Relaxing whilst doing Competition Law is not an Oxymoron

As efficient competitors in Case T‑612/17, Google Shopping: the principle and the conflations

with 13 comments

Learn English Words - CONFLATE - Meaning, Vocabulary Lesson with Pictures  and Examples - YouTube

It was inevitable that the Google Shopping judgment would require the General Court to engage with an illustrious principle of the case law: Article 102 TFEU is only concerned with the exclusion of rivals that are as efficient as the dominant firm.

As the Court put it in para 22 of Post Danmark Inot every exclusionary effect is necessarily detrimental to competition […]. Competition on the merits may, by definition, lead to the departure from the market or the marginalisation of competitors that are less efficient and so less attractive to consumers from the point of view of, among other things, price, choice, quality or innovation‘.

The corollary to this principle is that only anticompetitive effects that are ‘attributable‘ to the dominant firm’s conduct can trigger the application of Article 102 TFEU. Post Danmark II made an explicit reference to attributability (para 47). Crucially, this point is acknowledged in para 441 of Google Shopping (‘in order to find that Google had abused its dominant position, the Commission had to demonstrate the – at least potential – effects attributable to the impugned conduct of restricting or eliminating competition‘).

Showing that the effects are attributable to the dominant firm’s behaviour (that is, establishing a causal link between the conduct and its impact) demands, by definition, identifying a counterfactual. There is no way around it (I have come to understand that this idea is controversial in some quarters; it is a topic for another post, but I will definitely address it).

The Google Shopping judgment reveals that the principle tends to be conflated with related matters. Two conflations deserve to be discussed:

  • The principle is sometimes interpreted as meaning that the Commision (or any other authority or claimant) needs to show that specific rivals are as efficient as the dominant firm. I do not believe that interpretation is correct, and I struggle to find support for it in the case law.
  • The principle is occasionally used as synonymous with the ‘as efficient competitor’ test. They are different, and the former should not be reduced to the latter.

The principle in practice: what needs to be proved?

In para 514 of the judgment, the General Court explains that one of the interveners argued that the Commission had not established the anticompetitive effects of the practice as it had ‘failed to show that comparison shopping services competing with Google that had experienced difficulties were as efficient as Google or that they had exerted significant competitive pressure on prices or innovation‘.

The General Court rejects the argument as described above. There seems to be no basis for it in the case law. What cases like Post Danmark I and II demand is that a causal link be established between the practice and the effects. If the effects are not attributable to the dominant firm, but to other factors, then there is no abuse (think by analogy of the ‘failing firm defence’ in merger control).

Put differently: the implementation of the principle demands comparing the conditions of competition with and without the practice. It does not demand, however, establishing the relative relative efficiency of rivals in a reality that has already been ‘contaminated’ by the practice. It would not be possible to establish a causal link in such circumstances.

Suppose that a practice denies rivals a minimum efficient scale. It should not be possible for the dominant firm to then claim that rivals are less efficient and therefore that the practice is not abusive. If such an argument were accepted, then effects that are attributable to the dominant firm’s behaviour would fall outside the scope of Article 102 TFEU. The General Court makes a point along similar lines in para 540.

Instead, the question should rather be whether rivals being denied a minimum efficient scale is attributable to the behaviour of the dominant firm or to other factors. Simply put, the appropriate benchmark should be the world in the absence of the practice.

The principle and the ‘as efficient competitor test’

There is a point in the judgment that is arguably more controversial. In paras 538 and 539, the General Court appears to conflate the principle as described above and the ‘as efficient competitor test’ as used in relation to pricing abuses (such as margin squeezes and rebates).

It is sufficient to read the relevant passages in Post Danmark I (then reiterated in Intel) to realise that the Court of Justice lays down a principle that is broader than the ‘as efficient competitor test’. It is emphasising that exclusion that is not attributable to the dominant firm’s behaviour does not amount to an abuse of a dominant position within the meaning of Article 102 TFUE.

This principle can find many incarnations and can be implemented in a number of ways. The ‘as efficient competitor test’ is just one of them, and one that is particularly apt in relation to price abuses: if it turns out that a ‘margin squeeze’ would not require rivals to sell below cost, any exclusion would not be attributable to the dominant firm, but to the fact that the former are less efficient.

The wording of the judgment, which seemingly conflates principle and test, can be interpreted as meaning that only pricing abuses are concerned with as efficient competitors. According to this interpretation, the principle would not apply to practices such as tying or exclusive dealing. Which takes me to the last point.

Is the principle only relevant for pricing abuses? How could it be so?

According to a current of opinion, the principle laid down in Post Danmark I would indeed only be relevant in relation to pricing abuses. I struggle with this interpretation of the case law, but insofar as it has been debated, it is worth discussing.

There are several reasons why the principle laid down in Post Danmark I is applicable across the board. To begin with, the Court (both in Post Danmark I and Intel) did not confine it to pricing abuses. It was a general pronouncement. What is more, it made an explicit reference to other parameters of competition, namely ‘choice, quality or innovation‘.

The most powerful reason, in any event, is that confining the principle to pricing abuses would lead to outcomes that seem difficult to defend from an intellectual standpoint. Taken to its logical consequences, such an interpretation of Article 102 TFEU would mean that it is necessary to establish a causal link between practice and effects in relation to rebates and mixed bundling, but not in relation to exclusive dealing and tying.

If this interpretation of the case law were accepted, the two sets of practices would be subject to different analytical framework for a purely arbitrary reason (the fact that one set of practices relies on pricing mechanisms) even though they are interchangeable (and have the same object and effect).

I really look forward to your comments on this point (or indeed any of the preceding ones).

Written by Pablo Ibanez Colomo

19 November 2021 at 12:41 pm

Posted in Uncategorized

13 Responses

Subscribe to comments with RSS.

  1. Dear Pablo, I note that you took into account Tim’s comment on your earlier post regarding the efficiencies resulting from size that an access seeker cannot reach as a consequence of the alleged abusive behaviour. At the same time, should we not also highlight the problem created by the GC’s approach for follow damage claims. The GC’s approach includes CSS that ceased operating during the time period covered by the Google shopping decision and are less efficient than Google’s CSS, within the justification of the finding of the breach of article 102 TFUE. On the other hand, dammages will in principle only be granted on the basis of certain counterfactuals, i.e. after making hypotheses about the “but-for” performance of specific competitors, based on their individual capabilities, as well as the overall competitive evolution of an industry. However, in such assessment, the less efficient CSS will likely have ceased operating anyway, and their providers not be entitled to damage. This would lead to the paradox that the scope of the behaviour found to be in breach of competition law, would be broader than the behaviour leading to damages. Obviously, all behaviour found abusive does not entitle to damage. Indeed, a behaviour can be sanctionned only because it is capable of anticompetitive effects, even if it had no effects. But the GC now introduced a new category of abusive conducts that would not entitle to damages: conducts leading to the market exit of less effective competitors. In view of the (be it partial) EU harmonisation of damage claims conditions, would the GC not seek to ensure more consistency between the finding of antitrust abuses and the entitlement to damages?

    Christian Hocepied

    19 November 2021 at 2:46 pm

    • Thanks, Christian

      You are right that Tim understood my post as meaning that the principle entails that an authority would have to show that individual rivals are as efficient, and I explained to him that the two issues are not to be conflated. But Tim is certainly not the only one to make these points (including on the blog!).

      On the divergence between the substantive analysis and damages claims, I am frankly relaxed about it. They are different questions addressing different issues. The mere fact that an infringement may relate to potential anticompetitive effects (as opposed to actual effects), is already an indication that each of the stages seeks to answer very different questions.

      Pablo Ibanez Colomo

      19 November 2021 at 4:59 pm

      • I completely agree that both questions are different. First of all, one is EU law, the other is national law. Second, as you say, potential effect does not entitle to damages. But should there not be some consistency in the underlying methodology used in both frameworks? Can the GC say that potential effects include exclusion of less efficient undertakings, while national courts will in principle not grant damages to companies that would, even without the behaviour in question, have lost market share or disappeared because less efficient?

        Christian Hocepied

        19 November 2021 at 5:20 pm

  2. Pablo, your arguments about lack of causal link if you exclude a less efficient competitor (by whatever means, on the merits or not) makes me wonder if you may be planning to kill shortly some terminally ill patient in a hospital and use as defence that there is no causal link since the patient was dying for sure anyway.
    Jokes aside, many comments on some of you prior posts explain why the distinction between price and no-price cases makes perfect sense. The means used are relevant to define an abuse, and that’s the settled approach in the case law. The causal link and the counterfactual are just the latest variation of the traditional attempts to limit the rigour of competition law for dominant undertakings, after other arguments have failed (these ones in my view as well, but it appears that some may still survive in legal Zombieland). The debate on damages here is just noise.


    26 November 2021 at 7:21 pm

    • Thanks, Joan! Emotive metaphors aside, your comments get to the heart of the matter and are really useful. They invariably move the discussion forward.

      I would not say that demanding a causal link and a counterfactual are part of a grand scheme designed to leave dominant firms off the hook.

      The reality, in my view, is simpler and certainly less flashy. It is all just an inevitable consequence of the case law.

      The Court demands that the effects be “attributable” (the Court’s word, not mine) to the contentious practice for Article 102 TFEU to apply. The General Court, in Google Shopping, does not see things differently, as explained above.

      Establishing that the effects are attributable to the practice cannot mean anything other than showing that there is a causal link flowing from the latter to the former. It is not surprising that the Cambridge Dictionary defines ‘attributable’ as ’caused by’: The French word is ‘imputable’, which gets you to the same place.

      And how can a causal link be established if it is not by looking at the counterfactual? I fail to see how and will elaborate the point in a further post, as promised in this one. Your comments on that poist will be most welcome, as you certainly know.

      Pablo Ibanez Colomo

      26 November 2021 at 8:32 pm

  3. Pablo, a simple text search in the court’s website shows that the word “attributable” has not been used by the Court in the sense you use it (in 102 cases). The “conduct” must be “attributable” to the undertaking, for sure, and in that sense it is used many times (in parental liability or state action cases, for example). As regards “effects” the word is only used once in 60 years in Post Danmark II (but my research is probably not as thorough as yours), a price-cost case (where the ECJ rejects the test you propose), but the full reading of paragraph 47 shows that you are overstating its importance: “Lastly, should the referring court find that there are anticompetitive effects attributable to Post Danmark, it should be recalled that it is nevertheless open to a dominant undertaking to provide justification for behaviour liable to be caught by the prohibition set out in Article 82 EC.” Building a whole theory on this sentence requires a Quixotic dose of courage and imagination. The first part of paragraph 47 does not prescribe a standard, but is the narrative basis to introduce the notion that the undertaking can still provide a justification. The rest of the judgment does not use such expression or concept, let alone in prescriptive terms. The official summary prepared by the Court, typically with the main points of law, does not take over that sentence. Everything else you read in that word is just your own analysis, it is not in the judgment. But maybe the expression is used more prescriptively elsewhere in the case law and I have not seen it. You do not mention any other case.
    In any event, even grammatically, if the competitor is less efficient, the effect of the exclusionary conduct is still “attributable” to the (conduct of) dominant undertaking. Excluding a competitor earlier than would be the (always hypothetical) result of market forces (ie competition on the merits) is still relevant for the law. If the outcome would be identical with or without the exclusionary conduct, one wonders why the dominant undertaking would follow that conduct in the first place. One might say that dominant undertakings do unnecessary things sometimes, but, when one hears that argument in litigation, it normally smells of a poor (desperate) ex post excuse to escape liability.
    By the way, Post Danmark I does not support your view in any way. The quotation says something uncontroversial: being excluded on the basis of competition on the merits is unproblematic, ie the means to exclude matter in law. There is nothing on “attributability” in Post Danmark I (except for the attribution of costs, an entirely different matter). Nothing at all.


    26 November 2021 at 9:58 pm

    • Fascinating insights, thank you!

      The General Court makes repeated references to the need to show ‘effects attributable to the impugned conduct’ in Google Shopping (see inter alia paras 441, 456, 518, 541 and 543). It also deals extensively with causality.

      Is the General Court making stuff up? Did it get it wrong? I do not think so, for the reasons explained above. Is it the case that perhaps, just perhaps, it is applying an uncontroversial and widely accepted aspect of the case law?

      Other than that, I am particularly intrigued by the point on dominant firms’ motivation. Why does the dominant firm follow the conduct in the first place if exclusion would have happened anyway, you ask.

      We have long known (and the incorporation of this body of knowledge was one of the drivers behind the modernisation efforts of the mid-2000s) that most potentially anticompetitive practices can be implemented for pro-competitive reasons.

      Therefore, it would be incorrect to presume that most potentially abusive practices pursue an anticompetitive aim. They may very well be implemented for reasons that have nothing to do with the exclusion of rivals.

      The fact that there is no causal link between the effects and such conduct, far from making it more suspicious, suggests that it has a pro-competitive rationale. So no, the absence of a causal link does not mean that the practice is ‘unnecessary’ (as you seem to suggest); it means that it lacks an exclusionary motivation.

      Pablo Ibanez Colomo

      27 November 2021 at 10:22 am

      • The point in your post was that there is no causal link if the competitor is not efficient, a point you criticise in the judgment. You relied on Post Danmark II. Now you rely on the very judgment you criticise to move the debate elsewhere. Of course, any potential effect must come from the conduct, but the point you were making is a different one: that you must examine the efficiency of any excluded competitor (outside price-cost cases), as if competitor is not (as?) efficient there is no causal link. Again, there is no case law in support of that point.


        28 November 2021 at 1:54 pm

      • I suggest that you read the post again: I make, in fact, the opposite point. I wrote, in unambiguous terms (or so I thought), that the case law does not support the proposition that one must examine the efficiency of any excluded competitor.

        To quote the post above:

        The principle is sometimes interpreted as meaning that the Commision (or any other authority or claimant) needs to show that specific rivals are as efficient as the dominant firm. I do not believe that interpretation is correct, and I struggle to find support for it in the case law

        Pablo Ibanez Colomo

        28 November 2021 at 2:00 pm

  4. I have to support Joan here. I don’t think there is an “as efficient competitor principle” in the case law, universally applicable to Article 102 cases, and it does not help a fruitful discussion to pretend that it is “uncontroversial and widely accepted”. Of course it isn’t. Mostly because it would be make no economic sense; it would be “simple but wrong”. Generally speaking, the efficiency of excluded competitors says nothing about the welfare effects of a practice, and even less about its effects on consumer welfare.
    Causality and the counterfactual seem to me quite unrelated to this discussion.


    29 November 2021 at 12:04 pm

    • Thanks, Chichito for the contribution and thoughts. Always appreciated.

      Just to clarify: what I suggested was ‘uncontroversial and widely accepted’ is the idea that effects must be shown to be ‘attributable’ to the practice for Article 102 TFEU to come into play. The General Court seemed to embrace this idea as self-evident in Google Shopping (and even engaged with causality in an extensive manner). I pointed out that I fully agree with the General Court, which (again) repeatedly referred to attributability in its judgment.

      More controversial is the suggestion (if I got your point right, which I might not) that the (consumer) welfare effects of a practice is somehow relevant to evaluate its legality under Article 102 TFEU. As you certainly know, the Court has been explicit on this point and, bar perhaps Magill and IMS Health (new product condition), it has consistently rejected that the impact of conduct on consumer welfare is the criterion to evaluate its effects on competition.

      Causality and counterfactual are at the heart of this discussion, in my view. Or, put differently: rather than about efficiency (which can be distracting and triggering as a word), this discussion is in reality about causality and the counterfactual. Perhaps if we saw it this way, it would appear that we agree far more than it seems.

      The latter is a point that we can revisit in future posts, on which your comments would be most welcome, just as is true of this one. Thanks again!

      Pablo Ibanez Colomo

      29 November 2021 at 12:26 pm

  5. Hi Pablo, fully agreed that the statement in para. 133 of the Intel ECJ judgment is a general statement that does not apply to pricing practices only. But Google did make the argument that the anticompetitive effects were not attributable to Google’s conduct, but to the emergence of the merchant platforms, which the GC rejected.

    The GC found that the conduct was attributable to Google because the traffic from Google Search declined significantly as a result of Panda and that this traffic from Google represented a significant portion of the CSS traffic.

    The GC does not seem to assess is the bigger picture, i.e. whether CSS would have declined even in the absence of Panda due to the emergence of merchant platforms. The GC did accept in para. 391 that merchant platforms had grown in popularity and had inserted price comparison searches as a “possible explanation” for the decline of CSS, but ultimately rejected the argument because the emergence of these platforms were linked (in GC’s view) to the infringing conduct. More generally, the GC completely ignored merchant platforms in the analysis of anticompetitive effects by simply stating that they are not part of the same market, which seems a bit superficial, as you could have one-way competitive constraints between two neighboring markets, even if they are separate.

    This raises the question of what type of counterfactual analysis the Commission is required to do if the defendant puts forward evidence that its conduct was not capable of having anticompetitive effects due to objective factors such as the emergence of merchant platforms, just like a defendant in a pricing abuse raises the AEC test. This ultimately is a factual question that might also raise quantification issues if there is no concrete evidence. The GC has rejected on the basis that the EC is not required to prove actual effects, as mere potential effects are sufficient (para. 378). It would be interesting to see if the ECJ will have the same view if the judgment is appealed.


    1 December 2021 at 12:27 am

    • Thanks, Nikolaos, for contributing to the discussion with such a rich analysis of this aspect of the judgment.

      Please watch this space for further thoughts on the counterfactual!

      Pablo Ibanez Colomo

      1 December 2021 at 9:02 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 )

Google photo

You are commenting using your Google 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: