Relaxing whilst doing Competition Law is not an Oxymoron

Thoughts on today’s judgments in Lundbeck and Slovak Telekom: expected and valuable clarifications for the future

with 15 comments

Just as expected - Photos | Facebook

The Court of Justice has delivered an important set of judgments in the Slovak Telekom and Lundbeck cases (see in particular here and here). As widely anticipated, all appeals have been dismissed. This fact does not deprive the cases of interest, which always lied in the points of principle at stake in them.

The Slovak Telekom case raised issues relating to the scope of Bronner and the indispensability condition. The Court confirms that Bronner is alive and well and defines the scope of the ruling in a manner that is consistent with its spirit and logic (today’s judgments are very much in line with the Advocate General’s Opinion, which we discussed here). Crucially, the judgment also provides a valuable template for the evaluation of cases in digital markets.

The issues raised in Lundbeck, in turn, concerned the notion of restriction by object. It was an occasion to test the meaning of the ‘robust and reliable experience’ test laid down in Budapest Bank. Unsurprisingly, the Court concludes that the lack of experience in pay-for-delay cases is of no assistance to avoid a finding of a ‘by object’ infringement. As I explained here, cartel-like conduct is cartel-like conduct, irrespective of whether it is disguised as a pay-for-delay arrangement. And we have decades, if not centuries, of ‘robust and reliable’ experience concerning cartels and analogous practices.

In addition, the Court dismisses, in Lundbeck, the interpretation of the counterfactual advanced by the appellants. Again, this is hardly surprising. I explained in this post why the appellants’ understanding of the notion was not obvious to reconcile with the case law and why, in all likelihood, it would be rejected by the Court (in the same way it was rejected by AG Kokott).

Slovak Telekom: Bronner is alive and well, and has a definite scope of application

Following AG Saugmandsgaard Øe’s Opinion in Slovak Telekom, some commentators expressed concerns about the demise of Bronner. We can now safely say that Bronner is alive and well: in fact, its scope of application and the rationale underpinning it have never been clearer.

In essence, the Court has ruled that the indispensability and elimination of all competition conditions are part of the legal test where intervention would force a dominant firm to conclude a contract with a competitor. As the Court explained in the Slovak Telekom judgment:

’45. The imposition of those conditions [indispensability and elimination of all competition] was justified by the specific circumstances of that case [Bronner] which consisted in a refusal by a dominant undertaking to give a competitor access to infrastructure that it had developed for the needs of its own business, to the exclusion of any other conduct’.

In other words: the key lies in the remedy (in what a finding of infringement would entail in effect). By the same token, where intervention in a competition law case does not necessitate a duty to deal, indispensability would not be an element of the legal test. Such was the case in Slovak Telekom (and TeliaSonera, which is also confirmed), where remedial action only required the Commission to tackle the unfair nature of the practices involved.

The rationale behind the definition of the scope of Bronner is twofold. First, the Court explains that intervention mandating shared access to a facility interferes with the right to property and with freedom of contract (para 46). Second, it concludes (in para 47, and in line with AG Jacobs’ Opinion in Bronner) that promoting short-term competition by means of shared access may be harmful in the long run insofar as it would negatively impact firms’ incentives to invest and innovate.

These clarifications provide a useful template for the evaluation of cases in the digital arena. We now know that, when intervention interferes with a firm’s right to property and/or its freedom of contract, indispensability (and the elimination of all competition) will be a precondition for intervention.

What are the remaining pieces in the puzzle? Mandated shared access to a facility is not the only way in which intervention might interfere with a firm’s right to property. Such interference would also occur when a firm is asked to cease an activity or to sell its assets to a third party (a structural divestiture). Consistency would demand that such forms of intervention, which intrude as much as a duty to deal (if not more) with the right to property, be treated in the same way.

Where does this leave us for the future? My impression is that indispensability would be an element of the legal test, inter alia, in the following scenarios:

  • Intervention requires, in effect, a firm to remove some features from a product (say, a camera from a smartphone).
  • Intervention requires, in effect, a firm to redesign its product so that rivals have shared access to it (for instance, an interoperability duty).
  • Intervention requires, in effect, a firm to change its business model and adopt one that requires dealing with third parties with which it has chosen not to deal (for instance, by forcing a firm to start licensing a product).

Lundbeck: some refinements, and more clarity, about the scope of restrictions by object

Entire libraries have been written about the notion of restriction by object in the past couple of decades. Lundbeck confirms that we may be reaching the ‘end of history’ in relation to the notion. Most controversies have now been addressed and future rulings will provide, if at all, incremental refinements to the concept. Today’s judgments confirm the fundamentals about the Court’s approach. First, considering whether an agreement is restrictive by object demands a case-by-case, context-specific inquiry (see para 115 of Lundbeck).

Second, the fundamental question in this regard is whether the agreement can be explained on grounds other than the restriction of competition. If there is an alternative rationale for the practice, its treatment as a ‘by object’ infringement would not be warranted. In the specific context of pay-for-delay agreements, the analysis would revolve around the size of the payment (see for instance para 115 of the Lundbeck judgment). The qualification of the agreements in Lundbeck was particularly straightforward, as no plausible pro-competitive rationales had been advanced (para 118).

At the margin, however, the judgments provide useful clarifications. Reasonably, Lundbeck argued that there was no experience about the agreement and that, at the time of the infringement, it was not clear whether they were in breach of Article 101(1) TFEU. As a matter of substantive law, this argument was not deemed persuasive.

The Court’s reasoning comes across as sensible. There is certainly a great deal of experience about ‘naked’ market-sharing arrangements. Accordingly, once it is shown, to the requisite legal standard, that the object of the agreement is to restrict competition by paying a potential competitor to stay out of the market, we are in the realm of cartel-like conduct, which has long been known to be harmful, by its very nature, to the competitive process.

A second point concerned the counterfactual. Lundbeck argued in the case that the qualification of the agreement as a restriction by object demands the evaluation of a ‘counterfactual scenario’. This argument failed. According to the appellants, this analysis would have revealed that the generic producers would not have entered the market.

As explained in the judgment, the case law cannot be interpreted as supporting such an interpretation of the notion of restriction ‘by object’. Showing that a firm is a potential competitor is not the same as showing that its entry into the market would have occurred. In this sense, Lundbeck’s argument was found to relate more to the evaluation of the restrictive effects of the agreement than its ‘precise purpose’ (para 140). If such an interpretation of the ‘counterfactual scenario’ were embraced, the divide between object and effect would become blurred (ibid).

The above seems uncontroversial and fully in line with the case law. The point of the counterfactual at the ‘by object’ stage is not to assess the impact of the agreement, but its object. It is already well-established case law that the counterfactual can be relevant, at the ‘by object’ stage, in the following scenarios:

  • To show that there are ‘insurmountable barriers to entry’ and therefore that the generic producer is not a potential competitor (see in this sense Generics). In such circumstances, the agreement would not restrict competition, whether by object or effect.
  • To show that the agreement is capable of having pro-competitive (or at least ambivalent) effects and therefore its object is not the restriction of competition (see in this sense Generics and Budapest Bank).
  • To show, more generally, that the agreement is not liable to restrict competition in the economic and legal context of which it is a part (see in this sense Murphy).

Nothing in Lundbeck challenges the above aspects of the case law. Accordingly, these are factors to consider when evaluating whether an agreement is restrictive by object. Whether these steps are labelled counterfactual or otherwise, seems immaterial. As we have said many times here, fortunately competition law places substance above form.

I very much look forward to your comments. And please stay tuned for a more in-depth analysis of some aspects of the judgments.

Written by Pablo Ibanez Colomo

25 March 2021 at 2:09 pm

Posted in Uncategorized

15 Responses

Subscribe to comments with RSS.

  1. On Slovak Telekom, para. 57 is particularly enlightening, concluding that “the imposition of that obligation [to give access to the local loop] has the consequence that, during the entire infringement period taken into account in the present case, the appellant could not and did not actually refuse to give access to its local loop network.” There is no refusal, de facto or de jure; no “decision making autonomy” over access.

    The implication is that, where there is a regulatory duty to deal, a holder of an essential facility cannot use constructive refusal (unfair terms and conditions) to raise the evidentiary burden for proof of abuse (effectively imposing on the competition authority the indispensability requirement through their own actions). Reminds me of the “clean hands” equity doctrine.


    25 March 2021 at 3:27 pm

    • Good observation. As an economist, I find paragraph 57 of Slovak Telekom somewhat confusing. Effectively, it seems to imply that when there is a regulatory obligation to supply access, notwithstanding the fact that such obligation may be imposed by an ex ante regulator applying lower standards of proof compared to an ‘indispensability’ requirement under competition law, the Courts must operate as if the indispensability threshold has been met.

      This can give rise to the very odd scenario where, if there wasn’t a regulatory obligation to supply access and a firm refused such access, a competition authority or Court may rule in its favour when applying the high standard of proof of indispensability, but as soon as this same firm has an ex ante access obligation imposed on it, it is game over: the firm is legally barred from refusing access, and the Courts are barred from examining indispensability in such a scenario.

      Is this a legal loophole or am I missing something?

      Otherwise, it now becomes very clear why the European Commission is rushing in to impose regulatory access obligations in digital markets!


      25 March 2021 at 5:53 pm

      • @Felipe, I’m not sure I understand the loophole you’re identifying.

        In the absence of a regulatory framework, where the Court must strike the balance between protecting property and protecting competition in the specific circumstances of a case, the defendant can justify the refusal to grant access to its property on the grounds that access is not indispensable for competition.

        Under ex ante regulatory access, the legislator has struck the balance, finding that the infringement of property rights is necessary. Accordingly, the defendant cannot justify a refusal to provide access (unless there are exceptions in the legislation I suppose). It needs to provide access, though there may be some discretion in the terms that it sets. Where there is a complaint over those access conditions, the defendant can try to objectively justify them. Per para. 57, the enforcer still needs to find that the access conditions were abusive. It’s logical then that in such circumstances, access conditions that amount to a de facto refusal would be abusive.


        25 March 2021 at 7:47 pm

      • My point is precisely that an ex ante regulator will strike that balance but in doing so it will be applying standards and objectives that relate to its own powers and duties, and these are likely to be different from those of a Court applying competition law.

        Hence, if the Commission wanted to impose access obligations as a remedy on a dominant firm, it would have to prove indispensability under competition law, but not under ex ante regulation. It is easier to do so under ex ante regulation, especially where a regulator may have a duty to promote competition (no such duty exists in competition law).

        From an economics perspective an access obligation is an access obligation and it needs to be well founded, ideally on a standard that is equivalent to indispensability under competition law. Otherwise, you get regulatory arbitrage (or what I loosely called a loophole).


        25 March 2021 at 10:53 pm

    • Thanks, Kay and Felipe for the lively discussion! I made a similar point to Felipe’s in my discussion of Lithuanian Railways:

      On the other hand, the way the Court draws the line (whether intervention would compel a dominant firm to conclude a contract) seems sensible to me. Perhaps not the only sensible approach, but eminently sensible and arguably the most reasonable one all things considered

      Pablo Ibanez Colomo

      26 March 2021 at 10:52 am

      • Thanks for reminding, Pablo. Re-reading your Lithuanian Railways post does become interesting now. In the counter-factual (but for the abuse) scenario, one would logically assume a market with the sectoral ex-ante regulation in effect (i.e. access provided). But there is indeed this “gap”, or perhaps “loophole”, between the objectives of that ex-ante regulation, and its reality in practice. As you said “We would also be accepting that the objectives of competition law may change whenever it overlaps with a sector-specific regime.” This could lead to over-broad discretion for a competition authority to cover or fill gaps/limitations (objectives vs reality) of sectoral ex-ante regulation (i.e. prosecuting domco restrictions to achieving those ex-ante objectives). Makes me think some of these questions may need to be resolved in the upcoming ECJ Facebook v BKA case, and suggests that any violation of the DMA could then also be pursued as a 102 violation (as Felipe alluded to).


        26 March 2021 at 11:20 am

  2. It is interesting to see the Court speak in para. 47 about incentives and long-term v. short-term benefits. Normally, you have this kind of discussion in AG Opinions and not in the text of judgments. So this is very encouraging. It’s like reading para. 57 of AG Jacobs’s Opinion in Bronner.


    25 March 2021 at 3:40 pm

  3. I respecfully (and once again:)))) disagree with two points in your assessment re: Lundbeck and object.

    1. You argue that “the fundamental question in this regard is whether the agreement can be explained on grounds other than the restriction of competition. If there is an alternative rationale for the practice, its treatment as a ‘by object’ infringement would not be warranted.” This is never stated by the court and this is clearly not the “fundamental” test. Rather, the fundamental test (which is even repeated in paras 112, 131 and 139!) is “sufficient degree of harm” in terms of content and objective in the economic and legal context. The same is the test in the operative part of Budapest Bank. No mention of alternative explanations here, which may be one element among many, when trying to establish object (a case in point is Budapest Bank, where pro-competitive explanations were accepted as possibilities, but were clearly not regarded as being decisive for a finding by object, see para 69, 73).

    2. You yourself confirm that Lundbeck’s argument re: counterfactual in object was found to be irrelevant (“This argument failed.”), but still argue for the relevance of the counterfactual. Of course, Lundbeck (just like you!) tried to argue for the relevance of the counterfactual not at the effect (or the exemption…:) stage, but – certainly! – at the object stage. And this is exactly what was rejected by the court: “an examination of the ‘counterfactual scenario’, the purpose of which is to make apparent the effects of a given concerted practice, cannot be required in order to characterise a concerted practice as a ‘restriction by object’.” (para 140). The examples you state (eg insurmountable barriers to entry) are simply parts of the “economic and legal context”, but are surely conceptually different from a “real” counterfactual test as meant by the court, ie where you undertake a fully blow econometric assessment to show what the market would have been without the agreement.

    Would be sooooo great to discuss all these in Florence or elsewhere in person…:)


    25 March 2021 at 10:16 pm

    • Thanks for being such a faithful reader, Asimo! As you know, Chillin’ loves to debate

      As much I enjoy a discussion about the law, I do not believe we disagree all that much

      On your first point: it is indeed the case that the legal question is whether an agreement reveals a ‘sufficient degree of harm’. The practical issue is how we get there. ‘Sufficient degree of harm’ is not, as you know, an operational concept.

      The above is the reason why I focus not only on what the Court says but on what it does. And when I read Generics, Budapest Bank or Lundbeck (just to mention three salient examples), it seems clear to me that the fundamental (albeit not only) question is the one I outline.

      Any disagreement between us remains at the margins, as far as I can tell. Paras 69 and 73 in Budapest Bank are certainly good law!

      On the second point: it looks like we both agree that Lundbeck’s argument about the counterfactual was at odds with the case law. More importantly, it is not what I mean by counterfactual in my work (I even made this very point in a post published last year)

      If you agree that the factors that I label ‘counterfactual’ are relevant in the assessment of the economic and legal context, then the only remaining disagreement is semantic. And if it just about labels, it is not really a disagreement. I would happily drop the counterfactual label if it avoids misunderstandings

      Thanks so much again!

      Pablo Ibanez Colomo

      26 March 2021 at 10:43 am

      • Oh, isn’t “this is the right room for an argument?”…:) (see the old old sketch:

        On to substance, then!

        – Nr 1: you appear to say the key is that if there is a good alternative rationale, then an agreement cannot be a by object infringement. But doesn’t case-law appear to state that this is not THE decisive issue, but merely ONE of the many aspects? Eg, in Budapest Bank, there were several plausible pro-competitive rationales brought forward (balancing of system, stopping upward pressure), but the CJEU still considered the question of object open (see the “unless…” section of the operative part of the judgment).

        – Nr 2: Ok, just a semantic issue, then!


        29 March 2021 at 6:31 pm

      • Good one, Asimo, thanks so much!

        I am not particularly surprised that the Court left the issue open in Budapest Bank. As you know, it is not for the ECJ to apply the law to the facts of the case in the context of a preliminary reference procedure.

        And you also know from that case that the pro-competitive rationale behind an agreement needs to be ‘actually established’ (para 69), which is for the national court to evaluate. In the same vein, Generics noted that the pro-competitive effects of the agrement must be ‘demonstrated and relevant’ (para 105).

        Pablo Ibanez Colomo

        31 March 2021 at 8:35 am

  4. Pablo

    Excellent post as usual. But I think I disagree with two of your three bullet point examples at the conclusion of the Slovak Telekom discussion.

    “Intervention requires, in effect, a firm to remove some features from a product (say, a camera from a smartphone).”

    This would mean that the Court was wrong in Microsoft: the Media Player analysis – or a browser analysis – could lead to this remedy. (Though choice screens are better.)

    More generally, this seems to say that a bundled product can only be unbundled if the indispensability test is met. That seems wrong.

    More generally still, I suspect that the right to contract or not contract principle is more important than the property right principle. An awful lot of antitrust intervention could be construed as interference in property rights.

    (Not strictly on point, but it’s such a good quote that I’ll mention it anyway. The Court of Appeals for the DC Circuit in the US Microsoft case said that the proposition that a firm has ‘an absolute and unfettered right to use its intellectual property as it wishes [is] no more correct than the proposition that use of one’s personal property, such as a baseball bat, cannot give rise to tort liability.’)

    “Intervention requires, in effect, a firm to redesign its product so that rivals have shared access to it (for instance, an interoperability duty).”

    For interoperability, I think this will sometimes be true and sometimes not. Not all interoperability duties will require product redesign or contractual relations. Going back to Microsoft again – and this time the ruling goes against my argument – the interoperability remedy arguably required neither. The documentation of the interoperability information did not – IIRC – require any redesign of the interface, merely documenting it publicly. And the information has been freely available since then without entering into a contract. Samba and Apple, for example, have both implemented the protocols without – AFAIK – any contractual relationship.

    That type of interoperability remedy doesn’t require contractual relations, and doesn’t lead us down the path of at least two of Scalia’s risks of intervention in Trinko (antitrust as price regulation and the remedy as facilitating collusion). There’s still the third risk – to innovation – which is a related but much larger discussion.

    Once you get into looking at actual effects – whether it is innovation incentives or exclusionary effects – the analysis can vary massively if you are talking about rivalrous access to a port, or non-rivalrous access to an API on a market with network effects. I think this point has been under-emphasised in the caselaw; perhaps now that Slovak Telekom seems to clarify that the contractual relations point seems to be a factual threshold question, we can now focus more on effects.

    There are also cases that might be termed “predatory innovation” where a product redesign has no pro-competitive benefits, and causes only costs to rivals (or arguably where the costs outweigh any such benefits). Finding that redesign to be abusive implicitly requires a return to the older design but I don’t think would require indispensability: Racal / Decca is the EU example that’s usually cited. There’s also a US Court of Appeals case – Bard v M3: – which has the same pattern of product redesign serving no pro-competitive purpose. The “remedy” there was only damages, but I’m fairly sure that if the case were transplanted to the EU, a “return to your previous non-abusing product design” remedy could be imposed. (There were no contractual relations between the parties in that case.)

    With apologies for this turning into an essay rather than a blog comment!


    27 March 2021 at 3:15 pm

    • Thanks so much, Kevin, for the thoughtful comments!

      I agree that the Media Player case is the right starting point to discuss these matters. For better or worse, the issue never made it to the ECJ.

      Where do I come from when I make this point (and your thoughts would be most welcome)?

      In a scenario in which the Bronner conditions are applicable, the infringement can be brought to an end in three ways:

      – By asking the firm to deal with third parties.
      – By requiring a structural divestiture of the upstream and downstream activities.
      – By asking the firm to close down one of its divisions.

      Since these three remedies are equivalent, I fail to see why the law would treat them differently.

      If you ask me: the failure of the Media Player remedy proves my point. The fact that the remedy in question failred and the fact that the Commission moved to a duty to deal remedy (which is what the choice screen remedy is in effect) mean that it was a Bronner case all along.

      On the rest: I agree with you. The interoperability point is well spotted and is clear in my view that interoperability does not always demand dealing with a third party. On the predatory innovation cases: absolutely. You may have seen that Lithuanian Railways has been appealed.

      Thanks so much again!

      Pablo Ibanez Colomo

      31 March 2021 at 8:49 am

  5. It seems that the General Court didn’t get it so wrong after all… but maybe someone did at the time and doesn’t want to confess it? 😉

    anonymous lawyer

    2 April 2021 at 10:33 am

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: