Chillin'Competition

Last week gave us, among others, Advocate General Rantos’s Opinion in Unilever (see here for the French version). As Servizio Elettrico Nazionale, the preliminary reference comes from the Italian Consiglio di Stato and originated in abuse of dominance proceedings before the AGCM.

Unilever raises two points of law. One relates to the single economic entity doctrine. The second, on which this post will focus, to the meaning and scope of the Intel judgment of 2017.

In essence, the Consiglio di Stato asks whether the framework laid down in Intel is also relevant outside the specific context of loyalty rebates, and more precisely where a dominant firm imposes outright exclusivity obligations on its customers.

The question from Italy’s highest administrative court has two elements: one substantive and one procedural. The substantive aspect has to do with the need to assess, as a matter of law, the capacity of an exclusivity obligation to foreclose competition. The procedural side of things concerns the need for the administrative authority to engage with the economic arguments raised by the dominant firm.

AG Rantos’s view on the point of principle is clear: the framework laid down in Intel applies irrespective of the practice at stake, to the extent that the dominant firm provides evidence showing the absence of effects (para 71).[1]

Thus, whenever the ‘Intel test’ is triggered, the authority is required to assess the actual or potential impact of the practice in light of the five criteria identified by the Court.

This clarification would be relevant when assessing the legality of ‘by object’ conduct, such as exclusive dealing (at stake in the case), predatory pricing and tying, since a finding of abuse does not necessitate, in principle, an assessment of effects.

It is difficult to disagree with AG Rantos on this point. As the Opinion explains, the very letter of the Intel judgment appears to suggest that the framework applies to any dominant undertaking (paras 74-76).

In addition, a teleological interpretation of Article 102 TFEU would go to confirm this view. As the Court made explicit in Generics, a finding of abuse presupposes that the contentious practice is capable of restricting competition.

Thus, it is only logical that arguments pertaining to the absence of actual or potential effects are considered across the board, and not only in the narrow factual circumstances of Intel (paras 78-79).

The answer to the substantive aspects of the question already addresses, by and large, its procedural dimension. If the Intel framework is applicable irrespective of the practice, by necessity a competition authority is under a duty to consider the arguments of an economic nature raised by the dominant firm (para 84).

The Opinion makes several important points from a procedural perspective. First, AG Rantos reminds us not to lose sight of the fact that authorities have the legal burden of proving the infringement, even if it is for dominant firms to reverse the presumption of foreclosure/exploitation in ‘by object’ cases.

Second, competition authorities cannot reject outright the economic evidence put forward by the parties, except by showing that the methodology relied upon by the dominant firm is not capable of substantiating claims about the potential effects of a practice (para 85).[2]

This point addresses one of the gaps left by previous case law, namely the standard of proof that dominant firms would have to meet to trigger the ‘Intel test’.

AG Rantos appears to suggest that authorities would only be dispensed from the need to engage with the evidence provided by a dominant undertaking when such evidence is irrelevant for the purposes of the assessment (a very low threshold indeed). Importantly, the Opinion makes it clear that, even in this scenario, the authority would still be subject to a duty to state reasons (para 85).

All in all, the answer to the question raised by the Consiglio di Stato seems straightforward. Its significance for future Article 102 TFEU cases, on the other hand, cannot be overstated. Some of the themes addressed by AG Rantos, such as whether the evidence has been adequately considered by the authority, are likely to be relevant again soon.

---

[1] The French version reads as follows: ’71. Pour les raisons suivantes, et ainsi qu’il a déjà été indiqué au point 63 des présentes conclusions, j’estime que ce même principe vaut de manière générale, et indépendamment du type de restriction, lorsqu’une entreprise dominante avance des preuves visant à démontrer que le comportement en cause n’était pas susceptible de produire de tels effets‘ (emphasis in the original)’.

[2] ’85. Or, même si l’autorité de concurrence considère, comme en l’occurrence, que la méthodologie utilisée aux fins de l’étude économique n’est pas pertinente, elle ne peut pas exclure d’emblée la pertinence d’une telle étude, sauf à indiquer, dans la décision par laquelle cette autorité qualifie un comportement d’« abusif », les raisons pour lesquelles elle estime que la méthodologie sur laquelle repose cette étude ne permet pas de contribuer à la démonstration du fait que les conduites mises en cause ne sont pas aptes à exclure des concurrents aussi efficaces‘.

Exactly 7 years ago, on 16 July 2015, the CJEU rendered its Judgment in Huawei v ZTE (here are the comments I published that day).

The Huawei v ZTE Judgment essentially sought to clarify the circumstances under which the seeking of injunctions by a SEP holder could constitute an abuse of dominance. The Judgment confirmed the view, initially advanced in academic circles, and endorsed by the European Commission in Samsung and Motorola, (and vehemently opposed by many) that in certain cases patent hold up was a competition law problem connected to the leveraging of market power obtained through standardization. The underlying idea was that hold up could materialize in refusals to licence, excessive royalties or injunctions. In that Judgment the Court set up a procedural framework balancing the different stakes and incentives at issue.

7 years later many of these debates remain (and remain equally bitter). Interestingly, though, there appears to have been an effort to shift attention away from hold up and focus, instead, on hold out  (i.e. the situation where implementers would allegedly refuse to negotiate in good faith). The argument is that innovation on the part of SEP holders would be discouraged should their royalties not be high enough as a result of hold out.

Paying attention to potential hold out on case-by-case assessments might be, to some extent, natural because  implementing the procedural framework set out in Huawei v ZTE necessarily requires assessing whether implementers have entered into bona fides negotiations.

At the same time, however, the recent trend is to present hold out (aka “reverse hold up”) as the other side of the same coin. This view has made it from economic articles, to national litigation, to the “new Madison” policy in the US under AAG Delrahim. More recently, and more surprisingly, the European Commission’s draft horizontal Guidelines (recital 470) would appear to support this view:

“When the standard constitutes a barrier to entry, the undertaking could thereby control the product or service market to which the standard relates. This in turn could allow undertakings to behave in anti-competitive ways, for example by refusing to license the necessary IPR or by extracting excess rents by way of discriminatory or excessive royalty fees thereby preventing effective access to the standard (“hold-up”). The reverse situation may also arise if licensing negotiations are drawn out for reasons attributable solely to the user of the standard. This could include for example a refusal to pay a FRAND royalty fee or using dilatory strategies (“hold-out”)”.

Perhaps it is simply a drafting problem, but this paragraph appears to put hold up on the part of SEP holders and hold out on the part of individual users of the standard (and the concerns to which they both relate) at the same level, also from a legal standpoint. This is interesting for various reasons that we have often discussed on this blog. First, the shift in the focus of these debates is one more example of the pendulum oscillations that characterize competition law, but one where the swing would appear to be particularly wide. Second, this text would also appear to equate hold out practices with anticompetitive hold up practices on the grounds that both can affect the distribution of rents between the different parties, regardless of whether they involve the exercise of market power or not.

I would welcome your views on this point. Not having worked for clients on these issues, I have no view on the extent to which hold out may be a real-life concern. As a competition lawyer, however, I have trouble seeing how hold out practices could lead to genuine competition law concerns (i.e. how they could lead to foreclosure, anticompetitive leveraging, exploitation or otherwise restrict competition) absent dominance or a cartel/boycott-like arrangement at the level of would-be licensees. I see that others have expressed very similar thoughts (e.g. here or here).

Don’t get me wrong. As mentioned above, hold out considerations can be, and have been, relevant in case-by-case assessments under the Huawei v ZTE framework (under that framework injunctions remain legitimate in relation to implementers not acting in good faith). But to the extent that hold out concerns may be concerned with relative bargaining power (as opposed to market power) and with the distribution of rents between SEP holders and implementers (absent market power, exploitation or foreclosure), they would not appear to be a matter for competition law to address. In sum, while the narrative, the incentives, and perhaps even the economics, may be “the reverse” as those arising in hold up scenarios, this might not be accurate from a legal standpoint.

Once again, thanks to Lewis Crofts and his reporting abilities via Twitter, I have been able to get a sense of what has been going before the Court this week. If you have not done so, go check his tweeting on ISU and Super League. These two cases will have a major impact on the relationship between competition law and sports governance.

I have been following this topic closely for a while (the updated version of my paper on sports governance can be found here; I am really grateful, by the way, to those who reached out with comments).

Lewis’s reporting gives me the impression that much has been discussed about sports associations’ alleged ‘conflicts of interests’. According to a particular school of thought, it is concerning that governing bodies enjoy the power to regulate the sport and, at the same time, to authorise rival competitions.

I have never really understood why this idea has managed to gain so much traction. It is definitely an astute spin on the issues. As an outside observer, however, I am not sure it makes sense to frame discussions in terms of conflicts of interest. It is, in fact, a problematic way of looking into the underlying substance.

It is problematic, first, because it suggests that the fact that a firm protects its own economic interests is somehow a concern under competition law (of all disciplines). Second, because it is based on the (now discredited) idea that one can meaningfully distinguish between, respectively, sports-related and economic considerations.

Third, and finally, because it would be unfair to sports associations, in the sense that it would demand more from them than from any other entity engaged in an economic activity.

‘Conflict of interest’ is just another way of saying ‘protecting one’s economic interest’ (which has never been presumptively anticompetitive)

Discussions around conflicts of interests in cases like ISU and Super League give the impression that the situation is specific or unique to sports governing bodies, in the sense that it does not arise elsewhere in the economy (or only rarely).

In reality, the only thing that is unique to sports is the vocabulary used to frame the underlying issues. When it comes to the substance of these issues, there is nothing special, let alone exceptional, about the situations described in the abovementioned cases.

In reality, ‘conflict of interest’ is another way of saying that governing bodies have put in place mechanisms aimed at defending their economic interests. And we know from the case law that doing so is not necessarily or presumptively anticompetitive (not even when the firm enjoys a dominant position).

Just to illustrate how pervasive (and, sometimes, even prima facie pro-competitive) so-called conflicts of interest are, consider the following examples.

A franchisor finds itself in a position that is not fundamentally different from that of a sports governing body. It dictates the rules of the system (brand image, quality of the products, look and feel of the stores) and also limits competition: franchisees will typically be subject to a non-compete obligation preventing them from concluding similar agreements with other suppliers, or setting up rival shops themselves.

In spite of the blatant conflict of interest, franchising agreements are prima facie lawful under Article 101(1) TFEU.

Consider also the proverbial refusal to deal scenario. A vertically-integrated firm that produces an input and also manufactures the finished product would also be in a ‘conflict of interest’: this firm would be able to control competition against itself on the downstream market.

It is clear from the case law, however, that the vertically-integrated firm cannot be compelled to deal with rivals absent exceptional circumstances. This is so in spite of the fact that its dual status as supplier and competitor to its own (would-be) customers necessarily creates a conflict.

The ‘conflict of interest’ test would apply a stricter standard to sports governing bodies for no valid reason

To the extent that ‘conflict of interest’ is just another way of saying ‘undertaking acting as expected in a system based on undistorted competition’, there is no reason to make it presumptively anticompetitive.

What is more, seeing with suspicion this alleged ‘conflict of interest’ would lead to governing bodies being treated more strictly than any other undertaking in competition law. There seems to be no valid reason justifying this differential treatment.

As the Court held in Meca Medina, there is nothing that immunises sports associations from the application of competition law. They may be scrutinised under Articles 101 and 102 TFEU, just like any other economic activity.

Conversely, it is not because a case is about football or skating (as opposed to the manufacturing of aminobutanol, or the delivery of newspapers) that so-called ‘conflicts of interests’ should become problematic ipso facto.

As explained in my paper, cases like ISU and Super League are best understood when examined through the lenses of other horizontal co-operation agreements raising similar issues.

Think of Gottrup-Klim. The cooperative in that case faced a ‘conflict of interest’ just as much as the UEFA or the ISU do. In fact, members of the cooperative were prohibited from taking part in competing ventures.

The association decided who was entitled to compete with itself, while regulating the joint purchasing activities. Alas, the Court held that these non-compete obligations were ancillary and did not restrict competition, whether by object or effect (just like franchising).

The divide between economic and non-economic interests is more of an illusion than a reality

I can think of a final reason why the idea of the ‘conflict of interests’ of sports associations is problematic. It seems to be grounded on the premise that one can establish a clean divide between economic and non-economic measures. According to this understanding, governing bodies would combine functions relating to sport and then economic functions.

The problem with this understanding is that it is, at best, an oversimplification of what goes on in the sector. Scratch beneath the surface and you will realise that this clean divide between the economic and non-economic does not reflect the reality.

Think of salary caps, thinks of rules limiting how much money can be spent on transfers, and on how the revenues generated by a competition are to be allocated. Are these measures economic? Certaintly. Do they serve non-economic aims as well? Without any doubt: they will are typically introduced to achieve competitive balance.

The bottomline is the same: to the extent that they relate to the exercise of an economic activity, the rules adopted by governing bodies are best scrutinised, in the usual way, under Articles 101 and 102 TFEU.

As ever: nothing to disclose in this and, indeed, any other case.

On 15 June the General Court (“GC”) annulled the Commission’s decision imposing a close-to-1-billion euro fine on Qualcomm in relation to alleged exclusivity payments made to Apple in breach of Article 102 TFEU. In a previous post I discussed the Judgment focusing on procedure; this second post deals with substance, and concludes my comments on what this means for future cases.

What the case was about. The case concerns rebates (in the form of direct payments from Qualcomm to Apple) in exchange for Apple exclusively incorporating Qualcomm LTE chipsets in certain devices. The Qualcomm decision was the first Commission decision concerning exclusivity rebates following the CJEUs’ Judgment in Intel.

The Commission’s approach. The Commission largely built its case on the wording of the agreements, the extent of Qualcomm’s alleged dominance, the importance of Apple as a key client and certain (see below) internal documents from Apple. For additional background on how the Commission saw the case, see this guest post from Max Kadar that we published in 2020.

What the Judgment does not address. Throughout the case there were relevant discussions about market definition and dominance (including on the question of whether conducting a SSNIP test was necessary or not) and on the binding nature of the Guidance Paper on Article 102 (see here for my take; the Guidance Paper will likely be withdrawn before the Courts address this), but the Judgment did not need touch on those points as it annulled the decision on other grounds.

The Judgment’s substantive review.

The principles. The Judgment starts off recalling the principles that competition law is not concerned with the exclusion of competitors due to their inferior efficiency because dominant companies cannot be prevented from competing on the merits (349-351). The Judgment also recalls the special responsibility of dominant firms (352), retains the formal “presumption” that exclusivity arrangements are abusive (353) subject, however, to the “further clarification” in Intel (354). It recalls that for conduct to be abusive, it must be capable of producing exclusionary effects, and that this assessment requires examining all the relevant circumstances (355). The Court also recalls the Galp case law indicating that post-decision elements may be relevant (357) Tetra Laval standard of review applicable to complex economic assessments (358), and the rules on the burden of proof/presumption of innocence (359).

The F word. In my mind, the principle in 354 (the one set out by the CJEU in Intel) makes clear that, under Art. 102, effects=capability to foreclosure (that is also view that the Commission advanced in its Guidance Paper). The Court is saying that, even for conduct presumed abusive, whenever the dominant company offers evidence challenging capability to restrict, then it is for the Commission to carry out a foreclosure analysis. This is not new, but it’s relevant here, because the Commission’s decision does not refer to foreclosure (just run a Ctrl+Find search here; 373 shows that Qualcomm made the same point). The Commission’s argument is that there was no need to show foreclosure, and that the restrictive effect consisted in the reduction of Apple’s incentives to switch to competitors (381, 384). This debate is also key several pending cases (including some in which I represent clients), but I will reserve my views on those for now. Its also worth noting that the GC’s review looks both at “real world” foreclosure effects and at “hypothetical AEC world” foreclosure effects.

The relevant circumstances. The Court observes that while the decision defines a worldwide relevant market for LTE chipsets, the alleged abuse concerns a single important customer (380). It also notes that the decision did not allege the existence of a strategy to foreclose (383). It then goes on to examine several circumstances.

• Scope of the conduct. The Judgment observes a mismatch between, on the one hand, the Commission’s analysis and findings of abuse (which relate to LTE chipsets for iPhones and iPads) (389-391) and, on the other hand, the Apple documents and explanations invoked to support those findings, which only referred to certain iPad models which Apple planned to launch in 2014 and 2015 (395, see also 420 and 422).
• The counterfactual. The Court notes that, according to the decision itself (322), between 2011 and 2015 “Apple had no alternative as regards its requirements of LTE chipsets for its iPhone devices” (400); this, the Court underlines, is common ground (403, 405). Since iPhones represented approximately 90% of Apple’s requirements of LTE chipsets (408), this means that for a very large part of Apple’s requirements covered by the decision Apple could not have switched to competing LTE suppliers (409-410). The Judgment observes that, while the decision acknowledged this fact, it failed to consider it when analyzing whether Qualcomm’s conduct was capable of restricting competition (412, 415). The Judgment does not use the term “counterfactual”, but evidently applies that logic: Qualcomm’s payments could not have reduced Apple’s incentives to switch to rivals because, even absent that conduct, Apple could not have switched to rivals. In other words, there was no competition that Qualcomm could have restricted. For this reason, the Court concludes that the decision failed to take into account all the relevant circumstances (417).
• Conditions for granting the payment / exclusivity label. The Court does not dispute that the characterization of the payments as exclusivity payments, but explains that this is “not sufficient to conclude that those payments constituted an abuse” (424).
• Other circumstances invoked by the Commission. At 425 the Court finds that while the circumstances invoked by the Commission (extent of the dominant position, conditions for granting the payments, their amount and duration or the importance of Apple as a customer) “should not be disregarded, the fact remains that those factors (…) do not in themselves demonstrate, in the present case, anticompetitive effect and, in particular, foreclosure” (425). What this means is that “all the relevant circumstances” includes “all the relevant circumstances”, not only those that may support the finding of abuse.

The Judgment also engages, for the sake of completeness (442-444), in an analysis of whether Qualcomm’s conduct could have influenced Apple’s sourcing decisions concerning 2014 and 2015 iPads. In a nutshell, the Court observes that this section of the decision is based on Apple’s internal documents and explanations regarding only to certain versions of certain iPad models which were to be launched in 2014-2015 (439, 450), which do not necessarily match  the Commission’s conclusions (455-456); the Court dismisses both the Commission’s argument that this was a clerical error (458). For this reason, the Court finds that the evidence on which the Commission relied is “inconsistent, both internally within such evidence, and in relation to the findings which it seeks to support” (462-463).

The Court goes on to address Qualcomm’s argument that the Commission failed to take into account evidence that demonstrated that Apple did not select Intel’s chipsets for reasons other than the payments concerned. [This is related to both the procedural points we discussed in the previous post and to the discussion on the counterfactual above]. This has to do with the question of whether Intel’s chipsets met Apple’s technical and schedule requirements. The Commission, based on certain Apple documents and explanations, concluded that it did. But, in the Court’s view, the evidence obtained by Qualcomm through the Section 1782 application (that the Commission opposed) “gives rise to doubts in that regard” (467, 476). Para. 475, for example, shows that the decision had relied on an internal Apple email, but that Qualcomm obtained other follow-up emails suggesting the opposite. For these reasons, the Judgment concludes that the decision failed to carry out a “true examination” of whether alternative chipsets could have met Apple’s requirements (480, also 477) and, therefore, failed to take into consideration all the relevant circumstances.

Finally, the Judgment concludes that the evidence in the decision was not only inconsistent and incomplete, but also that it was incapable of substantiating the conclusions drawn from it (483). In essence, the Court finds that the decision relied exclusively on Apple’s statements and documents which either did not refer to the models at issue (492, 496), were not conclusive (494), or had been internally challenged by other Apple’s employees according to evidence obtained by Qualcomm (498, 499). In a harsh concluding recital (505), the Judgment concludes that the Commission “in the context of a general analysis mixing models and years, relied on evidence which is not relevant, which is contradicted by other evidence or which is not capable of substantiating its conclusions (…) and which, therefore, does not make it possible to demonstrate that the payments concerned actually reduced Apple’s incentives to switch to the applicant’s competitors to obtain supplies of LTE chipsets”.

Comments

This is a thorough and (despite the confidentiality challenges) very clear Judgment articulating the principles set by the Court of Justice in its 2017 Intel Judgment. It is probably noting that the Commission’s case was initially conceived in one world (after the 2014 Intel Judment) but born in a very different (after the 2017 Intel Judgment).

From a legal standpoint, I do not see anything innovative other than the important message that enforcers must consider all, not only some, relevant circumstances. Perhaps ironically, the message from the Court in this case is, in my view, not very different from that of the Apple State aid Judgment (see my comments here). This is what I wrote exactly 2 years ago when that Judgment was delivered; everything applies equally, word by word, to this case:

“The Judgment will attract more attention than other annulments, but in reality, it is not in any way groundbreaking from a legal standpoint. The reasons, and the reasoning, leading to today’s annulment are exactly the same as the one that has led to the recent annulment of other decisions, including in Frucona Kosice, FC Barcelona, Real Madrid, Naviera Armas, Valencia and Elche cases, among others. In recent years the Courts have consistently insisted on the Commission’s obligation to actively and impartially gather and assess all the relevant evidence in relation to issues where the burden of proof is incumbent upon it. It was all there.

You might think this is easy to say in retrospect, but we already anticipated all of this here and here. At the time, we said that “it won’t be difficult for the Commission to continue to win cases if it incorporates this logic into its day-to-day. If that does not happen, we are likely to witness a series of annulments based on this logic (…) My bet is that I will be making a few future cross references back to this prediction”. Here’s one more cross reference, probably not the last one”.

Ditto.

AG Rantos delivered today his Opinion in the Lithuanian Railways case (see here for Pablo’s comments on the General Court’s Judgment). While the case concerns what Pablo has described as perhaps “the most blatant abuse that the Commission has ever considered“, AG Rantos has managed to use the opportunity to shed some welcome light on some contentious issues.

The Opinion is of particular interest in relation to the interpretation of the Bronner Judgment and the indispensability condition, which has been the subject of much debate, including on this blog. This issue is also relevant to cases where I am acting for clients, so I will stay away from discussing the relevance of this Opinion to those cases.

As you will see, the Opinion is firmly rooted in the established case law of the CJEU and sets out a clear and clean analytical framework:

First, the Opinion observes that it would appear that the Bronner case law applies to situations where there has been a “request” and a consequential “refusal”, either explicit or implicit. At paras. 74-75 the Opinion explains that conduct that “could be perceived as an implicit refusal of access (constructive refusal to supply) (…) ultimately having de facto the same result as an (explicit) refusal of access” must also be analysed under the Bronner framework where its constituent elements share the meaning intended by the judgment in Bronner.

Second, the Opinion (para. 76) confirms, in line with Slovak Telekom and Van der Bergh Foods, that where a case does not involve an obligation to provide access but rather “the provision of services or the sale of products subject to unfair conditions, the Bronner conditions do not apply“.

Third, and this is in my view the key, the Opinion identifies the legal (paras. 64, 81 and 85) and economic (paras. 65, 86) logic that have always justified the application of the Bronner conditions in certain cases:

-The Opinion explains that, from a legal standpoint, the Bronner conditions are necessary in cases where putting an end to the alleged abuse would have the “consequence” of interfering with the dominant undertaking’s freedom to contract and right to property by requiring firms to dispose of an aseet or conclude contracts with person with whom it had opted not do so (paras. 64 and 81, both citing Slovak Telekom).

-For this reason, the Opinion posits that “any intervention, for the purposes of Article 102 TFEU, which consists in imposing on a dominant undertaking a (complete or partial) duty to supply to its competitors may clearly affect that right and should be carefully considered and justified”. AG Rantos explains that “any approach that involves a strict interpretation and application of that judgment would, in [his] eyes, disregard that underlying purpose“. Accordingly, he argues that Bronner “should therefore be the leading judgment, and the rule rather than the exception” (fn. 19).

-At para. 85 the Opinion argues that the Bronner criteria should apply in relation to infrastructure “of which the dominant undertaking is the owner and which, in principle, result from its own investment”; the accompanying footnote (38) distinguishes these scenarios from others where facilities were developed with public funding.

-The Opinion also recalls that, from an economic standpoint, the Bronner conditions are justified by the desire “to promote competition in the long term, in the interests of consumers, by allowing a company to reserve for its own use the facilities that it has developed“, thereby preserving its incentives to innovate and invest (para. 64). At para. 86, the Opinion endorses the view that the pre-existence of a regulatory duty to supply is a relevant factor to consider in as much as it already affects and takes into consideration those incentives.

For these reasons (in my view, the right reasons) AG Rantos’ Opinion proposes to endorse the General Court’s Judgment which, in turn, validated the Commission’s decision.

To be continued…