Cohen Det Meyer K Rahm Er 2001

Research Policy 30 (2001) 15631591
The theoretical and policy implications of knowledge codification

Patrick Cohendet a, , Frieder Meyer-Krahmer b
a
BETA, Universite Louis Pasteur, 61, Avenue de la Foret Noire, 67085 Strasbourg Cedex, France
Fraunhofer Institute for Systems and Innovation Research ISI, Breslauer Str. 48, 76139 Karlsruhe, Germany
Abstract
The purpose of the following contribution is to present the main consequences in terms of policy of firms and governments
that emerged from the theoretical and empirical work achieved under the TIPIK project. In a first part, the introduction is
devoted to a summing-up of the main theoretical dimensions used in TIPIK, that led in particular to a special issue of Industrial
and Corporate Change (June 2000). Then, the discussion envisages the main consequences of the theoretical debate in terms
of policy issues and addresses the new perspectives of a knowledge oriented policy of firms and governments. Finally, we
expose a selected series of policy implications, that are derived from the empirical studies presented in this special issue.
2001 Elsevier Science B.V. All rights reserved.
Keywords: RTD; Policy; Codification; Communities; Competencies; TIPIK project
1. Theoretical framework
1.1. The nature of the codification process
The codification of knowledge is a process that aims
at reducing and converting knowledge into messages.
These messages can then be processed as information
that will serve to reconstitute knowledge at a later
time, in a different place, or by a different group of
individuals. The main interest of this process of transformation of knowledge is to facilitate the treatment
of knowledge as an economic good, which can be exchanged, in particular on markets. 1
Corresponding author. Tel.: +33-3-90-24-21-89;
fax: +33-3-90-24-20-71.
E-mail address: cohendet@cournot.u-strasbg.fr (P. Cohendet).
1 The treatment of knowledge as a commodity is sometimes called
commodification of knowledge. A risk of confusion is to assume that codification of knowledge will automatically allow the
commodification of knowledge. Codification obviously facilitates
the exchange of knowledge, but in many situations, knowledge
can be transferred without codification (for instance by recreating
The process of codifying knowledge entails three

distinct but related steps (Cowan and Foray, 1997):
creating models; creating languages; and creating messages. 2 Each of these aspects has its own costs, and
at each level, as the process occurs, very typically new
knowledge is created. The first two steps of the process of knowledge codification the development
of a model of interpretation and of a language
generally involve high fixed costs. Indeed, it requires
time and efforts to implement standards of reference
the learning context as in the case of the driving lessons). Conversely, the codification of knowledge does not necessarily ensure
its commodification because it may leave the knowledge highly
specific (a difficult mathematical theorem, an industrial operating
procedure). Furthermore, if the codes are kept secret and this
can be done on purpose or not codification may even impede
the commodification of knowledge.
2 As an example, when referring to engineering drawing, Monges
descriptive theory expressed a model of representation. The language that was developed from it, based on the first angle projection, allowed a rapidly increasing community of practitioners in
engineering drawing to exchange messages.
0048-7333/01/$ see front matter 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 4 8 - 7 3 3 3 ( 0 1 ) 0 0 1 6 8 - 8
1564
P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591
(numerical, symbolic, geometrical languages and taxonomies of many kinds), standards of performance, a
vocabulary of precisely defined and commonly understood terms, and a grammar to stabilise the language.
Once these steps have been achieved, a code-book
becomes available, and agents are able to carry out
knowledge operations at low marginal costs, since
messages are reproducible. 3
It must be emphasised that knowledge is simultaneously an input and an output of the codification
process. In other words, some knowledge is needed
to codify knowledge, and furthermore, knowledge is
needed to exploit a given piece of codified knowledge. This recursive and dynamic structure of knowledge has two consequences: one is that knowledge
cannot be regarded as a mere stock resulting from
an accumulation of information. The second is that
knowledge cannot be considered separately neither
from its holder(s), nor from its time/space location.
Both arguments stress the importance of the (individual and collective) cognitive processes through which
knowledge is converted, combined, stored, retrieved,
exchanged, and interpreted.
1.2. The relationships between codified and tacit
knowledge in the codification process
The codification process alters the relationship between the codified and the tacit form of knowledge.
On the one hand, codified knowledge refers to a view
of knowledge as necessarily explicit, formal or systematic which can be expressed in words and numbers, scientific procedures, or universal principles.
This codified category of knowledge is easy to transfer, to store, to recall and to valorise. On the other
hand, the notion of tacit knowledge refers to that form
of knowledge that cannot be easily expressed. The
main forms of tacit knowledge are know-how (that
results from the accumulation of practice), the mastering of a language (that results from the accumulation
of the ability to communicate) and representations
3 As noticed by Steinmueller (2000), this simplified vision of the
process of codifying knowledge as a three step process, does not
encompass all the cognitive issues associated with the codification
process. In particular, there may be problems in aligning cognitive
understanding with the language by which models and messages
are constructed.
of the world (that results from the accumulation of

wisdom). 4 Whereas codified knowledge can easily
circulate and be exchanged, tacit knowledge is extremely difficult to transfer and to exploit. However,
when compared to codified knowledge, one of the
characteristics of tacit knowledge is that it allows
firms to solve specific problems, even when there is
no general understanding of the reasons behind these
problems or the optimal rational methods for their
solutions. Skills or know-how are associated with the
use of implicit routines or procedural rules that can
be shared via learning, imitation and practical examples, rather than explanations and manuals, repeated
practices. Nevertheless, there is one condition for
the use of tacit knowledge to be efficient: it must be
permanently activated. If tacit knowledge is not activated, after a time, knowledge will be forgotten and
lost. 5
Among the tacit forms of knowledge, some are fundamentally unarticulable, which means that it would
cost too much, at the current state of knowledge, to
try and articulate these pieces of knowledge. 6 Cowan
et al. (2000) argued that this category of unarticulable
knowledge is infra-marginal. Very little knowledge is
inherently tacit and impossible to codify. Most of the
tacit knowledge is just unarticulated, and the process
of codification can be viewed as an attempt to transform, at a finite cost, unarticulated pieces of knowledge into codified ones. However, reducing the process
of codification to mere conversion of tacit knowledge
into a codified form, where the new codified knowledge is just substituting for some of the tacit form,
would oversimplify the complex nature of the process
and lead to misleading interpretations. Nonaka and
Takeuchi (1995) showed that the codification process
appears as a complex conversion process, where the
codified and the tacit forms are not substitutes, but
rather complements. In most contexts, agents need at
least the tacit knowledge involved in mastering a lan4 Many economic decision makers can hardly formalise the representations they refer to in their decision process: it is not just
because they cannot explicitly describe the mental image they use,
it is also because it is impossible to explicitly retrieve all the bodies of knowledge such a mental representation is derived from.
5 This is another incentive to codify knowledge, and furthermore
to codify carefully how to use the code.
6 For example, some of the (tacit) know-how of the glass-blowers
are still considered as unarticulable.
guage in order that codified knowledge can be reconstituted as operational and generative. Moreover, as
more knowledge gets codified, the nature of the tacit
form is also generally changing. In other words, the
process of codification is a process of knowledge creation that alters both the codified and the tacit forms of
knowledge. 7 As new knowledge is codified, new concepts and terminology will inevitably be introduced so
that the codification of knowledge inherently involves
further creation of knowledge.
The combination and the composition of tacit and
codified knowledge depend strongly on the context
within which agents or organisations manipulate
knowledge. This means in particular that there are
contexts in which agents will be willing to invest
more on codification, and other contexts in which they
would rather use and reinforce their tacit knowledge.
Thus, the ability of a cognitive agent to exploit different categories of knowledge matters. On the one hand,
the existence of given tacit forms of knowledge (beliefs, languages, know-how), of accumulated learning
and habits, and of norms will shape the ways codified
knowledge is produced. On the other hand, the way
codified knowledge is produced (the nature of the
codes, the types of organisation, the nature of physical
carriers of knowledge) will also shape the ways learning processes are directed, focused and assimilated.
The context-dependence nature of knowledge clearly
suggests that tacit knowledge also refers to knowledge
which is not mobilised (at least consciously) when
conducting some activities in a given context. To this
extent, some codified knowledge can be made tacit by
an agent, who places part of her knowledge in a zone
of subsidiary awareness, whereas some other part is
put under her focal awareness (Polanyi, 1962).
The context-dependence of the codification process
is particularly important when considering the type
of tacit knowledge that is the point of focus of the
codification effort. The nature and the costs of the
codification process is not the same if one aims at
codifying a belief, a representation of the world, a specific know-how, or the mastering of a given language.
Cowan this issue showed that in the case of expert sys7 As Cowan et al. (2000) underlined in the case of the design
of expert systems, the act of codifying is not merely translating
the experts knowledge out of his head and onto paper, but is
typically an act of knowledge creation.
1565
tems analysed by Hatchuel and Weil (1995), different

types of know-how (the artisan, the repairer, the
strategist know-how) lend themselves with different
degrees of compliance to the codification process. 8
Moreover, even in the cases where the codification
processes fail to come at a reasonable cost to a usable codified form, and lead to only partially codified
knowledge, there could be some indirect advantages
of the codification efforts in terms of increase of
some form of tacit knowledge. For instance, there are
numerous cases in industry where ISO certification
efforts, as attempts to codify some tacit knowledge
embodied in collective practice, did not succeed or
just partially succeeded. But they generally lead to an
improvement of the strategic vision of the firms concerned (Benezech et al., this issue). More precisely
the analysis of such examples suggests, through a
codification process, the transformation of some tacit
knowledge (embodied in collective know-how) into
higher level types of tacit knowledge (embodied in
the strategic vision and representation of the firm).
When emphasising the importance of the context
in the analysis of the relationships between tacit and
codified knowledge, Polanyi (1962) showed that what
matters is the degree of attention of the cognitive
agent. In a modem economy, the rapidity of knowledge production and codification processes and the
low and decreasing costs of storing codified knowledge, make the problem of degree of attention more
and more acute. It is attention rather than information
that is becoming a rare resource, as screening and
selection of information become important functions.
Information abundance is generating a problem for
agents to discriminate between information which
are important to store and memorise, and information that can be simply put in the basket. 9 The
more knowledge is being stored and made available
8 Relatively linear direct processes with a known and fixed goal,
like those pursued by the artisan, can be easily codified into a
list of instructions or decision rules which can be implemented
by a machine. Activities that involve significant amount of pattern
recognition, generalisation and use of analogy are more difficult. . .
Forming strategies, trading different, conflicting goals off against
each other is harder yet (Cowan, 2001).
9 One should note that if you look at the web today, people are
storing too much, and putting too little in the basket. The same
is true for most people when they look at their old email we
keep way too much.
1566
in codified form, the more knowledge management

problem changes shape. To use Lundvalls taxonomy,
when the knowledge we seek (to understand) is tacit,
know-who is extremely important. When it is stored
in codified form, know-where becomes important.
A new skill must be cultivated, namely, how to find
things using, for example, search tools on the web. 10
As we will see later, the question of degree of attention is a key phenomenon to understand the process
of building competencies by firms.
1.3. The collective nature of knowledge
production, distribution and use
Up to now, we only referred to a cognitive agent
in isolation. But of course, the process of codification has also a strong collective dimension. 11 The
building of a common knowledge, the need to set
collective rules and languages to facilitate the formation of knowledge, or the need to mobilise socially
the dispersed forms of individual knowledge (Hayek,
1945), underline that knowledge results from a collective construction process. Social processes contribute
to shaping the way knowledge is produced and circulates. In particular, the codification process takes
place within specific communities, where the models,
languages and messages are built by agents sharing
a common understanding. 12 In particular, the role of
10 This has two aspects: the first is how to find likely documents
(this includes how to find things, and then how to reduce the number of things found to something manageable ask AltaVista for
documents containing the words knowledge and codification
and you get 2342 hits!). The second tool that has to be developed
is how to judge quality. Because the cost of codifying (and publishing) has fallen so much, it is too easy to diffuse knowledge.
There is a lot of codified knowledge available on the web that
is misleading or just plain wrong. How to teach people to filter
the good from the bad may be an important issue.
11 As Nonaka and Takeuchi (1995) emphasised, the process of
creation, distribution and use of knowledge in society relies on
the following two main dimensions:
the first one is the epistemological dimension which is the

critical assumption is that human knowledge is created and
expanded through social interaction between tacit knowledge
and codified knowledge knowledge conversion;
the second one is the ontological dimension which is concerned with the levels of knowledge creating entities (individual,
group, organisational and inter-organisational).
12 As mentioned by Cohendet and Steinmueller (2000), effective
communication cannot be separated from the influences govern-
communities appears as essential to understand the

growing phenomenon of virtual communities, that
work across space, time and organisational boundaries,
with links strengthened by information and communication technologies.
Among the main communities that participate in the
building of knowledge, Cowan et al. (2000) emphasised the role of epistemic communities, whose objective is the deliberate production of new knowledge.
Another relevant type of communities, the communities of practice, is suggested by Wenger (1998).
Communities of practice could also be active in the
construction of knowledge, even if for them the production of new knowledge is not a deliberate objective.
These two types of communities are detailed below:
Epistemic communities can be defined as groups
of agents working on a commonly acknowledged
subset of knowledge issues and who at the very
least accept a commonly understood procedural authority as essential to the success of their knowledge activities (Cowan et al., 2000). Epistemic
communities can, thus, be defined as a group of
agents sharing a common goal of deliberate knowledge creation and a common framework allowing
the shared understanding of this trend. An epistemic community may then be concretely defined
as a framework providing sufficient incentives for
members of the community to make, knowledge accessible to the group, through their contribution to
the process of codification. 13 What characterises
an epistemic community is also that the messages
emitted by the community can freely circulate to
the outside world. 14 What defines an epistemic
ing the coherence, homogeneity, and sustainability of particular
communities.
13 The concept of epistemic communities was developed in particular in the realm of international relations (Haas (1992); Adler
and Haas (1992)). Related concepts are also to be found in sociology of science. In this domain, one may mention Barber (1952)
who asserts that scientists tend to create self-regulated communities and Knorr-Certina (1981) who developed the concept of
scientific communities. One may also quote Beyssade (1998) who
studies linguistics and, in this field, uses the notion of epistemic
community and stresses the importance of a common language as
a cement for such communities.
14 Once these messages have been validated by reference to the
procedural authority of the community (for instance, once an article
has been published in the case of scientific communities). This
also obviously requires that the receptor of the message has the
community is the existence of a procedural authority

that can be explicit or not. Moreover, the procedural
authority conveys the idea of progress toward the
cognitive goal set by the community. 15 Epistemic
communities are, thus, structured around a goal to
be reached and a procedural authority endowed by
themselves (or with which they were endowed) to
fulfil that goal. Because of the heterogeneity of the
agents, the objective of knowledge creation for the
sake of knowledge, the first task of epistemic communities is to create a code-book. 16 Validation
of the cognitive activity of an agent is made with respect to the procedural authority. What is evaluated
is the contribution to the endeavour toward the goal
to be reached, according to the criteria set within
the procedural authority. Within an epistemic community, agents are bound together by their commitment to enhance a particular set of knowledge. The
recruitment rule is, thus, defined with regard to the
contribution an agent makes to fulfil this goal.
The concept of communities of practice was introduced by Lave and Wenger (1990) who, by
focusing on individual practices, identified groups
of persons engaged in the same practice, communicating regularly with one another about their activities. This goal is reached through the construction,
the exchange and the sharing of a common repertoire of resources, and through the circulation and
regular benchmarking of best practices (Wenger,
1998). Within communities of practice, the privileged knowledge is thus essentially the know-how
(Brown and Duguid, 1991), which is tacit and socially localised. Members of a community of practice essentially seek to develop their competencies
in the practice considered. As a result, the commuabsorptive capability to assimilate the message.
15 The belonging of members will, thus, be evaluated with respect
to this procedural authority. It should be noted that this procedural authority can a priori emerge from the interactions among
members.
16 Hence, knowledge circulating within epistemic communities
is explicit. At this point some authors (Nonaka and Takeuchi,
1995; Baumard, 1999) make a difference between explicit and
codified. The knowledge that circulates in a given community
is explicit not codified since it remains mainly internal to the
community (Baumard, 1999). Because of the lack of deeply shared
values, it appears that the knowledge creation mode is much like a
form of externalisation (conversion of tacit into explicit knowledge
(Nonaka and Takeuchi, 1995)).
1567
nity tends to send no messages toward the outer

world. Messages are almost exclusively exchanged
among the members of such a community. Wenger
(1998), Brown and Duguid (1991, 1998) state that
self-organisation is an essential characteristic of
communities of practice. 17 The self-consciousness
is visible in the mutual commitment of the community that binds agents in a social entity. This
mutual commitment ensures cohesion of the community and recruitment of new members. It is built
around activities commonly understood and continually renegotiated by its members. A communitys
member feeds it with his/her experience and relies
on the knowledge capitalised by the community to
carry out his/her activity. 18 Within the community
a jargon is, thus, progressively developed. This
jargon is understandable by the members of the
community only. The production of new knowledge
is, thus, not the deliberate goal of a community of
practice. But it could be an important by-product
of the functioning of the community. Furthermore,
though the process of codification is not the main
point of focus of the community, it is required at
many levels, in particular when the circulation and
benchmarking of best practices are at stake. The
technology of replication of best practices, as
Cohen et al. (1996) mentioned, usually implies (i)
learning a language within which to code successful routines, (ii) creating cognitive artifacts that can
be diffused, (iii) translating the high-level description contained into the cognitive artifact in actual
practice, generating a new routine adapted to the
new context.
Epistemic communities and communities of practice are self-organised communities, that usually cross
the boundaries of existing organisations. Within these
latter, one can find the traditional hierarchical communities, such as the functional work groups and the
multidisciplinary teams within firms. The ways communities self-organised ones, as well as hierarchical
17 More precisely, autonomy and identity of communities, the key
characteristics of self-organisation allow the collective acquisition
and processing of stimuli from the environment (Wenger, 1998;
Dibbiaggio, 1998). Identity and autonomy are essential for the
agent to define him-/herself with respect to his/her environment
and for the members of the community to behave collectively.
18 For instance, these processes take the shape of war stories
(Brown and Duguid, 1998) that members tell when they gather.
1568
ones interact will considerably influence the process of formation of knowledge. 19

An important aspect is that the nature and the type
of organisation of the community could change over
time. A community of practice could for instance
evolve into an epistemic community, as Callon and
Rabeharisoa (2001) showed in the case of an association of parents having children suffering from muscular dystrophy. 20 The other way round an epistemic
community could turn into a community of practice,
as this was originally the case for Linux. 21 In a first
phase, this free software was the mere project of a
student, Linus Torvalds, who clearly belonged to an
epistemic community. Then, after having written the
core, he published the source code on the Internet to
receive feedback in order to enhance his programme.
Thus, the second phase of Linux relied on the functioning a community of practice, a subset of the
community of hackers, where members were not
seeking to enhance the body of knowledge in computing science, but to improve their skills and to gain
reputation. Anybody having the required technical
skills and the desire to participate could bring his/her
contribution. The Linux community, thus, emerged
and organised itself spontaneously. 22
19
As underlined by Cohendet et al. (2001), in the traditional vision of the firm that dominated in the 20th century, there was a
clear delineation between the communities in charge of a deliberate production of knowledge (epistemic communities) and the
communities focused on the use of knowledge (these include the
hierarchical communities and autonomous communities such as
communities of practice). There is now less and less separation
between the communities in charge of producing new knowledge
and the communities in charge of using and transferring knowledge. The ability to proceed to on-line learning (as we shall see
further with the development of experimental learning) affects all
the types of communities that where not focused on the deliberate
production of knowledge.
20 In a first step, some parents gathered through a mutual commitment, by exchanging best practices, in order to improve the
situation of their children. The community (of practice) attracted
quickly new members, and succeeded in cumulating a critical mass
of knowledge that allowed in turn epistemic communities to function (in particular scientific ones) in order to continue the process
of knowledge building.
21 It is an operating system based on Minix, a small operating
system for PC inspired from Unix.
22 However, in order to cope with the flux of contributions, a
committee in charge of evaluating the incoming contributions was
created. This committee was then apart and above the bunch of
developers. There was thus the creation of a procedural authority
The development and expansion of the knowledgebased economy will enhance the role of communities. They are the nucleus of knowledge formation
and use. They encapsulate incentive mechanisms to
produce new knowledge, either through passion or
commitment to the collective action, or through the
adherence to a procedural authority as was the case
for the epistemic communities. The role of communities is also essential in the distribution and dissemination of knowledge throughout society. The process
of codification aims also at reaching one or several
communities (some codification processes aim at the
same community in which the building of knowledge
is taking place, in general codification processes are
directed towards specified communities, sometimes
the codification has a universal vocation). To a large
extent, the process of diffusion of knowledge can be
seen as a process of sequential transmission of knowledge through communities. This interactive process
supposes active means of mediation of knowledge,
but it also reveals the ambivalent role of the codification of knowledge. As noticed by a recent OECD
report (2000, p. 27), on the one hand, the production
and use of highly specialised codes or codes using
technical or local jargon would create an obstacle to
appropriation in the field by lay people or potential
users of the knowledge. On the other hand, a lack of
codification would also create an obstacle as users
would not have access to sufficiently explicit knowledge. This ambivalence indicates the importance
of designing and implementing metacodes or semicodes as mechanisms for developing compromises
between the need to make knowledge more explicit
and the need to avoid excessive technicalities and
local jargon. In this perspective, one can measure
how much the dynamics of social networks and the
co-evolution of codification methodologies and social
organisations are interrelated.
To a large extent, the process of codification of
knowledge will, thus, exhibit different modes and patterns according to the community within which the
process of codification is undertaken, and the community (or communities) which is (are) the target of the
codification process. The community at the origin of
to judge what constituted a progress toward the objective. It should
be noted that the objective evolved from the enhancement of a
student project to the building of a competitor for Windows NT.
the process and the community which is the target of

the process can be the same, as in the case of communities of practice. But this is not generally the case.
One can envisage the following two extreme cases:
On the one side, there could be a vertical processes of codification, where a given community
tries to codify a body of knowledge which is (tacitly) held by another community, generally to take
control over it. The development of Taylorian procedures could thus be seen as an effort of codification of knowledge by the communities of managers
to absorb and control the tacit knowledge held by
the communities of shop floor workers. In the same
perspective, the development of ISO certification
processes may be seen as an attempt by large firms
to control part of the knowledge held by small and
medium enterprises. 23
On the other side, there could be horizontal process of codification where the community at the
origin of the codification effort aims at reaching
the world, without any restriction on the type of
community concerned. The code-book produced
has, thus, a universal ambition.
In general, a process of codification reveals an intermediary pattern between these two extreme ones.
This has an important implication, which is the following: a process of codification carried out by one
community generally aims at explicitly reaching some
communities. But it may also exclude (implicitly or
explicitly) other communities. The building of codes
is selectively and carefully done (by selecting appropriate models and languages) to avoid the messages
to be understood by the excluded communities. A
process of codification has, thus, two faces. Numerous private organisations develop internal specific
codes (that they may eventually share with a selected
number of key suppliers) to prevent competitors from
accessing their organisational secrets.
23 In an interview carried out within the TIPIK project, an interesting case was observed in a company where the codification
process of engineering procedures in quality control, led to the
depravation of the use and control of knowledge by the engineers to the benefit of the lawyers of the company. These latter
were able to manipulate and control strategically the new codified
knowledge. The engineers emphasised how a (vertical) codification process is inherently associated with a shift of power to the
benefit of those who control the process.
1569
The above developments have deliberately emphasised two main dimensions of the transformation of
knowledge, the tacit/codified dimension and the individual/collective dimension. However, as underlined
by Malerba and Orsenigo (2000), the multiplicity and
the contextuality of other features of knowledge are often underestimated. There are many other dimensions
of knowledge (declarative knowledge versus procedural, procedural knowledge versus generic knowledge)
that require a careful examination which is beyond the
scope of this contribution.
1.4. Benefits and costs of the codification process
As it was mentioned in the introduction, the main
expected benefit from the process of codification
is that knowledge is picking up some properties of
a (private) commodity. Codification facilitates the
commodification of knowledge. Knowledge becomes transferable independently of the transfer of
other things, such as people, in which the knowledge is embedded (Cowan and Foray, 1997). This
implies that market transactions are facilitated by
codification, reducing uncertainties and information
asymmetries in transactions involving knowledge. 24
It was shown for instance that the production of
quality assurance standards (Gunby, 1996) reduced
the asymmetric information on the related markets.
In a similar perspective, codification also facilitates knowledge externalisation and allows a firm to
24 It should be emphasised that the use of knowledge is at the
origin of very specific asymmetries. As Bessy and Brousseau
(1998) noted: Regarding its value, knowledge is a highly uncertain
resource when it is transferred between two economic units. This is
partly due to the well-known information asymmetry about quality
inherent in every market transfer of information (Arrow, 1962).
But it is also due to the radical uncertainty over the potential use
of knowledge. Two different phenomena are in question: (1) the
receiving party can use knowledge in ways that are not easily
observable by the prior holder. This is typically a moral hazard
problem (hidden action), but in a specific informational context
where it is impossible to build ex ante incentive schemes or
(infaillible) supervision mechanisms because all the potential uses
of knowledge cannot be anticipated and made observable, and (2)
knowledge is itself an input in the process of knowledge creation.
The unit that benefits from a transfer can be conducive to the
creation of new knowledge, that will increase the value of the
initial knowledge. The question is then how to remunerate each
party for its contribution to the knowledge creation process. Again,
due to the uncertainty of such a process, it is very difficult to
solve it ex ante by an optimal incentive scheme. . . .
1570
acquire more knowledge than previously for a given

(but not necessarily lower) cost. Thanks to codification, organisations will find more opportunities to
purchase knowledge from outside (instead of producing it internally). This may be related to the current
trend of firms towards outsourcing. But besides these
main types of commodification of knowledge, many
other forms of benefits from codification can be
acknowledged:
The process of codification allows the modularization of knowledge. This has positive potential
impacts in terms of cost reductions, for it contributes to facilitating the division and the dispersion of knowledge in different phases or domains
(Machlup, 1983). The division of knowledge is
a result of the division of labour in the field of
knowledge production. The dispersion of knowledge is related to local situations in which knowledge is produced (a site, a workshop, a laboratory).
However, there could be also negative impacts
from knowledge being more specialised and dispersed. An increase in the division and dispersion
of knowledge makes it more and more difficult for
economic agents to locate and retrieve elements
of knowledge that would be particularly useful to
them. It is probably not disputable that the division
of knowledge is increasing over time (specialisation), raising the marginal cost of knowledge
integration. The dispersion trend is less clear, but
one can expect a higher dispersion as knowledge
production becomes more collectively distributed
(located in many places). And this increasing tendency of knowledge division and dispersion again
makes the problem of memory more acute. How to
build storage processes that are integrative; that is
to say which are not just reproducing the state of
division and dispersion of knowledge as it was at
the moment of its creation?
Through the process of codification, knowledge
may be used by agents as a signal. The signal could
indicate a specific capability or a competence for
an agent who wants to enter a specific commercial
relationship (such as a joint venture), a strategic
development partnership, or a given network. A
better signalling capability for agents facilitates
the formation of reputation and trust. This in turn
lowers the risks to enter into contractual rela-
tionships when co-operative schemes seem more

adapted.
Codification directly affects the speeding-up of
knowledge creation, innovation and economic
change. 25 Moreover, codification has significant
consequences for the technologies of learning
(David, 1994). It contributes to increasing the value
of on line learning (learning by doing, learning by using). It facilitates the emergence of new
forms of learning, such as experimental learning,
that make the differentiation between on-line
and off-line learning activities less and less relevant. 26 It contributes to increasing the speed and
decreasing the cost of developing prototypes. It
contributes to enhancing the ability to generate
technological options.
Codification reduces some of the costs in the process of knowledge acquisition and technology diffusion. It improves the reliability of information
storage and retrieval. However, when reflecting the
recent history of the use of information and communication technologies (ICT) to achieve direct
codification of knowledge that is related to individual and organisational memory, Steinmueller (2000)
25 For instance, in the case of pharmaceutical industry, Nightingale
(2000) showed that the ability to automate and accelerate dramatically the trial and error process, within a particular context chosen
by the researchers, is a powerful tool for knowledge creation.
26 Experimental learning takes place on-line, that is to say during the process of producing a good, and consists in deliberately
experimenting during the production process. By doing so, one
creates new options and variety. This form of learning is based on
a strategy whereby experimentation allows for collecting data, on
the basis of which the best strategy for future activities is chosen.
For example, a professor can undertake pedagogical experiments;
the craftman can seek new solutions to a problem even during the
fabrication process. The possibility of moving this type of learning
in many activities represents an important transition in the historical emergence of the knowledge-based economy. In effect, as long
as an activity remains fundamentally based on learning processes
that are routine adaptation procedures and leave no room for programming experiments during economic activity, there remains a
strong dichotomy between those who deliberately produce knowledge and those who use and exploit it. When an activity moves to
higher forms of learning, and where the individual can programme
experiments and obtain results, the production of knowledge becomes much more collectively distributed. . . With the emergence
of experimental learning, the feedback and reciprocal links that
tie on-line learning processes and in house R&D together- and
whereby a potential creative activity effectively contributes to the
production of knowledge become crucial (OECD, 2000, p. 25).
showed that the contribution of ICT has been relatively modest.

The potential benefits from the codification processes
that have been underlined, should be compared to
the costs of codification in order to carefully assess
the real value of codification for society. In fact, the
balance between benefits and costs of codification
is inherently context-dependent. For instance, in the
contexts of production processes, Balconi (2000)
showed that the cost of codification increases: (i) in
relation with the level of product differentiation, since
firms specialised in customised products will rely less
on codified procedures and specifications than firms
producing standardised goods; (ii) with the number
of variables that influence the performance of a process. The case study of Benezech et al. (this issue)
on completion of knowledge codification considered
ISO 9000 as a code which provides an opportunity
to communicate. The paper focuses on how this code
might be used within firms to produce codified knowledge. Consequently, this study explains the standards
implementation process and addresses the learning
process inside organisations which contributes mainly
to knowledge accumulation within firms.
More generally, Cowan and Foray (1997) suggest to
distinguish between stable contexts and contexts of
change in order to assess the net value of codification.
On the one side, stable contexts refer to contexts where
languages and models pre-exist (fixed costs have generally been sunk). In such contexts, Cowan and Foray
have identified classes of situation where the codification generally leads to net benefits. 27 On the other
side, when models and/or languages have to be built,
one can refer to contexts of change. In such contexts,
there could be situations of excess of codification,
when the accumulation of codes can prevent the development of radically new knowledge, simply because
27 In stable contexts, Cowan and Foray (1997) argued that on the
benefit side, the efficiency of codification would be greater in very
large systems having specific requirements regarding coordination
between agents, and identified five classes of situations: (i) systems
including many agents and many locations, (ii) systems strongly
based on recombination and reuse and which took advantage of the
existing knowledge (rather than of independent innovation); (iii)
systems that required recourse to detailed memory; (iv) systems
which needed particular kinds of description of (what (and how)
the agents did; and lastly, (v) systems characterised by an intensive
use of information technologies.
1571
explaining and understanding it would require entirely

new codes. As argued by Arrow (1974), codification
entails organisational rigidity and uniformity while
increasing communication and transaction efficiency.
The need for codes mutually understandable within
an organisation imposes uniformity requirements on
the behaviour of participants. They are specialised in
the information capable of being transmitted by the
codes, so they learn more in the direction of their activity and become less efficient in acquiring and transmitting information not easily fitted in the code. It
is also in these contexts of change, where codification
implies the burden of a fixed cost, that small entities
(SMEs, for instance) might find codification too costly
and prefer to take the risk to rely on the accumulation
and use of tacit knowledge.
In order to test the above theoretical hypotheses,
the TIPIK teams carried out a series of empirical case
studies. The 12 case studies, presented in this special issue, examine different forms of codification of
knowledge. They emphasise various perspectives of
the process of codification, from research activities
in high tech industries to the writing of basic ISO
procedures in traditional SMEs. They relied on different methods (postal survey, semi-structured interviews, reinterpretation of existing data, etc.). For all
these reasons, the results from the case studies cannot be considered as comprehensive or representative.
However, despite the variety of situations and perspectives, the case studies revealed a common main
conclusion. What appears as particularly remarkable
is that the focus on codification of knowledge is not
limited to the research area and high-tech industry. In
all social activities, whether in the production, mediation or use of knowledge, the process of codification of knowledge is becoming one of the main
drivers of evolution. An in-depth understanding of the
incentives to codify, of the nature and form of the
codification process, are thus, becoming essential for
analysing the process of innovation and growth of the
economy.
2. From theory to policy

The theoretical analyses that has been presented,
as well as the empirical case studies based on this
theoretical framework suggest a renewed vision
1572
of the traditional research, technology and development (RTD) policies. These traditional policies
were inspired by the vision of linear models of innovation, where the production of new knowledge
could be assimilated to mere information. The rationale for designing policy instruments in this traditional vision is based on supplying market failures due to externalities in the production of new
knowledge-reduced-to-information. Several ways of
intervention have been proposed, and often tested, in
order to move more closely to the optimal level of
research. David (1993) speaks about the three Ps:
public patronage (prizes, research grants, subsidies,
etc.), state procurement (or production) and the legal exclusive ownership of intellectual property that
shape the nature of government interventions to supply the failure of the market. The focus of traditional
RTD policies is thus essentially on the conditions of
production of new knowledge, and not on the ways
knowledge is assimilated and diffused through society, since, according to the traditional vision, agents
are supposed to be able to assimilate new knowledge
without significant costs.
As the knowledge based economy is growing, there
is a crucial need to think in terms of knowledge oriented policies (KOP) which could take into account
the specific characteristics of knowledge that have
been examined, in particular the specific impacts of
the processes of codification. Contrary to traditional
RTD policies, KOP refer to the non-linear interactive model of innovation, and assume that knowledge
and competencies are crucial factors for production,
innovation and competitiveness rather than physical
resources. Highlighting interactions that allow the generation and diffusion of knowledge, this approach differs from linear models (science-push, market-pull) for
which knowledge constitutes only one input among
others intervening in a sequential process. In particular, a key difference is that, for KOP, the way knowledge is assimilated and acquired by agents (the ways
competencies are built), is as important as the conditions of production of new knowledge (the ways innovations are produced).The KOP should promote ways
of codifying knowledge that contribute to enhancing
the generation of knowledge externalities, since the
process of knowledge generation is cumulative and integrative. Cumulative forms of knowledge are those
in which todays advances lay the foundation for suc-
ceeding rounds of progress. 28 Thus, the process of

knowledge generation produces positive learning externalities. The generation of a new piece of knowledge increases the probability of creating useful new
products, processes and ideas arising from novels and
unanticipated combinations. As Machlup (1983) put it,
The more is invented, the easier it becomes to invent
still more. 29 The rationale for KOP is that purely private forces cannot guarantee satisfying conditions for
the production, mediation and use of knowledge in the
economy. There are not only risks of market failures
when dealing with knowledge generation and diffusion
in society, there are also potential learning failures in
the ways knowledge is acquired, assimilated, and used.
Therefore, KOP stress the importance of government
interventions for knowledge generation and diffusion
as well as for the building of skills and competencies.
In order to compare KOP with traditional policies,
and to design policy recommendations to implement
efficient processes of codification, one has to recon-
28 Scotchmer (1991), when emphasising the cumulative nature of

inventive activity, compared it to the effect of raising the height
of the shoulders of giants from which one can see further, rather
than obscuring the view.
29 These positive learning externalities are based on various knowledge interactions, such as in the mining analogy proposed by
David and Foray (1995): First knowledge is like surveying: it
generates maps that raise the return to further investment in exploration and exploitation (David et al., 1994). Second, it has been
observed that mineral production in an area leads to localization
of exploration, so that at least for some time, mineral reserves
become larger in the territories where exploitation is underway
(David and Wright, 1997). When discovery increases the probability that others will undertake exploration in the neighborhood,
producing knowledge is likely to generate positive externalities
for the explorers and each agent has an interest in diffusing the
product of his discovery so as to profit from the results of others.
At the very least, information about where others have failed to
make a discovery will be valuable in guiding ones own search.
This third aspect of positive externalities deals with the migration
of young investigators into new fields-colonizing new areas with
tools and concepts developed elsewhere. As in the case of exploitation of minerals, this frontier expansion requires falling yields
in colonized areas to generate breakthroughs into new territory.
Many studies have shown that the introduction of new techniques
of codification of knowledge has positive impact on each of the
steps of the mining analogy (increasing the ability to design better
maps at a lower cost, increasing the diffusion of the results of the
mining, improving the possibility to reproduce elsewhere efficient
schemes of discovery), which implies significant possibilities to
increase positive externalities in general.
sider policies instruments in the following two main

domains.
in the domain of production of knowledge (when
knowledge is considered as an input for society);
in the domain of building of competencies (when
knowledge is considered as an output for the economy).
2.1. The production of knowledge: a renewed vision
and the role of codification
2.1.1. The traditional vision
The traditional vision of knowledge production
dates back to the seminal contribution given by Arrow (1962) 30 dealing with knowledge creation in the
firm. For Arrow, the process of invention can be interpreted as the production of new knowledge, which in
turn is comparable to information. Arrow stressed that
in such a context, the production of new knowledge
faces the key problem of appropriability. 31 He emphasises that it is difficult or even impossible to create
a market for knowledge once it is produced, so it is
difficult for producers of knowledge to appropriate
the benefits which flow from it. 32 Arrows proposals
rely on a body of several main hypotheses, some very
explicit, some rather implicit, but important to underline for our purpose. Therefore, we remind Arrows
hypotheses:
1. Knowledge treated as information possesses
the generic properties of a pure public good.
It exhibits the following two conditions for being considered as a public good (non-rivalry and
non-exclusion):
First, codified knowledge is a non-rival good,
that is, a good which is infinitely expansible
30 One can also mention the seminal contribution given by Nelson
(1959) on the production of knowledge in basic science.
31 First the problem of uncertainty (which means that knowledge
outputs are not predictable from inputs, and that classical market
mechanisms for sharing risks, such as insurance, will rarely apply
to research activities); second, the problem of indivisibilities (the
underlying knowledge must exist on a certain minimum scale
before any production can take place at all, and this necessary
minimum is independent of the rate of production).
32 There is a fundamental paradox in the determination of demand
for information; its value for the purchaser is not known until he
has the information, but then he has in effect acquired it without
cost (Arrow, 1962).
1573
without being diminished in quality, so that it

can be possessed and used jointly by as many as
care to do so. 33
The second property concerns the characteristics
of information as a non-exclusive good. A good
is exclusive if it is relatively easy to exclude
individuals from benefiting from the good once
it is produced. A good is non-exclusive if it is
impossible or very costly to exclude individuals
from benefiting from the good.
2. The only incentive that matters for the producer of
knowledge is to experience the full ownership of
the new piece of knowledge produced. There is no
trade-off between the incentive to be the sole owner
of the innovation and other forms of incentive that
could influence the behaviour of the producer of
new knowledge.
3. The producer of new knowledge is a solitary one.
In Arrows perspective, the producer of knowledge
acts in isolation. Nothing is said about the complementary forms of knowledge that have been necessary for him to invent. 34 Nothing is said about
the community of agents who supported him in the
process that lead to the invention. Nothing is said
about the interest to him of the new piece of knowledge that has been produced (is it an incremental
invention aiming at improving a current process. Is
it a radically new invention opening the perspective of new fields of research?). In such a solitary
perspective, an important consequence is that the
producer of knowledge is in a position to claim the
totality of the invention.
4. The producer of new knowledge is facing the market. More precisely, the agents who may capture
for free the new piece of knowledge are anonymous. The mechanisms of externalities generated
by the producer of knowledge, on which the dif33 From an economic point of view, the property of non-rivalry is
essential: it means that the goods (re)production costs are fixed:
once the good has been produced, there is no need for continuing
investment because there are no production costs in replicating it.
(Callon, 1994). Non-rival goods are often privately produced (for
example private swimming pools or highways that consumers pay
to use) as long as non-payers can be excluded from consuming
them. This category of goods is sometimes referred to as club
goods.
34 We could assume either that he had all the capacities to invent,
or that the complementary forms of knowledge, he needed to
achieve his invention, have been bought in a market.
1574
fusion process relies, are isotropic. As in a market mechanism, one can refer to a representative
agent, who benefits from the knowledge spillovers
emitted by the producer of knowledge.
5. The producer of knowledge is not supposed to have
emitting capacities, that is the ability to tune the
disclosure/secrecy dimension. He is just supposed
to try and avoid the loss of the integrity of the piece
of new produced knowledge.
6. All the agents of the economy have the full capability to absorb the innovative idea emitted by the
producer of knowledge. Any buyer of the knowledge can effectively destroy the market, since he
can reproduce the knowledge at very low cost.
7. The epistemic content of knowledge does not
matter. The content of knowledge exhibits a
cognitive equivalence, which means that in such
a context it is impossible for instance to distinguish between the generic and specific forms of
knowledge.
From these hypotheses, Arrow showed that the
characteristic of knowledge-reduced-to-information
as a public good implies the existence of some major positive externalities that prevent the effective
functioning of a market for knowledge. Any buyer
of the knowledge can effectively destroy the market,
since he can reproduce the knowledge at very low
cost. If the producers of knowledge cannot appropriate the benefits of knowledge, then they have no
incentive to produce it. Without an external intervention, the incentives for doing research are not high
enough and the level of research in society will be
under-optimal. The consequences of this broad traditional vision, that assimilates knowledge to mere
information, were considerable. Since the knowledge generated by the different research activities
possesses the generic properties of public good, it
cannot be optimally produced or distributed through
the workings of competitive markets. Here is the
justification for government subsidisation of science,
technological and engineering research, and for innovative activity more generally (the3P). This vision
shaped the conception of public intervention in R&D
during decades. It justified the role and creation of
public laboratories, of research centres, of public
programmes of R&D, of public institutions (patent
offices, for instance), of public infrastructure for trans-
fer of technology. It explained why public efforts in

R&D were generally disconnected from applications
and why arguments upon the existence of spillovers
from public research programmes were so important
to justify public money spent on R&D. It suggested
that scientific production was in fact considered as
exogenous to the economic sphere, and governed by
rules and behavioural norms (reputation effects, peer
reviews, etc.) that were drastically different from the
norms and behaviours of industry (seeking profit and
technical efficiency). In particular in this perspective,
the choices of research themes by academics were to
remain independent of the objectives of industry.
2.1.2. Questioning the traditional vision
The idea that research produces only codified information is increasingly questioned. Dosi (1988), Pavitt
(1984) amongst others, stated that research does not
produce information, but knowledge of which some is
coded and the rest tacit. Cohen and Levinthal (1990)
argued that the degree of spillovers and imitation depends on both the nature of knowledge and the absorptive capacity of firms. All things being equal, the
more knowledge is codified, the easier its absorption
will be. But even in the case of codified knowledge,
the user or imitator needs certain know-how, and technical ability to benefit from the knowledge. To appropriate the results of academic research, even if it is
codified, one has to know the code. 35
For all the above reasons, the different hypotheses of the traditional model of knowledge production
should be carefully reconsidered, within the theoretical framework exposed in the introduction, along the
following hypotheses concerning the production of
knowledge in a KOP context:
1. Knowledge is not a pure public good. There is a
range of situations varying from the completely
appropriable to the completely public. Thus, to
sum-up this discussion and try to categorise the
35 From this point of view, to quote Joly and Mangematin (1996)
research activity has two complementary facets: it naturally contributes to the creation of information and knowledge, but it is
also a learning process which helps to increase absorptive capacity.
Not only are externalities not evenly distributed, but they increase
when the knowledge bases of firms are similar. In such contexts,
external research cannot be substituted for internal research: the
two are complementary.
different kinds of economic forms of knowledge,

following Romer (1993), we can consider that
Knowledge expressed in codified statements
strings of bits constitutes the prototype of nonexclusive and non-rival goods, which exhibit
a completely public character. If one renders
them appropriable, which is always possible,
but which implies the costs of reconfiguring
them or giving them legal protection, one would
create a sub-optimal situation.
Knowledge expressed as information codes or
encrypted messages are intrinsically non-rival,
but exclusive.
Knowledge expressed through a tacit form is a
rival good, with a wide range of modalities from
pure personal tacit knowledge to shared tacit
forms of knowledge.
2. Appropriation is not the only incentive for knowledge production. Firms do have other incentives
than the direct exploitation of the monopoly rent,
the sale of licenses or the advantage in negotiations offered by patents. The willingness to maintain the firm on the technological frontier, 36 the
search for reputation, the objective of signalling, 37
the need to build an absorptive capacity, entrance
to networks, and more generally the endeavours
of agents in building competencies, are amongst
the main other incentives for the firms to invest in
R&D.
3. The production of knowledge is not a solitary venture. It is generally produced within a community.
The community could deliberately aim at produc36 As Schumpeter argued, competition is about new products, new
innovations. It is a dynamic process: in capitalist reality as distinguished from its textbook picture, it is not price competition which
counts, but the competition from the new commodity, the new technology, the new source of supply, the new type of organisation. . .
This kind of competition is much more effective than the others
as a bombardment is in comparison as forcing a door (Schumpeter, CSD, 1942, pp. 8485). In this way, it would be a suicidal
tendency for a firm not to invest in knowledge production.
37 The growing number of publications by firms can be interpreted (Hicks, 1995; Meyer-Krahmer, 1997) as an attempt to find
new access to external knowledge and to signal the existence of
tacit knowledge and other unpublishable resources. By becoming
member of the club of academic activities, by paying an implicit
fee to access the epistemic communities of researchers, the firm
clearly expects a right of access to the academic tacit knowledge
in a given field.
1575
ing new knowledge, as is the case of an epistemic

community (Cowan et al., 2000). However, the
building of knowledge could also be made within
other types of communities such as the communities of practice. Networking between academic
institutions and private enterprises is a growing
phenomenon that takes different forms. Networks
can offer a way to share knowledge complementarities. It is also a way to build collective forms
of knowledge, and a sufficient level of trust between partners to facilitate the collective creation
of knowledge.
4. The producer of knowledge does not face the
market, but a specific structure of interaction of
economic agents. As Nonaka and Takeuchi (1995)
mentioned, organisational knowledge creation
should be understood as a process that organisationally amplifies the knowledge created by
individuals and crystallises it as a part of the
knowledge network of the organisation. This
process takes place within an expanding community of interaction, which crosses intra- and
inter-organisational levels and boundaries.
5. The producer of knowledge has emitting capacities. An agent producing new knowledge will generally operate a selection between communities: on
the one side, he will consider to which communities the new knowledge is addressed, and on the
other side, those communities that he chooses to
exclude. This brings forward in particular the issue of the disclosure/secrecy dimension (David and
Foray, 1995). 38
6. The other agents do not have full absorptive capabilities to absorb the innovative ideas emitted by the
38 The extent of disclosure is a continuous variable bounded by
full disclosure at one limit and total secrecy at the other. The degree
of disclosure required is not uniform across intellectually property
rights regimes, and even with a given regime, different kinds
of text may be protected by varying completeness of disclosure.
Computer software, for example, may be copyrighted with-out
revealing the source code, and in some instance even the full
body of object code does not have to be disclosed. Standards
of disclosure may be defined not only by the statute laws and
intellectual-property agencies such as patent offices, but also by
the policies of the journals in which scientific papers are published.
For example, professional journals may or may not insist upon the
exact co-ordinates of complex proteins whose molecular structure
is being reported, or required disclosure of the computational
algorithm used in analysing experimental observations.
1576
producer of knowledge. First, firms cannot assimilate, absorb knowledge without any effort. 39 In
order to absorb new external information, the firm
needs to develop what Cohen and Levinthal (1990)
called an absorptive capacity. The principle is that
we cannot understand something if we know nothing. Firms need to build a knowledge background,
based on knowledge previously acquired, to be able
to absorb the external knowledge the two faces of
R&D. It not only creates new knowledge, it also
helps firms to assimilate external knowledge. Thus,
Cohen and Levinthal show that spillovers do not
necessarily have a negative impact on R&D.
7. The epistemic content of knowledge matters. As
noted by Callon et al. (1999), we should distinguish
the case of knowledge having a high degree of generality (knowledge which can be potentially used
in various contexts by a large variety of agents)
from the case of very specific forms of knowledge that can be absorbed and used by few other
agents.
This reconsidered set of hypotheses leads to the following remarks in terms of KOP.
2.1.3. New interpretations of traditional hypotheses
It must be emphasised that the above hypotheses of
the conditions of production of innovative ideas in a
KOP context do not invalidate the traditional hypotheses. In fact, the traditional context appears as a particular case of the more general one. This implies that the
traditional policy instruments (3P) elaborated within
the traditional context, will still be valid as instruments
in a KOP context. However, the ways to interpret them,
39 As Saviotti (1998) mentioned Traditionally, knowledge was
considered a public good because it is impossible for its creator to
prevent it being used by economic agents who do not pay anything
in exchange of it. However, even a completely codified piece of
knowledge could not be used at a zero cost by anyone. Only
agents who know the code can use the piece of knowledge at
zero (imitation) cost. Agents who do not know the code, if they
realise the economic value of a given piece of knowledge, have
to know the code before being able to retrieve and to imitate. The
cost of learning has to be included among the costs of imitation. . .
the concept of absorption capacity tells us that in order to be able
to access a piece of knowledge, we must have done R&D on
something similar. As a consequence, we have included among the
variables determining the degree of appropriability, not only the
degree of codification, but also the fraction of agents knowing the
code and the distribution of agents with respect to the frontier.
the ways to design and implement them, the ways to

use them will be different. The very difference relies
in the fact the KOP hypotheses take into account both
the tacit and the collective dimensions of knowledge.
Thus, any classical policy instrument has to be reconsidered in the KOP context. For example, when
considering the key question of appropriability, it is
solved in the traditional R&D policy through patents,
viewed as strong property rights instruments. In a KOP
context, patents naturally keep their (necessary) appropriability dimension to protect the innovator, but they
reveal other dimensions. Patents have for instance a
more and more important strategic role in negotiation.
Very often patents are the first side of a co-operation
or knowledge exchange. In this way, they determine
the balance of power between the members of the network. Patents may, thus, be used as signalling devices
to enter complex negotiations related to the building
of networks. On the other side, the collective dimension of the building of knowledge brings forward new
ways to look at appropriability. When incentives to
work for building knowledge within a given community are strong enough (such as for belonging to specific communities of practice), the necessity to decide,
at least temporarily, on the question of appropriation
could be marginal. An extreme example is the recent
case of free software development, such as Linux.
The members strongly believed that property rights
were a threat for the users freedom and the dynamics
of innovation in industry. In order to allow free use of
each software, they adopted the copyleft system by
opposition with copyright. In short, the above developments suggest that the understanding of these recent
issues supposes to analyse in-depth the problem of redefinition of incentives in a new mode of production
of knowledge.
2.1.4. Reconsidering incentives in a KOP context
The introduction, beside the appropriability dimension, of other characteristics of the production of
knowledge (signalling, voluntary disclosure, increased
absorptive capabilities) leads to key consequences in
terms of incentives for producing new knowledge.
It underlines the key role of institutional settings in
shaping the incentives to innovate. Institutional settings, as expressed by norms, rules and standards to
be followed, govern, to a large extent, incentives to
produce and distribute knowledge among members
1577
Fig. 1. The role of incentive mechanisms: differences in Romers and DasguptaDavids approaches concerning the production of knowledge
(Callon, 1999).
of different social organisations. Institutional settings

also contribute to shaping the nature of the codification processes that are taking place. In particular, the
modes of organising research activities strongly influence the costs of transferring the knowledge that has
been produced. As an example, with regard to the differences between fundamental and applied research,
Callon (1999) recently emphasised what he considers
to be two extreme visions among economists of the
main ways to produce knowledge (Fig. 1):
On the one hand, there is the vision proposed by
Romer (1993), for whom the main difference between basic research and applied research resides in
the difference in the contents of knowledge which
are manipulated (basic science consisting essentially in codified statements having a large degree
of generality, and applied research consisting essentially in manipulating private tacit knowledge
in forms of know-how incorporated in workers or
equipments).
On the other hand, the vision proposed by Dasgupta
and David (1994) is that the difference would not
be found in the contents of knowledge (which, a
priori, would exhibit a cognitive equivalence that

leads to a strong substitutability between the two
forms), but rather in the institutional settings. The
incentive schemes and norms edicted by the institutional settings are the main explanation for the
reasons why codified forms are preferred by some
agents (researchers who have incentives to publish
articles, theorems, treaties, etc.) while tacit forms
are preferred by others (engineers working in private firms). 40
The above considerations emphasise the role of incentives as shaping the nature of the production of
knowledge. For example, when considering the status of incentives for research in the US, Stephan and
Levin (1997) mentioned: why have researchers in the
US focused so extensively on individuals as opposed
40 As expressed by David and Foray (1995), The true nature of
new knowledge does not stem from any intrinsic differences between knowledge that is scientific rather than technological, nor
between basic and applied scientific knowledge. The critical factor
governing the distribution and the utilisation of new findings are
those regarding the rules structures and behavioural norms about
information disclosure that dominate in the particular social organisations within which the new knowledge is found or improved.
1578
to groups and why has this focus persisted despite

widespread evidence that science is becoming increasingly a collaborative effort? It is virtually impossible
for a scientist to survive and have a career at a university without becoming a principle investigator
(PI) and directing a lab. The research the PI directs
is collaborative, but the majority of the collaborators
are graduate students and post-docs statuses which by
their very definition are temporary. This individualistic
vision of incentives is in accordance with the Mertonian model of scientific activity, where the individual
trajectory of the researcher and his/her capacity to accumulate a stock of credibility is the main driver of the
academic domain. This leads for instance to the well
known Matthews effect: as public funding of scientific research is related to previous accomplishments,
the system may give disproportionate recognition to
scientists who attained early discoveries (Diamond,
1996). 41
Cowan and Foray (1997) indicated some other types
of issues related to the setting up of incentive mechanisms. In particular, they referred to a series of recent
works that raised the question of the difference between a system of awards (ex post recognition) and a
system of grants (ex ante stimulation), and its implication for the efficiency of research.
2.1.5. The role of trust
Besides the institutional settings, there is another
variable to be considered as influencing the behaviour
of the innovator in a KOP context: the degree of
trust between different agents involved in the production of knowledge. Taking into account the degree
of trust raises an important issue, which contributes
to influence the choice between specialisation and
41 A very different point of view on incentives has been recently
taken by other countries, for example the UK, where the lab
is the unit of reference for reputation and other mechanisms of
incentives. In some cases, one can see a tendency to reward
beyond the frontier of the lab, the network of research centres
and laboratories that have produced the new knowledge. A recent
example, quoted by Joly (1997), is the publication in Nature of
an article on the sequence of chromosome III in yeast. The article
was signed by 147 researchers from 40 different institutions of
research. It is one amongst many signs of the tendency to consider
research as a real collective enterprise. Depending on the types
of incentive mechanisms that will be promoted in the future by
public authorities that support research, this tendency has some
chances to become a dominant mode of production of research.
co-operation in the production of knowledge. As

mentioned by Zuscovitch (1998), Trust is a tacit
agreement in which rather than systematically seeking
out the best opportunity at every instant, each agent
takes a longer perspective to the transactions, as long
as his traditional partner does not go beyond some
mutually accepted norm. Sharing the risks of specialisation is an aspect of co-operation that manifests an
important trust mechanism in network functioning.
Specialisation is risky business. One may sacrifice the
horizontal ability to satisfy various demands in order to gain vertical efficiency in an effort to increase
profitability. Any specialising firm accepts this risk,
network or not. A risk-sharing mechanism is essential
because, while aggregate profits for participating firms
may indeed be superior to the situation where firms
are less specialised, the distribution of profits may be
very hazardous. To make specialisation worthwhile,
the dichotomous (win-lose) individual outcome must
be smoothed somehow by a co-operative principle of
risk sharing.
The choice for an agent A to specialise in one
domain of knowledge (and to accept to bear the
sunk-costs for that) in co-operation with other agents
that would accept to specialise in turn in the complementary types of knowledge that are necessary
for A (while As knowledge would be considered as
complementary viewed by the other agents) seems to
be one of the main lines of research to understand
the management of knowledge by organisations.The
question of trust in relation with the production of
knowledge brings forward a key topic in the KOP
context: the problem of access to knowledge. This
is a crucial variable in several of the studies within
this volume (e.g. Creplet et al., DAdderio et al.).
Within networks, there is a considerable amount of
negotiations decentralised at the level of economic
units that are taking place with the objective to determine the mutual degree of specialisation in order to
produce knowledge and the mutual rights of access
to knowledge which is compatible with the system of
specialisation. The key variable in this perspective is
the degree of trust between economic units. This is by
nature a very decentralised process, however, an adequate policy to favour an efficient co-determination of
specialisation and system of rights of access to knowledge in society is to favour as much as possible a
climate of trust between economic units. Networks
1579
Fig. 2. Trust, appropriability and incentives.
are, thus, the institutional framework, within which

economic units realise a trade-off between property
rights and rights of access on knowledge.
The above discussion on trust, in line with the analyses of incentives, leads to Fig. 2 that summarises four
types of cases of the use of property rights instruments,
when considering two dimensions: trust and nature of
incentives of economic units. In this table, the first
case (absence of trust, appropriability is the only incentive) is equivalent to the traditional context. 42 The
fourth case (existence of trust, different incentives) is
equivalent to cases studied by Creplet et al., Muller
et al. (this issue).
2.2. The building of competencies
Competencies could be understood as the ways
through which knowledge is elicited, used and applied
to specific contexts and domains by learning agents
engaged in the innovation process. As Malerba and
Orsenigo (2000) noted, the three notions of competencies aims at capturing the ways through which agents
structure their knowledge, manage their interactions
between differentiated fragments of (i) information,
(ii) knowledge codified in different codes, and (iii)
tacit knowledge. Thus, economic units organise themselves intentionally to enhance learning, in order to lay
the foundation for further rounds of progress. In the
traditional context, the notion of competence hardly
matters. The economic agents are supposed to absorb
42 In this figure, we only consider distinct economic units, we do
not consider members of some specific communities (for whom
the incentive mechanisms are driven by the functioning of the
community).
external knowledge at no cost, and do not exhibit

learning capabilities. An exception is the learning by
doing (Arrow, 1962), but this category of learning is
not deliberate and does not contribute to modify the
cognitive capabilities of agents. In a KOP context, the
building of competencies is fully part of the framework within which policy instruments are designed.
The interactive nature of the innovation process implies that the agents involved increase their competencies while engaging in the innovation process.
As repeatedly argued, (cf., for example, Nelson and
Winter, 1982; Dosi and Egidi, 1991), innovative activities involve a kind of learning quite different from
a passive learning by doing mechanism: it requires
agents to build new representations of the environment
they operate in (and which remains largely unknown)
and to develop new skills which enable them both to
explore and to exploit this world of ever-expanding
opportunities. Such representations are embedded in
the routines, which characterise the organisation.
The concept of competence has been first developed within the theory of the firm. The notion
of competence relies on that of routines and rules,
and refers to a view of the firm as a social institution, the main characteristic of which is to know
(well) how to do certain things. These competencies are coherent sets of capabilities to use them in
an efficient way. Some of these competencies are
strategic (core competencies according to Teece
(1988)) and constitute the main sources of the competitiveness of the firm. They are the results of a
selection process both internal and external to the
firm. The management, the construction and the
combination of these competencies are critical in
order to understand the limits of the firm and the
1580
co-ordination as well as the incentive structure of the

firm.
According to these approaches, the firm is conceived as an institution where competencies are continuously built, shaped, maintained, and protected.
This is a cumulative and strategic process that relies intensively on the management of knowledge:
the competence-based approach depends essentially
on the ways, for a given firm, to acquire, produce,
absorb, memorise, share and transfer knowledge. 43
But, and perhaps even more important, this focus on
knowledge issues brings about also the issue of how
such knowledge is generated, maintained, replicated,
and modified (and possibly also lost), i.e. the issue of
learning and its nature.
The concept of competence is inherently related to
the degree of attention, which is (as it has been emphasised in the introduction) the rare resource in a
knowledge-based context. As Amesse and Cohendet
(this issue) emphasised, since the degree of attention
is the key limiting factor, the firm will first focus on
its core competencies (what the firm does well and
better than the others), then as the degree of attention
decreases, the firm manages competencies (what the
firm does well), and finally, at the other end of the
spectrum, the firm manages its other basic activities,
the periphery. By definition the activities that belong to the periphery do not require a strong commitment in terms of knowledge management. The firm
just needs to be informed of the best practices of external firms and organisations that can offer equivalent
services and if it appears that these activities are too
costly to run within the firm (according to transaction
43 This has some important consequences in particular for the
issue of co-ordination by the firm: first, because knowledge of
complex production processes is necessarily distributed (cf. von
Hayek, 1937) and cannot be fully grasped and controlled by a
single individual. A primary role of an organisation becomes that
of co-ordinating this dispersed knowledge. Second, because coordination in this case generally involves the creation of commonly
shared bodies of knowledge: sets of facts, notions, models of the
world, rules, procedures which are at least partly-known to all
the members of the organisation involved in a given interaction.
In a sense this kind of co-ordination is a pre-condition for the
co-ordination of actions which is examined by most current literature and which implicitly assumes that all these mechanisms for
the co-ordination of dispersed knowledge are already in place. It
seems in any case most unlikely that mere incentive mechanisms
could alone be sufficient to promote this kind of co-ordination.
costs criteria), they will be outsourced. The degree of

attention of firms is, thus, shaping the ways they manage knowledge, the ways they build their cumulative
cognitive assets, and also the ways they interact with
other firms. To be more precise:
Within its set of core competencies, the firm functions as a knowledge processor giving full priority to the creation of resources. For a given firm,
in terms of exchange of knowledge, this zone is
characterised by partners or quasi integrated
suppliers, that produce high value components or
systems which are highly strategic. 44 What is essentially transferred in this zone are creative ideas
through multiple functional interfaces (manufacturing to manufacturing, engineering to engineering,
etc.). This requires a permanent capabilities benchmarking within the group of partners, a substantial
investment in inter-firm knowledge shared routines,
and regular activities of socialisation. The activities that belong to the core of the firm are not
considered as being tradable on the market: they
are disconnected from the make or buy trade-off
suggested by the transaction cost theory. 45
44
They could be wholly owned suppliers or partly owned affiliate

suppliers in which the firm holds an equity stake and typically
transfers personnel to work on a part-time or full-time basis. These
suppliers fully or partly participate in long term strategic plans,
capital investments and capacity planning, and personnel transfers.
The formal duration of the typical contract is a long term one, and
most of the contracts are renewed automatically. Moreover, they
tend to participate in the building of the knowledge base of the
firm, and to benefit from the absorptive capacities accumulated by
the firm. But it is also important for a firm to enhance the absorbing
capacities of the suppliers themselves. The firm provides assistance
to their suppliers not only in the areas of quality, costs reduction,
factory layout and inventory management, but also in terms of
increasing technological competencies and research facilities.
45 However, the scope of the set of core competencies is very limited, for managing core competencies is by definition very costly:
it requires specific sunk costs, forging and managing alliances and
other types of co-operation with institutions that have complementary forms of knowledge, accessing and absorbing the most recent scientific results related to the domain of core competencies,
etc. This is the reason why companies generally choose few core
competencies to develop, extend and protect in the long run. For
instance, Sony vies for command in the field of miniaturisation,
Benetton has two core competencies (design and marketing), and
in general it is rare for a company to possess more than five core
competencies.
In the domain of competencies, the firm holds significant pieces of knowledge, but needs to access
complementary forms of knowledge held by other
firms to be able to develop and use the knowledge
efficiently. The firm has, thus, incentives to participate to networks, in order to proceed to the mutual
exchange of complementary forms of knowledge.
As it has been underlined, networks offer precisely
a way to share and exchange knowledge complementarities.
In the peripheral zone, the firm does not hold any
specific advantage in terms of knowledge and the
risks for it in terms of asymmetries of information is very high. This is a zone of quasi market,
where the degree of supplierbuyer interdependence
is generally low. Products are standardised ones, and
require few interactions with other inputs. Contracts
are arm-length contracts, the duration of which depends on the classical transactional parameters. 46
In this zone, firms organise the process of on-going
allocation of resources and adaptation to the environment. The latter requires governance mechanisms that are well analysed by the transaction cost
approach. 47
The building of competencies by firms constitute
one of the major sources of heterogeneity between
firms and their respective competitiveness. It has
strong impacts on the processes of codification of
knowledge within and between firms:
In the core domain, firms will favour those processes allowing both a strong absorptive capacity
and a strong selecting emission, understandable
only by the firm and the small nucleus of partners.
It is a domain, where part of the codification effort
is devoted to code productive and organisational
46 For a given firm, in terms of supplier management practices,
this zone requires a minimal assistance to suppliers, with single
functional interfaces (sales to purchasing for instance), and the
practice of price benchmarking.
47 This hypothesis implies that there are two main governance
mechanisms to be considered by the firm: the mechanism for governing competencies, defined by the need to co-ordinate knowledge, and the mechanism for governing transactions, reliant on
the need to manage transactions (see also Amin and Cohendet,
2000). The organisation of the firm requires a dual structure of
governance, one for governing the domain of the core activities
of the firm, the other one to govern the remaining activities (the
peripheral domain).
1581
secrets. 48 Between integrated partners, strong

knowledge complementarities are present. This is
a context, as suggested by Malerba and Orsenigo
(2000), where organisational integration (and not
specialisation) is taking place because of the need
to have better co-ordination and integration of
different complementary capabilities in systemic
innovation. 49 Measures for reproducing knowledge are also taken to facilitate the integration of
scientific knowledge and practical know-how (such
as experimental learning).
In the domain of networks to regulate competencies,
the building of common languages and standards
between the participants of the network is the key
issue. In the domain of competencies, what matters
for firms is to have their competencies well recognised by other firms in order to enter the networks
they aim to participate in. The use of codification
in this domain will, thus, be strongly driven by signalling motives.
In the periphery, the firm will follow basic standardised contractual procedures. The firm will adapt to
existing meta-models and languages (for example,
the bar code system) that are used on a universal
scale.
3. Levels of policy implementation

What has been shown at the level of the firm is in
fact valid for any economic unit, any region, any state.
Each economic unit needs, in a knowledge based environment, to build its own competencies on a long
run basis. The progressive building of competitive advantages will allow economic units to adapt, survive,
develop, and learn in competitive environments. This
requires time and high fixed costs. Different levels of
48 These are contexts where generally, as Cowan et al., 2000 stated,
(The code-books are not manifest. It is not explicitly consulted,
nor in evidence, and an external observer, therefore, would have
no direct indication of its existence. (To this outside observer, this
group appears to be using a large amount of tacit knowledge in
its normal operations).
49 In fact a fundamental aspect of firms activities consists in the
integration of different pieces of knowledge (Pavitt, 1998) through
co-ordination and integration capabilities due to complementarities,
particularly in the case of systemic innovations (Malerba and
Orsenigo, 2000).
1582
policy implementation can be thought of for the realisation of KOP. Networks, clusters and other communities of firms and/or individuals may be starting-points
for knowledge-oriented policy measures. The aim of
KOP in this respect is the generation, transformation and diffusion of knowledge through learning processes.
3.1. KOP for firms
The building of competencies is the main driver
of the ways knowledge is used as an input by firms.
It will determine the ways they acquire, absorb and
use knowledge. In particular, it will explain the constitution of their absorptive capabilities, of the degree
and nature of interaction with others, and of their signalling strategies. KOP are characterised by a vision
of the firm in which knowledge-based competencies
constitute the main source of competitive advantages.
These advantages are based on learning processes
within the firm and beyond the boundaries of single
firms, namely, in inter-firm relationships. In this respect, RTD policies distinguish between competition
and co-operation whereas KOP consider opportunities
induced by situations of co-opetition.
Traditional RTD policies focusing on manufacturing industries are dominated by thoughts of specialisation. KOP on the contrary pay particular attention
to interfaces and overlapping between manufacturing
and service activities (e.g. special focus on knowledge
intensive services (KIBS). KOP should pay attention
of the fact that each firm, whatever its size or sector, could have the possibility to build and develop
competencies. This raises in particular the problem of
SMEs that do not always have the critical mass of internal knowledge to build a process of accumulation
of competencies, or cannot afford the fixed costs incurred by the building and managing of competencies.
This certainly encourages the setting up of new agents
of knowledge such as KIBS to make SMEs competencies explicit and to assist the further development
of these competencies. However, in this perspective,
public policy should progressively support absorptive
capabilities inside the small units themselves.
Considering policy instruments, grants and subsidies constitute typical tools of traditional RTD policies, according mainly to the picking the winners
principle. Policy instruments of KOP could be initia-
tives, contests enhancing leverage effects on interest

groups, epistemic communities, clusters, etc.
Between firms and other economic units, KOP
should make sure that meta-languages and meta-codes
exist to provide for the circulation of general knowledge within society. The search for the building of
competencies will put a high focus on private codification processes using local jargons and codes within
closed groups of partners. At a global scale, there is
a risk of too high a specialisation in the formation
of knowledge in the domain of competencies chosen
by private firms. The role of governments is, thus,
to ensure both compatibility between the different
specialised bodies of knowledge that lie behind the
building of private competencies, but also to ensure
a sufficient degree of diversity at society level. The
risk is high that private forces will not invest in some
domains of competencies that require public involvement in the long term.
3.2. KOP for regions
For regions one of the main issues is to participate in
the co-development of competencies at a global scale.
Whatever the private forces, the building of competencies at a global scale requires that private efforts and
public ones are complement. The formation of local
districts (Lissoni, this issue) is for instance a good example of the co-development at the regional scale of
local competencies between the private and the public
forces.
The study of Muller and Zenker (this issue) provides
a comparison of different regional patterns in terms
of innovation-related characteristics revealing regional
disparities in the generation and diffusion of knowledge by firms. Geographical proximity increases the
circulation of tacit knowledge, mainly between actors
of the same region, but at the same time impedes its
accessibility for external actors. Additionally, empirical research showed that small and large firms are
characterised by a different sensitivity to their regional
environment. Large firms often perform internal R&D
and maintain relationships with big players in other
regions whereas small firms do not have the (financial and human) means to perform internal research
and development on a large scale. Thus, small firms
are much more dependent than large firms on external knowledge sources and, therefore, more sensitive
to innovation-related knowledge offered by their environment.

The specificity of the regional scale of innovation
is the proximity effect that facilitates the exchange of
tacit knowledge. Proximity is not only considered in
a physical meaning, but also in the social and cultural dimension. Common settings and short distances
make the exchange of tacit knowledge much easier
and may therefore enhance the networks (and thus
the regions) knowledge base. This is based on the
fact that tacit knowledge is much more context-related
(and thus localised) than codified knowledge so that
knowledge flows more easily within common contexts. Codified knowledge, on the other hand, is less
context-dependent and can, therefore, be transferred
over large distances without loss. Regional schemes to
support interactions, common learning and knowledge
flows are an important policy measure in this respect.
Nevertheless, regions should be considered as open
systems, i.e. systems that are prepared to acquire and
to deliver knowledge to other regional systems.
In this respect, regions are of particular interest
since the development of skills and capabilities is related to local human resources and the exchange of
tacit knowledge. The circulation of tacit knowledge
is facilitated by geographical and cultural proximity.
Since firms develop their innovation capabilities in interrelation with their local knowledge environments,
the implementation of KOP should reflect regional diversity and consider region-specific innovation potentials.
In parallel, European institutions such as the
Commission, provide the opportunity to ensure
the co-ordination of different national players at
meta-level. In this perspective, the design of KOP at
meta-level would rather aim at exploiting national
and regional synergies and complementarities than at
increasing homogeneity between the Member States.
Specific innovation patterns and structures on the
sub-national level can be taken into account by regional institutions (innovation governance dimension)
which may contribute to the generation and better
functioning of a regional innovation system.
To sum-up, the focus on both the production of
knowledge and the building of competencies in a
knowledge-based concept leads to a deep reconsideration of the classical policy instruments in the domain
of research, technology and development policies.
1583
As we have seen, traditional policies rarely took into

account the tacit and collective dimensions of the
production and distribution of knowledge, nor the
building of competencies by economic units to acquire, absorb and use knowledge. That is the reason
why knowledge oriented policies should reconsider
classical policy instruments in the light of the above
evoked dimensions of knowledge.
4. Selected conclusions for knowledge oriented

policies 50
This special issue on knowledge codification is far
from being comprehensive, coherent and mature. This
is mainly caused by the fact that this research field is
in an early stage of its life cycle. Therefore, our policy conclusions are higly selective, preliminary and
miss very specific proposals of detailed policy actions.
Nevertheless, we feel that we could develop a conceptional framework which allows us to feed relevant actors, processes and their contexts with a broad and rich
set of empirical evidence. On the basis of this work
we would propose for a future agenda of research following the main lines of more specific and in-depth
research on the following themes:
accessibility of knowledge: driving factors and conditions, processes and their impacts;
communities and their dynamics;
cognitive structures and content of knowledge;
role of knowledge for macro economic growth;
the importance of technological breakthroughs versus comprehensive diffusion of knowledge;
the role of public infrastructures in a broad sense
(e.g. universities, R&D landscape, institutional connectivity, standards, etc.);
policy rationales (e.g. co-ordination failure) and political governance.
The above general discussion on KOP can be illustrated by referring to the different case studies that
are presented in this special issue. Most of the case
studies exhibit specific policy issues that we have proposed to present along the following main perspectives
50 This chapter partly is based on internal contributions of several
members of the TIPIK project (esp. Lissoni, Muller, Zenker, Cowan
and Foray).
1584
that illustrate some of the main directions for a future

agenda of research that we are suggesting.
4.1. Access to knowledge and co-evolution of
emittive and absorptive capacities: Technology
transfer revisited
The case study of Lissoni (this issue) comes to
the following conclusions: the results of the empirical research are relevant for outlining policy measures
directed at sustaining the innovation capabilities of
SMEs, both in the mechanical field and in a few traditional sectors, especially when located in local clusters. In particular, some re-thinking of recent trends in
technology transfer policies may be pushed up to the
top of policy-makers agenda.
The case study when placed within the conceptual framework derived from Steinmueller (2000) and
Cowan et al., 2000, goes against the typical description
of (Italian) clusters of SMEs (especially industrial districts) as homogeneous cultural settings, wherein technological findings are quasi-public goods. Rather than
flowing freely within the cluster boundaries, knowledge about those findings is shown to circulate within
a few smaller epistemic communities, each of them
centred around the machinery producers which the research chose to investigate, but often spanning outside
the cluster geographical boundaries. We then observe
that
Those communities are better seen as made of people, linked together by personal ties of trust and
reputation, rather than of inter-firm arrangements,
although they arise from successful commercial
partnerships and deals, and respect firms appropriation strategies.
The localisation of members of the epistemic communities is affected by the frequency of contacts
required for transmitting information effectively, as
well as by the size of the members companies.
Public labs and universities seem almost totally
absent from those communities.
From these observations, it follows that epistemic
communities can be a better policy target than either
firms or specific geographical units, which require specific policy actions: new firms may arise from community members seizing some technological opportunity,
as it had happened when many SME clusters were
born. Allowing community members to access knowledge from other sectors or from academic research
may help, although policy measures in this direction
may be in contrast with the appropriability measures
and staff management practices of the employees of
the episternic community members. It should be noted
that many technology transfer actions which currently
target existing SMEs as potential innovators, could
be instead directed at giving some members of local
epistemic communities the chance to found their own
start-up.
More generally, within a KOP concept one should
distinguish between traditional technology transfer
and knowledge transfer. The target groups of technology transfer are users and practical communities primarily, the target groups of knowledge transfer mostly
epistemic communities. Technology transfer policies,
which focus on specific sectors and locations, but do
not arise from an agreement with local members of
the existing epistemic communities, are very likely to
end up offering very generic, and possibly irrelevant
services (as many assessment of technology transfer
policies actually show). Since knowledge circulates
within a number of relatively close networks, policy
initiatives have to focus on access to knowledge and
inter-personal networks, the degree of geographical
dispersion of the relevant epistemic community, and
the extent to which knowledge can be considered
as public (i.e. shared by different communities) or
semipublic (i.e. circulating only within one community). Some of the links between SMEs and larger
firms which many technology transfer policies try to
set in motion, are already in place within the existing
epistemic communities.
The study by Amesse, Cohendet (this issue) views
the process of technological transfer as a process that
depends on the ways firms and other institutions deal
with knowledge. On the one hand, they underline the
role of absorptive capacities as essentially active along
the perspective suggested by Cohen and Levinthal
(1990). They show that the more groups, teams and
communities within the firm are receptive to new ideas
the higher are the chances of an efficient absorption
of technologies from outside. On the other hand, the
quality of the process of technology transfer is also
fundamentally dependent on the firms capabilities to
emit knowledge outside its frontier. When firms provide significant assistance to their strategic partners,
through multiplying functional interfaces and investment in knowledge sharing routines for instance, they
in fact deliberately contribute to enhancing the absorptive and emittive capacities of their key suppliers.
The authors also show, when negotiating the rights
of access to complementary forms of knowledge that
they need within networks, they carefully, and permanently assess the absorptive and emitting capacities
of the other members of the network. In other words,
the management of the technology transfer process is
essentially bi-directional. What matters is more the
co-evolution of the mutual absorptive and emitting capacities between partners, than the mere observation
of the technology flow between an emitter and a receiver.
Similar conclusions can be drawn in the case of
university/industry linkages (see Section 4.4) and in
the case of the dense interactions between knowledge
intensive services and their customers (see Muller
and Zenker and Creplet et al., this issue). The study
of DAdderio (this issue) on software development
demonstrates that standardised, coded procedures
and models are of little use unless they are locally
appropriated and effectively transformed into actional
routines and prototypes. Diffusion of standardised
practices, models and methodologies run the risk of
seriously miss-stating the organisational costs and
productivity effects of software adoption processes.
As a consequence the study shows that software
producers need to build greater flexibility and customisation potential into their systems in order to
facilitate the process of adaptation of generic systems
to local, context specific, circumstances and requirements. These emittive and absorptive capabilities lead
to specific requirements, to dynamic learning, translation routines etc. All these cases demonstrate that
policy has to go far beyond R&D and to focus more
on competencies.
4.2. Shifting from R&D to competencies: the specific
problems of small and medium-sized enterprises
The study by Muller and Zenker (this issue)
comes to the following conclusions: compared to
medium-sized manufacturers, small enterprises are
characterised by a lower level of knowledge-intensive
interactions. This means mainly that small firms acquire less innovation-related information from com-
1585
petitors, suppliers and also from research institutes

and thus have more limited access to external knowledge than large firms. This has two consequences:
first, small firms have less knowledge they can use
for innovation projects and second small firms benefit
from fewer opportunities to improve their absorptive
capacities. These aspects and the lack of personnel,
especially marketing personnel, are the major obstacles to innovation in comparison with large manufacturing firms. Since small firms carry out R&D on an
incidental basis while large manufacturing firms are
permanently engaged in development activities, they
have to a lesser extent the opportunity to codify the
knowledge produced in the frame of these activities.
Thus, through the development of routines, it can be
assumed that large firms may codify their knowledge
more easily and that therefore small firms, in comparison, benefit only from a lower level of codified
knowledge. Therefore, the strategic aims for policy
to achieve with regard to the situation of small firms
are: to raise awareness for the significance of knowledge and learning and to compensate knowledge
codification weaknesses.
One important measure to raise awareness
concerning the issues of knowledge codification and
innovation issues in general consists in providing
manufacturing firms with information on innovation
projects and on the importance of knowledge. This
also includes the introduction of routines such as
knowledge monitoring tools in the firm, comprising
organisational, technical, financial and human factors. In order to distinguish their competencies, firms
must define their specific strengths and consequently
the contents of their knowledge bases. This is an
important process which requires in-depth insights
concerning firm activities and equipment with capital
goods and with staff. Firms must then find a way to
organise their knowledge flows and to manage their
knowledge base. This is supported by available tools
such as computer networks or specific software tools,
and requires a certain technical standard. All these
mentioned aspects mainly refer to codified knowledge and rather to technical than to social or organisational knowledge. Knowledge monitoring forces
firms to trace their innovation activities which allows
for a better understanding of innovation processes
and the respective (knowledge) inputs in different
stages of the innovation process. Besides, this kind of
1586
self-monitoring forces firms to define and formulate

their problems in innovation matters which serves
as very important first step to solution-finding and
problem-solving. On the other hand, firms should
monitor their competencies, their skills and knowledge they can refer to. This might include codified
knowledge that is stored in databases, containing information about available literature, brochures, web
sites and further information, descriptions and comments about former experiences in innovation projects
and specific skills of the staff. Especially the later aspect is difficult to achieve, but is crucial in order for a
firm to use its opportunities and strengths optimally.
Another important additional effect is the codification
of knowledge internal to the firm that did not exist
in codified form before: once a monitoring tool has
been introduced and is maintained within the firm,
firms codify at least parts of their internal knowledge
in order to store it and use it for innovation projects.
Policy measures in this respect would include incentives and project support in introducing routines
of knowledge monitoring in small and medium-sized
firms that may lack own (financial) capacities to purchase equipment and to train their employees. Innovation policy should also support knowledge managers
i.e. persons that visit firms and raise firms awareness
for the importance of those knowledge subjects. Additionally, the development of at least partly
standardised knowledge monitoring tools would be
helpful for firms, since the barrier for each firm to
develop their own measurement tools is high. Besides
organisational skills and respective personnel this also
highlights technical skills and experiences that are important for performing research and development activities, for preparing and realising innovative projects
and for the absorptive capacity of firms. Recent studies mention the phenomenon of innovation without
research emphasising firm networks as knowledge
source. That is, individual firms are to a lesser extent
seen as research performers as pre-requisite for internal innovations, but the innovation networks in which
firms act and interact are emphasised. Nevertheless,
a non-researching, but innovating firm has to acquire external knowledge and apply it to its individual
problems, which implies that a firm has a knowledge
appropriating capacity. In this context, the absorptive capacity of firm employees, i.e. the capacity to
know-what, to get familiar with external knowledge
and to apply it internally is of crucial importance.

Since a certain level of skills and knowledge is necessary to acquire new knowledge, the skill levels of
firms employees are referred to. Thus, political measures should increasingly include support for fostering
skill levels in firms, for a qualified human capital in
firms and for human capital mobility, for instance by
exchange programmes with research organisations.
4.3. New agents of knowledge
The new agents of knowledge are not really new.
What is new is that in the meanwhile as economic actors they play a far greater role than in the past. The
new actors are mostly small firms characterised by
skilled knowledge workers. The study of Creplet et al.
(this issue) tries to analyse in more detail the role of
two specific new agents of knowledge from a cognitive point of view. The study by Muller and Zenker
(this issue) is focused on knowledge intensive business
services in general and their characteristics in creating, re-engineering and diffusing knowledge. The core
characteristics of the new agents are a very high degree of interactions with customers, a deep access to
the knowledge structure of these customers and significant capability of knowledge re-engineering. As a
consequence policy has to understand that this target
group shows very different characteristics compared
with the traditional target groups such as R&D performing firms within manufacturing industries.
The study of Creplet et al. demonstrates the differences between consultants and experts with respect to a cognitive dimension: summarised in a
very simplified manner, consultants contribute to
the problem-solving process of their customers by
standardised methods, routines and processes and by
their knowledge of best practices. The development
of their competencies is mostly based on links with
communities of practice. Experts mainly intervene in
complex situation and create and operate a relatively
new knowledge. The development of their learning
process is mostly based on links with epistemic communities. This study demonstrates how different the
new agents of knowledge behave due to their different roles and roots in epistemic communities or
communities of practice. As a consequence simple
policy conclusions cannot be drawn. Nevertheless,
it is obvious that several policy shifts are necessary.
This new target group of KOP has to pay attention far

more to skills, competencies and personnel (instead
of R&D), on soft factors such as management organisation, training (instead of hard factors), on changes
of behaviour (instead of reaching technological advances) and on knowledge management (instead of
R&D projects) than in a traditional policy perspective.
With regard to knowledge intensive business services (KIBS) the case study of Muller and Zenker
comes to the following conclusions: it shows that
KIBS play a role of knowledge processing, reengineering and diffusing for innovation. Fulfilling this
function implies knowledge transformation from its
generation to its application in client firms. KIBS,
thus, take on a go-between function between research
organisations that produce scientific results and firms
that use and apply this new knowledge. Since firms
mostly cannot directly apply new knowledge and
since KIBS know firm-internal processes and demands, they process new knowledge, diffuse it
among their clients and support its application in firm
innovation processes. This knowledge transformation
process consists to a large extent in a modification
and reengineering of codified knowledge made public by research organisations into tacit (or specific)
knowledge that is communicated to firms and applied
by them. Interactions between KIBS and SMEs lead
to a circle based on the exchange of knowledge in
both directions. This fosters innovations in both types
of firms and can be described as mutual activation
of knowledge resources. Co-operating manufacturing
SMEs and KIBS treat certain problems commonly and
they participate in shared learning processes. A very
close interaction, a far-reaching access to the knowledge structure of their customers and re-engineering
of knowledge are the very specific characteristics of
KIBS as new agents of knowledge. In this respect,
KIBS may also compensate for regional weaknesses
in research infrastructure since they approach scientific results and prepare those results for application
in manufacturing firms. As a consequence, the strategic aim will be to support the expansion of the KIBS
industry in Europe and to acknowledge their activities
in terms of innovation boosting, both internally and
in their clients.
This has two implication for innovation policy:
on the one hand, policy should devote more attention to these new agents of knowledge as a new
1587
target group. On the other hand, in order to stimulate co-operations between KIBS and other types of
firms, the visibility of the former firm type should be
raised, especially for small and medium-sized firms
that often lack information concerning co-operation
partners. A helpful means could be a special type of
certification of KIBS in order to label their competencies. This could be useful for KIBS marketing as well
as for manufacturing SMEs to get information about
KIBS offers. Furthermore, innovation policy can increase collaboration by giving incentives for using
KIBS services. The benefit from this kind of support
cannot only be expected in innovation activities of
manufacturing SMEs, but also in KIBS internal innovation that are nurtured by knowledge they gain
from co-operation with their manufacturing partners.
The emittive capacities of KIBS, the absorptive capacities of SMEs and the level of interaction between
both are the main targets of policy.
4.4. Redesign of incentives: dynamics of communities
The promotion of university/industry relationships
is a well-established science and technology policy
area. A long tradition of different institutional settings,
financial incentives and other tools of technology
transfer are used. Although there is a broad consensus
that the linear model of innovation is inadequate, the
concept of the one-way bridge from public research
to industrial research is still widespread in the discussion on technology transfer. The contributions of this
volume support the need to refocus the traditional
policy of technology transfer the establishment of
transfer institutions and incentives within the public
research world and to convert it into a concept
supporting a two-way bridge. A survey at German
universities (Meyer-Krahmer and Schmoch, 1998) has
shown for all fields examined that the central linking
element in the co-operation between universities and
industrial firms is the exchange of knowledge in both
directions. Although the institutional orientation of
academic and industrial researchers is different, the
exchange of knowledge can be considered a common
denominator where both interests meet. The mutual
exchange of knowledge in techno-scientific communities is obviously a broad phenomenon that is not
limited to some exceptional cases, but applies to
whole disciplines and sub-disciplines.
1588
Another major conclusion underlines a characteristic of a national innovation system which one
could call structural absorptive capacity. We call
the absorptive capacity (of firms, research institutes
or universities) structural, because it largely depends on meso-level characteristics. This structural
absorptive capacity depends, according to empirical
evidence, on the science intensity of technologies,
the industrial life cycle, and the firm size structure
of the industrial sectors. Furthermore, the absorptive
and emittive capacities depends on micro-level factors: internal R&D capacity of firms, and interaction
patterns to relevant technologies outside traditional
linkages, as well as formal co-operation and informal
networks.
The emergence of national systems of innovation
has to be considered as a path-dependent evolutionary
process where various economic, technological, social and cultural factors interlock and strengthen each
other mutually. The cognitive structures of universities are strongly influenced by the co-evolution of
industry structure, technology and institutional factors
(see also Nelson, 1994), e.g. the close interaction of
application-oriented university departments and industrial firms in the less science-based area of mechanical
engineering in Germany has to be interpreted in this
theoretical perspective. For many years, it has been a
successful model of co-operation, but in the present
situation, it implies risks of lock-in effects which
can be overcome only by deliberate efforts. There is
broad evidence that in addition to the factors already
mentioned, the specific cognitive structure in this area
supports the relatively strong orientation of universities on application and hampers the openness towards
new, more theory-based technologies. However, this
aspect has not yet been studied in a systematic way. So
we suggest that more research on cognitive structures
of different technologies is undertaken, because it can
contribute to an improved understanding of national
systems of innovation, but also of related problems
such as the organisation of inter-disciplinarity or the
relations between science and technology.
The paradigm of inter-disciplinarity affects these
instruments by a shift from inter-organisational
to intra-organisational linkages. The paradigm of
science-based technologies calls for a renewal
of traditional instruments to improve university/
industry co-operation through new instruments link-
ing academic communities with research communities in industry. Interaction and communication are
more important than financial issues. This leads to
the following set of conclusions.
An important consequence for science technology policy is the necessary change of organisation,
communication, interaction and incentives within
the public research world. The consequences for
science-based technologies are the need for new ways
of linkage between basic research and applied research. Also trans- and inter-disciplinarity need better
horizontal linkages between disciplines. Often in the
literature inter-disciplinarity is a misleading concept
of integration and mixing disciplines. It is essential to
understand that first-class trans- and interdisciplinary
research is highly dependent on first-class disciplinary
quality of the scientists involved in interdisciplinary
research. Therefore, an efficient linkage between (and
not the integration of) disciplines is crucial for the
establishment and dynamics of such epistemic communities. Possible tools and mechanisms of such
linkages are
Organisation of research: problem orientation in the
case of well definable social or industrial-technical
problems. This requires, in contrast to the currently
predominant internal objectives of science, new
ways of project organisation and management.
A better linkage of the long-term applicationoriented basic research with applied research would
meet future requirements. This could be achieved
for example by a renewed institutional network,
co-operation research with specific subjects, new
models of financing, improved communication and
other assessment criteria.
Team research: besides the currently predominant
orientation of academic research towards individualised research setting, interdisciplinary team
research must be strengthened by appropriate incentives.
Improved intra- and inter-sectoral mobility of researchers: on an international level and also between
science and industry.
Increased flexibility of research structures: more
rapid taking-up of new developments by flexibilisation of the present rigid public service rules and
budget laws, deregulation of the academic administration, and networking of research institutions for
a limited time, especially in an international framework virtual research institutes.

4.5. Solving the paradox of digital memory
There is a range of policy issues to be addressed
around the paradox of digital memory. This is a
paradox because on the one hand many public and
private organisations are keen to implement procedures and mechanisms for building organisational
memories (with appropriate incentive structures and
co-ordination devices); while on the other hand there
is a general concern about the memory capacity of
societies facing the challenges of storing and archiving huge amount of digital data and information.
The problems and difficulties related to digital
memory have three sources: firstly, we should note
that with information technologies, what we record
are not documents but sets of instructions (Cowan
and Foray, 1997). Because paper has a low acid content, printed books and journals can last for centuries.
Physically, it deteriorates slowly and because the language in which it is written evolves very slowly, the
interpretation of even very old texts remains relatively
easy. The problem is that with information technologies, what we record is not documents, but sets of
instructions that have to be interpreted and managed
by appropriate equipment and software before the
information they contain can be used.
Secondly, the rapidity of knowledge production
and codification processes and the low and decreasing
costs of storing codified knowledge make the problem
of attention more acute (Simon, 1982). The classical
formulation of that problem is that, more than ever, it
is attention rather than information that is becoming a
rare resource as screening and selection of information
become important functions. However, this second
problem can be formulated in a slightly different way:
given the reasonable assumption that the attention
management problem has always existed in some degrees, if we can make the claim that more knowledge
is being stored and made available in codified form,
the knowledge management problem changes shape.
As we have seen earlier, when the knowledge, we seek
to understand is tacit, know-who does matter. When
it is stored in codified form, know-where becomes
important. In a sense, know-where and know-who
should be combined to develop the real problematic of
1589
software agents: the best agents would not only be

efficient in finding all the information corresponding
to a certain question, but they will take into account
the peculiarities of the user and the situation.
Thirdly, knowledge is divided and dispersed
(Machlup, 1983). The division of knowledge is a
result of the division of labour in the field of knowledge production. It is probably not disputable that the
division of knowledge is increasing over time (specialisation), raising the marginal cost of knowledge
integration. The dispersion trend is less clear, but, as a
law, one can expect a higher dispersion as knowledge
production becomes more collectively distributed (located in many places). How to build storage processes
that are integrative; that is to say which are not just
reproducing the state of division and dispersion of
knowledge as it is at the moment of its creation?
Facing those challenges problems posed by
recording the set of instructions, by information abundance and by the division and dispersion of knowledge
an important policy objective should be to reach
a higher level of organisational procedures and capabilities in a way that could allow organisations and
agents to overcome this digital memory challenge.
4.6. Standardisation and codification
The case study on ISO standards implementation
of Benezech et al. (this issue) emphasises the importance of the way economic agents conceive the role
played by industrial standards in the organisation. If
standards are considered as constraints by the firm,
conformity process will follow the strict application
of the normative document. Then, quality code-book
is viewed as a technical requirement imposed by the
institutional environment of the firm. 51
A consequence of such a centralised vision of ISO
code-book leads to a higher inertia in terms of organisational change. Resistance to change was, indeed,
particularly stronger in SMEs sharing the vision of
the ISO code-book as an external constraint. Thus, an
important part of the failure in ISO certification pro51 France for instance is familiar with this normative culture
where standards are emanation of the central governmental authority. Countries such as Germany and UK exhibit more decentralized
forms of quality norms with a higher participation of industry in
the constitution of quality referential.
1590
cess comes from this attitude. In this perspective, the

following could be the relevant policy measures:
Extending the process of appropriation by economic actors through specific governmental actions.
It seems important to inform enterprises of the
true goal of organisational standards such as ISO
devoted to the improvement of the system of management in accordance to their internal objectives.
Opening the participation of economic actors such
as small firms in the work of norms determination
in order to facilitate the diffusion.
Creating a normative pre-competitive co-operation
at the European level. On the same order of magnitude than the big R&D pre-competitive European
programmes from the eighties, a more specific
action concerning now a normative framework
could be imagined. In a competitive environment
characterised by globalisation and a very high
rate of technical progress, an ex ante normalisation (before products, technologies and property
rights) could be a powerful tool in Europe for
collective codification of knowledge and construction of common references in order to stimulate
inter-enterprises relationships.
References
Adler, E., Haas, P., 1992. Conclusion: epistemic communities,
world order, and the creation of a reflexive research programme.
International Organization 46 (1) 367390.
Amin, A., Cohendet, P., 2000. Organisational learning
and governance through embedded practices. Journal of
Management and Governance 4, 93116.
Arrow, K.J., 1962. Economic Welfare and the Allocation of
Resources for Invention. The Rate and Direction of Inventive
Activity. Princeton University Press, Princeton, pp. 609625.
Balconi, 2000. Tacitness, Codification of Technological Knowledge
and the Organization of Industry. Santa Anna Working Paper.
Barber, B., 1952. Science and the Social Order. Free Press, New
York.
Baumard, P., 1999. Tacit Knowledge in Organizations. Sage,
London.
Bessy, C., Brousseau, B., 1998. Licensing of Technology: various
contracts for diverse transactions. International Review of Law
and Economics 18, 451489.
Beyssade, C., 1998. Sens et Savoirs: des Communautes
Epistemiques dans le Discours. Presses Universitaires de
Rennes, 1998.
Brown, J.S., Duguid, P., 1991. Organizational learning and
communities of practice. Toward a unified view of making,
learning and innovation. Organisation Science 2, 4057.
Brown, J.S., Duguid, P., 1998. Organizing knowledge. California

Management Review 40 (3), 90111.
Callon, M., 1994. Is science a public good? Science Technology
and Human Values 19 (4), 395425.
Callon, M., Rabeharisoa, V., 2001. The involvement of patients
in research activities supported by French Muscular Dystrophy
Association. In: Jasanoff, S. (Ed.), States of Knowledge:
Science, Power and Political Culture, Chicago Press.
Callon, M., 1999. Le reseau comme forme e mergente et comme
modalite de coordination. In: Callon, M., Cohendet, P., Curien,
N., Dalle, J.M., Eymard-Duvernay, F., Foray, D., Schenk (Eds.),
Reseau et Coordination. Economica, Paris.
Callon, M., Cohendet, P., Curien, N., Dalle, J.M.,
Eymard-Duvernay, F., Foray, D., Schenk, B., 1999. Reseau et
Co-ordination, Economica, Paris.
Cohen, W.M., Levinthal, D.A., 1990. Absorptive capacity: a new
perspective on learning and innovation. Administrative Science
Quarterly 35, 128152.
Cohen, M.D., Burkhart, R., Dosi, G., Egidi, M., Marengo, L.,
Warglien, M., Winter, S.G., 1996. Routines and other recurring
action patterns of organisations: contemporary research issues.
Industrial and Corporate Change 5 (3), 653698.
Cohendet, P., Creplet, F., Dupouet, O., 2001. Organisational
innovation, communities of practice and epistemic communities:
the case of Linux. In: Kirman, A., Zimmermann, J.B. (Eds.),
Economics with Heterogeneus Interacting Agents. Springer
Verlag, Berlin.
Cohendet, P., Steinmueller, W.E., 2000. The codification of
knowledge: a conceptual and empirical exploration. Industrial
and Corporate Change 9 (2), 195209.
Cowan, R., Foray, D., 1997. The economics of codification and
the diffusion of knowledge. Industrial and Corporate Change
6, 595622.
Cowan, R., David, P.A., Foray, D., 2000. The explicit economics of
knowledge codification and tacitness. Industrial and Corporate
Change 9 (2), 212253.
Dasgupta, P., David, P., 1994. Towards a new economics of science.
Research Policy 23, 487521.
David, P., 1993. Knowledge, property and the system dynamics
of technological change. In: Proceedings of the World Bank
Annual Conference on Development Economics. 1992.
David, P.A., 1994. Why are institutions the carriers of history?
Path dependence and the evolution of conventions, organizations
and institutions. Economic Dynamics and Structural Change
5 (2), 205220.
David, P.A., Foray, D., 1995. Accessing and expanding the science
and technology knowledge base. In: Special Issue on Innovation
and Standards, Science Technology Industry (STI), Review No.
16. Organisation for Economic Co-operation and Development
(OECD), Paris, pp. 1368.
David, P.A., Mowery, D., Steinmueller, W.E., 1994. University
Industry Research Collaborations: Managing Missions in
Conflict, Conference CEPRT/AAAS, University Goals,
Institutional Mechanisms and the Industrial Transferability of
Research, University of Stanford, Stanford, 1820 March.
Diamond, A., 1996. The economics of science. Knowledge and
Policy 9 (2/3), 649.

Dibbiaggio, L., 1998. Apprentissage, Coordination et Organisation
de lIndustrie. Ph.D. Universite de Nice-Sophia Antipolis.
Dosi, G., 1988. Sources, procedures, and microeconomic effects
of innovation. Journal of Economic Literature 16, 1120
1171.
Dosi, G., Egidi, M., 1991. Substantive and procedural Uncertainty.
Journal of Evolutionary Economics 1 (1), 145168.
Gunby, P., 1996. Explaining adoption patterns of process standards,
Ph.D. dissertation. Department of Economics, The University
of Western Ontario.
Haas, P., 1992. Introduction: epistemic communities and
international policy co-ordination. International Organization
46 (1), 137.
Hatchuel, A., Weil, B., 1995. Experts in Organizations. Walter de
Gruyter, Berlin.
Hayek, F.A., 1937. Economics and knowledge. Economica 4, 33
54.
Hayek, F.A., 1945. The pretence of knowledge. The American
Economic Review 76 (6).
Hicks, D., 1995. Published papers, tacit competencies and
corporate management of the public/private character of
knowledge, Industrial and Corporate Change 4 (2), 401
424.
Joly, P.B., Mangematin, V., 1996. Profile of laboratories, industrial
partnerships and organization of R&D: the dynamics of relations
with industry in a large research organization. Research Policy
25 (4), 901922.
Joly, P.B., 1997. Chercheurs et Laboratoires dans la Nouvelle
Economie
de la Science, Revue dEconomie Industrielle, No.
79. Leconomie Industrielle de la Science, pp. 7794.
Knorr-Certina, K., 1981. The Manufacture of Knowledge, An
Essay on the Constructivist and Contextual Nature of Science.
Pergamon Press, Oxford.
Lave, I., Wenger, E., 1990. Situated Learning: Legitimate
Peripheral Participation. Cambridge University Press, New
York.
Machlup, F., 1983. Knowledge, Its Creation, Distribution and
Economic Significance. Princeton University Press, Princeton,
NJ.
Malerba, F., Orsenigo, L., 2000. Knowledge, innovative activities
and industry evolution. Industrial and Corporate Change 9 (2),
289314.
Meyer-Krahmer, F., 1997. Public research/industry linkages
revisited. In: Barre, R., Gibbons, M., Maddox, J., Martin, B.,
Papon, P. (Eds), Science in Tomorrows Europe. Economica
International, Paris, pp. 155173.
Meyer-Krahmer, F., Schmoch, U., 1998. Science-based
technologies: universityindustry interactions in four fields.
Research Policy 27, 835851.
1591
Nelson, R.R., 1959. The simple economics of basic scientific

research. Journal of Political Economy 67, 323348.
Nelson, R.R., 1994. The co-evolution of technology, industrial
structure and supporting institutions. Industrial and Corporate
Change 3 (1), 4764.
Nelson, R.R., Winter, S.G., 1982. An Evolutionary Theory of
Economic Change, Harvard University Press, Cambridge.
Nightingale, P., 2000. Economies of scale in experimentation:
knowledge and technology in pharmaceutical R&D. Industrial
and Corporate Change 9 (2), 315359.
Nonaka, I., Takeuchi, H., 1995. The Knowledge-Creating
Company. Oxford University Press, New York, Oxford.
OECD, 2000. Knowledge Management in the Learning Economy:
Education and Skills, Paris. ISBN 92-64-1782-7.
Pavitt, K., 1984. Sectoral patterns of technological change: towards
a taxonomy and a theory. Research Policy 13, 343374.
Pavitt, K., 1998. Technologies products and organisation in
the innovating firm: what Adam Smith tells us and Joseph
Schumpeter does not . Industrial and Corporate Change 7 (3),
433452.
Polanyi, M., 1962. Personal Knowledge: Towards a Post-Critical
Philosophy. University of Chicago Press, Chicago.
Romer, P., 1993. Implementing a national technology strategy
with self-organizing industry boards. Brooking Papers
Microeconomics 2, 345399.
Saviotti, P.P., 1998. On the dynamics of appropriability, of tacit
and of codified knowledge. Research Policy 26, 843856.
Schumpeter, J.A., 1942. Capitalism, Socialism and Democracy.
Harper and Row, New York.
Scotchmer, S., 1991. Standing on the shoulder of giants: cumulative
research and the patent law. Journal of Economic Perspectives
5 (1), 2941.
Simon, H., 1982. Models of Bounded Rationality. MIT Press.
Cambridge MA.
Steinmueller, W.E., 2000. Does information and communication
technology facilitate codification of knowledge? Industrial and
Corporate Change 9 (2), 361376.
Stephan, P., Levin, C., 1997. In: Proceedings of the Importance
of Implicit Contacts in Collaborative Research and Scientific
Research, Conference on New Economics of Science. Notre
Dame, Indiana.
Teece, D.J., 1988. Technological change and the nature of the
firm. In: Dosi, et al. (Eds.), Technical Change and Economic
Theory. Pinter, London, pp. 256281.
Wenger, E., 1998. Communities of Practice; Learning as a Social
System, Systems Thinker.
Zuscovitch, E., 1998. Networks, specialization and trust. In:
Cohendet, et al. (Eds.), The Economics of Networks. Springer,
Berlin.

Cohen Det Meyer K Rahm Er 2001

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cohen Det Meyer K Rahm Er 2001

Uploaded by

Copyright:

Available Formats

Research Policy 30 (2001) 15631591

The theoretical and policy implications of knowledge codification

The process of codifying knowledge entails three

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

of the world (that results from the accumulation of

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

tems analysed by Hatchuel and Weil (1995), different

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

in codified form, the more knowledge management

the first one is the epistemological dimension which is the

communities appears as essential to understand the

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

community is the existence of a procedural authority

nity tends to send no messages toward the outer

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

ones interact will considerably influence the process of formation of knowledge. 19

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

the process and the community which is the target of

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

acquire more knowledge than previously for a given

tionships when co-operative schemes seem more

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

showed that the contribution of ICT has been relatively modest.

explaining and understanding it would require entirely

2. From theory to policy

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

ceeding rounds of progress. 28 Thus, the process of

28 Scotchmer (1991), when emphasising the cumulative nature of

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

sider policies instruments in the following two main

without being diminished in quality, so that it

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

fer of technology. It explained why public efforts in

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

different kinds of economic forms of knowledge,

ing new knowledge, as is the case of an epistemic

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

the ways to design and implement them, the ways to

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

of different social organisations. Institutional settings

priori, would exhibit a cognitive equivalence that

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

to groups and why has this focus persisted despite

co-operation in the production of knowledge. As

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

Fig. 2. Trust, appropriability and incentives.

are, thus, the institutional framework, within which

external knowledge at no cost, and do not exhibit

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

co-ordination as well as the incentive structure of the

costs criteria), they will be outsourced. The degree of

They could be wholly owned suppliers or partly owned affiliate

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

secrets. 48 Between integrated partners, strong

3. Levels of policy implementation

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

tives, contests enhancing leverage effects on interest

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

to innovation-related knowledge offered by their environment.

As we have seen, traditional policies rarely took into

4. Selected conclusions for knowledge oriented

P. Cohendet, F. Meyer-Krahmer / Research Policy 30 (2001) 15631591

that illustrate some of the main directions for a future