Legitimising RTI-policy: Market failure and systems failure Two sides of one coin? and beyond

Transcription

1 Legitimising RTI-policy: Market failure and systems failure Two sides of one coin? and beyond Paper presented at the EAEPE Conference, Porto, 1-3 November 2007 Authors: Klaus Kubeczko and Matthias Weber ARC systems research, Austria Abstract: Market failure and systems failure are the most widely used rationales to legitimise research, technology and innovation policy intervention. However, the systems failure argument often tends to be regarded as "just another type of market failure" argument. The goal of the paper is first of all to question this simplistic interpretation of systems failure and give the institutional and evolutionary arguments underpinning the systems failure approach the prominence it deserves. Secondly, by linking the discussion on systems failure to the one on functions of innovation systems, we propose a framework that allows using the systems failure argument in a consistent and operational way for underpinning policy. Thirdly, we will further develop our complex innovation system (CIS) approach, which is expanding the functional perspective on innovation systems by drawing on complexity arguments, in order to derive additional arguments to legitimate political intervention. This approach opens up a number of inroads for policy that can not be captured by the market failure arguments and extends the systems failure argument, to better legitimise demand side policies. Keywords: Innovation System, Technology Policy, Market failure, Systems failure, systemic failure, Functions of Innovation Systems

2 Introduction The legitimisation of public policy intervention to support innovation activity is becoming a growing issue. The reason is twofold. First, there is the growing importance of R&D in the political debate caused by the Lisbon strategy in general and, in particular, by the search for appropriate demandside policy measures that are discussed around Europe. Second, the growing need for innovative solutions for other societal challenges, like adaptation and mitigation strategies to reduce global warming, other environmental problems, the aging society, and so on. These societal problems likewise require R&D activities to back up transition process for a more sustainable development, but which while being widely recognised as major future challenges have so far met difficulties when it comes to deriving well-legitimised strategies for policy intervention and justifying the use of novel policy instruments. Most prominently the Kok Report (Kok 2004) and the Aho Report (Aho 2006) draw attention to the procurement policy and other measures to promote pioneer markets or Lead-markets (Georghiou 2007) for new research and innovation-intensive products. The Aho report, for example, suggests using procurement policy to foster innovation activity. Based on that report, the Commission intensifies its efforts to develop new instruments. At national level, in the United Kingdom, the Government's Innovation Report of 2003 proposed a series of measures, aimed at increasing the research and innovation impact of public procurement. Other countries like Spain, Ireland, Austria, and Germany also commissioned studies or workshops on procurement policy issues. Representing 16.3% of European GDP, public procurement is both a key source of demand for firms in sectors such as construction, health care, environment, security and transport, and a major area in which governments are striving to improve effectiveness. Procurement policy is also discussed down to the regional level as almost half of public procurement budget is spent by sub-central bodies (DG_Research 2005). To give an example from the regional level, a paper from a participatory strategy process in the City of Vienna suggest to develop new tools to link RTDI-policy activities to the city s procurement policy with an annual budget of over 3 billion EUR. Based on mainstream economic argument, market failure is probably the most widely accepted argument to legitimise policy intervention in economic systems, and in particular in the context of RTDIpolicy measures. Market failure arguments might be sufficient to legitimise certain types of policy measures, including public procurement and measures to face societal problems. However, we argue, that the implications of market failure are (1) very often too rough to give policy guidance and (2) not addressing all problems phased by policy makers in the field of RTDI. Systems failure is another failure-argument that legitimises policy intervention in respect to research, development and innovation based on the innovation system approach (Woolthuis et al. 2005; OECD 2002). While being increasingly accepted in RTDI policy debates, it is used rather diverse and inconsistent ways, with often ill-defined boundaries with some market failure arguments. A key question of this paper is therefore: Is there any reasoning added by system failure arguments that can not already be provided by market failure arguments? In other words: Can Innovation Systems heuristics provide a more profound rationale for public policy intervention than neoclassical approach using market failure arguments? And how can the different arguments be integrated into a consistent conceptual framework? Institutional and evolutionary economics, we will argue, already provides elaborated reasoning for legitimising public policy intervention and it is able to contribute to the further development of the systems failure argument in the light of the present debate on demand-side policy issues. However, in order to overcome conceptual inconsistencies and to broaden the range of arguments to legitimise

3 policy interventions in the face of major societal challenges, the theoretical basis of the system failure argument needs to be revisited and expanded. Prior to discussing the questions of systems and market failure being two sides of one coin, we will briefly explain both failure arguments. We will then explore the argument of market failure being not sufficient to legitimise present policy measure. We will continue by outlining a discussion, which is related to the discourse on systems failure, and which is dealing with functions of innovation systems. This debate bypasses the failure doctrine completely by providing arguments related to the functions or services that can (and should) be provided by an innovation system taking into account that there is a division of labour between different categories of actors (policy, performers, intermediaries). This relatively recent field of innovation systems research can lead the way to a new proactive doctrine that provides a framing to conceptualise demand side issues including societal challenges. We contribute to this debate by building on the theoretical approach of complex innovation systems (CIS).It stresses the complex dynamics of innovation systems, implying the need for an integration of anticipatory and adaptive elements into policy making (including proactive demand side measures), as well as the need to provide an orientation for the direction of search processes that are in line with societal needs and not only with the search for economic benefits alone. This might help to better legitimise the role of public technology procurement and mitigation and adaptation measures that might solve the dilemma of policy makers perceiving societal challenges that might require a more active (more risky, more experimental) role of policy making. Approaches to systems failure Systems failure is a term introduced into the debate on legitimisation of RTDI policy intervention. The system it refers to is the innovation system. Systemic imperfection and systemic problem are others terms used to describe basically the same thing. As innovation system approaches are based on evolutionary and institutional economics which draw attention to a wide range of phenomenon in the economic sphere, like the importance of institutions and networks, the problem of path dependencies, variation and selection mechanisms, etc., the corresponding literature always mentions several different types of failure arguments, that are said to go beyond the market failure perspective. Frequently Franco Malerba (1998) and Keith Smith (1999), Charles Edquist et al. (1998) and Carlson and Jacobson (1997)are cited to give an overview of systems failure arguments (see. E.g. Borras and Lundval (2003), Woolthuis et al. 2005, Gustafson, STEP Report 1999 (Hauknes and Nordgren 1999). For the early literature in the 90ies, eight failure arguments are being summarised.by (Woolthuis et al. 2005): 1. Infrastructural failures (Smith 1999; Edquist et al. 1998) 2. Transition failures (Smith 1999) 3. Lock-in/path dependency failures (Smith 1999) 4. Hard institutional failure (Smith 1999). 5. Soft institutional failure (Smith 1999; Carlsson and Jacobsson 1997). 6. Strong network failures (Carlsson and Jacobsson 1997) 7. Weak network failures (Carlsson and Jacobsson 1997) 8. Capabilities failure (Learning): (Smith 1999) and (Malerba 1998)

4 Similar to the above cited summary (Chaminade and Edquist 2006) speak of several systemic problems. They avoid talking of failures because failure arguments are related to questions of efficiency which can not be evaluated based on the concept of innovation system. Chaminade and Edquist (2006) argue that the term failure is not appropriate as there is no optimal system that shall be achieved, designed or constructed. Therefore they use the term systemic problems, that shall underpin the difference between market failure argument of neoclassic theory and emphasise the background of the IS approach in evolutionary economics. Similar to the categories by Woolthuis et al. (2006), they combine some categories and extend the list by two new categories: 1. Infrastructure provision and investment problems 2. Transition problem (to foresee the emergence of new paradigms, radically new pervasive technologies or significant changes in the markets that require new technological solutions) 3. Lock-in problems 4. Hard and soft institutional problems 5. Network problems 6. Capability and learning problems 7. Unbalanced exploration-exploitation mechanisms 8. Complementarity problems (in terms of competences shared between the actors in the system) (Metcalfe 2003) concentrates on three arguments to describe deficiencies in innovation systems and why policy intervention should tale place. These are: 1. Wrong system boundaries 2. Inappropriate organisations 3. Connections are not functioning Although the failure arguments are developed along the lines of evolutionary and institutional economics, there seems to be a lack of conceptual underpinning of the list provided, in particular with respect to demand-side policies. The systems failure arguments concentrate on the generic features of the innovation system which are widely covered and can profoundly support the justification of policy measures to improve the internal functioning of the innovation system up to the point of development of innovations that can be sold on a market. Reasoning based on market failure The practice of policy making is very much dominated by the market failure argument. However, what is rarely made explicit is: Which markets are policy makers referring to? We think that it is important to distinguish at least two different kinds of markets referred to in the literature. First, it is the market for knowledge which is most commonly referred to as a rationale for public policy intervention. This knowledge is expected to be transformed into innovation. If not, we speak of knowledge-market failure. Second, it is the market for products or services that demand for innovations. If a lack of innovative products or services is perceived, we speak of product-market failure. Researchers dealing with RTI-policy are very often referring to markets for knowledge. From this point of view, which is formulated by Metcalfe (2003), the problem is that of market failure to invest in

5 knowledge production. He outlines that [c]entral to this rationale is the idea that markets in relation to knowledge and information have an inherent tendency to produce socially inefficient outcomes, inefficiencies, which provide the justification for failure correcting public policies. This has proved to be a powerful set of ideas for shaping policy debate, particularly concerning the public support of university based science and technology that are far from market application (Metcalfe 2003). Knowledge-market failure is addressed by the Arrow-Nelson argument (Arrow 1962; Nelson 1959). There are three problems, leading to knowledge-market failure: uncertainty, inappropriability, inseperability (Hauknes and Nordgren 1999; Arrow 1962; Nelson 1959): Uncertainty is involved at the side of research activity that can not always produce innovations with calculated risk as well as through uncertainties of adaptation by the markets. The principle-agent problems related to information asymmetry is another problem faced by innovators which are related to discrepancies in risk assessment. Public goods and externalities are keywords for inappropriability. Public policy support is justified by expected knowledge spill-over in basic research. Inseperability is related to a minimum investment into knowledge which innovative firms might not be willing to share with others. Underlying assumptions leading to the knowledge market failure doctrine are: 1. A linear model of knowledge creation which was commonly used up to the 1990ies, dating back to Francis Bacon 2. The market mechanism, although considered to be the most efficient coordination mechanism, is not the only mechanism sufficient to support economic growth induced by technologically progress (otherwise there would be no reason for state intervention as another hierarchical and topdown way of coordination) Knowledge-market failure is not necessarily sufficient to legitimise policy intervention satisfying societal demand for environmental goods that need to be provided somehow. A first important question is: Who provides the knowledge? If the market can not provide it, the state may intervene to install other coordination mechanisms that can provide the organisational structures and incentives supporting the necessary function. Another and a more pressing question is: Who provides the products that are necessary for the state to fulfil its role as procurement agent aiming to satisfy important consumer needs, when markets can not provide solutions that are technologically satisfying? This holds true for societal challenges as health and environmental aspects. E.g.: There is a demand for more energy efficient public buildings as the state has to fulfil its obligations as consumer to reduce greenhouse gases. The state must be interested in products that are based on more efficient technologies for public hospitals to reduce the predictably increasing cost of the health system of an aging society. As such the state is a consumer with a large purchasing need of products that are not yet available on a market. The problems raised here can not be conceptually covered by the knowledge-market failure argument. It is to be conceptualised as another kind of market failure which we call the product-market failure. Product-market failure can be related to markets failing to efficiently provide or allocate goods and services and hence, in the context of RTI-policy goals, to the lack of innovations on existing markets and to the complete lack of markets. Also here the three categories used above - uncertainty, inappropriability, inseperability are relevant to identify the market-failure problem: Uncertainty is involved at the side of knowledge producers through uncertainties of adaptation by the markets. Inappropriability needs not always be related to basic research; it can also be related to the innovative products or services provided from which consumers can not be excluded easily (problems faced extensively by the software and music industry). Inseperability is related to large technological systems that are in most cases coordinated by oligopolistic or monopolisitic markets with few incentives to innovate due to the favourable market situation for the firms which provides hardly any incentive to innovate.

6 In all three respects, public policy intervention can provide incentives for innovation activity by means of regulation, public procurement policy, etc. Market failure is a doctrine that introduces a kind of subsidiarity principle into the discussion on legitimisation of political intervention in the economic system. Unlike the subsidiarity principle in the Maastricht treaty, which is meant to provide a legitimatisation of top down policy intervention from the highest administrative body (i.e. the European level), the subsidiarity principle underlying the market failure doctrine assumes a hierarchical relation between the economic and the political system in which the political system is subordinated to the economic system. Provided that policy making is following these assumptions, the market failure doctrine provides a simple argument to legitimise policy intervention and the underlying theory can provide some policy recommendations: Mitigate (1) non-desired externalities as well as (2) asymmetries of information, (3) eliminate barriers of entry to markets, (4) correct inefficiency in market structures, (5) secure intellectual property rights (Metcalfe 2003). Market failure and systems failure: Two sides of one coin? On the one side, many arguments for public policy intervention can be reduced to either, knowledgeor product-market failure arguments. On the other side, many arguments can also be reduced to systems failure. Instruments supporting science-industry relations are often legitimised using the argument of systems failure. One of Metcalfe s (2002) arguments is that connections are not functioning; Woolthuis et al. (2005), Smith (1999) and Malerba (1997) argue that capabilities failure of firms, especially SMEs which are not able to adapt to knew technologies, are important aspects of systems failure. Other system failure arguments might also be used to legitimise these kinds of instruments. However, it can also be argued that the market mechanism is not working appropriately in these cases, and that measures have to be taken by the state because of this failure. If all identified problems can be argued by market failure, the market failure doctrine would be sufficient to legitimise state intervention. To use market failure as argument legitimising policy intervention in promoting linkage of science and industry or other forms of network activities, however, is only possible when having a concept of Innovation System in mind. In a strict interpretation of market failure based on its neoclassical theoretical foundation such type of failure can not be identified as it is based on a linear innovation approach. In that case the failure arguments used by Arrows would have to be extended. The appropriate remedy from a neoclassical point of view could be to establish a working market mechanisms by means of financial incentives to firms (e.g. by tax grants) to use private consultancy services that find appropriate partners at universities. Having in mind evolutionary and institutional economic approaches to Innovation Systems, however, there might be other ways to eliminate the coordination problem between science and industry. The state can also support the creation of public-private partnerships, e.g. by financing the contribution of universities to shared industry oriented projects as well as the coordination costs of a network of private and public partners. This is, for instance, the case of the successful Austrian competence centres, with several partners from industry and research organisations working under a common privately organised enterprise. The state can also support by industry oriented research programmes or technology transfer centres etc. Both of the above cases address problem of knowledge-market failure. In the first case this is done by using instruments that support the creation of a market for knowledge. In the latter case, the Innovation System approach provides a heuristic that helps to find other effective solutions to the problem, by using networks as coordination mechanism to improve the failure.

7 First of all, one should distinguish between the kind of coordination type of problem (markets as opposed to hierarchies and networks) and the theoretical reasoning helping to identify the cause of the failure and the theory based strategy to solve the problem, which is either based on a linear model of the innovation process or its based on a cumulative, co-evolutionalry model of innovation activity as suggested by evolutionary and institutional economic theorising. The doctrine for knowledge-market failure when based on the linear model is very simple as it is based on a weakly differentiated concept of knowledge. Knowledge is seen as equal to information and the market mechanisms is understood to represent the dominating co-ordination mechanism which can only be left aside when no market mechanism can be established.. Given the linear model behind the original market failure argument which sees technology as an external factor falling from heaven as long as markets work as coordination mechanism, we get not much of an explanation for the failure argument. More so, we can not derive remedy to overcome the problem. In other words, there is no neoclassical macroeconomic theory based analysis helping to identify the original cause (problem in science industry relation) which is identified by systems thinking. Here we argue that the problem of science industry relation can be analysed as a problem of a coevolutionary process which is non-linear. This is an argument of evolutionary economics. When policy makers are looking for structural remedies the strategy adviser will look for solutions in his toolbox. Let s take a policy adviser with sophisticated tools to establish market coordination mechanisms with only weak expertise in using coordination mechanism provided by networks or hierarchies; he/she will go for an instrumental design supporting market coordination. Another policy adviser with an innovation systems expertise might go for other designs including networks and hierarchies to provide coordination. Assuming that perfect markets for innovative products exist, the neoclassical theory predicts efficient use of resources and even effective use in terms of a flow of incremental innovations developed for the market. If the market-failure doctrine includes that policy intervention requires some prove of efficiency in the outcome, other solutions than those related to market coordination will be in trouble as there are no established ways of measuring network efficiency. Measuring efficiency is well elaborated in economics as regards market coordination and can help to legitimise policy to establish markets which can then be evaluated, at least ex post. Hence under such a doctrine it will be hard to legitimise other than measures than those supporting market solutions. In several cases however efficiency will not be the most important criteria to legitimise policy intervention. Demand-side arguments that are mainly related to increasing the competitiveness of the European economies are not so much related towards efficiency of the innovation process but towards the effectiveness in terms of products that can be sold on the global market. Another demandside argument is the need to find solutions for societal challenges (environmental problems, defence, and health threats). Here the urgency to come to a solution is more important than the efficiency within the innovation system that produces the innovation as a remedy to the problem. Research on AIDS, mad-cow-decease, or environmental pollution problems are examples for an urgency problem that requires most importantly the effective use of knowledge to come to innovations that help solving the problem in time. This argument becomes very prominent today with the discussion about the short time that remains to keep the global temperature under a certain threshold. In this case the IPCC, the Stern Report and others warn of the short time span that remains to keep global temperature at a level that does not cause unpredictable change of global weather, water levels etc. causing uncertainties for the global economy and for humankind.

8 It therefore has to be decided by policy makers, if policy intervention in the field of RTDI follows a doctrine that asks primarily for economic efficiency criteria or if effectiveness has the first priority. We would argue that in those fields of demand-side arguments and especially in case of societal challenges with time constraints, the doctrine must primarily ask for effectiveness. Preliminary conclusion How to come to a new doctrine? Based on the concept of Innovation Systems, the OECD (2002) and others ask for a complementary approach between market failure and systems failure as a rationale for policy intervention. Market failure arguments provide reasoning why markets do not provide the most efficient way of coordination. Systems failure is related to the generic working of the innovation system in different ways (failure of provision of infrastructure, transition failure, lock-in failures and institutional failures) (OECD 2002) and hence addresses the performance and effectiveness of the system. Also Chaminade and Edquist (2006) argue for a kind of complementary approach to take into account both effectiveness and efficiency. Only if firms and market mechanisms fail to achieve the objective that public policy wants to address (i.e., systemic problem 1 ), and only if public policy has the ability to solve or mitigate the problem identified, they see an opportunity for public policy involvement. This is a wide approach in terms of services to be provided by the innovation system. They do not merely see the innovation system as a system to improve innovation performance (mostly by incremental innovations), but also as a system which is meant to solve societal problems providing more radical innovations. The underlying assumption behind this approach is that by providing innovative products and services on a market, policy makers expect solutions for societal challenges. This would include not only the field of RTDI-policy but also other policy fields such as defence policy, environmental policy, and health policy. The argument is based on a dichotomy between market and hierarchy, taking into account the existence of market failure and government failure. The argument of the systemic problem legitimises government intervention. The complementarity of the approaches lies in focussing on efficiency in the case of market failure arguments and the one on effectiveness by the systemic problems. This also makes clear the contradictions within the approaches. While the effectiveness criteria used by neo-classical economist, promotes strict IPR regulations as remedy by encouraging enterprises to bring innovations to the market, evolutionary economists argue for looser IPR regulations in cases where a diversity of innovations is necessary for the best one to be selected by the market. This is rather an effectiveness argument and contradicts the policy recommendation by a neoclassical economist. The contradiction lies in the theoretical underpinning of the policy argument that leads to the conclusion and can not strictly be related to the failure argument itself. That a knowledge-market mechanism is not the best coordination mechanism can lead to different conclusions, depending on the aim to be achieved, which is either related to efficiency or to effectiveness. In both cases knowledge-markets (whatever that may be) could be useful to overcome the problem identified. Therefore we further argue that knowledge-market failure can be seen as overlapping with the system failure argument, especially with respect to institutional aspects and learning. In general, systems failure can be understood as the general lack of performance of an innovation system in delivering and diffusing the required quality and quantity of innovations to meet societal needs. This means that systems failure is related to the core societal function of an IS, as defined by Edquist. As a consequence, the systems failure argument, on the one hand, mirrors the product-market failure argument, which is also based on the problem that firms do not provide innovations. The main difference between systems failure and product market failures is thus a historical one: the market failure argument is related to a linear approach to innovation processes, whereas the systems failure argu- 1 The authors prefer this term to systems failure as they see the legitimising argument based on effectiveness and therefore avoid the term failure which they relate to the effectiveness argument.

9 ment takes into account the complexity of the innovation process and concludes that a system fails as a whole rather than due to the lack of basic research results and innovation-averse firms having failed to provide the innovation. Obviously, the consequences to be drawn from the two arguments can be quite different. Knowledge-market failure on the other hand, is related to certain types of malfunctions within the innovation system that are rooted in the failure of coordination mechanisms on knowledge-markets. However, this does not mean that the entire system fails to provide innovations, as long as the failure of knowledge-markets is compensated other coordination mechanisms, be it self-organisation in networks or coordination by government (e.g. by insisting on the public good character of some piece of knowledge which allows also other innovators to use it and develop new products this is what is actually done when patents loose their protective effect after certain time this then helps to diffuse an innovation. The triangular concept of markets, networks and hierarchies takes into account three specific coordination mechanisms provided by different structural designs: market, hierarchy, and networks: Market coordination mechanisms can provide incentives to sell knowledge (IPR helps to establish such markets) Government can provide formal institutions (property rights for knowledge, standardisation to provide the cascadic structure for technology making innovation affordable, organisations that foster self organisation of networks, etc.) through governance mechanisms Coordination by networks can increase trust for the non-market exchange of knowledge, and it can provide standardisation for the actors in the network by means of self organising institution building In this interpretation, knowledge-market failure stands at the same level as network failure and government failure. These are all specific forms of coordination failure within the innovation system. The arguments of legitimising policy intervention may also be overlapping in some cases (e.g. cluster policy), but they represent the extremes of mechanisms which are widely used in innovation systems. Product-market failure is much closer to the notion that an innovation system is not able to deliver to society and economy what is needed (generalised systems failure). This expanded and integrated set of market/systems failure arguments provides additional rationales for policy intervention, even if in the case of governance failure the intervention may actually consist of reducing the level of policy actions. Figure: Triangular characteristic of coordination failure Network failure Market failure Government failure

10 We have elaborated on this argument in order to question the prevailing doctrine of knowledgemarket failure and to conceptualise it just as an aspect of coordination failure. This is based on the intention to support the development of a doctrine which both considers effectiveness principles (in response to socio-economic needs) and efficiency principles (in response to systemic coordination problems). We think that the issues of orientation towards socio-economic problems and of specific demand for innovation, be it in terms of innovation that shall lead to more economic growth or innovation to serve other societal challenges asks for another way of looking at the debate. The failure argument legitimizes public policy action only, if a failure or deficiency can acturally be proven, but it does not allow anticipating possible deficiencies and opportunities to come. A different and more pro-active way to approach the question of legitimisation starts from the functions of IS. We are optimistic, that this might open up new ways of legitimising necessary changes in the IS, including public policy intervention. Functional approaches to IS Innovation System approaches have been discussed for two decades. Starting with (Freeman 1987) and (Lundvall 1988) on national systems of innovation to sectoral systems of innovation (Breschi and Malerba 1997) and Regional Innovation Systems (Tödtling and Kaufmann 2001; Cooke 1992) most emphasis was given to look into the composition of actors, their interactions and the institutional aspects of innovation activity. The innovation systems approach as the dominant approach in innovation research while being useful for descriptive purposes and as a heuristic device has been rather weak when it comes to providing normative guidance for stimulating innovation systems. Over the last years some attempts have been made to develop the innovation system approach further in a way that allows for such guidance. In general terms the overall function of the IS for the economy is to provide innovations, to diffuse and to use them (Edquist 2004). Its focus lies in the role of the IS for the economic system. Since the late 1990ies some concepts were developed, which analyze the functions of innovation systems in more detail extending the innovation system approach (Hekkert et al. 2006; Kubeczko et al. 2006; Bergek et al. 2005; Chaminade and Edquist 2005; Borrás 2004; Jacobsson and Bergek 2004; Smits and Kuhlmann 2004; Johnson and Jacobsson 2003; Liu and White 2001; Rickne 2000; Galli and Teubal 1997; David and Foray 1995; David and Foray 1994). The value of the different approaches lies in their role as heuristics for the analysis of innovation systems in addition to the analysis of actors, their interaction and the institutions relevant for innovation activities. They are approaches for the analysis and identification of system failure. As describes by Chaminade and Edquist s (2005), the approaches are developed to serve different objectives, so the identified sets of functions of innovation systems have different focuses. Some approaches focus on the services provided by the innovation system as a whole that are supporting the innovation activity of actors in a value chain. Here, enterprises (private firm or enterprise co-operation) are in the centre of the anylysis. Others put the emphasis on the process of knowledge production, in particular on intermediaries between industry and science; another group of approaches emphasises the role of organizations and their achievement within the innovation system (see Table 1). The different perspectives synthesised permit a diverse picture on the role of innovation systems and can be applied to different types of innovation systems. For sectoral innovation systems Bergek et al. (2005) suggested a framework of analysis which is based on a pragmatic structuring of the functions described by various authors. Similar to that, Kubeczko et al. (2006) suggested a functional framework based on empirical evidence from an innovation survey (Rametsteiner

11 and Kubeczko 2003) that allows for an analysis of sectoral and regional innovations systems at the same time. Table 1: Focus of different functional approaches describing innovation systems Topic areas addressed Value chain, firm centred innovation activity Intermediary actors, Scienceindustry relations Actors role and impact in IS Role of policy actors Research focus on Innovation process (Liu and White 2001, Edquist 2004, Kubeczko and Rametsteiner 2006) Knowledge production process (Johnson and Jacobsson 2003) (Bergek et al. 2005) Organisational performance (Borràs 2004) Technology and innovation policy (Smits and Kuhlmann 2004, OECD 2002) To give an overview of the different approaches, without being able to go into details, we cite those of five papers: Chaminade and Edquist (2005) list four sub-functions : 1. Provision of knowledge inputs to the innovation process 2. Demand side factors 3. Provision of constituents 4. Provision of support services for innovating firms Johnson and Jacobsson s (2003) synthesis of functions of SIS as framework for policy: 1. Supply incentives for companies 2. Supply resources (capital, knowledge) 3. Guide direction of search 4. Recognise the potential for growth 5. Facilitate the exchange of information (and knowledge) 6. Stimulate and create markets 7. Reduce social uncertainty (on strategic behaviour) 8. Provide legitimacy for innovation

12 Borras (2004) functions of innovation systems in the economy and in the IS 1. Competence building and generation of incentives a. Production of knowledge b. Diffusion of knowledge c. Financial innovation d. Alignment of actors e. Guidance of innovators 2. Generation of incentives and reduction of uncertainty a. Appropriation of knowledge (IPR) b. Reduction of technological diversity (standards) 3. Establishment of limits and reduction of uncertainty a. Reduction of risk (incertitude) b. Control of knowledge usage Value Chain oriented functional approach by Kubeczko et al. (2006) 1. Provision of Resources 2. Management of the Complexity of the Innovation Process 3. Promotion of the Use of the Innovation Smits and Kuhlmann (2004) identified five functions: 1. management of interfaces 2. (de-) construction and organising (innovation) systems 3. providing a platform for learning and experimenting 4. providing an infrastructure for strategic intelligence 5. stimulating demand articulation, strategy and vision development. Although, the approaches have been developed addressing different objectives, many of the functions are overlapping and they can be useful to discuss legitimisation of public policy intervention. Still there is a lack of consistency in all those approaches.

13 Complex Research and Innovation System Approach and its consequences for legitimizing RTI-policy intervention Complex research and innovation systems As stated earlier, functional approaches put the emphasis on the dynamics of the innovation system which is meant to provide innovation and to diffuse it (Edquist 2004). Moreover, the functional approach is well suited to serve as a normative frame of reference for devising policy strategies. However, the theoretical underpinning tends to remain vague as it was developed as a synthesis of existing innovation systems approaches. They certainly take the evolutionary and institutional aspects of the innovation process into account, but mostly represent an amalgamation and synthesis of the insights of previous conceptual elements and empirical findings on innovation systems. Functional innovation system approaches are increasingly used to legitimise policy intervention in innovation systems. This is actually what makes them particularly interesting from a policy perspective because lines of argumentation can be built that avoid the negative connotations associated with the notion of failures. The approach of complex innovation systems (CIS) outlined subsequently is part of this functional tradition, but builds explicitly on a theoretical and conceptual framework that integrates elements of the lines of thought from innovation systems, science and technology studies, evolutionary theory and complexity research (Weber 2005). It aims to guide policy-oriented, applied research on innovation systems by providing on the one hand a heuristic for investigating concrete innovation phenomena and on the other a set of key functions that allow justifying and specifying policy interventions in innovation systems. A systems approach is regarded as a particularly useful basis for formulating such an integrative framework. The reasoning behind this is threefold. First of all, the systems language has already been used in several approaches to the study of innovation, mainly along the lines of innovation systems approaches and thus facilitates compatibility with earlier work of Edquist, Lundvall, Malerba, and others. Second, a systems language does not presuppose per se any specific theoretical perspective, but allows using different theoretical building blocks to establish the relationships and mutual influences between the constituent elements of the system. As such, a systems language can serve as a platform for integrating building blocks from different theoretical approaches to research and innovation and thus for inter-connecting different streams of research work. Third, the perspectives for dealing with innovation dynamics (i.e. in particular evolutionary and complexity concepts) are usually formulated in a systems language as well, and can thus be integrated into the CRIS approach quite easily. The underlying theoretical approach is formulated in a sufficiently neutral way to be applicable to national and regional as well as to sectoral or technological examples of innovation system. A complex research and innovation system is defined as an open, multi-level socio-technical system, embedded in a wider social context or system environment. In order to structure a complex research and innovation system internally, three constituent levels are distinguished: (i) the structural level of the innovation regime (i.e. structural characteristics, institutions and paradigms); (ii) the behavioural level of actors, interactions and networks; and (iii) the societal knowledge base in terms of knowledge, concepts and visions regarding the innovation domain in question.

14 The dynamic behaviour of complex innovation systems is based on the interactions within this threelevel model. The dynamics of complex innovation systems are driven by the interactions between and within the three levels. These interactions give rise to complex phenomena, such as selfreinforcing mechanisms, path-dependencies, etc.. The aim of this highly structured way of dealing with dynamism is to understand and describe major transformations of socio-technical systems. In order to capture the complex dynamics, it is necessary to look into the interplay between these levels. Any attempt at political intervention requires a deeper understanding of the possibilities and limitations that these dynamics in complex systems provide. This requires to trace back transformation processes of innovation systems to structural changes at the meso-level as well as to the behaviour of actors, their decisions, perceptions and interactions at the micro-level and as suggested by the three-level model above to changes at the level of the societal knowledge base. The linkages within as well as between these levels need to be understood in terms of feedbacks or circular causalities (Küppers 2001) as the characteristic features of complex innovation systems. The circular causalities are at the origin of self-reinforcing (or self-delaying, stabilising) mechanisms that drive system change, and help explain why and how collective choices and pathways in innovation systems take place. In other words, self-organisation is used as a key concept to capture the mechanisms that underlie the evolutionary, but sometimes also revolutionary processes of system change that can be observed at macroscopic level. In the CIS approach, we distinguish three key processes that capture the system dynamics and that can be regarded as co-evolving in a complex innovation system. They correspond roughly to the three levels of the complex innovation systems framework: 2 The transformation of the structure of the innovation system The formation of actors opinions and decisions at the behavioural level The formation of innovation hypotheses 3 at the level of societal knowledge base From reactive to proactive policies Policy interventions to transform the structure of innovation systems have been taking place since their emergence, which can be traced back even to the nineteenth century. It at least goes back to the foundation of university faculties for basic research and technical universities and the applied research institutes. Over the last two decades innovation systems research focussed on scienceindustry links promoting competence centres, clusters and networks to foster the behaviour of actors in respect to intensify interaction and the formation of networks as an increasingly important coordination mechanism. In essence, most of these types of interventions followed a reactive approach of correcting failures that have been observed. However, with the acceleration of the pace of technological change and the growing internationalisation of research and innovation, government policy is in need of a more strategic perspective for underpinning its interventions in research and innovation. The key point is not any more to improve the internal efficiency of research and innovation systems (although it continues to be an issue!), but to 2 Further details on the operation on the dynamics driving these three processes can be found in Weber (2005). 3 Weber 2005 Both problem pressure and opportunities can trigger search procedures for new, innovative solutions by actors. Problem pressure may emerge as a result of external developments in the system environment (e.g. societal concerns about environmental degradation, shortage of resources, etc.), but also as a result of perceived inefficiencies within the system (e.g. bad performance indicators). In terms of opportunities, one may consider also new scientific findings as factors standing behind the search for new solutions.

15 anticipate on future developments outside the national, regional or sectoral contexts and to devise strategies for dealing with them. Foresight has developed in many countries into an important tool providing inputs to such a more strategic orientation of research and innovation policy. This is still a relatively new policy instrument which takes into account the self organising mechanism within the innovation systems, designed as a kind of open method of coordination, with the intention to foster shares formation of innovation hypotheses and visions about innovation paths. They can hence contribute to the formation of actors opinions and decisions. While initially focusing mainly on technology foresight, socio-economic consideration and structural features of innovation systems have increasingly moved into the focus of most recent foresight initiatives. The functional innovation systems approach has in principle the potential to deal in a more strategic and forward-looking manner with the changing requirements of research and innovation systems, but the approaches presented so far concentrate mainly on their structural features and coordination mechanisms. In other words, the functions addressed allow tackling those aspects that can also be sub-sumed under the notion of coordination failures. Functions of Complex Research and Innovation Systems The empirical evidence of many European countries shows that most of these policy interventions already take place and support structures of an innovation system exist (research units, intermediary agencies, clusters, networks supporting innovative firms). They are most frequently legitimised by market failure and systems failure arguments. However failure arguments do not provide strategic guidance (Where to look at and what to do?). We argue that policy makers must be aware not only of the internal dynamics of the innovation system and the innovation activity in order to be able to use their resources effectively (internal function) and efficiently, but also of the overall function of the research and innovation system for other social systems, in order to get orientation with respect to the direction of change that the system should take.. The overall function of an innovation system in the wider societal context is to pursue innovation processes, i.e. to develop, diffuse and use innovations (Edquist 2004). As this function is provided to society as a whole, we want to call this the societal function of an innovation system. Society can make use of the innovation when it is on the market for products and services. Within the CIS approach (being a heuristic to describe both the functioning and the function of specific a societal subsystem) we get a slightly broader understanding of the function of research and innovation systems, namely that it is supposed to provide services to other sub-systems (economy, politics, civil-society, etc.), in response to the requirements raised by these. This can be conceptualised as the societal function or service provided to other social systems by the innovation system. This interpretation implies that innovation systems need to be assessed with respect to the amount of innovations they produce, but also with respect of the types and actual content of these innovations, i.e. whether they meet the requirements of other social systems (Chaminade and Edquist 2006). Similarly, the political system, as societal subsystem can provide services to the innovation system, e.g. it can institutionalise regulations, provide financial incentives or infrastructure. As this can not be directly translated into innovation activity of a single firm (embedded in an innovation system), we have to take into account the generic internal functioning of the innovation system that should be supportive of the societal function of the innovation system. The CIS approach thus suggests distinguishing three types of generic functions of innovation systems: Structuring: Innovation systems shall provide structures for innovation activities and support activities of the IS itself. This can be achieved by the introduction of actors, institutions (rules, norms, etc.), networks or artefacts (infrastructure). Orientation: Orientation can be given by means of Leitbilder, visions or other open methods of coordination, or more concrete as information flows or financial incentives.