On the Article 102 TFEU Guidelines (I): the three categories of practices make sense, but a fourth is missing
As evidenced by the many conversations it has initiated, the Draft Guidelines on exclusionary abuses issued back in the summer provide a good basis to think systematically about Article 102 TFEU and the evolution of the case law over the past couple of decades.
This post is the first of a weekly series engaging with the points addressed by the Commission in the document and discussing whether, and how, the review exercise can contribute to making enforcement both more effective and more predictable.
The Draft Guidelines are rich and offer a number of angles to take. It seemed to me that the obvious starting point is the tripartite categorisation of conduct, if only because much of the document (and many assumptions underpinning the analysis) revolves around it.
I assume every reader knows that the Commission identifies three broad families of potentially abusive practices: (i) so-called ‘naked restrictions’; (ii) those where the anticompetitive effects are presumed and (iii) those that require a case-by-case assessment of their impact.
The tripartite distinction makes sense and faithfully captures the essence of the case law. From an enforcement standpoint, it is consistent with the ambition of favouring the administrability of Article 102 TFEU.
That some practices are abusive by their very nature (whether we call them ‘naked restrictions’ or abuses by object) cannot be seriously disputed. It is an integral feature of the case law, and a useful one at that: if it did not exist today, the Court would create it.
By definition, conduct that cannot be explained other than as a means to restrict competition falls (and should fall) within the scope of Article 102 TFEU. There is no need for an authority or claimant to show any actual or potential anticompetitive effects in such instances.
My only suggestion about this first category is that predatory pricing within the meaning of AKZO, even though not qualified as such in the Draft, is emphatically a ‘naked restriction’. Arguably, it is the single most prominent example of the category.
It is sufficient to compare and contrast para 71 of AKZO and the definition of ‘naked restriction’ given by the Commission: the former is the direct inspiration of the latter. It would be illogical not to treat it as abusive by its very nature.
It is also difficult to dispute that the anticompetitive effects of some practices are presumed and thus that the creation of a specific category necessarily in relation to these practices is warranted.
Loyalty rebates (that is, rebates conditional upon exclusivity) feature prominently in this category. The Intel judgment of 2017 made a fundamental contribution to the case law in that it clarified that the presumption of anticompetitive effects underpinning Hoffmann-La Roche can be rebutted (it may be rebuttable, but it remains a presumption).
There is also a presumption of anticompetitive effects (and this is another good example in the Guidelines) where ‘margin squeeze’ conduct leads to negative spreads (that is, where the wholesale price the dominant firm charges to its downstream rivals is higher than the retail price it charges to end-users)
I would argue that the fundamental (and, I would add, necessary) contribution that the Draft Guidelines make to the above is that they acknowledge the difference between, respectively, ‘naked restrictions’ and Intel-type scenarios. The latter cannot be treated as ‘naked restrictions’, if only because they can be rationalised on pro-competitive grounds.
It is no less valuable that the Commission clarifies that the bar would be relatively higher if a dominant undertaking were to argue that a ‘naked restriction’ should not be prohibited as abusive in a particular instance. The question of where exactly one should place the bar in relation to these practices is one that I will address in the second post of this series.
The main comment I would make about the tripartite distinction is that it is missing a fourth category, that of presumptively lawful conduct.
The case law is unequivocal about the existence of this family of practices. Acknowledging its existence in the final version of the Guidelines would not only complete the codification exercise but would also contribute to the declared goal of enhancing legal certainty.
It has been clear since Hoffmann-La Roche that the category of presumptively lawful conduct encompasses a rebate that is genuinely conditional on the volume supplied (that is, ‘a simple quantity rebate linked solely to the volume of purchases‘).
The scope of this category would be clarified in Post Danmark II, where the Court held that it comprises rebates that are granted ‘in respect of each individual order‘ and which therefore correspond ‘to the cost savings made by the supplier‘. It is admittedly a narrow range of practices, but a range nonetheless.
Following Post Danmark I, moreover, there should be little doubt that above cost unconditional prices are presumptively lawful. If anything, the Court went further in its judgment, as it suggested that unconditional prices are unlikely to have anticompetitive effects where they would allow to cover ‘the great bulk of the costs attributable to the supply of the goods or services in question‘.
By extension, a practice cannot be qualified as an abusive ‘margin squeeze’, where the margin between the wholesale and the retail prices would not force the downstream division of the dominant firm to sell at a loss (in the sense that the spread between the former and the latter prices give the downstream division a sufficient margin to sell above cost).
The four categories in the case law are summarised in the table below. Next week, I will say a more detailed word about ‘naked restrictions’ (including whether it is more appropriate to call them abuses by object). In the meantime, I very much look forward to your comments.
| Category | Rationale | Examples | Case law |
| Presumptively lawful | Expressions of ‘on the merits’ competition | Prices x>ATC Volume rebates | Post Danmark I Post Danmark II |
| Case-by-case analysis of effects | Ambivalent effects on competition | Standard rebates Tying in tech | Post Danmark II Android |
| Anticompetitive effects presumed | Likely to have negative effects | Loyalty rebates Negative spread | Intel TeliaSonera |
| ‘Naked restrictions‘ | Only explanation is exclusionary | Predatory pricing Sham litigation | AKZO ITT Promedia |
Chillin'Competition 
I agree with your categorisation if one were to be made. However, I am not convinced the ‘presumptions’ framework is useful:
(i) We ‘presume’ things when we don’t have sufficient information to know them. By the time the rubber hits the road, the presumption quickly becomes secondary to the evidence.
(ii) It does not help administrative efficiency. The authority cannot save time putting something in the naked restrictions/presumption box in practice because they still need to prove the conduct qualifies for the box, and this eventually unravels into the full effects analysis paradigm the framework was supposed to avoid. For example, the ICERA predatory pricing case and the Intel case might (if you agree with the EC’s view) fall into the naked restrictions and presumed boxes – these cases have dragged on for over a decade!
(iii) For the same reason, it does not aid legal certainty. Also because the very classification of conduct into boxes inevitably shifts in ways companies cannot predict (no matter how good their advisors are at reading the runes of intentionally obtuse ECJ judgments). Exclusivity rebates any one?
Tom
12 November 2024 at 10:38 am
We do not “‘presume’ things when we don’t have sufficient information to know them”. Most of the presumptions are based on prior experience, and intuitive or logical evidential reasoning. The presumption spares the effort to look for the information, but it is not necessarily premised on its absence.
If a presumption were to be explained by lack of information (information “to know the things” as you put it), it would be a strange presumption, since we would never be able to test if the presumption is good or bad.
The idea that when we start having specific information (when “the rubber hits the road”) we can forget about presumptions (which are often the evidentiary formalisation of certain priors), defies basic principles about how we shape our beliefs. Please just read Kahneman or get familiar with Bayes. We do not examine case-specific evidence as if each case was premised on a cognitive tabula rasa.
Joan
26 November 2024 at 2:24 pm
Dear Pablo, I am not sure that it is useful nor necessary to distinguish four categories of conduct under Article 102. Instead, I would suggest that it is sufficient, as done under Article 101, to simply distinguish between ‘by object’ abusive conduct (indeed including pricing below AVC, as suggested by you) and ‘by effect’ abusive conduct.
Your third category of conduct where the anticompetitive effects are presumed, is based on the rebuttable presumption from the 2017 Intel judgment. However, in that judgment the Court merely clarified the order in which authority and dominant firm need to bring forward evidence to show anti-competitive effects. Later judgments are crystal clear that, once the dominant firm submits serious arguments and facts why its conduct may not have anti-competitive effects, the authority is required to investigate all relevant facts and circumstances, i.e. to do a full effects analysis. As one can expect in practically all cases that, firstly, dominant firms will make such arguments and, secondly, that an authority will not want to draft first an SO based only on the presumption after which it will have to (re)start its investigation once the dominant firm has submitted its ‘rebuttal arguments’, this third category is of no practical relevance. Unless of course, as the Commission seems to suggest in its draft GL, the rebuttable presumption would somehow reduce the standard of proof to show anti-competitive effects for this category of conduct, but for that suggestion there is no basis in the case law.
As to your fourth category of presumptively lawful conduct, I think adding this category is mainly useful in case one really wants to apply two separate steps for finding abuse, i.e. that conduct should be not on the merits and have likely anti-competitive effects. In that case one should indeed be able to describe scenarios/conduct which are not abusive because it concerns competition on the merits, even though the conduct may have anti-competitive effects. As suggested by you, pricing above ATC could fall within that category. This fourth category seems less useful if the two steps are seen as two sides of the same coin and conduct that is likely to have anti-competitive effects is automatically seen as conduct not on the merits. As I have argued for instance in my submission to the consultation of the Commission, cases like Post Denmark II seem to point in that direction, by making it clear that occasionally anti-competitive effects and abuse can also be found for pricing above ATC.
Luc Peeperkorn
14 November 2024 at 5:30 pm
[…] de dos columnas obtenidas de Chillin’ Competition, redactadas por Pablo Ibáñez Colomo: «On the Article 102 TFEU Guidelines (I): the three categories of practices make sense, but a fourth i…» y «On the Article 102 TFEU Guidelines (II): ‘naked restrictions’ (or ‘by object’ […]
CeCo | Borrador de Guía del Artículo 102 (Ibáñez Colomo)
8 January 2025 at 2:47 pm