EU Technology Transfer Draft Guidelines: Economic Analysis and Suggestions for Revisions Carl Shapiro 25 November 2003 I. Introduction and Qualifications I am Carl Shapiro, the Transamerica Professor of Business Strategy at the Haas School of Business at the University of California at Berkeley where I have taught since 1990. 1 I also am Director of the Institute of Business and Economic Research at U.C. Berkeley. I have served as the Editor of the Journal of Economic Perspectives, a leading economics journal published by the American Economic Association. I am also a Senior Consultant with Charles River Associates, an economics consulting firm. I am an economist who has been studying antitrust, innovation, and competitive strategy for roughly twenty years. During the 1980s, I published a series of well-known articles with Michael L. Katz on the licensing of intellectual property. 2 I am co-author, with Hal R. Varian, of Information Rules: A Strategic Guide to the Network Economy (Harvard Business School Press, 1999). Recently, much of my work has focused on intellectual property issues and competition policy. 3 Most relevant for the Commission s draft Technology Transfer Block I was asked by the Intel Corporation to communicate to the Commission my views on technology transfer agreements and the likely impact of the Commission s draft regulations and Guidelines. This statement is the result of that request. Many of the economic points made here regarding licensing restrictions can be found in my 2002 report to the OECD, which is available at http://www.olis.oecd.org/olis/2002doc.nsf/linkto/dsti-doc(2002)11. 1 My curriculum vitae is available at http://faculty.haas.berkeley.edu/shapiro. 2 On the Licensing of Innovations, Rand Journal of Economics, Winter 1985, How to License Intangible Property, Quarterly Journal of Economics, August 1986, and R&D Rivalry with Licensing or Imitation, American Economic Review, June 1987. 3 See, for example, Antitrust Issues in the Licensing of Intellectual Property: The Nine No-No s Meet the Nineties, with Richard J. Gilbert, Brookings Papers on Economics: Microeconomics, 1997, and Carl Shapiro, Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 1
Exemption Regulation ( Draft TTBER ) and accompanying Guidelines ( Draft Guidelines ), which were published 25 September 2003, is my work on the role of patents, licensing, and cross-licensing in the presence of a patent thicket, i.e., in industries where a great many patents are being issued and a single product can potentially infringe on many patents. See especially my 2001 paper, Navigating the Patent Thicket: Cross Licenses, Patent Pools and Standard Setting. 4 I am writing to provide comments on the Commission s Draft TTBER and Draft Guidelines. While there are many parts of the Draft Guidelines that I applaud, I respectfully submit that some portions are likely to be counterproductive and deter or impede technology licensing. In particular, I am greatly concerned that a number of provisions that the Draft Guidelines characterize as hardcore restrictions in fact frequently serve legitimate business goals and are necessary as part of pro-competitive licensing agreements. I focus my attention on three provisions that the Draft Guidelines treat as hardcore in agreements between competitors: (1) field-of-use provisions that are narrower than product markets; (2) reciprocal field-of-use provisions in cross licenses; and (3) running royalties in cross licenses. I also discuss the treatment of agreements that license future patents. Before turning to these four categories of provisions, however, I first identify several problems associated with the approach to analyzing licensing restrictions stated in 14b of the Draft Guidelines. II. General Principles A. Innovation and Dissemination Both Matter For nearly twenty years, my work has explored the pro-competitive aspects of licensing, while recognizing that some licensing agreements can harm competition. I am pleased to see that the Commission recognizes the important role played by licensing in encouraging innovation: Competition Policy and Innovation, Prepared for the Directorate for Scientific, Technology, and Industry, OECD, STI Working Paper No. 2002/11, April 2002, available at http://www.olis.oecd.org/olis/2002doc.nsf/linkto/dstidoc(2002)11. 4 This paper appeared in Innovation Policy and the Economy, Adam Jaffe, Joshua Lerner, and Scott Stern, eds., National Bureau of Economics, 2001 and is available at http://faculty.haas.berkeley.edu/shapiro/thicket.pdf. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 2
Licence agreements may promote innovation by allowing innovators to earn returns to cover at least part of their research and development costs. (Draft Guidelines, 15) In order the maintain the incentive to innovate, the innovator must therefore not be unduly restricted in the exploitation of intellectual property rights that turn out to be valuable. (Draft Guidelines, 8) Of course, the key issue is what unduly restricted means in practice; much of what I have to say below addresses this issue. As the Commission establishes rules for technology transfer and licensing agreements, in addition to the short-term and medium-term effects of various rules on the diffusion of technology, the Commission should consider the long-term effects of those rules on incentives to innovate. More restrictive rules regarding the licensing agreements into which a patent holder may enter will tend to reduce the return to obtaining patents, and thus, often, to innovation itself. While this is not the end of the story, effects on innovation (and not just dissemination) are important. I will not keep repeating this point, but it lurks behind the entire discussion of rules governing technology transfer and patent licensing. B. Proper Benchmark for Evaluating Effects of a Licensing Agreement The Draft Guidelines offer two benchmarks by which the Commission will evaluate provisions in technology transfer agreements. In 14a, the Draft Guidelines ask: Does the agreement restrict actual or potential competition that would have existed had no license been granted? As my published work makes clear, I strongly support this benchmark. For example, I wrote in my OECD report ( p. 18): On the one hand there are restrictions that simply limit the extent of the grant to use the licensor s intellectual property. Such restrictions will not typically prevent competition that would have taken place in the absence of the license. 5 This approach protects consumers and competition from agreements that abuse intellectual property rights to reduce competition while incorporating 5 More recently, I explain in detail how this same benchmark can be used to evaluate patent settlement agreements between rivals. See Antitrust Limits to Patent Settlements, Rand Journal of Economics, Summer 2003, and Antitrust Analysis of Patent Settlements Between Rivals, Antitrust Magazine, Summer 2003. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 3
the principle that an innovator is not obliged to create competition using its own patented technology. However, in 14b, the Draft Guidelines pose an alternative question: Does the agreement restrict competition that would have existed in the absence of its alleged restriction(s) on competition? This question relates to the issue of whether or not the restriction is objectively necessary for the conclusion of the agreement. The Draft Guidelines go on to make it clear that the Commission would be skeptical of claims by the parties to an agreement that any given provision is objectively necessary. The question is not whether the parties in their particular situation would have accepted to conclude a less restrictive agreement, but whether given the nature of the agreement and the characteristics of the market a less restrictive agreement would have been concluded by undertakings in a similar setting. Claims that in the absence of a restriction the supplier would have resorted to vertical integration are not sufficient. I am very concerned that, if adopted and used regularly without substantial limitations in the context of technology licensing, this approach would discourage innovation and impede a great deal of procompetitive licensing. Consider the following simple but illustrative example. 6 Alpha Company has obtained a patent on a process innovation that significantly lowers production costs for both widgets and gadgets. Alpha has a leading position in widgets, but only a moderate share of gadget sales. The reverse is true of Beta Company. Alpha seeks to license Beta for gadget production, but not for widget production. In other words, Alpha offers Beta a license with a field-of-use provision so that Beta only obtains the right to use the patented invention for gadget production. 7 As a profit- 6 This example reveals some major problems with the proposed 14b approach generally. The example happens to involve a field-of-use provision for illustrative purposes. As I discuss below, the branding of certain field-of-use provisions as hardcore violations would cause additional problems, above and beyond the general point about 14b made here. 7 Nothing in the economic analysis of this example turns on whether widgets and gadgets are in the same relevant antitrust market. One may think of four cases, distinguished by the cross-elasticity of demand between widgets and gadgets: (a) widgets and gadgets are close substitutes in the same market (the cross-elasticity is negative and large in absolute value); (b) widgets and gadgets are distant substitutes in separate markets (the cross-elasticity is negative but small in absolute value); (c) demand for widgets and gadgets are independent (the cross-elasticity of demand is zero, and widgets and gadgets are in separate and unrelated markets); and (d) widgets and gadgets and complements (the cross-elasticity of demand is positive, and widgets and gadgets are in separate but related markets). The analysis Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 4
maximizing firm, Alpha negotiates a royalty rate with Beta that reflects (i) the cost savings available to Beta from using Alpha s innovation, (ii) bargaining between Alpha and Beta over sharing the rents associated with those cost savings, taking into account (iii) the opportunity cost to Alpha of licensing Beta rather than producing gadgets itself, and (iv) any adverse impact on Alpha s own gadget and/or widget production as a result of Beta having lower costs. Does this license enhance or restrict competition? As I read the Draft Guidelines, the Commission would be concerned that this arrangement limits potentially increased competition between Beta widgets and Alpha widgets. And the Commission would view this competition as entirely feasible to achieve if it concludes that the field-of-use provision is objectively unnecessary because Alpha would have licensed its patent to Beta even in the absence of that provision. In my opinion, these concerns are not justified, at least as regards field-of-use provisions. To begin with, by allowing Alpha to follow the hybrid strategy of exploiting its invention internally for widgets while obtaining licensing revenues for gadgets, Alpha s return to innovation is greater than would be possible without the licensing agreement, so the agreement clearly promotes innovation. Furthermore, it should be clear that this license, including its field-of-use provision, is pro-competitive in the sense that it allows Alpha s patented technology to be used by the leading firm in gadget production (Beta) and thereby enhances competition in gadgets. For precisely this reason, Alpha and Beta could sign such a license confident that the field-of-use provision would be judged pro-competitive under the 14a benchmark. 8 However, under the 14b approach, the Commission would seek to determine whether the field-of-use provision was objectively necessary for the conclusion of the agreement. This example, while simple, is sufficient to reveal several major problems systemic to the 14b approach in the context of patent licensing. of licensing incentives and royalties is easiest in case (c) when there is no cross-elasticity of demand between widgets and gadgets, but my basic points here apply as well in the other three cases. 8 Nothing in the terms of the license would cause Beta to cease competing with Alpha in widgets, or Alpha to cease competing with Beta in gadgets, i.e., there is no inhibition on pre-existing competition. Of course, if Alpha and Beta allocated widget sales to Alpha and gadget sales to Beta, such a market division agreement would harm competition under the 14a approach. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 5
First and foremost, the 14b approach tends to impede diffusion of patented technology and lead to higher royalty rates. Clearly, both of these outcomes are antithetical to the goals of competition policy. How do these undesirable consequences flow from adopting the 14b approach to field-of-use provisions? In my example, which is illustrative of the general economic forces at work, economic principles tell us that Alpha will set a higher royalty rate in gadgets if not permitted to limit Beta s use of Alpha s patented technology to the gadget business. After all, Alpha s preferred strategy was not to license Beta in the widget business, so Alpha s profit-maximizing uniform royalty (across gadgets and widgets) is higher than its profitmaximizing royalty in gadgets alone. 9 This is especially unfortunate as Alpha s higher royalty rate will apply to gadgets, where Beta is the leading firm, so the higher royalty burden will apply to a relatively large volume of units and could well exacerbate Beta s pre-existing market power by raising the price of gadgets and further opening up the gap between the price of gadgets and the marginal production cost of gadgets. The pitfall in the 14b approach is viewing Alpha s decision as a dichotomous one: choosing to license Beta or not. In fact, however, Alpha is also negotiating a royalty rate (and very likely other terms and conditions) with Beta. So, if Alpha is forced to consider a license that applies to widgets as well as gadgets, inevitably Alpha will be less keen to license, will have a higher optimal royalty rate, and will have a more credible threat not to license. All of these factors tell us that we should expect a higher royalty rate for the gadget plus widget license than for the gadget only license. So, while it may be tempting for the Commission to think that it can take the gadget-only license and extend it to widgets, in fact the 14b rule will likely cause Alpha to raise its royalty rate. This pitfall is a fine example of the general rule of unintended consequences: regulators are well advised to consider how private firms will adjust their strategies and tactics in response to regulatory rules. Here the adjustment comes in the form of higher royalty rates, which impede technology transfer, raise the costs of licensees, and tend to lead to higher prices. These adverse consequences are concrete and foreseeable. 9 A refusal to license is economically equivalent to an infinite royalty rate. Note also that licensing Beta in the widget business, where Alpha has a leading position, may well entail greater opportunity cost to Alpha in terms of lost sales than does licensing Beta in the gadget business, where Alpha does not enjoy a leading position. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 6
Second, as a practical matter, we may ask how Alpha and Beta are to determine whether their proposed licensing agreement will be judged by the Commission to restrict competition under the 14b approach. According to the Draft Guidelines, even if Alpha concludes based on a reasonable, good-faith internal analysis that it would rather keep its invention to itself (not license Beta at all) than license to Beta in both widgets and gadgets, the Commission may still conclude that the field-of-use provision was not objectively necessary. Inevitably, the 14b approach would impose substantial business risks on a wide range of licensing provisions. Surely the result will be to discourage many license agreements. To see this, consider what happens if Alpha considers the field-of-use provision an essential part of a license to Beta, and thus includes that provision in its license with Beta, but the Commission subsequently reviews the license and concludes that the field-of-use provision was not objectively necessary. If the Commission then voids this provision alone, Beta will enjoy a windfall, namely the right to produce widgets using the patented technology. This could prove very costly to Alpha, a cost that appears as a business risk when Alpha is considering the original license agreement. Alpha may not be willing to bear this risk and may instead simply refrain from licensing at all. Or Alpha may charge a higher royalty rate to Beta to compensate for this risk. Either of these outcomes would be worse for competition than the license as proposed and as accepted by the Commission under the 14a approach. Alternatively, if the Commission voids the entire agreement, Beta will lose the right to use Alpha s patented process invention, even if Beta had made significant capital investments in reliance on the license from Alpha. In this situation, Beta bears a significant risk and will be less inclined to make the specific investments necessary to effectively utilize Alpha s invention. The result is that Alpha s invention is not utilized as effectively and consumers in the gadget market are harmed in comparison with the license as originally proposed and as accepted by the Commission under the 14a approach. The Draft Guidelines focus on, and then dismiss, the possibility that Alpha would vertically integrate in gadgets if not permitted to issue a gadget-only license: Claims that in the absence of a restriction the supplier would have resorted to vertical integration are not sufficient. Decisions on whether or not to vertically integrate depend on a broad range of complex economic factors, a number of Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 7
which are internal to the undertaking concerned. The initial choice to rely on cooperation rather than vertical integration anyhow already generally indicates that vertical integration was not practicable. (Draft Guidelines, 14b) While this logic may have some force in vertical distribution agreements (where a manufacturer or wholesaler is seeking distribution through retail outlets), I question whether the conclusion that vertical integration was not practicable applies generally to technology licensing situations. In my hypothetical, Alpha was already producing gadgets, so Alpha has some ability to exploit its innovation in gadgets whether or not it licenses to Beta. Furthermore, Alpha s original preference to grant a gadget-only license to Beta implies that Alpha has concluded that the opportunity cost of granting Beta a license in widgets (on the original terms) exceeds the royalty revenues Alpha would earn from Beta in widgets. Therefore, Alpha will either raise the royalty rate or refrain from licensing Beta at all if a gadget-only license is not permitted. Indeed, Alpha may decline to license Beta, or raise the royalty rate charged to Beta, even if vertical integration was not practicable for Alpha in the gadget business. To summarize, there is no free lunch here. The Commission cannot expect to force Alpha to license its patented invention to Beta for widget production without creating major economic problems. One major problem is that Alpha s incentives to innovate in the first place will be reduced. The other major problem is that Alpha will predictably respond by charging higher royalties. When we also account for the lack of predictability of the 14b approach, it becomes clear that 14a is both sufficient to protect competition and superior in terms of promoting innovation and diffusion. C. Limited Grants or Licensing Restrictions The Draft Guidelines tend to view provisions that limit how the licensee can use the licensed intellectual properties as restrictions on the license. But one may equally well view such provisions as merely defining the boundaries of the rights that are granted under the license. As I stated in my OECD report (p. 18): One of the very attractive features of intellectual property is that it can be divided up in various ways and thus efficiently combined with complementary assets. I explain there that field-of-use license grants are a way that the innovator can offer partial rights to licensees, presumably rights tailored to the licensee s Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 8
strengths. Economists widely recognise that such matching of complementary assets is highly efficient. (Shapiro OECD Report, p. 19, emphasis in original) D. The Pro-Competitive Role of Cross Licenses Cross licenses can be highly efficient arrangements whereby two companies trade their intellectual property. For example, broad cross licenses are widely used in the semiconductor industry. Many researchers, including myself, have shown how these cross licenses allow firms in the semiconductor industry to cut through the patent thicket and effectively commercialize new technology. 10 Indeed, it is hard to envision a good alternative to broad, forward-looking cross licenses in the semiconductor industry, where many firms possess substantial patent portfolios, where a single product can potentially infringe hundreds if not thousands of patents, and where manufacturers may be subject to hold-up when newly-issued patents are asserted against products that have been in the design and/or production phase for years. Given the impressive innovation displayed in this industry, great care should be taken not to impose rules that would impede pro-competitive cross licensing. Put differently, while it is true that crosslicenses can be abused, they also serve many crucial, pro-competitive purposes. For this reason, the Commission should be very cautious about labeling provisions in cross licenses as hardcore restrictions. III. Hardcore Restrictions The Draft Guidelines make it very clear that hardcore restrictions are to treated harshly by the Commission. Hardcore restrictions of competition only fulfil the conditions of Article 81(3) in very exceptional circumstances. They normally do not create objective 10 See p. 16 of my OECD report, which explains how rapid technological change, the importance of interfaces, a growing patent thicket, and the danger of hold-up or opportunism combine to make licensing, and cross-licensing, especially important in the information technology sector. My article on the patent thicket develops these ideas further. For more information about the semiconductor industry in particular, see also Bronwyn H. Hall and Rosemarie Ham Ziedonis, The Patent Paradox Revisited: An Empirical Study of Patenting in the U.S. Semiconductor Industry, 1979-1995, Rand Journal of Economics, Spring 2001; and Peter C. Grindley and David J. Teece, Managing Intellectual Capital: Licensing and Cross-Licensing in Semiconductors and Electronics, California Management Review, 1997 Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 9
economic benefits or benefits for consumers. Moreover, these types of agreements generally also fail the indispensability test under the third condition [of Article 81(3)]. (Draft Guidelines, 16) The classification of a restraint as a hardcore restriction of competition is based on the nature of the restriction and experience showing that such restrictions are almost always anti-competitive. (Draft Guidelines, 72) When a technology transfer agreement contain a hardcore restriction of competition the agreement as a whole falls outside the scope of the block exemption. There is no severability for hardcore restrictions. Moreover, the Commission holds the view that in the context of individual assessment hardcore restrictions of competition will only in exceptional circumstances fulfil the four conditions of Article 81(3), df. point 16 above. (Draft Guidelines, 73) I am deeply concerned that the Draft Guidelines classify several licensing provisions as hardcore restrictions, despite the fact that they frequently serve pro-competitive purposes. I do not believe that there is any economic basis either theoretically or empirically to conclude that the practices discussed below normally do not create objective economic benefits or benefits for consumers. Nor do we have experience showing that such restrictions are almost always anticompetitive. Therefore, I urge the Commission to reconsider classifying these licensing provisions as hardcore restrictions, and instead subject agreements containing these provisions to individual assessment. 11 A. Field-of-Use Provisions Narrower Than Product Markets The economic analysis of whether field-of-use provisions are pro- or anti-competitive does not hinge on whether the permitted uses correspond to a relevant product market. In my example above with Alpha and Beta, in which Alpha granted Beta a license only for gadgets, my analysis did not turn on whether gadgets were in the same relevant market as widgets: a gadget-only license enhances competition in gadgets and does not hinder competition in widgets. If gadgets compete with widgets, some effect can be expected on widget prices (depending upon how 11 My understanding is that certain restrictions will be treated as hardcore when contained in licenses for the manufacture of contract products, but not when contained in settlement and non-assertion agreements. See, e.g., Draft Guidelines, 35. It is not clear to me what the economic distinction is between these classes of agreements, but it appears that similar reasoning lies behind branding certain restrictions as hardcore when contained in licenses for contract products and the strict language about how such restrictions would be treated when contained in Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 10
differentiated gadgets are from widgets), so the pro-competitive effects of the gadget-only license are not confined to gadgets. If gadgets compete with widgets, Alpha might be less inclined to grant a gadget-only license, or to charge a higher royalty. But all of my conclusions above about the pro-competitive nature of the gadget-only license apply, whether or not gadgets are sufficiently close substitutes to widgets to belong in the same relevant market. The Draft Guidelines do not appear to appreciate these sound economic principles: The decisive factor is whether the products incorporating the licensed technology belong to separate product markets. A restriction on the license based on the use made by the buyer of the products incorporating the licensed technology is not a field of use restriction for the purposes of these guidelines. Such a restriction is a customer restriction. For instance, licence agreements whereby one licensee is licensed to produce vehicle transmissions for incorporation into cars having engines with four cylinders or less and another licensee is licensed to produce the same transmissions for incorporation into cars having engines with more than four cylinders allocates customers between the licensees and is not treated as a field of use restriction. (Draft Guidelines, 170) I fail to see why the example provided constitutes a customer restriction, since the provision at issue does not appear to limit competition, either generally or for specific customers, that would have occurred in the absence of the license. I presume in the example that both licensees make transmissions using non-infringing technology; the terms and conditions of the two licenses do nothing to hinder that competition. The licenses enhance competition by one licensee using the licensed technology for cars with four cylinders or less, and by another licensee using the licensed technology for cars with more than four cylinders. Whether one looks at competition for sales of cars as a whole, for sales of cars with four cylinders or less, or for sales of cars with more than four cylinders, competition has been enhanced. Further, there is no economic basis for treating intra-market licenses more harshly than other field-of-use provisions. Indeed, if Firm A is granted an exclusive license restricted to high-end products and Firm B is granted an exclusive license restricted to low-end products (in the same relevant market), Firm A and Firm B still compete with each other in the same relevant market settlement or non-assertion agreements. My analysis below does not distinguish between these two classes of agreements; in either case, I question whether the provisions discussed here are almost always anticompetitive. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 11
using the licensed technology. In contrast, if Firm C is exclusively licensed to sell in Market X and Firm D is exclusively licensed to sell in Market Y, neither Firm C nor Firm D faces any competition using the licensed technology. Yet the Draft Guidelines would treat the high-end and low-end field-of-use provisions, which enable some competition using the licensed technology, as hardcore customer restrictions, and the Market X and Market Y field-of-use provisions, which do not enable any competition using the licensed technology, more favorably. It is possible that the Commission is hoping that, by treating intra-market field-of-use provisions as hardcore customer restrictions, it will induce licensors to enter into the same licenses without such provisions. For reasons I explained above (see my discussion of the effect of prohibiting field-of-use provisions in the context of my example with Alpha and Beta selling widgets and gadgets), I do not believe this is a sound basis for competition policy or a reasonable expectation. First, such a policy seeks to force the patent holder who chooses to license to create competition in its own technology (akin to mandatory licensing), either between licensees or between itself and its licensees. As is well known, mandatory licensing undermines the value of patents and hence the incentive to innovate. Second, such a policy would undoubtedly impede licensing, especially licensing designed to permit the licensee selectively to serve part of the market that the licensor did not, or could not, serve efficiently. Put simply, some licenses that are issued today containing field-of-use provisions will not be issued under approach taken in the Draft Guidelines. Third, such a policy will predictably lead to higher royalty rates in those licenses that today contain the offending field-of-use provisions and would still be entered into, especially in those cases where the licensor sells products that generate a higher gross margin than the competing products sold by the licensee. 12 The net effect of such a policy would be to hinder the diffusion and use of new technology, and to prevent firms who are not at the cutting edge of technology from obtaining licenses from their more innovative rivals who sell higherend, higher-margin products. 12 Put differently, the licensor will clearly have an incentive to raise the royalty rate it charges if the Commission prohibits it from employing a field-of-use provision that would otherwise be included. In some situations, the licensee will accept the higher royalty rate, although this will tend to raise costs and prices. In other situations, the licensee will not accept the higher royalty rate, in which case the license agreement will not be concluded. Either outcome is undesirable from the perspective of efficiency and competition. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 12
Additionally, as a purely practical matter, treating intra-market field-of-use provisions as hardcore violations would require licensors and licensees to carefully define relevant antitrust markets simply in order to determine which licensing provisions are likely to be acceptable to the Commission. Inevitably, this approach would impose great uncertainty on firms seeking to sign license agreements. Modifying slightly the example from the Draft Guidelines, how would a field-of-use provision limiting a license to compact and sub-compact cars be evaluated? I know from personal experience that defining the contours of the market(s) for cars is a difficult and potentially controversial exercise. More generally, I know from years of experience as an antitrust economist, including my work on dozens of mergers, that defining the scope of relevant markets typically is a complex and time-consuming exercise. Worse yet, even if two companies engage in a good-faith effort to determine the scope of the relevant market, they cannot be confident that the Commission would reach the same conclusion. These complexities and uncertainties are especially pronounced in markets experiencing technological change; of course, licensing agreements are important in many such markets. Again, I know from considerable personal experience that determining which generations of technology belong in the same relevant market can be a complex and controversial exercise. So, while the prohibition on intra-market restrictions might greatly increase the need to define relevant markets, and thus the demand for antitrust economists (!), I urge the Commission to abandon this approach in part because of the uncertainty and costs it will impose on licensors and licensees. B. Reciprocal Field-of-Use Provisions in Cross Licenses Draft Guidelines state ( 83): Article 4(1)(c) is applicable where competitors cross licence competing technologies and impose reciprocal field of use provisions covering distinct product markets or technical fields of application. For instance, if A and B cross licence competing technologies and A limits the licence to product market X and B limits his licence to product market Y, the agreement amounts to a sharing of product markets. To see why hardcore treatment of such provisions is unsupported by economics, consider the following reasonable hypothetical. Sigma is a large company with many patents that holds a leading position in Market X but only a moderate market share in Market Y. Product Y is a Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 13
complement to Product X. Tao is a much smaller company that has a handful of patents and also has a moderate market share in Market Y, where it competes directly against Sigma. Tao is considering entering Market X, but may not do so, in part because products in Market X are very likely to infringe many of Sigma s patents, in part because Tao lacks other assets valuable or necessary to compete in Market X. Sigma has some concern that Tao will assert its patents against Sigma s products in Market X, where Sigma earns substantial revenues. Sigma believes that Tao s sales in Market Y infringe some of Sigma s patents. Sigma hopes to use its own extensive patent portfolio defensively to protect itself in Market X by signing a cross-license with Tao covering all of Tao s patents. Tao is prepared to license its patents to Sigma in Market X in exchange for a license to use Sigma s patents in Market Y. To make sure that it protects itself, Tao seeks a license to all of Sigma s patents. Tao does not, however, want to license its patents to Sigma in Market Y, as doing so might give up an important competitive advantage that Tao enjoys in Market Y, where Sigma enjoys other advantages based on its leading position in Market X. So, Sigma and Tao enter into a cross-license covering all of their patents with the provision that Sigma can only use Tao s patents in Market X and Tao can only use Sigma s patents in Market Y. Perhaps there is some lump-sum balancing payment, but no running royalties in either direction. How should we evaluate this cross license? From the perspective of competition policy, this cross license is a wonderful outcome. Sigma can continue to offer its products in Market X without bearing any royalty costs to Tao. Indeed, the cross license eliminates the danger that Sigma would be forced to pay royalties to Tao based on sales in Market X, which would tend to elevate prices in Market X. The cross license also eliminates the danger that Tao might obtain an injunction to force Sigma to reduce output in Market X. Plus, the agreement in no way limits competition in Market X: nothing in the agreement makes it harder for Tao to enter Market X. What about Market Y? The agreement allows Tao to continue to compete, with no royalty burden, and eliminates the risk that Tao would be forced to pay Sigma royalties on Tao s sales in Market Y. And the agreement in no way restricts the ability of Sigma to compete in Market Y. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 14
It is difficult to understand why the Commission would conclude that this type of agreement is almost always anti-competitive, but that appears to be what the Draft Guidelines intend ( 83, as cited above). There is no reason to consider that such arrangements almost always amount to sham license arrangements intended to reserve to each of the parties a specific delineated market. Nevertheless, the Draft Guidelines take the view that reciprocal field-of-use provisions amount to market allocation. If Sigma and Tao entered into an agreement that flatly prohibited Sigma from competing in Market Y and flatly prohibited Tao from competing in Market X, regardless of whether they were infringing the others patents, I would agree that such an agreement on its face restricts competition that was taking place (or might have arisen) in the absence of the agreement. But that is simply not the case for the cross license in my hypothetical, and for many cross licenses in practice that involve field-of-use provisions. 13 For example, under my hypothetical cross license between Sigma and Tao, there is no reason to expect Sigma to withdraw from Market Y, or Tao to be any less significant a potential rival in Market X. 14 The Draft Guidelines go on to say ( 83): The agreement also falls outside Article 4(1)(c) where the field of use restriction is symmetrical in the sense that both parties can exploit the licensed technology only within the same field(s) of use. I fail to understand why there is anything special about symmetrical field of use restrictions, given that the two parties to a cross license are often highly asymmetric in terms of their pre-existing market positions (as in my hypothetical example above) or in terms of their patent portfolios. Competition is not harmed when Sigma lets Tao practice Sigma s patents in Market Y and Tao lets Sigma practice Tao s patents in Market X. 13 My understanding is that the Racal/Decca license agreement was found, after a full competitive analysis, to have restricted competition that would have taken place in the absence of the agreement. I strongly suspect but cannot prove that such agreements, which use the cover of an intellectual property license to achieve a collusive outcome, are relatively rare in the universe of all cross-license agreements. But my point here is far weaker than my hypothesis in the previous sentence; I am merely asserting that there are many pro-competitive cross licenses involving field-of-use provisions that would automatically be branded hardcore violations under the Draft Guidelines. 14 The Draft Guidelines recognize elsewhere that failure to obtain additional rights does not mean that a company will compete any less than it did before. See, for example, 84: If the licensee is unrestricted in the use of its own competing technology then it cannot be presumed that the field of use restriction will cause the licensee to withdraw from the markets excluded by the restriction. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 15
Evidently, the Commission hopes or believes that, by treating certain field of use provisions as hardcore restrictions, companies will simply eliminate those provisions and not otherwise modify the terms and conditions of their cross licenses in response to the Commission s rules. While I can see that such an outcome might appear at first glance attractive (holding aside the long-term impact on innovation incentives), there is no reason to believe it is obtainable as a general matter. Rather, companies are likely to respond to the Commission s treatment of reciprocal field-of-use provisions in cross licenses as hardcore restrictions in two ways that are undesirable from the perspective of competition policy: (1) by entering into fewer cross licenses, and (2) by raising the royalty rates in those cross licenses. 15 Response (1) would be very unfortunate, precisely because cross licenses with field-of-use provisions can be so clearly pro-competitive, as in my example above. Response (2) also is clearly worse for consumers, since higher royalty rates tend to lead to higher final prices. My hypothetical cross license example involving Sigma and Tao can be used to illustrate responses (1) and (2). From Sigma s perspective, licensing Tao to use Sigma s entire patent portfolio in Market X could prove very costly indeed. Sigma might well prefer to take its chances defending against Tao s handful of patents rather than make all of its patents available to Tao in Market X. If some patents were clearly usable only in Market X, and others in only Market Y, Sigma could try licensing only its patents suitable for Market Y, but that may well not be the case if some patents are useful or even essential in both Market X and Market Y. 16 Further, Tao will be very wary of licensing only a portion of Sigma s patent portfolio for fear of later being sued by Sigma for infringing patents not covered by the license. If Sigma has a large patent portfolio, the cost to Tao of making sure that the license covers all of the Sigma s patent needed by Tao may be considerable, especially since Tao knows that Sigma will be better informed about Sigma s patent portfolio than Tao. 17 Even if Sigma is willing to grant Tao a license to all of Sigma s patents for use in Market X as well as Market Y, Sigma would very 15 I discuss running royalties in cross licenses below. If the Commission also limits the use of running royalties in cross licenses, companies will be even more inclined to reduce their use of cross licenses altogether. 16 The inability to cleanly assign patents to specific markets, or products, is especially pronounced in the semiconductor industry. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 16
likely seek a higher royalty rate in the absence of the limitation to Market Y. Apart from Sigma s strong position in Market X, and possibly higher gross margins in Market X than Market Y, Sigma s patents may simply be more valuable in Market X than in Market Y in terms of cost savings or performance improvements. Of course, Tao may decline the pay the higher royalty rate, which is another reason why the approach in the Draft Guidelines will lead to fewer cross licenses. C. Running Royalties in Cross Licenses The Draft Guidelines exhibit a general skepticism regarding running royalties in cross licenses. Where competitors conclude an agreement providing for cross licensing and reciprocal royalty payments based on the sales of the final product, the Commission will treat the arrangement as price fixing where the agreement does not lead to a significant integration of complementary technologies. The Commission will only consider that the arrangement constitutes price fixing where the parties could reasonably have chosen a less restrictive payment scheme such as lump sum payments or one-way payment of net royalties and where the amount of the royalty is such that it is likely to have a not insignificant impact on prices. (Draft Guidelines, 77) In the context of cross-licensing between competitors royalty obligations may amount to price fixing and thus constitute a hardcore restriction of competition. (Draft Guidelines, 147) Licenses [in the context of settlement and non-assertion agreements] should be royalty free or, where the objective value of the technologies in question is different, provide for one-way (lump sum) royalties, reflecting the objective difference in value. (Draft Guidelines, 199) It is true as a matter of economics that rivals can use excessive two-way running royalties in cross-licenses to elevate prices and replicate the monopoly or cartel solution. Michael L. Katz and I showed how this could occur in our 1985 Rand Journal of Economics article, On the Licensing of Innovations. The hallmark of such anti-competitive licenses is that they raise, not lower, the cost of production for one or both firms, e.g., by imposing royalties that exceed the cost savings flowing from a process innovation. One warning flag indicating the presence of 17 Again, this problem is especially severe in the semiconductor industry, where hundreds of patents can potentially read on a single product. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 17
such an arrangement would be royalties that are based on total production, not just production using the licensed invention. 18 (If royalties only apply when the licensee uses the licensed technology, the licensee has an incentive to produce products not using the licensed technology if the royalty rate in fact exceeds the costs savings associated with the licensed technology). Other indicators of such an anti-competitive agreement would be evidence that the license in fact raised the cost of production to one or both firms, along with an observed price increase following conclusion of the agreement. However, the fact that cross-licenses can be abused to coordinate pricing and restrict competition does not imply that such a practice is widespread. Nor does the danger of abuse take away from the fact that cross-licenses with running royalties serve other pro-competitive purposes. Indeed, the Draft Guidelines acknowledge as a general matter that running royalties are common in procompetitive licensing agreements. It is a common feature of licence agreement that the consideration payable to the licensor depends on the licensee s use of the licensed technology. (Draft Guidelines, 45) Economists generally recognize that using running royalties allows firms to enter into licensing agreements when they are highly uncertain over the future stream of royalty-bearing revenues, and especially when the two parties have very different expectations regarding those revenue streams, and thus very different ex ante valuations on the licensed technology. Certainly oneway running royalties in cross-licenses are often used just as are running royalties in nonreciprocal licenses, namely to permit payment based on the realized value of the licensed technology. I see no reason to believe that lump-sum payments can work smoothly as an alternative to one-way running royalties in cross-licenses when we know running royalties are common, and necessary, in many standard licensing agreements. 19 18 I do not mean to suggest that total-sales royalties are themselves typically anti-competitive; they often can reduce the transaction costs associated with licensing. I am merely pointing out that total-sales royalties can be abused in conjunction with two-way running royalties in cross licenses. 19 Prohibiting running royalties would make it much harder for smaller, capital-constrained licensees to negotiate license agreements with licensors. Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 18
When two-way running royalties are involved, the Commission might like to see the two parties net out the royalties running in one direction against royalties running in the other direction. This would indeed be logical if the royalties running from Firm A to Firm B were calculated on the same revenue base as those running from Firm B to Firm A. However, this is typically not the case: revenues on Firm A s products are used to calculate royalties owed by Firm A to Firm B, and revenues on Firm B s products are used to calculate royalties owed by Firm B to Firm A. Since running royalties serve an important and legitimate purpose in many licensing agreements, prohibiting running royalties in cross licenses surely would have significant costs, including discouraging the use of cross-licenses. Given that running royalties are generally an important element in licensing agreements, I encourage the Commission to evaluate running royalties in cross licenses using individual assessment rather than hardcore treatment. IV. Cross Licenses That Include Future Patents The Draft Guidelines may impede cross licenses that include future patents as well as current patent portfolios. Where under the agreement the parties are entitled to use each other s technology and the agreement extends to future developments, it needs to be assessed what is the impact of the agreement on the parties incentive to innovate. In cases where the parties have an appreciable degree of market power and where the agreement prevents the parties from gaining a competitive lead over each other, the agreement is likely to be caught by Article 81(1) and is unlikely to fulfil the conditions of Article 81(3), in particular the condition of indispensability. (Draft Guidelines, 198) This approach concerns me, since it does not appear to recognize some of the important procompetitive aspects of cross licenses that include future patents, at least in industries, such as the semiconductor industry, where firms run the risk that patents not yet issued will later be asserted against products currently being designed or even produced. In such industries, a broad portfolio cross license including future patents allows both parties to the cross license to obtain design freedom, to achieve genuine patent peace, and to avoid being held up by the other with patents Statement of Carl Shapiro on Draft Technology Transfer Guidelines, Page 19