The European Commission’s reform of Article 82 EC (now 102 TFEU) on abuse of dominance began in 2005. It culminated in a 2008 Guidance paper which emphasised a more effects-based approach to abuse of dominance.1 However, the General Court’s Intel judgment of 2014 slammed the door on effects-based approaches to abuse of dominance under Article 102 TFEU, at least with regard to rebates granted by a dominant firm on the condition of exclusivity.2 An article on this judgment by Wouter Wils, a senior Commission official, reopened the whole debate on form-based versus effects-based approaches.3 We are back to square one. Back, therefore, to this article we wrote in May 2005.
1 European Commission (2008), ‘Guidance on the Commission’s Enforcement Priorities in Applying Article 82 EC Treaty to Abusive Exclusionary Conduct by Dominant Undertakings’, December. 2 Case T-286/09, Intel Corp. v EC Commission, judgment of 12 June 2014. 3 Wils, W. (2014), ‘The Judgment of the EU General Court in Intel and the So-Called “More Economic Approach” to Abuse of Dominance’, World Competition, 37:4.
The European Commission’s review of its policy on abuse of dominance is one of the main policy issues in EU competition law in 2005. This review follows the substantial changes made in recent years to the other two pillars of competition law: merger control, and restrictive agreements under Article 81 of the EC Treaty. One outcome of these previous reforms is that EU policy on mergers and agreements has been brought more into line with current economic thinking.
The general expectation is that the abuse of dominance provisions in Article 82 will follow a similar path. EU case law on abuse has for years been criticised as legalistic and interventionist, and the current review is seen by many as an opportunity for change. Various commentators have stressed the desirability of moving towards an approach that emphasises the actual or expected economic effects of allegedly abusive behaviour by dominant firms, rather than its form.
Perhaps less attention has been paid thus far to where it is exactly that the current approach fails. Many have argued that Article 82 protects competitors rather than competition—but how does this come about? This article explains that one of the fundamental shortcomings in the current case law is the virtual per se prohibition of certain practices once a firm is deemed to be dominant—in other words, dominance is used as a shortcut to infer anticompetitive effects. This can be misleading, as the implied link between dominance and effects does not always hold.
Bulls in a china shop
Article 82 policy has traditionally been influenced by the ‘ordo-liberal school’. In essence, this school of thought emphasises individual freedom as the primary objective for competition policy, and considers that the presence of dominant firms weakens the competitive process and reduces the economic freedom of other market participants.
The notion that the mere existence of dominant firms is dangerous for competition is still deeply embedded in EU law. A dominant firm is in effect regarded as the proverbial bull in a china shop—it must be restrained to prevent it from inflicting further damage to its already fragile surroundings. As formally established in Michelin (1983), a dominant firm has a ‘special responsibility not to allow its conduct to impair genuine undistorted competition on the common market’.
This view on how competition works appears somewhat outdated. Economic theory and practical experience over the past 30–40 years have shown that competitive dynamics can function well even if a market has some very large players. (Indeed, large players and/or temporary positions of market power can improve competitive dynamics.) The theory has also established that certain behaviour can have positive efficiency effects, even if practised by dominant firms. In other words, using the bull analogy to describe dominant firms does not fit well with current thinking on how markets work.
The shortcomings of the current approach to Article 82 arise from a combination of the following three policy aspects:
EU case law has only one threshold for dominance, regardless of the type of practice at hand;
this threshold is set relatively low; and
the threshold constitutes the basis for a form-based per se prohibition of certain behaviour.
With regard to the first of these, it is important to bear in mind that market power is a matter of degree—Microsoft has market power, but so has a small corner shop (provided there are few other shops nearby). Dominance can be interpreted as a very high degree of market power—one that enables a firm to sustain prices above the competitive level without inducing customer switching or competitor entry.
The assessment of dominance can be a useful intermediary step in the analysis of an alleged abuse of dominance. In particular, many types of behaviour are likely to be of little competitive concern if the accused firm is not dominant—in other words, dominance is a necessary condition, and can be used to filter out cases below the threshold. However, in EU law, dominance is also used the other way around—i.e. to apply virtual per se prohibitions to firms above the threshold.
This policy approach can be problematic. From an economics perspective, the competitive effects of any business practice will depend on, first, the type of practice in question, and second, the degree of market power of the firm in question. The current ‘one-size-fits-all’ dominance threshold—the traditional market share rule of thumb of 40–50% (combined with some indications of entry barriers) still seems to be the norm—is arguably set too low to allow for any strong inferences of anticompetitive effects, as discussed in this article.
This is not to say that a per se prohibition above a certain dominance threshold can never be the correct policy approach. Such a prohibition might be justified if the threshold were set sufficiently high—some types of behaviour (such as refusal to deal or margin squeeze) are often invariably anticompetitive if engaged in by a firm with a near-monopoly. In this respect it is worth noting that some recent EU cases have referred to the concept of ‘super-dominance’, which might be a reasonable threshold above which the behaviour under investigation can be presumed to have anticompetitive effects.
However, if the current, lower threshold for dominance is maintained, this can only be used as an intermediate step in the analysis of actual or likely anticompetitive effects—i.e. as a necessary but not sufficient condition. A further assessment of those effects is required.
Predation and targeted discounting
Predatory pricing and targeted discounting are practices where the link between dominance and effects may not hold. US antitrust law has established the ‘recoupment test’ for predation cases, which places strong emphasis on whether market structure is such that predation is feasible (and hence the initial losses from predation can be recouped through subsequent monopoly profits).
While sometimes criticised as overly harsh on complainants, the recoupment test has consumer welfare at its heart, as it essentially allows any aggressive price cut as long as there is no prospect for successful monopolisation of the market.
In contrast, EU law has explicitly rejected the recoupment test—it considers pricing below the acceptable level (see below) by dominant firms as abusive in itself if there is a risk that competitors are eliminated, and therefore sees no need to establish the feasibility of predation. Another (more economic) justification for rejecting the recoupment test might be that the possession of dominance in itself means that feasibility of predation is likely. However, the conditions for recoupment are typically more stringent than those for dominance, as defined in EU law. (For example, a very high market share is needed from the start, and greater emphasis is placed on the possibility of successfully maintaining monopoly prices in future.)
Instead, predatory pricing cases in the EU focus primarily on the relationship between prices and costs. In the AKZO judgment (1993), the European Court of Justice determined that predation can be presumed if a dominant firm sets prices below average variable costs (AVC). As such, this test is reasonably in line with economic theory, which also identifies marginal cost (or some variant) as a relevant price floor for predation cases. Yet the risk is that pricing below AVC by dominant firms is outlawed per se, without any consideration of whether such pricing:
actually has negative effects on competition, which is not always the case—for example, below-AVC pricing during one month only (the basis for the UK Office of Fair Trading’s finding of predation in Aberdeen Journals) is arguably not sufficient to infer an anticompetitive effect;
may be justified on efficiency grounds—for example, if there are strong network effects, below-AVC pricing may be required to gain critical mass, even in the absence of any competitors.
The link between dominance and effects on competition may also not hold in targeted discounting and ‘fighting brands’ cases. Targeted discounting arises where a dominant firm sets low prices only for those customers who are using, or likely to switch to, a competitor, but keeps other prices unchanged (the AKZO case is an example). Fighting brands—or fighting ships in Compagnie Maritime Belge and fighting titles in Aberdeen Journals (both cases referred to above)—are also specifically targeted at competitors. They are priced low, while the price of the main brand remains unaffected.
In ‘standard’ predation cases, economic theory establishes a link between market share and the likelihood of success of predation—the higher a predator’s market share, the quicker it can depress market price, and hence the greater the chance that rivals will leave the market soon. With targeted discounting and fighting brands, however, this link does not hold because the predator is not reducing the prices on all its products in the market. Take the example in Figure 1. A dominant firm with 90% of the market pitches a fighting brand (representing 10%) against the competitor with 10% market share. The firm therefore only sets below-AVC prices for the fighting brand, keeping the price of its main brand unaffected. In other words, the predator is not using the full weight of its dominance (i.e. its 90% share), and hence the possession of such dominance is not directly informative on the actual or likely effects on competition.
Figure 1 Predation through a fighting brand
This is not to say that complaints about these forms of predation should be rejected outright. The point is that the actual or likely effects of targeted discounting or fighting brands need to be assessed more carefully, and cannot be inferred from the position of dominance in the broader market as such.
Form-based rules have also been applied to practices that may result in vertical foreclosure by dominant firms, such as exclusivity requirements and loyalty incentives on distributors. Under EU law, dominant firms are virtually prevented from engaging in any such practice. Again, such a per se approach may be overly intrusive and unrelated to the actual or likely effects on competition. The existence of dominance is a necessary but not sufficient condition for this to apply. Whether a significant part of the distribution channel is indeed foreclosed must also be assessed. For example, if a dominant firm with 60% of the market imposes exclusivity requirements on 10% of all distributors, the foreclosure effect is probably limited. In this case, a prohibition seems less appropriate than if, say, 60% of the distribution channel were foreclosed, particularly if such exclusivity generates certain efficiency benefits (which would also need to be assessed as part of the effects-based test).
The treatment of loyalty rebates illustrates the shortcomings of the form-based approach in EU law. In two judgments in 2003—Michelin II and British Airways—the Court of First Instance confirmed the long-established EU policy that dominant firms are only allowed to offer discounts that relate to cost savings, but not to encourage loyalty. Any considerations of the actual or likely effects of such discounts on competition are deemed irrelevant.
For example, in British Airways, the concern was that the loyalty incentives offered by the airline to travel agents were retrospective—i.e. paid on all ticket sales above the performance target, and not just on the incremental sales above that target—and could thus induce agents who are close to their sales target to promote British Airways rather than rival airlines. The form of this incentive scheme was considered an abuse, without much consideration of the effects. One such effect could have been the foreclosure of sales channels to Virgin Atlantic, a new competitor and complainant in this case. However, Virgin had continued to gain market share throughout the affected period, which indicates that the competitive effect of the loyalty schemes was probably limited. For the US court that reviewed the same facts, this was one reason (among others) to reject the complaint that Virgin had filed in that jurisdiction. In contrast, the European Commission simply made the point that Virgin would have had even more success in the absence of the loyalty schemes.
Abuse in a related market
A final illustration of where the link between dominance and effects may not hold refers to the principle that dominance in one market can also be abused in a related market. This principle means that, in practice, dominant firms can be subject to the same per se prohibitions in those related markets. In terms of the analogy used earlier in this article, the bull is feared not only inside the china shop, but is also considered capable of inflicting damage as soon as it sets foot outside. Such concerns may overstate the power of dominant firms.
For a more accurate assessment of the competitive effects of an alleged abuse in a related market, it is necessary to establish the precise link between the two markets and to identify the mechanism through which such abuse would work. In some cases there may well be links that enable this type of abuse—for example, where markets are vertically related and one provides an important input to the other, or where the two products are complements. Such links may facilitate leveraging practices such as refusal to supply, tying and discrimination. Making the existence of such links between the two markets explicit is an important part of any effects-based test, but this is not always done in practice—it is often simply taken for granted that a dominant firm can do harm in neighbouring markets.
Furthermore, an assessment of this type of practice would need to consider the actual or likely effects on competition in the related market—is the practice in question likely to result in monopolisation of the related market, or does it merely give the dominant firm some competitive edge in that market? If the latter holds, there is arguably little cause for concern. The likely effects will depend on the relative strength of the dominant firm and other competitors in the related market. Dominance in the first market in itself is no guarantee of success in the second market. Even a company like Microsoft might find it hard to monopolise every market it ventures into.
The Commission’s review of its policy on Article 82 will lead to a greater consideration of the current economic thinking on abuse of dominance, and there will be much debate on how far this should go. Admittedly, there has as yet been relatively little analysis of the potential negative effects on consumers of the current form-based approach—most of the criticisms that have been made are based on anecdotal evidence, or simply on ideological grounds. This article has set out one of the main shortcomings of the current approach, namely its reliance on the dominance determination as a shortcut to inferring anticompetitive effects. This can lead to incorrect, and probably too much, intervention against many types of business practices by firms that are considered dominant.
A move to some form of effects-based test has been advocated. From an economics perspective, this would be a more appropriate approach to Article 82 as it would allow greater emphasis on efficiency and consumer welfare. Such an effects-based test does not necessarily mean—as some may fear—fully quantifying and weighting all the costs and benefits of the alleged abusive behaviour. Often, it may simply involve the consideration of a range of economic indicators of actual and likely competitive effects that go beyond the assessment of dominance, such as the likelihood of success in excluding competitors and possible efficiency benefits. Dominance would then merely be one of the indicators of relevance to the assessment of allegedly abusive practices.
It is not clear why the as-efficient competitor test should be considered appropriate for margin squeeze but dismissed for other forms of abuse (such as Intel’s rebates that are conditional on exclusivity). The debate on effects versus form also rages in the context of Article 101 TFEU, which deals with restrictive agreements. In another 2014 judgment, Cartes bancaires, the Court of Justice clarified the restrictions on the extent to which competition authorities can treat certain agreements as restrictions ‘by object’—i.e. as inherently anticompetitive and therefore not requiring a full effects-based analysis. This judgment means that, going forward, it is likely that more Article 101 cases will require an effects-based analysis.3 Agenda will continue to lead the debate on the issue, and reassess the situation in 2025…