Thursday, January 10, 2013
In Laws of Creation: Property Rights in the World of Ideas, Ronald A. Cass and Keith N. Hylton provocatively discuss the relationship between competition policy and intellectual property rights, presenting a bird's eye view of how the antitrust laws and patent law do or should interact. While commentators have seen numerous actual or potential conflicts between IP and antitrust, Cass and Hylton believe that this tension "largely dissolves" when one considers the differences between static and dynamic costs, with antitrust focusing on the static and intellectual property law on the dynamic. This review examines that claim.
Most of Cass and Hylton's discussion of competition policy is related to problems of patent scope -- for example, the fact that a patent excludes, while antitrust generally abhors exclusion that results in lower market wide output. Patents can create structural monopoly (although most do not), and antitrust generally hates monopoly. They then focus on a short list of practices: setting a price, refusal to license, misuse including tying, reverse payment settlements, and collusion in the context of patent pooling.
Setting a price is a power inherent in nearly all property rights, including patent rights, and in any event unilateral price setting is not an antitrust violation in the United States either, not even when the firm is a monopolist. The only exception is predatory pricing, which is rarely proved and which concerns short run below cost prices rather than high ones. So there is no conflict here.
Cass and Hylton note that a simple refusal to license offends neither patent law nor antitrust law. However, they do not discuss more complex scenarios that test the boundaries of that proposition. For example, should there be a difference between acquired patents and internally developed patents? What of the dominant firm that acquires patents from nascent rivals for the purpose of keeping their technologies off the market, as in the Paper Bag case that the authors discuss? (Continental Paper Bag Co. v. Eastern Paper Bag Co., 210 U.S. 405 (1908); compare the discussion in Bohannan and Hovenkamp, Creation Without Restraint: Promoting Liberty and Rivalry in Innovation 295-299 (2011)). And what of the firm that has established a course of dealing with rivals but then pulls the plug after the rivals’ market commitments have been made? United States antitrust law has found liability in such a case (Aspen Skiing), as did the EU in the Microsoft/server case.
Cass and Hylton briefly mention the Supreme Court's Walker Process doctrine (382 U.S. 172 (1965)), stating only its sixties' era version that enforcing a patent obtained by "fraud" can violate the antitrust laws. Through a long and rich history, nearly 1000 federal decisions have discussed the boundaries of Walker Process, which presents a relatively rare situation where antitrust reaches into the inner workings of the patent granting process, particularly in the Federal Circuit's development of the law of inequitable conduct.
I also found the two-paragraph discussion of patent ties, stating mainly that ties should be governed by the rule of reason, to be too thin in relation to the manifold practices and economic effects that are encompassed under that term. Historically, patent ties were condemned by both patent law and antitrust law even if they were not exclusionary, on the theory that the patentee could “leverage” a second monopoly on an unpatented tied good. See, e.g., Carbice, 283 U.S. 27 (1931) (patent law); and International Salt, 332 U.S. 392 (1947) (antitrust law). Cass and Hylton’s insistence on a rule of reason, with which I agree, very likely means that they would limit illegal tying to situations involving market foreclosure. But that still leaves a great deal of unexplored territory, including technological ties (Microsoft and Internet Explorer, Kodak’s Instamatic camera and film cartridge; Lexmark's printers and brand-specific print cartridges) and ties of unique aftermarket parts. They also do not address questions such as whether their rule of reason should be purely structural, addressing mainly market share and foreclosure percentages, or should it look to some other criterion, such as ability to exclude an equally efficient rival? In tying law, the tension between static and dynamic concerns is fairly obvious. If tying is profitable, then permitting it will induce more patenting by increasing the returns to doing so, at least in those situations where profitable tying is possible. Prohibiting it accordingly reduces those returns. In this case the statutes state that tying of goods "whether patented or unpatented" is unlawful, provided that the tie may substantially lessen competition (15 U.S.C. §14). This suggests a Congressional preference for the static concern, which is short run monopoly profits.
Reverse payments settlements, which Cass and Hylton discuss at some length, are a special creature of the Hatch-Waxman Act. Under that provision the first generic drug manufacturer to enter into competition with a pioneer manufacturer receives a limited period of exclusivity vis-a-vis other generics. Once the first generic is identified this creates a Coasean bilateral monopoly in which the joint maximizing outcome for the two parties, pioneer and generic, is typically to share the monopoly profit stream from the pioneer's drug rather than to compete against each other by pitting the generic's output against that of the pioneer. The generic manufacturer in particular can often earn much more by sharing the pioneer's market position than by producing in competition; as a result, its interests are a very poor surrogate for the public's.
For example, suppose manufacturing costs to both parties are 50 cents per unit. The monopoly price is 90 cents per unit. When the generic enters, if the two firms behave competitively the price will drop to 50 cents and they will each earn only a competitive return. By contrast, if they settle via a payment for delayed entry, the two firms will share the 40 cents in monopoly profits for a time, at consumers’ expense. The ironic result is that it is more profitable for the generic to settle than even to win the lawsuit outright, which would make the market competitive. The parties might of course achieve a similar result if the generic produced and the two firms colluded on the product price. The statute does not permit price collusion, however, and as a result it would be per se unlawful and perhaps even a criminal violation of the antitrust laws. So the Hatch-Waxman settlement somewhat resembles the story of two price-fixers who shut down one of their plants and produce the cartel output from the remaining plant.
Cass and Hylton see the settlements as frequently being devices for addressing the risk of a legal outcome that is unfavorable to the pioneer patentee, such as a finding of invalidity. That is the justification for most settlements of infringement cases: the patentee discounts the risk of losing the lawsuit into an agreement that typically includes a license to the infringer to produce at a specified royalty. One significant difference between the two is that this ordinary settlement is an output increasing event, making both patentee and licensee into producers, while a reverse payment settlement presumptively reduces output by preserving production only by the pioneer while raising its costs.
Cass and Hylton argue that patents in these cases are often valid and (about to be) infringed, citing data from Bessen and Meurer's Patent Failure (2008) which concludes that pharmaceutical patents are among the most durable and robust patents. While that is true, most Hatch-Waxman settlements are not on original pioneer molecules. They are typically on "evergreened" extension patents for new uses, new dosages, new forms of delivery, and the like. The failure rate of these patents is much higher, and the incentives to profit from the bilateral monopoly accordingly greater. Indeed, while the invalidity rate of litigated patents is an already-too-high 40%, the invalidity rate of pharmaceutical patents litigated under paragraph IV of the Hatch-Waxman process is nearly double that, 73% (FTC, Generic Drug Entry Prior to Patent Expiration (2002),, available at http://www.ftc.gov/os/2002/07/genericdrugstudy.pdf. In the Watson Pharmaceuticals case on the Supreme Court's docket this term, the drug patent is on a particular gel formulation of a drug that was established and widely available but whose patent had expired. (FTC v. Watson Pharma, Inc., 677 F.3d 1298 (11th Cir. 2012), cert. granted, 2012 WL 4758105 (Dec. 7, 2012). Further, the formulation very likely did not meet patent law's novelty requirement. The drug itself was in the public domain, and gel formulations of drugs have been well known for decades. The delayed payments, which would run to more than $200 million over the course of the settlement agreement, were far larger than the generic could have anticipated by making competitively priced sales.
On collusion, Cass and Hylton focus mainly on patent pools and the differences between substitutes and complements. They conclude, as most of the literature has, that a patent pool of substitutes is more likely to be anticompetitive than a pool of complements. Once again, however, the devil is in the details. In the hypothetical case of a two-patent pool where each patent has a single claim, complements and substitutes may be relatively easy to distinguish. But pools today often include several thousand patents, and each of them has multiple claims. In that case one finds a range of pairwise relationships in nearly every pool, running from complements to substitutes. For example, in the Federal Circuit's Princo decision (616 F.3d 1318 (2010)) the licensed technology included an "analog" and a "digital" method for identifying locations on a recordable optical disc. One would ordinarily think of these methods as complements; that is, a manufacturer would choose one but not both. But in this case the analog method infringed at least one claim in the digital patent. To the extent one needed both patents to produce, the patents operated as both complements and substitutes. Over the century-long history of antitrust challenges to patent pools the complement/substitute distinction has been well known, but it has generally been of little use in identifying specific pools as anticompetitive.
Quite aside from the complements/substitutes issue, the real problem with the Harrow case (Bement v. National Harrow, 186 U.S. 70 (1902)), which Cass and Hylton use as a patent pool model, is not the technology sharing but rather the price fix. The defendants cross licensed their patents for producing spring tooth harrows but also set the product price. That price fixing agreement would be anticompetitive in most circumstances no matter whether the patents in question were substitutes or complements.
For example, suppose A has a patent on technology that enables an electric fan to oscillate up and down as well as side to side. B has a patent on technology that enables the fan to blow air in two directions at once, making oscillation unnecessary. These patents are substitutes, although it is possible that they are complements as well. That is, a fan maker that wanted to increase the directions of air flow would use one or the other, or perhaps both. Suppose A and B are both fan makers and that they cross license the patents to each other and fix the product price at $30, which is double the competitive price. The antitrust problem with this arrangement has nothing to do with the substitute/complement relationship of the patents. Rather, it is a function of the fact that A and B have "assigned" a value to their patents through the price fixing process that is equal to the entire available monopoly markup on the fans themselves. The fact is that many people might not want either of these technologies, or if they did want them they might be willing to pay slightly more than the competitive price, but certainly not the full difference between competitive and monopoly prices. Most patents, whether substitutes or complements, add much less value to a product than the difference between competitive and cartelized output, but the price fix permits the firms to capture that entire difference as the return on their patents.
It also seems clear that the ability to fix the price of patented goods is a dynamic inducement to patenting, because it increases the potential for patentees to obtain the full monopoly returns to their inventions, something that the patent act itself does not guarantee. So here the "dynamic" position (permit the price fix) and the "static" position (prevent it) are clearly in tension, and we need to make a further policy judgment about how to meter one against the other. Congress certainly could have decided this issue of patent scope by either explicitly permitting or explicitly prohibiting product price fixing of patented goods, as it did for tying, but it did neither. Under the Supreme Court's rule in United States v. General Electric, 272 U.S. 476 (1926), the hypothetical fan maker's price fix is probably lawful. Congress has never overruled the GE decision, but the Antitrust Division has consistently opposed it and sought to narrow its reach.
What Cass and Hylton say about patent law's focus on dynamic gains and antitrust law's focus on static gains is indeed true much of the time. But antitrust's common law approach to post-issuance patent restraints properly requires a great deal of focused, case-specific analysis in order to sort out the sheep from the goats. As the pooling illustration above indicates, the most fundamental difference between patent scope and antitrust scope is that patent law defines its scope in terms of property boundaries, with little thought about whether these boundaries create economic monopolies. For example, questions of infringement, claim construction, or even the doctrine of equivalents are addressed without reference to any economic market in which the patentee operates or the amount of market power that a particular outcome might produce. By contrast, post-issuance patent restraints are frequently intended to enable firms to create or prolong economic market power.
Indeed, the fact that patents do not create economic monopolies explains why there is so much room for antitrust law in cases involving patent practices. If patents did create monopoly power then there would be less opportunity to leverage even more monopoly by means of a restraint, at least not in the market in which the patent created the power. Antitrust acquires its role when the patentee seeks to obtain monopoly returns in an anticompetitive manner that the patent itself did not create and that the Patent Act does not authorize.
Cass and Hylton provocatively suggest that the tension between antitrust and patent law dissolves when one steps back far enough. They are also correct that the economic concerns of patent law are mainly dynamic, while the core concerns of antitrust are with static efficiency, principally competitive pricing. However, stepping back from the details impairs our ability to resolve specific disputes, and there is a large variety of them. Most cannot be resolved simply by naming the efficiency interests at stake.