Post by Erik Hovenkamp.
When will private settlements advance the normative objectives espoused by the relevant area of law, as opposed to straying away from them? One sensible metric is to assess the similarity between (1) the positions the parties maintain post-settlement; and (2) the positions the parties would have maintained had they litigated (accounting for possible post-trial contracting). Given that the outcome of litigation can rarely be predicted with certainty, we can interpret the second prong as the expected result of litigation—as weighted by the relative likelihoods the parties assign to the different possible outcomes. To illustrate, suppose my doctor’s treatment of me causes me to suffer $1000 in harm, and there is a 50% chance that he will be held liable for negligence. Then we will settle for about $500, for we cannot mutually agree on anything else. This is precisely the expected result of litigation, and hence the settlement is representative of how the courts would apply the law to our dispute.
If we ignore things that might skew negotiations in one party’s favor (e.g. asymmetric ability to cover litigation costs), the above result is essentially a corollary of the Coase theorem. If the court’s judgment won’t affect the final allocation of rights, then the parties just bargain over the distribution of welfare. Neither party will accept less than it expects to get through litigation, which is a function of the legal standards that would be applied by a court. Thus, under these circumstances, private settlements will “emulate the law.”
But as I have emphasized previously, one important feature of the IP-antitrust interface is that the parties to a dispute are often prohibited from contracting out of a judgment that fails to maximize their (joint) profits. Competitors in a patent dispute cannot “undo” a holding that the patent is invalid or noninfringed in order to revive the patent’s exclusionary power; they are obligated to stick with open competition. The result is that rival firms often have a shared interest in settling an IP dispute on terms that deviate substantially from their expectations about litigation. Specifically, if they expect litigation to produce a relatively competitive result (e.g. because the patent is likely invalid), they prefer to evade that outcome by settling on terms that restrain competition substantially, resulting in higher prices and larger profits. The result is that these settlements do a very bad job of promoting IP and antitrust policy objectives; they do not emulate the law.
In my recent paper with Jorge Lemus, Proportional Restraints and the Patent System, we propose a comprehensive standard for evaluating patent settlements under the antitrust laws, along with a set of economic tools for administering it practicably. In what follows, I provide an overview of the underlying problem and our proposed solution.
On one hand, if the patent presents a potential entry barrier, then a settlement with a rival often must restrain competition to some degree if it is to be mutually-acceptable. That is because litigation will substantially curtail competition (via injunctive relief) with some probability, and that would provide much larger total profits than unfettered competition. Hence for a settlement to provide the same expected value to the patentee, the rival’s ability to compete must be restrained to some extent. On the other hand, the firms can jointly benefit from restraining competition more than necessary–beyond the level needed to match the joint profits expected to accrue from litigation.
There are many ways the firms might go about doing this; the antitrust case law is rife with examples, and we discuss many in our paper. A simple example involves the patentee offering the rival a large “reverse payment” (usually cash) in exchange for the rival’s agreement to delay its market entry for some number of years, after which it can enter and compete freely. This arrangement is known as “pay-for-delay,” and is most commonly associated with pharmaceutical markets. The Supreme Court’s 2013 Actavis decision held, in effect, that such agreements are illegal whenever they involve a nontrivial payment that lacks any reasonable procompetitive explanation. The opinion was a clear step forward. But it is also narrow. Pay-for-delay is just one problematic tactic that firms might employ, and it may not be the firms’ first choice outside the pharmaceutical context. A broader policy is needed if antitrust is to address the problem comprehensively.
We propose that antitrust’s underlying policy objective should be to promote settlement proportionality, which means that the settlement’s competitive effects are commensurate with the parties’ expectations about litigation. Of course, there are clear reasons to worry about the administrability of an antitrust standard that depends on expectations about hypothetical patent litigation. But we show that one can in fact discern a settlement’s proportionality by simply looking at what kinds of terms the firms negotiated. That is, even without knowing the firms’ true expectations about litigation, we can identify efforts to evade them.
More concretely, certain kinds of restraints on competition (and certain ancillary provisions, like reverse payments) always result in disproportionately restrictive settlements, because the firms cannot agree on anything less. By contrast, other kinds of restraints keep the firms honest; they make it impossible to mutually agree on terms that materially deviate from the anticipated result of litigation. As such, our proposal would have antitrust law focus less on ascertaining the extent to which competition is restrained, and more on evaluating the manner in which it is restrained.
Under our proposal, we focus on two distinct classes of settlement provisions. First, we look at the restraint imposed by the patentee on its rival(s). A restraint is anything that inhibits the rival’s ability to compete during the remainder of the patent term. For example, in pay-for-delay, the relevant restraint is the delay of the rival’s entry. But there are countless other examples, such as cost-raising royalties, price or output restrictions, territorial restraints, or field-of-use limitations, to name a few. Second, we look for certain problematic “side-deals,” which is just a catch-all for things that do not inherently restrain the patentee’s rival, but which skew bargaining by making the rival less “uncomfortable” being restrained. The clearest example of this is a reverse payment; it does not inherently affect competition, but it persuades the patentee’s rival to stay out of the market for a disproportionately long period of time.
We show that different kinds of restraints vary systematically in terms of how proportional the resulting settlements are. The details are somewhat technical, but the intuition is that restraints differ in terms of how they affect the distribution of oligopoly profits. Some restraints always give a comparatively large share of these profits to the restrained rivals, so that it is systematically much less “painful” to the rivals being restrained (and simultaneously less lucrative for the patentee), and in these cases mutual agreement requires that competition be restrained excessively. In fact there are some kinds of restraints that the patentee’s rivals actually like, meaning they’d rather be restrained a little than not at all. For example, if the restraint is a cap on the rivals’ sales, then it can make them act as if they are colluding with each other. We ultimately find that price and output restraints are generally the worst offenders in terms of proportionality.
One interesting example is becoming increasingly relevant in the aftermath of the Actavis decision. In the majority opinion, the Court remarked that competitors resolving a patent dispute can still enter into delayed entry settlements, so long as they don’t include a reverse payment. This settlement format, which we call “pure delay,” is becoming more and more common in pharmaceutical markets. So is pure delay good or bad at producing proportional settlements? In fact, it is one of a select few restraints that we designate as “perfectly proportional,” which is our highest normative benchmark. This means that, ignoring bargaining over avoided litigation costs, the firms can never agree on anything other than the exactly proportional result—there is no wiggle room whatsoever. (However, it is worth noting that some well-known flaws in the Hatch-Waxman Act will undermine this proportionality in the pharmaceutical context; Jorge and I discuss that problem in a separate paper.)
There are at least three justifications for our proposed standard. First, because a proportional settlement provides the same profits the firms expect to get as a result of litigation (with the added benefit of avoiding litigation costs), it is mutually-acceptable, and therefore sufficient to resolve the dispute. And by extension a settlement that restrains competition disproportionately is unnecessarily restrictive. As such, our proposal is consistent with antitrust’s broader treatment of horizontal restraints (those applied in contracts between rivals), which is that they are permissible only to the extent reasonably necessary to effect some cognizable efficiency.
Second, focusing on the validity prong of the dispute, our proposal helps to align patentees’ rewards with the significance of their improvements over the prior art: all else equal, a patent on a more innovative technology is more likely to be valid; this implies the expected result of litigation—and hence a proportional settlement—is commensurately more profitable.
Third, our proposed approach is much more administrable than some alternative approaches we might adopt. For example, in principle we could employ a “screening” process under which the antitrust agencies estimate a potential settlement’s proportionality empirically, with only those clearing a certain benchmark being approved; this is analogous to merger review. But this approach would be very difficult and expensive, which would chill enforcement. Further, courts would likely be uncomfortable relying on estimates about how hypothetical patent litigation would have played out. Our proposal avoids this by merely focusing on how different kinds of provisions distort bargaining outcomes for the worse.