Grasping the Slim Tail of Innovative Success: Biotechnology Patenting from 1990 to 2004
David E. Adelman & Kathryn DeAngelis1
The current alarm over the dramatic rise in patents granted each year by the United States
Patent and Trademark Office (“PTO”) springs from a deceptively simple question—do we have
too much of a good thing?
In at least one respect, the answer is an unequivocal “yes.” No one doubts that the PTO
is overwhelmed by the torrent of patents that inventors are filing.2 Beyond this point of
agreement, however, views of legal commentators diverge. Two opposing camps standout in
this battle: the anti-commons contingent, which carries the banner of restraint;3 and the
transactionalists, who are bullish about patents and favor the status quo.4 The two camps come
to diametrically opposed conclusions about the effect of patents on transaction costs. Anticommons proponents believe that the surge in patents issued by the PTO risks fatally raising the
transaction costs of innovation.5 Whereas optimism among transactionalists is leavened by their
view that, far from impeding transactions, patents are essential to fostering efficient market
transfers of valuable inventions.6
1
David E. Adelman is an Association Professor at the Rogers College of Law, University of Arizona;
Kathryn DeAngelis is a 2005 graduate of the Rogers College of Law, University of Arizona, and an association in
the San Diego office of Gray Cary Ware & Freidenrich.
2
The Patent System Today and Tomorrow Before Subcommittee on Intellectual Property of the Senate
Committee on the Judiciary, 109th Cong. __ (2005) (statement of John W. Dudas, Under Secretary of Commerce for
Intellectual Property and Direct of the U.S. Patent and Trademark Office); Mark A. Lemley, Rational Ignorance at
the Patent Office, 95 NW. U. L. REV. 1495, 1495-96 (2001) (commenting on the numerous criticisms of the PTO’s
patent prosecution process).
3
See, e.g., Michael A. Heller & Rebecca S. Eisenberg, Can Patents Deter Innovation? The Anticommons in
Biomedical Research, 280 SCIENCE 698, 698 (1998); Rebecca S. Eisenberg, A Technology Policy Perspective on the
NIH Gene Patenting Controversy, 55 U. PITT. L. REV. 633, 640 (1994).
4
See, e.g., Robert R. Merges, A Transactional View of Property Rights (March 10, 2005) available at
<http://ssrn.com/abstract=707202>; Robert R. Merges, Contracting into Liability Rules: Intellectual Property Rights
and Collective Rights Organizations, Cingal. L. Rev. 1293 (1996).
5
Heller, supra note 3, at 698 (arguing that when numerous patent rights are broadly dispersed among
different entities no single entity hwill have access to the technology needed to conduct research and development).
6
Merges, supra note 4, at 8-9 (claiming that strong intellectual property rights are necessary to facilitate
market transactions, making research and development more efficient and plentiful).
Much of this battle has been waged with anecdotal evidence, although several recent
empirical studies have begun to fill this void.7 This paper presents a comprehensive empirical
analysis of U.S. biotechnology patents, which, because of their perceived social and economic
importance, have provoked exceptionally heated debate.8 Beyond its high profile, biotechnology
offers a rich context for examining patent policy. Rapid scientific advances during the 1990s
stimulated large influxes of private funding for research and rapid growth in the number of
biotechnology patents issued.9 At the same time, universities and research institutes increased
their patent filings, blurring the line between commercial and basic-science research.10 These
economic and scientific developments have focused attention on biotechnology patents and
propelled a parallel debate over the prudence of allowing patents on basic-science research tools,
or “upstream technologies.”11
7
Very few data of any kind were available until the late 1990s when several major studies were initiated,
including the following: Bronwyn H. Hall, et al., The NBER Patent Citations DATA File: Lessons, Insights and
Methodological Tools, Working Paper 8498, available at http://www.nber.org/papers/w8498; John R. Allison, &
Mark A. Lemley, Who’s Patenting What? An Empirical Exploration of Patent Prosecution, 53 Vand. L. Rev. 2099
(2000); Kimberly A. Moore, Judges, Juries, and Patent Cases—An Empirical Peek Inside the Black Box, 99 Mich.
L. Rev. 365 (2000-01); Jonathan A. Barney, A Study of Patent Mortality Rates: Using Statistical Survival Analysis
to Rate and Value Patent Assets, 30 AIPLA Q.J. 317 (2002); Jean Lanjouw & Mark Shankerman, Enforcement of
Patent Rights in the United States in Patents in the Knowledge-Based Economy (Wesley M. Cohen & Stephen A.
Merrill, eds., 2003); John R. Allison, Valuable Patents, 92 Geo L.J. 435 (2004); Josh Lerner, Patenting in the
Shadows of Competition, 38 J.L. & Econ. 463 (1995).
8
See, e.g., John P. Walsh, et al., Working Through the Patent Problem, 299 Science 1021, 1021 (2003); John
P. Walsh, Ashish Arora, and Wesley M. Cohen, Effects of Research Tool Patents and Licensing on Biomedical
Innovation, in PATENTS IN THE KNOWLEDGE-BASED ECONOMY 285 (Wesley M. Cohen and Stephen A. Merrell eds.,
2003); Arti K. Rai, Regulating Scientific Research: Intellectual Property Rights and the Norms of Science, 94 NW.
U. L. REV. 77, 90-92 (1999); Rebecca S. Eisenberg, A Technology Policy Perspective on the NIH Gene Patenting
Controversy, 55 U. PITT. L. REV. 633, 640 (1994).
9
Id. at 227 n.15.
10
Walsh et al., supra note 8, at 295.
11
Id.; NATIONAL ACADEMIES OF SCIENCE, A PATENT SYSTEM FOR THE 21ST CENTURY 17, 20 (Stephen A.
Merrill et al., eds., 2004) [hereinafter NAS REPORT]. Archetype examples of upstream technologies are the famous
Cohen-Boyer patent, which covered the canonical methods for replicating and expressing foreign genes in
microorganisms, and the polymerase chain reaction (“PCR”) for copying DNA. Rebecca S. Eisenberg, Bargaining
Over the Transfer of Proprietary Research Tools: Is This Market Failing of Emerging?, in EXPANDING THE
BOUNDARIES OF INTELLECTUAL PROPERTY: INNOVATION POLICY FOR THE KNOWLEDGE SOCIETY 229-30 (Rochelle
Cooper Dreyfuss et al., eds., 2001).
Our empirical study is based on a dataset comprised of biotechnology patents granted
from January 1990 through the end of 2004, about 53,000 patents in all.12 This work represents
the first time that a comprehensive dataset of biotechnology patents has been compiled. The
dataset is unique in that it covers the period of most dramatic rise in biotechnology patenting, as
well as important shifts in PTO policy towards more stringent standards for patents on genetic
sequences and a significant retrenching of the biotechnology financial markets.13
The data expose several new characteristics of biotechnology patenting so far unresolved
by existing empirical studies of U.S. patents. At the broadest level, we find that the number of
biotechnology patents issued per year increased by more than 750 percent between 1990 and
1998. More surprisingly, despite a forty-six percent increase in biotechnology applications
during the past five years,14 we observe a twenty-nine percent decline in the number of
biotechnology patents issued over roughly the same period. These opposing trends reveal the
otherwise latent power of the PTO to cull patents during the prosecution process. The data are
revealing another respect as well. We find that biotechnology patenting is dominated by patents
on methods or processes, as opposed to patents on genetic or protein sequences.15 This finding is
significant in itself given that much of the controversy surrounding upstream biotechnology
patents has centered on the patenting of genes and proteins.16
12
See infra Part I.
Ashish Arora, et al., MARKETS FOR TECHNOLOGY: THE ECONOMICS OF INNOVATION AND CORPORATE
STRATEGY 67 (2001) (observing that equity financing of biotech companies declined significantly between 1997 and
1998); see infra Part II.A. (describing the PTO changes to its regulations regarding certain types of biotechnology
patents during the 1990s).
14
Dudas, supra note 2, at __.
15
Other commentators have predicted this as well. Arora, supra note 13, at 160-61 (claiming that
biotechnology companies have focused on tool development).
16
See, e.g., Rebecca S. Eisenberg, Intellectual Property at the Public-Private Divide: The Case of LargeScale cDNA Sequencing, 3 U. CHI ROUNDTABLE 557, 569-70 (1996), citing David Dickson, 'Gene Map' Plan
Highlights Dispute over Public vs Private Interest, 371 Nature 365 (1994); Rebecca S. Eisenberg, Public Research
and Private Development: Patents and Technology Transfer in Government-Sponsored Research, 82 Va. L. Rev.
1663 (1996); Maxwell J. Mehlman & Jeffrey R. Botkin, ACCESS TO THE GENOME: THE CHALLENGE TO EQUALITY
(1998).
13
One of the most interesting results of the study is the degree to which ownership of
biotechnology patents is diffuse. We find that even the largest biotechnology companies are on
average granted fewer than twenty biotechnology patents per year. Our data also show that the
number of entities obtaining biotechnology patents has monotonically increased over the fifteenyear period covered by the dataset. Interpreting these trends is necessarily impressionistic, but
the lack of concentrated control, rising number of patent applications, and the continuous record
of new market entrants are positive signs that biotechnology patenting is not adversely affecting
innovation. Further, while the large number and broad spread of patents among different entities
raises the specter of patent anti-commons, we find little support for this theory in our analysis of
the distribution of patents across distinct biotechnology subfields.17
Beyond these descriptive statistics, our study reveals the stark analytical barriers limiting
the PTO’s options for addressing the growing backlog of patent applications.18 Among legal
commentators, a seemingly obvious target for reversing the patent backlog is the substantial
majority of patents—some data suggest more than ninety-five percent19—that have little or no
economic value.20 Their logic is simple. Because the slim tail of high-value patents accounts
disproportionately for the success of the U.S. patent system, valuable patents should receive
special consideration from the PTO. This strategy plainly requires that valuable patents be
identifiable ex ante. However, as Harvard economist F.M. Scherer has shown, the “innovation
17
Many commentators point to the rapid increase in patent filings since the mid-1980s as prime facie
evidence that patent policies are failing in a more fundamental way. They infer from this increase and anecdotal
accounts that the number and uncertainties of patent rights—so called patent anti-commons and thickets—threaten
to retard innovation. By fixating on the gross numbers and anecdotal evidence, these accounts fail to consider
whether the biotechnology scientific commons is congested. As it turns out, biomedical science appears to have a
high capacity for patenting because it is still relatively undeveloped and research opportunities far exceed existing
scientific resources. David E. Adelman, A Fallacy of the Commons in Biotech Patent Policy, 20 BERKELEY TECH.
L.J. 985 (2005).
18
Dudas, supra note 2, at 2.
19
Lemley, supra note 2, at 1507 (estimating that “the total number of patents litigated or licensed for a
royalty (as opposed to cross-license) is on the order of five percent of issued patents”).
20
See, e.g., Allison, Valuable Patents, supra note 7, at 439, 464-65.
lottery” is chaotic and impervious to standard statistical methods that would otherwise allow the
PTO to prioritize its patent review process along these lines.21
A recent empirical article, “Valuable Patents,” adopts a refreshingly defiant view of
Scherer’s notorious finding.22 It rejects Scherer’s model of the innovation lottery and, at least
implicitly, the empirical studies on which it is based.23 While the authors acknowledge that
“[t]here is something to [Scherer’s] idea,” they counter that:
. . . if valuable patents can be reliably identified at the time of application, or at least at
the time of issue, the lottery theory runs into difficulty. At best, it becomes only a partial
explanation—patentees may identify some clearly valuable patents, and may also apply
for other patents in the hope that they might pay off. That many valuable patents can be
identified early may also suggest that an opposition system is feasible because
competitors can rely on objective characteristics to determine early on which patents they
should challenge.24
There is an air of hedging in this declaration against randomness, but the logical premise of their
opposition—causal consistency and chaos are incompatible—is certainly correct. Extending this
logic, the authors also rightly acknowledge that rational patent policies are contingent on
predictive patent characteristics existing. By contrast, the rare innovative strikes contemplated
by Scherer are difficult, if not impossible, to manage systematically.
21
Fredrick M. Scherer, Firm Size, Market Structure, Opportunity, and the Output of Patented Inventions, 55
AM. ECON. REV. 1098, 1098 (1965); Fredrick M. Scherer & Dietmar Harhoff, Technology Policy for a World of
Skew-Distribution Outcomes, 29 RESEARCH POLICY 559, 559-60 (2000) (several studies “confirm[] that the size
distribution of patent values is indeed quite skew, most likely conforming either to a log normal of Paretian
distribution law.”). The skew of these distributions “implies that it is difficult or impossible to achieve stable mean
expectations and hence to hedge against risk by supporting sizeable portfolios of projects.” Id. at 563. Stated
otherwise, in the tail-wags-dog universe of innovation, statistical measures, such as averages and standard
deviations, are highly erratic and unreliable. This failure arises because statistical analyses are driven by the center
of the distribution, not its periphery; consistency—and predictability—emerge because the large number of small,
random events cancel out each other, exposing the systematic influences on the population as a whole. The
uniqueness of valuable patents defies this basic model because a small number of unique, extremely valuable patents
have a disproportionate effect on the statistics.
22
Allison, Valuable Patents, supra note 7, at 462.
23
Id.
24
Id. (citations omitted). In this quotation, “opposition system” refers to procedures that permit third parties
to challenge the validity or scope of a patent either while it is being prosecuted or within a limited period of time
after it issues. Under existing law, third parties have limited opportunities to challenge the validity of scope of a
patent. See 35 U.S.C. §§301-318 (2005). The authors go on to argue that “the PTO could quite easily create an
objective composite or algorithm based on the number of claims and prior art citations in an application . . . . [to]
focus[] its resources on the patents that are most likely to matter in the real world.” Id. at 464-65.
The authors’ equivocal tone is more than stylistic, though. As we will show, it portends
several fundamental shortcomings of their analysis and, more importantly, the analytical
impediments posed by the anemic predictive power of the patent characteristics on which they
rely. Standard patent characteristics, such as length of PTO review of patent applications (i.e.,
patent prosecution), number of claims, and citations made, prove to be only weakly associated
with patent value. We evaluate the reasons for the muted insight these factors provide for
refining patent policies and predicting patent value. This analysis underpins our critiques of
three recent empirical studies of U.S. patents, including the Valuable Patents paper.25 It also
forms the basis of our pessimism that prioritizing patents applications has the potential to revive
the PTO’s flagging patent prosecution process.
These analytical constraints are not be grounds for despair. Instead, they should motivate
legal scholars to reconsider the mass-data methods that have dominated empirical studies of
patents. We favor selective narrow investigations of patenting in specific areas of research and
development. Only narrowly focused studies have the potential to provide the level of detail
needed to understand localized patent dynamics—one can learn only so much from studies at
twenty thousand feet. We close the paper by outlining the elements of this approach.
The paper is divided into four parts. Part I describes how the database was constructed,
particularly the criteria used to classify patents under the rubric of biotechnology, and the general
outline of our data analysis.26 Part II describes general trends in biotechnology patenting and
discusses the implications of these findings for patent policy. Part III examines the statistical
limits of our study and three closely related existing studies. In this section, we reassess the
25
The three studies are found in Hall, supra note 7, Allison & Lemley, supra note 7, and Allison, Valuable
Patents, supra note 7.
26
Those people not interested in the details of the data and the data analysis should feel free to skip Part I; the
subsequent sections do not rely on it.
statistical constraints that are likely to preclude the PTO from effectively prioritizing the
thousands of patents its examiners review each year. Part IV returns to the obstacles hampering
empirical studies of U.S. patents, focusing on Scherer’s model of the innovation lottery, and
concludes with a brief discussion of opportunities for refining empirical methods given the
analytical challenges posed by the unpredictable nature of innovative success.