آیا اختراعات ثبت شده موثر باید در پکیج های کم حجم ارائه بشود؟تجزیه و تحلیل رفاهی مالکیت معنوی

Intellectual property owners often hold the rights to several patents, each of which is essential to make or use a product. We compare the welfare properties of package licenses, under which a licensee pays the same fee regardless of the number of technologies licensed, with component licenses, under which each technology is licensed separately and there is no quantity discount. A central finding is that a long-term package license can induce incentives to invent around patents and invest in complementary assets that are closer to their socially optimal levels than are those induced by a long-term component license. We also identify settings in which a short-term license is a partial substitute for a package license and a prohibition on package licensing induces parties to adopt contracts that result in less efficient complementary investment because of hold-up.

It is common for an intellectual property owner to hold several patents covering technologies that are valuable only if used together. Although it would seem natural to offer the rights to use these patents under a single license, such “package licensing” has long been greeted with skepticism under antitrust policy.1 Two common objections to package licensing are claims that the practice: (a) forces licensees to purchase intellectual property rights that they do not want or need, and (b) discourages attempts to innovate around specific patents or have specific patents declared invalid, not infringed, or unenforceable, because eliminating the need to license a single patent would not change the package price for the remaining patents in the bundle.2 The first argument against package licensing is readily dismissed given that the overall exchange between the licensor and licensee is voluntary and the marginal costs of including additional patents in a license are zero or nearly so. The flaws in the second argument are more subtle and are the subject of the formal analysis below. Antitrust concern with the packaging of two or more distinct products comes up in other settings, under names such as bundling, tying, and block booking. It is helpful to identify what distinguishes package licensing in our model from these other practices and the related literature. Package licensing is of particular interest for a number of reasons: • Package licenses often contain patents that are strongly complementary in the sense that the underlying intellectual property covered by each patent can be put into application only if one also makes use of the intellectual property covered by the other patents. In these instances, there is no sense in which users have separate valuations of the different patents. Thus, the motive to use bundling or block booking to “average out” valuations across different units does not arise.3 • The technologies covered by patents in a package often are used in fixed proportions. Consequently, packaging complementary patents is not motivated by metering or Ramsey pricing considerations that may arise with other goods.4 • The inclusion of additional patents in a package license typically has near-zero incremental cost. From a purely static perspective, even small transaction costs associated with licensing individual patents can make combining patents in a package both socially and privately desirable. • A licensee may desire a package license to reduce the potential for hold-up. Separate licensing on a patent-by-patent basis exposes a licensee to high royalties for any additional patents that are necessary to produce a commercial product after the licensee has agreed to pay fees for the rights to an initial subset of the necessary patents. A package license that covers all present and future patents owned by a given licensor can reduce this hold-up risk. Several papers, including Whinston (1990), Choi and Stefanadis (2001), and Carlton and Waldman (2002), have explored how tying can serve as a form of entry (or innovation) deterrence, but the forces at work for package licensing are very different. The earlier models require a commitment to bundled pricing. In contrast, package licensing of complementary patents is a contractual, not technological tie, and renegotiation is feasible. Furthermore, in Whinston's model, for example, technological tying is a means of committing to a low price in response to entry. In contrast, in our model package licensing serves a function equivalent to raising the post-entry licensee fees for those technologies where entry has not occurred. A related difference between our model and the entry-deterrence literature is that the buyer in our model enters the input market itself, so that it internalizes the gains in consumer surplus due to lower prices. Below, we examine the welfare properties of equilibrium licensing contracts and how these contracts are affected by the antitrust treatment of package licensing undertaken by a single intellectual property owner.5 In our model, a tie in the form of a package license creates pricing flexibility following the entry that may occur through innovation. This flexibility discourages invent-around. At the same time, the package license enables a commitment not to charge prices that exploit sunk complementary investments made by the licensee. A central finding is that, in many settings, package licensing does not have adverse efficiency effects on invent-around incentives. Intuitively, this is so for two reasons. First, although package licensing may attenuate invent-around incentives in comparison with other types of licensing, this reduction can bring private incentives closer to their socially optimal level. To the extent that R&D is motivated solely by the desire to avoid paying royalties for intellectual property that would otherwise be licensed, the innovation gives rise to private benefits, but makes no contribution to total surplus. Second, the argument about incentive effects neglects to account for the fact that a licensor could adjust the license fees for its remaining patents should one of its patents become ineligible for licensing. As we will show, such adjustments can allow the licensor to achieve the effects that critics ascribe to package licensing, and can do so in ways that result in lower welfare than would package licensing. Our analysis confirms in some respects the standard intuition of the one-monopoly-rent theorem, but in other respects our results differ sharply. If a manufacturer requires a license to use one or both patents after investing in R&D and complementary assets, then the intellectual property owner can extract all of the available rent by charging for the use of only one patented technology. This is consistent with the one-monopoly-rent intuition. However, before the manufacturer invests in R&D and complementary assets, the allocation of license fees to individual patents can affect the profit that the intellectual property owner is able to derive from its patents because the allocation influences the manufacturer's investment incentives. The problem of license contract design in our model is related to research on technology transfer with asymmetric information such as Gallini and Wright (1990), Beggs (1992), and Anton and Yao, 1994 and Anton and Yao, 2002. The focus of the first two papers is on the design of separating contracts that signal the value of the technology when either the licensor or the licensee has private information. We also derive separating contracts but our focus is on the welfare implications of licensing both patents as a package. The IP incumbent's intellectual property is perfectly protected by patent in our model, which eliminates the risk of expropriation analyzed in Anton and Yao, 1994 and Anton and Yao, 2002. The paper proceeds as follows. Section 2 lays out the baseline model in which a single incumbent owns a pair of complementary patents that can be used by a single potential licensee to produce final output. Section 3 characterizes the equilibrium when package licensing is allowed, while Section 4 characterizes the equilibrium when component licensing is mandatory. Section 5 examines several extensions, and the paper closes with a short conclusion.

The economic effects of package licenses are more subtle and complex than popular intuition asserts. Although based on strong assumptions, our simple formal model illustrates several broader points that apply to package licensing when the licensor knows the licensee's values of the patents. First, prohibiting package licensing can result in equilibrium contracts that induce inefficient invent-around R&D in situations where package licensing would forestall it. In other cases, the licensee's incentives to innovate may be socially excessive even under package licensing. Second, shorter contracts can be a (partial) substitute for package licensing. For some settings in our model, the IP incumbent can use short-term contracting to adjust its contract offer after the uncertainty about innovation has been resolved and can thus fully exercise whatever market power remains. More generally, this analysis indicates that – absent the ability to engage in packaging – licensors may shorten the length of pricing commitments and thus retain flexibility to respond to changes in the ability to collect fees for specific patents. One of the effects of prohibiting package licensing can thus be to induce the private parties to adopt contracts that result in less efficient complementary investment because of hold-up problems. Third, when there are multiple production periods, a long-term component contract under which some payments are made before any innovation has taken place can also serve as a (partial) substitute for package licensing in some circumstances. Lastly, although our formal model takes the IP incumbent's technology as given, a clear implication of this analysis is that package licensing can increase the original innovator's initial incentives to undertake R&D. There clearly remain a number of extensions that must be addressed before we have a full understanding of package licensing. A model with a more general set of bargaining institutions would be useful, as would one in which Condition A is relaxed. Results without Condition A are difficult to obtain for general convex R&D cost functions because the IP incumbent's profit need not be a concave function of the license fees. It also would be worthwhile to examine additional institutional settings, including those in which the potential innovator were a third party, the potential licensee had better information about its valuation of the patented technology than did the potential licensor, or the parties were asymmetrically informed about the costs of innovation. We note that this type of model could also be extended to analyze the effects of package licensing on a potential intellectual property user's incentives to pursue litigation to have patents declared invalid, uninfringed, or unenforceable, or even extended to the more general phenomenon of contractually tying complementary goods in order to deter two-stage entry in markets where buyers may sponsor entry. Lastly, the analysis of the present paper could be extended to the case of package licensing by patent pools in situations where design-around innovation by non-pool members is possible.