By John P. Walsh, Ashish Arora and Wesley M. Cohen
October 9, 2000

PRELIMINARY DRAFT FOR COMMENTS; DO NOT QUOTE OR CITE WITHOUT AUTHORS’ PERMISSION

Over the last two decades changes in technology and policy have altered the landscape of drug discovery. These changes have led to concerns that the patent system may be creating difficulties for those trying to do research in biomedical fields. Based on interviews and archival data, we examine the changes in patenting in recent years and how these have affected innovation in pharmaceuticals and related biotech industries.

We find that there has in fact been an increase in patents on the inputs to drug discover ("research tools"). However, we find drug discovery has not been substantially impeded by these changes. There is some evidence of delays associated with negotiating access to patented research tools, and there are areas where patents over targets limit access. There are also cases where research is redirected to areas with more IP freedom. However, the vast majority of respondents say that there are no cases where valuable research projects were stopped due to IP problems.

We do not observe as much breakdown as one might expect because firms have been able to develop "working solutions" that allows their research to proceed. These working solutions combine taking licenses, inventing around patents, infringement (often informally invoking a research exemption), developing and using public databases and challenging patents in court. In addition, changes in the institutional environment, particularly new PTO guidelines and some shift in the courts' views toward research tool patents, appear to have further reduced the threat of breakdown. Finally, the very high technological opportunity in this industry means that firms have a surplus of potential targets for drug development, so that the walling off of some by patent holders, while shifting the focus, does not prevent firms from discovering drugs.

We conclude with a discussion of the potential social welfare effects of these changes in the industry and the adoption of these working solutions for dealing with a complex patent landscape. While there are social costs associated with these changes, there are also important benefits. Overall, we are optimistic about the industry’s ability to accommodate the increased complexity of intellectual property.

There is widespread consensus that patents have long benefited biomedical innovation. A forty year empirical legacy suggests that patents are more effective, for example, in protecting the commercialization and licensing of innovation in the drug industry than any other.¹ There is broad agreement (though not much statistical evidence) that the profits protected by drug patents constitute an important incentive for firms to invest in R&D. Over the past twenty years, however, fundamental changes have revolutionized the science and technology underlying product and process innovation in the drug industry. Moreover, and roughly in parallel, policy changes have encouraged university patenting and the patenting of new kinds of biomedical discoveries. This combination of changes in policy and science has provoked concerns that patent policy may now be imposing unnecessary costs upon and consequently impeding -- perhaps even blocking -- biomedical innovation in some instances. Heller and Eisenberg [1998] argue biomedical innovation is especially susceptible to what they call a "tragedy of the anticommons," which can emerge when there are numerous property right claims to separate building blocks for some prospective product or line of research. When these property rights are held by many different claimants, the negotiations necessary to their combination may fail, quashing the pursuit of otherwise promising lines of research or product development.

In this paper, we will report the results of interviews with executives and researchers at biotechnology and pharmaceutical firms and research personnel and administrators at several universities to help us further evaluate whether the "anticommons tragedy" pervades biomedicine, and, specifically, whether patent rights over certain types of research tools are retarding innovation in this field (cf. National Research Council [1997], Eisenberg [1999]).

We focus on the costs to biomedical innovation associated with the kinds of transactions and negotiations over rights entailed by current patent policy and patenting practices. The property rights also, however, come with benefits, especially in the form of greater incentives to undertake innovation. While any ultimate policy judgment requires a consideration of both social benefits and costs, an examination of the benefit side of this calculus is beyond the scope of our current study.

In Section II of the paper, we provide background to the anticommons problem. Section III describes our data and methods. In section IV, we will provide an overview of the results from our interviews, and assess the extent of the "anticommons" problem reported to us. To prefigure the key result, we find little evidence of either routine breakdowns in negotiations over rights, or significant impediments to progress in biomedical research due to such negotiations. We do find that research tool patents can impose a range of social costs, and the potential for problems exists. In Section V, we describe the mechanisms and strategies employed by firms and other institutions that have allowed innovation to move forward and avert serious breakdowns. Although our interviews suggest that breakdowns in negotiations over rights are uncommon, there are a range of other issues to consider, which are reviewed in Section VI. Section VII discusses our findings and we conclude the paper in Section VIII.

Perhaps the defining feature of a research tool is that its social value is realized, for the most part, when it is used to discover a drug or therapy that is eventually commercialized and used.² This implies that a research tool is an input. Further, use of patented research tools will require potentially costly transactions between the patent holder and user of the research tool when the patent holder does not have the capabilities to develop therapeutics in house.

As molecular and cell biology, genetics and related fields of inquiry have come to underpin innovation in the drug industry, innovation in biomedicine has become more cumulative (Drews [2000]). Since 1981, patent policy has also significantly enhanced the range of patentable subject matter and the nature of patenting institutions. In addition to the 1981 Diamond vs. Chakrabarty decision that permitted the patenting of life forms, in the 1980’s, gene fragments, markers and a range of intermediate techniques and other inputs key to drug discovery and commercialization also became patentable. Moreover, Bayh-Dole and related legislation encouraged universities and national labs, responsible for many such upstream developments and tools, to patent their inventions.

These changes in the underlying science, and in what can be patented and who can patent have raised concerns over the implications for biomedical innovation. When property rights held by different parties have to be combined, the transaction costs may be higher than the total value from the combination. Similarly, breakdowns could arise if the stacking of license fees make the entire project commercially unviable. (Heller and Eisenberg [1998], Shapiro [2000]).

Merges and Nelson [1990], Barton [2000] and Scotchmer [1991] have highlighted other problems that might emerge in what Nelson and Merges [1990] call "cumulative-system technologies." In such cases, "unless licensed easily and widely," patents--especially broad patents--on early, foundational discoveries may either discourage or limit subsequent use, which in turn would impede subsequent innovation. Barton [2000] suggests that, in the tradeoff between the interests of pioneers versus those who wish to build on prior discoveries, the current balance "is weighted too much in favor of the initial innovator." Scotchmer [1991] has suggested that ex ante deals between pioneers and follow-on innovators can be structured to mitigate the problem although her concerns are largely about providing adequate incentives for the pioneer.³

The possibility that negotiations over patent rights might break down to the detriment of technical advance--even when a successful resolution would be in the collective interests of the parties concerned--is not a matter of conjecture. There is historical precedent. Merges and Nelson [1990] and Merges [1994], for example, consider the case of radio technology where the Marconi Company, De Forest and DeForest’s main licensee, AT&T, arrived at an impasse over rights which lasted about ten years and was only resolved in 1919 when RCA was formed at the urging of the Navy. Merges and Nelson argue that the refusal of the Wright brothers to license their patent significantly retarded progress in the aviation industry. The problems caused by the initial pioneer patent (owned by the Wright brothers) were compounded as improvements and complementary patents, owned by different companies, came into existence. Ultimately, World War I forced the Secretary of the Navy to intervene to work out an automatic cross licensing arrangement. "By the end of World War I there were so many patents on different aircraft features that a company had to negotiate a large number of licenses to produce a state-of-the-art plane." (Merges and Nelson [1990, p. 891])

Although breakdowns in negotiations over rights may therefore occur, we also know of many industries in which rights over essential inputs to innovation are routinely transferred. Levin [1982], Hall and Ziedonis [2001], and others document the experience of the semiconductor industry where rights are routinely cross-licensed. Licensing is routine in the drug industry itself.⁴ In Japan, where there are many more patents per commercializable innovation across the entire manufacturing sector, licensing and cross-licensing is commonplace (Cohen et al. [2001]).

Heller and Eisenberg [1998] and Eisenberg [1999] try to address the following question: if there is a cooperative surplus to be realized in biomedical innovation, why isn’t it? They argue that biomedical research and innovation may be especially susceptible to breakdowns and delays in negotiations over rights for one or more of the following reasons. First, the transactions costs of bundling rights belonging to numerous rights-holders may be greater than the value of the deal. Moreover, different kinds of institutions--large pharmaceutical firms, small biotechnology firms that specialize in research, and universities--are involved, and this heterogeneity can increase the difficulty and cost of reaching agreement. Complicating the story further is that these institutions--especially universities--are themselves comprised of units and individuals who may have different interests. Uncertainty over the value of rights, which is acute for upstream discoveries and research tools, spawn asymmetric valuations that contribute to bargaining breakdowns. Finally, cognitive biases may further undercut incentives to bargain successfully.

Eisenberg [1999] considers whether, "bargaining failure in the market for intellectual property licenses is a hypothetical problem that sophisticated institutions and well-functioning markets are likely to avoid, or is it something to worry about?" She states: "there is evidence that the problem is real" and that within the communities of scientists, university technology transfer professionals, and private firms in the pharmaceutical and biotechnology industries, "there seems to be a widely shared perception that negotiations over the transfer of proprietary research tools present a considerable and growing obstacle to progress in biomedical research and product development." (Eisenberg [1999, pp. 3,4]) She does suggest, however, that breakdowns and undue delays tend not to afflict the higher value innovations. She points out, however, that under conditions of uncertainty, what may be expected to be lower value innovations ex ante may turn out to be something much more significant ex post--suggesting that breakdowns afflicting only lower expected value transactions may represent a considerable social cost. She also examines negotiations over material transfer agreements, and observes that such negotiations have become more costly and time consuming, due at least partly to the prospect of patentable discoveries originating from them. The National Research Council [1997] also considers the challenges for biomedical innovation posed by the patenting of research tools and upstream discoveries more generally. In a series of case studies, the National Research Council [1997, Ch. 5] documents pervasive concern over limitations on access due to the price of intellectual property, and concern over the prospect of blocking of worthwhile innovations due to intellectual property (IP) negotiations, but no actual instances of worthwhile innovations being blocked.

To address these issues, we conducted 45 interviews with representatives from nine pharmaceuticals firms and 14 biotech firms, as well as university researchers and technology transfer officers, patent lawyers and government and trade association personnel. These interviews averaged over one and half hours each. The interviews focus on changes in patenting, licensing activity and the relations between pharmaceuticals, biotechnology firms, and universities, and how patent policy has affected firm behavior.

We used the interviews to probe whether there has been a proliferation and fragmentation of patent rights, and whether this has resulted in the failure to realize mutually beneficial trades, as predicted by the theory of anti-commons. We also looked for instances where patenting of research tools has harmed research and development in this field. Thus, we asked our respondents about the extent to which there has been a shift in the number of patents for a given drug, whether the negotiations over IP rights have changed and how they currently are operating, and have there been cases of innovations stopped because of breakdowns in negotiations or stacking of license fees.

Do the key conditions that might foster an "anticommons" exist? We find that there has been in an increase in patenting, in the number and type of patent holders, and that conditions conducive to an "anti-commons"-like breakdown obtain.

Our interviews confirm the observation that there has been a proliferation of patents associated with the drug development process. As shown in Chart 1, the number of biotechnology patents has grown from less than 2000 issued in 1985 to 9,000 in 1999. The number of US biotechnology patent applications has jumped from less than 2,000 in 1986 to approximately 30,000 in 2000 (Wolff [2001]). Preceding the recent growth in patenting in biotechnology, we also witnessed rapid growth in the number of biotechnology firms in the 1980’s, (Cockburn, Henderson, Orsenigo and Pisano [2000]). In the 1990’s, we have also seen the growth of firms specializing in research tools as distinct from drug discovery.

Universities themselves have now become major players in biotechnology, as sources of patented biomedical inventions and startup firms often founded on the strength of a patent position. Several respondents noted that this new role of universities is one of the significant changes over the last two decades in the drug and related industries. Universities have increased their patenting dramatically over the last two decades, and while still small, their share of all patents is significantly higher than before 1980. Furthermore, much of the growth in university patents tend to concentrate in a few utility classes, particularly those related to life sciences. In three of the key biomedical utility classes, universities’ share of total patents increased from about 8% in the early 1970s to over 25% by the mid-1990s.

Contributing to this rise in patenting, particularly in genomics, is the intensification of defensive patenting characteristic of complex product industries such as semiconductors or telecommunications (cf. Cohen et al. [2000], Cohen et al. [2001]).⁵ Overall, about a third of our respondents claimed to be increasing their patenting of gene sequences, assays or other research tools to ensure freedom to operate. This increased patenting is one response to the increased patenting of others, a situation that echoes the patent races observed in semiconductors and other complex product industries. These other industries have for the most part managed, however, to negotiate the "anti-commons" problem thus far. (Hall and Ziedonis [2000], Cohen, et al. [2000]).

Thus, the conditions appear conducive to a tragedy of the anticommons. We have many patents, owned by different parties, with different agendas. In short, the patent landscape has indeed become more complex.⁶

Such conducive preconditions notwithstanding, we find only limited evidence of breakdown.

The case of beta-carotene enhanced rice ("golden rice") is suggestive of the impact of numerous rights-holders on negotiations over IP. The innovation involves using as many as 70 pieces of IP and 15 pieces of technical property, spread over 31 institutions (Kryder, et al. [2000]). While there was strong interest in this product from international aid agencies, they required general IP clearance before the product could be developed. After about a year of negotiations, the result was an agreement by Monsanto, Zeneca and others to provide royalty free licenses for the development and distribution of this innovation in third world countries. In terms of IP complexity, this is the most extreme case we encountered, and even in this case, a solution was found. This experience may, however, not be generalizable since, with only the poorest countries needing this technology, the commercial value was perceived to be low.⁷

Beyond the case of golden rice, we asked respondents and searched the literature to find cases where projects were stopped due to an inability to obtain access to all the necessary intellectual property rights. In brief, respondents reported that negotiations over access to necessary IP rarely led to a project’s cessation. Of the 16 respondents who addressed this issue, 14 said they had never had a projects stopped because of problems obtaining the necessary IP. For example, one respondent indicated that about a quarter of his firm’s projects were terminated in the past year. Of these, none were terminated due to any difficulties with in-licensing of tools. Instead, the key factors leading to the termination of projects included pessimism about technical success and the size of the prospective market. One biotechnology executive stated: "I am hard pressed to think of a piece of research that we haven’t done because of blocked access to a research tool. We have dropped products because others ahead in proprietary position, but that is different." Numerous respondents reported that they did not initiate or have dropped projects if they learned another firm had a proprietary position on a drug they were considering developing.

Only one of our respondents suggested that they would not initiate projects sometimes because the "tangle" of rights they would have to obtain on tools was too great. Over a third of these respondents did, however, note that dealing with research tool patents did bog things down, and added to the cost of research.⁸ Thus, while projects being stopped is quite rare, delays were not unusual and settlements could be costly.

A particular concern raised by Heller and Eisenberg [1998] and by the National Research Council [1997, Ch. 5] was the prospect that, by potentially increasing the number of patent rights corresponding to a single gene, patents on expressed sequence tags (EST’s) would proliferate the number of claimants to a prospective drug and increase the likelihood of a bargaining breakdown. Our respondents suggested that this has not occurred. A key concern in this context was that patents on the partial sequence might give the patent holder rights to the whole gene or the associated protein, or at least, the patent might block later patents issued on the gene or the protein (as Doll [1998] of the USPTO suggested). In contrast, Genentech's Dennis Henner testified last year before Congress that his firm's position is that such EST patents do not dominate the full gene sequence patent, nor the protein, nor the protein's use; that these are separate inventions.⁹ Our respondents from industry and from the USPTO reflected this latter view. The existence of large numbers of EST patents may have had the potential to create anti-commons problems. However, the new utility and written description guidelines will, from now on, likely prevent many EST patents from issuing and will give those that issue only a narrow scope of claims. In addition, already issued EST patents will likely be construed narrowly by the PTO and the courts. Thus, the consensus is that the storm over EST’s has largely passed.

One important way in which progress on drug discovery and development might be blocked is the stacking of license fees to the point of overwhelming the commercial value of a prospective product. Most of our respondents reported that license stacking did not represent a significant or pervasive threat to ongoing R&D projects. We only heard of one specific instance where a project was stopped due to license stacking. We were told, however, that, in this instance, there were too many claimants to royalty percentages due to carelessness by a manager, who had given away royalty percentages without carefully accounting for prior agreements.¹⁰ One of our other biotechnology respondents suggest, however, that, "the royalty burden can become onerous," and that the stacking of royalties, "comes up pretty regularly now" with the proliferation of IP. Even here, the respondent said that no projects had ever been stopped because of royalty stacking. Overall, about half of our respondents complained about licensing costs for research tools, though nearly all of those concerned about licensing costs also went on to say that the research always went forward.

License stacking does not represent a significant threat to ongoing R&D projects because it tends to be anticipated ahead of time.¹¹ Indeed, one firm we interviewed said they had a corporate-level committee that reviewed all such requests to make sure such cases do not occur. Others said that while generally not the most important factor, they may choose one promising project over another based on royalty rates.¹² One respondent said while stacking is a consideration, "I can’t think of any example where someone said they did not develop a therapeutic because the royalty was not reasonable." He then went on to suggest that where stacking becomes an issue for an ongoing project, then compromises tend to be struck, often in the form of royalty offsets across the various IP holders. He states: "All are sensitive and aware of stacking phenomenon so there is a basis for negotiation, that you can’t have excessive royalties."

We find only limited support for the idea that negotiations over rights stymie pre-commercial research conducted in universities. Industrial respondents all claim that university researchers, to the extent they are doing non-commercial work, are largely left alone. In fact, firms often welcome this research, as it helps further develop knowledge of the patented technology. A prominent molecular biologist from a leading research institution confirmed this. Also, many of the firms interviewed expressed the view that the negative publicity that would attend to an aggressive assertion of rights against a university was not worth it. One university technology transfer officer reports that their university will indeed receive letters of notification of infringement. They indicated that their response was to effectively ignore such letters and inform the IP holder that the university was engaged in research, did not intend to threaten the firm’s commercial interests, and would not cease its research.¹³ Other university personnel, while acknowledging their researchers commonly infringe firms’ IP, are more risk averse and have occasionally shut down research projects in the face of such assertions.

This is particularly the case for clinical research based on diagnostic tests using patented technologies. Merz and his colleagues have recently conducted several studies of the incidence of clinical labs that have been affected by patents on diagnostic tests. One study found that 25% of laboratory physicians reported abandoning a clinical test due to patents. They also reported royalty rates ranging from 9% for PCR to 75% for the human chorionic gonadotropin (hCG) patent. In a follow-up survey of 112 labs capable of performing hemochromatosis testing, they found that many had adopted the test immediately upon publication. When the patent issued a year later, it was licensed exclusively to SmithKlineBeecham. Nearly all respondents said they knew of the patent. About half had received letters from SKB. Nineteen percent said they did not develop the genetic test for hemochromatosis and another 4% said they abandoned the test, in part because of the patent. Merz argues that: "There is no clear line to be drawn between clinical testing and research testing, because the state of the art of genetic tests is such that much more clinical study is necessary to validate and extend the early discovery of a disease gene. Thus, the restriction of physicians from performing clinical testing will directly reduce the knowledge about these genes."¹⁴ Another respondent, a former medical school dean, echoed these remarks, saying that he had been shown letters from medical school researchers and that programs had been stopped. He noted that the fact that the universities charge for these tests complicates the matter, but that clinical work is critical for the research process.¹⁵

Thus, we see some evidence that firms are willing to assert their patents against universities doing diagnostic testing and charging a fee without licensing the patented tests, and that at least some labs are stopping their testing as a result. However, the majority continue with the testing. In the case where the university is not generating revenue based on the patented technology, although some firms will send letters, universities appear to be largely left alone.

While negotiations over IP and licensing fees surely affect access, and sometimes choice of projects, our conclusion is that it is rare, with the possible exception of targets, for research and development on promising therapies not to proceed due to an inability to obtain the necessary rights to research tools. In this section, we review the various private strategies adopted by firms (and universities) that allow research and commercialization to go forward despite the proliferation of biomedical intellectual property and claimants over the past decade or so.

Before doing so, however, it is important to note that breakdowns tend not to occur simply because it is not that difficult to contract. Although most R&D executives report that the number of licenses they must obtain in the course of any given project has increased over the past decade, that number is considered to be manageable. To probe this issue, we asked respondents to consider a particular project and tell us how many pieces of IP had to be in-licensed. They said that there may be a large number of patents to consider initially--sometimes in the hundreds, and that this number is surely larger than in the past. However, respondents then went on to say that in practice, there may be, at most, only a moderate number, about a half dozen, that they have to seriously address, but that more typically the number was zero.¹⁶ However, the process of reducing the stack of relevant patents from hundreds to a handful, as well as addressing that handful, is time consuming (though it is often done in parallel with the research effort).

Many of our responding firms suggested that if a research tool was critical, they would buy access to it. Several companies that produce targets noted that, in addition to trying to develop their own therapeutics, they include the liberal and broad licensing of those targets to others as part of their business model, reflecting a belief on the part of some holders of target patents that by giving several firms a non-exclusive license, they increase the chances that one will discover a useful drug. We also observe that what might be called "general purpose" tools--tools that cut across numerous therapeutic and research applications that tend to be nonrival in use--tend to be licensed broadly. Thus, many of the more fundamental (general purpose) research tools, such as genomics databases, DNA chips, recombinant DNA technology, PCR, etc., are made widely available through non-exclusive licenses. Incyte, for example, licensed its genomics database to over 20 pharmaceutical firms (who together account for about 75% of total private pharmaceutical R&D). They have also begun expanding their licensing program to include biotech firms and universities as well.¹⁷ Celera licenses their database to firms for about $5 to $15 million per year and to university labs for about $7500 to $15,000 (Service [2001]). Taq polymerase and thermal cyclers for PCR are available from a variety of authorized reagent and equipment vendors (Beck [1998]).

Liberal licensing practices are also encouraged to the extent that the inventing around of certain tools is viable. Under such circumstances, patent holders are more willing to license on reasonable terms assuming the prospective user does not invent around to begin with. The ability to invent around puts an upper limit on the value of the rival’s patent, and gives the buyer the leverage to back out of the negotiations if the price is too high. Indeed, our respondents frequently pointed to their ability to invent around a patent as one component in their suite of solutions to blocking patents, although this is more difficult for certain targets. Firms have also occasionally developed technologies that, it was claimed, made it possible to circumvent a number of the patents in the field.¹⁸

Aside from these conventional methods for coming to terms, we find that firms have adopted a set of complementary strategies that create "working solutions" to the anti-commons problem. They routinely ignore patents (sometimes invoking an informal research exemption), go offshore, create public databases, and challenge patents in court. One pharmaceutical executive summarizes the range of strategies that firms tended to employ:

"If someone has a patent on genes, when the gene encodes a therapeutic product and they are ahead of us, we drop those projects. That is different than case of a gene as a target for a small molecule screen. There we don't drop the project. If it is just an application, it is not till the patent issues that it is infringing. Lots of these patents are pretty thin. It is an issue whether it is valid. Third, you can do things offshore. Fourth, it may be available for license and fifth, they don't tend to enforce them."

A solution to the anti-commons problem is simply to ignore research tool patents. Several respondents noted that infringement of tool patents is very hard to detect.

University researchers have a reputation for routinely ignoring IP rights in the course of their research [Seide and MacLeod, 1998]. Respondents note that many research tools are "do-it-yourself" technologies and therefore they do not feel they should be required to pay royalties for the work. In fact, there is a strong belief by some that these patented technologies were well-known in the scientific community and therefore the patents are not valid (see for example, Kornberg [1995]). University researchers will often informally invoke a "research exemption," although the legal research exemption is actually quite narrow. Some reagent suppliers facilitate this practice by supplying "unlicensed" (and less expensive) materials, also invoking the research exemption. Promega, for example, sells Taq polymerase for about half of what many licensed vendors charge, and asserts that many of its customers in university and government labs do not need a license under the experimental use exemption (Beck [1998]).

Many firms claim to be reluctant to enforce their patents against universities to the extent that the university is engaging in noncommercial research, because of the low damage awards and bad publicity that suing a university would entail.¹⁹ For example, William Haseltine of HGS said that they were ready to give academics access to data and reagents related to their patented CCR5 HIV receptor: "We would not block anyone in the academic world from using this for research purposes" (Marshall [1999]). In fact, several respondents noted that they appreciated universities using their patented technologies because if the university discovered a new use, the patent holder is best positioned to exploit the innovation, whether it is included in their patent claims or whether they have to negotiate a license from the university (since any rival licensee would need a further license for the original patent). If the university becomes a competitor, however, firms feel they then have a right to assert their patents. As noted above, this is particularly evident in cases where university physicians use patent protected discoveries as the basis for diagnostic tests. Finally, as a rule, universities do not assert their rights against one another.²⁰

Infringement of research tool patents by firms also appears to be pervasive. About a third of firms noted that one solution to the problem of research tool patents is infringing, and many respondents suggested that the practice is widespread.²¹ The firms felt that much of their research would not yield commercially valuable discoveries, and thus saw little need to spend money to secure the rights to use the input technology, particularly since it is very difficult to police such infractions. If the research looked promising, then they would get a license, if necessary. In addition, since many of these patents are of debatable validity, they felt that if a license is not available, they could challenge the patent in court. Finally, not only is use of a patented research tool hard to detect, but because of the long drug development process, the statute of limitations may expire before infringement is detected.

Consistent with this behavior, we also find that firms feel that it is not worth their while to assert their patents on all other firms that might be infringing. They may send a letter, offering terms, but will not aggressively pursue infringers on their more marginal patents. Respondents point out that the cost of pursuing these cases greatly outweighs their value in most cases. "The average suit costs millions of dollars. The target is worth $100,000. Even with treble damages, it doesn't pay to sue." There is an additional cost, and that is the risk of the patent being invalidated by the court. These firms note, however, that if the patent is central to the firm’s strategy, they will aggressively defend it. Barring that, with reference to research tool patents, there is a sense that the industry practices "rational forbearance." (NRC [1997, Ch. 6]).

Respondents also pointed out that patents are national, while the research community is global. Thus, another solution observed to the existence of research tool patents is to take the knowledge and apply it off shore. Though similar to the solution of ignoring the patent, in that it involves using patented technologies without securing the rights, this case differs in that firms are not violating the legal rights of the patent owner, at least not until there was a product developed and the firm tried to import the product. One reason why both this solution as well as a strategy of infringement is available is that many of these tools do not involve acquiring actual technology from the patent owner, but only the knowledge of the research findings.

Over the past five years, we have observed firms (especially larger pharmaceutical firms), the courts and the U.S. PTO undertake initiatives and policies that have had the effect (if not always the intent) of broadening and easing access to research tools. For example, with substantial public, private and foundation support, public databases (e.g., GenBank) and quasi-public databases (such as the Merck Gene Index and the SNPs Consortium) have been created, making genomic information widely available. Similarly, Merck has sponsored an $8 million program to create 150 patent free transgenic mice to be made available to the research community at cost, without patent or use restrictions. According to our respondents, these efforts partly represent an attempt by large pharmaceutical firms to undercut the genomics firms’ business model by putting genomic information into publicly available databases, and then competing on the exploitation of this shared information to develop drug candidates (see also Marshall [1999, 2001]).²² These initiatives represent a partial return to the time before the genomics revolution, when publicly funded university researchers produced a body of publicly available knowledge that was then used by pharmaceutical firms to help guide their drug candidate searching.²³

Some of our respondents have suggested that recent court decisions have also mitigated the "anti-commons" problem by limiting the scope of tool patents, or, in some cases, invalidating them. Thus, while patent-holders have the right to sue for infringement, the perception is that they are increasingly likely to lose such a suit. One case that comes up frequently is University of California v. Eli Lilly and Co. Here the University of California tried to argue that their patent on insulin, based on work on rats, covered Lilly’s human-based bio-engineered insulin production process. The CAFC ruled that California did not in fact possess this claimed invention at the time of filing and therefore the claim was not valid, and Lilly was not infringing. Another case is SIBIA v. Cadus, where the court initially ruled in favor of SIBIA and awarded then $18 million, but on appeal the CAFC ruled that SIBIA’s patent was not valid. The result was SIBIA, having lost its main asset, was bought out by Merck for $87million. As one respondent put it: "These are good times for a patent infringer and not great times for a patent holder." In fact, some of our respondents expressed a fear that early investments in biotech firms may not pay off because these firms will not be able to capture the expected rents from their patent portfolios. From the point of view of those using research tools, however, this seeming change in the court's attitude may represent a shift toward more freedom to conduct research without undue concern over research tool patents.

Partly responding to concerns expressed by NIH, universities and large pharmaceutical firms, the U.S. PTO has also adopted new policies that diminish the prospect of an anticommons. Specifically, last January the PTO adopted new utility guidelines that have effectively raised the bar on the patentability of tools, particularly for ESTs. These guidelines are designed to reduce the number of "invalid" patents and help clear some of the underbrush in the patent landscape (cf. Barton [2000]).

NIH has also taken the lead in pressing for greater access to research tools. For example, since 1997, NIH has negotiated with DuPont to provide more favorable terms to transgenic mice for NIH and NIH-sponsored researchers. NIH (under the direction of Harold Varmus) spent over a year negotiating with DuPont to get restrictions on publication and sharing of animals lifted and to eliminate reach through provisions (Marshall, 2000). NIH has also begun a "mouse initiative" to sequence the mouse genome and create transgenic mice. One of the conditions of funding is that grantees forgo patenting on this research. Similarly, when Celera published their human genome map findings, Science's editors were able to negotiate largely unrestricted access for academic users of Celera's proprietary database.²⁴ Thus, large institutional actors have been able to act as advocates for university researchers to increase their access to necessary research tools.

Although biomedical research does not appear to be especially vulnerable to cessation due to breakdowns over IP negotiations, our interviews do suggest that there may be other public policy concerns associated with patents on research tools and upstream discoveries. The absence of breakdown does not imply that the current system is not imposing social costs.

The area where complaints were most widespread from universities, biotechnology and pharmaceutical firms was where patent holders asserted exclusivity over "targets," either in the form of a firm excluding all others from exploiting its target (in the anticipation of doing so itself), or licensing it exclusively. In no case, however, did we hear of any target not being exploited (i.e., breakdown)--at least one party (sometimes the owner) was using it. Thus, though not a problem necessarily associated with the need to combine rights from a large number of claimants, patents on targets highlight the importance of the policy tradeoff between the rights of an initial innovator as reflected in the scope of the original patent and those who wish to build upon or use that initial innovation considered by Nelson and Merges [1990], Scotchmer [1991] and others. To the extent that firms assert exclusivity in developing drugs for a particular target, innovation may suffer.

We speculate that targets are a source of particular complaint because, to the degree that their use eventuates in a product, they are more difficult to use without ultimately notifying the owner; infringement of a target patent is more difficult when the property turns out to be valuable. Moreover, some respondents claimed that, compared to other technologies such as semiconductors, it is more difficult to invent around biological targets.²⁵

Patents on targets, if broad in scope and licensed on an exclusive basis, may forego the benefits of having different firms with distinctive capabilities and perceptions pursue different approaches to the problem ([cf. Nelson [1961]; Cohen and Klepper [1992]). For example, big pharmaceuticals firms have libraries of compounds that might affect the target, these libraries vary by firm, and are either kept secret or patented. Thus narrowing access to the target potentially entails a social cost. The following quote from a large biotech firm summarizes the issue:

"The problem is, a target is just that, a target, say, a receptor on a cell. If we did an exclusive license, and we've had that opportunity, the only compounds tested would be those in the chemical library of our licensee. Generally there is no chemical relation among the compounds that [act on this target]. The drugs work by occupying the reception site, for example. They [licensee] throw all the compounds in their library and they may or may not have one that works well. The libraries vary a lot across firms. A lot are patented. Also large pharma companies have huge collections of compounds they've synthesized over the last 100 years that have never seen the light of day. With an exclusive license, the odds of finding an active drug, let alone the best, are not good. Therefore, we [the target owner] want the target technology broadly available. Broad licensing only makes economic sense in our view."

There are a number of prominent examples of firms asserting exclusivity over a target, denying the use of such tools to others. Myriad, for example, will not permit any other firm to license its BRCA1 and BRCA2 patents, in the hope of exploiting it itself. Myriad has also required that physicians at university research hospitals not even do diagnostic tests for the BRCA1 and BRCA2 genes themselves--diagnostics which produce data valuable for research on breast cancer. Chiron has also developed a reputation for aggressively enforcing its patents on research targets. Chiron has filed suits against four firms that were doing research on drugs that block the HCV protease, and some have claimed these suits are deterring others from developing HCV drugs (Cohen [1999]). Responding to complaints about restricted access to their patented targets, a respondent from a pharmaceuticals firm stated: "Your competitors find out that you’ve filed against anything they might do. They complain, ‘How can we do research?’ I respond, ‘It was not my intent for you to do research.’" Others also defended their rights to exclude rivals from their patent targets.

Related to the complaints over the scope of claims are "homology" claims. Firms complain that their rivals are blocking them using patents that are unreasonably broad due to their use of homology arguments. In particular, some IP lawyers complained about rivals receiving claims to sequences of X length with Y% homology, which would include an enormous number of potential sequences.

In addition, critics complain that homology claims allow firms to claim inventions they do not yet posses. Henner from Genentech made the following assertion before the House Judiciary Committee: "In our view, a computer-based homology analysis should not be regarded as being a reliable indicator of a biological activity with regard to most nucleic acid or protein sequences. Accordingly, where a particular biological activity is the only basis for the utility of a particular gene or expression product, a homology-based prediction should not be capable of satisfying the requirement of our law in a majority of situations." The issue of patent scope is particularly problematic if the patent holder did not in fact "possess" the claimed invention, because this claim reduces the incentives of others better able to develop these "latent" inventions. The complaint about HGS asserting its patent over the HIV receptor is of this form (Marshall [1999]). At the time of the patent application, HGS knew only that they had found the gene for something that was a chemokine receptor. Later work published by NIH scientists detailed how the CCR5 receptor worked with HIV, making this a very important drug target. HGS will pursue this target itself and also license it to some pharmaceutical partners. University of California’s claiming of all mammalian (including human) insulin based on their work on rats is another example of an attempt to assert rights over a broadly construed target patent--albeit an example that failed.

We have found little direct support for the claim that large numbers of claimants have led to breakdowns. However, social costs may be manifest in other ways. For instance, firms may avoid derailing in-house R&D projects but only by engaging in long and costly negotiations or litigation with IP holders. Firms may also invent around or conduct the R&D overseas, possibly at the cost of reducing R&D efficiency. Finally, IP holders may have to invest in monitoring the use of their IP, which, from a social welfare perspective, may also constitute a waste of resources.

Litigation costs are likely to be a significant component of such social costs. While estimates of litigation costs varied, and the costs depend significantly on the facts of the case, it was common for the estimates to be in the range of $1-10M for each side. In addition to these out of pocket expenses, we tried to obtain an estimate of the opportunity cost of engaging in patent litigation. Again, estimates varied widely. However, most respondents suggested that being involved in litigation put a significant burden on the managers and the scientists involved in the case. In terms of actual work time, estimates were usually in terms of a few weeks over the course of a year. Respondents also underscored the time spent worrying about the progress and outcome of the case.

Our interviews do not comprehensively address the question of negotiation delays or litigation. Several respondents did, however, characterize the costs of sifting through a large number of potentially relevant patents and subsequent negotiations as substantial. One characterized the process as "complex, ongoing and labor intensive," but a cost of doing business. Another stated: "All these patents makes research more expensive. It can slow it down, while you secure licenses." One biotechnology executive suggested that about a third of his firm’s R&D projects suffer delays while licensing and related agreements are worked out. Another respondent gave more precise estimates, suggesting that it commonly took about three or four weeks to sift through the patents potentially relevant to a project, often identifying somewhere between five and twenty that may be worth investigating intensively over the course of another three to six months. The respondent noted, however, that the research itself would typically be moving forward during this time. At this point, it would be typically determined that there are about three to six patents where agreements were required, and these negotiations could affect the progress and direction of the firm’s R&D.

Although suggesting that IP reviews and negotiations are costly and time consuming, our Interviews--though limited on this point--provided no clear evidence that such delays or costs have increased over the past decade. To arrive at a quantitative estimate, we supplemented our interview data with data from AIPLA and BIO. The AIPLA’s Report of Economic Survey (AIPLA [1995, 1997, 2001]), however, reports the number of responding attorneys working in the area of biotechnology, and the median percent of effort dedicated to biotechnology by each respondent. Assuming the AIPLA’s data are representative (which they may not be with only a 18-20% response rate), they suggest just more than a 10% jump in the number of attorneys working on biotech between 1995 and 2001, and a 25% jump in the amount of time (at least per the median) that each attorney commonly dedicates to biotechnology. Therefore, there is roughly a 30% increase in resources devoted to what one might broadly construe as the "transaction costs" of filing, enforcing and contracting for patents. Ernst & Young LLP’s Annual Biotechnology Industry Reports suggest, however, that in nominal terms, R&D expenditures by biotechnology firms has increased over 80% during the 1994-2000 period. If we assume an annual R&D costs deflator as high as 5%, a lag of one year between the conduct of R&D and the generation of a patentable invention, and that the jump in biotechnology firm R&D expenditures reflects the growth in patentable inventions industrywide, then attorney activity per patentable invention appears to have actually declined in the recent past. We take this as further evidence that the patenting of research tools has not dramatically increased demand for legal resources, and by extension, that the transaction costs have not increased disproportionately.

Also bearing on the impacts of IP on transactions costs are the costs and delays associated with material transfer agreements, a particular focus of Eisenberg [1999] who identifies delays in negotiating access to research materials (MTA’s) to be an important cost. Although our interviews did not focus on such transactions, to the extent we considered them, they also suggested that MTAs are a source of some concern and vexation.²⁶ While the material being transferred may or may not be patented, the delays often involve negotiating an allocation of patent rights over the discoveries that may build on the material. Several respondents from universities and from industry confirmed that when dealing with MTAs, especially those involving university technology transfer offices, delays could be substantial. One industry respondent suggested about a third of the firm’s projects involved some research agreement with a university and that such negotiations result in substantial delays (on the order of months).

Our limited interviews on the subject suggest that the prospect of acquiring IP on biomedical discoveries accounts for little change in the likelihood of sharing materials since experimental biology has long been a competitive field where scientists were somewhat reluctant to share materials with competing scientists [see Hagstrom [1974], Campbell et al. [2000]). What has changed is that, when there is a willingness to share materials, processing that transfer had gone from taking a day or two, to taking months. Eisenberg, [1999] finds a similar result. This negotiation can delay the research since the research on the material cannot go forward until the material is transferred. One solution reported by our respondents is that some firms have a standard, take-it-or-leave-it agreement (cf. Eisenberg [1999]). This does reduce delays, but probably also reduces the number of transactions.

We heard repeated complaints about the inequality in the ability to bear the legal costs and risks of litigation affecting the terms of licensing contracts between large pharmaceutical and small biotechnology firms. While large firms frequently complained about small patent holders trying to hold them up with exorbitant demands, the large firms felt they could settle these by paying a small license fee, as insurance, or, failing that, they could say, "sue me," knowing that they have the ability to bankrupt a small firm through legal expenses. Indeed, respondents from larger firms expressed confidence that they could enforce lower terms on small patent holders, because the threat of litigation that underpins such negotiations is less credible when it comes from someone who is not in a position to invest millions of dollars in litigation. The small firms we talked to complained that, from their perspective, they were being "bullied" by large firms and were not able to collect the full value of their patented technology. Because they could ill afford to pay the high costs of litigation (either the out-of-pocket costs or the loss of managers’ or scientists’ time), they were forced to accept lower terms for their technology than they felt was appropriate. So, while contracting is a solution, the terms of the contract may reflect the relative bargaining leverage of the parties involved, and that leverage at least appears to bear some relation to firm size.²⁷

Without further study, it is difficult to assess the importance or implications for biomedical innovation of the asymmetric bargaining power between smaller biotechnology and large pharmaceutical firms in the industry. These characterizations of IP negotiations raise several questions. First, is it a matter for patent policy to redress such asymmetries? For example, to the extent that biotechnology firms and universities develop a disproportionate share of research tools, do these observations imply that patent scope should actually be expanded? Second, might such asymmetry, perhaps by diminishing uncertainty over bargaining outcomes, actually diminish transactions costs or the likelihood of bargaining breakdowns?

A related issue that was raised by universities, small firms and the lawyers who represent them was that access research tools might be prohibitively expensive. Large pharmaceutical or biotechnology firms may be able to afford such tools, but small firms or universities were denied access. This issue of access due to high prices was also prominently raised in the National Academy’s workshop on IP and biomedical innovation (National Research Council [1997]). One of our respondents suggested that, for example, "DNA chips are a high investment technology. Very small labs can't afford to do it. When the technology is out of reach of small labs, they have to collaborate. But, this collaboration generally means giving up IP rights. The technology forces collaboration because barriers to entry are high." This sentiment was echoed by some university researchers we talked with. This was one justification for the "do-it-yourself" solution of making patented laboratory technology without paying royalties (Marshall, [1999]). Similarly, the manager of a small biotech startup told us that Incyte's licensing terms for access to their gene database was several times the firm’s whole annual budget. They were forced to rely on the public databases, a viable but second best solution. One solution for universities has been the development of core facilities to share expensive resources such as chip-making facilities. Some firms (particularly genomics firms), however, offer discounted terms for university and government researchers.

As noted above, firms occasionally told us that they typically would not undertake or continue with projects once they learned that another firm had a patent on an end product (i.e., a therapeutic), and occasionally would redirect their research away from areas where there was a "tangle" of patents on research tools (cf. Lerner [1995]). Several respondents expressed the view that the technological opportunities in molecular biology and related fields were so rich and varied, that such redirection of research effort toward areas less encumbered by patents was not terribly costly for their firm or others. As one respondent put it "There are lots of targets, lots of diseases." Some have even suggested that the value of targets has declined substantially because companies can't exploit all of the ones they have, and so firms are more willing to license some of their targets, or even abandon some of their patents and let the inventions shift to the public domain.

We could well imagine that the social welfare costs of redirection of research effort to avoid IP entanglements may be less due to the presence of numerous research opportunities. Under plausible conditions, there can be excessive correlation in research portfolios to the degree that research bandwagons emerge around the mining of what may be considered the most promising veins. Under such circumstances, a shift to less crowded areas of art would be socially beneficial. One can also weave tales that suggest that such shifting, even in the presence of rich opportunities, may be socially inefficient. Yet, our conjecture is that the social welfare loss from the redirection of research to areas less encumbered by IP is not as great as it might otherwise be due to the high levels of technological opportunity. As a practical matter, it is, however, difficult to measure the extent to which projects were not started due to patent related concerns, and we have certainly not done that.

Although the conditions conducive to an anticommons tragedy exist, there is little evidence that such a tragedy itself has taken place on any significant level. The effective number of patent holders is rarely very large, and for the most part, they appear to be able to contract to realize the joint value of their intellectual property. These contracts take place in the shadow of the various options open to the potential licensee, such as inventing around or mounting a court challenge to the patent. Court challenges and even the contract negotiations themselves can, however, impose significant social costs. Litigation can be expensive and the non-out-of-pocket costs, in the form of the time devoted to the matter by researchers and management, can be substantial. Even when there is no court challenge, the negotiations can be long and complex and may impose delays, with implied costs. The complexity of the negotiation and contracting is not unexpected. However, it seems likely that in a substantial fraction of the cases, the development of standard contracts and templates may be helpful and funding agencies can play an important role in developing and encouraging the use of such standards (cf. Eisenberg [1999]).

In a substantial number of cases none of these costs apply. Where there is only a small likelihood of a valuable discovery, particularly in cases involving pre-commercial or academic research, the problem tend to be solved differently. Universities and firms simply infringe, and patent holders tacitly tolerate this. Infringing uses of research tools are often difficult to police, particularly when the research tool is not embodied in a physical product like a gene chip. Consequently, users contemplating non-commercial or low value uses are likely to ignore the patent on the tool altogether. One can rationalize the failure of the IP holder to aggressively monitor infringement as a form of price discrimination. Economic theory suggests that such price discrimination can improve social welfare. As long as the infringing uses do not reduce the value of the tool to the users with a high willingness to pay, such price discrimination is likely to be privately profitable as well.

Of course, there are always cases where access to the tool is restricted because the IP holder either charges a price higher than a particular user is willing to pay, or simply refuses to license the tool (cf. National Research Council [1997]). Should we consider such denial of access as an additional social cost of extending the patent system to research tools?

In thinking about this, it is helpful to distinguish between "general purpose" tools such PCR, microarrays, and combinatorial libraries, versus special purpose tools such as genes and receptors that are specific to particular diseases or particular therapeutic approaches to a disease. General purpose research tools are more likely to be non-rival in use in the sense that the use of the research tool by one firm will often not reduce others’ profits from using the tool. Here, the primary issue is how effectively the IP holder provides access to the tool, and the key concern is the ability and willingness of the IP holder to price discriminate in an efficient manner. There are cases where the IP holder cannot or does not develop a pricing strategy that allows low value and academic projects access to the tool, as for instance in the previously cited case of SmithKline-Beecham, which appears to have set too high and inflexible a royalty rate on its patent covering hemochromatosis testing, resulting in nearly one fifth of the labs surveyed by Merz not carrying out this particular test.

From a social welfare perspective, a general purpose tool is like a public good. It has a high fixed cost of development and zero marginal cost in serving an additional user. Thus, maximization of social welfare requires that the tool be made available to as large a set of users as possible. One way of achieving this entails charging higher prices to users with higher willingness to pay and lower prices to others such as researchers using the tool for non-commercial purposes--a strategy which approximates at least some of the current practice.

On the other end of the continuum, where the tools is specific to a therapeutic or diagnostic function, uses tend to be rival. For instance in the case of a receptor which is specific to a particular therapeutic approach to a disease, if one firm finds a compound that blocks the receptor, it undermines another firms ability to profit from its compound that blocks the same receptor. Where the IP holder also participates in the downstream use of such a special purpose tool (i.e., in developing a therapeutic or diagnostic product), it may refuse to provide access to.29 anyone else. The social welfare analysis is, however, not straightforward in this case. Even though knowledge, once developed, can be shared at little additional cost, it does not follow that social welfare would be maximized by providing broad access to the tool. Consider an extreme example. A firm patents a receptor which is specific to a particular disease. Insofar as there is a unique molecule that blocks this receptor and the patent holder is as well qualified as any to search for the molecule, there is no loss of efficiency implied by a failure to license the receptor.

This example also suggests why narrow uses of special purpose research tools may entail social welfare losses, especially if the patent on such a special purpose research tool is broad in scope. It is likely that there are a variety of lines of attack, not a single one. It is also likely that no single firm is capable of mounting all the potentially fruitful lines of attack, or for that matter, even be able to conceive of all the varieties. Indeed, Rosenberg has persuasively argued that the history of technical advance shows that it is often impossible to anticipate all or even the most important uses of a discovery. As Merges and Nelson [1990, 1994] suggest, the problem--which is more acute the broader is the scope of the patent--is that the holder of a patent on a special purpose tool such as a target may not be able to identify or come to terms with all those able to use the research tool in ways that add value, consequently denying society its best chance of realizing the full impact of that discovery on technical advance.²⁸ Thus, commonly observed licensing practices for special purpose research tools pose a problem, though not of an anti-commons character.²⁹

In summary, we find that patenting of research tools has made the patent landscape more complex. There are more patents and more patent holders involved in any commercializable innovation in biomedicine. Many of these patents relate to research tools, some of which are general purpose while others are specific to an innovation. However, we do not find evidence that this increased complexity has given rise to a tragedy of the anti-commons.

Rarely was a research project stopped because of difficulties of access to a research tool. The effective number of patent holders is not as large as it appears on the surface, and firms appear to have used contracts effectively. In general, fears about inability to contract and transaction costs overwhelming the value from contracts have been largely unrealized.

We do not observe as much breakdown as one might expect because firms have been able to develop "working solutions" that allows their research to proceed. These working solutions combine taking licenses, inventing around patents, infringement (sometimes under an informal and typically self-proclaimed research exemption), the development and use of public databases and court challenges. In addition, changes in the institutional environment, particularly new PTO guidelines and some shift in the courts' views toward research tool patents, as well as pressure from powerful actors such as NIH, appear to have further reduced the threat of breakdown. Finally, the very high technological opportunity in this industry means that firms can readily shift their research to areas not encumbered by intellectual property claims, and, therefore, the walling off of some areas of research may not exact a high toll on social welfare.

Increased complexity imposes social welfare costs in a variety of ways. Negotiations and contracting are costly and sometimes can impose substantial delays. Disagreements can and have lead to litigation, which are especially costly for small firms and universities. In some cases, firms redirected their research efforts away from projects where the firm anticipated that it would be difficult or expensive to obtain access to all the intellectual property, including patent protected research tools. Firms also tried to navigate the increased complexity by circumventing patents, using substitute research tools, inventing around or going offshore. These responses, while privately rational, constitute a social waste. Limits on access to targets and other special purpose research tools are also a source of concern.

Having dwelt on the costs of current patent policy and practices in biomedicine, we should be mindful of the benefits. Patents benefit drug innovation broadly, providing important incentives that have called forth enormous investment. Respondents suggest patenting benefits generation of research tools in particular and also claim that others’ patents facilitate the research enterprise by increasing the availability of useful tools. Moreover, a broad consensus in the literature (e.g., Henderson et al. [1999]) as well as among our respondents suggest that patented research tools such as the Cohen-Boyer technology for recombinant DNA, PCR, genomics, and so on have also increased the productivity of biomedical research enormously.³⁰

Thus, there is some cause for optimism. We cannot rule out future problems resulting from patents currently under review or new shifts in technology or in court decisions. Yet, our conclusion is that, despite difficulties that should be attended to, the industry seems to be succeeding in developing an accommodation that incorporates both the need to provide strong incentives to conduct research and the need to maintain free space for discovery.

American Intellectual Property Law Association (AIPLA), 1995, 1997, 2001, "Report of the Economic Survey," AIPLA, Washington, D.C.

Barton, J. 2000, "Intellectual Property Rights. Reforming the Patent System," Science, V. 287, pp. 1933-34.

Beck, S. 1998. "Do you have a license?: products licensed for PCR in research applications." The Scientist 12 [12]: 21, June 8, 1998.

Campbell EG, Weissman JS, Causino N, Blumenthal D. 2000. Data withholding in academic medicine: characteristics of faculty denied access to research results and biomaterials. RESEARCH POLICY 29 (2): 303-312 FEB 2000

Cockburn, I., R. Henderson, L. Orsenigo, and G.P. Pisano, 2000, "Pharmaceuticals and Biotechnology," in National Research Council (D.C. Mowery, ed.), U.S. Industry in 2000: Studies in Competitive Performance, Washington, D.C.: National Academy Press Cohen, J. 1999. "Chiron stakes out its territory." Science 285 [2 Jul 1999]:28.

W.M. Cohen and S. Klepper, 1992, "The Tradeoff between Firm Size and Diversity for Technological Progress," Journal of Small Business Economics, December.

Cohen, W.C., Nelson, R.R., and Walsh, J.P. 2000. "Protecting their intellectual assets: appropriability conditions and why U.S. manufacturing firms patent (or not)." NBER Working Paper 7522.

Cohen, W.C., Goto, A., Nagata, A., Nelson, R.R., and Walsh, J.P. 2001. "R&D spillovers, patents and the incentives to innovate in Japan and the United States." Carnegie Mellon University Working Paper.

Doll, J.J., 1998, "The Patenting of DNA," Science 280, May 1, pp. 689-690.

Drews, J. 2000. "Drug discovery: a historical perspective." Science 287 [2000 March 17]:1960-1962.

Eisenberg, R.S., 1999, "Bargaining over the Transfer of Proprietary Research Tools: Is this Market Failing or Emerging?" University of Michigan, mimeo.

Evenson, R. and Y. Kislev 1973, "Research and productivity in wheat and maize," Journal of Political Economy, V. 81, pp. 1309-1329.

Hall, Bronwyn H. and Rosemary Ham Ziedonis. 2001. "The Patent Paradox Revisited: An Empirical Study of Patenting in the U.S. Semiconductor Industry, 1979-1995" RAND Journal of Economics. 32:1, pp. 101-128.

Heller, M.A. and Eisenberg, R.S. 1998. "Can patents deter innovation? The anticommons in biomedical research." Science 280 [1998 May 1]: 698-701.

Henderson, R., L. Orsenigo, and G.P. Pisano, 1999, "The Pharmaceutical Industry and the Revolution in Molecular Biology: Interactions among Scientific, Institutional and Organizational Change," in D.C. Mowery and R.R. Nelson, Sources of Industrial Leadership: Studies of Seven Industries, New York: Cambridge University Press.

Jewkes, J., D. Sawers and Stillerman 1958, The Sources of Invention, London: Macmillan Kash, D.E. 2000. "Patents in a world of complex technologies." Paper presented to NAS STEP Board Conference on Intellectual Property Rights: How Far Should They Be Extended?, Washington, DC, February 2-3, 2000.

Kornberg, A. 1995. The Golden Helix. Sausilito, CA: University Science Books.

Kryder, R. David, Stanley P. Kowalski and Anatole F. Krattiger 2000. The Intellectual and Technical Property Components of pro-Vitamin A Rice (GoldenRiceTM): A Preliminary Freedom-To-Operate Review. ISAAA Brief No. 20-2000.

Lerner, Josh. 1995. "Patenting in the Shadow of Competitors" Journal of Law and Economics. 38:2, pp. 463-495.

Levin, R. 1982, "The Semiconductor Industry," in R.R. Nelson, ed., Government and Technical Progress: A Cross-Industry Analysis," New York: Pergamon Press.

Levin, R., Klevorick, A., Nelson, R.R. and Winter, S.G. 1987. "Appropriating the returns from industrial R&D," Brookings Papers on Economic Activity, 783-820.

Mansfield, E. 1986. "Patents and innovation: an empirical study." Management Science 32:173-181.

Marshall, E. 1997. "Ethics in science: is data-hoarding slowing the assault on pathogens?" Science 275 [7 Feb 1997]: 777-780.

Marshall, E. 1999. "Drug firms to create public database of genetic mutations." Science [16 April 1999] 284:406-407.

Marshall, E. 1999. "Patent on HIV receptor provokes an outcry." Science 287.

Marshall, E. 2000. "Property claims: a deluge of patents creates legal hassles for research." Science 288 [2000 April 14]: 255-257.

Marshall, E. 2001. "Bermuda rules: community spirit, with teeth." Science 291 [16 Feb 2001]: 1192.

Merges, R. 1994, "Intellectual Property Rights and Bargaining Breakdown: the Case of Blocking Patents," Tennessee Law Review, V. 62, No. 1, pp. 74-106.

Merges, R. 1999. Remarks to the Board on Science, Technology, and Economic Policy, National Research Council, Workshop on Intellectual Property Rights Research. Berkeley, CA, October 19, 1999.

Merges, Robert P. and Richard R. Nelson. 1990. "On the Complex Economics of Patent Scope" Columbia Law Review. 90:4, pp. 839-916.

Merges, R.P. and R.R. Nelson, 1994, "On limiting or encouraging rivalry in technical progress: the effect of patent scope decisions," Journal of Economic Behavior and Organization, V. 25, pp. 1-24.

National Research Council, 1997, "Intellectual Property Rights and Research Tools in Molecular Biology," Washington, D.C.: National Academy of Sciences.

NSF [National Science Foundation]. 1998. Science and Engineering Indicators. Washington: U.S. GPO.

Nelson, R.R., 1961, "Uncertainty, learning, and the economics of parallel research and development efforts," Review of Economics and Statistics, Vol 43, pp. 351-364.

Nelson, R.R. 1982, "The role of knowledge in R&D efficiency," Quarterly Journal of Economics, V. 97, pp. 453-470.

Pollack, A. 2000 "Battling Searle, university gets broad patent for new painkiller." New York Times, April 12, 2000, C1-2.

Regalado, A. 2000. "The great gene grab." Technology Review September/October: 49-55.

Science. 1998. "Biochip-makers do battle in court." Science 279 [1998 January 16]: 311b.

Scherer, F.M. et al., 1959, "Patents and the Corporation," 2^nd edn., Boston, privately published.

Scotchmer, Suzanne. 1991. "Standing on the Shoulders of Giants: Cumulative Research and the Patent Law" Journal of Economic Perspectives. 5:1, pp. 29-41.

Seide, R.K. and J.M. MacLeod. 1998. "Comment on Heller and Eisenberg." ScienceOnline. http://www.sciencemag.org/feature/data/980465/seide.shl.

Service, R.F. 1998. "DNA analysis: will patent fights hold DNA chips hostage?" Science 282 [1998 October 16]: 397.

Service, R.F. 2001. "Can Data Banks Tally Profits?" Science 291 [16 Feb 2001]:1203. USPTO, 2001. http://www.uspto.gov/web/offices/com/speeches/01-06.htm.

Shapiro, Carl. 2000. "Navigating the Patent Thicket: Cross Licenses, Patent Pools, and Standard-Setting," in A. Jaffe, J. Lerner and S. Stern, eds., Innovation Policy and the Economy, Cambridge: MIT Press.

Williamson, O.E. 1979. "Transaction cost economics: the governance of contractual relations." Journal of Law and Economics 22:233-261.

Wolff, G., 2001, The Biotech Investor’s Bible, New York, John Wiley and Sons.

1 | See Scherer et al [1959], Levin, et al. [1987]; Mansfield [1986] and Cohen et al. [2000]. For pharmaceuticals, there is near universal agreement among our respondents that patent rights are critical to providing the incentive to conduct R&D. Indeed, data from the Carnegie Mellon Survey of Industrial R&D (cf. Cohen et al. [2000]) show that the average imitation lag for the drug industry is nearly 5 years for patented products, while for the rest of the manufacturing sector, the average is just over 3.5 years (p<.01). Our interviewees also noted that in the case of small firms, patent ownership is often their critical asset and that access to funding (particularly from venture capital) hinges on their patent portfolio.

2 | The qualifier reflects the fact that tools may also be used to rule out unsuccessful lines of research and thereby increase the efficiency of research and reduce research costs.

3 | One must carefully distinguish between constraints on access to research tools that we would expect to accompany self-interested use of patents on research tools and other upstream discoveries (considered by Merges and Nelson [1990] and others), versus inefficiencies that arise from failures to effect mutually beneficial trades that would provide access to research tools (emphasized by Heller and Eisenberg [1998]). The former reflects the rational playing out of private incentives, while the latter reflects an apparent frustration of such.

4 | From 1990 to 1997, there were an average of 379 such licenses each year in the drug and chemicals industries (SIC28). For comparison, during this same period there were an average of only 276 such licenses per year in electronics (SIC36).

5 | An executive with a biotechnology firm explicitly compared its patenting strategy with Japanese firms in complex product industries: "We have a defensive patent program in genomics. It is the same as in the Japanese electronics industry. There they patent every nut and screw on a copier, camera, and build a huge portfolio, so Sony never sues Panasonic and Panasonic never sues Sony. There is a little of that going on in genomics. That way, if an IP issue ever arose, we have some cards in our hand." A respondent from a large pharmaceutical firm made a similar comment: "I supposed because we see everyone else doing it [patenting research tools] in part. Sort of like the great Oklahoma Land Rush. If you don't do it you’re not going to have any place to set up a tent, eventually. That would be true for 20 years. Then you come back and everything will be in the public domain."

6 | A biotechnology executive responsible for IP states:

The patent landscape has gotten much more complex in the 11 years I've been here. I tell the story that when I started and we were interested in assessing the third party patent situation, back then, it consisted of looking at [4 or 5 named firms]. If none were working on it, that was the extent of due diligence. Now, it is a routine matter that when I ask for some search for third party patents, it is not unusual to get an inch or two thick printout filled with patent applications and granted patents. With the growth of biotech around the world, you have greater numbers and they overlap to a greater extent. The difference between different patents can be very subtle. This is a result of many different entities, universities, for profits, nonprofits, often doing research in the same areas. The results are overlapping patents that may be similar. The inter-relationship between different patents is very complex and it takes sophisticated analysis to discern the differences. In addition to dealing with patents over the end product, there are a multitude of patents, potentially, related to intermediate research tools that you may be concerned with as well. Consequently, you run the risk of stacking royalties on products you are trying to develop. The royalty burden can become onerous."

This example is interesting for several reasons. The first is that the respondent notes the substantial increase in relevant patents to be considered (and the growing number of patenting institutions). Second, he notes there is an increase in overlap among patents, such that several apply to the same innovation. Third, he notes the problem with royalty stacking.

7 | Moreover, detailed study of the proprietary landscape noted that, depending on the country, and the technologies that are used, the number of patents in fact could vary from 40 (in the U.S. or Europe) to zero (in, for example, Thailand, Bangladesh, Myanmar, Malaysia, Iran, Iraq, Saudi Arabia, or Nigeria).

8 | The following respondent from a pharmaceutical firm expressed their frustration:

We do have frustration internally because we can't do what we consider basic research with a cloned gene, not selling the gene, just using it to make another discovery. To be cut off from that, it sits badly. Because, at the end of the day, you are cut off from tools, from making a breakthrough discovery. Because there is a patent on the human gene, you work with the guinea pig gene, but it is not the best approach. That’s very frustrating. In a number of cases, we can't work with this protein or this gene and it slows things down. We are looking at ways to get around this. How to not infringe their IP. And, we are coming up with ways to do that, but it involves some labor and time.

9 | Testimony before House Judiciary Committee, 7/13/00, http://www.house.gove/judiciary/henn0713.htm.

10 | We also had one respondent, an IP lawyer, who said such cases where projects were stopped existed, but client privilege prevented giving details.

11 | In response to a question of whether their firm ever had a case of a project being stopped for problems with license stacking, a biotechnology respondent stated: "No. It would be hard to find such a case, given the reality of how decisions made. It is not a late stage decisions. At the preclinical stage, you find you have 10 candidates, and you can afford to continue work on 3. The decision is a complex prediction based on the potential for technical success, the cost of manufacturing, the size of the market, what you can charge, what you need to put in for royalties. I am not familiar with royalty stack being the deciding factor. The probability of technical success and the size of market are key. The numbers on royalty stacking are fairly predicable, compared to the size of the market, where there is much more variance around the median." This response emphasizes high technological opportunities, with the firm facing many more targets than it can afford to pursue.

12 | One respondent gave the following account, "[The royalty] is a factor to consider. I don't want to say a worthwhile therapeutic was not developed because of stacking problems. But if we have two equally viable candidates, then we choose based on royalties. Also, if there is a patent on manufacturing in different host cells, and certain others that don't have a royalty, then, on the last step, you don't make it in this cell, but over here. But that incurs some technical costs. It is a different system, you are not as familiar with the technology, but you go there because you don't want to pay the royalty."

13 | Recently, however, this university did agree to engage in negotiations over the use of a research tool over which a firm had rights.

14 | Testimony before House Judiciary Committee, 7/13/00, http://www.house.gove/judiciary/merz0713.htm.

15 | One controversial case was the diagnostic test for the Canavan disease gene mutation. Miami Children’s Hospital held the patent and was charging a $12.50 per test royalty, even though the doctor did the test himself. Washington University’s Michael Watson was among those complaining that this royalty hurts research and patient care: "We would be happy to pay for some kind of test kit that is faster, better, cheaper. But they are trying to control manual testing, which is not appropriate." (Regalado [2000, p. 55). This quote reflects the opinion of many academics that they should not be forced to pay royalties for "do-it-yourself" technologies. In this particular case, we also have a sense of the scale of the royalty demanded.

16 | For example, Heller and Eisenberg [1998] use the case of "adrenergic receptor" claims as an illustration of the anti-commons problem and find over 100 patents that might require a license to do research in this area. In a response to the Heller and Eisenberg article, Seide and MacLeod [1998] did a search on "adrenergic receptor" and, indeed, find 135 patents using this term. They then do an (admittedly cursory) patent clearance review and find that the vast majority would not in fact be infringed by an assay to screen for ligands against this receptor and that, at most, only a small number of licenses might be required.

17 | Furthermore, Incyte's license requires users to "grant back" non-exclusive rights to use of genes discovered from its database, providing freedom of operation for firms in the network and creating what Incyte refers to as an "IP Trust." Similarly, the Cohen-Boyer patent was frequently noted by respondents as an example of a research tool that was widely licensed on a non-exclusive basis at what are often characterized as "reasonable" terms.

18 | For example, Athersys, a Cleveland based biotech firm, advertises its RAGE technology as one where one use automated techniques to create protein expression libraries (i.e., activate and express every gene and therefore produce every protein) without using any knowledge about the location and structure of the corresponding gene. Published reports indicated that some established pharmaceutical firms had licensed this technology.

19 | As one university technology transfer officer put it, "Asserting against a university doesn't make sense. First, there are no damages. You can get injunctive relief and/or damages. What have you gained? You've just made people mad. Also, these firms are consumers of technology as well. No one will talk to you if you sue. We all scratch each others' backs. You will become an instant pariah if you sue a university." Similar, from the industry side Leon Rosenberg from Bristol-Myers Squibb said, "Frankly, we all know it is not good form to sue researchers in academic institutions and stifle their progress." (NRC [1997, Ch.6, p. 3]).

20 | Some universities feel that they have the moral right to assert against another university if, however, it is commercializing their innovation. "Universities have a general agreement of sharing results freely. There is a research ethic. But, if a university were making money off of our technology, then there would be trouble."

21 | One respondent stated, "Sometimes we take a license and sometimes we don't. I think there is a lot of infringement out there. The scientists are not telling their patent counsel." One respondent was explicit: "If you are confronted with a patent on a target you need, you have to decide what to do. You can infringe, and take the risk of getting sued. They would have to know your practices. If you keep it secret, then they may only find out when you release a product. Then they may know you used it. But, the statute of limitations may have run out. Some research tool owners are very aggressive. If they get a hint you are using their tool, they sue. You take all this into account." Another stated, "I think all the firms in the industry take on some infringement risk, because the behavior in the industry is that you have to try a million things to find one that is promising. Once you identify the promising candidate, then you look into licensing the research tools or sequences you used."

22 | For example, the firms in the SNPs Consortium include Bayer, Bristol-Meyers Squibb, Glaxo Wellcome, Hoechst Marion Roussel, Monstanto, Novartis, Pfizer, Roche, SmithKline Beecham and Zaneca. Each firm contributed $3 million, and Wellcome Trust added another $14 million to the effort.

23 | One could speculate that the shift of HGS away from the database business and toward the drug development business may be a response to both the higher returns available to drug companies and the lower returns available to genomics companies that are competing with increasingly developed public databases.

24 | Academics have the right to access the data at no charge, do searches and download segments up to 1 megabase, publish, and patent. They can download the whole database if the university signs an agreement not to redistribute the data. There are no reach through provisions or restrictions on publication. Science also kept a copy of the database in escrow to ensure compliance.

25 | One R&D executive stated: "In lay terms, when you are looking for a therapeutic to treat a particular disease, there are certain immutable facts of the disease and the human body that you have to deal with, that you can’t circumvent. In electronics, apart from violating fundamental laws of physics, there are probably lots of different ways to develop a semiconductor. When you deal with disease, the disease is what it is. That limits the number of interventions, based on nature. You are more limited in the number of approaches when you treat a disease that when you engineer a bridge, or a semiconductor, or a piece of software. Given that, if someone had a patent that covered most of the fundamental aspects of a disease, it is hard to invent around. For example, you identify that a particular cancer is caused by over-expression of a particular cell surface receptor. Something goes wrong in the DNA and makes a lot of a receptor, a protein on the surface of cells, and then when that happens cancer develops. The most obvious way is to intervene at the protein receptor. If you had a patent covering that protein receptor broadly or drugs that interact with that receptor, you've got the field covered for drugs for that form of cancer. It's hard to get around your patent. Those sorts of patents are not so frequent, in part because the patent office, in recent years, adopted new written description and utility guidelines."

26 | The following quote reinforces Eisenberg's findings: "Things are becoming more bureaucratic. MTAs, they are crazy. Before, whenever someone wanted a plasmid from my lab, I would just send it. Now, the university says they own it and I have to go through the IP office. It goes back and forth between the two offices and it takes a long time. Before, we would just send it in the mail, and you would have it and could use it. Basic science is now becoming interested in "value." The university is particularly interested in value."

27 | In response to the question of why they might litigate, a university technology transfer officer stated: "The realpolitik involves a certain calculus. How much time, cost, and pain you can inflict on the other guy. I’ll give you a clean example that I know well. University of Minnesota had a set of patents on carbovir, an antiviral drug that is used for AIDS treatment. Somehow, GlaxoWellman ends up with the license to these patents. The license is at a high royalty rate, 10% or so. The lifetime royalty was estimated to be about $600 million. Glaxo says, "We have the license, but we won’t pay. Sue us." They know that for the university to sue a big pharma, or anyone, is traumatic and counter to their mission. Glaxo want to renegotiate the royalty. It costs them a couple of million to litigate. If they can get a fraction reduction in the royalty, they will save money. The case got settled. They got a 50% reduction. They spent 2 to 5 million and saved 300 million."

28 | The premise of this argument, well recognized in the economics of innovation (Jewkes, Sawers and Stillerman [1958], Evenson and Kislev [1976], Nelson [1982]) is that, given a technological objective (e.g., curing a disease) and uncertainty about the best way to attain it, that objective will be most effectively met to the extent that a greater number of approaches to it are pursued.

29 | While one might argue that such a case also reflects a failure to enter into mutually profitable contracts, it is also true that the problem is not one of a large numbers negotiation.

30 | Incyte's Randall Scott offered the following examples of the productivity benefits of genomics: "In 1999, CV Therapeutics, an Incyte collaborator, was able to use Incyte gene expression technology, information about the structure of a known transporter gene, and chromosomal mapping locations, to identify the key gene associated with Tangiers disease. This discovery took place over a matter of only a few weeks, due to the power of these new genomics technologies. ... An Incyte customer stated that it had reduced the time associated with target discovery and validation from 36 months to 18 months, through use of Incyte's genomic information database. Other Incyte customers have privately reported similar experiences. ... One Incyte customer stated that by using Incyte's database, it quickly discovered a new histamine receptor gene which had long eluded researchers, and which is being used to develop an effective drug that is specific for brain tissue. In fact, after isolating the gene and using high-throughput screening, a candidate drug was identified in less than a month. Again, by making new targets available to the pharmaceutical industry, Incyte helped the company go from picking a target receptor to developing a potential drug in just 18 months, a process that typically takes five years or more, clearly accelerating the drug discovery process by three-fold or more." [testimony before House Judiciary Committee, 7/13/00 http://www.house.gov/judiciary/scot0713.htm].