Вы находитесь на странице: 1из 9

Research Policy 35 (2006) 1122–1130

Profiting from innovation and the intellectual property revolution

Gary Pisano ∗,1
Harvard Business School, Soldiers Field, Boston, MA 02163, USA
Available online 12 October 2006

This paper reviews the contribution of Teece’s article [Teece, D., 1986. Profiting from technological innovation: implications
for integration, collaboration, licensing and public policy. Research Policy 15, 285–305.]. It then re-examines the core concept
of appropriability in the light of recent developments in the business environment. Whereas twenty years ago the appropriability
regime of an industry was exogenous and given, today they are often the product of conscious strategies of firms. And as open
source software and other industries show, advantageous appropriability regimes are not always “tight” or characterized by strong
intellectual property protections. The strategies adopted by firms that have successfully profited from their innovative activities cast
into new light old questions about the impact of intellectual property protection on the rate and direction of innovation.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Technological innovation; Teece; Appropriability

1. Introduction In this essay, I have two objectives. First, I want

to summarize what I see as the main contributions of
Twenty years ago, David Teece published his highly PFI to the fields of strategy and innovation and how
influential article entitled “Profiting from Innovation.” these ideas have shaped the intellectual dialogue on the
Since that time, this paper has drawn extensive attention topic. Second, I want to re-examine some of the core
in the literature. It is one of the most cited papers in concepts presented in PFI – particularly the concept of
the field of innovation, and it has been the most cited “appropriability regimes” – in light of recent changes in
paper in “Research Policy” over this time period. The the nature of intellectual property. In PFI, the appropri-
attention and praise aimed at this article is well deserved. ability regime was exogenous. The challenge of strat-
“Profiting from Innovation” (PFI) has had a profound egy was to develop appropriate vertical integration and
influence on the field of innovation and strategy. In short, complementary asset positions given the extant appro-
this paper initiated a process of convergence between priability regime. Increasingly, appropriability regimes
two fields that had essentially (and surprisingly) lived are not givens but are the product of conscious strategies
apart: innovation and strategic management. Many of of firms. And, surprisingly, an advantageous appropri-
the ideas presented in PFI continue to shape the way ability regime is not always “tight” or characterized
scholars, and increasingly practitioners, think about the by strong intellectual property protections. Phenomena
role of intellectual property in strategy. such as open source software and forms of deliberate
intellectual property sharing are increasingly being uti-
lized by for-profit enterprises as a rent seeking strategy.
∗ Tel.: +1 617 495 6562.
Such strategies cast into new light old questions about
E-mail address: gpisano@hbs.edu. the impact of intellectual property protection on the rate
1 Harry E. Figgie, Jr. Professor of Business Administration. and direction of innovation.

0048-7333/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
G. Pisano / Research Policy 35 (2006) 1122–1130 1123

2. PFI and the fields of innovation and strategy cal framework for explaining and predicting why (and
when) innovators would generate sustained profits from
Like most good theory, PFI addressed a puzzle that their innovation and when they were vulnerable to dis-
had not been well explained in the previous literature, placement by later entrants. I will not bother to recite
namely: why is it that innovators often fail to capture the the details of the argument here, as they are clearly
economic returns on their innovations? At the time PFI explained in the original PFI. In essence, the ability of
was written, there were many such examples, and the an innovator to profit from its innovation over time was
paper cites these (e.g. EMI in CAT scanners, Bowmar a function of its “complementary asset” position and
in calculators). Perhaps more impressively, during the the “appropriability regime” in which it found itself. In
ensuing twenty years after PFI was published, we can see “weak” appropriability regimes – where imitation was
many more examples of such failures. The phenomenon relatively easy from both a technical and legal stand-
endures. The first generation PC manufacturers all but point – protecting rent streams from innovation required
disappeared from the scene (and even IBM, while not a privileged access to what Teece called complementary-
first-mover in PCs, recently exited the business by selling specialized assets (or “co-specialized assets”). In strong
its PC business to a Chinese company, Lenova). Apple regimes, in contrast, firms could rely on licensing
invented the graphical user interface, but Microsoft Win- and other contractual arrangements to extract rents
dows dominates the PC market. Apple invented the PDA from their innovation without access to such assets.
(the bricklike Newton) but Palm became the dominant In short, strategy is contingent on the appropriability
player. Netscape invented the browsers, but Microsoft regime.
captured the market. Merck was a pioneer in cholesterol This is an extraordinarily powerful insight, and this
lowering drugs (Zocor), but Pfizer, a late entrant, grabbed leads to the broader, perhaps even more enduring, con-
the dominant market position (Lipitor). Excite and Lycos tribution of the paper. Previous to PFI, the field of
were the first real web search engines, but they lost out strategy was disconnected from the field of innovation,
to Yahoo. And Yahoo then lost out to Google. at least academically (practitioners had always had to
At first glance, it is tempting to say that these exam- deal with these issues!). The field of innovation was
ples simply reflect the normal rough and tumble of typi- focused on understanding such (important) issues as the
cally Schumpeterian competition. Winners do not stay on rate and direction of technical progress (at both industry
top for long; in highly dynamic settings, new entrants are and national levels), the sources of innovation (e.g. von
always ready with disruptive innovations.1 I will return Hippel, 1978), and a host of questions about the orga-
later to the “why established firms fail” literature and its nization and management of R&D. In the mid-1980s,
relation to PFI. However, it is worth noting that there the strategy field was being transformed by Michael
is ample variance in the phenomenon. There are many Porter’s seminal work on the competitive forces (Porter,
cases where first or early movers captured and sustained 1980). Strategy was focused on understanding the impli-
significant competitive advantage over time. Genentech cations of industry structure on competitive choices
was a pioneer in using biotechnology to discover and and positioning. Innovation was not really a central
develop drugs, and 30 years later is the second largest actor.
biotechnology firm (and, the most productive from an Industrial organization economics came closest to
R&D point). It is second only to Amgen, another early providing a link. Theoretical and empirical work explor-
entrant. Intel invented the microprocessor and contin- ing the link between R&D, innovation, and market
ues to dominate that market more than 30 years later. structure was the main focus of industrial organization
Dell pioneered a new distribution system for personal economics at that time. That work itself could be classi-
computers and, despite numerous attempts to imitate its fied into two broad categories: traditional neo-classical
highly successful business model, remains the dominant microeconomics work (including game theoretic explo-
“supplier” of computers (and increasingly a wide range ration of “patent races”) and evolutionary models of
of electronics). Nelson and Winter (1982). The more traditional eco-
The contributions of PFI can be viewed at various nomic work was not really concerned with strategy; strat-
levels. At a very specific level, PFI offered a theoreti- egy involves choices, and in traditional economics of the
mid-1980s, choice was about optimizing. The idea that
similar firms (facing similar prices) might make different
1 There is a long literature on the role of new entrants in dislodging choices was something most of economic theory did not
established firms. See for instance, Anderson and Tushman (1990), really want to think about. In addition, game theoretic
Clark (1985), Henderson and Clark (1990), Christensen (1997), etc. approaches that might have led to insights about strate-
1124 G. Pisano / Research Policy 35 (2006) 1122–1130

gic interaction generally assumed perfect patents. There nization mattered to innovation. And, appropriability
is nothing wrong with simplifying assumptions; they are regimes mattered to strategy. In bringing these issues to
simply a way to hold constant one set of parameters in the same table of debate, Teece introduced to the innova-
order to explore the implications of changes in others. tion and strategy field new theoretical perspectives such
However, by holding constant the assumption of per- as transaction cost economics, evolutionary economics,
fect patents, these models could not provide insights into and legal theories of intellectual property.
how competition under different appropriability regimes The influence of PFI on subsequent research is pro-
might operate. found and far reaching. An enumeration of specific arti-
An alternative line of work that emerged in the late cles or topics would be well beyond what even the most
1970s and early 1980s was the evolutionary economic comprehensive review piece could accomplish. Let me
approach to technical change. Pioneered by Nelson and just highlight a few of the most critical areas where the
Winter (1977, 1982) this approach took a more grounded PFI framework seems to have the most impact.
approach to innovation. It recognized that innovation was
an often messy and unpredictable process. Firm strate- 2.1. Alliances and networks
gies were viewed as sets of heuristics (that may or may
not be consciously articulated by management). More The first would clearly be the mass of work on
importantly, Nelson and Winter recognized that firms strategic alliances and networks. While this work is
operated in potentially different environments when it extraordinarily broad, a significant part of the alliance
came to both the opportunity for innovation and the and network literature is concerned with innovation.
ability to block would-be imitators. In a very interest- Prior to the 1980s, the lion’s share of innovative activ-
ing empirical investigation, Klevorick, Levin, Nelson, ities (R&D and the complementary activities needed to
and Winter (1995) surveyed a large sample of senior bring innovation to market) was conducted inside the
R&D managers to understand how they thought about boundaries of firms. In essence, innovation and verti-
different mechanisms of appropriability (patents, trade cal integration went together. Beginning in the early to
secrets, etc.). This work demonstrated that appropri- mid-1980s, amidst growing global competition, large US
ability was a multi-dimensional concept. It was not, as enterprises began to experiment with alternative orga-
economic theorizing had generally assumed, only about nizational approaches to innovation. In particular, they
patents. Firms could seek to protect their innovation in began to source technology through alliances, licensing
various ways. agreements, and other contractual forms of collabora-
This was an important insight, and, as is widely rec- tion with outside (often smaller, entrepreneurial) firms.
ognized today, evolutionary economic approaches have This trend first began in electronics, computers and
had a powerful influence on the field of strategy. But, at software, and telecommunications (in the wake of the
that time, evolutionary economic theorizing was focused re-structuring of the industry after the AT&T divesti-
squarely on “fixing” the discipline of economics and was tures), but has clearly spread throughout the economy.
not looking to influence the field of strategy.2 However, Pharmaceuticals/biotechnology, automobiles, entertain-
a predecessor to the kind of issues found in PFI can be ment and media, and financial services all involve mas-
found Nelson and Winter’s (1982, chapter 7) simula- sive amounts of inter-firm collaboration. Innovation
tion models exploring how “imitative” versus “innova- occurs through both internal hierarchies and through
tive” strategies might perform over time under different markets.
assumptions of appropriability. The academic literature exploring such arrangements
In building on this thread, Teece triggered a deeper has grown accordingly. This literature is vast and thus
exploration of the connection between firms’ strate- falls into multiple theoretical categories. Teece’s PFI
gies, innovation, and appropriability. Strategy and orga- has influenced the strand of literature concerned with
strategic positioning in these networks and the perfor-
mance of different network strategies. The basic PFI
2 As a personal side note, the field of strategy also appeared to be provides a potentially useful way to understand these
completely unaware of evolutionary economics at that time. When I alliances and to make normative predictions about per-
interviewed for a job in the strategy department at a leading business formance. In order to help innovators specialize (safely),
school in 1988, I was asked by a top scholar in the field of strategy markets for know-how must work effectively. Networks
what I thought the most important book published in the strategy over
the past 10 years had been. When I replied, “Nelson and Winter’s
of innovation thus depend partly on intellectual property
‘Evolutionary Economic Theory’,” I received a blank stare (and no job regimes. Strong intellectual property regimes would sup-
offer!). port broader and more diffuse networks of innovation.
G. Pisano / Research Policy 35 (2006) 1122–1130 1125

2.2. Sustainability of innovation-based first-mover remembering to attach the right files. There is no tacit
advantage knowledge in taking his manuscript and printing it.
But in weak appropriability regimes, complementary
Ever since the 1970s, the strategy field has been con- assets – often in the form of capabilities – matter. Thus,
cerned with the normative implications of so-called first “core capabilities” become less meaningful in those set-
mover strategies. When will the first entrant into a new tings. Capturing value on innovation requires the firm to
market sustain their advantage? When will a company have mastered complementary capabilities (e.g. manu-
with a novel value-creating business model sustain their facturing, distribution, etc.).
initial dominance? Again, this is a vast literature. Not The case history of Intel is an example of the basic
all first mover advantages have to do with innovation, concepts of PFI applied to the idea of capabilities.3
but innovation represents an important potential source Intel was a pioneer in DRAMs but lost its competitive
of first mover. The PFI framework provides a theoret- advantage to Japanese rivals in the early 1980s because
ical underpinning for examining when and where first it lacked complementary capabilities in process devel-
mover innovators may sustain their advantage and when opment, manufacturing ramp-up, and manufacturing.
they will not. Operating in a weak appropriability regime These lessons were not lost on Intel senior manage-
does not mean a first-mover strategy will not work. It ment when they began their ascension in microproces-
simply means that a firm will need to protect its position sors. Intel invented the microprocessor. Initially, they had
by securing access to complementary specialized assets. strong intellectual property protection on the basic archi-
tecture of their microprocessors. Over time, this IP posi-
2.3. Capabilities-based approaches to strategy tion eroded (partly due to licensing that enabled a rival,
AMD, access to the micro-code). Yet, Intel was able
Over the past decade, there has been a swelling inter- to maintain its dominance; why? Part of it was invest-
est in the role that various kind of organizational capabil- ments in complementary specialized assets; throughout
ities play in competitive advantage (see e.g. Pisano and the mid-1980s and early 1990s, Intel invested heavily in
Teece, 1994; Teece et al., 1997). One of the most impor- building world-class process development and manufac-
tant, interesting, and yet vexing management challenges turing capabilities. By the mid-1990s, Intel was one of
concerns investments in building appropriate organiza- the best performers in ramping up high volume produc-
tional capabilities. While ex post case studies can often tion of new chips (Iansiti, 1997). While it was possible
identify (or at least rationalize) missing capabilities that for AMD to enter the market with architecturally com-
doomed a competitor, ex ante it is extremely difficult to patible designs (X86), Intel consistently won the market
identify what capabilities matter to competitive advan- by ramping up production more quickly. We might not
tage. Much of the general writing on the subject is vague. normally think of Intel’s “core capabilities” as manu-
Managers are urged to focus on building their organiza- facturing. It was not historically. It was, in the DRAM
tion’s “core capabilities.” But, how one can identify what period, a weakness. But, recognizing the need for this
these core capabilities should be is another matter. The complementary asset, the company built these comple-
basic theoretical constructs offered in PFI begin to give mentary capabilities. Intel may create value through its
us some traction around this question. designs, but it captures them through its complementary
Following PFI, the answer really begins with under- process development and manufacturing capabilities.
standing the appropriability regime. In a very strong
appropriability regime, the firm can specialize. It needs 2.4. Why established firms fail
only a narrow range of “core capabilities” and can cap-
ture returns on their innovation via the market. It does One of the most productive lines of research in the
not need complementary capabilities. Take an extreme innovation field has concerned the question of why
example: Stephen King is an extremely talented writer. established firms fail in the wake of major technolog-
He does not need to own a book publisher, a printing ical changes. This literature, of course, has its roots
press, or distribution network to capture the rents on in Schumpeter and the notion of creative destruction.
his considerable talent. He can use contracts because he Several decades elapsed before scholars of innovation
operates in a strong appropriability regime. Books are picked up the question in the 1980s (Anderson and
well protected by legal mechanisms. It is hard to imitate
his talents. And transferring his “technology” to produc-
tion, particularly in this day and age of electronics, is 3 See “Intel Corporation: 1968–1997” (D.J. Collis, G. Pisano, P.
a trivial matter of hitting the “send” key on email and Botticelli, HBS case number 737-137).
1126 G. Pisano / Research Policy 35 (2006) 1122–1130

Tushman, 1990; Clark, 1985; Henderson and Clark, In a weak appropriability regime, complementary assets
1990; Christensen, 1997). This body of work tackles the should matter. Asset position of incumbents (manufac-
issue from two basic angles. On the “supply side” are turing, distribution, etc.) have not been a major focus
writers such as Anderson, Tushman, Clark, and Hen- of the “why established firms fail” literature. If PFI the-
derson who have explored the ways in which a firm’s ory is correct, complementary assets would also need
existing repertoire of capabilities may either blind it from to become somewhat obsolete by the new technology to
seeing novel opportunities to innovate or acting upon enable entrants to gain a protected foothold.
those opportunities when they see them. For instance, In the disruptive innovation framework of Chris-
Henderson and Clark (1990) advance the idea that some tensen, “disruptive” innovations (by definition) take hold
innovations are “architectural” in nature and thus require initially in small niches that are of less interest to incum-
a reconfiguration of existing organizational and techno- bents. Mini-mills got their foothold in the “low end”
logical capabilities to pursue. Organizations have diffi- market for rebar; 5.25 disk drives got their foothold in
culties with these types of reconfiguration due to iner- the then tiny market for desktop computers. In essence,
tia and kindred forces. As a result, they predict, and disruptive innovations take hold because the innovators
demonstrate in the case of photolithography, that exist- are given an opportunity to build the complementary spe-
ing organizations are more likely to fail when faced with cialized assets needed to enter the market.
architectural innovation (than when they are faced with A contrasting example would be that of biotechnol-
“component” innovations which require change in one ogy and pharmaceuticals.4 For more than 30 years, many
organizational sub-unit, but not across units). observers have predicted that new biotechnology firms
On the “demand side” of this literature is a writer would displace the established pharmaceutical compa-
like Christensen who sees lack of response rooted not nies that have dominated the sector for more than 60
so much as in constrained organizational capabilities years. This has not happened. Incumbent firms remain
but in the constraints created by connections to exist- dominant. Only a tiny fraction of biotechnology firms
ing customers and markets. Christensen theorized and that have entered the industry since 1976 have entered
demonstrated for various industry cases (but most deeply the league of top companies (Amgen is approximately
in disk drives) that new innovations that challenge an 5th in size by market capitalization; Genentech, the sec-
organization’s existing business model and customer ond most profitable firm, is 60% owned by an established
segmentation are more likely to give an established com- firm, Roche). Why did not biotech follow the same pat-
pany difficulty. tern as say disk drives, photolithography, and others?
Interestingly, PFI theory has largely been outside the PFI theory helps: in essence, while biotechnology cre-
domain of the “established firm failure” literature. I ated novel approaches to discovering, developing, and
would argue that PFI theory has something to add to manufacturing drugs, by and large the key downstream
this literature. Established firms fail in the wake of inno- complementary assets needed to commercialize drugs
vation when, by definition, new entrants succeed. Part are very similar between “old” and “new” technologies.
of the story is clearly rooted in the kinds of inertias dis- Drugs still need to go through clinical trials; one still
cussed by the writers above. However, we know from PFI needs deep regulatory expertise; and, both new and tra-
theory that there is more to the story than just success- ditionally derived drugs are by and large sold through the
fully innovating. New entrants that successfully innovate same marketing and distribution channels. In essence,
can still fail if they fail to capture sustainable returns. the biotech firms needed access to the co-specialized
PFI theory points out certain key features of the envi- assets controlled and owned by incumbents. These assets
ronment and strategies that might matter a great deal. were costly to build; as such, most biotech firms were
Clearly, again, appropriability regimes matter. If appro- forced to collaborate with incumbents. Large pharma-
priability regimes are weak, not only can a new entrant’s ceutical companies have been able to use their control
innovation be imitated by other new entrants, they can be over co-specialized assets to appropriate the lion’s share
imitated by established firms. In many of the case studies of value from the biotech revolution. PFI theory would
explored in the above literature, appropriability regimes predict that only once we see changes downstream (e.g.
seemed to be weak. That is, established players did not in terms of how drugs are marketed or distributed in the
fail because they got legally “locked out” of the new tech- wake of changes in the health care system) will we see
nology (indeed, there is ample evidence that established new entrants gain a foothold.
firms often have very strong access to the relevant tech-
nologies and technical talent). PFI theory would then
lead us to consider the role of complementary assets. 4 For a discussion of these issues, see Pisano (2006).
G. Pisano / Research Policy 35 (2006) 1122–1130 1127

3. Why PFI matters today distribution of rents from innovation moves away from
traditional end product innovators and toward innovators
One of the tests of the significance of any piece of in chips. This happened first in PCs (with the emergence
research is whether the questions and issues raised can of the Windows and Microsoft standard) but has contin-
stand the test of time. At the time Teece wrote PFI, ued in others such as cell phones, televisions, and digital
issues of collaboration and novel organizational forms cameras. For instance, if a company wants to get into the
for innovation were just coming to the fore of both aca- cell phone business, it can now buy the relevant chips
demic and practitioner attention. It is clear that twenty sets from a company like Qualcom. In these settings,
years after PFI, these issues continue to be of enormous rents flow to the suppliers of specialized chips that create
importance. Today, globalization, and in particular the lock-in via compatibility or control over standards. Note,
emergence of China and India as formidable competi- this does not mean that innovation is impossible outside
tors on the global stage, means that companies across the the chip. A cell phone supplier, for instance, can inno-
world increasingly find themselves embroiled in com- vate in key pads, displays, software, and aesthetic design.
plex webs of collaborative and contractual relationships. However, appropriating the value on this innovation is
Whereas in the mid-1980s, companies were beginning difficult where control over the key system functionality
to experiment with novel forms of partnerships, today, is dictated by the chip.
collaborative arrangements are woven into the fabric of The PFI framework helps us consider the kinds of
corporate strategies worldwide. Moreover, companies strategies that may or may not be appropriate in the
in most industries face extremely able competitors and context of digitization. Prior to digitization, innovation
innovators across the world. competition centered around high level system design.
At one level, global competition has raised the impor- Traditional system producers (e.g. computer manufac-
tance of innovation once again. A perusal of any major turers, phone manufacturers) innovated by architecting
business publication, such as Business Week, shows that unique designs from a combination of proprietary (self-
innovation is in vogue again as companies look for ways designed) and non-proprietary components. In these con-
to grow. Once seemingly mature industries – like televi- texts, systems could be reverse engineered and rents
sions and consumer electronics – have become hotbeds flowed across horizontal competitors (computer man-
of innovations. At the same time, it has become perhaps ufacturers competed with other computer manufactur-
more difficult than ever to profit from innovation for all ers). As more of the functionality shifted to chips, those
the reasons Teece outlined in 1986. In many contexts, companies, often specialized chip suppliers like Intel or
particularly in China and India, appropriability regimes Qualcom, captured a growing share of the rents from
are weak. Innovation draws rapid imitation. In others, innovation at the system level. Using PFI terminology,
technically sophisticated competitors are able to respond the chip design (including embedded software) became
very quickly to innovations, and develop equivalent, if the key co-specialized asset. And because these designs
not superior, products of their own. A PFI framework could be well protected by patents, it became nearly
helps organizations to think through systematically the impossible for system producers to capture the rents.
kinds of assets they need to foster internally and those The PFI framework would suggest that in these set-
that they can safely outsource. tings, the best alternative strategy for a system producer
Perhaps one of the most important factors reshap- would be to establish another set of co-specialized assets,
ing the terms of innovation based competition is the such as downstream marketing and distribution. This
impact of digitization. Over the past decade, a wide seems to be what happened in the PC industry. As Intel
range of product categories – such as phones, cameras, (and Microsoft) gained control over the rents from PC,
televisions, music systems, music distribution, entertain- most PC manufacturers were reduced to suppliers of
ment, etc. – once based on mechanical, optical, and a commodity product. Only Dell was able to maintain
analog technologies have been transformed through the super-normal profits largely because it created a highly
application of digital technologies. Beyond changing the unique and difficult to imitate direct-sales distribution
technology of products, digitization has had two criti- system.
cal impacts on the landscape of innovation. First, it has
altered the locus of innovation. As more functionality 4. PFI in the age of endogenous appropriability
is embedded in chips, innovation increasingly occurs in regimes
the digital components of the system. Increasingly, the
system is the chip and the chip is the system. Second, as I want to conclude this paper by briefly suggesting an
the key intellectual property shifts to the chip system, the important avenue for future research that was not con-
1128 G. Pisano / Research Policy 35 (2006) 1122–1130

sidered in the original Teece PFI framework, but which legal circles. What were the implications for privately
recent history suggests has become a critical compo- owned genes? Was this legal? Was this ethical? What
nent of innovation strategy. In the original formulation, impact might this have on biomedical research progress?
appropriability regimes are taken as a given. They are Interestingly, equally concerned with the patenting of
determined exogenously by a confluence of legal forces genes were a group of companies who are normally
(e.g. the scope and potency of patent protection) and the among the staunchest advocates of strong patent rights:
nature of the technology itself (e.g. ease of imitability). pharmaceutical companies. The concern among phar-
The PFI framework sees the firm’s strategic problem as maceutical companies was that they could essentially
choosing the best complementary assets positions given be held hostage by another entity that claimed owner-
the appropriability regime it faces. I would like to sug- ship of a key gene or genes associated with a disease
gest that recent history suggests that this view may miss where they had strong commercial interest. Take a firm
a critical element of intellectual property rights strategy. like Merck. Merck had established a very strong research
Increasingly, appropriability regimes are endogenously program in cardiovascular disease and cholesterol lower-
influenced by the behaviors and strategies of firms them- ing drugs in particular. Moreover, they had built a strong
selves. And, in some cases, firms take their complemen- downstream asset positions in the sales and marketing of
tary asset positions as given and then attempt to shape such drugs. Merck’s R&D and marketing capabilities in
the appropriability regime to optimize the value of those cardiovascular drugs represented strong co-specialized
assets. Let me provide two examples: the first from the assets positions to use PFI terminology. If some other
field of genomics and the second from the field of open private firms were able to identify and claim intellec-
source software. tual property ownership over the genes associated with
cardiovascular disease, this could potentially lead to a
4.1. Genomics hold-up situation (at an extreme). If Merck could not con-
tinue to conduct certain research programs, it might not
During the late 1980s and 1990s, there was a ver- be able to leverage its existing co-specialized assets posi-
itable revolution in the field that came to be known tions. The value of these assets would become severely
as genomics. With enormous advances in the scien- impaired.
tific instruments used to “read” DNA code, it became One strategy for a large company with existing co-
possible to identify genes on a mass production scale. specialized assets positions would be to move aggres-
Where it once might have taken a researcher ten years sively to secure rights to the genes that might impact
of dedicated work to sequence a single gene, it became its future research. This strategy was followed by a
possible by 1990s to identify thousands of genes on a number of large pharmaceutical companies who signed
monthly basis. The US government funded a project expensive deals with genomics firms for access to their
– the Human Genome Project – to sequence all the proprietary genetic databases. These deals were both
genes found in the human body. And a competing pri- a way to open avenues for new research, but to also
vately funded effort was launched by a company called protect the firm against future lock-out. Another strat-
Celera Genomics. The potential of mass sequencing of egy, one followed by Merck, was to attempt to alter the
genes for biomedical research was immense. For the appropriability regime. Once something is made pub-
first time, researchers could begin to explore the genetic lic, it can not be patented. In September 1994, Merck
bases for a variety of diseases such as cancers, diabetes, announced plans to collaborate with Washington Uni-
Alzheimers, and many others. The potential economic versity to create a database (the Merck Gene Index) of
impact was not lost on the financial community. If genes expressed human gene sequence and to put these data
were valuable to biomedical research and drug discov- into the public domain. The stated goal of this effort,
ery, and drug discovery was lucrative, then it stands to and in particular the decision to make all findings pub-
reason that genes had enormous economic value if they licly available in 48 hours, was to stimulate biomedical
could become intellectual property. Following this logic, research. This sounds highly altruistic, and I do not want
venture capitalists funded dozens of firms (e.g. Celera, to discount the potential for Merck to be engaged in a
Incyte, Human Genome Sciences, etc.) to exploit the public service. However, it is also easy to see a strate-
commercial potential of genomics through the selling gic motive: by making expressed human gene sequences
of proprietary genomic databases to pharmaceutical and publicly available, Merck was essentially preventing a
biotechnology firms. privatization of genes that could block its future research
The potential for firms to take ownership of specific objectives. In essence, Merck was keeping the upstream
genes caused some consternation in public policy and appropriability regime “loose.”
G. Pisano / Research Policy 35 (2006) 1122–1130 1129

An interesting point of the Merck Gene Index story any capable person, and allows anyone to utilize the
is that it shows how a profit seeking intellectual prop- open source software (as long as they abide by the
erty strategy does not necessarily mean creating a tighter norms), open source communities tend to be fairly broad.
appropriability regime. In discussions of IP strategy, it Depending on the project, literally thousands of develop-
is generally assumed that private, profit seeking firms ers around the world may be involved in an open source
would always desire a “tighter” appropriability regime. project.
However, viewed through the lens of complementary Open source poses an interesting strategic challenge
asset theory and PFI, we see the flaw in this logic. A firm for firms. How do they respond? Should they embrace
with very strong downstream complementary assets may open source? Should they resist (if possible)? This
have strong strategic incentives to make “upstream” IP is an issue firms like Microsoft and Sun are dealing
weak to prevent foreclosure of opportunities. The value with today. The emergence of Linux clearly represents
of those assets is higher in a weak appropriability regime a threat to Microsoft’s server operating system busi-
than in a strong one. This is consistent with PFI theory. ness (Windows). And it represents a threat to Sun’s
server business which rests on a proprietary version
4.2. Open source of Unix called Solaris. Some firms, like IBM, have
clearly embraced open source and promoted it. Figur-
Open source refers to a movement in software devel- ing out how to respond partly depends on a company’s
opment to make publicly available the source code for co-specialized asset position. A weakening of the appro-
computer programs so that other developers can build priability regime through the emergence of open source
upon on the code base. Under various open source licens- operating systems can be beneficial to companies with
ing arrangements (such as the GPL), any developer can strong downstream asset positions in middleware, appli-
use open source code and build upon it, as long as they cations, hardware, and services. In essence, as the server
do not try to appropriate (i.e. claim intellectual prop- operating system becomes a commodity, the locus of
erty ownership) of the previously disclosed code. This value capture in the innovation chain shifts downward.
is designed to prevent anyone from essentially privatiz- This is another example where the weakening of an
ing the intellectual commons. Well known examples of appropriability regime can be economically beneficial
open source development include Linux and Apache, to some firms (while hurting others). It may well be in
but there have been literally tens of thousands of other the interest of firms with strong downstream complemen-
lesser known (and often much less successful) efforts. tary asset positions to proactively weaken the upstream
Open source clearly represents a shift in the appropri- appropriability regime (e.g. via code contributions or
ability regime of software. Traditional software devel- public announcements of support).
opment followed the model we see in other industries;
development was proprietary and developers do every- 5. Conclusion
thing possible to protect the designs from imitation or
un-compensated use. In software, this included using To conclude briefly, twenty years after its publication,
legal mechanisms (e.g. copyrights and patents) but also PFI has continued to have a profound impact on the inno-
secrecy (e.g. refusing to make available source code). vation field, and its relevance today in a world of global
With open source, the opposite logic holds. Developers competition, open source, and other forces is greater than
contribute code with an understanding that it can be used ever. Understanding the source and evolution of appro-
freely by others (including additional development). In priability regimes has been a topic of interest in legal
essence, open source leads to the creation of a commonly scholarly communities, but to date has not been the sub-
shared base of technology. ject of deep research in strategic management. Many
Robert Merges draws a fascinating analogy between interesting questions remain to be explored. For instance,
open source and medieval guilds (Merges, 2004). While how much can firms proactively shape an appropriability
he notes important differences, he sees important simi- regime in their favor (either by weakening or strength-
larities: both utilize shared norms in a community about ening it)? What are the mechanisms firms can use to
what can and can not be appropriated. Open source shape an appropriability regime? What are the alterna-
communities are generally often depicted as mass move- tive institutional arrangements? As intellectual property
ments of independent developers who share some com- continues to play a central role in the health and growth
mon goal around building a particular piece of software. of firms, these questions will continue to be of as much
To some extent, this is true. Since the open source importance over the next twenty years as they have been
community invites the participation and contribution of over the past twenty.
1130 G. Pisano / Research Policy 35 (2006) 1122–1130

References Merges, R., 2004. From medieval guilds to open source software:
informal norms, appropriabability institutions and innovations. In:
Anderson, P., Tushman, M.L., 1990. Technological discontinuities Conference on the Legal History of Intellectual Property, Novem-
and dominant designs: A cyclical model of technological change. ber, University of Wisconsin, Madison.
Administrative Science Quarterly 35, 604–633. Nelson, R.R., Winter, S.G., 1977. In search of useful theory of inno-
Christensen, C., 1997. The Innovator’s Dilemma: When New Tech- vation. Research Policy 5, 36–76.
nologies Cause Great Firms to Fall. Harvard Business School Press, Nelson, R., Winter, S., 1982. An evolutionary theory of economic
Boston, MA. change. Harvard University Press, Cambridge, MA.
Clark, K.B., 1985. The interaction of design hierarchies and mar- Pisano, G.P., 2006. Science business: the profits, the reality and the
ket concepts in technological innovation. Research Policy 14, future of biotech. Harvard Business School Press, Boston, MA.
235–252. Pisano, G.P., Teece, D., 1994. Dynamic capabilities: an introduction.
Collis, D.J., Pisano, G.P., Botticelli, P., 1997. Intel Corp. —1968–1997. Industrial and Corporate Change 3, 537–556.
Harvard Business School Publishing, Boston, MA, HBS Case No Porter, M.E., 1980. Industry structure and competitive strategy: keys
797-137. to profitability. Financial Analysts Journal 36, 30–41.
Henderson, R.M., Clark, K.B., 1990. Architectural innovation: the Teece, D., 1986. Profiting from technological innovation: implications
reconfiguration of existing product technologies and the failure for integration, collaboration, licensing and public policy. Research
of established firms. Administrative Science Quarterly 35, 9–30. Policy 15, 285–305.
Iansiti, M., 1997. Technological integration: making critical choices in Teece, D.J., Pisano, G.P., Shuen, A., 1997. Dynamic capabilities
a turbulent world. Harvard Business School Press, Boston, MA. and strategic management. Strategic Management Journal 18,
Klevorick, A.K., Levin, R.C., Nelson, R.R., Winter, S.G., 1995. On the 509–533.
sources and significance of interindustry differences in technolog- von Hippel, E.A., 1978. Users as innovators. Technology Review 80,
ical opportunities. Research Policy 24, 185–205. 30.