Part IV Analytic of the Concept of Dominance, 11 Collective Dominance
From: The Foundations of European Union Competition Law: The Objective and Principles of Article 102
- Collective or joint dominance
Article 102 prohibits any abuse of a dominant position ‘by one or more undertakings’. The concept of collective dominance, first recognized by the Court of First Instance in Italian Flat Glass,1 has given rise to a considerable degree of uncertainty and complexity in the development of the legal tests.2 The reason is that collective dominance has been interpreted as covering tacit oligopolistic collusion3 as well as References(p. 360) other instances in which two or more undertakings are linked in such a way that they adopt the same conduct4 or present themselves as a collective entity5 on the market because of commercial or structural links or direct or indirect contacts between them. However, oligopolistic collective dominance and non-oligopolistic collective dominance are fundamentally different concepts. The problem of whether undertakings in a concentrated market have the ability and the incentive to restrict output and raise prices is different from the question of whether members of an association of undertakings or a consortium have the ability to harm competition byadopting a common policy on the market. Failure to draw this distinction has led to unnecessary complications in what is already a complex area of the law.
After examining the emergence of the concept of collective dominance, this chapter demonstrates that there are two separate concepts of collective dominance governed by different legal tests. This chapter then goes on to identify the principles that define the scope of abusive conduct by undertakings that are collectively dominant. Finally, conclusions are drawn.
Article 102 explicitly provides for the possibility that more than one undertaking can commit an abuse of a dominant position. However, it was only in Italian Flat Glass that the Court of First Instance recognized that a dominant position could be held by two or more undertakings. In that case, the Commission had found that three undertakings had engaged in agreements and concerted practices prohibited under Article 101 and had abused their collective dominant position under Article 102.6 The undertakings concerned controlled more than 80 per cent of the relevant market.7 They systematically sold products to each other so as to allow each undertaking to offer a full range of products. There was evidence that the market was transparent8 and that the product was homogenous.9 The Commission's case under Article 102 was that the position of collective dominance resulted from the agreements and concerted practices which were caught by Article 101.10 Before the Court, the applicants and the United Kingdom argued that Article 102 was References(p. 361) not applicable to the facts of the case as a matter of law.11 The Court disagreed. It emphasized that Article 102 envisaged the possibility of an abuse by ‘one or more undertakings’.12 ‘Undertaking’ in Article 102 must have the same meaning as in Article 101. Therefore, collective dominance could not refer to situations in which the links between different business entities are such that they present themselves as an economic unit on the market, as is the case when separate companies belong to the same corporate group and have no decisional independence.13 Collective dominance refers to circumstances in which two or more separate undertakings are united by such economic links that, by virtue of that fact, together they hold a dominant position vis-à-vis the other operators on the relevant market while remaining separate undertakings.14 The Court added that this could be the case when two or more undertakings have, through agreements or licences, a technological lead affording them the power to behave to an appreciable extent independently of their competitors, their customers, and ultimately the consumers within the meaning of the definition of dominance in Hoffman-La Roche.15 Ultimately, the Court annulled the decision of the Commission under Article 102 for failure to prove that the undertakings in question presented themselves on the market as a collective entity16 but held that a finding of collective dominance could not be ruled out as a matter of law.17
The Court of Justice confirmed the possibility of a collective dominant position in Almelo. In that case, under the relevant Dutch legislation, regional energy distributors were granted a non-exclusive concession for the distribution of energy to local distributors and end-users within a given territory.18 The general conditions for the supply of electric power by a regional distributor to local distributors contained an exclusive purchasing obligation.19 This reproduced a clause contained in the model terms and conditions for the supply of electricity of the Association of Operators of Electricity Undertakings in the Netherlands.20 The Court of Justice held that the clause in question was an agreement between undertakings having a restrictive effect on competition21 and that the same clause could be an abuse of a collective dominant position by the regional electricity distributors in the Netherlands.22 The Court stated that for such a collective dominant position References(p. 362) to exist, the undertakings in the group must be linked in such a way that they adopt the same conduct on the market.23 It was, however, for the national court to consider whether there existed between the regional electricity distributors in the Netherlands links which were sufficiently strong for there to be a collective dominant position in a substantial part of the internal market.24 Almelo may be seen as a case that establishes, as a matter of law, the possibility of a collective dominant position among undertakings operating at the same level of the distribution chain in different markets.25 However, the Court of Justice did not adopt any market definition. It is arguable that the market for the supply of energy to local distributors may have been national. Regional distributors had a non-exclusive concession for the supply of energy in a given territory and, in the absence of the exclusive purchasing obligation, it is plausible to assume that local distributors and end-customers could have purchased energy from other Dutch regional distributors. The market definition issue was not raised before the Court. Therefore, given that the case law is clear that dominance can only be assessed once a relevant market has been properly defined, the issue must have been left to the national court by necessary implication.26
The Court of First Instance further refined the concept of collective dominance in Gencor.27 The Commission had prohibited the integration of Gencor's References(p. 363) and Lonrho's platinum and rhodium businesses because the concentration would have created a collective dominant position between the merger entity and another undertaking (Amplats).28 The merger would have reduced the main suppliers of platinum from four to three and, within two years from the decision, to two (because of the exhaustion of Russian stocks). The merged entity and Amplats would have accounted for 60 to 70 per cent and then 80 per cent of the market, with respective market shares of 30 to 35 per cent and then 40 per cent.29 The Commission took the view that mere oligopolistic interdependence could amount to a collective dominant position.30 A number of market characteristics were held to make it likely that the duopoly resulting from the concentration and the exhaustion of Russian stocks would have had no incentive to compete effectively but, on the contrary, would have had the incentive to reduce output and raise prices.31 The Court of First Instance upheld the Commission's decision holding that the creation or strengthening of a collective dominant position between the merged entity and one or more undertakings could be incompatible with the system of undistorted competition that the 1989 Merger Regulation intended to preserve.32 The Court defined collective dominance as the ability of two or more undertakings acting together, in particular because of factors giving rise to a connection between them, to adopt a common policy on the market and to act, to a considerable extent, independently of their competitors, their customers and, ultimately, of consumers.33 In rejecting the applicant's argument that the Commission had not established the existence of structural links as required by the case law, the Court stated that the case law (mainly, the Italian Flat Glass case) required the existence of economic links, which need not be structural. The concept of economic links includes the relationship of interdependence between the members of a tight oligopoly which, in a market with the appropriate characteristics in particular in terms of market concentration, transparency, and product homogeneity, are able to anticipate one another's behaviour and have therefore a strong incentive to align their conduct on the market, in particular in such a way as to maximize joint profits by restricting production with a view to increasing prices.34 In Compagnie Maritime Belge, the Court of Justice confirmed that agreements or other legal links are not necessary to establish a collective dominant position. Collective dominance may result from ‘other connecting factors and would depend on an economic assessment and, in particular, on an assessment of the structure of the market in question’.35 Thus, the case law in the 1990s gives full effect to the text of Article 102, which prohibits References(p. 364) any abuse of a dominant position by one or more undertakings. However, the early case law on collective dominance conflated two fundamentally different concepts of dominance: the ability to harm competition by two or more undertakings acting as a collective entity by virtue of structural or commercial links or direct or indirect contacts and the ability to harm competition of oligopolistic undertakings tacitly coordinating their behaviour. Instead, these two concepts of collective dominance must be separately examined and are subject to completely different legal tests.
(1) Horizontal non-oligopolistic collective dominance
(a) The two-pronged structure of the test
Horizontal non-oligopolistic collective dominance arises when two or more undertakings act as a collective entity on the same market and have the ability to harm competition because of structural or commercial links or direct or indirect contacts between them. Structural or commercial links or direct or indirect contacts must be sufficient to arrive at the conclusion that two or more undertakings act as a collective entity. Their role must not be limited to facilitating tacit oligopolistic coordination.
The Court of Justice in Compagnie Maritime Belge held that the test for non-oligopolistic collective dominance is two-pronged. The first question is whether two or more undertakings act as a collective entity. The second question is whether they are dominant.36
(b) Collective entity test
The collective entity test requires a thorough analysis of the links between the undertakings in question. Structural links may include interlocking directorships, cross-shareholdings, participation in joint ventures, or participation in trade associations. Commercial links may include cross-licences of relevant technologies or exclusive reciprocal distributorship agreements. Direct contacts include regular exchanges of confidential information relevant to the parameters of competition, such as information on costs, prices, and output. Indirect contacts include the exchange of the same type of information through a third party, for instance in the case of retailers who share information through their common suppliers. Structural or commercial links or direct or indirect contacts need not be unlawful under Article 101. In Compagnie Maritime Belge, the Court of Justice upheld a finding of abuse of collective dominance notwithstanding the fact that the arrangements among the members of a dominant liner conference were block-exempted from the application of Article 101(1).37 However, the links giving rise to non-oligopolistic collective dominance may be also an infringement of Article 101. Therefore, the References(p. 365) same evidence can be relevant under both Articles 101 and 102 but a competition authority or claimant cannot simply ‘recycle’ the evidence substantiating a finding of infringement under Article 101 in order to establish the existence of a collective dominance position. In Italian Flat Glass, the Court firmly rejected such an approach.38 The Court, however, did not declare the evidence that is relevant under Article 101 inadmissible under Article 102. The injunction against the ‘recycling’ of the evidence relates to cases in which a competition authority or claimant relies exclusively on proof of an infringement of Article 101 to establish collective dominance under Article 102.39 This was confirmed in Compagnie Maritime Belge, where the Court of Justice held that the mere fact that undertakings are linked by an agreement, a concerted practice, or a decision of an association of undertakings within the meaning of Article 101 is not ‘a sufficient basis’ for a finding of collective dominance.40 However, explained the Court, the nature and terms of an agreement, concerted practice, or decision of association of undertakings and the way it is implemented may result in the undertakings concerned presenting themselves as a collective entity on the market.41 On the facts, the Court held that a liner conference agreement whereby the participant undertakings coordinated their pricing policy gave rise to a collective dominant position.42
Compagnie Maritime Belge further illustrates what kind of links may give rise to a non-oligopolistic collective dominant position. Associated Central West Africa Lines (Cewal) was a liner conference operating between Zaire and Angola and the ports in the North Sea. It targeted its main competitor by insisting that an agreement concluded with the Zairean authorities and granting Cewal exclusive rights on the shipping routes between Zaire and Northern Europe be strictly complied with, by implementing a system of selective price cuts, and by loyalty contracts and anti-competitive rebates.43 Before the Court of First Instance, the members of the conference disputed that they held a collective dominant position. Liner conferences were, at that time, exempted from the application of Article 101 TFEU. Regulation (EEC) 4056/86 defined liner conferences as follows:44
…a group of two or more vessel-operating carriers which provides international liner services for the carriage of cargo on a particular route or routes within specified geographical limits and which has an agreement or arrangement, whatever its nature, within the framework of which they operate under uniform or common freight rates and any other agreed conditions with respect to the provision of liner services.
In the light of the definition of liner conferences in EU law, three factors were considered relevant to the finding that Cewal was a collective entity. The first was structural: the members of the conference participated in a number of committees References(p. 366) in which they discussed their commercial policy. The second related to the absence of effective competition between the members of the conference: the purpose of the liner conference was ‘to define and apply uniform freight rates and other common conditions of carriage’. The third was behavioural: the allegedly abusive conduct of the conference members demonstrated an ‘intention to adopt together the same conduct on the market in order to react unilaterally to a change, deemed to be a threat, in the competitive situation on the market on which they operate’.45
The intention to adopt the same anti-competitive conduct is, however, not necessary for a finding of non-oligopolistic collective dominance. Structural or commercial links or direct or indirect contacts may suffice. In Atlantic Container Line, the Court of First Instance upheld the Commission's decision that the members of the Trans-Atlantic Conference Agreement were collectively dominant based on four factors: (a) the members applied a common or uniform tariff, which meant that they did not compete on price; (b) the conference agreement contained enforcement mechanisms and penalties to ensure compliance with the conference tariff; (c) the conference had a secretariat which represented the members vis-à-vis third parties; and (d) the conference members agreed and published annual business plans of the conference.46
Structural or commercial links or direct or indirect contacts must eliminate all effective competition among the undertakings concerned. Effective price competition between two or more undertakings is incompatible with a finding that they are a collective entity.47 There is, however, no requirement that the parties of a collective entity must have eliminated all competition between them. Some marginal competition may survive as long as it does not result in effective competition.48 When the elimination of effective price competition is proven, internal non-price competition rebuts a prima facie case that two or more undertakings are a collective entity only if it is of such intensity that it precludes reasonable reliance of the parties on their common price policy.49
Because all effective competition among the undertakings concerned must have been eliminated, the mere presence of structural links not amounting to control50 is not sufficient to establish collective dominance or even to give rise to a presumption of collective dominance. In Airtours, the Court of First Instance clarified that the fact that the same institutional shareholders hold not insignificant References(p. 367) shareholdings (on the facts, between 30 and 40 per cent) in the major players on a concentrated market ‘cannot be regarded as evidence that there is…a tendency to collective dominance in the industry’.51
(c) Dominance test
Once it is established that two or more undertakings act as a collective entity on the market, the dominance test is essentially the same as in single dominance. In Compagnie Maritime Belge, the Court of First Instance relied on the principle that ‘in the absence of exceptional circumstances, extremely large market shares’ are in themselves evidence of a dominant position.52 Cewal's market share exceeded 90 per cent of the market during the relevant period.53 A number of other factors were relevant to Cewal's dominant position: the difference between Cewal's market share and the market share of its main competition, an exclusivity arrangement with the Zairean authorities, the large size of Cewal's network, its capacity and frequencies, and its market experience.54 The Court carried out the assessment of collective dominance in the same way as it would have examined single dominance. The Court of First Instance adopted the same approach in Atlantic Container Line, where the relevant factors sufficient to establish non-oligopolistic collective dominance were market shares, evidence of price discrimination, the limited ability of customers to switch to alternative suppliers, the weak position of competitors, and substantial barriers to entry.55 In response to the argument that market shares had a different weight in the assessment of collective dominance, the Court held that an entity with a market share above 50 per cent is capable of behaving independently ‘whether it is an individual entity or a single entity’.56
The approach to collective dominance adopted in Compagnie Maritime Belge and Atlantic Container Line is justified insofar as it applies to non-oligopolistic collective dominance only. The point can be clearly understood by contrasting Compagnie Maritime Belge and Atlantic Container Line with the analysis of oligopolistic collective dominance in Kali and Salz.57 In that case, the Commission relied on two types of factor to prove that the merged entity and the other leading European undertaking in the market, Société Commerciale des Potasses et de References(p. 368) l'Azote (SCPA), would hold a collective dominant position after the merger. The first type of factor was in the nature of commercial and structural links. Kali and Salz and SCPA had a joint venture in Canada which was expected to start exporting to Europe in the future and was likely to be of key importance to SCPA.58 Furthermore, Kali and Salz and SCPA participated in an export cartel relating to sales outside the European Union. There was evidence concerning the behaviour of another undertaking suggesting that the latter had only started competing with Kali and Salz and SCPA after it left the cartel.59 Finally, Kali und Salz exported to France through SCPA and not through its own distribution network as it did elsewhere in the European Community.60 The second type of factor was purely economic. The Commission found that the market was mature, transparent, and lacking innovation. The product was homogenous and the market shares of Kali and Salz and SCPA had remained stable over four years.61 Furthermore, Kali and Salz and SCPA had been in a cartel in the past. After the termination of the cartel, and notwithstanding overproduction in Germany, where Kali and Salz was mainly active, most of the latter's sales in France were channelled through SCPA.62 The Court of Justice annulled the Commission's decision. It held that, because of errors of assessment that led the Commission to overestimate their significance, the structural and commercial links between Kali and Salz and SCPA were not sufficient to establish a collective dominant position.63 The purely economic factors relied on by the Commission could not ‘be regarded as lending decisive support to the Commission's conclusion'. Of particular relevance was the defective analysis of the degree of competitive pressure that other undertakings were able to exercise on the duopolists. A foreign competitor imported 11 per cent of potash-salt based products in the European Union, while a Spanish undertaking had a 10 per cent market share and a spare capacity of 70 per cent.64 These findings should also be read in conjunction with the principle that a duopoly with market shares of 23 per cent and 37 per cent cannot be found to be collectively dominant based on market shares alone.65 In Kali and Salz, therefore, the Court established the important principle that when commercial or structural links or direct or indirect contacts are insufficient, in themselves, to establish than two or more undertakings act on the market as a single entity, it is necessary to carry out a systematic examination of oligopolist interdependence.
(2) Vertical non-oligopolistic collective dominance
Vertical non-oligopolistic collective dominance may arise when two undertakings at different levels of the supply or distribution chain act as a collective entity. In Irish Sugar, the Commission had found that Irish Sugar and its distributor Sugar Distributors Ltd (SDL) had a collective dominant position on the market for the retail of white granulated sugar in Ireland.66 The abuse consisted in Irish Sugar References(p. 369) and SDL taking action to restrict imports of sugar from France and Northern Ireland.67 SDL on its own had agreed with certain customers to swap Irish Sugar's products for a competitor's products68 and had granted fidelity rebates to customers.69 The rebates were funded by Irish Sugar.70 The Court of First Instance upheld the Commission's finding of vertical collective dominance. The test for non-oligopolistic vertical collective dominance was the same as for non-oligopolistic collective dominance in general: whether structural or commercial links or direct or indirect contacts between two or more undertakings give them the ability to act on the market as a collective entity and, if so, whether such a collective entity was dominant.71 On the facts, it was undisputed that Irish Sugar did not have management control of SDL and that, as a consequence, Irish Sugar and SDL were not a single undertaking even if Irish Sugar held 51 per cent of the shares in SDL's holding company.72 The Court nevertheless ruled that the two undertakings were collectively dominant because of sufficient structural and commercial links and contacts between them that gave them the ability to adopt the same policy on the market. These links were:73
1. Irish Sugar's shareholding in SDL's holding company;
2. The fact that half of the members of the board of SDL's holding company were Irish Sugar's appointees;
3. Irish Sugar's representation on the board of SDL;
4. SDL's exclusive role, subject to availability of supply, as distributor of Irish Sugar's products;
5. Regular sharing of information relating to sales, marketing, and financial matters;
6. Monthly meetings to discuss key aspects of sugar trading; and
7. The fact that Irish Sugar financed all rebates and promotions to customers.
The Court held that the mere fact that Irish Sugar and SDL were in a vertical relationship was not incompatible with a collective dominant position. Had it been otherwise, there would have been a gap in Article 102. It is possible that undertakings in a vertical relationship are not so closely integrated as to constitute a single undertaking but have the ability and the incentive to behave as a single entity on the market. If they also have the ability and the incentive to harm competition, they would be allowed to do so if Article 102 did not apply to them.74
Commentators have puzzled at this concept of vertical collective dominance, stressing that the very premise of collective dominance is that undertakings are active on the same market, that Irish Sugar and SDL were probably a single undertaking, and that Article 101 would probably apply to most of the conduct under review in that case.75 This confusion comes from the failure of the case law to References(p. 370) articulate clearly the distinction between oligopolistic and non-oligopolistic collective dominance. There is no doubt that oligopolistic collective dominance presupposes that the oligopolists are active on the same market. Multi-market contacts may be relevant if, for instance, they facilitate the exchange of information or make retaliation more credible but undertakings that are not active on the same market have no ability or incentive to coordinate their behaviour through repeated market interactions. However, vertical collective dominance is a form of non-oligopolistic collective dominance. There are circumstances in which undertakings in a vertical relationship are closely connected but are not a single undertaking. In Irish Sugar, the upstream undertaking held, albeit indirectly, 51 per cent of the shares in the distributor but did not have management control over the latter. Still, the links between the two undertakings were such that they acted as a collective entity on the market. Nor are the structural and commercial links or direct or indirect contacts that give rise to a vertical collective dominant position necessarily prohibited under Article 101. Article 101 may not apply to structural links at all unless such links are enshrined in an agreement between undertakings. Even when they are, in principle holding a significant shareholding in a company at a different level of the supply chain is unobjectionable. Nor are exchanges of information between a supplier and its distributor unlawful under Article 101. Nevertheless a combination of these factors may align the incentives of vertically-related undertakings in such as way that they act as a collective entity. Furthermore, even if two undertakings in a vertical relationship act as a collective entity, this is still not sufficient to hold that they are dominant. In non-oligopolistic collective dominance, dominance must be separately proved. Finally, even holding a collective dominance position is unobjectionable. Only the abuse of a dominant position is prohibited. It seems, therefore, that while cases of vertical non-oligopolistic collective dominance may well be rare, they cannot be ruled out and Irish Sugar provides a sound legal basis for assessing them, providing for sufficient safeguards against the risk of false convictions and over-deterrence.
Oligopolistic collective dominance arises when two or more undertakings have the ability and incentive to harm competition, provided that such an ability and incentive result, to a decisive extent, from the undertakings' repeated interactions on the market and not from structural or commercial links or direct or indirect contacts between them. Structural or commercial links or direct or indirect contacts may facilitate tacit coordination but must not be sufficient for tacit coordination to be sustainable. If they are, the case is one of non-oligopolistic collective dominance. The difference is an important one. Proof of non-oligopolistic collective dominance does not require an analysis of oligopolistic interaction. Once it has been established that two or more undertakings are a collective entity, the analysis of dominance is the same as for single dominance. On the other hand, if structural References(p. 371) or commercial links or direct or indirect contacts are not sufficient to hold that two or more independent undertakings act as a collective entity on the market, it is necessary to examine whether they have the ability and incentive to harm competition as a result of oligopolistic interaction. If they do, they are also dominant. The members of a collectively dominant oligopoly cannot act as a collective entity and yet not be dominant.
The concept of oligopolistic collective dominance is heavily reliant on the economics of tacit oligopolistic coordination.76 The complexity of the economic theories in this field77 has meant that the EU courts have struggled to articulate a clear test, at least until the Court of Justice judgment in Impala.78References(p. 372) An economic test of tacit coordination would be structured around three elements: the ability of the oligopolists to coordinate, their incentive to do so, and the absence of incentives to deviate from the terms of coordination.
For tacit coordination to arise, firms must have the ability to coordinate their behaviour. The structural conditions for firms to be able to reach a tacit understanding of the terms of coordination are market transparency, a stable and inelastic demand, and an inelastic supply. Because oligopolistic collective dominance is defined as the ability of firms to coordinate their competitive behaviour based on repeated interactions on the market and not, or at least not to a decisive extent, because of structural or commercial links or direct or indirect contacts, the required level of market transparency must be reasonably high. A cartel may work at lower levels of market transparency than tacit coordination because competitors agree on prices or output directly or exchange, directly or indirectly, key information on price and output. In oligopolistic dominance, coordination must be feasible based mainly or exclusively on the observation of the relevant competitive parameters on the market. As regards the characteristics of demand, demand volatility or high elasticity makes tacit coordination unfeasible. If demand is volatile, firms must change their output frequently and would have to reach new terms of coordination each time. Equally, if the elasticity of the demand of the coordinating firms is high because customers are able to switch to other suppliers, coordination is unfeasible for the simple reason that the oligopolists lack sufficient market power so that they are unable to raise prices and restrict output or reduce investments. The same applies if supply is elastic because a contraction of output attracts new entry or leads firms outside the coordinating oligopoly to increase output. In practical terms, this means that three conditions must be met. First, the market share of the oligopolists on the relevant market must be very high and firms outside the coordinating oligopoly must be capacity constrained or supply inferior substitutes. Secondly, there must be no ‘maverick’ firm. A maverick firm is a firm which, being indifferent or almost indifferent between coordination and competitive behaviour, has a history or reputation for output-expansive conduct even in the face of price increases on the part of other market players.79 Thirdly, there must be substantial barriers to entry to the market.
The ability to reach terms of coordination is not sufficient for oligopolistic dominance to arise. Firms must also have the incentive to coordinate their behaviour. This is inherent in the very definition of oligopolistic collective dominance, which provides that the members of a tight oligopoly must be ‘strongly encouraged to align their conduct on the market in such a way as to maximise their joint profits by increasing prices, reducing output, the choice or quality of goods and services,References(p. 373) diminishing innovation or otherwise influencing parameters of competition’.80 For firms to have the incentive to engage in tacit coordination, they must have either a direct interest in each other's profitability through cross-ownership or cross-shareholdings or must have sufficiently symmetrical cost and demand functions. Cross-ownership and cross-shareholdings effectively internalize the externalities of the competitive process. However, in oligopolistic collective dominance, cross-ownership and cross-shareholdings cannot be the decisive factor that gives the undertakings in question the ability and incentive to behave as a single entity on the market. Otherwise, the case would be one of non-oligopolistic collective dominance. The role of structural links, if any, must be limited to facilitating the functioning of tacit coordination. Also relevant is the symmetry in the oligopolists' cost functions. Symmetrical cost functions mean that the terms of coordination, for instance a given level of output or price, will have the same effect for all firms. A misalignment of incentives resulting from different profit levels for different firms at the same industry output makes coordination less likely.81 The same applies to different demand functions facing different firms, which may be reflected in the asymmetry of market shares or product differentiation. Demand prospects as well as actual demand are relevant.
In an oligopoly where firms have the ability and incentive to coordinate, each firm nevertheless has the incentive to lower its price and make a profit at the expense of the other oligopolists. This conduct is likely to be profitable if the other firms maintain their supra-competitive prices. If, however, they also lower their prices and they do so to the extent that the cheating firm's behaviour yields no profits, deviation is not rational. Therefore, for coordination to be sustainable, there must be a sufficiently deterrent punishment mechanism that the oligopolists have the ability and the incentive to impose on a deviating firm. Three conditions must be fulfilled for a punishment mechanism to exist and be a sufficient deterrent: (a) a monitoring condition; (b) a credibility condition; and (c) a deterrence condition. As regards monitoring, the market must be sufficiently transparent and concentrated for any significant deviation to be likely to be detected. These conditions are essentially the same as the concentration and transparency conditions relating to the ability to coordinate. As regards the credibility and the deterrent effect of the punishment, the other oligopolists must have the ability and incentive to punish the cheating firm so as to neutralize any benefits that the latter may derive from the deviation.
The test of oligopolistic collective dominance developed by the case law is in line with the economic theories of tacit coordination. In Impala, the Court of Justice held that for two or more undertakings to be collectively dominant based on oligopolistic interaction, they must have the ability to reach terms of coordination References(p. 374) that negatively affect the parameters of competition, they must have the incentive to do so, and they must have no incentive to deviate from the tacit coordination.82 At paragraph 123, the Court then set out four conditions that are relevant to the test:83
1. Coordination is more likely to arise if the oligopolists can easily arrive at a shared understanding of the terms of coordination and of the focal point of such coordination.
2. Because each oligopolist has an incentive to deviate from the common policy so as to increase its profits by increasing its market share, coordination must be sustainable over time. To this end, the oligopolists must be able to monitor to a sufficient degree whether the terms of coordination are being adhered to. This requires that the market be sufficiently transparent for each undertaking concerned to be aware, sufficiently, precisely and quickly, of the way in which the market conduct of each of the other participants is evolving.
3. There must be a credible deterrent mechanism, which can be used if deviation is detected.
4. The reactions of outsiders, ie actual or potential competitors and customers, should not be able to jeopardize the results expected from the coordination.
The Court in Impala was at pains to clarify that a mechanical approach to the factors that may be conducive to tacit coordination is to be avoided. What matters is the proof of the overall economic mechanism of tacit coordination rather than the proof of each criterion in isolation.84 It is in this perspective that the relationship between the test in Impala and the conditions in paragraph 62 of Airtours must be understood. In Airtours, the Court of First Instance set out three conditions that must be met for collective dominance to be established:85
1. Each member of the dominant oligopoly must have the ability to monitor whether the others are implementing the terms of coordination. This means that the market must be sufficiently transparent for all the members of the oligopoly to become aware, sufficiently, precisely and quickly, of the others' conduct.
2. Tacit coordination must be sustainable over time. There must be an incentive not to depart from the terms of coordination. Retaliation is inherent in this condition. Each member of the oligopoly must be aware that its departure from the terms of coordination to gain market share will trigger identical action by the others. As a consequence, the former would derive no benefit from departing from the terms of coordination.
3. The foreseeable reaction of current and future competitors and customers must not be such as to jeopardize the results expected from the coordination.
References(p. 375) The Court of Justice in Impala said that the conditions set out in paragraph 62 of Airtours are not incompatible with the criteria laid down by the Court of Justice itself in paragraph 123 of Impala.86 Therefore, the significance of the judgment of the Court of Justice is not to overrule Airtours but to clarify that it would be wrong to decide cases simply by verifying whether the three conditions set out in paragraph 62 of Airtours or indeed the four conditions set out in paragraph 123 of Impala are met by looking at each factor in isolation. It is necessary to identify a plausible hypothesis of oligopolistic collective dominance and to prove it on the facts of the individual case. In Impala, the lack of a comprehensive analysis of market transparency ‘having regard to a postulated monitoring mechanism forming part of a plausible theory of tacit coordination’ meant that the Court of First Instance had committed an error of law.87 Such an error consisted in the analysis of market transparency in isolation and not as an element of the assessment of the ability and incentive of the oligopolists to harm competition on the particular facts of the case.
Given the different but compatible formulations of the oligopolistic collective dominance test in Airtours and Impala and the fluidity of structural and behavioural factors88 that are relevant to the analysis, a systematic and integrated assessment of the market is always required before oligopolistic collective dominance may be established. Following the approach mandated by the Court of Justice in Impala, the relevant factors must be examined not in isolation but in relation to the three elements of the legal test: the ability to coordinate, the incentive to coordinate, and the absence of incentives to deviate. In any oligopolistic collective dominance scenario, what ultimately matters is the proof of the overall coordination mechanism rather than the presence of a number of factors that, in theory, are conducive to tacit coordination.
In oligopolistic collective dominance, the case law repeatedly relies on the concept of a ‘tight’ oligopoly.89 It is, therefore, not sufficient that the market has an oligopolistic structure. The oligopoly must be particularly concentrated or, in the rather imprecise language of the case law, ‘tight’. Two factors are relevant to the definition of a ‘tight’ oligopoly: the collective market share of the oligopolists References(p. 376) and the number of oligopolists. There are no clear rules on these two issues. Guidance may be derived from the case law and the decisional practice of the Commission.
The case law and decisional practice suggests that the collective market share of the oligopoly must be significantly above the 40 per cent market share that constitutes a safe harbour for single dominance,90 and even above the 50 per cent market share that may give rise to a presumption of dominance.91 In Airtours, the post-merger market share of the oligopoly would have been 80 per cent.92 In Impala, post-merger, the four major recorded music suppliers would have held market shares ranging between 72 and 93 per cent on most relevant markets.93 In Gencor, the post-merger market share of the duopoly was between 60 and 70 per cent but it would have been ‘in all likelihood’ 80 per cent within two years of the decision because of the exit of a significant competitor.94 In Kali and Salz, the Court of Justice held that the duopolists's post-merger market shares of 23 per cent and 37 per cent could not, because of their size and the way in which they were distributed, ‘point conclusively to the existence of a collective dominant position’.95 However, the Court also said that the presence of competitors with market shares of 11 per cent and 10 per cent was prima facie evidence of effective competitive pressure on the duopoly.96 In the E.ON commitments decision, the three oligopolists had a joint market share between 75 and 67 per cent during the relevant period.97 The Commission merger control practice is to investigate coordinated effects at market shares of well above 70 per cent.98 A higher market share requirement than in single dominance is consistent with the observation that, at market shares towards the lower bounds of single dominance (between 40 and 60 per cent),99 outsiders are more likely to be able to grow and benefit from demand-related efficiencies on the market and exert competitive pressure on the oligopolists.
As far as concentration is concerned, the key observation is that, as the number of oligopolists increases, the number of competitive interactions that References(p. 377) must be taken into account and monitored increases. However, the number of oligopolists is only one of the factors relevant to the ability to coordinate. The more transparent the market, the more undertakings would be able to coordinate because the increased difficulty in monitoring competitive interactions may be compensated by the ease of monitoring the terms of coordination. While the case law and the decisional practice of the Commission suggest that in a purely oligopolistic setting with no structural or commercial links or direct or indirect contacts between undertakings a finding of collective dominance is unlikely if the oligopolists are more than five,100 it is impossible, and would be completely arbitrary as a matter of law, to extrapolate a clear-cut rule from the case law or economic theory.
Finally, stability of market shares makes coordination easier because it provides a historical focal point for future output decisions.101 When the size of an undertaking is a relevant parameter of competition, growth by acquisition must be taken into account when assessing the stability of market shares.102
In order to be able to coordinate, the oligopolists must be able to reach an understanding of the terms of coordination. This requires that the market be sufficiently transparent.103 Transparency, however, is not the same as the availability at all times of costless and perfect information about prices or the other relevant parameters of competition.104 The sufficient degree of transparency must be assessed in relation to the form of coordination and the retaliatory mechanisms that are postulated on the facts of the case. Between the case of markets where prices are formed on public exchanges and output statistics are regularly published105 and markets where prices are negotiated on a confidential basis with each customer,106 there is a considerable grey area.
In Airtours, the Commission had found that the UK market for short-haul foreign package holidays was sufficiently transparent. Tour operators planned capacity in advance. They did so based on previous years' offerings that were References(p. 378) known to the other operators. Changes in capacity would be known to the other oligopolists, because they all used the same hotels and each other's airline companies and because major capacity expansions were public, for instance as they required the long-term lease of additional aircraft.107 The Court of First Instance disagreed. The planning decisions of tour operators were not simply a ‘renewal’ of past offerings but involved a year-by-year assessment of the factors that determine demand for short-haul foreign holidays.108 Furthermore, aggregate capacity was the result of a myriad decisions relating to individual destinations, dates, airports of departure, accommodation, length of stay, and price. In order to forecast the planned capacity of other market players, each oligopolist would have to know and aggregate these decisions taken at the micro level.109 Nor could the oligopolists monitor each other's overall capacity. It was not feasible to obtain accurate information on the capacity of other market players through hotels because of their large number and their preference to let their rooms to at least two tour operators.110 Discussions between the tour operators on seat requirements on each other's airlines took place after capacity had been set and were too limited in scope to result in a sufficient degree of market transparency at the time when capacity decisions were taken.111
In Impala, the question of whether the market was transparent gave rise to some considerable difficulties. The major recorded music suppliers operated a system of published prices to dealers (PPDs) available in the suppliers' catalogues. There were generally more than 100 PPDs but most discs were sold at the top 5 PPDs. The Commission took the view that PPDs could be the focal point of coordination and that net prices were closely aligned with gross prices. However, the major suppliers granted customers campaign discounts, which varied from album to album and from customer to customer. There was evidence that hit charts increased transparency because they allowed the parties to indentify the titles that generated the most sales and that the major suppliers had close relationships with their customers. However, these factors were not such as to overcome the opacity of campaign discounts.112 The Court of First Instance annulled the Commission's decisions on two grounds. First, it found that the Commission had failed to provide adequate reasons for the conclusion that campaign discounts made the market opaque. The Commission should have examined the nature of the campaign discounts, their size, the circumstances in which they were granted, and their impact on price transparency.113 Secondly, the Court held that the Commission's decision was vitiated by a manifest error of assessment. The Court identified the following factors that were conducive to market transparency: (a) the public nature of gross prices; (b) the limited number of reference prices; (c) the limited number of albums to be monitored; (d) the publication of weekly charts; (e) the long-term relationships between suppliers and customers; (f) monitoring of the market; and (g) the small number of References(p. 379) players on the market.114 The only element of opacity was the campaign discounts. However, the Court held that, while the campaign discounts varied by customer, by album, and over time, they could be based on a set of rules that would not be excessively difficult for a ‘market professional’ to understand.115 Transparency must be sufficient to make coordination possible and to give rise to a serious risk that deviation from the terms of coordination would be detected.116 The Commission had, instead, considered that all price components had to be transparent, including the list price and discounts. This was a ‘high level of transparency’.117 On appeal, the Court of Justice held that the Court of First Instance had committed an error of law by misconstruing the criteria for the assessment of market transparency. The Court of Justice defined transparency as a relative concept. There is no level of transparency that is sufficient as such. Transparency must be sufficient in relation to ‘a postulated monitoring mechanism forming part of a plausible theory of tacit coordination’. The reference by the Court of First Instance to a set of rules that might be comprehensible to an industry professional was not sufficient. Those hypothetical rules should have been identified and the degree of market transparency tested against that benchmark.118
The Court of Justice ruling in Impala brings welcome discipline to the analysis of market transparency by requiring that a plausible theory of tacit coordination must be identified on the facts of each individual case. It goes without saying that such plausible theory of tacit coordination need not be consistent with a model published in the economic literature or constructed ad hoc to suit the facts of the case. However, it must be consistent with the evidence, including any economic evidence. Airtours and Impala provide two instructive examples. In Airtours, the theory of coordination related to the setting of capacity. Market transparency in relation to the prices of packaged holidays was, therefore, irrelevant. Transparency had to relate to capacity at the time when the planning decisions were taken. In Impala, the theory of coordination related to prices. However, actual prices were not transparent. A plausible theory of coordination should have explained how the oligopolists could learn each other's actual prices in the absence of a direct exchange of information, given that only list prices were published.
Finally, market transparency also depends on product homogeneity. Differentiated products are valued differently by consumers and, by definition, prices reflect these different valuations. Therefore, coordination is more difficult because it must focus on a number of different prices. This does not mean that any product differentiation automatically rules out tacit coordination. In Impala, discs were a highly differentiated product reflecting the very different and well defined preferences of consumers. However, the pricing structure was relatively standardized with the vast majority of high selling albums been sold at the top five published list prices. This pricing structure was held in principle to be capable of giving rise to References(p. 380) tacit coordination.119 Before the EU courts, the issue of market transparency was highly controversial but it was accepted that, if actual prices had been transparent, they could have been the focal point of tacit coordination.120
Because Article 102 applies to past or present conduct, tacit coordination must have occurred or be occurring on the market. There is no doubt, therefore, that actual price parallelism is relevant evidence of oligopolistic collective dominance. It is, however, not sufficient evidence, as undertakings with similar cost structures on a homogeneous market may well be charging the same prices as a result of effective competition between them. Parallel prices in themselves are not, therefore, evidence of oligopolistic collective dominance. The case law allows undertakings to adapt themselves intelligently to market conditions, including by taking into account the current or expected conduct of their competitors.121 In Hoffmann-La Roche, the Court distinguished dominance from ‘parallel courses of conduct which are peculiar to oligopolies in that in an oligopoly the courses of conduct interact’.122 This broad definition of oligopolies that are outside the concept of dominance under Article 102 is no longer an accurate statement of the law but does indicate that not all oligopolies are collectively dominant. The key is in the Court of Justice's definition of collective dominance in Impala as oligopolistic interaction that increases prices and reduces output or diminishes innovation.123 The emphasis on negative effects on price, output, or investment is important. In the case of Bertrand competition with homogeneous products, the oligopolists still interact with each other and make their output decisions based on the expected decisions of the other market players. The result is, however, a price that equals MC.124 This form of oligopolistic interaction does not fall under the definition of oligopolistic collective dominance.
If mere price parallelism is, in itself, not sufficient evidence of oligopolistic collective dominance, price parallelism carries more weight if corroborated by other evidence. For instance, if prices increase in parallel in a way that is not related to either cost or demand, this may be evidence of tacit coordination. Another example may be a situation in which there are significant and temporary price reductions that are neither related to costs nor to demand in an environment where prices have been stable or increasing for a significant period of time before and after the price reductions. This structural break in a stable price environment may indicate a deviation from the terms of coordination that triggered an industry-wide output expansion as a retaliatory measure inducing the parties to revert to coordinated conduct. This analysis is relevant evidence of oligopolistic collective References(p. 381) dominance, although it will not be sufficient evidence thereof unless there is no other plausible explanation for the structural break.125
Tacit coordination is not feasible if the demand faced by the oligopolists is elastic. A price increase would lead consumers to look for substitutes, making the price increase unprofitable. In Airtours, the reaction of consumers who could switch to smaller tour operators or to different holiday packages outside the relevant market was deemed capable of counteracting an oligopolistic restriction of output.126 Airtours also clarifies that elasticity of demand is not the same as countervailing buyer power.127 A fragmented demand can still be sufficiently elastic to deny the oligopolists the ability to raise prices and restrict output to the detriment of long-term social welfare. A fortiori, the presence of a few large customers capable of exercising countervailing buyer power is incompatible with oligopolistic collective dominance. In SNECMA/TI, a merger in the landing gear sector, the Commission held that oligopolistic collective dominance was unlikely, given ‘the considerable buying power of the major customers’.128 In Sony/BMG, the Commission found that the concentrated supply of the recorded digital music market with only four major suppliers would be counterbalanced by the concentrated structure of demand where iTunes and the other large digital music service providers had sufficient buyer power to ‘react to potential coordination among the majors’.129 A volatile demand makes coordination more difficult because price and output must change frequently in response to unpredictable changes in demand.130 For the same reason, slow growth or stability in demand facilitates tacit coordination.131 The Court of Justice in Kali and Salz said that a falling market is also generally considered to promote competition rather than coordination.132
As with single dominance, for oligopolistic collective dominance to arise, actual and potential competitors must be unable to exercise an effective competitive constraint on the oligopolists. Substantial barriers to entry and expansion are a necessary condition of tacit oligopolistic coordination. In Airtours, smaller tour References(p. 382) operators in the United Kingdom were held to be able to respond to a restriction of output by the oligopolists by increasing their capacity.133 In Kali and Salz, competitors with market shares of 11 per cent and 10 per cent, the latter with considerable excess capacity, were held to be prima facie capable of exercising effective competitive pressure on a duopoly with a joint market share of 60 per cent.134 In Syniverse/BSG, a duopoly on the market for GSM roaming data clearing services to mobile network operators was held to be unable to coordinate because the likely entry of competitors would make coordination unprofitable.135 Furthermore, customers would have the ability and incentive to sponsor new entry.136 In Gencor, the Court of First Instance held that the slow growth of the market would not have encouraged new entry or aggressive competition by existing competitors.137 On the contrary, a fast growing market is conducive to entry because post-entry output is expected to be larger than in a stable market where the entrant must capture a share of the existing demand.
Markets characterized by a high rate of innovation are incompatible with tacit coordination. In Kali and Salz, the Court of Justice quoted without criticism the Commission's finding that the potash market was characterized by a lack of technological innovation.138 Tacit coordination is generally incompatible with a dynamic and unstable market. In Syniverse/BSG, the Commission found that ‘significant modifications to the industry’ were expected ‘as a result of ongoing technological change’. This would ‘reduce the ability of market participants to reach and sustain’ tacit coordination.139 In addition, technological changes could trigger new entry from undertakings already operating neighbouring technologies.140
For tacit coordination to arise, the oligopolists' incentives must be sufficiently aligned. This means that the parameters of supply and demand of each oligopolist must be similar. Undertakings with significantly different demand or cost structures would not have the incentive to coordinate because their respective profit-maximizing prices and output would be significantly different. In Gencor, the Court of First Instance considered two parameters: the allocation of market shares, including the shares of reserves, and the cost structures of the duopolists. Given the similarity in these parameters, the Court concluded that the dupolists' interests ‘would have coincided to a higher degree and that this alignment of References(p. 383) interests would have increased the likelihood of anti-competitive parallel behaviour, for example restrictions of output’.141
Certain factors that are relevant to the ability to coordinate are also relevant to the incentive to do so. A fast-growing demand might make coordination more difficult but, in addition, enhances the oligopolists' incentives to coordinate because the future gains from coordination are higher. Differentiated products make coordination more difficult and, at the same time, misalign the incentives of the oligopolists as each of them faces a differently sloped demand.
For coordination to be sustainable, the oligopolists must have no incentive to deviate. For instance, a large and powerful buyer can induce an oligopolist to deviate by offering to buy at a lower price. Such an inducement would be strong because the buyer's requirements are large and may destabilize oligopolistic coordination.142
The condition of the absence of incentives to deviate requires that a credible deterrent mechanism be available.143 A credible deterrent mechanism means that the oligopolists must have the ability and incentive to punish the deviator. The ability to punish the deviator depends on whether the oligopolists can obtain information about each other's conduct ‘sufficiently precisely and quickly’.144 Market transparency is a necessary condition. Following the judgment of the Court of Justice in Impala, it is now clear that it is not sufficient to prove that the market is transparent in the abstract but transparency must be assessed in relation to a plausible monitoring mechanism.145 Generally, the degree of transparency necessary for the oligopolists to have the ability to coordinate will be the same as the degree of transparency necessary for the oligopolists to have the ability to detect deviations. A factor that makes detection of deviations more difficult is demand volatility. If demand is volatile, it is difficult for the oligopolists to distinguish deviations from capacity adjustment responding to changes in demand.146 Airtours appears to establish a presumption, based on economic theory, that volatile demand is incompatible with oligopolistic collective dominance, unless the contrary is proved on the facts of an individual case.147
As well as being able to detect deviations, the oligopolists must have the ability to punish the deviator. The most obvious form of punishment is an increase in output that makes the deviation unprofitable. This form of retaliation has been an element of the definition of oligopolistic collective dominance since the early case law. In Gencor, the Court of First Instance said that in oligopolistic collective dominance each undertaking ‘is aware that highly competitive action on its part designed to increase its market share…would provoke identical actions by others, so that it would derive no benefit from its initiative’.148 The oligopolists must be able to increase output to the extent that it makes deviation unprofitable. This References(p. 384) would not be possible if, for instance, they are capacity constrained149 or they sell a differentiated product. Equally, if the market is characterized by few large buyers and lumpy orders or long-term contracts, timely retaliation would be difficult because of the need to wait for the next bidding round or to renegotiate existing contracts. The time lag between the deviation and the punishment may result in the punishment being interpreted as competitive behaviour.
Output expansion is not the only available retaliatory mechanism. In Impala, a potential retaliatory mechanism was the exclusion of the deviator from compilations. The major recorded music suppliers derived a significant part of their revenues from compilations but compilations were generally only attractive to consumers if they included artists from different suppliers. Therefore, suppliers licensed each other their repertoire or entered into compilation joint ventures. An oligopolist excluded from compilations would suffer a significant reduction of revenue that could have offset the benefits resulting from the deviation.150 Punishment need not take place in the same market where the deviation occurs. In Gencor, the Court of First Instance held that multi-market contacts multiplied the risks of retaliation as punishment could also take place in other markets.151 Multi-market contacts can take place between the same legal entities or between companies belonging to the same group.
The mere existence of a credible deterrent mechanism is sufficient for a finding of oligopolistic dominance. The retaliatory mechanism need not have been used in practice. Proof that the postulated retaliatory mechanism has never been used can rebut a prima facie case of oligopolistic collective dominance if two cumulative conditions are met: (a) it is established that one oligopolist deviated from the terms of coordination; and (b) it is established that the postulated retaliatory mechanism was not implemented.152
For a retaliatory mechanism to be credible and, thus, sufficiently deterrent, the oligopolists must have the incentive to implement it. As with all business decisions, this means that the benefits of carrying out the retaliation must outweigh its costs. In Airtours, one of the potential retaliatory mechanisms was directional selling at the distribution level. Tour operators were vertically integrated. As well as selling their own products, they also sold each others' package holidays, earning a commission. Retaliation could be in the form of conduct aimed at lowering the sales of the deviator's products. However, the Court found that ‘such retaliation would entail economic loss for its perpetrators’. As a result, this form of retaliation was not credible.153 The incentive to implement lower prices for retaliatory purposes is also reduced when customers are few and powerful and make lumpy orders or negotiate long-term contracts. In these circumstances, it is less likely that the oligopolists will be able to revert to tacit coordination and higher prices.154
References(p. 385) E. Abuse of collective dominance
Article 102 prohibits the abuse of a collective dominant position. In Irish Sugar the Court of First Instance held that ‘the abuse does not necessarily have to be the action of all the undertakings in question. It only has to be capable of being identified as one of the manifestations of such a joint dominant position being held’.155 This means that the conduct of one or more undertakings holding a collective dominant position can be abusive only if the undertakings in question would not have the incentive to engage in the behaviour under review but for the collective dominant position. The language of the Court in Irish Sugar leaves room for little doubt in this respect. Furthermore, any other interpretation would give rise to a disproportionate risk that normal competitive behaviour by an oligopolist could be characterized as an abuse of a collective dominant position.
Joint abuses of a collective dominant position occur when some or all the collectively dominant undertakings engage in abusive conduct in concert. Generally, this form of abuse occurs in the case of non-oligopolistic collective dominance when undertakings communicate directly with each other or have established stable structural links. In Compagnie Maritime Belge, the members of a dominant liner conference engaged in exclusionary conduct jointly and under the coordination and supervision of the conference committees.156 Atlantic Container Line shows that joint abuses may also consist in practices aimed at limiting or eliminating competition between the members of a collective entity157 or in the provision of incentives by the collective entity to potential competitors to enter the market only as members of the collective entity.158 This latter form of abuse is problematic as it could be interpreted to include all activities by a dominant collective entity aimed at increasing its membership. However, Atlantic Container Line must be read in conjunction with the Irish Sugar principle that the abuse must be a manifestation of the collective dominant position. Activities aimed at increasing the membership of a collective entity can only be abusive if the collective entity would not have any incentive to carry out the activities in question but for its collective dominant position. This could be the case, as was alleged, but ultimately not proved, in Atlantic Container Line, if the members of the collective entity were prepared to forgo significant revenues in order to induce a competitor to join the collective entity only because of the market power rents that they would have lost if the competitor had become an effective competitive constraint.159References(p. 386) A single abuse of a collective dominant position was found in Irish Sugar, where the jointly dominant supplier and distributor of white granulated sugar in Ireland engaged in a number of abusive practices individually.160 Even when the parties carried out the same type of practice on the same market, there was no requirement that the abuse had to be jointly planned or agreed.161 Furthermore, certain practices were carried out by the distributor alone and still found abusive.162 In E.ON, the Commission was concerned that E.ON, one of the three collective dominant undertakings on the German wholesale electricity market, could have deterred competitors' investment in the generation market by entering into long-term supply contracts and offering competitors a participation in an E.ON generation plant. In this way, E.ON could have been able to maintain high prices in the long-term.163 There was no suggestion in the decision that the other two oligopolists could have engaged in similar practices. However, it appears that E.ON's alleged abuse was a manifestation of the collective dominant position because it would have been unprofitable if there had been effective competition instead of tacit coordination among the three main German electricity suppliers.
A potential abuse of oligopolistic collective dominance is the adoption of facilitating practices. Certain practices may facilitate an oligopolistic collective dominant position while not being sufficient on their own to give rise to it. For instance, any behaviour that enhances market transparency is clearly capable of making it easier to reach an understanding on the terms of coordination and to monitor deviations. An example would be the widespread use of most favoured nation clauses, which oblige the supplier to grant a customer the best terms it grants to any other customers. Most favoured nation clauses make deviation less profitable because the deviating firm must lower the price to all its customers and not only to the additional customers targeted by the price decrease.164
Finally, collectively dominant undertakings may commit exploitative as well as exclusionary abuses. In E.ON, the Commission took the view that E.ON could have abused its collectively dominant position by withholding available generation capacity that would have been profitable to sell in order to raise overall prices. The price of electricity is determined by the price of the short-term MC of the marginal plant. The marginal plant is also the most expensive to run as electricity suppliers use the cheaper plants first (for instance, hydroelectric and nuclear plants) before using the more expensive gas and oil plants. E.ON had substantial capacity in cheaper energy. Withholding capacity resulted in substantial profits on the energy generated by the more efficient plants. This had severe effects for consumers and long-term effects on the price of energy because of the impact of short-term prices References(p. 387) on the forward markets.165 The commitments decision does not discuss the position of the other two oligopolists. The practice could have been more profitable for E.ON than for the others if E.ON had a larger share of more efficient energy on which it would earn a higher margin. However, the profitability of E.ON's conduct could have been dependent on the other oligopolists not increasing capacity. If this was actually the case, the test that the abuse must be a manifestation of the collective dominant position would have been met.
Collective dominance arises when two or more undertakings acting together have the ability to harm competition. So far, collective dominance has been analysed as a unitary concept. This has hindered the development of robust legal tests. A closer analysis of the case law and its economic foundations shows that there are two distinct types of collective dominance that are governed by different legal tests.
Non-oligopolistic collective dominance arises when structural or commercial links or direct or indirect contacts are sufficient for two or more undertakings to act on the market as a collective entity. The analysis of the links or contacts that justify a finding that two or more undertakings are a collective entity is, however, only the first step in the test. The second step is to apply the test of dominance based on the same analytical structure that applies to single dominance. The same principles apply to undertakings in a vertical relationship.
Oligopolistic collective dominance is inherently different from non-oligopolistic collective dominance. Oligopolistic collective dominance arises when undertakings in a tight oligopoly have the ability and incentive to coordinate their behaviour by raising prices, restricting output, or reducing investment through repeated market interactions. Structural or commercial links or direct or indirect contacts may contribute to oligopolistic collective dominance but must not be sufficient. Otherwise, the non-oligopolistic collective dominance test applies. The oligopolistic collective dominance test in the case law provides that for two or more undertakings to be collectively dominant based on oligopolistic interaction, they must have the ability to coordinate, they must have the incentive to do so, and they must have no incentive to deviate. This test is consistent with economic theory and ensures that Article 102 can apply to address harm to long-term social welfare arising from anti-competitive conduct by the members of a collusive oligopoly.
Abuse of a collective dominant position can take various forms. It may be a joint abuse or an individual abuse. It may be exclusionary or exploitative. Or it may consist in facilitating practices that appreciably contribute to the maintenance of the collective dominant position. The guiding principle must always be that conduct is only capable of being a manifestation of the collective dominant position if, in the absence of such a position, the undertakings or undertaking in question would not References(p. 388) have the incentive to carry out the conduct under review. There must, therefore, be a causal link between dominance and abuse.
Overall, the law on collective dominance is consistent with the objective of Article 102 and strikes the right balance between the risk of false convictions and over-deterrence and the need to ensure the effectiveness of Article 102 in addressing competitive harm caused by two or more undertakings which are jointly dominant. The law should, however, be clearer in distinguishing between non-oligopolistic and oligopolistic collective dominance as two different forms of dominance subject to different legal tests and in requiring a causal link between dominance and abuse, especially in cases of abuse of a collective dominant position by one undertaking only.
2. The concept of collective dominance has generated a vast literature: BJ Rodgers, ‘Oligopolistic Market Failure: Collective Dominance Versus Complex Monopoly’ (1995) 16 ECLR 21; T Soames, ‘An Analysis of the Principles of Concerted Practice and Collective Dominance: a Distinction without a Difference’ (1996) 17 ECLR 24; EJ Morgan, ‘The Treatment of Oligopoly under the European Merger Control Regulation’ (1996) 41 Antitrust Bull 203; JS Venit, ‘Two Steps Forward and No Steps Back: Economic Analysis and Oligopolistic Dominance after Kali & Salz’ (1998) 35 CML Rev 1101; SB Bishop, ‘Power and Responsibility: The ECJ's Kali-Salz Judgment’ (1999) 20 ECLR 37; E Kloosterhuis, ‘Joint Dominance and the Interaction Between Firms’ (2000) 21 ECLR 79; C Gordon and R Richardson, ‘Collective Dominance: The Third Way?’ (2001) 22 ECLR 416; M Jephcott and C Withers, ‘Where to Go Now for EC Oligopoly Control?’ (2001) 22 ECLR 295; G Niels, ‘Collective Dominance: More than Just Oligopolistic Interdependence’ (2001) 22 ECLR 168; O Black, ‘Collusion and Co-ordination in EC Merger Control’ (2003) 24 ECLR 408; GJ Werden, ‘Economic Evidence on the Existence of Collusion: Reconciling Antitrust Law with Oligopoly Theory’ (2004) 71 Antitrust LJ 719; O Black, ‘Communication, Concerted Practices and the Oligopoly Problem’ (2005) 1 European Competition Journal 341; R O'Donoghue and AJ Padilla, The Law and Economics of Article 82 EC (Oxford, Hart Publishing 2006) 137–165; D Parker, ‘A Screening Device for Tacit Collusion Concerns’ (2006) 27 ECLR 424; G Monti, EC Competition Law (Cambridge, CUP 2007) 334–338 and 341–342; CE Mosso et al, ‘Article 82’ in J Faull and A Nikpay (eds), The EC Law of Competition (2nd edn Oxford, OUP 2007) 313, 338–344; v Korah, An Introductory Guide to EC Competition Law and Practice (9th edn Oxford, Hart Publishing 2007) 124–127; MB Coate, ‘Alive and Kicking: Collusion Theories in Merger Analysis at the Federal Trade Commission’ (2008) 4 Competition Policy International 145; P Roth and v Rose (eds), Bellamy & Child European Community Law of Competition (6th edn Oxford, OUP 2008) 939–946; R Whish, Competition Law (6th edn Oxford, OUP 2008) 556–567; J Goyder, Goyder's EC Competition Law (5th edn Oxford, OUP 2009) 376–382; FE Mezzanotte, ‘Tacit Collusion as Economic Links in Article 82 EC Revisited’ (2009) 30 ECLR 137; M Filippelli, ‘Collective Dominance in the Italian Mobile Telecommunications Market’ (2010) 31 ECLR 81; FE Mezzanotte, ‘Using Abuse of Collective Dominance in Article 102 TFEU to Fight Tacit Collusion: The Problem of Proof and Inferential Error’ (2010) 33 World Competition 77.
3. Economic theory moved from the uniform understanding of oligopoly, to be found in the classic works by EH Chamberlain, The Theory of Monopolistic Competition (Oxford, OUP, 1966); JS Bain, Industrial Organisation (2nd edn New York, John Wiley and Sons 1968); and W Fellner, Competition Among the Few (London, Frank Cass & Co 1965), to recognizing a distinction between coordinating and non-coordinating oligopolies. The modern approach starts with the seminal study by GJ Stigler, ‘A Theory of Oligopoly’ (1964) 72 Journal of Political Economy 44, and is now mainstream: see Werden, ‘Economic Evidence on the Existence of Collusion’, 719.
14. Joined Cases T-68/89, T-77/89, and T-78/89 Italian Flat Glass, para 358; Case T-193/02 Laurent Piau v Commission  ECR II-209, para 110; Joined Cases C-395/96 P and C-396/96 P Compagnie Maritime Belge, para 36.
22. ibid paras 40–45. The Court cited Case 30/87 Corinne Bodson v SA Pompes Funèbres des Régions Libérées  ECR 2479. Bodson, however, does not explicitly address the question of a collective dominant position. It is true that the Court of Justice consistently refers to a ‘group of undertakings’ in the part of the judgment dealing with Art 102 but this seems to be on the assumption that the group in question is a single undertaking for the purpose of Arts 101 and 102. The Court had already ruled that if the undertakings in question constitute an economic unit, any agreements between them cannot fall under Art 101. The part of the judgment discussing the issue of dominance opens with the following statement: ‘Any anti-competitive behaviour on the part of a group of undertakings holding concessions which constitute an economic unit as defined in the case-law of the Court must be considered in the light of Article  of the Treaty.’ It seems, therefore, that the Court of Justice in Almelo incorrectly cited the Bodson case as an authority on collective dominance.
23. Case C-393/92 Almelo, para 42. The definition of collective dominance in Almelo clearly implies that there must be no competition between the collectively dominant undertakings. See Case C-96/94 Centro Servizi Spediporto Srl v Spedizioni Marittima del Golfo Srl  ECR I-2883, paras 33–34 (finding that national legislation providing that a committee with a minority representation of road haulage contractors had the power to recommend road haulage tariffs to be approved and made mandatory by the relevant Minister did not confer on the undertakings concerned a collective dominant position because it did not eliminate competition between them) and Joined Cases C-140/94 P to C-142/94 P DIP SpA v Comune di Bassano del Grappa  I-3257, paras 26–27 (finding that national legislation requiring a licence to be obtained before opening a shop in a municipality by a procedure envisaging the advisory opinion of a committee in which established traders were represented, albeit in the minority, did not confer on the established traders a collective dominant position because it did not eliminate competition between them).
26. Case 6/72 Europemballage Corp and Continental Can Co Inc v Commission  ECR 215 (Continental Can) para 32; Case 27/76 United Brands Co and United Brands Continentaal BV v Commission  ECR 207, para 10; Case 85/76 Hoffmann-La Roche & Co AG v Commission  ECR 461, para 21.
27. Case T-102/96 Gencor Ltd v Commission  ECR II-753; GP Elliott, ‘The Gencor Judgment: Collective Dominance, Remedies and Extraterritoriality under the Merger Regulation’ (1999) 24 EL Rev 638; v Korah, ‘Gencor v Commission: Collective Dominance’ (1999) 20 ECLR 337. Gencor was a case under Council Regulation (EEC) 4064/89 of 21 December 1989 on the control of concentrations between undertakings  OJ L395/1, corrected version  OJ L257/13, as amended by Council Regulation (EC) 1310/97 of 30 June 1997 amending Regulation (EEC) 4064/89 on the control of concentrations between undertakings  OJ L180/1, corrigendum  OJ L40/17 (‘1989 Merger Regulation’). It is established, however, that the definition of collective dominance under Art 102 is the same as in merger control: Case T-193/02 Laurent Piau v Commission  ECR II-209, para 111, concerning an alleged abuse of a collective dominant position under Art 102, where the Court of First Instance adopted the test set out in Case T-342/99 Airtours plc v Commission  ECR II-2585, para 62, concerning a case under the 1989 Merger Regulation. The judgment of the Court of First Instance in the Piau case was upheld in Case C-171/05 P Laurent Piau v Commission  ECR I-37 but the issue of collective dominance did not arise on appeal: see para 33. The different nature of the analysis, prospective in merger control and retrospective under Art 102, does not impinge on the definition of collective dominance but only on the type of evidence that may be relevant and on the standard of proof.
43. Joined Cases T-24/93 to T-26/93 and T-28/93 Compagnie Maritime Belge Transports SA v Commission  ECR II-1201 (Compagnie Maritime Belge) para 15, upheld in Joined Cases C-395/96 P and C-396/96 P Compagnie Maritime Belge.
45. Joined Cases T-24/93 to T-26/93 and T-28/93 Compagnie Maritime Belge, para 65. This analysis was upheld on appeal, Joined Cases C-395/96 P and C-396/96 P Compagnie Maritime Belge, paras 46–59. The quoted phrase is from para 65.
47. ibid paras 650 and 695. The language used by the Court in para 695 strongly suggests that structural and commercial links and direct or indirect contacts may be sufficient evidence that the undertakings are a collective entity but the inference may be rebutted by evidence of effective price competition between them. This principle only applies to non-oligopolistic collective dominance.
50. In which case the different companies would be a single undertaking: see eg Case T-102/92 Viho Europe BV v Commission  ECR II-17, paras 47–55, upheld on appeal in Case C-73/95 P Viho Europe BV v Commission  ECR I-5457.
52. Joined Cases T-24/93 to T-26/93 and T-28/93 Compagnie Maritime Belge, para 76. This principle is well established in relation to single dominance: see Case 85/76 Hoffmann-La Roche, para 41; Case 322/81 NV Nederlandsche Banden Industrie Michelin v Commission  ECR 3461(Michelin I), paras 32–61; Case 62/86 AKZO Chemie BV v Commission  ECR I-3359, para 60; Case T-30/89 Hilti AG v Commission  ECR II-163, paras 92–93, upheld in Case C-53/92 P Hilti AG v Commission  ECR I-667; Case T-228/97 Irish Sugar plc v Commission  ECR II-2969, paras 71–104; Case T-321/05 AstraZeneca AB v Commission  OJ C221/33, paras 243–253, under appeal in Case C-457/10 P AstraZeneca AB v Commission  OJ C301/18.
55. Joined Cases T-191/98 and T-212/98 to T-214/98 Atlantic Container Line, paras 903–942, 953–998, and 1009–1037. These factors correspond to the standard analysis of single dominance: see Mosso et al, ‘Article 82’, 320–335.
77. The references to economic theory or theories in the text distil the key principles from a vast literature: see JS Bain, ‘Output Quotas in Imperfect Cartels’ (1948) 62 Quarterly Journal of Economics 617; GJ Stigler, ‘A Theory of Oligopoly’, 44–61; J Friedman, ‘A Non-Cooperative Equilibrium for Supergames’ (1971) 38 Rev Econ Stud 1; R Porter, ‘A Study of Cartel Stability: The Joint Executive Committee, 1880–1886’ (1983) 14 Bell Journal of Economics 301; C Davidson and R Deneckere, ‘Horizontal Mergers and Collusive Behavior’ (1984) 2 Intl J Ind Org 117; EJ Green and RH Porter, ‘Non-Cooperative Collusion under Imperfect Price Information’ (1984) 52 Econometrica 87; WA Brock and J Scheinkman, ‘Price Setting Supergames with Capacity Constraints’ (1985) 52 Rev Econ Stud 371; MK Perry and RH Porter, ‘Oligopoly and the Incentive for Horizontal Merger’ (1985) 75 American Economic Rev 219; D Abreu, ‘Extremal Equilibria of Oligopolistic Supergames’ (1986) 39 Journal of Economic Theory 191; D Abreu, D Pearce, and E Stachetti, ‘Optimal Cartel Equilibria with Imperfect Monitoring’ (1986) 39 Journal of Economic Theory 251; D Fudenberg and E Maskin, ‘The Folk Theorem in Repeated Games with Discounting or with Incomplete Information’ (1986) 54 Econometrica 533; M Osborne and C Pitchik, ‘Price Competition in a Capacity-Constrained Duopoly’ (1986) 38 Journal of Economic Theory 238; J Rotemberg and G Saloner, ‘A Supergame-Theoretic Model of Price Wars during Booms’ (1986) 76 American Economic Rev 390; R Gilbert and M Lieberman, ‘Investment and Coordination in Oligopolistic Industries’ (1987) 18 RAND Journal of Economics 17; VE Lambson, ‘Optimal Penal Codes in Price-Setting Supergames with Capacity Constraints’ (1987) 54 Rev Econ Stud 385; R Schmalensee, ‘Competitive Advantage and Collusive Optima’ (1987) 5 Intl J Ind Org 351; J Harrington, ‘Collusion Among Asymmetric Firms: The Case of Different Discount Factors’ (1989) 7 Intl J Ind Org 289; BD Bernheim and MD Whinston, ‘Multimarket Contact and Collusive Behavior’ (1990) 21 RAND Journal of Economics 1; C Davidson and R Deneckere, ‘Excess Capacity and Collusion’ (1990) 31 International Economic Review 521; DL Booth et al, ‘An Empirical Model of Capacity Expansion and Pricing in an Oligopoly with Barometric Price Leadership: A Case Study of the Newsprint Industry in North America’ (1991) 39 Journal of Industrial Economics 255; J Haltiwanger and J Harrington, ‘The Impact of Cyclical Demand Movements on Collusive Behavior’ (1991) 22 RAND Journal of Economics 89; TW Ross, ‘Cartel Stability and Product Differentiation’ (1992) 10 Intl J Ind Org 1; RW Staiger and FA Wolak, ‘Collusive Pricing with Capacity Constraints in the Presence of Demand Uncertainty’ (1992) 23 RAND Journal of Economics 203; S Martin, ‘Endogenous Firm Efficiency in a Cournot Principal-Agent Model’ (1993) 59 Journal of Economic Theory 445; WN Evans and IN Kessides, ‘Living by the “Golden Rule”: Multimarket Contact in the US Airline Industry’ (1994) 109 Quarterly Journal of Economics 341; VE Lambson, ‘Some Results on Optimal Penal Codes in Asymmetric Bertrand Supergames’ (1994) 62 Journal of Economic Theory 444; S Martin, ‘R&D Joint Ventures and Tacit Product Market Collusion’ (1996) 11 European Journal of Political Economy 733; LR Christensen and R Caves, ‘Cheap Talk and Investment Rivalry in the Pulp and Paper Industry’ (1997) 45 Journal of Industrial Economics 47; M Armstrong, ‘Network Interconnection in Telecommunications’ (1998) 108 Economic Journal 545; JJ Laffont, P Rey, and J Tirole, ‘Network Competition: I. Overview and Nondiscriminatory Pricing’ (1998) 29 RAND Journal of Economics 1; KU Kühn, C Matutes, and B Moldovanu, ‘Fighting Collusion by Regulating Communication Between Firms’ (2001) 32 Economic Policy 169; O Compte, F Jenny, and P Rey, ‘Capacity Constraints, Mergers and Collusion’ (2002) 46 European Economic Review 1; I Marc et al, ‘The Economics of Tacit Collusion’ Final Report to DG Competition, European Commission (Toulouse, March 2003).
79. There is, of course, no legal definition of a ‘maverick’ firm. However, the Guidelines on the assessment of horizontal mergers under the Council Regulation on the control of concentrations between undertakings  OJ C31/5 (‘Guidelines on Horizontal Mergers’) refer to maverick firms twice: at para 20(d) as a ‘firm with a high likelihood of disrupting coordinated conduct’ and at para 42 as a ‘firm that has a history of preventing or disrupting coordination, for example by failing to follow price increases by its competitors, or has characteristics that give it an incentive to favour different strategic choices than its coordinating competitors would prefer’.
81. A Jaquemin and ME Slade, ‘Cartels, Collusion and Horizontal Merger’ in R Schmalensee and RD Willig (eds), Handbook of Industrial Organisation (Amsterdam, North Holland 1989) 418; J Harrington, ‘The Determination of Price and Output Quotas in a Heterogeneous Cartel’ (1991) 32 Intl Economic Rev 767; R Rothschild, ‘Cartel Stability when Costs are Heterogeneous’ (1999) 17 Intl J Ind Org 717.
88. For an analysis of some of these factors, see I Kokkoris, ‘Assessment of Mergers Inducing Coordinated Effects in the Presence of Explicit Collusion’ (2008) 31 World Competition 499, 501–507.
91. Case 85/76 Hoffmann-La Roche, para 41; Case 322/81 Michelin I, paras 32–61; Case C-62/86 AKZO, para 60; Case T-30/89 Hilti, paras 92–93; Case T-228/97 Irish Sugar, paras 71–104; Case T-321/05 AstraZeneca, paras 243–253.
98. Nestlé/Perrier (Case IV/M.190)  OJ L356/1, recital 42 (collective dominance found at 82%); Syniverse/BSG (Case COMP/M.4662)  OJ C101/25, recital 104 (further investigation was warranted on markets where the oligopolists had joint market shares between 70 and 90% or 90 and 100% depending on the geographical market definition, with symmetrical individual market shares); Pilkington-Techint/SIV (Case IV/M.358) recitals 25–27 (joint market share of the two leading producers of over 50% did not warrant investigation of collective dominance as there was clearly sufficient competitive pressure from outsiders but joint market share of the five main players of over 96% did).
100. eg in the following cases the number of the oligopolists was held to be in principle compatible with a collective dominant position even if some of these cases resulted in a finding that such a position was not proved: Case T-342/99 Airtours, paras 66–67 (four players before the merger and three players after the merger); Sony/BMG (Case COMP/M.3333) (19 July 2004), recital 156 (5 players before the merger and 4 players after the merger); E.ON, recital 13 (3 players); Case T-102/96 Gencor, para 207 (3 players before the merger and 2 players after the merger). In Price Waterhouse/Coopers & Lybrand (Case IV/M.1016)  OJ L50/27, recital 103, the Commission stated that ‘collective dominance involving more than three or four suppliers is unlikely simply because of the complexity of the interrelationships involved and consequent temptation to deviate: such a situation is unstable and untenable in the long term’.
121. The case law under Art 101 is well established: Case 48/69 Imperial Chemical Industries v Commission  ECR 619, para 64; Joined Cases 40/73 to 48/78, 50/73, 54/73 to 56/73, 111/73, 113/73, and 114/73 Coöperatieve Vereniging ‘Suiker Unie’ UA v Commission  ECR 1663 (Suiker Unie), para 174.
125. See, by analogy, Joined Cases C-204/00 P, C-205/00 P, C-211/00 P, C-213/00 P, C-217/00 P, and C-219/00 P Aalborg Portland A/S v Commission  ECR I-123, para 57: ‘In most cases, the existence of an anti-competitive practice or agreement must be inferred from a number of coincidences and indicia which, taken together, may, in the absence of another plausible explanation, constitute evidence of an infringement of the competition rules.’
130. Case T-342/99 Airtours, paras 134–147, where the Court addressed the issue from the perspective of the ability to detect deviations. Demand volatility, however, is also relevant to the ability to coordinate in the first place.
149. Case T-342/99 Airtours, para 203, where the Court said that the oligopolists would be able to add capacity during the same season only to a limited extent so that the retaliation would not be effective.
156. Joined Cases T-24/93 to T-26/93 and T-28/93 Compagnie Maritime Belge, para 105 (referring to action by ‘the members of Cewal’), paras 139 and 151 (although the judgment is not explicit on this point, the loyalty contracts were drawn up by the conference: Cewal, Cowac and Ukwal  OJ L34/20, recital 28), and paras 182–186.