To begin with, despite all this talk about cost definitions, this is not actually a case about predatory pricing. For that, you need either pricing below average variable cost or anti-competitive intent. Cf. par. 27 of the Post Danmark judgement or par.72 of Akzo v. Commission, which says:
72. Moreover, prices below average total costs, that is to say, fixed costs plus variable costs, but above average variable costs, must be regarded as abusive if they are determined as part of a plan for eliminating a competitor. Such prices can drive from the market undertakings which are perhaps as efficient as the dominant undertaking but which, because of their smaller financial resources, are incapable of withstanding the competition waged against them.Since neither is shown here, predatory pricing is off the table.
So then why are we talking about definitions of costs? Well, your guess is as good as mine.
To be clear, there is a possible issue of unlawful cross-subsidisation here. If Post Danmark charged prices below average total costs in the unaddressed mail sector, that money has to come from somewhere. And if that money comes from whatever profit it is making in the sector where it has a legal monopoly and a universal service obligation, there may well be a legal problem. I'm just not so sure why that would be a competition law problem. In general, I don't see why it would be an abuse of dominance for an undertaking to cross-subsidise between its divisions absent a showing of predatory pricing, and I also don't see why the presence of a universal service obligation would change that, as far as competition law is concerned. The Advocate General cites a Commission Decision against Deutsche Post where an infringement was found in similar circumstances, but no one seems to cite any actual case law.
So logically, this is simply a case about price discrimination. So then why are we talking about definitions of costs? If the goal is to talk about exclusionary effect (cf. par. 121 of the AG's opinion), rather than exclusionary intent, surely the relevant cost data concerns the cost of Post Danmark's main competitors, in this case Forbruger-Kontakt.
AG Mengozzi went another way. He proposed a third way between an effect-based analysis and an intent-based approach, whereby it would be an abuse to offer prices below Average Total Cost if this is financed by cross-subsidisation from a universal service obligation. For the reasons stated above I wouldn't necessarily favour this, but it at least explains why we should be talking about Post Danmark's costs structure, and it has the merit of being more or less objective and workable. Its only problems are that it unnecessarily imports sector-specific regulation into competition law and that it treats companies more strictly than the interests of consumers require.
The Court, on the other hand, spends all this time talking about Akzo, only to observe that it doesn't help much (par. 29). Then it spends a lot of time talking about cost definitions, only to conclude that a case for abuse of dominance still requires a showing of exclusionary effect, which is exactly where the analysis also started (par. 21). The only novelty? In the zone between Average Total Costs and Average Incremental Costs, an abuse of dominance is possible even without exclusionary intent - this contra Akzo - as long as the exclusionary effect is "actual" or "likely". If this rule applies to all abusive pricing, it would essentially be an overruling of Akzo, but is it? The Court describes the behaviour in question as "a policy by which a dominant undertaking charges low prices to certain major customers of a competitor". To me, this sounds more like a description of the facts of the case than a limitation of the legal rule of the case, but I could be wrong.
Now about this ATC, AVC and AIC thing: A reasonable person with an average command of the English language might object that a cost measure can refer to average costs or incremental costs but not both, but that would be a petty way to criticise the Court of Justice. After all, there is nothing in the Treaties or in the Court's case law that requires it to obey the laws of arithmetic. Joking aside, it is of course easy to see the difficulty that this new concept is intended to resolve. They were trying to separate the variable costs attributable to unaddressed mail from the variable costs that should be attributed to the rest of Post Danmark, particularly the sector of its legal monopoly. But none of that explains what is going on here:
33. In those circumstances, it emerges from the case-file,(...) that for the purpose of estimating what it described as ‘average incremental costs’, the Konkurrencerådet included, among other things, not only those fixed and variable costs attributable solely to the activity of distributing unaddressed mail, but also elements described as ‘common variable costs’, ‘75% of the attributable common costs of logistical capacity’ and ‘25% of non-attributable common costs’.Fortunately it does not look like the CJEU actually endorsed this approach, but if that is the best anyone can do, perhaps we would be better off without quite so many cost definitions in competition law. In the meantime, lawyers will have a field day trying to figure out how bad it is that the Court seems to have overruled Akzo, and the economists will have a field day trying to come up with a better way to work out whether there is any cross-subsidisation going on. And everybody gets to be employed for another day...
UPDATE FEBRUARY 2013: Cf. here for a discussion of the subsequent litigation in Denmark.