Working Papers in Technology Governance and Economic Dynamics no. 66 the other canon foundation, Norway Tallinn University of Technology, Tallinn Ragnar Nurkse School of Innovation and Governance CONTACT: Rainer Kattel, [email protected]; Wolfgang Drechsler, [email protected]; Erik S. Reinert, [email protected] How to Organize for Innovation: Entrepreneurial State and Organizational Variety Erkki Karo & Rainer Kattel Nurkse School, Tallinn University of Technology Contact: [email protected] January 2016 Abstract Research on how to effectively organize innovation policy has seemingly come to a consensus that decentralized, often peripheral, flexible and specialized organizations are superior to monolithic and hierarchical bureaucracies. These agencies are expected to both support innovation in markets through effective implementation of government interventions and, if necessary for the former task, also innovations in public policies, institutions and organizations. Yet, it is also recognized that policy inno- vations and effective implementation of policies may require organiza- tions with different routines and capacities. We propose that we might gain a more systematic understanding of this governance challenge by focusing on the organizational morphology of innovation policy design and implementation and its evolutionary variety, or ability to sustain dif- ferent policy and administrative capacities deemed necessary for effec- tive innovation policy. We propose an analytical taxonomy that system- izes the diversity of organizational routines and capacities using the organizational configurations proposed by Henry Mintzberg. We illustrate through three critical cases that effective innovation policies are based on a variety of organizations with different routines and capacities. We also discuss the policy implications and avenues for further research. Introduction In recent years, many prominent innovation scholars (Mazzucato 2013; Block and Keller 2011; Fagerberg et al. 2013; Zysman and Breznitz 2012; Weiss 2014) have argued that if technological development and innova- tion is to significantly contribute to solving current economic and societal challenges, the thinking on the role of the government in innovation should be extend beyond the narrow perspective of governments mostly fixing market failures. The so called entrepreneurial, or modern mission- oriented state should build policies and institutions that proactively take on the uncertainties of technological development and innovation (Maz- zucato 2013). This argument can also be taken as a self-criticism of the innovation policy scholarship and practice. Freeman (1987) argued already in the early years of the systems of innovation thinking that radi- cal technological innovations and changes in technology systems and techno-economic paradigms (see also Perez 2002) require complemen- tary institutional and social innovations. Thus, the role of the state is to both support innovation in government policies, institutions, organiza- tions (this is now labelled as public sector innovation) and innovations in markets through government actions (this is what we usually label as innovation policy). 2 Yet, our knowledge of how to do this seems to be somewhat limited. Empirically, the Asian developmental state (Johnson 1982; Evans 1998) and the Western mission-oriented innovation policies of the Cold War era (Mowery et al. 2010; Foray et al. 2012) are probably closest to the cur- rent entrepreneurial state thinking, even if current the techno-economic context (in terms of policies, institutions, markets and diffusion path- ways) may be more complex (Karo and Lember 2016; Wong 2011; Yeung 2013). Thus, in her recent paper, Mazzucato (2014: 8) argues that one of the crucial questions for the innovation research is to understand the internal workings of what we call in this paper ‘innovation bureaucra- cies’, i.e. ‘how should public organisations be structured so they accom- modate the risk-taking and explorative capacity, and the capabilities needed to envision and manage contemporary challenges?’ Further, ‘key concern should be to establish which skills/resources, capabilities and structures are useful to increase the chances that organizations will be effective both in learning and establishing symbiotic partnership with the private sector – and ultimately succeed in implementing mission-oriented and transformative policies’ (Mazzucato 2014: 17). These questions are also highly topical in development policy research (Easterly 2014; Reinert 2007) and public sector innovation research (see de Vries et al. 2015; Kattel 2015). In this paper, we propose that these questions are best answered by an analytical framework with an explicit organizational focus – looking at innovation bureaucracies through the lens of organizations and organiza- tional morphologies (or, systems of organizations) – as opposed to indi- vidual or institutional level analyses. This is one of the core premises of the evolutionary economics research on technological change and innova- tion (see Nelson and Winter 1982). We show that from this perspective, we can find two almost juxtaposing views on innovations bureaucracies, associated with two great social scientists: central meritocratic and hier- archical expert organizations (labeled as Weberian bureaucracies) deliver innovations versus small, agile and often peripheral or decentralized orga- nizations (labeled as Schumpeterian organizations) do a better job at innovations than others. The debate on the role of the state in innovation often gets stuck just at this juncture searching for the definitive answer to the question: should we still stick to modernizing Weberian meritocra- cies, or move radically towards experimental, start-up like governments? We aim to show that innovative bureaucracies are much more complex phenomena and require a more elaborate framework and thinking. On the theoretical level, we show that the arguments in favor of Weberian vs Schumpeterian agencies are in fact not mutually exclusive, but highlight the complexity of how government organizations need to be structured 3 and organized to support innovations both in government (policies, insti- tutions, organizations) and also through government (innovation) policies in firms and industries (to achieve desired policy impact, e.g. tackling societal challenges and increasing productivity and economic growth). In other words, well performing innovation bureaucracies consist of a variety of organizational configurations (see also Mintzberg 1989) and capacities that foster both constant search for better policy ideas and practices and implementation of these ideas and practices in an effective manner and with a desired systemic impact. In section 1 we briefly review the existing literature on how to organize the design and implementation of innovation policies – here, we under- stand innovation policy in the widest possible sense including all public policies that consciously aim to promote innovations and technological change (see also Lundvall 2013). In section 2 we provide our analytical taxonomy of organizational configurations and elaborate the concept ‘organizational variety’. In section 3 we apply this taxonomy to stylized case studies (based on literature review of three ‘critical cases’) of how governments have organized innovation policy in different contexts. In the concluding section we provide guidelines for future research and policy. 1. Classic and modern debates on innovative bureaucracies 1.1. The classics As with most social science research, we can dissect research on innova- tion bureaucracies into three inter-related levels of research – individual, organizational and institutional – pursued by sociologists, economists, organizational theorist and others. Most of the questions on how to orga- nize innovation bureaucracies posed by Mazzucato (2014) were among the core topics also in the works by Weber (1922), Schumpeter (from 1912 to 1942), Merton (1940), Hayek (1945), Simon (1952). These scholars were among the first to provide modern systematic scientific inquiries into such questions as what are the potential trade-offs between organizations/bureaucracies as the key characteristics of modern societ- ies (see also Cohen 1970) and individual, interests, motivations, wants?; and how organizations and societies in general maintain dynamism, change, and innovation? Weber and Schumpeter were among the first trying to build comprehen- sive evolutionary/dynamic perspectives on these questions (see e.g., MacDonald 1965 and Ingham 2003 for comparisons of their perspec- tives), but their thinking is often simplified into opposite views: Weber’s work is related with hierarchical organizations and bureaucracies and 4 Schumpeter’s with individual entrepreneurship. Of course, Weber’s ideal- types and broader analysis of economy and society recognized both ‘char- ismatic’ and ‘rational’ (and ‘traditional’) forms of authority underpinning the organization of social life (Weber 1922) and Schumpeter moved from emphasizing entrepreneurs (and their personal qualities) to organizations (firms) as crucial sources of innovation (see Schumpeter 1912, 1942). Witt (2002) argues that this shift in Schumpeter’s work partly shows that his approach lacked the necessary traits of an evolutionary theory, espe- cially the ‘self-transformation explaining’ aspects, or ‘endogenous drivers’ of change of the system in focus.1 Somewhat similar criticism is also raised regarding Weber’s work (see MacDonald 1965). Indeed, both Weber and Schumpeter gave most emphasis to exogenous factors – unique individuals, both charismatic entrepreneurs and financers; and specific cultural-religious aspects – as drivers, or triggers of change. Per- haps the key common feature in the work of Weber and Schumpeter is the understanding that conflict between incumbent and new (political, business, etc) ideas always also takes organizational shape. Later evolutionary attempts have tried to include also technology as important variable in itself into the analysis noting co-evolutionary ties between technological, organizational and institutional developments (see also Nelson 1994). Litwak and Figueira (1968: 468) were among the first to explicitly link the debates on the trade-offs between individual vs bureaucratic ways of organizing social life and specific impact of tech- nologies and technological development: Bureaucratic structures are ideally suited to deal with problems requiring technical knowledge or large-scale capital investments. Primary group structures are most able to handle problems requiring little technical knowledge, for example, where knowledge is so simple the ordinary person can do it as well as the expert, where knowledge is lacking so experts cannot be trained, where knowledge is so complex it cannot be put together in time to make a decision. In principle, technology is as likely to take tasks now handled by experts and simplify them so the ordinary person can deal with them as it is to take tasks now handled by ordinary individuals and show how they can be more effectively handled by experts. Therefore, in principle, technology is not likely, after its first stage, to reduce functions of either the primary group or the bureaucracy. More characteristic will be stress on continuous change. 1 According to Witt (2002: 10) an evolutionary theory in whatever field is a) dynamic, b) his- torical (deals with historical processes that are irrevocable and path dependent), and c) self- transformation explaining (includes hypotheses relating to the source and driving force of the self-transformation of the system, be it be it firm, industry, government). 5 In this context, and departing from the understanding that innovation does not equate invention or just any kind of change, but denotes a more complex process of ‘successful’ applications of new technical or social knowledge (or application of existing knowledge in new context) (see OECD 2005), evolutionary economics and innovation research has taken an explicit organizational focus. The crucial contribution is Nelson and Winter’s (1982) work on evolutionary economic theory and their focus on organizational ‘routines’ as the key factors explaining firm and industry performance (see also Becker 2008; Nelson and Nelson 2002). As suc- cinctly defined by Hodgson (2008: 23): ‘Routines are not behavior; they are stored behavioural capacities or capabilities. These capacities involve knowledge and memory. They involve organizational structures and indi- vidual habits, when triggered, lead to sequential behaviours’. The com- plexities and uncertainties of innovation require organizations to ‘store’ (routinize) existing (tacit) knowledge and mediate between institutional and individual level drivers (environmental feedback, individual motiva- tions and wants) that influence (drive and constrain) innovation process- es. Thus, modern private sector innovation research focuses to large extent on organizational performance (capabilities) and implementation of strategies (e.g., how to keep creativity in an organization? what products and business strategies are feasible in specific institutional contexts?; see Lam 2006 for an overview). 1.2. Modern debates on organizing innovation bureaucracy Nelson and Winter also argued that ‘If one views policy making as a con- tinuing process, the organizational and institutional structures involved become critical. Public policies and programs, like private activities, are embedded in and carried out by organizations. And, in a basic sense, it is the organizations that learn, and adapt. The design of a good policy is, to a considerable extent, the design of an organizational structure capable of learning and of adjusting behavior in response to what is learned’ (Nel- son and Winter 1982: 384). Still, innovation policy (but also development policy and public sector innovation) research has done much less work on this level of analysis. Instead, one can find ample research on institu- tional level (what are the best institutions or governance systems for supporting technological and institutional innovation and economic devel- opment in general? – see Fagerberg et al. 2013 as the most recent recap of the system of innovation research and its contributions; Easterly 2014 and Reinert 2007 on development policies more broadly) and individual level (who are public sector entrepreneurs, how do they emerge and sur- vive in bureaucracies and how they support innovation in policies and markets?; see Leyden and Link 2015). As a result, the analysis of orga- nizational capabilities has been substituted mostly with references to 6 neo-institutional approaches to state policy and administrative capacities treating institutions mostly as constraints and not as enables, or drivers of innovation (see Nelson and Nelson 2002; Karo and Kattel 2014). One crucial exemption here is the debate started by the East Asian devel- opmental state scholars in the 1980s (Johnson 1982, Amsden 1989, Evans 1995 and 1998, Haggard 1990, 2004, Wade, 1990). According to Evans (1998; also Haggard 2004), a common assumption across dif- ferent theoretical focuses (on ‘market-friendly’ policy rationales, ‘indus- trial policy’ rationales, or ‘profit-investment nexus’) explaining East Asian development since the 1960s was that ‘highly capable, coherent eco- nomic bureaucracy, closely connected to, but still independent of the business community, has been essential institutional prerequisite for suc- cessful innovation policy’ (Evans 1998: 66).2 While in these different models governments follow diverse policies with different degrees of intervention and economic bureaucracies have diverse definitional and task-related borders (from generic regulation to sector specific targeting, finance and regulation), the assumptions of policy and administrative capacity have been rather uniform. Namely, it has been assumed that whatever the policy and institutional variety between specific economies, bureaucratic capacities can be best developed and best talent recruited and motivated via Weberian (in the sense of rational authority) means of meritocratic recruitment and career management to make working for government either financially competitive to, or culturally even more rewarding/prestigious than, working in the private sector. Quantitative studies have sought to solidify this position (see Evans and Rauch 1999; Rauch and Evans 2000; more recently Nistotskaya and Cin- golani 2014), but they test the importance of some more easily measur- able Weberian elements (merit-based recruitment and career systems) on system level without explicitly looking at and into innovation bureaucra- cies. Furthermore, also the qualitative studies following the pilot study of Johnson (1982) who studied the organization of Japanese MITI in great detail (structure, recruitment strategies, evolution of tasks etc) have taken a more institutional perspective (assuming the existence of general Weberian structures also in specific policy domains and organizations). 2 This view is still most elaborately captured by Chalmers Johnson’s concept of the develop- mental state and research on Japan: a country with predominant policy orientation towards development supported by a small and comparatively inexpensive elite bureaucracy centered around a pilot organization, such as MITI, with sufficient autonomy (limited intervention by the legislative and judiciary) to identify and choose the best industries and technologies to be devel- oped as well as the best-fitting policy instruments (from administrative guidance to control over finance and regulation of competition) while still maintaining market-conforming methods of state intervention, and public-private cooperation in state-business relations (Johnson 1982, 305-320). 7 Only since the late 1990s have some studies tried to replicate the original claims of the broader developmental state research (see Cheng et al. 1998; Kang 2002a; 2002b). They have revised the original claims and related assumptions and highlighted how the Weberian elements have in fact varied and also from time to time been overlooked in different coun- tries. Thus, a more revised approach on how to create bureaucracies supportive of innovation emphasizes developing minimally some islands or pockets of excellence in government – as insulated agencies – that can design and implement policies supportive of complex tasks of innovation and development (see again Evans 1998). This thinking has permeated much of the innovation and development policy research from Latin America (Schneider 1992) to Eastern Europe (Suurna and Kattel 2010) and beyond (OECD 2005; Edquist and Hommen 2008). At the same time, Western insights from the so-called neo-developmental state research (O’Riain 2004; Block 2008) and mission-oriented innova- tion policies (Mowery et al. 2010; Foray et al. 2012) and also more recent studies of East Asian Tigers trying acting closer to the techno- economic frontier (in ICT and biotech) (Yeung 2013; Wong 2011; Zys- man and Breznitz 2012) have provided somewhat different interpreta- tions questioning the validity of the developmental state thinking. For managing the uncertainties of innovation and development at the techno- economic, institutional set-ups of innovation policies need to find a bal- ance between centralized planning of priorities and policy goals, and decentralized management and implementation of specific programs and measure to allow sufficient flexibility and space for learning given the uncertainty of technological development and diffusion trajectories (espe- cially given the competing pull factors of different global value chains). Further, Breznitz and Ornston (2013), who analyze the evolution of the Israeli and Finnish innovation policies, argue that peripheral Schumpeteri- an agencies may be the sources of such flexibility and learning, or for policy innovations necessary for promoting rapid innovation-based com- petition, given that these agencies have sufficient managerial capacities (or, slack). Arguably, the peripheral status (and little prestige and resourc- es) is important to reduce the likelihood of political interference and to allow space and to create organizational need for policy experimentation (and innovation), but also for new forms of public-private interactions (while avoiding capture by special interests), as these agencies are unable to tap into existing political, financial and institutional resources. This current critique of the earlier argument of the developmental state research tallies also somewhat with findings in public management research that autonomous agencies with large managerial autonomy com- bined with strict performance controls – in another words, new public 8 management style agencies emerging in the 1990s – generate a rather innovation-oriented culture (Wynen at al. 2013). Thus, also in the context of public sector innovations, we see a somewhat similar trend to move beyond the so called Weberian rational expert bureaucracies whereby orga- nizations tasked with innovating within public organizations or services (innovation or design labs, ilabs in short) tend to be established as at arm’s length institutions, with low budgets and political profiles, but with highly charismatic leaders, broad independence in agenda setting and with high level of experimentation (see Puttick et al. 2014, Tõnurist et al. 2015). In sum, while developmental state research can be described to have a ‘Weberian’ bias, then the more recent innovation research seems to move towards ‘Schumpeterian’ bias. Although, a more correct assessment would be that developmental state research has had a bias towards ‘ratio- nal’ bureaucratic organizations whereas more recent innovation policy research seems to have a ‘entrepreneurial’ or ‘charismatic’ bias. Still, both approaches show some evolutionary insights. The capacities for innovation of the developmental state emerge from the complicate sys- temic relations (often conceptualized as ‘embedded autonomy’ – Evans 1995) between the rational-Weberian type policy organization (with its specific organizational routines), its relations with political system (provid- ing strategic direction and autonomy for the bureaucracy) and business system (providing input and feedback to policies). Yet, already they early critics of the approach argued that this is a very fragile institutional set- ting prone to politicization, capture by business sector, or dominance of instrumental goals of the bureaucratic organizations. Breznitz and Ornston (2013) recognize that also the Schumpeterian organizations may easily (often due to their success) become more ‘central’ and politicized as politicians either get interested in them and try to capture and gain polit- ically from their legitimacy, or simply give them too many tasks. While building a more evolutionary theory of how different types of orga- nization in innovation bureaucracy emerge and evolve is a fruitful avenue for further research (see also Karo and Kattel 2014, 2015), this is not our goal in this paper. Suffice here only to mention that we assume that the driving forces of the evolution of innovation bureaucracies are often con- flictual (different expectations) leadings to punctuated positive feedback processes in public sector (see more in Karo and Kattel 2014, 2015). This explain why policy rhetoric tends to spread faster than policy practices (see Pollitt 2001), why the convergence and spread of political/institu- tional blueprints is often slower than that of technological (Kogut, 1991), and more generally why differences in social systems of innovation and production may persist even in times of globalizations and global value chains (Amable 2000). 9 In this paper, we focus on another weakness of the current debates. As seen above, these debates on innovation bureaucracies tends to narrow down from system level analysis (where evolutionary theories could be developed further) to analysis of single organizations: crucial innovation agencies are either rational in the Weberian sense or entrepreneurial/char- ismatic in the Schumpeterian/Weberian sense. Yet, if we look carefully at the empirical foundations of these arguments, we can notice that the abovementioned studies do not define what an innovation or developmen- tal agency (the core bureaucratic entity analyzed and used as explanatory factor) actually is. Johnson (1982) looked at a ministry; later analyses of South Korea and Taiwan have emphasized planning and policy coordina- tion boards (Cheng et al. 1998), often set up on purpose outside usual career system and examinations. Evans and Rauch’s study (1999) does not differentiate systematically between ministries, development boards and other government organizations (nor does it in fact contain any ques- tions about institutional or organizational structures and capacities). Neo- developmental state research has looked at a research funding agency (DARPA in the US – Block 2008) and at an industrial development agency (IDA in Ireland – O’Riain 2004). Breznitz and Ornston (2013) looked at a ministerial department, or office (Office of Chief Scientist in Israel) and a foundation supervised by a central bank and later by parlia- ment (Sitra in Finland). These organizations have highly diverse tasks and positions within the broader governance and innovation systems; they differ in structure, size, skill-sets etc. Thus, it seems that the selection of these agencies as cases to be analyzed (and used as crucial explanatory factors) is determined by their importance as change agents within spe- cific innovation bureaucracies and systems with specific bottlenecks and failures that these agents have helped to overcome (either in innovations within government or in markets). In other words, their definition and selection as crucial innovation agencies is determined by their contribu- tion to the system or overall policy performance. In essence, we have two different conjectures stemming from research in two different contexts: first, Schumpeterian-entrepreneurial character- istics of agencies (often with peripheral status) allow for internal experi- mentation and design of new policy approaches in the context of uncer- tainty of innovation and development (innovations in policy); second, Weberian-rational characteristics of agencies (often with central status) provide policy space and access to policy resources to actually implement innovation policies (also new innovative instruments and institutions) in a systemic way. Still, these streams of research do not provide an organi- zational level framework to study innovation bureaucracies as systems of organizations that exist in reality, their specific organizational routines and resulting capacities. There is scant theoretical and systemic empirical 10