The long-awaited Intel judgment on rebates, released on 6 September, has established, following years of debate, a standard for regulatory intervention that requires a focus on the likely effects of individual rebate schemes in their particular context. In doing so it has overruled a more rigid approach taken by the General Court (GC) in the same case in 2014, annulled the GC’s judgment and remitted the case to it on the facts. The ruling continues the Court’s move towards the path first forged by the Commission. It confirms what many of us have long argued, and most companies in the real world (both dominant firms and complainants) have long recognised - that the rules are not black and white, and an assessment of the features, context and actual and likely effects of particular schemes is required in each case.
As with our previous articles at the various stages of this case – see links below – this article will focus on the rebates aspects of the judgment. The issues of procedural irregularity and jurisdictional reach are separate and for discussion elsewhere.
A brief recap: eight years ago the European Commission issued its mammoth decision in the Intel rebates case, following a complaint by rival AMD. The Decision was hailed as the first major implementation of the Commission’s new “effects based” approach announced in its abuse of dominance guidelines earlier that year. The facts concerned (i) exclusivity arrangements for microchips (a market in which Intel was held dominant) in contracts with computer hardware manufacturers and (ii) payments conditional on restrictions on distribution of computers containing rival AMD chips. Both practices were found abusive and Intel was fined over €1 billion. The GC upheld and the ECJ has struck down and remitted.
But the relevance of the case for businesses is not about the particular facts. It is about the broader question of how conditional pricing practices by dominant firms - and in particular loyalty rebates - are treated under competition law.
The oldest Court of Justice case on this topic – Hoffman-La-Roche – prohibited a dominant firm from making discounts or rebates conditional on purchasing “all or most” of a customer’s requirements from the dominant firm over a period. Intel, without expressly overruling Hoffman-La-Roche, modifies it by saying that where the dominant firm puts forward arguments that the scheme was not likely in practice to have anti-competitive effects these arguments need to be analysed. Since any well-advised defendant will argue its case that must surely amount to a re-setting of the standard for intervention.
As to what needs to be looked at, the Court is clear. First, a general set of factors – the extent of the firm’s dominance, what the terms and features of the rebate are – including the size of the discount and how long the reference period lasts. What proportion of the market the scheme covers is another feature here. This may seem obvious, but this last one in particular has been highly controversial: the GC said it did not matter and the ECJ sent mixed messages in Post Danmark II, its other leading rebates judgment handed down two years before Intel. The Intel GC, in short, had said that none of these features mattered: loyalty rebates were treated as unlawful irrespective of the facts, subject to a defence of objective justification interpreted so narrowly as not to be meaningful in practice. That approach has clearly now been overturned.
Second, a central feature of the Court’s judgment is the relevance of the comparative efficiency of rivals. The nub of this issue is that the Commission, in its guidelines, focused on the (likely) exclusion of rivals who were at least as efficient as the dominant firm as a key benchmark for likely harm. In its Decision the Commission rode two horses: it set out a 150 page analysis seeking to demonstrate that the rebates in that case meant that a rival which was as efficient as Intel would have to offer uneconomic prices (in the jargon, the schemes failed the “as efficient competitor” (or “AEC”) test). But at the same time it argued that under the law it did not need to carry out that analysis at all because the schemes were inherently unlawful.
The Court of Justice’s decision to annul the GC judgment rests in part on the failure of the GC to analyse Intel’s arguments on the AEC test. The Court is quite careful to say that it is sending it back because the arguments must be analysed – rather than saying that only schemes which exclude “as efficient competitors” are unlawful. And elsewhere it says that Article 102 (the prohibition on abuse of dominance) prohibits “among other things” practices that exclude equally efficient rivals. But the point features repeatedly in the judgment, and the thrust does seem to be that this issue is relevant and needs to be looked at by a court or regulator faced with examining one of these schemes.
Finally, the Intel judgment is not an answer to all problems of rebate analysis. Far from it: the relevant passage is extremely short. It deals with what it needs to deal with to remit the case and no more. Many questions about the categorisation of different types of rebates (e.g. target and volume structures, retroactive reward arrangements) – discussed by both the General Court and Advocate General – are not touched upon here. But the direction of travel is clear.
The judgment comes 17 years after the initial complaint by AMD. With the remittal to the GC the saga seems certain to exceed 20+ years. Whether the long slog will have been worth it for any of the parties involved is not clear. But if, as the dust settles, we have greater clarity around the assessment that needs to be carried out to determine where the competition law boundaries lie for rebate schemes, it will have been worth it for the wider business community.