On Wednesday 25 October 2023, the General Court of the European Union ruled in Case T-136/19 on the application of the prohibition of abuse of dominance under Article 102 TFEU. The judgment concerns the annulment of a Commission decision condemning Bulgarian gas holding company BEH and subsidiaries for an unauthorised refusal of access. The judgment shows the difficult balance between regulation and competition law, as demonstrated in the attributability question of abuse, the absence of potential competition the bandwidth of regulatory access obligations on competitors, and the role of national obligations as a lightning rod of Article 102 TFEU. Regarding the latter, we elaborate on the exceptional acceptance of the State action defence in this case.
The disputed Commission decision followed a complaint by Overgas, a gas company jointly controlled by Gazprom until 2021. The complaint, and, hence, the initial decision, was directed at BEH and Bulgargaz. The latter, at the time of the infringement, was (i) the main importer of Russian gas into Bulgaria, (ii) the public gas supplier on the Bulgarian market (namely, under public service obligation), and (iii) the holder of an exclusive right of a Romanian gas pipeline which, through its connection to an essential storage hub in Chiren, granted effective control over gas transmission through most distribution networks in Bulgaria.
Within this context, both the holding company (BEH) and subsidiaries for supply (Bulgargaz) and infrastructure (Bulgartransgaz) were sanctioned for a refusal to grant access in violation of Article 102 TFEU. The abuse of BEH (and its two subsidiaries) would consist of:
- Preventing and delaying access to the transmission network in Bulgaria;
- Introducing and enforcing discriminatory access rules for storage capacity, resulting in priority access for Bulgargaz; andHoarding storage capacity, resulting in scarcity for third parties.
Exclusionary effect, potential competition and burden of proof
A first point of interest is the exclusionary behaviour test used by the Commission in its decision. It assumed that the abuse analysis does not require proof of a concrete effect on the market. It is sufficient to show that a behaviour attempts to produce anti-competitive effects or is at least potentially capable of doing so (para 448 of the Commission decision). Where this has been shown, the company would be free to rebut on the basis of an objective justification (See TeliaSonera, Case C-52/09).
Probability as a capstone for the legal test of exclusionary conduct seemed somewhat bon ton after a.o. British Airways, but comes with a number of difficulties, as this judgment shows. First, the Court reiterates that being capable is more than a purely hypothetical possibility, an important distinction that raises the bar for the counterfactual analysis (para 954). How one can distinguish potential from hypothetical in that light, without indeed requiring concrete consequences of exclusionary behaviour to be demonstrated, is not an easy question to answer. In line with Generics (Case C-307/18) and Lundbeck (Case C-591/16), a seemingly similar test requires one to establish that market entry is possible and likely, with potential competition being a real fact (para 281).
Yet, the analogy does not fit entirely, at least not in fact. The bottleneck issue, where one company possesses a hub essential for market entry, is even more stringent in this context (gas supply, storage as well as transmission) than generally is the case in the pharmaceutical industry (the question is also slightly different, as infrastructure and patents obviously represent different challenges for competition). The threshold for qualifying as a potential competitor seems significantly higher in the gas industry, as entry is also more difficult and, thus, potential competition represents a more remote phenomenon. Testament to this is the GC’s characterisation of the difficult entry that exists in gas markets (paras 448-450).
In addition to the burden of proof of exclusionary effects, the behaviour underlying anti-competitive behaviour must also be attributable to the firm, implying that there is a demonstrable causal link between the conduct of the undertaking and the anti-competitive effects on the market. While this attributability requirement often plays a rather marginal role in substantive assessment, it does matter in refusal abuse cases, and even more so where the government is closely involved in the prosecuted conduct.
Firstly, the attributability requirement requires the Commission to ascertain in concrete terms whether there were, in fact, requests for access from (one of) the downstream competitors. At this level, the attributability somewhat increases the burden of proof for the notion of refusal. Secondly, the same question comes into play in the interpretation of the Bulgarian regulatory regime, which imposed, inter alia, storage and capacity requirements on BEH.
What is a refusal?
For there to be a refusal of access prohibited under Article 102 TFEU, there must be a refusal in the first place. In this respect, the GC concludes that the Commission wrongly equated the absence of access with an actual refusal. Thus, a simple letter of complaint from competitor Overgas about restrictions in the transmission network does not count as sufficient evidence of a refusal (para 942), nor did the Commission demonstrate a modus operandi leading to an actual refusal over the entire period (paras 949-950), as access was indeed given within a limited period of time, in line with the applicable access rules at the time (para 946).
The case law of recent years has taught that refusal abuses, whether refusal to deal, refusal to supply or refusal of access, are approached differently and can be subject to a different burden of proof. This judgment shows that such can happen on two levels. The first way, which we saw above, is to stretch what is understood by refusal. A functional approach to this concept, by referring, among other things, to actual conduct rather than assessing compliance with the regulations in force, achieves little in a utility’s context. The GC is quite formalistic in its assessment—for instance by referring to the rules of access and the procedures to be followed for that purpose (paras 708 et seq.)—thus imposing a rather strict burden of proof. This approach should be welcomed: in regulated sectors, where formal rules of access exist, it is desirable to follow the regulatory guidelines as much as possible, and not to extend the application of Article 102 TFEU unnecessarily—at the risk of operating contra legem. Of course, such a more formalistic approach does not account for manifest and obvious abuses or restrictions (Case C-32/11), albeit that, even then, it does not absolve the Commission from considering the economic and regulatory context (Cases C-211/22 and T-680/14, in particular, para 80).
Furthermore, the Bronner criteria are again an element of discussion. As such, the infrastructure or product to which access is sought should be necessary (indispensable) for the exercise of the competitor’s economic activity, in a sense that no viable alternative is available (see case C-241/91 P and C-457/10 P). Second, the refusal must lead, or be likely to lead, to the removal of all competition in the market, and third, it is not eligible for objective justification pleas.
Over the years, the application of Bronner became increasingly unclear, and various sidetracks emerged within refusal abuses (see this paper by Pablo Ibáñez Colomo for a complete overview). For example, such included the division between constructive and outright refusals, with only outright refusals falling under the application of Bronner. For the former, under which we can count margin squeezes, for example (see Servizio Elettrico Nazionale and Others, Case C-377/20), the essential facilities doctrine does not apply. Just recently, for example, the GC and the Court of Justice ruled that the removal of a piece of track, which de facto boycotts market access for a competing railway company, does not follow Bronner’s indispensability analysis, but is assessed as an autonomous form of abuse in itself (see Lietuvos geležinkeliai AB, Cases T-814/17 and C-42/21 P). The main reason is that the destruction of one’s own infrastructure is not the same as a refusal, which is always done with the aim of exclusive use by the incumbent. In that case, it was sufficient to demonstrate a competition infringement sufficiently serious to restrict market entry (Case T-814/17, para 99).
Bronner’s analysis and (non-)application in this case, which is also in line with the latest decisions in Deutsche Telekom (Case C-152/19 P) and Slovak Telecom (Case C-165/19 P), is a welcome clarification. Where access obligations exist based on EU regulation, Bronner finds no application, and the indispensability of infrastructure by downstream competitors does not need to be demonstrated. As in Slovak Telecom, which concerned access to the local loop in telecommunications (para 39), this judgment goes in the same direction. The GC follows this consistently and does make Bronner’s necessity analysis for the case of the pipeline for which there was no regulatory access obligation (para 447). An additional clarification made is the relationship that the dominant undertaking must have in relation to the infrastructure in question. For example, as regards the first Romanian pipeline, which was necessary for the transmission of gas into Bulgaria, there was no ownership but only an exclusive right for Bulgargaz. The GC underlines that ownership is not decisive nor necessary, but control (such as an exclusive right of disposal) is sufficient to apply compulsory access as a remedy (paras 256-262).
Finally, the judgment again emphasizes the need for restraint in refusal abuse cases. This reservation is not new, as the consequences are very far-reaching for a company. Hence, the GC sees compulsory access as “a serious infringement of the dominant undertaking’s freedom of contract and right to property, since an undertaking […] is in principle free to refuse to conclude a contract and to use the infrastructure it has built up for its own purposes” (para 256). Therefore, it should be applauded that this decision does not further fragment the legal test of refusal abuses, thereby avoiding (more) confusion.
Last but not least: state action as a valid defence in Article 102 TFEU?
Truly exceptional about this judgment is the application of the state action defence, a defence that companies can invoke when they are required by national law to engage in anti-competitive behaviour. This defence exists, at least in theory, under both Articles 101 and 102 TFEU. Although we often find the defence, even in forms other than just in case law, the plea is hardly ever accepted in a way that it exempts companies from the application of the competition rules (see e.g., Eric Blomme, ‘State Action as a Defence Against 81 and 82 EC’, (2007) 30 World Competition 243).
A State action defence succeeds, in principle, where the company can show that it was subject to a statutory scheme that deprived it of all autonomy, or that the scheme exerted irresistible pressure (para 572). In either of these two cases, there must be a clear connection between the distortion of competition, in which the company in question is involved, and the content and effect of the statutory scheme.
From the case law, we distinguish three possible scenarios that follow from the acceptance of a state action defence. First, where national law restricts the autonomy of the company, and therefore its autonomy to act in accordance with the competition rules, the competition rules can be disapplied, or at least not directly applied to the company’s conduct. Not only is a causal link required between the arrangement and the infringement but logically also between the arrangement and the loss of autonomy of the company (see e.g., Cases C-359/95 P and C-379/95 P Commission and France v Ladbroke Racing, paras 33-34). This step is predominant in the GC’s reasoning, as it extensively analyses the demand and dependence of Bulgargaz with respect to these capacity obligations (paras 575-577). A second case exists where national law neither exerts irresistible pressure nor suppresses the company’s autonomy. In that case, competition law is fully applicable, and the public authorities can be partly involved in the proceedings, through, inter alia, the application of Article 106 TFEU and the principle of sincere cooperation in Article 4(3) TEU. A third situation arises when there is a certain degree of autonomy and margin of discretion, but there is nevertheless pressure exerted by the government, either in its actions or through existing legal obligations. In this scenario—even where the autonomy of the company exists—a defence can only be accepted where “objective, relevant and consistent evidence that that conduct was unilaterally imposed upon them by the national authorities through the exercise of irresistible pressures, such as, for example, the threat to adopt State measures likely to cause them to sustain substantial losses” (Case T-399/19 Polskie Górnictwo Naftowe i Gazownictwo t. Commission, para 55).
A final controversial issue is whether a state action defence can survive when the pressure comes from a third state. Such a question is even more topical in times of geopolitical tensions—such as supply problems due to high dependence on Russian gas. Here, the GC recently ruled that it “[cannot] be ruled out that state coercion […] may be exercised by a third state” (PGNiG, para 64). Given the exceptional and restrictive nature of the state action defence, it is yet unlikely that such a context could give rise to justification.
While the case law seemingly assumes that state action defences are a clearly developed doctrine, it is still more of a theoretical phenomenon than a reliable part of the analysis of Article 102 TFEU. In the rare cases where the defence is accepted, it is likely to result in a softer approach towards liability, such as fine reductions, instead of withdrawing competition law’s application (see, for example, the impact of state involvement in fine calculations, para 29, last sentence of the Fining Guidelines).
What we know is that the indicators of autonomy and irresistible pressure are central to a successful defence. Within the facts of BEH v. Commission, the Bulgarian government was largely dependent on both the supply and storage of BEH’s gas. Therefore, part of the access refusal was caused by agreements concluded with the Bulgarian government, and the minimum capacity obligations that followed. In doing so, the Court appreciated BEH’s statements and constructive position at the time of these negotiations (paras 543-562). Finally, the geopolitical context, and the strong dependence on Gazprom, undoubtedly play an important role in the Court’s analysis (e.g. paras 567-570). This, and a concrete assessment of the minimum obligations imposed on BEH to perpetuate security of supply, induce that BEH’s conduct cannot be explained otherwise than by reference to the legal framework within which it had to provide its services (para 616).
The judgment is an interesting application of the interplay between industry regulation, on the one hand, and competition law, on the other. Where both approaches have different dynamics and approaches (ex-ante vs ex-post, competition by design, etc.), this saga shows how the two influence each other. For instance, for the Commission, regulatory access and non-discrimination obligations play a role at the level of dominance assessment (see Commission decision para 360), but also as a proxy to demonstrate, even suspect, the exclusionary effects of BEH’s conduct (Commission decision, para 449). The decision shows how the GC sets some boundaries to this proxy effect, and, thus, expects an autonomous assessment of a competition law infringement.
Furthermore, the judgment, unfortunately, leaves open a number of questions about the state action defence. It is important to draw attention to these, especially since the judgment proves that it does have practical relevance. First, it is unclear what place there really is for the state action defence within the legal assessment of Article 102 TFEU. It is well established that autonomy and irresistible pressure are the determining criteria for its outcome, yet the case law appears to give the wrong impression that they are alternative criteria, whereas the facts follow a cumulative, at least interconnected, assessment. Second, at the risk of falling into anecdotalist and inconsistent additions to the Article 102 TFEU acquis, it is desirable to treat the state action defence under the heading of objective justification (weighing the anticompetitive effects against efficiency considerations; Case C-680/20, Unilever Italia Mkt. Operations Srl, para 49). Alternatively, the defence could be considered under the exception provisions of Article 106(1-2) TFEU. However, this solution could be at odds with existing case law, which separates any justification from the role played by the state within the challenged conduct (see e.g. C-437/09).
Inevitably, the complexity of this decision is (partly) due to the regulatory nature of gas industries, which can significantly complicate the interpretation (and co-existence) of competition law. Proof of this can be found in the absence of potential competition, and the bandwidth of regulatory access obligations for competitors, among others. Still, it is safe to say that despite the particular facts and (geo)political context, the judgment offers numerous questions of principle and clarifications, which is why it is compulsory reading for every competition lawyer.