Welcome to the Competition Stories – a bimonthly exploration of recent courts and competition law agencies’ decisions. Authored by Makis Komninos, a renowned expert in the field, this new column aims to go through the latest and most important developments in competition law of the last two months. We call them “stories” because Makis has promised to include some anecdotes from time to time, and not just stay at the black letter. Enjoy!
The intense activity of the EU Courts and the European Commission reached a peak in May-June, so these “stories” are probably the longest ever. That gives us some nice summertime reading… The Qualcomm judgment of the General Court clearly takes my first prize. I also have a few things to say about the much-awaited Servizio Elettrico Nazionale ruling of the Court of Justice, as well as Volvo/DAF. The remaining judgments and opinions are interesting but not consequential. On the Commission front, we need to bear in mind the new Vertical Block Exemption Regulation and Vertical Guidelines, which entered into force on 1 June 2022.
I. Court of Justice
This was a judgment eagerly awaited after Advocate General Rantos’s Opinion, in my view, one of the best Opinions in competition law. I have already described the facts of the case and praised the Opinion in my Stories of November-December 2021. The case took place in the context of the gradual liberalisation of the Italian energy market. The former vertically integrated monopoly unbundled two subsidiaries, one of which, Servizio Elettrico Nazionale (SEN), operated on a protected market, and the other, Enel Energia (EE), operated on the free market. The conduct involved SEN transferring data to EE, so that EE could make commercial offers to customers that were going to switch onto the free market when the protected market was abolished. The Italian competition authority found that the conduct was an abuse of dominance and imposed a fine of EUR 27 million. There was an appeal and several questions were referred by the Italian Consiglio di Stato to the Court of Justice.
After AG Rantos’s Opinion, the main question was: will the Court follow the AG and will it espouse as clearly as the AG the effects-based analysis and the consumer welfare standard in Article 102 TFEU? The answer to that question is yes and no. Servizio Elettrico Nazionale is a mixed judgment. In my view, the Court attempted to reconcile the irreconcilable. So the competition law specialists – depending on their angle and credo – will read the judgment with mixed feelings. There are passages that will delight them and there are other parts that will disappoint them. That’s what you get when there is an attempt to reconcile economics and the effects-based approach with and the old ordoliberal and structuralist approach.
I can start with the judgment’s positive elements (from my own angle):
- First, for the first time, the Court of Justice explicitly held that consumer welfare is the ultimate objective of Article 102 TFEU (para. 46). We have had before occasional references to the objective of the competition rules being the “protection of competition and consumer welfare” (e.g. Post Danmark II), but never before had the Court referred to consumer welfare in such absolute terms. The Court did retain its existing position that an authority only needs to show harm to the structure of competition, not harm to consumers (para. 47), but “effective competitive structure” is seen more as a proxy for the broader ultimate objective of consumer welfare. Thus, it can be seen as subservient to the latter. The change in framing towards harm to competition being a concern because it harms consumers is positive.
- Second, a positive element is the Court’s explicit confirmation in para. 72 that the abusive nature of conduct does not depend on the form that it takes but rather on its capacity to foreclose (i.e. likely effects) after having regard to “all circumstances”. This is a general pronouncement and it’s good to see it repeated in such clarity, because there have been attempts in the past to limit the scope of Intel to exclusivity rebates, which is clearly wrong.
- Third, the Court found that (a) departure from competition on the merits and (b) exclusionary effects are two separate conditions that must both be established (para. 103). This is a very welcome reminder. Indeed, an Article 102 TFEU violation can be found only if there are actual or potential exclusionary effects, but these effects must be “anticompetitive”, in the sense of stemming from conduct other than competition on the merits. The merits of the conduct and its effects are separate concepts that cannot be conflated. If certain conduct became anticompetitive simply because it caused actual or potential exclusionary effects, procompetitive conduct that leads to the “the departure from the market or the marginalisation of competitors”, in the words of Intel, would be wrongly outlawed.
- Fourth, another helpful development was the Court’s endorsement of the “as efficient competitor” principle. While the principle has already been introduced in previous case law (Post Danmark I, Intel), the judgment made extensive references to it in paras 71, 73, 76, 78-82, 91, and 101. Interestingly, the Court relied on it to develop a “replicability test” for abuse of dominance cases. In paras 79-84, the Court held for the first time that the ability of an as efficient competitor to replicate the conduct of the dominant company is relevant to establishing that the conduct departed from competition on the merits. This is a test that AG Rantos had recommended to the Court and the Court took up. The “as efficient competitor” principle has to be combined with references to what we could call “performance competition”, which is emphasised by the Court in para. 85. The Court also held that a situation in which consumers benefit from lower prices, better quality, and more choice is, in principle, competition on the merits.
That was the good side of Servizio Elettrico Nazionale. However, the judgment, also includes statements that reflect the old structuralist-ordoliberal approach and I find problematic. In particular:
- First, while pleasing to see the references to consumer welfare, competition on the merits and the “as efficient competitor” principle, one cannot but be disappointed with the return to some of the more interventionist language that the Court was using in its pre-Intel formalistic case law. For example, the Court referred to conduct making it more difficult for competitors to penetrate or maintain their position on the relevant market (paras 61, 76). This type of language was absent in Intel. Similarly, the Court cited some of its earlier, more formalistic, case law, such as Hoffmann-La Roche (paras 44, 68), Tomra (para. 62), and Post Danmark II (paras 70-71, 84) – the last two cases had not been cited even once in Intel.
- Second, the Court, unlike AG Rantos, was not ready to recognize the importance of evidence showing no actual exclusionary effects. In paras 53-55, it rejected the relevance of such evidence and also seemed to go against the Commission’s Guidance Paper on that point (para. 54) – that’s not entirely clear, the text here is somewhat confused. Then, in paras 57-58, the Court was totally unreceptive to the argument that exclusionary effects had not, in reality, occurred. In particular, it rejected the argument that the conduct was not abusive because it only led to 0.002% of customers switching to EE. The specific language on this point is rather harsh.
- Third, while the Court stresses the importance of competition on the merits and of performance competition and efficiency (e.g. para. 85), it includes some confusing statements, which seem to indicate that these elements can only be taken into account at the objective justification / efficiency defence stage, which, if true, would amount to a reversal of the Intel case law. For example, in paras 47 and 86, the Court seems to confuse the opportunity of the dominant company to produce evidence showing that its conduct does not have the “intrinsic capacity to foreclose” (per Intel, paras 139-140) with the possibility to rebut an already-established finding of foreclosure by proving efficiencies. My sense is that this is a regrettable confusion in the drafting of the judgment. I am 100% certain that the Court did not intend to reverse Intel.
- Fourth, while stressing the importance of the “as efficient competitor” principle, at the same time, the Court seems to accept a sort of “no economic sense” test to determine whether conduct departs from competition on the merits (para. 77) – I have no objection to that. The problem is that it supplements this with a test of whether the conduct would be rational for a non-dominant as efficient competitor (para. 79). For the Court, if a non-dominant as efficient competitor would not have adopted the conduct, that indicates a departure from competition on the merits. The Court also stresses that the hypothetical as efficient competitor must not be dominant and must not enjoy advantages and resources that are the product of its dominance (paras 78-79). I find the latter references confusing and prone to be exploited as a back door to undermine the “as efficient competitor” principle. While this is not likely to make any major difference in terms of the cost measure that is taken into account when the AEC test is performed in a pricing case, it may have a negative bearing on other cases. In my view, these statements should be limited to and be seen in the very specific context of the case. Indeed, it would sound rather ironic for a current or former State monopoly to rely prominently on the “as efficient competitor” principle in order to protect its monopoly in perpetuity.
For the rest, contrary to what the AG’s Opinion could be interpreted as suggesting, the Court of Justice strongly implied that the specific conduct was an abuse of dominance. In particular, it took a near zero tolerance approach to using a former legal monopoly to maintain a dominant position on a newly liberalised market (paras 91-92). In light of that, the Court took a very strict approach to discrimination, particularly in the context of SEN’s use of customer data, and held that any discrimination or bias by SEN when collecting or sharing data would be abusive (paras 98-100, 102). In particular, the Court found that, when gathering consent for data use by the ENEL group, SEN had to also gather consent for use by other competitors. That consent-gathering process had to be non-discriminatory and not create any bias that was likely to lead to more customers agreeing to internal than external data use (para. 96). The Court noted that the enforcer has to adduce evidence, such as behavioural studies, showing that customer consent was more likely for use within the ENEL group than external use (para. 98). The Court noted that if there was any bias, then it would be impossible to tell whether EE’s success on the free market was due to the problematic conduct at stake or to superior performance (para. 99). Thus, transferring the list of customers internally would create an advantage for EE’s activities on the free market (para. 100). Indeed, the Court’s application of the replicability test was quite strict. Nonetheless, again, para. 101 is explicit that this was a result of the former legal monopoly, which may have been the rationale for the rather strict approach.
Drawing general principles from this part of the judgment is difficult. The Court is explicit about its concerns around conduct that aims at defeating the market liberalisation process. A plausible reading is therefore that this zero tolerance approach is only appropriate for current or former legal monopolists. Indeed, in paras 91-92, the Court seems to have been inspired by cases referring to State measures distorting competition (when Article 102 TFEU is applied in conjunction to Article 106 TFEU).
As to the Court’s more general pronouncements, the espousal of different approaches that are inconsistent with each other does not make this judgment a monument of clarity and somewhat reduces its precedential value. This is, after all, a judgment of a chamber and not a judgment of the Grand Chamber, as was Intel. We also need to bear in mind that the totally academic and theoretical nature of the reference questions of the Italian Consiglio di Stato did not help the Court of Justice (indeed, in paras 28-39, the Court had some difficulty on admissibility because of the totally theoretical nature of the questions). Clearly, not one of the leading judgments in Article 102 TFEU. But perhaps understandable because of the specific context.
This is another preliminary reference judgment following on from the trucks cartel case. The plaintiff in this case had purchased, during 2006 and 2007, three trucks manufactured by Volvo and DAF and brought a follow-on action for damages. His claim was upheld in part by the first instance Spanish court, and Volvo and DAF were ordered to pay damages amounting to 15% of the purchase price of the trucks. The court rejected the plea raised by the defendants that the action was time-barred, applying the five-year limitation period provided in the new Spanish legislation, which transposed the Damages Directive. In addition, in accordance with that legislation, the national court applied the presumption of harm caused by the infringement at issue and used its power to assess the harm, as provided for in the Damages Directive. The two defendants brought an appeal against that judgment and submitted, first, that the action was time-barred because the one-year limitation period of the Spanish Civil Code, which in their view was applicable, began to run from the publication of the Commission’s press release (as opposed to the later date of the publication of the summary decision in the OJ). Secondly, they argued that there is no proof of the causal link between the infringement and the price increase.
The appellate court referred to the Court of Justice a number of questions on the scope ratione temporis of certain provisions of the Damages Directive concerning the limitation period, the assessment of harm, as well as the compatibility of the national legislation applicable to actions for damages with Article 101 TFEU and the principle of effectiveness (effet utile). I have already commented on the Advocate General Rantos’s Opinion in my Stories of July-October 2021.
In effect, the Court followed AG Rantos’s Opinion almost entirely, apart from one point. So the Court held that Article 10 of the Damages Directive on limitation periods is a “substantive” rule, as opposed to “procedural”. Following its AG, the Court also considered that Article 17(1) of the Damages Directive, which provides that the burden and standard of proof required for the quantification of harm cannot render the exercise of the right to damages practically impossible or excessively difficult, is a “procedural” rule. However, Article 17(2) of the Damages Directive, which introduces a rebuttable presumption that cartels cause harm, is a “substantive rule”.
These are important questions because, under Article 22 of the Damages Directive, procedural rules apply to actions that have been brought after the Directive entered into force (on 26 December 2014), but substantive provisions do not apply retroactively to “situations existing” before that date. By the way, what is a “substantive” and what is a “procedural” rule, the Court said is a matter of EU, not national law (para. 39). The Court also followed AG Rantos as to the time that the limitation period begins to run (dies a quo). So, in the present case, the Court agreed that this cannot be before the publication of the summary decision in the OJ, as a matter of effectiveness (para. 72). That being said, the Court did also hold that this is not set in stone. It can be that the elements necessary for bringing an action for damages may be known to the injured party well before that publication or even before the publication of the press release concerning the adoption of the cartel decision but that has to be proven (para. 64). I think this is more likely to happen in an Article 102 TFEU exclusionary case.
While both the AG and the Court thought that the action in the specific case was not time-barred, where the Court did not follow the AG and went further was in the introduction (in a somewhat cryptic way in paras 73-79) of a transitory rule that the five-year limitation period of Article 10 of the Directive applies, if an action for damages, though relating to an infringement of competition law that ceased before the entry into force of the Directive, was brought after the entry into force of the national transposition measures, and the time limit for bringing that action under the old rules (1-year Spanish limitation period) did not expire before the time limit for the transposition of the Directive. In other words, if the national limitation period has begun to run after the expiry of the time limit for the transposition of Directive 2014/104 and continues to run after the date of entry into force of national transposition measures, Article 10 of the Directive extends in effect the limitation period. As a reminder, the Advocate General had considered that the temporal scope of the Directive does not cover an action of damages that was brought after the entry into force of the Directive and the national transposition measures but referred to “prior facts and penalties”. In the Court’s words, “it therefore appears that the situation at issue in the main proceedings continued to produce effects after the date of expiry of the time limit for transposition of Directive 2014/104, and even after the date of entry into force of Royal Decree-Law 9/2017 transposing that Directive” (para. 74). So the conclusion is that “[i]n so far as that is the case in the dispute in the main proceedings, which it is for the referring court to verify, Article 10 of that directive is applicable ratione temporis in this case” (para. 75).
I said before that the text here is rather cryptic, because the above solution given by the Court appears “out of the blue”. It also seems inconsistent with paras 76 and 77, where the Court reiterated its long-standing case law on the non-horizontal direct effect of directives and on the limits of the doctrine of the interpretation of national law in conformity with EU law (no contra legem interpretation). When one reads that part of the text, one has the impression that some paragraphs are missing… All these problems create some doubt: is the Court of Justice really creating new law (in the sense of a new EU legal principle) or is it merely stating what it understands to be a position of Spanish law? I am told by Spanish legal experts that, under Spanish law, where the law introduces a new limitation period, an action submitted under the prior rules enjoys the longer duration of the new limitation period. Could it be that the Court is merely stating the legal position under Spanish law here? Time will tell…
- Optical Disk Drives Judgments
These were four appeals (Sony, Sony Optiarc, Quanta Storage and Toshiba) in the optical disk drives cartel case. In my Stories of May-June 2021, I have already commented on Advocate General Pitruzzella’s Opinion, especially the part on the concept of “single and continuous infringement”. The AG proposed that the General Court’s rulings that dismissed the appeals against the Commission’s decision (see e.g. here) should effectively stand, but considered that the General Court made some legal errors in supporting the wording used by the Commission in its decision. In particular, the Commission had said that the “single and continuous infringement” consisted of “several separate infringements”. The addressees of the decision had challenged this finding but the General Court held that the Commission did not infringe their rights of defence by referring to the “several separate infringements” point for the first time in the decision, as opposed to the Statement of Objections.
The Court of Justice followed the AG, set aside the judgments of the General Court and partially annuled the decision of the Commission, while upholding the amounts of the fines imposed (amounting in total to EUR 116 million). The Court found that the Commission failed to satisfy its obligation to state reasons by finding that, in addition to their participation in a “single and continuous infringement”, the undertakings concerned also participated in “several separate infringements”. In that regard, the General Court erred in law in holding that the Commission had not breached the rights of defence of those undertakings and that it had satisfied its obligation to state reasons. The Court, however, rejected the other pleas of the appellants and no doubt disappointed them by holding that none of the elements relied on by the participants in the cartel, nor any ground of public policy, justified it making use of its unlimited jurisdiction to reduce the amount of the fine. So a complete loss for the appellants, but a victory in terms of a due process element…
This was an appeal against a General Court judgment that upheld the Commission’s rejection of a complaint, in a case probably involving one of the most persistent complainants ever! Remember the Velux case that gave rise to an interesting article written by Commission officials in the now defunct (and greatly missed) EU Competition Policy Newsletter in 2009? The article was explaining how the Commission had applied the AEC test to prove that the rebates at stake were not anticompetitive (those were the days…). Well, that’s it. It’s the same case.
The fact that there was no AG Opinion is indicative of what the Court really felt about the case. I will not spend too much time on this judgment. The funny element was that the (Polish) complainant included some colourful arguments about the Commission being biased against the “newer” Member States, because statistics showed that there was a tendency towards more frequent rejections of complaints from the “newer” EU Member States in comparison to what Donald Rumsfield famously once called “old Europe”… The Court dutifully rejected that particular argument, saying that even if there is such a quantitative tendency, it cannot explain the reasons behind the rejection of these complaints, nor can it prove that such a rejection was unjustified (para. 143). In short, a funny case…
This is an Opinion of Advocate General Szpunar, the Court of Justice’s first AG, and relates to the Damages Directive, which has been giving rise to a rich case law. The issue at stake was whether a national court can order disclosure of evidence in a suspended damages claim, where a European Commission’s investigation is still on-going. The facts relate to an Article 102 TFEU investigation in the Czech Republic against railway incumbent České dráhy, which was opened in 2012 by the Czech competition authority but was later superseded by a Commission investigation of the same scope (in 2016). In the meantime, in 2015, RegioJet had filed an action for damages against České dráhy. The facts are quite complicated and reference is made to paras 13-23 of the Opinion. The first legal question examined by the AG was admissibility and the temporal application of the Damages Directive. He held that the Directive was applicable even though it was transposed into Czech law after the action was filed.
Then, he addressed the referring court’s first question: can a national court order disclosure of evidence in a case where the action has been suspended, while the Commission is investigating the matter and contemplating a decision? The AG, relying on Article 16 of Regulation 1/2003, clarified that the suspension of proceedings by the national court is not a strict obligation imposed by EU law. True, Article 16(1) of Regulation 1/2003 invites the national courts to suspend proceedings when there is a risk that they take a decision that runs counter to a decision contemplated by the Commission, but this does not mean that suspension is the unqualified norm. In addition, the AG stressed that the Damages Directive does not preclude, in principle, the national courts from ordering discovery of documents that are included in the file of a competition authority before the closing of the proceedings (para. 49). Certainly, this is the case for the so-called “white list” of Article 6 of the Damages Directive, while for the so-called “grey” and “black lists” there are special rules. However, the national court must comply with the conditions of the Directive and, in particular, it must order the disclosure of documents that are critical and necessary, pursuant to the principle of proportionality. In that context, according to the AG, the national court must duly take into account that the civil proceedings have been suspended (para. 51). But this does not mean that a national court cannot order discovery simply because it has suspended proceedings because of the pendency of a Commission proceeding. So the answer is that a national court that has suspended proceedings is not precluded from ordering discovery.
The AG then dealt with the third reference question. Under Article 6(5) of the Damages Directive, a national court can order the disclosure of “grey-listed” documents and evidence “only after a competition authority, by adopting a decision or otherwise, has closed its proceedings”. The question here was whether the withdrawal of the competence of the Czech competition authority as a result of Article 11(6) of Regulation 1/2003, because of the Commission’s opening of proceedings in the same case, meant that the Czech authority had “closed its proceedings” per Article 6(5) of the Damages Directive. The AG relied on Toshiba and reminded that Article 11(6) of Regulation 1/2003 does not deprive the national competition authorities of their competence in a definitive and permanent manner. Instead, their competence can revive when the Commission completes its proceedings. Therefore, it could not be said that the Czech authority had “closed its proceedings”, if it was conceivable that it could regain its competence after the Commission completed its own proceedings.
The second reference question, which the AG dealt with in the third place concerned a point where Czech law was going further than the Damages Directive in restricting access to evidence. In particular, while Article 6(5)(a) of the Damages Directive protects “information that was prepared by a natural or legal person specifically for the proceedings of a competition authority”, the Czech transposition law went further than that and protects all information that is “submitted” for the purposes of a competition authority’s proceedings. AG Szpunar found that this was too strict and therefore not compatible with the Damages Directive, which states (Article 5(8)) that Member States can introduce rules that lead to wider (but not narrower) disclosure.
Finally, the AG concluded that the Damages Directive does not preclude a national court from asserting if the documents at stake contain information prepared for the purposes of an investigation and if they are covered by the “grey list” of Article 6(5) of the Damages Directive. However, if the national court decides to place the evidence under sequestration (custody), as was the case here, it has to ensure that the other party to the proceedings does not have access to it.
This is the first competition law Opinion of Advocate General Emiliou. For those of us with some grey hair, the name Emiliou rings a bell. The AG, in a previous life, was an academic with significant contributions in the area of the general principles of EU law, such as the principles of proportionality and subsidiarity.
The Opinion relates to the Euro Interest Rate Derivatives (EIRD) cartel case, which has a rather tarnished procedural history. The reason is that the Commission opted for a staggered hybrid cartel settlement procedure, in other words the settlement decision and the ordinary decision were not adopted at the same time but were staggered over time. In this case, the settlement decision was adopted in 2013 (against Barclays, Deutsche Bank, Société générale and RBS – all were imposed a EUR 1.71 billion fine) and the ordinary decision was adopted in 2016 (against HSBC, Crédit Agricole and JP Morgan – they were imposed a EUR 485 million fine). HSBC appealed on a number of grounds, most of which were rejected and the General Court confirmed HSBC’s participation in the cartel. However, the General Court annulled HSBC’s EUR 33.6 fine for insufficient reasoning. AG Emiliou’s Opinion relates to the appeal against the part of the General Court judgment that left the Commission’s findings intact. In the meantime, the Commission readopted in 2021 a decision against HSBC and re-imposed a slightly-decreased fine of EUR 31.7 million. It does not stop here… In addition, the Commission also adopted another decision amending the 2016 decision against Crédit Agricole and JPMorgan correcting the same irregularity (lack of sufficient reasoning of the fines imposed) identified by the General Court in the HSBC case, although those two undertakings’ appeals (see here and here) to the General Court were still pending. A procedural imbroglio…
Enough with that. AG Emiliou’s Opinion touches upon two very important questions: First, the famous (or some would say, infamous) “harmless error” case law, whereby a procedural irregularity will lead to the setting aside of the challenged act only if that irregularity has or may have influenced the outcome of the procedure and, second, the characterization of certain practices as anti-competitive “by object”.
The Opinion is quite didactic on the first question. The AG makes a distinction between a “first body of case-law”, involving the violation of an “essential procedural requirement” (such as the obligation to state reasons) and a “second body of case-law”, involving “procedural irregularities in respect of which no prevailing public interest in setting aside the act in question can be identified” (paras 41 and 45). In the first category, “the EU Courts have, without further ado, set aside decisions adopted following a procedure in which the Commission had infringed an ‘essential procedural requirement’. In those cases, the Court considered it unnecessary for the undertakings invoking a procedural error to demonstrate that such an error might have influenced the course of the proceedings and the content of the decisions in question to their detriment” (para. 41). In the second category, “it is for the undertakings concerned to invoke the breach, prove it to the requisite standard, and – usually – explain the possible consequences flowing from it. In fact, in those cases, the EU Courts have generally adopted what could be referred to […] as a ‘harmless error’ test. Put simply, a procedural irregularity leads to the setting aside of the challenged act only if that irregularity has or may have influenced the outcome of the procedure” (para. 45).
It is in that second category where the AG focuses. He identifies a lack of precision in the case law and a certain fluctuation “depending on the specific circumstances of each case – between a lighter form and a more stringent form” (para. 46). He then identifies “three distinct strands of judgments” in this second category:
“47. In a first strand of judgments, the Court followed the most stringent form of the harmless error test that, as mentioned above, was first formulated, at least in the field of competition law, in Suiker Unie. It is also the test that was applied by the General Court in the judgment under appeal. According to this formulation of the test, a procedural error leads to the annulment of the Commission decision only where the applicant is able to prove that, absent the irregularity, the outcome of the procedure would have been different. Under this test, the onus of proving the ‘what if’ scenario is on the applicant, and the threshold triggering the annulment of the challenged decision is one of ‘near certainty’, or at least of ‘high probability’.
- Subsequently, in a second strand of judgments, the harmless error test appears to have evolved into a lighter form. Indeed, in several cases, the EU Courts stated that the decisions challenged were to be annulled if the applicants were to establish that, without the procedural irregularity, the outcome of the procedure might have been different. In these cases too, at least initially, the burden of proof rests with the applicant. However, the standard of evidence required is lower (a mere ‘possibility’ normally suffices) with the consequence that, once met by the applicant, the burden shifts to the Commission.
- Finally, in a third strand of judgments, the Court, having detected an irregularity of the procedure that resulted in a serious and manifest infringement of some procedural right of the undertakings in question, presumed that that irregularity affected, or is likely to have affected, the outcome of the procedure. Accordingly, the Court did not require the undertakings in question to provide any evidence in that respect.”
The AG then proposes to the Court to follow the “lighter form” of the “harmless error test”, which has become “the ‘standard’ one, at least when there is a genuine fundamental right issue in the case” (para. 57). Instead, the more stringent form of that test should only be applied to “situations in which the procedural errors invoked appear to concern irregularities of a lesser nature” (para. 58). In the case at hand, AG Emiliou sided with HSBC and found that the General Court “erred in subjecting their allegations concerning an infringement of the duty of impartiality and of the principle of the presumption of innocence to a stringent harmless error test (the Suiker Unie test), thereby requiring specific proof from them that the content of the contested decision would have differed if that irregularity had not occurred” (para. 60). He added that “a failure by the Commission to act as an impartial public administration when assessing a given case – because it was biased, or because it had a pre-conceived idea about the guilt of the undertakings under investigation – would constitute a serious infringement of its duties that is likely to have repercussions on the outcome of the procedure” (para. 61).
On the second question, i.e. on what constitutes a “by object” restriction of competition under Article 101(1) TFEU, the AG had some interesting things to say. While rejecting most of HSBC’s arguments about the main thrust of the cartel, he proposed to the Court to uphold HSBC’s third ground of appeal, which claimed that the General Court erred in law by concluding that two discussions on mid-point/average of the bid and offer prices (“mids”) were infringements “by object” and by stating that the pro-competitive nature of those discussions could only be taken into account in the context of either the ancillary restraints doctrine or under Article 101(3) TFEU (para. 91). The AG here followed the “recent case-law” (para. 92) and stressed that cases such as UK Generics establish that possible pro-competitive effects of an agreement must be duly taken into account for the purpose of its characterisation as a “restriction by object” under Article 101(1) TFEU, as elements of the context of that agreement. Indeed, such elements “are capable of calling into question the overall assessment of whether the concerted practice concerned revealed a sufficient degree of harm to competition” (para. 92). This does not amount to a rule of reason, but “is merely intended to assess the objective seriousness of the practice concerned” (para. 93). In this case, the General Court examined only whether the exchanges of information on “mids” could be justified under the ancillary restraints doctrine and therefore there was a legal error.
For the rest, the AG proposed to the Court to reject the appeal. As to the legal consequences of the above two errors of the General Court, the Opinion proposed to the Court not to set aside the judgment under appeal, because its operative part was well-founded on other legal grounds. So the General Court’s errors were not capable of leading to the annulment of its judgment and the AG invited the Court of Justice simply to make a substitution of grounds (substitution de motifs).
II. General Court
This was clearly the judgment of the summer! In 2018, the European Commission imposed a EUR 1 billion fine on the world’s largest supplier of baseband (LTE) chipsets. The Commission found that Qualcomm abused its dominant position by paying Apple to exclusively source its iPhone and iPad chipsets from Qualcomm during 2011-2016. The payments were in the order of billions. In other words, this was a case of conditional rebates/payments. The Commission also found that Qualcomm’s exclusivity payments denied its rivals (Intel – the world is indeed small) the chance to compete for Apple’s business, which comprised a third of baseband chipset demand. It further asserted that the loss of Apple as a customer had further ripple effects on the market, due to the importance to other customers of Apple’s decisions on procurement and design.
The Commission failed miserably before the General Court and its decision was annulled in its entirety – one of the very rare occasions. In fact, unless I am wrong, this is the first ever full annulment of a Commission Article 102 TFEU infringement decision based on procedural and substantive reasons relating to the abuse findings (i.e. not market definition, dominance or inter-State trade effect). (In Intel we had a full annulment of the fine but a partial annulment of the findings.) The annulment was due to both procedural and substantive reasons, but the General Court was particularly harsh with the Commission’s procedural failures.
Let me start with the first major procedural flaw. Following the decision, it was revealed that the Commission had failed to inform Qualcomm of a number of meetings with third parties. For some of these, the Commission had also omitted to maintain a written record that could be disclosed to Qualcomm.
In the footsteps of the Court of Justice’s judgment in Intel, the General Court confirmed that when the Commission interviews third parties to collect information relevant to an investigation (be it in a meeting or a call (para. 185), it must record a meeting’s content in a form of its choosing (para. 190). The Commission does not discharge this obligation simply by means of recording the participants and agenda items, but must at least “provide an indication of the content” and “nature of information” discussed. The General Court explained that the Commission’s failure to provide Qualcomm with concrete evidence on these meetings – which the Commission itself characterised as “regrettable” (para. 237) – left Qualcomm exposed to the possibility of a third party mentioning evidence, be it inculpatory, exculpatory, or neutral, that Qualcomm would not be aware of (thus, not being able to use it in its defence). The General Court accepted that the interviews at hand “could have related” to matters wholly pertinent to the investigation, such as the competitors’ capacity to supply Apple, their willingness to challenge the agreement with Qualcomm, and the relative merits of Qualcomm’s chipsets. Likewise, knowledge of these interviews “could have proved relevant” to Qualcomm’s defence and “could have enabled” it to tweak its defence strategy accordingly. The General Court further rejected as “speculation” the Commission’s argument that the information provided during these meetings was comparable to the input provided by the third parties in response to the Commission’s requests for information (RFIs) (para. 142). The General Court also rejected the Commission’s argument suggesting that Qualcomm should have asked the Court to hear witnesses during the proceedings to remedy the Commission’s failure to provide meeting notes.
It is interesting that the General Court takes a relatively asymmetric approach with respect to “speculation” as to the content of a meeting. On the one hand, it rejected the Commission’s argument that the third parties would presumably not have produced in the meetings any concrete evidence on top of what they provided in their written responses to the RFIs. On the other hand, it accepted the possibility of Qualcomm fishing out from these meetings evidence potentially useful to its defence. Such an asymmetry makes sense, in my view. Had the Commission prevailed here, it would have essentially secured for itself a blank check for conducting virtually undocumented meetings. Put differently, the General Court recognised that it is not for the Commission to determine which meeting is prone to generate “concrete evidence” and thus merits a detailed account. This also makes sense given we are in quasi-criminal law territory. The judgment is also remarkable in that it reaches a different conclusion than in Intel: in that case, the Court of Justice, after having found that the Commission failed to take minutes of a meeting with an executive of Dell, considered that Intel’s rights of defence had not been breached. As the General Court notes in this case, the Intel case was different because the Commission had given Intel access to an internal note about the meeting and a subsequent submission reflected the content of the meeting.
A final point to be made here relates to the exceptional number of procedural violations. Unlike in Intel, where the failure to keep record of a meeting concerned only one meeting, here the Commission failed to properly record seven meetings, with six different third parties. Notes of all these meetings were sketchy at best and thus useless to Qualcomm (and to the General Court). And, regardless of the absence of notes, the very existence of these meetings was concealed until after Qualcomm was fined (and for one meeting even after the Commission’s Rejoinder, forcing the General Court to adopt measures of organisation and hear the parties on the newly revealed documents).
The second major procedural flaw had to do with certain differences between the Statement of Objections and the decision. The Commission was initially investigating an abuse on two distinct relevant markets, LTE chipsets and UMTS chipsets. It alleged an abuse in both markets. In response, Qualcomm submitted a “critical margin analysis”, a lighter version of the AEC test done in Intel, covering both of these markets. However, the Commission’s ultimate decision dropped the charges on the abuse on the UMTS segment and found an abuse only on LTE chipsets. Qualcomm thus complained against the Commission’s failure to inform it of the change and allow it to amend the data and the analysis used in its economic analysis. The General Court found that, by submitting the critical margin analysis on LTE and UMTS chipsets, Qualcomm exercised its Intel right to produce economic evidence that its conduct was not capable of restricting competition. The Commission actually examined this analysis, rejected it, and carried out a revised analysis in the contested decision. By then, however, the analysis had become obsolete on account of the reduced scope of the alleged abuse. The General Court thus held that, although the Commission was perfectly entitled to reduce the scope of its investigation to the defendant’s favour, without explaining this, it should have nevertheless brought that to the defendant’s attention, giving it the opportunity to adapt its economic analysis (para. 340). By failing to do that, the Commission had made a similar sort of error as seen in UPS, said the General Court.
The General Court’s conclusion was that these above procedural flaws sufficed to annul the Commission’s decision (para. 345). However, the judgment examined also the substantive plea on absence of anticompetitive effects “[i]n the circumstances of the present case” and “in the interests of the proper administration of justice” (para. 346). In that respect, Qualcomm claimed that the Commission had not proven that its payments to Apple were capable of having anticompetitive effects by reducing Apple’s incentives to source LTE chipsets from Intel.
First, the General Court noted that the analysis of whether a conduct is capable of foreclosing as-efficient competitors must take account of “all the relevant factual circumstances” and “cannot be purely hypothetical” (paras 396-397). It then found that, with respect to iPhones, which accounted for approx. 90% of Apple’s LTE chipset requirements, there was no contestable demand; the Commission itself acknowledged that, between 2011 and 2015, Qualcomm was the sole supplier capable of meeting Apple’s technical requirements. Even so, the Commission consistently referred to Apple devices in its decision, without distinguishing between iPhones and iPads. The General Court thus found that the Commission failed to “make a link” between the “undisputed fact” of iPhone chipset demand being uncontestable and the alleged lessening of Apple’s incentives to switch to Intel (paras 410, 412). More boldly, the General Court acknowledged that, in light of the absence of alternatives, Apple’s decision to single source from Qualcomm “could fall within competition on merits” rather than an “anticompetitive foreclosure effect” (para. 414). To put it differently, there was no competition in the first place that was restricted, since Apple would have sourced from Qualcomm anyway. This is effectively a counterfactual argument that was successful. Interestingly, the Court does not refer to “counterfactual” by name but prefers to view this under the “all circumstances” test. That’s not material. The fact remains that the counterfactual argument was successful.
Of course, readers might ask themselves: if that is the case and Apple would have bought in any event only from Qualcomm, why did Qualcomm pay billions, if the exclusivity was unnecessary? Surely, Qualcomm’s managers were not making a donation to the richest company in the world? I have come across this argument before. What we sometimes fail to recognise ex post, as enforcers or as competition law specialists, is that running a business is much more complicated. I have seen cases many times where rebates, discounts, and other payments are given based on conditions that make no sense, because the customer would still have bought from the dominant company. However, such payments may be necessary in terms of building a relationship and of ensuring a win-win result. Sometimes – interestingly – such payments simply mean that a customer has substantial buyer power.
Moving on to iPad chipsets, the Commission had sought during the General Court hearing to argue that, even if the contestable demand were to be confined to them, Qualcomm had leveraged its non-contestable demand for iPhone chipsets to foreclose competitors on the contestable iPad chipset demand. The General Court said that this change in the Commission’s theory of harm was not acceptable, and there was no evidence brought forward in the contested decision to support it (para. 421). In short, the General Court says that the Commission cannot rewrite the decision at the appeal stage.
Then, Qualcomm disputed the Commission’s finding that the exclusivity payments had an effect on Apple’s procurement incentives and strategy with respect to iPads actually launched in 2014 and 2015. Here, the Commission admitted a “clerical error”, revealing that its analysis on the actual effects of payments only concerned a certain type of iPad models “to be launched” and not “actually launched” in 2014 and 2015 (paras 394, 457, 458), whereas the decision referred to Apple devices, including iPhones. The General Court agreed with Qualcomm that it was not legitimate for the Commission to make a conclusion on Apple’s incentives without first exploring its actual alternatives for chipsets covering the iPad models concerned, under the same technical and scheduling requirements (para. 479).
Finally, the General Court reviewed whether the Commission had proved an actual effect on Apple’s incentives to switch chipset supplier for certain iPad devices. The General Court accepted that Apple had indeed taken into account the exclusivity payments from Qualcomm, but that did not suffice. The General Court was not convinced by the internal documents presented that, absent the payments, Apple would have switched (para. 488). It further rejected the alternative assertion that, absent actual effects, these payments were at least capable of foreclosure under the circumstances. Because the Commission’s theory of harm covered exclusivity payments for the iPhone and iPad chipset requirements of an entire six-year period, it was not fit for the fringe part of iPad chipset demand that the Commission was using it for.
On these multiple counts, the General Court quashed the Commission’s analysis of anticompetitive effects, since the conduct’s capacity to foreclose was not analysed “in the light of all the relevant factual circumstances,” while the alleged actual effects of such conduct were not verified based on the evidence brought forward (para. 511).
Finally, I note with satisfaction that Qualcomm amounts to another Article 102 TFEU judgment that strongly supports the “as efficient competitor” principle (e.g. paras 349-351). The General Court actually goes even further. In response to a Commission argument, it says in para. 416 that the Commission never held that Qualcomm’s conduct foreclosed less efficient competitors. In any event, the Court adds, the Commission could not have gone there, since Article 102 TFEU does not protect “competitors which are less efficient and so less attractive to consumers from the point of view of, among other things, price, choice, quality or innovation”. This emphasis is very welcome.
The General Court was also active in the merger control area. It delivered three judgments, which rejected the appeals against the Commission’s decisions and which I can treat in a summary fashion.
In Canon, the General Court dismissed the appeal against a decision imposing a EUR 28 million fine on Canon for gun-jumping in the context of its acquisition of Toshiba Medical Systems Corporation (TMSC). The Commission had taken issue with the warehousing structure that Canon had put in place and held that this in effect amounted to gun-jumping because Canon had partially implemented the acquisition, as this structure contributed to a lasting change of control in the target. The General Court agreed and held that a breach of the standstill obligation does not require that control be acquired in full or in part. Indeed, the implementation of a concentration is not limited to the acquisition of control, but can also cover any operations which contribute to a lasting change of control of the target.
In Wieland-Werke, the General Court dismissed Wieland’s against the Commission’s decision that blocked Wieland’s acquisition of its rival Aurubis. The appeal centered on the Commission’s market definition, theory of harm, and substantive assessment but the General Court thought that the Commission made no error. There was nothing wrong with the Commission’s market definition, which was based on conventional methods. In addition, the Commission was justified to define the market in a certain way but then focus its substantive analysis on specific segments of that market. Finally, among other things, the General Court dismissed Wieland’s arguments suggesting that the Commission’s theory of harm was erroneous, in that it mixed horizontal and vertical effects. Furthermore, the Commission had not overstated Wieland’s relative competitive position and the closeness of competition between the two parties. So, a good result for the Commission.
Finally, in thyssenkrupp, the General Court dismissed the appeal against the Commission’s decision that blocked a joint venture between Tata Steel and thyssenkrupp, because the JV would have reduced competition and increased prices for different types of steel. This was again a “gap” case, i.e. the European Commission prohibited a transaction that did not create or strengthen a dominant position but nonetheless risked causing a “significant impediment to effective competition” by changing the market structure. In other words, a case like CK Telecoms! As a reminder, the CK Telecoms judgment of the General Court is currently under appeal and on 14 June we had the hearing before the Court of Justice. In no other case has the Commission been so vocal against a General Court judgment. The thyssenkrupp judgment is quite long and the General Court’s analysis painstaking. The General Court thought that the Commission had done a good job. There is a lot of interesting language on the non-necessity of the SSNIP test for market definition, closeness of competition, etc. The absence of two words, two adjectives in particular (“high” and “strong”), has excited the commentators. Just compare paras 280 and 288 in thyssenkrupp, which refer to a “sufficient degree of probability”, with paras 118 (“strong probability”) and 275 (“sufficiently high degree of probability”) in CK Telecoms. Does the thyssenkrupp text indicate a certain retreat on the part of the General Court? It could be. Does it mean anything for CK Telecoms? No, because a three-member chamber judgment in the General Court will not have any impact on the appeal pending before the Court of Justice’s Grand Chamber! Anyway, this is all speculation now. We need to be patient for a few more months – probably early 2023.
III. European Commission
A notable case was the commitments decision that ended the Commission’s investigation against Insurance Ireland, an association of Irish insurers. According to the Commission, the commitments will ensure fair and non-discriminatory access to Insurance Ireland’s Insurance Link information exchange system, which contains important data for companies offering motor vehicle insurance services in Ireland. The Commission’s concerns were that Insurance Ireland arbitrarily delayed or in practice denied access of non-members to its platform, thereby placing them at a competitive disadvantage vis-à-vis its members who had access to the platform. This acted as a barrier to entry, particularly for insurers based in other Member States, ultimately reducing the possibility of more competitive prices and choice of suppliers for consumers seeking motor vehicle insurance in Ireland.
The commitments made binding by the Commission oblige Insurance Ireland: (i) to make the access to the Insurance Link information exchange system independent from membership to Insurance Ireland; (ii) to change the access criteria to Insurance Link and make them fair, objective, transparent and non-discriminatory and to apply them uniformly to all applicants, from Ireland and other Member States; (iii) to establish a new Insurance Link application procedure with a defined timeline that will be handled by an operationally independent Application Officer, who is of a sufficient level of seniority and has experience in the insurance sector acquired in a professional capacity; applicants who have been refused access will be able to appeal to the Oversight Committee, an independent appeal body; (iv) to establish a cost and usage-based fee model and to ensure that a fair, transparent and non-discriminatory fee will be charged to Insurance Link users; and (v) to ensure that the criteria for becoming member of the Insurance Ireland association will be fair, objective, transparent and non-discriminatory. The commitments will remain in force for 10 years.
- New Vertical Block Exemption Regulation, Vertical Guidelines, and other developments
Clearly the most important development has been the adoption of a new Vertical Block Exemption Regulation (VBER – Regulation 2022/720) and new Vertical Restraints Guidelines, which entered into force on 1 June 2022. The Commission’s ambition was to shed more light on how companies should assess distribution agreements, especially in the context of e-commerce. The new package has brought a number of new developments.
On dual distribution, the Commission has opted to extend the protection to cover wholesalers and importers (Article 2(4) VBER) but online intermediation services, including hybrid platforms, would not be protected as set out in Recital 14 and Article 2(6) VBER. In addition, to avoid infringing competition law, according to Article 2(5) VBER, an exchange of information in the framework of a dual distribution will not be protected if it is not directly related to the implementation of the vertical agreement and if it is not necessary to improve the production or distribution of the product in question. This clarification is important since the Commission has now dropped the exemption on exchanges of information between parties with up to 10% combined market share at the retail level.
On exclusive distribution agreements, it is noteworthy that suppliers now have the possibility to appoint up to a maximum of five distributors per exclusive territory or customer group and, moreover, suppliers can require distributors to pass on active sales restrictions to their customers (Article 4(b)(i) VBER). On selective distribution agreements, the new VBER notes that suppliers may prohibit buyers and their customers from selling to unauthorized distributors located in a territory where the supplier operates a selective distribution system (Article 4(c)(i)(2) VBER).
In the field of parity or so-called MFN obligations, the new package brings some novelties. Under the previous regime, all types of parity clauses were exempted, regardless of whether they were wide or narrow. Now, under the new regime, wide parity clauses are no longer exempted and must be assessed individually under Article 101 TFEU. In contrast, when it comes to the rest of the parity clauses, these are still exempted. Having said that, it must not be forgotten that the VBER includes in its Article 6 the possibility of withdrawal of an exemption, specifying that when the alleged parity clauses are used by platforms covering a significant share of users and there is no evidence of efficiencies, the exemption is likely to be withdrawn.
Finally, with regard to the practices covered by the concept of “hardcore restrictions”, resale price maintenance continues to be categorised as such, whereas dual pricing is not considered anymore a hardcore restriction and thus, under certain circumstances, suppliers are allowed to charge different prices for products sold offline and online. So clearly a new regime that is stricter for digital platforms…
Finally, on 30 June, the Commission launched a consultation on the performance of Regulations 1/2003 and 773/2004, which lay out the procedures for the application of EU competition rules and were adopted 20 years ago – at the modernisation time. So this is the precursor to adopting a new “Regulation 2” that will amend Regulation 1/2003. My sense is that the Commission is focusing on the “second part” of Regulation 1/2003, which regulates its investigative powers, the procedural rights of parties to investigations and third parties, and generally its enforcement powers. While I agree that this part needs a revisiting (and the Qualcomm debacle is a good reminder), I still think that the “first part” of Regulation 1/2003, which defines the relationship with national competition law and with NCAs and national courts, though politically tricky, deserves attention, too. I believe we have serious issues with Articles 3, 11, and 16 of Regulation 1/2003. Unfortunately, there is a lot of fragmentation and re-nationalisation of competition law enforcement in Europe lately. Interested parties can respond until 6 October 2022.
Citation: Makis Komninos, Competition Stories: May & June 2022, Network Law Review (August 8, 2022)
Read all the competition stories over here: link