaking in a dominant position
the EU’ (2018) 6 Journal of Antitrust Enforcement 173, 187: ‘The great challenge is to restrict competition, that conduct is liable to have such an effect.’ For a
for the jurist, and in particular for the judge, is to turn economic theories into different view—relying on the wording of Art. 102 TFEU as compared with the
solid legal criteria, capable of securing the clarity of the concepts and their wording of Art. 101 TFEU—see Akman (n 81), p 11.
adaptability to a complex reality, as well as to enhance legal certainty and 223 2008 Priorities Guidance, para 22.
predictability in the application of the law’. 224 See Commission Decision in Intel (n 18), recitals 1641 ff. See also:
216 2008 Priorities Guidance, para 22. Intel (Case AT.37990) Commission Decision of 22 September 2023 (imposing
217 2008 Priorities Guidance, para 22. fines for the naked restriction after the GC, in Case T-286/09 RENV Intel (n
218 See Ibáñez Colomo (n 62) 413–414. Others have objected, pointing to the 152), had fully annulled the Commission’s 2009 Decision in Intel (n 18), but
wording of Art. 102 TFEU which does not differentiate between abuses by object had simultaneously confirmed that Intel’s payments to the three computer
or effect—see Akman (n 81), p 11. manufacturers amounted to an abuse of dominance).
 Journal of European Competition Law & Practice, 2024 | 19

commercialisation of certain AMD-based products was cancelled be taken into account in order to determine that a dominant posi-
or delayed, customers were deprived of a choice they would tion has been abused’.232 However, it is not sufficient in itself.233
otherwise have had.225 The same is true for a showing of an anticompetitive strategy.234
However, beyond Intel, the Commission has been cautious in Judgments in which the Union Courts have found that a relevant
relying on the category of ‘naked restrictions’ so far. For instance, type of conduct was capable of harming competition because it
in Baltic Rail, the dominant undertaking dismantled its own rail- was part of a plan to harm competition235 or ‘part of a plan for
way infrastructure to stop a competitor from offering services eliminating a competitor’236 is no proof to the contrary. At a closer
based on that infrastructure. Even though this seemed like a text- look, this case law237 collapses with a ‘no economic sense’ inquiry:
book example of a ‘naked exclusion’,226 the Commission exten- neither the Commission nor the Courts have relied solely on direct
sively analysed the effects.227 (subjective) evidence of intent. Rather, they have inquired into the

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

The above discussion suggests that the ‘no economic sense’ economic rationality of the conduct in a given market context.238
test may be the right criterion to identify other cases of ‘naked In some settings, the finding of a ‘naked exclusion’ under
exclusions’. The Commission would then need to determine EU competition law may be strongly inf luenced by the interde-
whether a practice holds ‘no economic interest for a dominant pendence between EU competition law and the internal market
undertaking, except that of eliminating competitors’.228 In Super goal. In particular, a dominant firm’s conduct impeding parallel
Bock, the ECJ has redefined the scope of application of the exports of goods from low-price to high-price countries in the
concept of ‘restrictions by object’ under Art. 101(1) TFEU, however. internal market may qualify as abuse with no further effects
According to that judgment, any pro-competitive benefits of an analysis being required.239 In this case, it is the conf lict of the
agreement that, prima facie, come with a sufficient degree of harm conduct with a fundamental Treaty goal that justifies the finding
to competition will preclude its qualification as ‘restriction by of a prima facie case of illegality.
object’ only if the pro-competitive benefits are “demonstrated,
relevant, intrinsic to the agreement concerned and sufficiently 2.3. Curtailing consumer choice as a case of a form of
significant ...”.229 If one applies the same principles to the concept ‘naked exclusion’?
of ‘naked exclusion’ under Art. 102 TFEU, unilateral conduct may In some Art. 102 cases, not least in the context of tying, the notion
also qualify as ‘naked exclusion’ if it has some pro-competitive that a dominant undertaking reduced consumer choice with-
justification, but causes harm to the competitive structure that out producing countervailing benefits for consumers has figured
is clearly disproportionate to its benefits. prominently.240 If the reduction of choice is guided by no economic
It is less clear whether a qualification as a ‘naked exclusion’ rationale other than restricting competition, such conduct may
is possible and appropriate where substantial pro-competitive amount to a ‘naked exclusion’. No further effects analysis would
effects can be demonstrated, but could also be achieved by a then be required.241 In fact, such conduct interferes with one of
less exclusionary line of conduct. The principle of proportionality the most basic operating principles of performance-based com-
suggests that less exclusionary alternatives to an exclusionary petition, namely the freedom and ability of the demand side to
practice matter. But if the category of ‘naked exclusion’ is about choose between different products or services according to their
presuming effects, and thereby ‘radically’ facilitating enforce- individual preferences.
ment, ‘naked exclusions’ may not be the right category for this Nonetheless, the Commission frequently performs a full effects
analysis. analysis in these cases.242 An effects analysis will be indispensable
What is clear from the Union Courts’ jurisprudence is that indeed where the dominant undertaking comes forward with a
proof of anticompetitive intent as such will not suffice to establish plausible pro-competitive justification, e.g. related to significant
a case of ‘naked exclusion’. According to a long-standing line benefit for the demand side. In such a case, the concept of ‘naked
of case law, the abuse of a dominant position is an ‘objective exclusion’ is no longer applicable. However, a suggestion that tying
concept’.230 While a showing of anticompetitive intent is not practices in consumer-facing markets will maximise the users’
required,231 evidence of such intent ‘constitutes a factor that may convenience of use may not suffice where the same convenience

232 Servizio Elettrico Nazionale (n 38), para 63 (with further references). See
also: 2008 Priorities Guidance, para 20, where the Commission considers direct
evidence as one of a number of factors that the Commission will consider in
finding an abuse.
233 Servizio Elettrico Nazionale (n 38), para 63 (with further references). Fur-
225 Commission Decision in Intel (n 18), recital 1672. ther on the role of intent: Mariateresa Maggiolino, ‘The role of intent in abuse of
226 dominance and monopolization’, in Pınar Akman, Or Brook and Konstantinos
See Michele Giannino, ‘Lithuanian Railways: The Court of Justice Narrows
Stylianou (eds) Research Handbook on Abuse of Dominance and Monopolization
Down the Scope of Application of the Doctrine of Essential Facilities’ (2023) 7
(Edward Elgar 2023) ch 13; Nazzini (n 82) ch 6: ‘The Test of Intent’.
European Competition and Regulatory Law Review 260, 263; Kadar, Holzwarth 234
and Pereira (n 33), p 4; Pablo Ibáñez Colomo, ‘GC Judgment in Case T-814/17, See 2008 Priorities Guidance, para 20.
235 Case T-111/96 ITT Promedia v Commission, EU:T:1998:183, para 55; AKZO
Lithuanian Railways—Part I: object and indispensability’ (Chillin’Competition,
1 December 2020) <www.chillingcompetition.com/2020/12/01/gc-judgment- (n 125), para 72.
236 AKZO (n 125), para 72.
in-case-t%E2%80%91,814&#x2013;17-lithuanian-railways-part-i-object-and-
indispensability/> accessed 31 March 2024. 237 Michelin II (n 222), paras 241–242; Case T-340/03 France Télécom (n 222),
227 See Baltic Rail (n 18), pp 65–94: Roughly one-third of the total length of para 195.
the decision relates to the effects analysis. On this see Kadar, Holzwarth and 238 Nazzini (n 82) 206.
Pereira (n 33), p 4, insisting that the Commission was not required to engage 239 See Sot. Lélos kai Sia (n 46), para 66.
in a detailed effects analysis in this case, considering the seriousness of the 240 See, for example, Case T-201/04 Microsoft (n 89), paras 650–652, 1046,
restriction and the lack of plausible efficiency justifications. 1069; also: Commission Decision in Intel (n 18), recitals 1679, 1747; also: Castillo
228 See Servizio Elettrico Nazionale (n 38), para 77—which uses this definition de la Torre (n 42) 181–182.
to identify practices that do not come within the scope of ‘competition on the 241 See Castillo de la Torre (n 42) 181, pointing to Unilever Italia (n 54), para
merits’. For a discussion of the ‘no economic sense’ test, see U.S. Department 57.
of Justice, ‘Competition and Monopoly: Single-Firm Conduct under Section 2 of 242 For the effects analysis with a view to the closing of the interface for
the Sherman Act’ (2008) 39. competing work group server operating systems to interoperate with Windows,
229 Super Bock (n 184), para 36. see: Case T-201/04 Microsoft (n 89), paras 565 ff. For the effects analysis with
230 Servizio Elettrico Nazionale (n 38), para 60 (with further references). regard to the pre-installation of the Windows Media Player see Case T-201/04
231 Servizio Elettrico Nazionale (n 38), para 62 (with further references). Microsoft (n 89), paras 976 ff.
20 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

can be achieved by allowing undertakings that offer competing on the merits’ is sometimes deeply intertwined with an effects
products or services to access a relevant API. analysis. The application of the AEC test is the most important
example (see above, III. B. 4. 1.). These are not the cases dis-
3. A ‘quick look’ effects balancing in the case of cussed here.
disproportionally harmful conduct Conduct may be found to fall outside the scope of ‘competition
Where the Commission can establish a ‘naked exclusion’, no on the merits’ irrespective of its effects mainly in two types
effects analysis is required. As has been argued above, the cat- of settings: first, where competitive advantages of a dominant
egory of ‘naked exclusion’ extends to cases where prima facie undertaking do not result from superior performance, but from
harm to competition is clearly out of proportion to some relatively some sort of state privilege (4.1.); secondly, in case of conduct
minor pro-competitive benefits that may f low from the exclu- that is illegal on non-competition grounds—for example, because

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

sionary practice. To ensure that the category of ‘naked exclusion’ it violates the GDPR (4.2.). In both cases, the conduct, in and of
retains its benefit to radically simplify and speed up the enforce- itself, does not indicate anticompetitive effects: given the goal of
ment of Art. 102 TFEU, it must be limited to those cases where the Art. 102 TFEU to protect competition, potential anticompetitive
disproportionality is immediately obvious. effects need to be established before an abuse of dominance can
In other cases, the potential harm to competition of a given be found.247 However, an error-cost-oriented framework suggests
practice may still be disproportionally large compared with its that, with a view to such conduct, the evidentiary threshold for
alleged pro-competitive benefits, but a somewhat more com- anticompetitive effects can be lowered: prohibiting conduct that
prehensive analysis may be required to establish the dispropor- is clearly outside the scope of ‘competition on the merits’ does
tionality. Still, the disproportionality may follow from a ‘quick not come with a risk of ‘false positives’ or a risk to undermine
look’. In particular, there may be some settings where some pro- desirable incentives to compete. Consequently, a mere plausibility
competitive benefits are plausible, but the same pro-competitive of anticompetitive effects may suffice for establishing an infringe-
benefits can, as a general matter and based on sufficient experi- ment. The lowering of the evidentiary threshold from ‘concrete
ence in the market, be realised by a less anticompetitive line of likelihood’ to a more abstract ‘plausibility’ may facilitate the
conduct.243 For example, some efficiencies related to conditional enforcement of Art. 102 TFEU.
rebates can likewise be realised by way of quantity discounts.
The principle of proportionality is deeply engrained in the 4.1. Competitive advantages that result from a statutory
structure of Art. 102 TFEU. This is not only true when it comes monopoly or exclusive rights
to the defences that a dominant undertaking can raise: in order There are some indications that the Union Courts are willing
to prevail with an efficiency defence, the dominant undertaking’s to lower the evidentiary threshold for anticompetitive effects, at
conduct must be indispensable and proportionate to the goal pur- least in those cases where competitive advantages follow not
sued.244 Likewise, the notion that dominant undertakings are sub- from an undertaking’s investments and performance, but rather
ject to a ‘special responsibility’, i.e. that they must ‘not [ . . . ] allow from privileges conferred upon it. The ECJ’s judgment in Servizio
their conduct to impair genuine, undistorted competition on the Elettrico Nazionale is an example in point: in anticipation of the
internal market’,245 f lows from the principle of proportionality. full liberalisation of the Italian electricity market for households,
What is more, the analytical templates that the ECJ has developed and with the goal to prevent a large-scale departure of cus-
for specific types of conduct—like the test for tying, for refusal to tomers to competing suppliers, SEN—a subsidiary of the former
give access to an essential facility and even for predatory pricing— state monopolist ENEL—had obtained the consent of customers
are informed by that principle. Whenever the Commission can in the still regulated part of the market to receive commercial
show, upon a quick look and based on established economic offers for unregulated electricity supply from companies of the
wisdom,246 that the potentially anticompetitive conduct is not ENEL group—but not from ENEL’s competitors. The ECJ found,
necessary to realise the alleged efficiencies, a general showing of first, that an undertaking that holds a statutory monopoly and
plausible anticompetitive effects should suffice, with no concrete uses resources that result from these exclusive rights and are
in-depth inquiry being required. therefore inaccessible to a hypothetical as-efficient competitor for
Similarly, where, upon a quick look, the potential anticompet- the purpose of extending the dominant position in the protected
itive harm is much greater, or of much graver quality, than the market to another market is not competing ‘on the merits’.248
expected efficiencies, no in-depth proof of the scope and size of Secondly, the existence of a bias in collecting customers’ consent
the anticompetitive effects should be required. Rather, a rough would establish an irrebuttable presumption that any difference
demonstration of the disproportionality should suffice. in the quantity of customer data transferred would not result
from ‘competition on the merits’.249 Thirdly, given that access
4. No competition on the merits—lowering of the to such customer data likely conferred a comparative compet-
evidentiary threshold for anticompetitive effects itive advantage, the ECJ concluded that the subsequent use of
Finally, evidentiary shortcuts may be justified where the allegedly such data should be considered to be capable of producing an
anticompetitive conduct lies outside the scope of ‘competition on exclusionary effect on the free market.250 Consequently, if it were
the merits’. The analysis of what is, and what is not, ‘competition shown that the ENEL group, in seeking customers’ consent to
receive offers for energy supply in the free market, discriminated
243 Where a competition authority relies on the existence of less restrictive against competitors, ‘that fact alone would be sufficient to show
alternative practices, the defendant may challenge the viability of this alterna-
tive or show that such an intrusion into the freedom to conduct one’s business
is disproportionate. 247 For an apparently further-reaching position see Ibáñez Colomo (n 62)
244 British Airways (n 56), para 86; TeliaSonera (n 83), para 76; Post Danmark (n 403–404: Conduct that is, by its very nature, at odds with competition on the
69), paras 41–42. See also: 2008 Priorities Guidance, paras 28 ff. merits shall be considered presumptively abusive, with no need to analyse
245 See Michelin (n 126), para 57; Case C-413/14 P Intel (n 61), para 135; Servizio effects. However, the dominant undertaking can show that the conduct is
Elettrico Nazionale (n 38), para 74. incapable of having restrictive effects.
246 248 Servizio Elettrico Nazionale (n 38), para 91.
This analytical shortcut does not amount to an empowerment of the
249 Servizio Elettrico Nazionale (n 38), para 99.
Commission to engage in creative market engineering. The ‘less restrictive
alternative’ test should be limited to some well-established standard settings. 250 Servizio Elettrico Nazionale (n 38), para 100.
 Journal of European Competition Law & Practice, 2024 | 21

that the conduct [of the dominant undertaking] at the very least different sources) and harm to the other party to the contract is
was capable of impairing effective, undistorted competition’.251 clearly established. More specifically, a showing of a plausible,
In short, a discriminatory use of competitively relevant resources abstract risk to adversely affect competition will then be
resulting from a statutory monopoly was irrebuttably presumed enough.257
to have potential anticompetitive effects. As of now, it is not yet clear whether the ECJ would be willing
to adopt a similar approach. However, where the relevant conduct
4.2. Conduct that is illegal on non-competition grounds by a dominant undertaking is illegal under non-competition rules,
There is somewhat conf licting case law on whether the illegality the error costs of prohibiting it (also) under competition law are
of a relevant line of conduct under non-competition norms should low. As such, this would not justify activating the competition law
be considered when determining whether conduct is, or is not, enforcement machinery. A link to the goals of EU competition law,

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

‘competition on the merits’ within the meaning of Art. 102 TFEU, and hence some potential to adversely affect competition, is an
and whether it leads to a lowering of the threshold for proving indispensable precondition. But for the sake of a more effective
anticompetitive effects. Clearly, the legality of a practice under enforcement, it may well be justified to drop the requirement to
non-competition laws does not pre-empt a finding that such prove a concrete risk to competition and to let a plausible abstract
conduct constitutes an abuse under Art. 102 TFEU.252 In Servizio risk suffice.
Elettrico Nazionale, the ECJ opted for a broader wording, thereby
insinuating that the illegality of a practice under non-competition
laws is likewise irrelevant for the finding of an abuse: according to
V. Tentative conclusions
Servizio Elettrico Nazionale, ‘[t]he illegality of abusive conduct under The Commission’s initiative for drafting guidelines on exclu-
[Art. 102 TFEU] is unrelated to the characterisation of that conduct sionary abuses comes at the right moment in time. In the
in other areas of law’.253 Commission’s enforcement practice, the belief that the 2008
That statement stands in some tension to the ECJ’s preliminary Priorities Guidance, with its commitment to intervene only based
ruling in Meta Platforms, however. According to that judgment, on concrete proof of consumer harm, has quickly given way
‘the compliance or non-compliance of [the] conduct with the to the recognition that the quality of competition law depends
provisions of the GDPR may, depending on the circumstances, not only on the economic accuracy of the rules, but also on
be a vital clue among the relevant circumstances of the case the administrability of the enforcement regime, including the
in order to establish whether that conduct entails resorting to provability of the relevant facts at reasonable cost and within a
methods governing normal competition and to assess the conse- reasonable timeframe. As Joskow once observed: ‘The test of a
quences of a certain practice in the market or for consumers’.254 good legal rule is not primarily whether it leads to the correct
Consequently, in examining an abuse, it might be necessary for decision in a particular case, but whether it does a good job
a competition authority to inquire ‘whether that undertaking’s deterring anticompetitive behaviour throughout the economy
conduct complies with rules other than those relating to com- given all the relevant costs, benefits, and uncertainties associated
petition law’.255 Indeed, the relevance of the compliance or non- with diagnosis and remedies’.258
compliance with the GDPR for establishing whether the relevant Consequently, the focus of the new initiative now is on estab-
conduct constituted ‘competition on the merits’ seems intuitive. lishing a ‘workable effects-based approach’.259 In order to ensure
However, it is not, as such, indicative of whether that conduct Art. 102 TFEU’s effectiveness, including its deterrent effect and
potentially harms the competitive structure. Therefore, the ECJ’s the Commission’s ability to impose remedies that undo the com-
finding that the compliance or non-compliance with the GDPR petitive harm, the duration of Art. 102 proceedings must be sig-
may also be a ‘vital clue’ when it comes to the assessment of nificantly reduced.
the ‘consequences of a certain practice in the market or for While one may think of several different causes for the increase
consumers’ comes as a surprise: There is neither an empirical nor of the median duration of Art. 102 proceedings over time, a
a conclusive normative link between an infringement of the GDPR high degree of uncertainty surrounding the substantive law on
and anticompetitive harm. abuse and tightened requirements regarding the proof of anti-
The German Federal Court of Justice (BGH), in its Facebook competitive effects are arguably among them. Simultaneously,
decision,256 has therefore adopted a different line of argument: the evidentiary rules regarding the proof of effects are particularly
under the German prohibition of abuse of dominance (§ uncertain. In such a setting, a risk-averse Commission tends to
19(1) GWB), an abbreviated analysis of anticompetitive effects invest an excessive amount of time and resources to meet the
may suffice in cases where a dominant undertaking adopts highest possible standard in order not to lose a case that they have
unfair contractual conditions (in Facebook: a clause requiring determined to be of special importance.
consent to the merging of personal user data obtained from Guidelines on exclusionary abuses can contribute to deduc-
ing, from a partly inconsistent case law of the Union Courts, a
(more) coherent legal framework. They can thereby help to clarify
251 Servizio Elettrico Nazionale (n 38), para 102. the substantive law on abuse of dominance—albeit within the
252 Case C-457/10 P AstraZeneca (n 83), para 132. See also para 133: The fact interpretative framework defined by the Union Courts. For exam-
that the objectives of those other rules are complied with does not suggest that
the application of Art. 102 TFEU is no longer required to ensure effective and ple, they can explain the concept of ‘competition on the merits’
undistorted competition within the internal market. and its relation to the AEC principle; and they can specify what is
253 Servizio Elettrico Nazionale (n 38), para 67.
254 Case C-252/21 Meta Platforms and Others, EU:C:2023:537, para 47.
required, under Art. 102 TFEU, for showing ‘potential anticompet-
255 Meta Platforms (n 254), para 48. In his Opinion in the Meta Platforms case,
AG Rantos had taken a somewhat more cautious position, arguing that the
compliance or non-compliance of that conduct with the GDPR, ‘not taken in
isolation but considering all the circumstances of the case, may be a vital clue
257 BGH, 23 June 2020, KVR 69/19—Facebook, para 76.
as to whether that conduct entails resorting to methods prevailing under merit-
based competition’—see Opinion of AG Rantos in Case C-252/21 Meta Platforms 258 Paul L Joskow, ‘Transaction Cost Economics, Antitrust Rules and
and Other, EU:C:2022:704, para 23. Remedies’ (2002) 18 Journal of Law, Economics, and Organization 95, 99 f.
256 BGH, 23 June 2020, KVR 69/19—Facebook. 259 Policy Brief, title.
22 | Heike Schweitzer and Simon de Ridder • How to Fix a Failing Art. 102 TFEU

itive effects’. Although the Union Courts have explicitly refused to In other cases, an abbreviated effects analysis may suffice. This
require a specific degree of risk, they have repeatedly found that may be the case where the pro-competitive effects of a practice
an abstract risk of anticompetitive effects will not suffice, but that may likewise be achieved by less exclusionary alternatives, or
a concrete risk to the competitive structure must be established. where conduct is at issue that lies outside the scope of ‘compe-
In fact, this appears to be the core feature of the Union Courts’ tition on the merits’. In the latter case, the showing of an abstract
‘effects-based approach’ to Art. 102 TFEU. risk of anticompetitive effects may suffice.
Arguably, an equally important contribution of future guide- Ultimately, all this may not be enough. More may be needed
lines to a more effective enforcement of Art. 102 TFEU may lie to arrive at an effective enforcement of Art. 102 TFEU. The
in developing a clear and coherent framework with regard to the Commission’s initiative to re-evaluate Regulation 1/2003 has
evidentiary requirements for proving anticompetitive effects— already been mentioned. As an increasing number of Member

Downloaded from https://academic.oup.com/jeclap/advance-article/doi/10.1093/jeclap/lpae033/7690940 by guest on 26 July 2024

requirements that are currently utterly confusing. This is true States is about to introduce some sort of ‘market investigation’
in particular with regard to the types of inferences and other into their competition law toolbox,260 the debate about intro-
analytical shortcuts on which the Commission can rely, and with ducing a ‘New Competition Tool’261 that would empower the
regard to the preconditions on which these inferences or shortcuts Commission to intervene independently of the finding of an abuse
can be based. may see a revival also at the EU level. Possibly, a legal framework
From the ECJ’s 2017 Intel judgment onwards, the Union for Art. 102 proceedings without fines is required, which would
Courts have rejected the use of conduct-based presumptions of allow for an enforcement under a lower standard of proof and
anticompetitive effects. But future guidelines on exclusionary with somewhat different evidentiary requirements. Also, a better
abuses could set out the preconditions for when ‘factual and clearer framework for more effective remedies is called for.
inferences’ are justified, i.e. when potential anticompetitive Nonetheless, future guidelines that reduce the uncertainties
effects can be inferred, on a sound empirical basis, from a surrounding the substantive interpretation of exclusionary
limited number of easily observable facts. Simultaneously, abuses and establish a coherent and administrable evidentiary
presumptions with a partly normative, partly empirical basis have framework are a key element of any more comprehensive reform
not been banned from the law on abuse altogether: potential of Art. 102 TFEU. Compared with the alternatives—more sector-
anticompetitive effects may be presumed in case of ‘by object’ specific regulation à la DMA or a legal specification of the
exclusions or so-called ‘naked exclusions’—a concept that should conditions for the existence of an abuse in a legislative act by the
be further developed in the guidelines. Arguably, the category of Council, based on Art. 103 TFEU262 —guidelines on exclusionary
‘naked exclusion’ extends to cases where the pro-competitive abuses are the most attractive and realistic option to contribute
justification is obviously insignificant compared with the to a regime within which Art. 102 TFEU can gain new vitality in
anticompetitive potential. protecting competition.

260 See, for example, Germany (see Nada Ina Pauer, ‘The 11th Amendment
of the German Act Against Restraints of Competition’ (2023) 14 Journal
of European Competition Law & Practice 354, 355–357); a novel market
investigation tool is currently being considered in Sweden (see e.g. Helene
Andersson, ‘Main Developments in Competition Law and Policy 2023—Sweden’
(Kluwer Competition Law Blog, 19 February 2024) <www.competitionlawblog.
kluwercompetitionlaw.com/2024/02/19/main-developments-in-competition-
law-and-policy-2023-sweden/> accessed 31 March 2024).
261 On this, see Massimo Motta, Martin Peitz and Heike Schweitzer (eds),
Market Investigations: A New Competition Tool for Europe? (CUP 2021).
262 On this, see Laitenberger and Kröger (n 19) 633 ff.