Innovation Systems: A Survey

s are available only for journal articles. Therefore, it has not been possible to review this whole literature in detail. Beyond the statistical summary and classification presented here, there are many topics to explore in further research; some of these are indicated in the analysis that 7 follows. But the analysis of the more specific content of the literature remains somewhat impressionistic and superficial at this time; much more could be done. National Innovation Systems (NIS) Of the 381 publications classified as NIS studies, 147 (38 %) are focused on individual countries. 55 of these study European countries, 47 Asian countries, 22 Latin America, 14 North America, and 9 the rest of the world. Japan is the most frequently studied country (17 studies), followed by China (11), Finland and Germany (9 each). The most common orientation of these studies is toward policy discussion (66), general description of national innovation systems (21), and focus on a particular sector or industry (19). About one-third of these studies deal with developing or transition economies. 51 NIS-studies are comparative in nature (comparing one country or set of countries with another). 164 (43 % of NIS studies) are not focused on any particular country or group of countries but discuss concepts/theory (56), policy (43), issues having to do with globalization (42), or other issues without reference to country. These could also be classified as general innovation system studies. Regional Innovation Systems (RIS) 201 studies are focused on regional innovation systems (RIS). Slightly more than half (103 studies) are empirically oriented, focusing mostly on a particular region (62) or on multiple regions (24). More than half of these studies deal with regions within Europe. The other empirical studies are case studies of various sorts involving innovation surveys, patent analyses, 8 globalization issues, or innovation policy. Of the 93 non-empirical RIS studies, 70 are conceptual in nature and 11 are policy-oriented. Sectoral Innovation Systems (SIS) There are 49 published studies of sectoral innovation systems (SIS). 30 of these focus on individual sectors or industries (the service sectors and the biomedical/pharmaceutical industry being most heavily represented). 9 studies are conceptual, three are comparative, four are policyoriented. The remaining three fall into a miscellaneous category. Technological Innovation Systems (TIS) The technological innovation system studies differ from others not only in that they are more narrowly focused (being defined by a particular technology or set of technologies rather than a geographic region or industry) but also in that they are more conceptual/theoretical in nature. This is largely a result of the need to establish both the core and the boundaries of the systems before the analysis can take place. These issues are much less problematic in other approaches. Also, technological innovation systems have three dimensions (cognitive, institutional/ organizational, and economic – see Carlsson 2002), while other approaches focus primarily on institutions. Thus, of the 149 studies of technological innovation systems, more than one-third (57) are conceptual in nature. The remaining two-thirds are either case studies of various sorts or otherwise classified. The biotech/biomedical/pharmaceutical sector is the most frequently studied (17 studies), followed by agriculture (8), factory automation (6), and information technology (5). 9 Other Innovation Systems The ‘Other innovation systems’ category contains 30 publications. 19 of these are conceptual in nature without specific reference to any of the types of innovation systems previously mentioned, or refer to innovation systems in general. 11 focus on corporate innovation systems and related management issues. As shown in Figure 1, the number and focus of innovation studies have varied over time. After the first few studies on NIS (Freeman, Lundvall, Nelson et al.) and technological innovation systems (particularly focused on agriculture) in the late 1980s, the numbers increased dramatically in the early 1990s, peaking at 175 in 2000, and then declined sharply. Regional and sectoral innovation system studies began to appear in the late 1990s. The large number of studies published in 2000 appears to be a coincidental result of several books being published in the same year. The number of publications of RIS and SIS studies was particularly large that year compared with other years. Several books on RIS were published by Dunning, Holbrook, Boekema and others. Similarly, half of the SIS studies published in that year are chapters in books on the service sector (edited by Metcalfe & Miles, Boden & Miles, and Andersen et al., respectively). Overview of Topics and Themes 206 (27 %) of all the innovation systems publications are conceptual/theoretical in nature. As indicated already, the definition of boundaries and core activities is more problematic in regional and technological innovation systems than in others. This is reflected in the fact that a larger share of the regional (36 %) and technological (34 %) innovation system studies are conceptual 10 than is the case for other systems. The corresponding numbers for SIS and NIS are 21 % and 16 %, respectively. Of all the innovation system publications, 11 % have a sector focus. As one would expect, the SIS studies are the most sector-oriented: 58 %. (Other SIS studies are primarily conceptual in nature.) It is perhaps more surprising that as many as 9 % of both NIS and TIS studies and only 4 % of regional studies are focused on a particular sector or industry. To some extent this reflects difficulties of appropriate labeling. For example, studies of the role of particular sectors in a national innovation system are generally classified as both NIS and SIS. They are often parts of edited volumes focusing on a particular national innovation system and its components. In other cases the terminology used in the studies refers to national innovation systems, even though a sectoral designation would be more appropriate. Similarly, some TIS studies use the term ‘technological’ when ‘sectoral’ would be more appropriate. These difficulties are an unavoidable result of the procedure used to identify entries into the database. It is interesting to note, however, that the sector focus has shifted markedly over time. All innovation system studies have become much more sector-oriented (18 % in 2000-2002, compared with only 11 % in 19871999). The shift has been particularly dramatic in NIS studies: from 6 % in 1987-1999 to 16 % in 2000-2002. This suggests that as more has been learned about innovation systems at all levels (and especially at the national level), there is a greater need for more detailed, micro-based studies. Only a small subset (about 60 studies) can be considered ‘dynamic’ in the sense that they focus on a historical process or development over time rather than on a snapshot of a system in a 11 particular time period. There are even fewer studies dealing with new system formation, leaving an as yet wide open area for future research. It is tempting to conclude that Schumpeter’s vision of the dynamics of what he called the “economic system” is not yet fully developed: most studies still adhere to a static view of the world. Schumpeter distinguished sharply between invention (the original idea for a new product or process), innovation (its conversion into a commercializable product), and the diffusion of innovations. The innovation systems literature is heavily oriented to the earlier (invention) stage and to some extent diffusion, with relatively little emphasis on the innovative (entrepreneurial) stage. This is somewhat surprising, given the prominence of entrepreneurship in Schumpeter’s work, and the Schumpeterian origin of innovation system studies. Only about 20 studies address entrepreneurial issues. Thus, it appears that innovation systems are more deeply rooted in Schumpeter’s later work (Capitalism, Socialism, and Democracy) than in his earlier work (The Theory of Economic Development) that features the individual entrepreneur more prominently. It also appears that to the extent that entrepreneurial activity is necessary to convert innovation into economic growth, there is a missing link in the innovation systems literature. This is reflected also in the discussion and analysis of public policy in the literature. 190 of all the publications (25 %) deal with policy issues. The NIS studies tend to be the most policyoriented (34 %), while 24 % of RIS and 13 and 12 % of sectoral and technological innovation system studies, respectively, have a policy focus. Again, this state of affairs is no surprise. To a large extent it reflects the fact that it is easier to identify the relevant policy makers with respect 12 to nations and regions than in sectoral and technological systems. It is also easier to identify policy measures at the national level than at other levels. As one would expect, the policy discussion in the NIS studies tends to focus on national policies with respect to the technology infrastructure: promotion of R&D, intellectual property rights (especially, patent laws), the role of public and private research and technology institutes (particularly university-industry collaboration, technology transfer, and the role of science parks), as well as trade policy and the role of foreign direct investment. This reflects the fact that public policies in all these areas form an important part of the infrastructure for all innovation systems within nations (including regional, sectoral, and technological innovation systems). The lower the level of aggregation, the more qualitative and specific the policy analysis becomes, focusing more on interaction among actors and on institution building. It is therefore difficult to summarize briefly. However, it can be safely said that throughout the innovation systems literature, the primary policy concern is to improve the technology infrastructure and therefore increase the supply (and to some extent improving the diffusion) of innovations rather than stimulating entrepreneurship. Even though institutions are deeply imbedded in innovation systems and are the primary focus in many studies, it should be noted that the definition ‘institutions’ varies among studies and that, as a result, there is considerable confusion about what institutions are and what role they play. Some authors, (e.g., Freeman 1987 and Nelson & Rosenberg 1993) refer to institutions as networks or organizations supporting technical innovation, while Lundvall (1992) stresses the “institutional set-up” in the sense of rules or regimes that determine behavior. Carlsson & 13 Stankiewicz (1991) refer to the set of institutional arrangements in the form of both regimes and organizations. What is clear is that most innovation system studies use the notion of supporting organizations and that there is not much analysis or discussion of the specific mechanisms through which institutions work. One consequence of this lack of in-depth analysis of institutional mechanisms is a relative neglect of the role of financial institutions, mechanisms, and arrangements. Only five studies have finance as their primary focus. This is in sharp contrast to Schumpeter’s thinking. As Freeman has observed, Schumpeter devoted far more attention to the financial side of business cycles (in his Business Cycles, published in 1939) than to inventions and innovations. “More important was his preoccupation with the individual entrepreneur and the individual innovation, and his reluctance to conceptualize invention, innovation, and technology accumulation as a social process. This is related to his theory of diffusion with its sharp distinction between truly original entrepreneurs and routine managers and imitators” (Freeman 1990, p. 24). Of course, another reason for the relative lack of emphasis on the finance of entrepreneurial enterprise is the limited attention given to entrepreneurial activities in innovation systems. More or less in parallel with innovation system studies there has emerged another branch of economic analysis that has many similar features, namely the study of industry clusters. A lot of this work has been inspired by Porter (1990) and colleagues. What is the difference between a cluster and an innovation system? If a cluster is defined as a set of closely related business activities in a certain geographic region, the difference would be that an innovation system 3 For further discussion of these definitions, see Edquist and Johnson (1997). 14 differs from a cluster in that it takes into consideration the whole set of factors (especially institutional ones) that are conducive to the formation of a cluster. Most cluster definitions in the literature thus far ignore institutions (Porter being a notable exception). Probably mostly for this reason there is surprisingly little overlap between “cluster-focused” and innovation systemfocused publications: only 63 out of 752 innovation system publications reviewed here mention clusters. But the overlap between the two strands of literature has increased over time, most of it involving publications in 1996 and later. As mentioned earlier, about 50 NIS studies are focused on individual developing or transition countries. Many of these are cross-referenced as SIS or TIS studies also. Beyond these, there is an additional handful (about 10) publications dealing with innovation systems in developing/ transition economies but not focusing on any particular country. It seems fair to say that this is a relatively undeveloped part of the innovation systems literature. But there seems to be increasing interest in innovation systems in developing or transition economies; the vast majority of publications in this area have appeared in 1999 or later. Many of these studies deal with the problem of catching up with more advanced countries and importing technology, knowledge, and ideas, particularly via direct foreign investment and repatriation of nationals educated abroad. Another area that has not received much attention in the innovation systems literature is the performance of various systems. Only about 20 studies are aimed at assessing the performance of innovation systems. There may be several reasons for this. One is certainly the difficulty of measuring performance: what indicators should be used? (Only 11 studies discuss measurement issues specifically.) What indicates high or low performance – i.e., what should be the standard? 15 Relative to a different time period? This requires historical data that are difficult to obtain. Comparisons with other systems? Given the detailed and complex data requirements, such analyses are also extremely difficult. One consequence of the lack of performance data and analyses is that there is still no connection between innovation and economic growth. Through the study of innovation systems we have learned a lot about the contents of the ‘black box’ that converts innovation into economic growth, but there are still missing links: As already indicated, the role of entrepreneurship connecting invention via innovation to successful commercial application and diffusion is poorly understood. While there has been a lot of recent work on entrepreneurship, it has not generally been integrated with innovation systems. Also, there has not been much theoretical work explicitly connecting innovation systems to economic growth. As a result, there is little formal modeling in the innovation systems literature. Only 10 studies involve modeling; six of these pertain to technological innovation systems. Beyond a few simulation studies there is no empirical testing of hypotheses. Thus, in spite of hundreds of innovation system studies we have not really advanced much (yet) beyond the endogenous growth model. We still lack understanding of how to measure success and what makes innovation systems successful. There is still much to be done. What Have We Learned, and What Difference Does Research on Innovation Systems Make? Perhaps the most important insight gained from the study of innovation systems is a better understanding of how complex innovation systems are – and how complex the growth process is. 16 There is much more to innovation – and to economic growth – than an aggregate production function captures. Even though there are still missing elements in our understanding of the links between innovation and economic growth, the study of innovation systems has already resulted in a deeper and more comprehensive view of economic growth. This is certainly consistent with Schumpeter’s ideas about growth originating within the system and about the role of history and institutions. The new insights are limited but they are still useful in that (1) they help economists better understand how to think about innovation and its role in economic growth, and (2) they put industrial/technology policy in a broader framework than was the case previously. The questions raised are different and more qualitatively oriented, with attention given not only to the end results but also to the mechanisms involved. Even though the policy recommendations may differ, there is certainly consensus that more attention than in the past needs to be given to institutions and institution building. Policy makers have responded at all levels, from international organizations such as the OECD and national governments (by re-organizing their technology policies and agencies to focus on innovation systems as distinct from more piecemeal policies) to regional and sectoral agencies. The various systems approaches are complements, not substitutes, each focusing on a particular domain with its own issues, problems and opportunities. The policy recommendations vary among the various systems approaches, but they are not necessarily inconsistent. They basically reflect the fact that different systems address different questions. Though the study of innovation systems has charted a new course in economic analysis, it is not a smooth and easy one. There are many obstacles and bumps in the road ahead: how to formalize the theoretical insights that have already been gained, how to link microeconomic phenomena 17 with macroeconomic outcomes, and how to correctly measure both inputs and outputs are just a few. There seems to be no escaping building the micro foundations (i.e., micro dynamics) for understanding macroeconomic growth. Innovation system studies represent an important step in the right direction.