Reason and Fiat in the Google Shopping Case
Interpreting and applying Article 102 TFEU – at least in all difficult cases that typically reach the courts – requires a combined reliance on both ‘reason’ and ‘fiat’. It requires reliance on ‘reason’, by which I mean substantive reasoning about how a case should be decided, because no case is exactly like any other, and ‘existing positive law’ hardly ever mechanically determines the outcome of a novel and complex factual matrix. But it also requires reliance on ‘fiat’, which I define as past decisions by courts that carry precedential weight regardless of whether they were substantively correct. No matter how unique and novel the facts, applying Article 102 TFEU to a new case is not and should not be entirely a matter of ‘case-by-case’ analysis unbound by pre-existing law.
This may sound like a truism, and probably it is. Nonetheless, it seems to me that competition law scholarship about recent or pending cases often tends towards either of two extremes: a narrow focus on whether the case can be subsumed under a pre-existing rule or precedent, or a narrow focus on whether the merits of the case make sense from an economic point of view, regardless of what pre-existing rules and precedents have to say. Both extremes, in my view, are mistaken.
In other words, while competition law decision-making cannot and should not function without pre-existing legal rules, standards and principles, at the same time it almost always extends beyond them – thus creating new legal rules by applying old ones. The Opinion of Advocate General (AG) Kokott in the pending Google Shopping case, which is the focus of this contribution, offers a particularly good example.
On 11 January 2024, AG Kokott delivered her Opinion in the Google Shopping appeal case at the Court of Justice. Much has already been written about the General Court’s judgment (e.g. in this special issue), which AG Kokott advises the Court of Justice to uphold. The core legal question may be summarised as whether and under what conditions Google’s favouring of its own comparison shopping service (Google Shopping Service) in the Google Search results display, to the detriment of competing comparison shopping services, constitutes abuse of a dominant position. There is no precedent offering a clear substantive legal test to answer this question. The facts of the case in many ways illustrate the novelties of digital markets and online platforms, to which the rules and standards of competition law from the proverbial stone age will somehow have to be applied.
The Opinion of AG Kokott, as I argue in this contribution, exemplifies how reason and fiat are always integrated in complex competition law cases, and it paves the way for the Court of Justice to translate the Article 102 TFEU case law into the age of digital platforms.
Reason and fiat: a primer
The title of this contribution refers to the 1946 article Reason and Fiat in Case Law by legal theorist Lon Fuller. In this article, Fuller aims to show how judge-made law comprises an inescapable synthesis of reason and fiat. By ‘reason’ he refers to substantive or merits-based arguments about what the law means or how it should apply to a case. By ‘fiat’ Fuller refers to case law-based rules that are legally binding even if they ‘might easily have been otherwise, a fiat intended to fill the space left blank by defaulting reason’.
Legal philosophers have typically resorted to either of the two options. Natural lawyers would have it so that lawfulness in the end can only depend on reason, not fiat. By contrast, so-called legal positivists – notwithstanding great varieties among them – typically insist on separating ‘law as it is’ from ‘law as it ought to be’ and infer that legal validity depends ultimately on social practices of legal officials, irrespective of their moral merits.
For Fuller, however, the antinomy between ‘reason’ and ‘fiat’ is misleading; in reality legal practice cannot function but through an ‘apparently illogical acceptance of both branches of the antinomy’. Using the example of the common law concept of ‘ownership’, Fuller shows that attempts to reduce law to either fiat or reason result in paradoxes:
‘[W]hen the owner of Blackacre is permitted recovery against a trespasser we say that his “ownership” is protected by the suit in trespass; his “ownership” is the thing that “gives rise to” the cause of action. Asked what his “ownership” consists of we respond that it is made up of certain legal rights, including the right to sue trespassers. In brief, he can sue because he is owner; he is owner because he can sue’.
In competition law, similar circularities are easy to find. For example, conduct by a dominant undertaking is ‘abusive’ when it deviates from ‘competition on the merits’. When asked what sort of conduct deviates from ‘competition on the merits’ we may be referred to a list of types of conduct about which the EU courts have held that, insofar as they are capable of having anti-competitive effects, they constitute ‘abuse’. But asked why the EU courts held such conduct to be abusive we say that these types of conduct deviate from ‘competition on the merits’.
Many of the complex issues in EU competition law reveal that competition law almost naturally demands integrating reason and fiat. Google Shopping offers a particularly pertinent example.
Self-preferencing as an independent form of abuse and the Bronner criteria
A substantial part of Google’s appeal to the Court of Justice focuses on the question of whether the stringent Bronner criteria for refusals to deal should apply to Google’s conduct in this case. According to the General Court, the Bronner criteria do not apply to the Google Shopping case because, unlike in Bronner and a range of subsequent cases involving ‘refusals to deal’, there was no explicit request for access and a subsequent refusal by Google (General Court, paras. 232–240). The fact that these criteria are nowhere to be found either in Bronner or in any of the other cases cited (as I pointed out elsewhere) is not worrisome: I take it that the General Court simply aimed to integrate reason and fiat by offering its best constructive interpretation of the indeterminate rule provided by Bronner.
AG Kokott agrees, albeit seemingly for slightly different reasons. According to her, ‘the Bronner criteria should be applied within narrow limits and only to comparable cases of refusal of access or supply’ (para. 81) and there is no reason why ‘unequal treatment through self-preferencing, has to have such strict criteria applied to it in order to be capable of supporting a finding of abuse’ (para. 88). The functionalist crux can be found in the subsequent sentence: ‘[Applying the Bronner criteria in this case] would, moreover, unduly restrict the practical effectiveness of Article 102 TFEU’ (para. 88). In other words, a key reason why AG Kokott finds the situation in Google Shopping more comparable to Article 102(c) TFEU (paras. 75–76) and margin squeezes (para. 95) than to the refusal to deal in Bronner is because she believes that the latter analogy will result in underenforcement of Article 102 TFEU, both in this case and probably more generally too.
While I agree with that assessment, admittedly there is no rule in the case law that mandates this conclusion as a matter of positive law. Reliance on ‘fiat’ alone cannot determine that Google’s self-preferencing is an ‘independent form of abuse’ to which the Bronner criteria do not apply.
Neither does reliance on reason alone: one may be able to show that Google’s self-preferencing is capable of having anti-competitive effects, but it does not follow that the conduct is abusive because such capability is not always sufficient to establish ‘abuse’. Examples include not only refusals to deal but also loyalty rebates, which since Intel escape liability if they do not meet the as-efficient-competitor test, even if they are capable of foreclosing (less efficient) competitors.
So what is the right way for the Court of Justice to decide this case? Functionalism is definitely one side of the coin: the (re-)shaping of existing legal rules and precedents in the process of applying them to a new factual situation is obviously guided by concerns about underenforcement and overenforcement. If applying the Bronner criteria in Google Shopping is believed to make effective enforcement of Article 102 TFEU to digital markets excessively difficult, that is a road better not taken.
The other side of the coin is administrative and judicial discretion. In the end, it is the Court of Justice that decides whether self-preferencing is abusive, and under which circumstances. Even before such judicial settlement, however, the task of authoritatively applying Article 102 TFEU is primarily delegated to the European Commission, as I argued here. The Commission has considerable discretion in determining whether specific types of conduct qualify as ‘abusive’ under Article 102 TFEU, albeit within the boundaries of the legal obligation to show that the conduct deviates from ‘competition on the merits’ and to prove potential anti-competitive effects.
Thus, the Commission has a leading role in devising new theories of law and creating new substantive legal tests. In doing so, it too integrates fiat (binding legal precedents that constrain its law-making discretion) and reason (the Commission’s best assessment of whether enforcement is warranted in this case).
In other words, if the Commission found that Google’s self-preferencing is abusive because it leverages Google’s dominance in general search to the specialised product search services market, and if it demonstrates that this conduct is capable of having anti-competitive effects, for the most part, that is the end of it. The Commission was not legally bound to apply the Bronner criteria, but rather had an important role in deciding whether or not – on the basis of substantive reasons – the Bronner criteria should apply here. Accordingly, the AG affirms that both the Commission and the General Court were allowed – ‘without committing an error of law’ – to characterise Google’s self-preferencing conduct as unreasonable conditions of access by which Google gained a competitive advantage on the specialised product search services market (para. 92).
‘Competition on the merits’ and the ‘special responsibility’
If self-preferencing is not subject to the Bronner criteria, what test governs its assessment under Article 102 TFEU? In part because of the general guidance offered by the Court of Justice in Servizio Elettrico Nazionale, recent attention has focused on how to distinguish between ‘competition on the merits’ and ‘conduct deviating from competition on the merits’ (see e.g. here and here). We know that there is no exhaustive list – case law-based or otherwise – of types of conduct that fall within either category. Since fiat does not provide clear-cut answers, substantive reasoning often needs to supplement pre-existing rules and precedents.
Central to AG Kokott’s Opinion is her explanation, in paragraphs 141–144, that the General Court correctly concluded that the Commission had not made any legal error when it ‘treated the importance of the data traffic from Google’s general search pages and the fact that that traffic is not effectively replaceable as relevant circumstances capable of characterising conduct falling outside the scope of competition on the merits’ (para. 142). This explanation seems to allude to the ‘replicability’ test from Servizio Elettrico Nazionale (para. 78) but does not quite clarify how exactly the requirement that the conduct deviates from competition on the merits constrains the Commission’s decisional discretion.
Indeed, the Opinion keeps the relationship between demonstrating that Google’s conduct is not ‘competition on the merits’ and proving potential anti-competitive effects somewhat obscure. In paras. 142–143, AG Kokott confirms the General Court’s conclusion that ‘[i]f the Commission validly demonstrated the favouring and its effects, it was therefore entitled to assume that that favouring was a departure from competition on the merits’. In this regard, it is particularly important that ‘the impact of data traffic which is diverted from Google’s general search results […] accounts for a large proportion of the traffic to competing comparison shopping services and cannot be effectively replaced’ (para. 141). As I read it, this is essentially an analysis of (potential) anti-competitive effects. But at the same time, AG Kokott concludes that ‘the question as to whether a practice departs from the means of competition on the merits must be distinguished conceptually from that as to whether it is also capable of restricting competition’ (para. 144). Even though they are ‘conceptually’ distinct, in practice these criteria may often be two sides of the same coin (see also here at page 223).
Intuitively, a deviation from competition on the merits may refer to conduct by which a dominant undertaking makes use of its dominant position – instead of the price, quality, etc. of its products or services – to provide itself with a competitive advantage over its competitors. For that reason, certain types of abusive conduct that deviate from competition on the merits are, as we know, perfectly lawful for a non-dominant undertaking; not because these are necessarily expedient or ‘good’ ways to compete but because there is no dominance to make use of in the first place. Tying or loyalty rebates are good examples, self-preferencing is probably another one.
This brings me to the ‘special responsibility’ of dominant undertakings not to further weaken competition. AG Kokott mentions this special responsibility only in passing, in the context of establishing the general criteria for assessing unequal treatment of competitors (para. 77). But the special responsibility may be more important than this single reference suggests. While a non-dominant undertaking may favour its own products or services to compete with others, the same conduct by a dominant undertaking may be abusive if it gives the dominant undertaking a competitive advantage in another market that is not due to the merits of its products or services, but which is merely a consequence of its dominant position. These two situations are perfectly consistent if the starting assumption is that only dominant undertakings have a special responsibility not to further weaken competition.
The ‘special responsibility’, of course, has been often criticised for its lack of clarity and perhaps the tension it could create with certain varieties of the consumer welfare standard. Regardless of the merits of these discussions, the special responsibility is a good example of a fiat rule that is part of the ‘law of the land’ of Article 102 TFEU, regardless of whether one believes it makes sense. Even though it may not always be easy to ascertain what obligations follow from the special responsibility, this rule has normative weight in assessing whether conduct by dominant undertakings is unlawful. Therefore, it cannot be ignored or downplayed in assessing the merits of self-preferencing by a dominant undertaking.
The question remains whether this viewpoint results in over-inclusiveness with respect to self-preferencing. The consequence of AG Kokott’s Opinion may well be that self-preferencing by dominant undertakings presumptively deviates from competition on the merits, insofar as it gives the dominant undertaking a competitive advantage on an adjacent market. This would be unlawful as soon as potential anti-competitive effects can be established.
Interestingly, neither the General Court nor AG Kokott seem particularly concerned about a ‘limiting principle’ that would confine abusive self-preferencing to a more specific substantive legal test. Among the alternative roads not taken are relying on the MEO judgment concerning Article 102(c) TFEU (see e.g. here) (though I am not confident that this would make a substantial difference). But it does not follow that the Commission and the General Court were legally obligated to follow this route, which is the legal question that confronts the Court of Justice in this case. Like Bronner, Article 102(c) TFEU is but one possible analogy to Google’s conduct in this case, and unless there is an imperative legal reason why the Commission ought to have applied it, the EU courts cannot quash the decision for not having followed either analogy.
Concluding remarks: the need for both ‘reason’ and ‘fiat’ in Article 102 TFEU
Google Shopping is only one of several cases in which the Court of Justice will need to decide how to apply general and sometimes vague principles and rules, as well as precedents from another era, to the novel and peculiar characteristics of digital markets. Existing positive law does not provide all the answers, nor does it offer a consensus about how the law should apply to cases like this one. At the same time, no court can decide Google Shopping without relying on fiat: the existing and binding rules of law that were created by past jurisprudence. Reason and fiat are both part and parcel of the law of Article 102 TFEU.
AG Kokott’s Opinion, overall, properly integrates ‘reason’ and ‘fiat’ in Article 102 TFEU; combining the positive law principles of Article 102 TFEU with a reasoned elaboration on how they should translate into this case. Importantly, Kokott recognises, as did the General Court, that the Commission’s enforcement action cannot and should not be confined to applying pre-existing substantive legal tests and underlying theories of harm. In EU competition law, the Commission is a law-maker as much as it is a law-applier.
