Chillin'Competition

This post is the second instalment of the series dedicated to the Court of Justice’s ruling in Servizio Elettrico Nazionale (see here for the first instalment). This time, I turn to what is potentially the judgment’s main innovation: the replicability test.

The Court suggests in its ruling that a large fraction of the case law, from rebates to refusal to deal and margin squeeze, can be analysed under the umbrella of replicability. The idea of an all-encompassing framework to scrutinise potentially abusive conduct has always been attractive. This judgment’s is certainly not the first attempt and is unlikely to be the last.

Alas, creating an all-encompassing framework is as tempting as it is difficult. Servizio Elettrico Nazionale goes to show, in this vein, that capturing the essence of the case law under a single test will always be a major challenge. It seems to me, after a careful reading of the judgment, that the replicability test does not quite work (probably not even in its own terms).

I have the impression that replicability will be remembered alongside its illustrious predecessors such as the ‘no-economic sense’ and the ‘profit sacrifice’ tests (that is, as an approach that is sometimes useful and potentially illuminating but that might lead to enforcement errors if applied mechanically or across the board).

There are three main reasons why I have come to this conclusion:

• The case law that the Court cites in support of the test (including TeliaSonera and Bronner) is not exactly about the replicability of the practice, but about something else.
• The meaning and scope of the replicability test fluctuates: para 78 suggests that it is about the practice, but the analysis that follows suggests it is about something else (the assets, or the effects).
• The reasoning applying to the specifics of the case does not seem to fit the replicability test as defined in para 78.

TeliaSonera and Bronner are not about the replicability of the practice (nor is the AEC test)

As mentioned in my previous post, para 78 sets out the replicability test: under this framework, behaviour that cannot be replicated by an equally efficient rival is not in keeping with competition on the merits, and this insofar as the said behaviour flows from the firm’s dominant position.

The ‘as efficient competitor’ test is not about replicability

The Court goes on to support its claim in light of a broad range of precedents, in particular TeliaSonera and Bronner. According to the judgment, the ‘as efficient competitor’ test is essentially about whether an equally efficient firm would be able replicate the conduct implemented by the dominant firm.

This interpretation of the test is not easy to reconcile with TeliaSonera (paras 41-43, cited in the judgment). That judgment shows that the ‘as efficient competitor’ test is not about the practice, but the effects of the practice: more precisely, it seeks to ascertain whether an equally efficient firm would be able to offer its products or services otherwise than at a loss.

Put differently: the point of the test is not to assess the replicability of the practice, but whether, given the practice, an as efficient competitor would have the ability to compete on the market where the effects are manifested (which, by the way, is not necessarily the market in which the practice is implemented, as one might infer from para 78).

The refusal to deal case law is not about replicability, but indispensability

The same is true as far as non-pricing practices are concerned. Para 83 of Servizio Elettrico Nazionale interprets Bronner as revolving around replicability, and more precisely around whether the dominant firm’s assets can be replicated by an equally efficient rival.

However, Bronner (and Magill and IMS Health) is not about replicability, but about indispensability. It may sound similar, but it is not quite the same thing. According to Bronner, the question is not whether an equally efficient rival can build a replica of the dominant firm’s infrastructure, but whether the infrastructure is indispensable to compete on the relevant adjacent market.

Accordingly, an infrastructure may well be impossible to replicate and still not be indispensable. As explained by the Court in Bronner and IMS Health, if it is possible to enter the adjacent market by other means, even if less advantageous, the indispensability condition would not be met. Whether or not the infrastructure could be replicated is not a relevant, let alone decisive, consideration in this regard.

This point is crucial, as there seems to be a growing tendency to equate indispensability and replicability, even though the latter does not necessarily imply the former.

What is the replicability test about? The practice or the assets?

If one reads para 78 of Servizio Elettrico Nazionale, there seems to be little ambiguity about the nature of the replicability test: it is about the practice, and about whether it can be implemented by an equally efficient competitor.

The issue becomes less clear, however, when one takes a look at para 83 (which refers to Bronner and the exceptional circumstances test). This paragraph suggests that the replicability test relates to the infrastructure (or assets), not the behaviour (that is, the refusal to give access to the infrastructure).

The question is further complicated if one considers margin squeeze conduct. In a margin squeeze setting, replicability cannot be about the practice (since the practice is implemented on a market other than the one where it displays its effects) or the assets (since the relevant question is not whether the infrastructure can be replicated, but whether the spread between wholesale and retail prices allows equally efficient rivals to compete on the adjacent market).

The fact that the meaning (even the relevance) of replicability seems to fluctuate from one paragraph to the next suggests that, as much as we would want it to, one size does not fit all. It also means, by the same token, that the replicability test most probably does not work as an overarching framework.

The case seems to be decided on grounds other than replicability

Arguably, the single most reliable indicator suggesting that the replicability test may not work as an overarching framework is that the dispute itself is ultimately decided on grounds other than those laid down in para 78.

As already mentioned, para 78 refers to a practice that flows from the dominant position (implying, as explained in the previous instalment, the need to show a link between the dominant position and the practice). Para 101, however, suggests that the crucial consideration is another one, namely the fact that the firm has a unique status and set of assets as a former legal monopoly.

In the following paragraph (102), the Court holds that the replicability of the assets by other means is not a relevant consideration. This point, wholly sensible in the context of the case, is not easy to reconcile with para 83 and the refusal to deal case law: under the Bronner doctrine, there would be no abuse if there are other means, even if less advantageous, to compete on the adjacent market where the refusal displays its effects.

As can be seen, the case is not decided on replicability grounds in the end, but based on the fact that the dominant firm’s unique assets had not been obtained on the merits (but rather as a by-product of the exclusive rights granted by the State) and that, as a former monopoly, the undertaking had a particularly stringent duty not to impair the competitive process.

On those (somewhat narrower) grounds, the judgment is as illuminating as it is uncontroversial. A most useful addition to the growing case law applying to incumbents in liberalised markets.

I very much look forward to your comments.