Rebates Revisited (Again) – The Continuing Article 82 Debate

John Kallaugher[1]

In June 2004, Brian Sher and I published a paper on the treatment of rebates under Article 82.[2] In that paper, we criticised the Article 82 approach to rebates as it stood following the judgments of the Court of First Instance in Michelin II and British Airways.[3] Our critique had two principal elements: (i) the current approach incorrectly assumes that any rebate scheme that has a “loyalty-enhancing” effect will lead to competitive harm; and (ii) the current approach fails to recognise the welfare benefits of rebate schemes that go beyond transaction-specific cost savings. We argued for a case-specific assessment of likely competitive harm and for a more open approach to assessing consumer benefit. Since the publication of our paper, the treatment of rebates has been subject to extensive discussion, in part as a result of the broader discussion of Article 82.[4] The purpose of this paper is to summarise the current position regarding rebates under Article 82 and to comment on some points raised in the continuing debate.

  1. Discounts or Exclusivity? --The Special Position of Rebates in Competition Law

The reason that rebate schemes play such a central role in the case law can be attributed to their ambiguous position in the traditional theology of Article 82. Exclusive dealing arrangements between dominant firms and their customers have been condemned as virtually per se exclusionary since the Sugar Cases in the mid-1970’s. Volume discounts, in contrast, are traditionally viewed as pro-competitive unless they are either “predatory” or “discriminatory”. This distinction goes back to the concept of “performance-based competition.” Price reductions are the classic form of “competition on the merits” and lead to short-term consumer benefits. In contrast, exclusive dealing commitments (in the traditional analysis) bar rivals without direct benefits for consumers. A rebate scheme falls somewhere between these extremes. As a retrospective discount for a block of purchases, a rebate scheme can function as a volume discount. A rebate scheme may also, however, function as a “quantity-forcing” or exclusive purchasing obligation if it operates to encourage or even require a purchaser to obtain its supplies from the seller operating the scheme.

The case law largely reflects the tension between these two ways of viewing rebates. In the traditional approach to Article 82, a rebate scheme that functions as a volume discount ought to be challenged (if at all) under the rules applicable to discounts more generally. A rebate scheme that functions as an exclusive dealing obligation, in contrast, ought to be challenged on the same basis as other exclusive dealing arrangements. This has led to an emphasis on the “loyalty-enhancing” aspects of rebate schemes, as the element that makes a scheme more equivalent to an exclusive or quasi-exclusive purchasing commitment for participating customers. The approach outlined by Gyselen in his 2003 paper, largely followed by the CFI in Michelin II, attempts to give more content to consideration of where a rebate scheme may be deemed loyalty-enhancing.

The relationship between rebates and quantity discounts is important for assessing the competitive harm of rebate schemes, particularly in the context of some of the more “imaginative” legal rules now under discussion. The traditional view that treated quantity discounts as presumptively lawful under Article 82 made a lot of sense. Requiring dominant firms to justify any price differentiation on the basis of objective cost differences would almost certainly lead to higher overall prices and consumer welfare losses in the great majority of cases. As long as this view prevails, the economic effects of a rebate scheme must be assessed by comparison to the effects inherent in any quantity discount scheme. The unique effects of a rebate scheme derive from the fact that it aggregates otherwise separate purchasing decisions over time. To the extent that the competitive impact of a rebate scheme is attributable to the size of the discount or to the threshold for granting that discount, the impact of a rebate scheme is the same as the impact of an advance commitment to purchase the same amount or tranches of the same amount in proportionate quantities and should ber assessed on the same basis.[5]

The dichotomy between rebate schemes that function as volume discounts and rebate schemes that function as exclusive dealing arrangements ignores the dynamic aspect of many rebate programmes. In contrast to an exclusive dealing requirement, a rebate programme creates an incentive for the purchaser to increase its own sales efforts and thereby increase its own total sales. This incentive can have the effect of increasing total sales in the relevant market. It is thus possible for competitors to maintain or even expand total sales in the face of rebate programmes operated by dominant firms. The impact of rebates as tool to motivate resellers to make more active sales efforts is a principal reason for the popularity of rebate schemes for both dominant and non-dominant firms. The potential of rebate schemes to increase total output while lowering overall prices is simply ignored in the traditional assessment.

  1. The Current Legal Position

The current legal position can be summarised as follows:

  • Fidelity Rebates. A rebate that is available where the purchaser has not made purchases from a rival firm during the relevant reference period have been regarded as abusive since Hoffmann-LaRoche. The same rule applies where a rebate is provided on condition that the customer purchase a set percentage of its requirements from the dominant firm.
  • Target Rebates. In principle, rebates paid by a dominant firm where a customer achieves an individually set target are abusive following British Airways. This rule applies in particular to targets based on sales in the equivalent period in the previous year (sometimes referred to as “growth rebates”). It is based on an application of Article 82(c) -- where two customers purchase the same amount but had different sales in the previous year, one customer will qualify for a growth rebate and the other will not.
  • Volume Rebates. Following Michelin II, a grid of rebates – where any customer can qualify for a rebate by achieving sales in the reference period in excess of the specified threshold may be abusive unless:
  • The reference period is short or
  • The purchaser obtains a rebate only on purchases above the threshold (sometimes described as a “back to zero” system) or
  • The system is objectively justified (Michelin II leaves room for objective justification – but suggests that this may require transaction-specific cost justification)

Despite the considerable legal and administrative attention to rebates over the last twenty five years there are still some significant unresolved issues in applying the traditional approach:

  • How long is long? Michelin II condemned volume rebates where reference period for rebate was “long”. In resolving airline cases post-decision in British Airways, Commission accepted six month periods (corresponding with airline traffic seasons). Commission accepted 3/5/4 month reference periods in settling Michelin II.
  • Do all customers need to be in the same grid? Since the British Airways judgment is based on Article 82(c), it should be possible to have different rebate grids for different customer groups as long as those customers are not similarly situated.
  • What happens if there are only two customers? The requirements of Michelin II and British Airways become a bit onerous where there are a limited number of customers with very different demand patterns. Creating a single rebate grid has the effect of creating de facto individual targets. Again this should not matter if there is no discrimination – even a pure target is arguably permissible with short reference periods.
  • What about lawful exclusive dealing arrangements? Since rebate schemes are principally objectionable because they are equated to exclusive dealing arrangements, using rebate thresholds to set prices in a tender arrangement or other lawful exclusive arrangement should not be a concern. Arguably this should also be the case where de facto exclusivity is the norm (i.e., where the customer would source from a sole supplier regardless of whether there is a rebate scheme in place).
  • Rebutting “Loyalty Effects.” The current case law puts the principal evidence in “proving” unlawful rebates on the existence of a “loyalty-inducing” effect. The “suction effect” of back-to-zero rebates in Michelin II is the best example of judicial consideration of “loyalty induction” Since the proof of loyalty-inducing effects is the central requirement of the current case law, it should therefore be open to a defendant to show that a specific scheme does not have such effects. It could be argued, for example that evidence shows that there is no correlation between purchasing decisions and rebate reference periods, even where the purchase might have qualified for a rebate with limited additional purchases. Anecdotal (unattributable) evidence suggests that such arguments have indeed been made and even accepted by regulatory authorities.
  1. The Critique of the Current Position

The principal criticisms leveled at the current case law have already been mentioned in the introduction to this paper. The most important criticism is that current approach to rebates does not require any assessment of the actual effects of the rebate scheme at issue in an individual case. If there is a significant loyalty inducing effect for an single customer, then showing that effect suffices to render the scheme unlawful.[6] The failure to consider real economic effects is, of course, an issue that applies more generally to application of Article 82. It is particularly significant in the rebate context, however, because some rebate schemes clearly do not have anti-competitive effects and some clearly lead to significant consumer welfare benefits. Thus the possibility of Type One errors (prohibiting conduct that is not harmful – false positives) seems particularly great in the rebate context if some attention is not given to the effects of a specific scheme.

The second important criticism of the current case law involves the failure to recognize that rebate schemes may have efficiency benefits that go beyond transaction-specific cost benefits. As noted in the previous section, this aspect of the law is less defined. The reliance of the CFI in Michelin II on precedents involving objective justification in discrimination cases suggests, however, a restrictive approach to economic justification that would lead to further Type One errors.

The rebate cases also illustrate general problems with application of Article 82 that transcend the specific issue of rebates. One such problem is the mechanical application of Article 82(c) as a substitute for reasoned analysis. This reflects, however, the broader problem that the courts can no longer articulate how Article 82 fits into the Treaty’s competition law enforcement system. The discussion of effects in Michelin II is particularly telling in this regard. The judgment asserts that there is no need to assess effects under Article 82, because under Article 82 the object and effect of restricting competition are the same.[7] catches any conduct that has the “object” of restricting competition. The judgment then equates a restriction of competition (which should involve a detrimental effect on consumer welfare) with an increase in the “loyalty” of an individual customer.[8] Until the goals of Article 82 are properly articulated, the abuse rules under Article 82 will remain arbitrary and inconsistent.

  1. The Sher/Kallaugher Proposal

Brian Sher and I proposed in our paper an alternative approach to assessing rebate schemes. This would involve (i) assessment of the likely anticompetitive effects of a specific rebate scheme; and (ii) a more open-ended consideration of consumer benefits. This proposal still appears to be a sensible approach. Without revisiting the previous discussion, some issues regarding the assessment of effects under this test merit brief mention here.

A central criticism of a test that requires assessment of anticompetitive effects is that they will be difficult to apply because of the complexity and cost of a proper economic analysis. It is also suggested that requiring proof of economic effects would lead to Type Two errors because serious competitive harm could occur before the requisite competitive harm could be shown. This argument would clearly have merit if application of Article 82 (or meaningful self-assessment) required a full economic analysis of the sort encountered in a Phase II merger case. Identifying likely anticompetitive effects in a conduct case does not, however, require econometric analysis. The real requirement should be that the Commission (or a Member State authority or national court plaintiff) establish a theory of competitive harm that is based on real market conditions.

The facts in Michelin II provide a good example of what this could mean. That case involved a rebate scheme for distributors of heavy goods vehicle tyres in France. Assuming that the rebate scheme actually had a “loyalty-enhancing” effect, the central question ought to be whether this scheme is likely (or reasonably likely) to lead to higher prices for heavy goods vehicle tyres in France or prevent prices from going lower.[9] To answer this question would require identification of the principal competitors and potential competitors and the extent to which they rely, or are likely to rely, on the Michelin distributors as a channel to market. Put another way, would exclusion from the Michelin distributors raise the distribution costs for rival manufacturers in a way that lead them to exit the market or charge higher prices than would otherwise apply? If the Commission could show that alternative distribution methods (such as direct sales or focusing on non-Michelin outlets) would give rival manufacturers insufficient sales to achieve distribution economies[10] or would involve significantly higher costs per sale, then it would have a theory of competitive harm based on foreclosure. It may be that the Michelin II facts would actually support this conclusion. The decision simply does not consider these issues, however, since it assumes that any loyalty enhancing rebate scheme is anticompetitive.

  1. Alternative Proposals and Current Issues

The current debates on application of Article 82 have given rise to a variety of arguments or proposals that are relevant to consideration of rebate schemes. Some of these points are discussed below as they relate to rebate schemes. This brief survey makes no claim to be comprehensive.

The Need for Administrable Rules. Some commentators argue that any effects-based system creates legal uncertainty and prevents effective enforcement activity. The German Federal Cartel Office is identified with this view, although careful consideration of the practice of the FCO indicates a less formalistic approach than that currently applicable to rebates under Article 82.[11] Clear rules that can be easily applied have definite advantages. The problem of achieving the right balance between “administrability” or predictability on the one hand and correct results from a policy perspective on the other raises issues that go beyond rebates. If administrability or predictability become governing critera, however, it must be evident that these rules cannot take a maximalist position designed to cover any situation where competitive harm is theoretically possible. The resulting Type One errors would mean that the competition law enforcement system itself was leading to competitive harm. Generally applicable rules should therefore be set at a point where the likelihood (and extent) of Type One and Type Two errors should balance out.

The test for rebates prior to Irish Sugar, Michelin II, and British Airways arguably met this test. In essence, the prior law prohibited fidelity rebates, but allowed target rebates (with a short reference period) and did not regulate volume rebates. On the traditional view that rebate schemes become objectionable where they become more like exclusive purchasing arrangements, the prior test caught the clearly abusive conduct. The benefits of rebate schemes (which are primarily associated with target and volume rebates) were preserved to a substantial extent.[12] The subsequent case law destroyed this balance. In short, if predictable quasi-per se rules are desirable, the current rebate rules do not satisfy the criteria that should apply to such rules.

“Super-dominance” and Rebates. Some commentators have proposed that abuse rules under Article 82 should vary depending on the degree of market power possessed by the dominant firm. Thus where a firm has real monopoly power (particularly where the firm in question is a present or former State-owned monopoly) the firm would be subject to more stringent obligations, in particular with respect to discrimination or obligations to supply. This approach is attractive for many reasons, not least because it would preserve the value of Article 82 (in conjunction with Article 86) as a tool for opening up protected markets. From the perspective of Article 82 itself, this approach is attractive because the likelihood of competitive harm is usually far greater where the levels of market power are high.

Rebate rules could be suitable candidates for a “super-dominance” assessment.[13] Anecdotal evidence (again unattributable) suggests that where firms with market shares in the 65% or higher range implement rebate schemes there are very real risks of significant foreclosure effects. (This may reflect the high residual demand for the products in question – the higher the residual demand the more difficult it becomes for customers to forego the back-to-zero benefits of the rebate scheme.) Below the 55-60% range, serious foreclosure effects are actually very unlikely, particularly by comparison with the position if the dominant firm institutes volume discounts at a similar scale. It could be sensible to deploy a modified per se rule where market shares exceed a specified percentage, while requiring a showing of economic harm in cases below that threshold.