Why Catalonia cannot be considered

as a Regional Innovation System


Scientometrics, forthcoming



Marta Riba Vilanova* and Loet Leydesdorff **


* European Patent Office, P.O. Box 5818, 2280 HV Rijswijk, The Netherlands.



** Science and Technology Dynamics, University of Amsterdam

Dep. of Communication Studies, Oude Hoogstraat, 1012 CE Amsterdam, The Netherlands

E-mail: ;





We present a model to assess the systemness of an innovation system. Patent and citation data with an institutional address in Catalonia (1986-1996) were analyzed in terms of relational linkages and the development in these distributions over time was evaluated using methods from systems dynamics.  Relational linkages are extremely scarce.  A transition at the system’s level could be indicated around 1990 when using institutional addresses, but not when using cognitive categories. The institutional restructuring has led to changes in the pattern of linkages (coauthorship, etc.), but the reproduction of the system’s knowledge base has remained differentiated. We conclude that although a system in several other respects, Catalonia cannot (yet) be considered as a (knowledge-based) innovation system.  The existence of a mechanism for the integration could not be indicated at the regional level.



Keywords: Regional Innovation Systems, Systemness, Path-dependency, Catalonia




In 1988, Baden-Wurttemberg, Catalonia, Lombardy, and Rhône-Alpes concluded the so-called “four motors agreement.”  These four regions aspired to take the lead in organizing and innovating the European Union at the regional level (Cooke, 1998; Lash and Urry, 1994).  Does the process of European unification enable the region to function as the relevant level for organizing innovations, technological developments in SMEs, and wealth and job creation? 


The Maastricht Treaty (1991) assigned an advisory role to the European Committee of Regions (Council of the European Communities, 1992).  This role was further strengthened by the Treaty of Amsterdam in 1997, which envisaged direct consultations between this Committee and the European Parliament. Among these regions, Catalonia holds a special place because this region is an example of what has been called "re-emergent nations without state", in which a significant nationalist movement has turned cultural demands into political claims (Guibernau, 1999). Catalonia has its own language and culture, and it enjoys considerable autonomy within the Spanish state (Castells, 1997).


The Generalitat, that is, the government of Catalonia, has developed its own S&T policies during the 1980s.  The Spanish constitution (of 1978) stated explicitly that the promotion and coordination of scientific and technological research belonged to the competence of the central government.[1] Yet, the Catalan Statute of 1979 claimed that the Generalitat in Barcelona would have exclusive competencies on research.[2]  In practice, the balance has been precarious. The admission of Spain to the European Union in 1986 has made it possible for Catalonia to strengthen its autonomy further with respect to S&T policies.  Science, technology, and innovation have also been core policy objectives at the European level since the Single European Act of 1986. 


Catalonia has thus become a region of confluence of regional, national, and European science, technology, and innovation policies.  In this study, we focus on the question of whether a regional innovation system (RIS) has indeed emerged in Catalonia since 1986.  For this purpose, we have used all patent, publication, and citation data for the decade 1986-1995 with at least one Catalan address, and we analysed this data in terms of co-authorship relations, linkages across sectors (Nauwelaers and Reid 1995), and by using probabilistic entropy measures for emerging systemness and path-dependencies (Leydesdorff & Oomes 1999).  Before turning to this data, however, we need first to specify further the theoretical expectation of a regional system of innovation in relation to these measurement instruments.



Regional versus national innovation systems


Lundvall (1988) based his original concept of “national systems of innovations” on user-producer relations.  Localized user-producer relations entail interaction mechanisms different from abstract market forces, such as mutual confidence rooted in geographical and cultural proximity. The concept of Regional Innovation Systems (RIS) builds upon several turns in economic geography and economic theory (Cooke et al., 1997; Cooke et al., 1998).  ‘New regional science’ has stressed “the importance of geographical proximity or agglomeration characteristics, for facilitating innovative tacit-knowledge exchange and other externalities; and [...] an institutional and organizational learning propensity to regional economic performance” (Ibid. at p. 1563). Evolutionary economic theory has pointed to the major role played by innovation as a source of economic growth (e.g., Nelson & Winter, 1982).


In our opinion, an innovation system should not be considered as a physically existing system, but as the specific dynamics of change in systems of production and distribution. While the latter can be defined in institutional terms, innovation systems can be considered as complex social dynamics and as such cannot be apprehended in their entirety from a single perspective. Innovation systems have been studied more often as a collection of institutions and policies than as the socio-economic dynamics of generation and appropriation of technology (Nelson, 1993; Borras Alomar, 1995). In particular, the systemness of such dynamics has not been empirically addressed. As Cooke points out:


In the literature on innovation the term "system" is not analyzed in great detail. … Clearly, an innovation system is a social system, and innovations are the result of social interaction between economic actors. (1998, at p.11)


In this study, we approach the Catalan innovation system using data collected from patent publications and scientific articles. Indicators of technological activity can be constructed from patent data (Pavitt, 1985; Grupp & Schmoch, 1999). Scientific activity can be traced by using scientific publications (Carpenter & Narin, 1981; Leydesdorff, 1995). Both patent and science publications are the result of knowledge-intensive activities and register mainly knowledge-based innovation. Organizational innovations, innovation in the services sector, or in marketing or in labour relations, to name a few examples, remain out of focus. On the other hand, not all knowledge produced by scientists and engineers is codified in science communications or patents. In the case of patents specifically, some skills may prove difficult to acquire, and organizational and financial thresholds may be difficult for small organizations and universities to cross (Webster & Packer, 1996).


The patent and science journal data inform us on the changes in the knowledge base underlying the systems of production and distribution. Of the five types of innovation in production described by Schumpeter (1934, at pp. 100-101), namely, the production of a new good or a new quality of a known good, the conception of a new production process, opening a new market, accessing new sources of raw or basic materials, or the re-organization of a market structure, the first four draw heavily on the production of knowledge.  The industrial revolution relied heavily on the systematization of change by creating, for instance, research laboratories in industry, scientific management, and the protection of inventions with modern patent (law) systems. However, in the last decade terms such as ‘information age’ and ‘knowledge economy’ are being used increasingly by academics and practitioners alike in the field of political economy. These concepts try to capture the changes in society that have given knowledge a heavier weight in the creation and redistribution of economic growth, employment, and welfare (DTI, 1998; OCDE, 1999; Beck, 1999; Lash & Urry, 1994).


In another context, we have developed the model of a Triple Helix of University-Industry-Government relations (Etzkowitz & Leydesdorff, 1998; Leydesdorff & Etzkowitz, 1998).  Crucial to the argument about “innovation systems” is the development of an overlay of expectations that drive the processes of institutional change. Expectations can be interfaced and communicated across the boundaries of different function systems (Luhmann, 1984).  For example, the expectation of market profit can be used to legitimate an investment decision in the knowledge infrastructure.


In the case of a reinforcement of expectations across different sectors, the (partial) consensus may drive the institutional reorganization of underlying structures. From this perspective, the institutional layer serves as a retention mechanism, among other things, for the creation of wealth and jobs.  The institutional (e.g., sectoral) arrangements, in turn, guide the further formulation of expectations and heuristics as a feedback.


Historically, national systems of production and distribution have thus functioned as competing units in organizing systems of innovation.  The operational mechanisms, however, are price expectations (that is, markets) and cultural expectations. At the level of expectations one is able to translate and to trade-off expected costs and benefits. The exchange of expectations can be organized within subsystems of communication recursively or across (sub-)systems using different codes interactively.


Within each subsystem, the recursive axis is expected to prevail, while across systems the interactive terms (“weak ties”) generate another non-linear dynamics (Blauwhof, 1995). In principle, the relative size of the recursivity and the interactivity of the communications determine the degree of radicality of innovations. Recursive improvements, for example, can be expected to be integrated within existing practices. Radical innovations will involve cross-boundary spanning mechanisms. 


An innovation system may be decomposable in terms of relatively independent subsystems (mainly recursive operation of functions, little interaction) or it may develop more as a single (integrated) organization.  In a later section, we will test our data for distinguishing between these two extremes.  In general, the innovative potential of a system can be defined in terms of the distribution of communicative competences that it contains for translating and thereby adjusting the expectations within and among different domains. This exchange of expectations has additionally to be matched with institutional arrangements that have been shaped historically.


The interfacing of two layers that is, of functional communications and institutional arrangements can be mediated by culturally different (e.g., national) systems of innovation.  The cultural and historical differences thus affect the role of academia and the configuration of trans-sectoral (“Mode 2”) research in national, supranational (e.g., Scandinavian) or regional systems of innovation (Hirasawa et al., 1998; cf. Shinn, 1998).  As noted, systems of innovation can then be considered as competing at the next-order (that is, global) level in terms of the adaptability of their knowledge infrastructures. 


How are communicative competences distributed in each of these local solutions? While the so-called “productivity growth puzzle” is continuously generated by uneven technological developments across sectors (Nelson & Winter, 1975; Nelson & Winter, 1982; Nelson, 1994), the infrastructure conditions the processes of innovation which are possible within and among the sectors.  In particular, the distribution of relevant actors contains an heuristic potential which can be made reflexive by a strategic analysis of specific strengths and weaknesses (Pavitt, 1984; Patel & Pavitt, 1994).



Catalonia, Spain, and Europe


The historical configuration of post-Franco Spain made it necessary to allow for regional autonomy. While the discussion in the Basque country has primarily focused on the issue of independence, the strength of the Catalan region has traditionally been its cultural identity within the Spanish state, with a strong claim for political and economic autonomy.  As a relatively well developed industrial region, whose industrialisation started in the nineteenth century, Catalonia was able to express its cultural identity notably in terms of industrial leadership, orientation towards Europe, and innovation (see Guibernau 1999, at p. 41).  Thereby, it could position itself conveniently between the central government in Madrid and the European Union, which since 1986 focused increasingly on research, technology, and development (RTD) as instruments for further integration across national borders.


Within the region, this integration cannot be fully achieved given the central position of the national governments in the European unification process, the relative weak position of the Catalan language in Europe, and the wish to open up the economy given the remnants of the corporatist past.  In this configuration, there is only a limited room for further developing a regional identity. Thus, contrary to the strong (and symbolic) integration in the case of national systems, the regional system can be expected to remain incomplete and subsymbolic.  Analytically, it has the status of a hypothesis, however seriously the actors involved may be interested in pursuing its realization.


The subsymbolic character of this identity can further be strengthened in terms of networks of relations.  Have specific densities of relations been maintained at this level of integration?  Among other things, an orientation towards Europe (where Spanish and Catalan are both secondary to English and French) may have been helpful in strengthening the network character of this cultural identity vis-à-vis Spain.  However, in the case of conflict, the national identities can be expected to prevail.


For example, when the Spanish government passed its (first) Science Bill in 1986, it reserved a place for the Autonomous Communities only in an advisory body of the coordinating interministerial commission CICYT.  Neither economic resources nor institutions were transferred from the Spanish to the regional administrations.  Only the centre of agricultural research (Institut de Recerca i Tecnologia Agraries, IRTA) in Catalonia was left to the region. All other centres of research remained under the control of the Spanish national government.  In 1987, the Catalan government (and parliament) filed a case of “unconstitutionality” against this law.  The Constitutional Court, however, ruled in favour of the Spanish government in 1992. 


In the meantime, the Catalan government had further developed its own S&T policies (CIRIT, 1997). During the period 1986-1991 both the Catalan and Spanish economies flourished. As a developed industrial region, Catalonia had been hit severely by downward cycles of the economy. The period of 1982-1986 was a period of reform policies by social-democratic governments in Madrid that intensified the democratic reforms of the post-Franco period.  The economy gained momentum only after 1986. As noted, Spain also joined the European Union in this year. During these years, foreign capital invested heavily in Catalonia, while Catalan industries increased their exports to other EU countries. Nevertheless, the rest of Spain remained its major economic exchange partner. The expansive cycle ended in 1992 and a crisis started which lasted until 1996 (Gasoliba, 1996).


Even during expansive cycles, Catalan industry has been characterized by few medium and big enterprises in the field of advanced technology, low R & D expenditures, and a shortage of staff working in knowledge-intensive areas (Petitbo & Bosch, 1990; Barcelo, 1993). The Catalan economy is structurally dominated by SMEs. Since SMEs are believed to create new jobs in Europe, Catalonia strongly qualified for European support. In reality, the share of Catalonia in the II and III Framework Programs, which ran respectively in the years 1988-1991 and 1990-1994, was of 15 and 18 percent of the Spanish total. Thus, it remained approximately proportional to the demographic and GNP weight of Catalonia within Spain. In contrast, the region of Madrid (with a share of above 50 percent) and the Basque country (with a share of almost 10 percent) had a much higher relative return than Catalonia in terms of EU funding of science, technology and industrial competitiveness (Vence 1998). Nevertheless, EU funds outweighed the Spanish and Catalan funds for R&D activities, as can be seen in Table 1 (IEC, 1997). The levels of investment in R&D activities (e.g., GERD/GDP) and levels of staff employed in the knowledge production sector are lower than the averages of the EU countries and similar to the Spanish averages. Historically, the Catalan industry has relied on importing technology rather than invested in endogenous technological development (Bacaria & Borras Alomar, 1998).


In his categorization of Regional Innovation Systems, Cooke (1998, at p. 22) classified Catalonia as an interactive and grassroots innovation system.  The characteristics of grassroots governance are, according to this author, diffuse funding, near-market research, initiation at a local level, and little supra-local coordination.  The features of an interactive RIS are a balance between large and small firms, indigenous capital or FDI, mixed public and private research laboratories, and a high degree of associationalism. These characterizations accord with our assumption about the distributed nature of the (network) data in the case of the Catalan innovation system (cf. de la Mothe & Paquet, 1998).


Is the historically emerging network increasingly developing into a coordinated system?  We shall test whether the distributions have changed, and if so, in which respects.  How strong is the regional organization and what role can it play vis-à-vis the national one?  What difference can the European dimension make for changing these relationships?







We used bibliometric information for studying the Catalan innovation system.  To this end, we searched the on-line patent databases (EUREG, EPODOC) of the European Patent Office (EPO) for patents with at least one address in the Catalan region (using the corresponding postal codes), and the CD-ROM version of the Science Citation Index using the same keys.[3]  All searches were done in January 1998. 


1318 European patent applications with a Catalan address were retrieved for the period 1986-1995,[4] and 15,968 scientific articles for the period 1986-1996.  Note that we used only scientific articles as a document type in the latter set. This was done partly for pragmatic reasons (that is, limiting the size of the dataset), but also because scientific articles can be considered as the prime indicator of developments at the research front while they are at the same time highly codified.  Patents are systematically codified during the certification process, both in terms of their contents and in terms of references to previous patents and scientific journal literature.  A summary of the data can be found in Tables 2 and 3.


Other statistical information (for example, about funding, Spanish patent data, and European comparisons) was used in the background of the study and also for purposes of normalization.





The data was organized in relational databases. This enables us to study both the multivariate complexity in each year and to pursue time-series analysis over the years. While multivariate analysis of, for example, citation relations provides us with network information for each of the years (or for the total set), comparisons of results for different years enables us to proceed only towards so-called comparative static analysis.  However, trend lines for the various indicators cannot systematically inform us about (multivariate) interaction effects. 


Entropy statistics enables us to relate the multi-variate and longitudinal analyses systematically: the expected information content of the message that a distribution has occurred can be considered as an update of the expectation contained in a previous distribution (Leydesdorff, 1991). The measures can be elaborated into tests for the emergence of systemness in the distributions and for irreversible (“path-dependent”) transitions. These complex dynamics cannot always be observed by visual inspection of a set of trend lines (Leydesdorff, 1995). 


As a measure of systemness, we used the expectation that a system is able to counteract a disturbance by maintaining the prevailing uncertainty in the distribution.  If the distribution is not changed, this is also called the Markov property.  The assumption of the Markov property can be tested against the assumption that the units (or subsystems) carrying the links of the network develop along their own historical trendlines.  The two expectations will be tested against each other using the next year’s data as a yardstick. The respective deviations from the expectations can be expressed in bits of information after proper normalization. (See Leydesdorff & Oomes (1999, pp. 79 ff.) for the derivation of the algorithm.)


Another measure that we will use is that of a path-dependent transition.  While the test for systemness focuses on the relative systemness in contrast to a (loose) aggregate of elements, this test evaluates the development of systemness in the datasets.  Is the system active in reorganizing the data according to a logic of its own? (Frenken & Leydesdorff, 2000). We use 'path dependency' here as a descriptive category for the historical event (cf. Foray, 1998) while in the literature 'path-dependency' has sometimes been defined mathematically in relation to the development of a non-ergodic system (e.g. Arthur, 1989).


As noted, the current state of a system provides us with a best prediction of its next state (given our above definition of a system).  In the case of an active system, one expects the historical trajectories of the elements and subsystems to be “overwritten” by the system’s (re‑)organization in the present.  Thus, previous states are no longer expected to matter for the prediction of the system’s next state.  The current state of an operating system boosts the signal from a previous state for the future. Thus, an active system is (by definition) path-dependent upon its current state.


Using the triangle in Figure 1, one can write the following inequality:[5]


I(C:B) + I(B:A) < I(C:A)



From this perspective, the informational distance along the two upper sides of the triangle is smaller than the informational distance between state (t+1) (at C) and state (t-1) (at A).  In this case, the signal is expected to travel first through the in-between state (B) and, rather like an auxiliary transmitter, the previous state is then boosted by the revision of the prediction given this current state of the system.


Note that this test is contrary to the notion of geometrical (that is, Euclidean) distances. (The sum of the two sides of a triangle is always larger than the hypotenuse.)  Information theory provides us with a calculus which enables us to compute algorithmically (as in a movie) what cannot be evaluated from comparing multidimensional (but static) geometries.  In addition to the question of how the units add up to a distribution, one is now able to ask questions about what the interactions may mean for the further development of the system.  We will apply these measures to the different distributions that we will, however, first analyze descriptively.



Descriptive statistics


Percentage share


In the period under study, Catalonia has spectacularly increased its share of international publications and patent applications at the European level.  While Spain has almost doubled its relative share of publications in the period 1986-1996 (from 1.14 to 2.23 percentage points of world share of publications), Catalonia has more than tripled its contribution (from 0.14 to 0.47 percentage points) during this same period. The proportion of Catalan articles in the total of Spanish articles has increased in the decade under study from 12 to 21 percent. (Figure 2; cf. Leydesdorff, 2000).


The trends in patent applications are exhibited in Figure 3.  In this case, both Spain and Catalonia as one of its regions exhibit spectacular growth rates.  While the total number of patent filings per year flattened in Europe during the first half of the 1990s, Spain and Catalonia were still growing in terms of absolute numbers and therefore in terms of percentage shares during this period.  The Catalan contribution to the Spanish portfolio has remained of the order of 60%, but one is able to observe a tendency of relative decline in the first half of the1990s.


The decomposition of the Catalan sets


The 15,968 articles of the period 1986-1996 containing at least one Catalan address appeared in 2122 different journals.  Since the CD-ROM version of the Science Citation Index has a coverage of approximately 3600 journals, this means that the relatively small contribution of articles with a Catalan address appears in this journal literature in a very dispersed distribution. 1564 of this set of journals (that is, 43.4% of 3600 journals) contained more than a single publication with a Catalan address.


The 1318 patents with at least one Catalan address during the period 1986-1995 are grouped in 31 categories, as defined by the European Patent Office in its annual reports. These 31 groups are based on the technical units of the International Patent Classification (IPC). We have further grouped these 31 groups in 10 groups –using the specific expertise in this field of one of the authors– in order to enable us to summarize and compare the information relating to technical fields in which Catalan applicants operate (Table 4).  However, we performed all tests on both tables.


The distribution of Catalan patents across the technological landscape is uneven. The main sectors of activity have been organic chemistry, human necessities, and transport technologies. The percentage shares of these three fields in the Catalan portfolio are above those of the corresponding shares for the world average. Organic, notably pharmaceutical, chemistry is heavily over-represented (Table 5).


The percentage share of patents in textile technology has continuously declined during the period under study, while the sectors of computing, electronics, telecommunications, instrumentation, and electrical engineering have remained below both the Spanish and world relative percentages in the respective portfolios. However, the biochemistry share (including biotechnology) is approximately the same as the world average.


The technological profile of Catalonia is concentrated in fewer fields than the Spanish profile or the world average. Furthermore, the changes from year to year are more pronounced, [6] like the relative decline of patent applications in the textile field and the increasing share of applications in the transport technology field for the years under study. These results suggest a specific industrial development.



The institutional carriers


The corporate addresses of the scientific articles indicate 62.1 % university research groups, 21.5% public research institutes, 12.7% hospitals, and only 3.6% industrial companies.[7] Table 5 provides an overview of how the data have been organised into institutional categories. During the period under investigation, the share of the non-university research institutes has increased, but the share of contributions with a hospital address has remained approximately stable.


University researchers take a central position in the pattern of co-authorship relations. Figure 4 exhibits these patterns for 1986 and 1996, respectively.  In our opinion, this data suggests that as in other OECD countries (Godin & Gingras, 2000 ) the academic research system has strengthened its position as the main carrier of the contribution to the international publication system of scientific communications. 


Patents do not always carry an institutional address: individual inventors can apply personally for a patent. Catalan patents are filed almost exclusively by individual inventors and firms (approximately 30% and 70%, respectively). Although individual inventors cannot be dismissed as producers of gadgetry, they usually face great difficulties to raise funds to develop and market their inventions (Macdonald, 1986).  The number of patents filed by universities has been zero during the period 1986-1989, and in the first half of the 1990s this number has grown to approximately three per year.  The number of patents filed by research institutes in the public sector exhibits a similar pattern.


We found only a single patent co-application with a university address in 1995. This was a patent with the CSIC, the national public research institute in Madrid.  The public research sector itself had seven cross-sectoral applications during this ten-year period, of which five were with industrial partners. For example, CSIC had one patent application with Prodesfarma, a pharmaceutical company in Barcelona, in 1992. [8] This patent concerns novel heterocyclic compounds, structurally related to ptheridines, with potential therapeutic properties, such as diuretics, bronchodilatators, and platelet antiaggregants. The search report established by the EPO contained three citations to articles published by one of the inventors, indicating a strong science base for this application.


Co-applications between individual inventors and industry have been more common. These co-applications may also indicate inventors who work in their (self-owned) SMEs.



The Science-relatedness of Catalan patents


Narin and Olivastro (1988, 1992) have developed an index of science-related patents by counting the number of international journals cited in the patent applications as a percentage of total citations within patents.  If this percentage is high, the patents draw heavily upon scientific literature and can therefore be considered as relatively science-based (Narin et al., 1997).  Grupp (1996) developed these measures further for European patents.


We analyzed patent applications with at least one Catalan address for the occurrence of non-patent literature citations (after correction for references to search reports of other patents). The numbers are provided in Table 6.  While 66% of the patent applications contain this type of references across the file, we found 50% as the value for patent applications with a Spanish address and only 40% for those with at least one Catalonian address.  This suggests that the Catalan industry might be less science-based than the Spanish average.


Inspection of the patents taught us that the patents that cite scientific literature are exclusively found in the pharmaceutical and in the clinical domains. The otherwise sparse co-authorship relations and co-applications are also found exclusively in these domains (that is, including biotechnology).  Hospitals have a large share of co-authorships in the collected SCI articles and companies authoring scientific articles are almost exclusively pharmaceutical ones.  Thus, the medical sector seems to be integrated differently from the other areas of RTD. The organization of public health provides a framework for a specific kind of integration (Blume, 1992).


Outside these domains we could not find substantive collaboration or indicators of mutual influencing through citations (see Tables 6, 7). Most of the companies with more than a single patent during this decade belong to either the chemical-pharmaceutical sector (science-based) or mechanical engineering (non-science-based). Patents in other science-based sectors like laser technology, microelectronics, optics, surface technology and inorganic chemistry are virtually non-existent in the Catalan set.



Dynamic analyses


Test for systemness


We tested the various distributions of patent applications and scientific publications for systemness, both in terms of the substantive categories (journal titles, patent classifications, etc.) and in terms of institutional addresses.  (The attribution of papers was based on “integer counting”, that is, each unit obtains a full count in case of a co-authorship relation.)


All these various distributions are found to be systemic using the tests described above.  However, this systemic character is generated by the absence of change. The distributions are reproduced from year to year with minor changes.  The assumption of the Markov property is always better than the prediction based on the individual time series of the respective subsystems, that is, the various scientific specialties and (technological) patent classes.


Note that this does not mean that the level of Catalonia is relevant for the integration. On the contrary, the descriptive results above suggest that both the academic system and the industrial system are integrated at other levels and that these different subsystems hardly disturb one another at the level of Catalonia.  The Catalan organization serves as a carrier of an otherwise differentiated system.



Tests for path-dependencies


Since the system has grown rapidly during the period under study, one can wonder whether the nature of the systemic distributions has changed.  As noted, the test for path-dependency provides us with a measure of the systemic integration at this level of the dataset.  If a system operates at the regional level, it can be expected to produce a path-dependency.


We always found path-dependencies in the time series of institutional distributions of addresses of authors and applicants.  However, path-dependency was (almost) never found in the distributions of patent classifications and journals. Most of the institutional path dependencies occurred around 1990/1991. Thus, the measure of path-dependency allows us to distinguish between the two types of systems: the institutional system is active, while the cognitive ones are not actively operating at this level of integration. (See Table 8 for a summary of these results.) 


This difference in path-dependency between the institutional and the cognitive categories can be considered as an indicator that the cognitive dimension of the system is less localised at the regional level than the institutional dimension. The institutional carriers of the system are anchored geographically and in the region, but the knowledge, once it is produced, can be published in peer-reviewed journals, or a patent application can be filed. International journals and patent classifications operate in global markets. Our results indicate that the institutional structures in Catalonia (as in other advanced regions) enable authors and inventors to participate in this global level of communication without restrictions stemming from their institutional locations.


Note that we do not deny the local dimension in the production of knowledge, but we draw attention to the possibility that scientific and technological knowledge is codified at another level. Knowledge production and control can be differentiated (Whitley 1984).  Knowledge control is achieved through codification (Luhmann, 1990; Cowan & Foray, 1997).  The cognitive dimension of scientific and technological communication is subject to a process of codification in journal articles and patent applications, respectively. Among other things, codification enables users to accelerate the transfer of the knowledge-based contents.


In summary, our results indicate that the Catalan system went through a rearrangement in institutional terms around the turn of the decade.  New patterns of collaboration have emerged and thereafter co-applications of patents and co-authorship relations in journal articles have become more common (that is, less rare) than before. Thus, the system has been put into another gear, but this interaction has remained marginal in relation to the overwhelming stability of the core (sub-)systems.  These latter are not firmly integrated with each other at the regional level. Differentiation has prevailed indicating integration at other levels (cf. Cimoli, forthcoming).





It is important to distinguish between the political discourse in the arena of science, technology, and innovation policies, and the practices to which these discussions refer. We do not wish to disclaim the legitimacy of a regional innovation policy, but we have investigated how the Catalan innovation system has been affected by the intense discourses about its restructuring during the latter half of the 1980s and the early 1990s. 


Undoubtedly, a new self-understanding of the role of regions has emerged in Europe (and elsewhere), and this perspective can be related to the emergence of the knowledge-based economies.  The role of the nation-state as a relevant level of integration for a system of innovations has come under pressure from the tendencies toward globalization and European unification.  Regional governance aspirations as in the case of Catalonia have legitimated efforts to shape an alternative system of innovations.


Our results indicate that the RIS of Catalonia has gone through a relatively short period of institutional change.  This can be considered as a consequence of the opening up of the Spanish economy, its specific position in relation to France given European economic integration across borders, and the Catalan aspirations.  However, the innovative activities have otherwise remained unchanged in character: they have remained differentiated at the Catalan level, and therefore integration is supposedly taking place at other (national and/or international) levels.  One can, for example, imagine that the regional universities are important as a window for the local industry on scientific developments at the international level.  In this case, one would not expect regional patents to cite regional articles.


The geographic distribution of science and technology in Europe is much more concentrated than population or economic activity in general, and amplifies the differences between centres and peripheries. However, the pace of concentration seems to be slowing down for science activities. Science and technology exhibit a strong pattern of co-concentration. Barcelona is classified as one of the 40 EU poles of science and technology, in the group of areas whose ranks are between 22 and 40 and contribute between 0,7 and 1,0 percent to the total EU value of science and technology (Zitt et al., 1999). Our results suggest that the geographical coincidence of science and technology activities is not reflected in an integrated regional system dynamics. On the other hand, the position of Barcelona in Spain, and to a certain extent in Europe, is neither that of a centre nor that of a periphery. Depending in the network dynamics of Spanish and European S&T activities, the intermediate position of Catalonia could hamper the development and/or integration of science and technology activities at the regional level.


Our research question was not whether Catalonia is a regional system, but whether it could be considered as a regional innovation system.  The answer to this question must be negative.  In our opinion, one must either modify the concept of  “regional innovation system” or conclude that this region is not a relevant level of the system of innovations.


With hindsight, one can argue that the model of “regional innovation systems” was shaped too much by analogy with the earlier model of national systems of innovation (Lundvall 1988; Nelson 1993). In pursuit of this analogy, scholars have tried to overcome the lack of hierarchical authority of national governments by introducing new concepts and models like regional innovation organizers (Gebhardt, 1997).  Can universities, for example, sometimes take the role of a local government in a region when regional authorities fail to provide the necessary guidance (da Rosa Pires & de Castro, 1997)?


It makes a difference in the local culture when knowledge-interests are coordinated, and when there is an organized exchange of expectations in terms of knowledge in addition to the mechanisms of exchange and coordination provided by the local economy but the envisaged development at the regional level underestimates, in our opinion, the complexities of global transitions towards a knowledge-based economy.  When the hierarchical model of state control is relativized, other models of integration have to be appreciated reflexively.


We have argued that the Catalan innovation systems can be expected to remain subsymbolic in character. Such regional innovation systems cannot be expected to replace the national innovation system, but they add a supplementary mode. Among other things, these innovation systems can be expected to remain juxtaposed to other networks of integration because of their network character. 


For example, not only regions, but also sectors, disciplines, problem fields, state apparatuses, etc., can be considered as relevant frames of reference for communication and interaction. One can expect the complex system to counterbalance control at the level of subsystems (de la Mothe & Paquet, 1996). In such cases, policies at one of the network levels are expected to be only moderately successful. Policies are also formulated reflexively, for instance, through the articulation of concepts such as complementarity and subsidiarity. As it is, policy initiatives emphasize the need for inter-relationship between the public and private sectors (CIRIT, 1997). Yet, the self-organization of the complex dynamics also promotes the further differentiation of communication because the system can then handle more complexity in its relevant environments.


Our results indicate that Catalonia as a region experienced a period of reorganization in the institutional basis of its innovation system during the early 1990s.  But perhaps the momentum has faded away since then.  We would guess that the ICT revolution integrates industrial regions in a different manner than was the case before the emergence of the Internet.  Knowledge-based communication systems have nowadays become complex and thereby non-trivial as mechanisms for integration.  From the perspective of the network, national and regional identities can be considered as  functional attributes (e.g., “made in Spain” or “”) rather than as essential characteristics.




An earlier version of the paper was presented at the NECTS-99 Conference about Regional Innovation Systems in Europe (San Sebastian, Spain, 30 Sep.- 2 Oct. 1999).  The authors acknowledge partial funding by the program for Targeted Social-Economic Research of the European Commission, project NR. SOE1-DT97-1060 (“The Self-Organization of the European Information Society”). We thank the European Patent Organization (EPO) for allowing one of  the authors to collect the patent data.






Arthur, W. B. (1989) Competing Technologies, Increasing returns, and Lock-In by Historical Events Economic Journal 99, 116-31


Bacaria, J., S. Borras Alomar (1998) The Catalan innovation system. Governing rapid changes, in: H. -J.  Braczyk, P. Cooke, and  M. Heidenreich (Editors). Regional Innovation Systems (University College London Press, London, Bristol PA) pp. 72-97.


Barcelo i Roca, M. (1993) Innovacio tecnologica i industria a Catalunya. (La Llar de Llibre, Barcelona).


Blauwhof, G. (1995) The non-linear dynamics of technological developments: an exploration of telecommunications technology, Unpublished Ph.D. Thesis, University of Amsterdam.


Blume, Stuart S. (1992) Insight and Industry: The Dynamics of Technological Change in Medicine (MIT Press, Cambridge, MA).


Borras Alomar, S. (1995) Governing Systems of Innovation: Regions and Technology in Europe. The case study of Catalonia in the 1980s and 1990s. Ph.D. Thesis, mimeo (Instituto Universitario Europeo, Florence).


Beck, U. ( 1999) Schöne neue Arbeitswelt (Campus Verlag, Frankfurt, New York).


Carpenter, M.,  F. Narin (1981) The adequacy of the Science Citation Index (SCI) as an Indicator of International Scientific Activity, Journal of the American Society of Information Science 32, 430-439.


Castells, M. (1997) The Power of Identity (Blackwell, London).


Cimoli, M. (Editor), forthcoming, Developing Innovation System: Mexico in the Global Context. (Cassell Academic, London).


Commissio Interdepartamental de Recerca i Innovacio Tecnologica (1997) II Pla de Recerca de Catalunya 1997/2000 (Generalitat de Catalunya, Barcelona).


Council of the European Communities, and Commission of the European Communities (1992) Treaty on European Union (Office for Offical Publications of the European Communities, Luxembourg).


Cooke, P. (1998) Introduction. The origins of the concept, in: Regional Innovation Systems, H.-J.  Braczyk, P. Cooke, and  M. Heidenreich (Editors) (University College London Press, London, Bristol PA) pp. 2-25.


Cooke, P., M. G. Uranga, and G. Etxebarria (1997) Regional innovation systems: Institutional and organisational dimensions, Research Policy 26, 475-491.


Cooke, P., M. G. Uranga, and G. Etxebarria (1998) Regional systems of innovation: an evolutionary perspective, Environment and Planning A 30, 1563-1584.


Cowan, R., D.Foray (1997) The Economics of Codification and the Diffusion of Knowledge, (MERIT, Maastricht).


Department for Trade and Industry (1998) Our Competitive Future: Building the Knowledge-Driven Economy: Analytical Background (DTI, London).


Etzkowitz, H., and L. Leydesdorff (1998) The Endless Transition: A “Triple Helix” of

University-Industry-Government Relations, Introduction to a Theme Issue, Minerva 36, 203-8.


Etzkowitz, H., and L. Leydesdorff (2000) The Dynamics of Innovation: From National Systems and “Mode 2” to a Triple Helix of University-Industry-Government Relations, Research Policy 29, 109-123.


Frenken, K., and L. Leydesdorff (2000) Scaling Trajectories in Civil Aircraft (1913-1997), Research Policy 29, 331-348.


Foray, D. (1998) Errors and Mistakes in Technological Systems: From Potential Regret to Path Dependent Inefficiency, in: Jacques Lesourne and Andre Orlean (Eds.) Advance in Self-Organisation and Evolutionary Economics (Economica, London) pp. 217-239 


Gasoliba Böhm, C. A. (1996) 10 anys d’integracio europea (Vicens Vives, Barcelona).


Gebhardt, C. (1997) Die Regionalisierung von Innovationsprozessen in der Informationstechnologie. Staatliche Forschungsfoerderung im Zeitalter der Globalisierung (Deutscher Universitaets Verlag, Wiesbaden).


Godin, B., Y. Gingras,  (2000) The Place of Universities in the System of Knowledge Production, Research Policy 29, 273-278.


Grupp, H. (1996) Spillover effects and the science base of innovations reconsidered: an empirical approach, Journal of Evolutionary Economics 6, 175-197.


Grupp, H., U. Schmoch (1999) Patent statistics in the age of globalisation: new legal procedures, new analytical methods, new economic interpretation, Research Policy 28, 377-396.


Guibernau, M. (1999) Nations without States. Political Communities in a Global Age (Polity Press, Cambridge).


Hirasawa, R., T. Ijichi, H.Tomizawa, and Y. Fujigaki (1998)  A Comparative Survey on Science and Technology Policy System and on Policy Formulation Processes. Report for the STA: Special co-ordinating program 1998.


Institut d’Estudis Catalans (1997) Reports de la recerca a Catalunya. Tecnologies de la informacio i de les comunicacions (IEC, Barcelona).


Katz, J. S. (1999) The self-similar science system, Research Policy 28, 501-517.


Lash, S., J. Urry (1994) Economies of Signs and Space (Sage, London).


Leydesdorff, L. (1991) The Static and Dynamic Analysis of Network Data Using Information Theory, Social Networks 13, 301-345.


Leydesdorff, L. (1995) The Challenge of Scientometrics: the development, measurement, and self-organization of scientific communications (DSWO Press, Leiden University, Leiden).


Leydesdorff, L. (2000) Is the EU Becoming a Single Publication System? Scientometrics (47) 2, 265-280.


Leydesdorff, L., and H. Etzkowitz (1998) The Triple Helix as a model for innovation studies, Science and Public Policy 25(3), 195-203.


Leydesdorff, L., and N. Oomes (1999) Is the European Monetary System Converging to Integration?  Social Science Information 38(1), 57-86.


Luhmann, N. (1984) Soziale Systeme: Grundriss einer allgemeinen Theorie (Suhrkamp, Frankfurt a. M.) [1995, Social Systems (Stanford University Press, Stanford, CA)].


Luhmann, N. (1990) Die Wissenschaft der Gesellschaft (Suhrkamp, Frankfurt a. M.).


Lundvall, B.-Å. (1988) Innovation as an interactive process: from user-producer interaction to the national system of innovation, in: G. Dosi, C. Freeman, R.R. Nelson, G. Silverberg and L. Soete (Editors), Technical Change and Economic Theory (Pinter, London), pp. 349-369.


Macdonald, S. (1986) The distinctive research of the individual inventor, Research Policy 15, 199-210.


Mothe, J. de la, and G. Paquet (Editors) (1996) Evolutionary Economics and the New International Political Economy (Cassell Academic, London).


Mothe,  J. de la, and G. Paquet (1998) Local and regional systems of innovation (Kluwer, Boston).


Narin, F., and D. Olivastro (1988) Technology Indicators Based on Patents and Patent Citations, in: A.F.J. van Raan (Editor), Handbook of Quantitative Studies of Science and Technology (Elsevier, Amsterdam), pp. 465-507.


Narin, F., and D. Olivastro (1992) Status report: linkages between technology and science, Research Policy 14, 237-249.


Narin, F., K. S. Hamilton, D. Olivastro (1997) The Increasing Linkage Between U.S. Technology and Public Science, Research Policy 26, 317-320.


Nauwelaers C., and A. Reid (1995) Methodologies for the evaluation of regional innovation potential, Scientometrics 34, 497-511.


Nelson, R. R. (Editor) (1993) National Innovation Systems: A comparative study (Oxford University Press, Oxford and New York).


Nelson, R.R. (1994) Economic Growth via the Coevolution of Technology and Institutions, in: L. Leydesdorff and P.Van den Besselaar (Editors), Evolutionary Economics and Chaos Theory: New Directions in Technology Studies (Pinter, London and New York), pp. 21-32.


Nelson, R.R., Winter, S.G. (1975) Growth Theory from an Evolutionary Perspective: The Differential Productivity Growth Puzzle, American Economic Review 65, 338.


Nelson, R.R., Winter, S.G. (1982) An Evolutionary Theory of Economic Change (Belknap Press, Cambridge, MA).


Organisation for Economic Co-operation and Development (1999) OECD Science, Technology and Industry Scoreboard 1999. Benchmarking Knowledge-based Economies (OECD, Paris).


Patel, P., and K. Pavitt (1994) The nature and economic importance of national innovation systems. STI Science Technology Industry Review 14,  9-32.


Pavitt, K. (1984) Sectoral patterns of technical change: towards a theory and a taxonomy, Research Policy 13, 343-73.


Pavitt, K. (1985) Patent Statistics as Indicators of Innovative Activities: Possibilities and Problems, Scientometrics 7, 77-99.


Petitbo, A., and J. Bosch (1990) Estructura Industrial, in: M. Parallada (Editor), Estructura economica de Cataluna, (Espasa-Calpe, Madrid), pp. 225-250.


Rosa Pires, A. da, and E. Anselmo de Castro (1997) Can a Strategic Project for a University be Strategic to Regional Development? Science and Public Policy 24, 15-20.


Shinn, T. (1998) The impact of research and education on industry: a comparative analysis of the relationship of education and research systems to industrial progress in six countries, Industry & Higher Education 12 (5), 270-89.


Schumpeter, J.A. (1934) Theorie der wirtschaftlichen Entwicklung. Eine Untersuchung über Unternehmergewinn, Kapital, Kredit, Zins und den Konjunkturzyklus (Duncker & Humblot, Berlin).


Theil, H. (1972) Statistical Decomposition Analysis (North-Holland, Amsterdam)


Vence, X. (1998) La politica tecnologica comunitaria y la cohesion regional. Los retos de los sistemas de innovacion perifericos (Civitas, Madrid)


Webster, A., K. Packer (1996) Intellectual property and the wider innovation system in: A. Webster and K. Packer (Editors), Innovation and the Intellectual Property System (Kluwer, London), pp. 1-19.


Whitley, R. R. (1984) The Intellectual and Social Organization of the Sciences.

(Clarendon Press, Oxford).


Zitt, M., R. Barre, A. Sigogneau, and F. Laville (1999) Territorial concentration and evolution of science and technology activities in the European Union: a descriptive analysis, Research Policy 28, 545-562.



Figures and Tables


Table 1


R&D Funding for all S&T fields in the year 1994. In millions pesetas. Percentages of the total for Catalonia in brackets.  Source: IEC,1997








Catalan funding





Spanish funding





European funding












Table 2

Summary of the patent and publication data collected with at least one Catalan address



Number of EP patents

Number of occurrences  of patent applicants

Number of different patent applicants

Number of SCI articles

Number of occurrences of corporate sources

Number of different corporate sources





















































































Table 3

Overview of the data





SCI articles


EP patents


1564 Journals, 857 Journals

31 patent class groups,

10 patent class groups


·       Universities

·       Enterprises

·       Research Institutes

·       Hospitals



·       Universities

·       Enterprises

·       Research Institutes

·       Individual inventors





Table 4

Concordance table among the 10 groups defined by the authors the 31 groups as defined in the annual reports published by the EPO




10 Groups

31 Groups

IPC sections


General necessities


A01, ex. A01N



Foodstuffs and tobacco




Personal and domestic articles




Health and amusement







Treating materials

Preparing and mixing




Shaping I




Shaping II








Transporting I




Transporting II








Fermentation, sugar, skins







Inorganic chem., metallurgy

Inorganic chemistry











Organic chemistry

Prep. for med., dent. or toiletry




Organic chemistry




Macromolecular compounds




Dyes, petroleum, oils







Textiles, paper, printing





Textiles and flexible materials




















Engines and pump




Engineering in general




Lighting and heating




Weapons, blasting







Instruments, computers

Instruments I




Instruments II




Instruments III







Electricity, telecomm., electronics

Nuclear physics




Electric techniques




Electronics and telecommunication







Not classified

Not classified


Table 5

Percentage of patent applications in each of the 10 patent classification groups

for Catalan, Spanish and European data, in the period 1986-1996






General necessities




Treating materials












Inorganic chem., metallurgy




Organic chemistry




Textiles, paper, printing








Instruments, computers




Electricity, telecomm., electronics









Table 6

Non-patent literature cited in search report






All EP applications with priority year 1986-95




EP applications with  search report  1




EP applications citing non-patent literature 2





Ratio of occurrence of non-patent literature to patent literature








1 Established by August 1999, as retrieved in the EPODOC database.

2  As proxied by the field CTNP (in the same EPODOC database) which contains references to publications in (scientific) journals, abstracts of those and/or abstracts to patent applications and granted patents published in a non-official language of the European Patent Office (for instance, Japanese patent applications or Soviet/Russian patent publications). Thus, this figure is an overestimation of the number of European patent applications really referencing to non-patent literature. Nevertheless as the overestimation is expected to be equal in the three geographical areas, the ratios are reliable relative indicators.    



Table 7

Number of patent applications with references to non-patent literature classified per technical field



Number of patents

Medical technology and medicinal chemistry


Macromolecular chemistry, paints




Inorganic chemistry


Shaping materials




Internal combustion engine


Control technology


Electrical power technology




Display technology, TV, sound and image




Computerised design



Total of patents with non-patent literature



Total of patents with a search report



Ratio NPL/Search report

0.31 1

1 This number is lower than the ratio in table 6 (0.40) because 0.31 is an exact ratio, and not an overestimation.  

Table 8

Summary of results on testing for  “systemness” and “path-dependency”






























Figure 1

“Sender,” “receiver” and reorganization of a signal by an “auxiliary transmitter”


Figure 2

Percentage of publications by Catalan authors in science journal as a percentage of world total publications.






Figure 3

Percentage of patent applications by Catalan inventors and/or applicants before the European Patent Office (EPO) journal as a percent of total European patent applications.












Figure 4

Share of  Sectors’ Papers Written with Universities (1986 and 1996). (The figure is adapted from Godin & Gingras, 2000)









[1] Article 149, point 1.15.

[2] Article 9, point 7.

[3] The journal data analysis was based on Barcelona province only, while the patent data analysis was based on the Catalan region. The latter includes three smaller (Catalan) provinces in addition to Barcelona. The share of Barcelona province in the Catalan journal data set in the years under study is approx. 80 percent.

[4] These include world patent applications which are made valid for the European region. The dates refer to the priorities of the patent applications, and not to the publication dates.

[5] I is the expected information of the message that the distribution has changed.  It can be defined as follows (Theil, 1972; Leydesdorff, 1995): I  = - S qi 2log qi/pi.. In this formula, S qi denotes the posterior distribution, while S pi  represents the probability distribution before the event that caused the revision of the prediction.

[6] The variations from year to year are larger in the Catalan case because of the smaller sets.

[7] Katz (1999, p. 503) found in the set of 43,000 UK research papers indexed in the 1994 Science Citition Index, 65% containing university addresses, 10% research council researchers, 27% hospitals, 8% from industry, and 2 % from nonprofit sectors.  He noted that 50% of the latter two categories of publications (from industry and non-profit organizations) were coauthored with scientists from the public sector institutes. 

[8] PN-EP580916; Publication date- 19940202; Priority date- 19920722.

* The title to this classification group was given in 1978 as description of ‘traditonal’ biochemistry, but it also includes genetic engineering.