Chillin'Competition

We had this morning at the GCLC a half day conference on pay for delay arrangements in the pharmaceutical sector.

This was a great event, which triggered the following thoughts.

On the law, the problem is not the conduct in itself, i.e. the payment arrangement (also referred to as a “settlement” agreement). Any agreement which consists in paying to seclude a competitor can be unlawful. The FTC apparently talked of a “classic violation” in old documents. We see such agreements all the time in the area of vertical restraints, when firms pay distributors in exchange for exclusivity so as to keep rivals off the market.

The problem lies elsewhere, in the proof of the anticompetitive impact of the agreement, actual or potential (the second component of an infringement). Economists talk about a credible theory of competitive harm.

Now, a settlement payment from originator to generic can only be deemed to restrict competitive entry if the judge would invalidate the patent. In other words, a settlement payment can only be anticompetitive if the patent is invalid. Otherwise (if the patent is valid), competitive entry could just NOT occur, and the payment question is irrelevant.

If I understand correctly, the Commission apparently assumes invalidation. In other words, it assumes that judges would invalidate litigated patents. This stance on what the judge is poised to do is similar to that found in the abuse cases involving Standard Essential Patents, where the Commission seems also to assume that the judge will grant injunctions.

My take on this: this is a meritorious assumption to make if only because patents are deemed presumptively valid once granted. And this weak assumption is not good practice when one has to formulate new policy. As J. Wright, FTC Commissioner, explains beautifully in a recent speech: the formulation of new theories of harm should be based on empiricism, not on guesswork.

Now, the one funny thing here is that to overcome what I may call the weak assumption problem (and to make the case for enforcement), the Commission relies on classic Chicago school reasoning (or if you prefer on basic rational incentives theory). The idea is as follows: an originator cannot rationally be paying a generic firm to stay out of the market if its patent is valid. The sole rational explanation is that the patent is invalid.

This line of reasoning has some teeth in conventional EU competition law. The AKZO “sacrifice test” in predation cases is one example of this: a dominant firm cannot be rationally pricing below costs. The sole rational explanation is that it is seeking to predate.

I believe, however, that one should err on the side of caution when it comes to crafting such “rules of inference”. Indeed, inferential equations of the AKZO kind are not mundane in EU competition law. They are exceptional, based on rich doctrinal debate as the AKZO ruling shows (it transposes the AREEDA and TURNER test) and on years of judicial precedents. Moreover, the problem is that there are many other possible explanations to the originator payment (irrationality, aversion to risk, etc.), including a whole raft of behavioral economics reasons.

So for patent settlement cases, I would actually apply a different type of “rule of inference”, such as the one applied in Woodpulp or CISAC, whereby the Commission can only find infringement if it proves the conduct has no other purpose than the restriction of competition, and thus must dispel all alternative explanations.

Funnily enough, most of the discussion on patent settlements has not touched upon this issue, and has instead focused on the “object-effect” distinction. This debate is, however, an uneasy one, given the lax state of the case-law. Maybe discussing about patent settlements under a “rules of inference” angle could help reach new ground.