|
COPYRIGHT 2007 Stanford Law School
INTRODUCTION
I. A BRIEF INTELLECTUAL HISTORY OF PATENT AUCTIONS A. Barzel on the Optimal Timing of Innovations B. Kitch on Prospect Theory C. Duffy on Demsetzian Auctions D. Bar-Gill and Parchomovsky on Auctioning Ideas II. THE UNEASY CASE FOR PATENT RACES OVER PATENT AUCTIONS A. The Effects of an Auction System 1. Enablement of expansive patent scope a. The time dimension b. The technology dimension 2. Effects of expansive patent scope a. Effects on invention and development incentives b. Effects on deadweight loss 3. The best case scenario B. Administration of an Auction System 1. Property right delineation 2. Timing 3. Cost C. Additional Objections and Caveats 1. International cooperation 2. Technology suppression 3. Irrational and mistaken bids 4. Small and independent inventors III. THE DESIGN OF PATENT AUCTIONS A. Auctions to Stimulate Development B. Price Auctions 1. Least-patent-term auctions 2. Price auctions IV. MORE MODEST PROPOSALS A. Auctions with Multiple Winners B. Patent Scope Auctions CONCLUSION
INTRODUCTION
The government awards intellectual property rights based on a principle analogous to that of first possession. The author who creates a literary work enjoys copyright protection, (l) the entrepreneur who first uses an indication of origin receives a trademark, (2) and the first inventor of a useful machine or process is entitled to apply for a patent. (3) This approach is not inevitable. Other property rights initially conferred by the government are assigned by administrative flat or through auction. For example, the Federal Communications Commission decides who may use some parts of the broadcast spectrum and auctions fights to other frequencies. (4) Economists, meanwhile, have suggested that auctions be used to limit participation in research tournaments, such as competitions to design the best fighter jet for the government. (5) Despite the similarities between research tournaments and patent races, (6) commentators have paid virtually no attention to the possibility that intellectual property rights might be auctioned. (7) In theory, the government could auction a particular class of intellectual property fights encompassing copyrighted works, trademarks, or inventions yet to be created. (8) The winner of such an auction would then have incentives to create intellectual property within the scope of the right, without needing to worry that competition from third parties might make the investments worthless.
This Article considers whether auctioning patents, instead of (or in addition to) granting them to the first inventors, could accelerate innovation and improve efficiency. The debate will turn on a variety of empirical considerations, many of which will be familiar from the debate on Edmund Kitch's prospect theory of patent law. (9) The prospect theory suggests that patents should generally be granted relatively early, even well before the patents are ready to be commercialized, and should nonetheless cover a broad range of subject matter. (10) This approach reduces the possibility of redundant efforts at technological and commercial improvement and provides a single patentee with the assurance of being able to profit from further research and development. Critics of prospect theory argue that prospects might harm innovation by muting competitive pressures (11) and centralizing decisionmaking on which projects to pursue, (12) but proponents identify countervailing considerations. (13) If patents are granted sufficiently early, they might lessen the inefficiency associated with patent races, (14) as the certainty of being able to capitalize on any successful research might lead an auction winner to invest more and earlier in research and development. (15) Similarly, broader patents reduce inefficiencies associated with "inventing around." (16)
Precisely the same arguments may be made about auctions, which will tend, for better or worse, to centralize in the winner of the patent the power to coordinate invention. The possibility of auctioning intellectual property rights may thus be seen as an extreme version of prospect theory. This Article will not attempt a resolution of the difficult empirical questions about the relative virtues of competition and centralized control in the inventive process, let alone announce a verdict on prospect theory. For two reasons, however, the prospect approach is strengthened when it is realized through a well-run auction rather than through some version of the existing patent system. First, an auction system can allow for prospects that are granted earlier and more broadly than is feasible in the existing patent system. (17) An auction system can thus increase the benefits of a prospect approach in encouraging early invention and avoiding duplicative invention. For example, one criticism of the prospect approach is that it encourages excessive competition in the race to a patent, (18) but an early auction can eliminate this inefficiency. Second, an auction system can be designed to avoid some of the potentially negative consequences of the prospect approach. We will see that auctions can be designed to reduce a patentee's power over price and thus deadweight loss, or to counteract a patentee's incentive to develop only some of the technologies within the scope of a patent.
Even so, the existing patent system probably is superior to an auction system alternative. An auction might work well only if it is well-run, but there are formidable obstacles to efficient governmental execution of patent auctions. Ultimately, this Article seeks only to make uneasy the conclusion that the existing system granting rights to successful inventors necessarily dominates an auction approach. A substantial literature considers whether the patent system is superior to the alternative of a reward system, (19) but the literature ignores that there is a third possible strategy for encouraging inventive activity, the granting of exclusive rights through auctions. This Article will seek to describe the range of ways in which patent auctions might work. Plainly, this analysis is of greater theoretical than practical interest. The auction idea has heuristic value as an idealization of the prospect approach and serves to emphasize more broadly how bidding and racing can be substitute methods for awarding property rights. With this background, the Article seeks its practical payoff by describing much more modest auction proposals.
The more theoretical portion of this Article extends a recent analysis by John Duffy, (20) though this Article differs on some key points. (21) The Duffy article, while not considering the possibility of actual auctions of patent rights, argues that the patent system functions in some respects like an auction, in which a patent is awarded to one of the inventors who agree to start racing earliest and thus implicitly consent to place the invention in the public domain first. The analogy highlights that a patent auction need not award a patent to the bidder who offers to make the highest cash payment to the government. Considering the possibility of explicit auctions enables analysis of several different approaches for selecting the winning bidder. A patent auction, for example, might award the patent to the inventor who commits the most resources to research and development, (22) who accepts the shortest patent term, (23) or who agrees to sell or license the patented invention for the lowest amount of money. (24) A significant potential benefit of patent auctions is that prospective inventors compete to offer as much as possible to the public, rather than solely to earn as much as possible through monopoly pricing from the public. (25) One reason that patent auctions may have received so little attention from scholars is that pure cash auctions appear to be little more than an inefficient tool for generating government revenue. Patent auctions may become more attractive, however, once alternative forms for auction bids are considered, and the prospect theory in general becomes more attractive as system design can better overcome some of its limitations.
A significant obstacle, however, remains: no single form of auction bid can capture the social welfare that a prospective patentee expects to provide. The more the winning bidder offers in the form of a shorter patent term or a reduced price, for example, the less incentive the patentee will have to develop inventions within the patent scope. At the limit, anyone could offer a zero patent term or zero price and then do nothing to develop the patent. With a shortest patent term or lowest price auction, then, the winning bidder would need to be required also to make some commitment to developing the patented product. This might come in the form of a payment to the government, or payment into a fund to be used only for future development costs. More complex auction designs might require the government to choose the bidder who offers the public the best overall package, considering variables such as patent term, price, payment to the government, and commitment of money to a development fund. (26) There is no obvious formula for the government to use in comparing bid packages, however. (27) Whether the government selects how it will weigh different components of a bid in advance or evaluates bid packages case-by-case, it risks placing excessive weight on some goals of patent law (such as encouraging development of patented inventions) at the expense of others (such as limiting deadweight loss). This trade-off between static and dynamic efficiency is familiar to patent scholarship, (28) though in the case of auctions, the question is whether there will be enough incentives to invent after a patent is issued.
The analysis proceeds as follows: Part I reviews past proposals that, in passing, have noted the possibility of auctioning patent rights, as well as Duffy's proposal and another proposal integrating an auction mechanism into the patent system. A principal purpose of this Part is to identify assumptions, the relaxation of which might strengthen the case for an auction system. Part II then evaluates patent auctions, focusing on cash auctions for blocking rights in particular fields. The principal benefit of a patent auction is that it can allow for broader patent rights than are feasible in the existing patent system. By reducing inefficient duplication, a patent auction can increase a property right holder's incentives to engage in inventive activity. Part III considers the design of patent auctions. It shows that incentives to engage in inventive activity also can be promoted by an auction design specifically geared to that goal, and that patent auctions also might be designed to minimize concerns about deadweight loss. Finally, Part IV considers alternative proposals for implementing patent auctions and for overcoming the difficulties associated with governmental selection of rights to auction. One proposal would be to have auctions that reduce the number of patent racers in a technological field, but do not give a single party exclusive fights. The second proposal is for patent scope auctions, in which a patentee could call for an auction of expanded patent scope, but could win the auction only by substantially outbidding competitors. With a sufficiently high markup demanded, a patentee will have incentives to call for patent scope auctions only in situations in which concerns about inefficient duplication of resources are particularly salient, thus making patent auctions socially beneficial.
In the end, the most important practical lesson from the theoretical explanation of patent auctions is that the analysis reinforces the classical prospect theory position that there might be benefits from patents with considerably broader scope than is now possible with the existing patent system. The literature on patent scope considers only relatively marginal policy choices, such as those presented by the doctrine of equivalents. (29) Administrative convenience demands that patents hew closely to what the inventor has already achieved, but theory suggests that patent scope should depend on analysis of alternative inventions targeting the same consumer need, in particular the additional benefits that these inventions would provide and the cost of developing them. Patent scope auctions provide one mechanism for selectively broadening patent scope in cases in which this analysis suggests that broad scope would be more efficient. Admittedly, other hypothetical approaches might be more politically palatable because the patent system, indeed the legal system generally, does not currently integrate auction mechanisms. Perhaps the ultimate goal of this Article, however, is to make the possibility of integrating auction and other mechanisms into the patent system less alien and to open a conversation that might lead to genuinely modest proposals for using market mechanisms to incorporate information not easily accessible to governmental officials in a patent office.
I. A BRIEF INTELLECTUAL HISTORY OF PATENT AUCTIONS
This Part will provide an intellectual history of patent auctions. This history will be brief for two reasons. First, no one has ever considered in more than a footnote whether it might be useful to prospectively auction intellectual property rights to all innovations in a field rather than grant rights to creators for specific innovations once the technological possibility of those innovations becomes apparent. Those who casually imagined auctions presumably envisioned only auctions for particular innovations, rather than auctions for technological fields. Because it is difficult to auction specific innovations before they are developed, no one has viewed the possibility of an auction as anything other than a heuristic device. There is thus little history to recount. Second, my purpose is analytical rather than historical. Instead of describing each of the four articles that constitute the intellectual history, I will offer only the barest summary and then identify a particular explicit or implicit assumption about the inventive process that underlies each article's position on patent auctions. Understanding these assumptions requires a recognition that because patents are limited in both time and technology, inventors can only partly internalize the benefits of research and development and may thus have inadequate incentives to research and develop patented products. Patent auctions may be able to provide for broader property rights and more robust innovation incentives, though potentially at the cost of greater deadweight loss.
A. Barzel on the Optimal Timing of Innovations
In Optimal Timing of Innovations, (30) Yoram Barzel argued that sometimes innovations may be produced inefficiently early. Assume that there is some cost to a particular innovation. Some private party will decide to produce the innovation once the net present value of the innovation just exceeds zero, (31) i.e. once the rents that the private party can obtain from the innovation will be just enough to offset the cost of producing it. The inventor will not wait until later than that because if the inventor does so, somebody else will have an economic incentive to produce the innovation first. If the inventor could prevent others from producing the innovation in the meantime, however, the inventor would have an incentive to wait. Assuming that the cost of the innovation is constant and the demand for it will grow over time, (32) the inventor would be better off delaying invention, and earning a normal rate of return on the funds that would be used for the innovation in the interim. If the inventor is able to capture privately the full social benefits of the invention, then invention will be too early from a social as well as private perspective. Intuitively, if there is some moment of invention that would produce a positive rent for an inventor, then inventors will dissipate that rent by inventing earlier because no inventor owns a property right in determining when to invent.
In a footnote, Barzel suggested that this inefficiency could be overcome by granting monopoly rights on innovations, for example through an auction. (33) The conclusion that invention would be too early from a social perspective in the absence of such a grant, however, depended on an assumption that Barzel made explicitly, (34) but acknowledged was "quite strong" (35): that the inventor is able to appropriate the full social benefits of the invention. When society receives benefits from an invention that do not accrue to the inventor, invention might occur too late, and so, Barzel recognized, in the absence of property rights, the net effect of these opposing forces was indeterminate. (36) Barzel's argument thus identifies a countervailing consideration to what he acknowledged as the conventional wisdom that there generally are insufficient incentives to innovate. (37) The economic literature on intellectual property, however, suggests that this conventional wisdom is powerful, and that inventive activity generally produces significant spillovers that patentees cannot capture. (38) That suggests that when a patent leads to delay in completion of inventive activity, that delay should count as a social cost, rather than as a social benefit.
This might at first appear to undermine Barzel's footnote suggestion for granting rights through auctions, but in fact it might suggest a benefit of auctions. In general, spillovers and other forms of positive externalities exist when property rights are ill-defined. (39) Until a patent is issued, no one has a property right in the underlying invention, and inventors will be able to capture only a portion of the value of their efforts, for example if they are fortunate enough to win the patent race. As a result, the total level of effort will be suboptimal. Providing property rights at an earlier stage, for example through an auction, could allow inventors to capture a greater proportion of the social benefits that their inventive activity creates, leading to greater investment in research and development. Certainly this would be true if the auctioned property right is granted sufficiently early and with sufficiently broad scope to eliminate spillovers.
Auctions thus might be justifiable, but the social calculus may be the exact opposite of what Barzel's analysis would appear to suggest. Auctions may be justified because they make delay less rather than more likely. Auctions will be valuable to the extent that they reduce spillovers and thus increase the overall level of inventive activity within the scope of the property right, but there is a danger that because some spillovers will still exist, property rights might lead to inefficient delays in inventive activity. The case for auctions will thus be strongest the greater the extent to which an auction helps consolidate property rights that would remain fragmented in a traditional patent system. The patent system can leave property rights fragmented across two dimensions: time, because the patent term is limited; and technology space, because a particular patent in the existing system will ordinarily protect only one particular means to solving a given problem, yet might provide information helpful to both inventors and investors in a technology area. Parts I.B and I.C will use other works on patent auctions to consider these dimensions, respectively, before Part I.D turns to a concern that broad property rights inevitably would create increases in consumer prices and thus deadweight loss.
B. Kitch on Prospect Theory
Although Barzel's footnote proposal for auctioning innovation rights was undeveloped, Edmund Kitch seized upon it in announcing his seminal prospect theory of patent law, (40) acknowledging that his "ideas first crystallized in response to Barzel's essay." (41) Barzel, Kitch wrote, recognized that a claim system could prevent inefficiently rapid depletion of the common pool of technological information. (42) "What Barzel did not realize," Kitch wrote, "is that a patent system can be such a claim system and, indeed, that it is a more sensible system than an auction system would be." (43) Kitch backed up the claim that the patent system would be "more sensible" with only a brief single footnote: "Because the patent creates private incentives for the identification and definition of claims and puts the claim identification and the claim development process in the same hands." (44) This must be among the most perceptive phrases in law and economics, though we will see that the comparison between an auction system and a patent prospect system is more complicated than Kitch allowed.
The patent system, Kitch argued, not only provides useful ex ante incentives, but also provides benefits ex post. Among other virtues, the system "puts the patent owner in a position to coordinate the search for technological and market enhancement of the patent's value," "lowers the cost for the owner of technological information of contracting with other firms possessing complementary information and resources," and "reduc[es] the amount of duplicative investment in innovation." (45) Kitch's article remains arguably the most important and the most controversial in patent theory. A principal criticism is that the possibility of prospect patents will simply shift duplicative investment to the pre-patent stage. (46) Auctions provide a potential antidote because there is no sense in racing to invent if exclusive fights are to be (or have been) distributed through auction. Auctions also may be responsive to a second criticism, that even after patents are granted inventors will compete to improve patents, (47) if the property rights granted through auction are sufficiently broad. Thus, while Kitch's identification of ex post benefits to property rights advance the case for auctions, the limits of a patent prospect's approach to providing those benefits might strengthen the case for auctions still more.
Despite the great deal of attention that Kitch's analysis has received, there is one assumption in his analysis that commentators appear to have ignored, and relaxing this assumption can point both to an unappreciated benefit of the auction approach and to a subtle risk. Kitch acknowledged that "[o]peration of the prospect function requires that the owner have most of the present value of the invention for the investment period." (48) Kitch reasoned that given a five-year investment cycle, the then-seventeen-year term should "give the owner a large part of the present value." (49) But if the inventor does not begin development right away, less of the patent term would be left. Indeed, perhaps so little of the patent term will be remaining that the inventor might well not develop the invention at all.
This might seem to be an implausible concern, for why would the inventor not begin development right away? Barzel's analysis, however, provides a partial answer: it might be in the interest of the patentee, even if not in the social interest, to delay inventive activity. Even so, why would an inventor have gone through the trouble to acquire a patent, only to later abandon it? The simple answer is that the receipt of a patent provides the patentee with an option to develop and commercialize the patent, but options often turn out to be not worth exercising. (50) Kitch might implicitly be assuming that the benefits and costs of commercializing a patent are predictable in advance, but market conditions might change after a patent is issued. It will thus sometimes make sense to patent an invention that seems likely to be worth commercializing only in the last years of a patent term, or even that seems unlikely to be worth commercializing at all in the patent term, if there is some chance that market conditions will in fact make commercialization worthwhile.
The danger of nondevelopment thus accentuates that property rights in inventions are fragmented over time, with the patentee owning the right only for a period of time and the public owning the right in common thereafter. Conceivably, an auction approach could reduce the risk of patent nondevelopment by allowing for longer patent terms. Longer terms would reduce the extent of spillovers, providing inventors greater incentive to invent and to perform research and development within the scope of the patent right. A trade-off thus inheres in adoption of auctions: if auctions provide long patent terms, incentives to invest in research and development increase, but so too does the social cost of monopoly pricing, deadweight loss. (51) Because auctions can be structured in different ways, (52) this presents not only a trade-off inherent in switching to an auction system from a patent system, but also a trade-off in auction design. If the auctioned patent term is sufficiently short, deadweight loss will fall, but the remaining spillovers might be so great that the auctioned right will increase the extent of inefficient delay and lower social welfare.
C. Duffy on Demsetzian Auctions
Though he too does not consider the possibility that uncertainty about the benefits and costs of developing the invention might end up resulting in effective patent terms so short that inventors fail to develop inventions, Professor Duffy, in Rethinking the Prospect Theory of Patents, (53) does recognize the centrality of patent expiration to an assessment of the prospect system. Indeed, Duffy argues that a consideration of patent expiration answers a central criticism of the prospect theory (54)--that prospect patents cannot eliminate inefficient duplication of effort because they instead will tend only to intensify races to obtain patents in the first place. (55) If patents produce positive rents, those rents will be dissipated by some form of competition. Duffy's observation is that competition in invention is not the only other means of dissipating rents. Rents also can be dissipated by earlier patenting. Such early patenting, Duffy maintains, is necessarily efficient because it will result in the invention being placed in the public domain sooner rather than later. (56)
Duffy analogizes the patent system to a Demsetzian auction. (57) Harold Demsetz argued that instead of regulating the price that a utility natural monopoly could charge, the government could simply hold an auction for the right to serve as the monopolist, with the winner offering the best package of price and quality. (58) The competition among potential utilities would result in the winning bidder offering service relatively close to marginal cost, thus reducing deadweight loss. (59) Similarly, potential patentees compete by offering to place the invention in the public domain sooner, thus also reducing deadweight loss. Because the prospect system by definition grants patents at a relatively early stage, potential patentees will begin racing early enough so that profits during the effective patent term are just enough to cover the costs of racing. (60) In effect, the patent system is a shortest-patent-term auction, with the patent awarded to one of the inventors who are willing to attempt the invention first. (61) The patent system differs from a pure shortest-patent-term auction only because the patent system does require some level of accomplishment to obtain a patent prospect, and thus there might be a race to determine who accomplishes the invention and thus receives the patent.
Duffy points out that because the issuance of a prospect patent ends the race, it will reduce duplicative efforts. The result is that the race will start earlier than it would have otherwise. Because inventors can always engage in other activities--other patent races or other work altogether--they will not begin any given patent race until they expect to profit from entering. Entry should become more attractive over time, for example because the cost of producing a particular invention falls with general technological development or because society's demand for the invention rises. But each inventor considering when to enter must contemplate not only the cost of completing the invention and the benefit of the patent, but also the possible cost of losing the patent race. This cost will be less if the prospect is granted early, and so a prospect approach will lead inventors to start racing earlier than they would have if they expected that all racers would incur costs after the time that a prospect patent is issued. (62) The combination of this earlier racing and the earlier award of the patent means that the invention will be committed to the public domain earlier as well. This point poses a significant problem for commentators who believe that the competition the patent system creates is generally efficiency-enhancing. (63) Competition might well provide extra motivation to participants during the race, but the race might start later, the longer potential inventors are allowed to remain in the race after it begins.
An assumption of Duffy's model is that once one racer obtains a patent, the patent race ends. The assumption might be problematic, however, because different racers might be working on different means to the same goal, and the prospect patent might not be broad enough to cover all of...
Read the full article for free courtesy of your local library.
|
Bookmark this article
Print this article
Link to this article
Email this article
Digg It!
Add to del.icio.us
RSS
