Firms’ innovation benefiting from networking and institutional support: A global analysis of national and firm effects

Research Policy 45 (2016) 1233–1246

Contents lists available at ScienceDirect

Research Policy journal homepage: www.elsevier.com/locate/respol

Firms’ innovation benefiting from networking and institutional support: A global analysis of national and firm effects Thomas Schøtt ∗ , Kent Wickstrøm Jensen University of Southern Denmark, Department of Entrepreneurship and Relationship Management, Kolding, Denmark

a r t i c l e

i n f o

Article history: Received 7 July 2014 Received in revised form 8 March 2016 Accepted 8 March 2016 Keywords: Firms Networks Collaboration Innovation Institutions National innovation systems Multi-level

a b s t r a c t Firms’ networking for innovation is embedded in institutions of society, where national policies are increasingly designed to provide institutional support for firms’ networking and thereby benefit innovation. But, globally, what are the quantitative and qualitative effects of institutional support for networking and, in turn, for innovation? 68 countries with 18,880 firms were surveyed in the Global Entrepreneurship Monitor, enabling generalization to the firms in the countries around the world. Two-level modeling shows that firms’ networking benefits both process and product innovation. Institutional support does not significantly affect quantity of networking, but greatly enhances quality of networking in the sense that support for networking in a country enhances the benefits of networking for both process and product innovation. Contrasting low and high support for networking leads to estimating that institutional support for networking can increase the benefits of networking considerably for both process and product innovation. © 2016 Elsevier B.V. All rights reserved.

1. Introduction During the last two decades much research has considered enterprise networks as the “locus of innovation” (Powell et al., 1996; Ahuja, 2000). Research examines how firms in complex and dynamic business environments have shifted innovation towards inter-firm endeavors in which collaborative networks bridge complementary and increasingly specialized firm competencies and provide for fast and flexible responses to market demands and opportunities (Das and Teng, 1998; Ahuja et al., 2008). In accordance with this view, several reviews confirmed that innovation flourishes within inter-firm networks (Rogers, 2004; Powell and Grodahl, 2005). However, findings on the impact of firms’ collaborative networking on their innovation are ambiguous. Study results have disagreed on the impact of different types of networks; some not being significantly associated with firms’ innovativeness, and some network types even showing negative impacts on firms’ innovativeness (Nieto and Santamaría, 2007; Lhuillery and Pfister, 2009). Similarly, research on the firms’ innovation benefits from occupying different network positions have been inconclusive on the impacts of network size, centrality, cohesion, and other proper-

∗ Corresponding author. E-mail address: [email protected] (T. Schøtt). http://dx.doi.org/10.1016/j.respol.2016.03.006 0048-7333/© 2016 Elsevier B.V. All rights reserved.

ties of networks (Zheng, 2010; Rost, 2011). Such inconsistencies have led several researchers to suggest a contingency approach to study the impact on networks on innovation (Tsai, 2009; Rost, 2011; Zheng, 2010). Contingency studies of innovation networks have so far primarily been focused on moderating impacts from firm characteristics (Tsai, 2009; Zheng, 2010), and difference between industries (Rowley et al., 2000). However, only little empirical research has attended to the potential moderating impact from the institutional environment surrounding firms and their collaborative networks. Such scarcity is surprising given the large interest in examining cross-national differences in innovation, and given the interest of policy makers in designing framework conditions for facilitating inter-firm collaboration through geographical industrial clusters, incubator milieus, etc. play a significant role in boosting innovation within and between regional firms (Lundvall, 1992; Freeman, 2002; Autio et al., 2014). Knowledge of how the impact of networking on innovation is contingent on institutional structures would provide recommendations for the structuring of such framework conditions. The awareness that firms’ networking for innovation is potentially contingent on the surrounding institutional environment has previously been advocated by Owen-Smith and Powell (2004). Since then, several studies have provided important knowledge of how collaborative relationships among firms depend on institutional structures. For example, Chua et al. (2009) found that

1234

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

institutionalized arrangements, as they differ among countries, entail different levels of cognition-based trust. Also, countries vary in their development and enforcement of formal institutions such as contract law (Stephan and Uhlaner, 2010). Whether such institutions are culturally grounded or established by formal law, they contribute to a social order holding specific properties that guide and direct the exchange of resources among firms, including exchange of resources for innovation. Following this reasoning, some institutional environments will expectedly be more supportive of innovation though networks, while some will be less supportive. This embeddedness of innovation in networks along with the embeddedness of networks in institutions have received only scant attention, as lamented in recent reviews (Phelps, 2010; Autio et al., 2014; Stam et al., 2014). A notable exception is a study of the alliance networks of 109 firms in nine countries (Vasudeva et al., 2013a,b). The study found that country differences in underlying norms for collaboration, as reflected in national institutional arrangements for inter-organizational collaboration, was associated with different patterns for partner selection in strategic alliances, and with different innovation potentials from occupying structural holes. Additionally, research on national and regional systems of innovation has generated many case studies and some comparisons, insightfully perceiving local systems, but little evidence on the world’s variability and little possibility for globally disentangling and assessing joint effects of institutions and networking on innovation (Andersen, 2012; Guan and Chen, 2012; Rodríguez-Pose and Di-Cataldo, 2014; Guan et al., 2016). The gap is thus a lack of knowledge about the joint effects of networks and institutions upon innovation. This frames our research question, How are firms’ innovation affected by their networking, as these endeavors are embedded in institutions in society? The contribution here is to assess not only the separate benefits of networking and institutional support, but to ascertain how benefit of networking for innovation differs around the world depending on institutions. By using a large sample of firms in many countries which are approximately representative of the world, our results can be generalized to the world. We use two-level modeling to ascertain direct and moderating effects of institutions on firm-level behavior, modeling that is increasingly used in research on entrepreneurial and innovative activity (Bosma, 2013; Stenholm et al., 2013; De Clercq et al., 2011; Levie et al., 2014). The following Section 2 reviews research and argues for a twolevel approach to examine direct effects of institutional support for networking at the country level and of networking at the firm level on innovation, and also the moderating effect from a country’s institutional support for networking on the benefit of networking for innovation. Section 3 describes our design and data for analyses at the country and firm levels. Section 4 presents the results and discussion. Section 5 summarizes the results, considers implications for theory and practice, proposes directions for future research, and considers limitations.

2. Theoretical background and hypotheses Theorizing combines the macro-level of institutions and the micro-level of firms pursuing networks and innovation, as depicted in Fig. 1. First we review firm-level effect of networking upon innovation, the horizontal arrow. Then we consider national-level institutional support for networking as it affects networking at firm-level, as a cross-level effect, the sloping arrow. Finally, we argue for a moderating effect of a country’s institutional support upon benefit of networking, the vertical arrow.

Fig. 1. A two-level perspective on institutions, networking and innovation.

Institutional support for networking is expectedly promoting innovation indirectly, by promoting networking that benefits innovation, but is not expected to directly affect innovation. 2.1. Network effects on innovation During the last decades, networks have become prominent in the innovation literature (Pittaway et al., 2004; Ozman, 2009; Parmigiani and Rivera-Santos, 2011; Leyden et al., 2014). This turn was driven by increasing dynamic and uncertain business environments and changes in inter-firm dynamics around increasing specialization and new management logics favoring inter-firm cooperation (Zaheer et al., 2000; Helfat and Peteraf, 2009; Chesbrough, 2003). The increase in scientific and productive knowledge makes knowledge develop faster outside than inside firms (Huber, 2004). To keep abreast, firms pursue external relationships to gain timely access to new knowledge and to exploit new opportunities within shortened windows of opportunity. The basic proposition is that networks benefit innovation by linking ideas and resources held by otherwise unconnected actors and thereby bring novelty through processes of recombination (Burt, 2000; Obstfeld, 2005). From this perspective several theoretical approaches have contributed to the understanding of the mechanisms for information and knowledge transfer among actors. Challenges may include high transaction costs and difficulty of acquiring tacit knowledge (Dhanaraj and Parkhe, 2006). Important social control mechanisms in the form of trust and reciprocity may reduce risk of malfeasance and hence transaction costs (Dyer and Singh, 1998). Also, inter-firm complementarity and shared understandings, which are enhanced by relational experience, trust and reciprocity, seem to decrease cognitive barriers to knowledge transfer and to increase benefits of inter-firm relationships (Powell and Grodal, 2005; Jensen and Schøtt, 2015). Evidence shows how firms’ innovation can benefit from collaboration with diverse partners such as clients, customers, suppliers, distributers and even competitors. However, innovation benefits seem to be different from different kinds of partners (Schøtt and Sedaghat, 2014; Zeng et al., 2010), and innovation partnerships may mal-function (Lhuillery and Pfister, 2009; Lokshin et al., 2011). The use of external sources not only enhances combinatory potential, but also enables tailoring products to customer requirements (Lipparini and Sobrero, 1994). Likewise, networks help overcoming liabilities of newness and smallness in commercialization of innovative products (Partanen et al., 2011). Interactions with different types of partners are likely to facilitate various stages of the innovation process (Love et al., 2011), and different types of collaborative partners vary in benefits for product and process innovation (Fitjar and Rodríguez-Pose, 2013). A study found that cooperative arrangement for innovation made with suppliers, customers, clients, competitors, universities, consultants, private research institutes, government institutes and research associations, and research and technology organizations—increased novelty of products for market (Tether, 2002). Another study found that collaboration with suppliers, collaborators, and research organizations benefitted innovativeness, but also found that collaboration with competitors was detrimental (Nieto and Santamaria, 2007). This challenge

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

of managing the tension between collaboration and competition is well illustrated by a finding that Norwegian firms collaborating with competitors had less product innovation (Fitjar and Rodríguez-Pose, 2013). At a more aggregate level, the collaborative benefits from partnerships have been assessed from considerations of the strength of the ties between the focal firm and its partners (Wang, 2016). The strength of ties has trade-offs for quality. At the one end, weak ties are found to be conducive of information that is new, but less easy for acquiring tacit knowledge (Rost, 2011). At the other end, strong ties carry less novelty, but have higher capacity for bringing tacit and sensible knowledge (Tortoriello and Krackhardt, 2010). At the dyadic level, research has found tie strength measured by frequency of interaction with clients and customers to be positively associated with the novelty in innovations (Tether, 2002; Amara and Landry, 2005). Furthermore, trust as a property of tie strength seems to have a positive impact on innovation in inter-firm collaboration (Das and Teng, 1998; Chua et al., 2009). Also from a structural perspective, studies confirm benefits of firms’ networking measured by size, strength and diversity for innovation in firms (Ahuja et al., 2008; Bell, 2005). This evidence, however, is contested. Notably, a meta-analysis of the relationship between SME innovation and SME performance found that while greater investment in internal R&D increase innovation and performance, innovation projects with collaboration partners did not significantly enhance performance (Rosenbusch et al., 2011). This finding challenges the view that networking with external partners benefits innovation. But perhaps this research is influenced by the design that measured the collaboration that firms direct specifically at innovation projects. Collaborative networks that are not specifically formed around joint innovation projects may still spur innovation, more serendipitously. As such, any collaborative relationship may expose a firm to new information and alternative interpretations and thereby challenge the firm’s existing world view and potentially lead to new opportunities. Anyway, this controversy leaves the question open as to the impact of collaborative networks on firm innovation, both process innovation and product innovation. We consider this issue in the following hypotheses 1a and b: Hypothesis 1. (a) Higher intensity of a firm’s networking typically brings higher level of process innovation in the firm. Hypothesis 1. (b) Higher intensity of a firm’s networking typically brings higher level of product innovation in the firm. These hypotheses about effects of the quantity of networking are our preamble for hypothesizing, in a later section, about the quality of networking. 2.2. Institutions and firms’ collaborative networking New institutional theory explains organizational practices as responses to demand for legitimacy. Practices become legitimate as they are diffused throughout institutional fields, promulgated by institutional agents such as governments and professional associations, and adopted by organizations susceptible to coercive, mimetic and normative pressures (DiMaggio and Powell, 1983). Several accounts have explained how such pressures promulgate management practices such as scientific management (Shenhav, 1995), financial reporting practices (Mezias, 1990), retail sales compensation (Eisenhardt, 1988), and ISO 9000 Quality Certification (Guler et al., 2002). Institutions function as rules of society, as scripts for meaning making (Powell and Colyvas, 2008), and for reproducing social patterns or order, where reproduction is based on “relatively selfactivating social processes” (Jepperson, 1991: 145). The legitimacy

1235

of such “rules” is based on generalized perceptions or assumptions of the desirability and appropriateness of ideas and actions (Suchman, 1995; Powell and Colyvas, 2008). When scripts acquire a taken-for-granted character, they serve as shared interpretations and justifications for pursuing various activities and practices, including networking. The emphasis on the role of logics in shaping such generalized perceptions and assumptions of desirability and appropriateness vary between different strands in institutional theory. Scott (2003) joins these strands in what he terms an ‘omnibus’ conception of institutions. Here three pillars represent three accounts of how regulative, normative, and cultural-cognitive elements of social structures bring stability and meaning to social life. We follow a series of recent studies, which have used this conception to analyze institutional effects on various entrepreneurial phenomena (Kshetri, 2009; Stenholm et al., 2013). We here consider how mechanisms from these three pillars may invoke social structures that are more or less supportive of inter-firm networking. Specifically, we conceptualize institutional support for networking in terms of generalized perceptions and assumptions of the desirability and appropriateness of collaborative networking among firms. Following Scott (2003), these perceptions and assumptions are considered reflections of how regulative structures impose, authorize and induce networking among entrepreneurs, how normative structures work to incorporate specific networking practices as standard procedures, and how cultural-cognitive mechanisms effectuate imprinting taken-forgranted understandings of networking into inter-organizational practices. The concern with the regulative elements of institutions has its roots in an economic perspective and focuses on the effects of public policies, regulations, laws and rules (Scott, 2014). From this perspective, behavioral patterns are widely influenced by a coercive mechanism along with logics of instrumentality. As individuals engage in specific behaviors, they are guided by perceptions of incentives and constraints in the rules and the governance structures prescribed by institutional elements (North, 1990; Herstad et al., 2014). Thereby, individual perceptions and behaviors come to reflect macro-level structures such as legal, cultural, economic and educational structures (Baumol et al., 2007). Individuals respond to these regulative institutions as they weigh their interests in given situations under considerations of institutional pressures to conform and the risk of legal sanctioning that may result from non-conformity (Hoffman, 1999). Trade laws, tax structures, property right laws, contract laws, regulations of competition, etc., are all regulative elements that shape conditions for interfirm collaboration and influence which firms become attractive business partners and which forms of collaboration to pursue. But also the overall stability of the regulative institutions may impact inter-firm collaboration. As an example, Hitt et al. (2004) describe how institutional instability in Russia, partly caused by decentralized political control, prompted Russian managers to take short term views in their inter-firm collaboration. By contrast, Chinese managers, who operated under a more stable regulatory system, facilitated by centralized political authority, favored a long term perspective for inter-firm collaboration. The normative pillar takes a primary stand in sociology and emphasizes how prescriptive expectations associated with norms, values and roles form a basis for individual behavior. From this view, legitimacy of action is evaluated upon internalized norms and what is considered as appropriate according to those norms (Scott, 2014). The focus is here on the normative structures and particularly the identities, roles, and rules that are carried by social systems such as religious, educational and professional communities (March and Olsen, 2006). As individuals or firms identify themselves with, for example, professional communities, they will

1236

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

out of moral obligation and professionalization weigh their opportunities for action according what is expected from the role they play within their community (Feldman and March, 1981; Hoffman, 1999). Thus decisions to participate in a collaborative partnership will take on a rule-like character. Choices of action will here be based on standard procedures, repertoires of practices, standard ways of doing things, and rules of thumbs which are explained and justified by the norms and expectations that have evolved within their community (March and Olsen, 2006). Finally, the cultural-cognitive pillar advances the perspective that social order, and thus behavioral patterns, arise out of shared understandings and taken-for-granted beliefs (DiMaggio and Powell, 1983). From this perspective, social actors enact schemas, scripts and routines without giving much thought to their economic rationale or associated role expectations and obligations (Scott, 2014). Instead, their adherences to specific actions are subconsciously guided by symbols and cultural frameworks which shape their understanding of the world in which they navigate (Hoffman, 1999). Choice of action is often driven by procedural mechanisms in which social actors have no substantive or moral commitment (Hirsch, 1997). Conformity to rules relies on the legitimacy that is ascribed to these rules by the extent to which others, with whom the social actor interacts, follow them. A much used example of such unquestioned cognitive frameworks is the belief structures embedded in national cultures, which constitute what has been described as a collective programming of the mind (Hofstede, 1980). As an example, firms in collectivist countries can be expected to form networks by building close relational ties, whereas firms in individualistic countries will tend to establish contract-based exchange-relations (Tiessen, 1997). Also, uncertainty avoidance affects the propensity of SMEs to pursue technology alliances where uncertainty is high, and SMEs in countries with high levels of femininity are more prone to engage in technology alliances (Steensma et al., 2000). While the categorization into regulative, normative and cultural-cognitive elements of institutions is conceptually helpful, the distinction is analytical, and in reality the three pillars are highly intertwined (Hirsch, 1997; Hoffman, 1999; Scott, 2014). The intertwining is illustrated by a study showing how institutions are shaped by sequences of change, and how different pillars may take dominance in different stages (Hoffman, 1999). Change in institutions starts from a questioning of existing beliefs, which transplants first into changes in regulative structures, then normative structures, and finally more slowly diffuses into cognitive structures. As this development progresses new institutions emerge when there is a convergence towards new beliefs across the three pillars, and new “rules of the game” arise, sustained by new beliefs that provide legitimacy and give meaning to new organizational practices (Hoffman, 1999). In this sequencing of institutional developments, the culturalcognitive elements of taken-for-granted beliefs are highly inert and occur at a more subconscious level (Hoffman, 1999). By contrast, the regulative and normative structures are more a result of human design by purposive and instrumentally motivated institutional actors (DiMaggio and Powell, 1983). This ties the regulative and normative elements of institutions more closely to regional policy, and to policy-makers and professional groups as influential institutional agents (Scott, 2003). Dominant beliefs are here reflected in policy initiatives and institutional arrangements (Stenholm et al., 2013). Institutionalized beliefs about inter-firm networking are reproduced when policy makers attempt to induce inter-firm collaboration by sponsoring and otherwise supporting business incubators, science parks, industry clusters and other forms of administrative structures to facilitate networking among firms (Chaminade et al., 2012; Larédo, 1998; Nishimura and Okamuro,

2011). Formal laws and rules for contracting, laws for protecting intellectual property rights (Autio and Ács, 2010), for reducing corruption (Anokhin and Schulze, 2009), and consistency and strength in law enforcement (Aidis et al., 2008), are also manifestations of dominant institutionalized beliefs about the regulation of transactions among firms. In the normative realm, we see such manifestations for example in the educational system, where business students are taught of the dominant theories about how to understand competitive and cooperative dynamics, and are also seen in professional organizations that develop norms and practices for inter-firm collaboration. When considering the manifestation of institutions in the cultural-cognitive realm, we often search for shared conceptions of problems and ideas about how to solve such problems as revealed in coordinated practices (Scott, 2014). A study describes how consensus and cohesion characterize inter-firm collaboration in corporatist countries such as Germany and Japan, which favor coordination and inclusiveness; in contrast, firms’ collaboration networks in less corporatist countries such as United States and UK tend to be more sparse and unconnected. Such differences in perceptions of benefits and appropriateness of specific forms of collaboration impact practices for acquiring external knowledge and selecting alliance partners (Vasudeva et al., 2013a). When considering institutional support for inter-firm networking as a means to innovation, we consider the combined support from the regulative, normative and cultural-cognitive realms. In doing so, we see institutional support for networking as expressed by entrepreneurs’ perceptions of how well these realms are supportive of inter-firm networking. Thus, the extent to which inter-firm collaboration networks are rationalized as legitimate solutions and adopted by firms within a country is a combined effect of the perceived desirability and appropriateness which are promulgated by regulative, normative and cultural-cognitive spheres within a country. In accordance with this view, we state our hypothesis 2: Hypothesis 2. A country’s institutional support for networking typically enhances its firms’ networking. Hypothesis 2. Concerns the effect on quantity of networking. Next, we consider quality of networking, quality in terms of its benefit. 2.3. Institutional support moderating benefit of networking for innovation We argued above that the institutional context within a country is likely to affect networking of firms. We now turn to argue that the institutional context impacts the effect of networking on innovation by affecting the quality of relationships. We here discuss trust and accessibility of resources as two relational mechanisms that impact the innovation process (Powell et al., 1996), which can also be considered elements of the institutional context (De Clercq et al., 2011). At the institutional level, trust reflects a general tendency of people to believe that they will not be taken advantage of even if others get the opportunity to do so (Putnam, 2000). A general propensity to trust others has a positive effect on the willingness to engage in partnerships when uncertainty is high (Lorenz, 1999). During the exchange of resources among firms, a high level of trust is a mechanism that to some extent substitutes for more formal and contractual control (Tsai and Ghoshal, 1998; Bodas-Freitas et al., 2011). Such informal control allows for higher flexibility and speed in the innovation process. Further, the prevalence of trust increases the intensity and quality of resources that firms allow to exchange across their boundaries (Molina-Morales et al., 2011).

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

We expect that general trust will be higher in countries where institutions support networking. This contention that trust in interfirm relationships has a primary base in institutional structures has previously been presented (Zucker, 1986). In line with the three institutional pillars considered above, the level of trust in a society is likely to be a function of the strength of legal, political and social systems in governing behavior, both legally, morally and culturally (Zucker, 1986). These multiple institutional sources of trust to support cooperative behavior in inter-firm relationships are evidenced in several empirical studies. The stability and consistency in a country’s institutional frameworks provided by legal norms, and industry standards for technology and market conduct are associated with higher levels of trust and closer inter-firm collaboration (Welter et al., 2005). Particularly strong institutionalized practices encourage cooperative behavior among Japanese firms (Hagen and Choe, 1998). National culture is influential, e.g. individualist societies such as the US and Australia promote trust and cooperation with out-group members, more than collectivistic societies such as several Asian societies (Huff and Kelley, 2003; Chen and Li, 2005). Another way in which institutions may affect innovation is by providing for access to valuable resources (Ács et al., 2014). Again here, there is reason to assume a high association between the emphasis on networking within a country, and the accessibility to resources which may promote innovation. Among the most prominent examples are the use of intermediary technology and innovation centers, and the involvement of research institutions in industry and firm innovation (Zeng et al., 2010). Empirical studies have shown that intermediary institutions take an important role in facilitating innovation (Nieto and Santamaria, 2007). Similarly, there are several studies that support the important role that research institutions and universities have for innovation when they cooperate with private firms for innovation (Liefner et al., 2006). The institutional frameworks that govern the relationships between industry and universities vary between countries (OwenSmith et al., 2002). For example, US firms have been found to have closer integration with universities when compared to firms in Europe and Japan. Similarly, technology transferred more effectively between universities and the biotechnology industry in the US than in Japan, UK, and Germany (Bartholomew, 1997). This supports a number of policy recommendations for National Innovation Systems depending on regional characteristics (Tödtling and Trippl, 2005). Among those recommendations, several are directed at giving regional firms access to valuable resources and improving the conditions for their transfer: building up providers of new skills, attracting leading global companies, assisting business start-up and spin-offs, stimulating networking at national and international levels, improving international visibility of local clusters, encouraging industry-university cooperation, etc. As we can infer from these recommendations, both cognitive and motivational aspects are at play. Making the resources available is a necessary but not a sufficient condition. To support innovation through networking, institutions must enhance the availability of valuable resources, and at the same time be supportive of perceptions that their transfer is beneficial and appropriate. Thus, regulative inducements from policy makers should work in concert with institutionalized normative and cultural-cognitive structures. To summarize, we have argue that both trust and accessibility of resources as embedded in institutional structures may increase the amount of networking activity. However, further than this, and as emphasized above, the regulative, normative, and culturalcognitive elements of institutions may additionally increase the quality of networking at any given level of activity. We have here argued for effect on networking quality with respect to firm innovation. As countries will tend to vary in the extent to which their institutional environment supports the quality of collaborative relationships we will therefore expect variations in the extent to

1237

which given networking activity of firms in different countries will impact firm innovation, both process innovation and product innovation (Hullova et al., 2016). We state this expectation as Hypotheses 3a and b, Hypothesis 3. (a) The benefit for process innovation of a firm from its networking is moderated by the country’s institutional support for networking, typically, such that the benefit is enhanced by high institutional support for networking. Hypothesis 3. (b) The benefit for product innovation of a firm from its networking is likewise moderated by the country’s institutional support for networking, typically, such that the benefit is enhanced by high institutional support for networking. 3. Data and method Our conception of firms acting in the context of society refers to the universe of all firms in all societies. For studying institutional support, the unit of analysis is a country, and for studying behavior such as networking and innovation, the unit of analysis is a firm. Data on countries and firms have been collected in the Global Entrepreneurship Monitor, GEM, which is increasingly used for policy-relevant academic research (Bosma, 2013). The GEM consortium annually conducts two surveys in many countries around the world (Reynolds et al., 2005). The Adult Population Survey samples and interviews adults about their involvement in entrepreneurial activities. The National Expert Survey assesses the institutional framework conditions in each participating country. In 2012 the Adult Population Survey adopted questions for owner-managers about their firms’ networking (repeated in some countries in 2013), and the National Expert Survey asked about institutional support for networking in each country. A few years after each survey, the common questionnaire and data are placed in the public domain (www.gemconsortium. org). 3.1. Sampling GEM samples in two stages, first selecting countries and then sampling people. Countries are included by self-selection, by formation of a national team that joins GEM GEM uses two-stage sampling, first sampling countries and then sampling people. Countries are sampled by self-selection, by formation of a national team that joins GEM. Both firms’ networking and national institutional support have been measured in a sample of 68 countries comprising Algeria, Angola, Argentina, Austria, Barbados, Belgium, Bosnia and Herzegovina, Botswana, Brazil, Chile, China, Colombia, Costa Rica, Croatia, Denmark, Ecuador, Egypt, El Salvador, Estonia, Ethiopia, Finland, France, Germany, Ghana, Greece, Hungary, India, Iran, Ireland, Israel, Italy, Jamaica, Japan, Korea, Latvia, Lithuania, Macedonia, Malawi, Malaysia, Mexico, Namibia, Netherlands, Nigeria, Norway, Pakistan, Palestine, Panama, Peru, Poland, Portugal, Romania, Russia, Singapore, Slovakia, Slovenia, South Africa, Spain, Sweden, Switzerland, Taiwan, Thailand, Trinidad and Tobago, Tunisia, Turkey, Uganda, United Kingdom, Uruguay and Zambia. This diversity of countries is imperfectly but approximately representative of the countries around the world. Within each country, adults were sampled approximately at random, in many countries by random-digit-dialing or randomly drawing from a national database of private telephone numbers and in many other countries by sampling location and then, within these locations, sampling a quota of adults for face-toface interviews (Bosma, 2013). Within each country, this yields an approximately representative sample of adults from which we use the sub-sample of those owning and managing an operating

1238

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

Table 1 Firms’ process and product innovation (N = 18,880 firms).

High innovation Medium innovation Low innovation Total

Process innovation

Product innovation

6% 13% 81% 100%

11% 19% 70% 100%

business. The approximate representativeness at each stage implies that results are generalizable to the universe of all firms in all societies. The Adult Population Survey in the 68 countries yielded a sample of 18,880 operating firms, for which no data are missing on any of the variables used in the analyses. By far most businesses are very small (In GEM, 45% have no other worker than the owner(s), and only 5% have more than ten workers, apart from the owner(s)). The major other survey is the Community Innovation Survey, CIS, which, however, samples only firms with at least ten employees. GEM and CIS are thus largely complementary in their “populations” of businesses. 3.2. Innovation in a firm Innovation in a firm comprises process innovation and product innovation. The GEM Adult Population Survey asked the ownermanager about process and product innovation, respectively, - Have the technologies or procedures required for this product or service been available for less than a year, or between one to five years, or longer than five years? - Do all, some, or none of your potential customers consider this product or service new and unfamiliar? The conceptualization in GEM differs somewhat from the conceptualization in the Oslo Manual and the conceptualization in the Community Innovation Survey. Earlier CIS and the Oslo manual limited innovation to technological innovation (Smith, 2006). Recent CIS and OECD guidelines, however, encompass other kinds of innovation, e.g. innovation in marketing and organizations (Tavassoli and Karlsson, 2015), also in the public sector (Arundel et al., 2015). The CIS and GEM surveys are somewhat similar, in that both encompass process-innovation and product-innovation, and in that newness does not mean new to the world, but may be quite local. The conceptualization and measurements in GEM, that is smallscale and local, are quite different from measurements based on R&D expenditures and on patenting. It is therefore not surprising that some countries rank high on the GEM measures, but low on measures based mainly on R&D expenditures and patenting, e.g. as used in the Global Innovation Index (Smith, 2006). The kind of innovation that is measured entails a limit on the generalization that can be made. It is more precise to say that GEM-based results generalize within the domain of this kind of small-scale local innovation (World Economic Forum and Global Entrepreneurship Research Association, 2015). GEM, by the above questions, measures process and product innovation on a three-point scale, low, medium or high innovation, which is often modeled linearly (Hovne et al., 2014; Jensen and Schøtt, 2014; Jensen et al., 2016; Schøtt and Sedaghat, 2014; Schøtt et al., 2014; Schøtt and Cheraghi, 2015). Most firms are low in process and product innovation, Table 1. The variance among firms in the process innovation comprises the variance within the countries among and the variance between the means of the countries. The variance among firms in their process innovation comprises 84% within countries and 16% between countries (the intraclass coefficient is 0.16). The variance among firms in their product innovation comprises 80% within coun-

tries and 20% between countries (the intraclass coefficient is 0.20) (Snijders and Bosker, 2012, Ch 3). The two measures of process and product innovation are correlated positively, of course. Their correlation is 0.20. This correlation is not so high that it calls for viewing innovation as an underlying factor, uni-dimensional with manifestations in processes and products. Instead, process innovation and product innovation are considered distinct and treated separately in the analyses of effects on process and product innovation (Hullova et al., 2016). 3.3. Network around a firm The network around a firm here denotes its collaborative relations. Relations are distinguished by their contents or substance of collaboration, namely as collaboration about seven endeavors: production, supplies, marketing, creation of new products for current market, search for new markets for current products, development of new product for new markets, and improving effectiveness of the business. The first three substances concern exploitation and the next three contents refer to exploration (Yamakawa et al., 2011). These kinds of relations were measured by asking, Is your business working together with other enterprises or organizations to produce goods or services? Is your business working together with other enterprises or organizations to procure supplies? Is your business working together with others to sell your products or services to your current customers? Is your business working together with others to sell your products or services to new customers? Is your business working together with others to create new products or services to your current customers? Is your business working together with others to create new products or services to new customers? Is your business working together with others about how to make your business more effective? Furthermore, when a relationship was reported to exist, a follow-up question asked whether the collaboration was intense or not so intense. The relationship is thus measured on a scale, going from no relation, through a weak or not so intense relation, to a strong or intense relation, coded 0, 0.5 and 1. Each kind of relation was mostly absent, Table 2. Most firms, however, had at least one kind of relation. The seven kinds of relations are correlated positively, of course. Their inter-correlations are between 0.3 and 0.8, Cronbach alpha is 0.86, and an exploratory factor analysis shows first eigenvalue 3.9 and second eigenvalue 0.9, with all factor loadings above 0.6. The seven relations thus manifest one underlying factor, a firm’s tendency to network. A firm’s networking can thus be measured as the mean of its seven relations (Jensen and Schøtt, 2014). The variance among firms in their networking comprises variance between countries and variance within countries. 87% of the variance among firms is within countries and 13% is between countries (the intra-class correlation coefficient is 0.13) (Snijders and Bosker 2012, Ch 3). 3.4. Institutional support in a country Institutional support for networking was measured by GEM in the National Expert Survey in 2012. National experts were selected based on their expert knowledge of entrepreneurship in their country. Typically experts were identified among entrepreneurs, consultants, academics, and also politicians or administrators. Questions about institutional support were designed to measure regulative, normative, and cultural-cognitive elements of a country’s institutions. In that way, the questions include elements from all three institutional pillars conceptualized by Scott (2014), and

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

1239

Table 2 Collaborative relations around firms. (N = 18,880 firms).

Strong relation Weak relation No relation Total

Production

Supply

Marketing

New customers

New products

New prod. New cust.

Effecti-veness

19% 10% 71% 100%

22% 11% 67% 100%

7% 10% 73% 100%

15% 10% 75% 100%

12% 7% 81% 100%

11% 7% 82% 100%

18% 11% 71% 100%

thus operationalize institutional context. The experts were asked to assess truthfulness versus falseness of the following statements about their own country, where the second and third indicate regulation, the fourth and fifth indicate norms, and the sixth and seventh indicate cognition,

- Public institutions often organize fairs and events where entrepreneurs meet and form contacts. - The government has a policy for promoting and supporting collaboration among businesses. - The local public authorities promote and support collaboration among businesses. - The educational system teaches that businesses ought to collaborate. - Training courses for entrepreneurs include training in collaboration. - Business owners believe that informal agreements are more effective than contracts between businesses. - Business owners believe they gain advantages through collaboration.

Truthfulness was rated on a scale from 1 to 5, as a measure of support of a specific kind. For each statement, the ratings were across the experts in the country, as seven measures of specific kinds of support in the country. An exploratory factor analysis revealed a single factor, showing that the institutional elements are intertwined rather than distinct regulatory, normative and cognitive factors. The seven kinds of support covary across countries; the correlations among them are all positive, and Cronbach alpha is 0.82. The seven measurements thus express one underlying factor, a national tendency to support networking. The seven measures of support were averaged as a measure of support for collaboration in the country, on a scale from 1 to 5. Institutional support differs considerably among countries, Table 3. Institutional support was highest in the Netherlands and lowest in Greece. This measure of institutional support is used for analyzing the effects of institutional support on networking and innovation.

Table 3 Support for networking in most supportive and least supportive countries. Netherlands Jamaica Singapore Switzerland Macedonia

3.77 3.61 3.57 3.54 3.51

Iran South Africa Egypt El Salvador Greece

2.67 2.66 2.51 2.49 2.32

3.5. Control variables Networking and innovation are influenced by many characteristics of firms and their owner-managers. GEM Adult Population Survey measures several characteristics, here used largely with the original full scales from GEM (questionnaire is at www.gemconsortium.org), Proprietorship refers to whether the firm has sole or shared ownership and is measured as a dichotomy, coded 0 if shared and 1 if sole proprietorship. Owners refers to number of owners including the respondent, is skewed, so for analyses the count is transformed logarithmically. Firm-age, as number of years for which compensation has been paid, is also skewed, and therefore transformed logarithmically. Firm-size, as number of people working for the business, including the respondent, is also skewed, and therefore transformed logarithmically. Motive for owning-managing the business, is measured in five categories: opportunity in that the owner-manager’s reason for running the firm was pursuing a business opportunity, necessity in that the reason for running the business was the need to make a living, both opportunity and necessity, pursuing an opportunity while having a job, and another motive. Self-efficacy denotes the responding owner-manager’s confidence in own ability to start and run a business, and is in GEM

Table 4 Means standard deviations, and correlations (N = 18,880 firms).

Support Networking Process innovation Product innovation Proprietorship Owners (log) Firm-age (log) Firm-size (log) Self-efficacy Opportunity-alertness Risk-propensity Education Income Age Gender: male * ** ***

P-value < 0.05. P-value < 0.01. P-value < 0.001.

Mean

Std. dev.

Support

Networking

Process innovation

Product innovation

3.1 0.21 1.3 1.4 0.75 0.24 1.9 0.77 0.80 0.55 0.70 9.8 2.2 40.6 0.60

0.30 0.28 0.57 0.69 0.43 0.48 1.0 0.91 0.40 0.50 0.46 5.2 0.80 11.5 0.49

0.06*** 0.08*** 0.14*** 0.06*** −0.04*** −0.04*** 0.05*** 0.01 0.13*** 0.05*** −0.02** 0.01 −0.01 −0.05***

0.08*** 0.17*** −0.14*** 0.16*** 0.02** 0.31*** 0.06*** −0.01 0.00 0.27*** 0.18*** 0.01* 0.10***

0.20*** −0.01 0.02* −0.18*** 0.02* 0.04*** 0.05*** 0.00 0.06*** 0.01 −0.11*** −0.02***

0.01 0.01 −0.08*** 0.11*** 0.04*** 0.03*** 0.01 0.10*** 0.06*** −0.03*** 0.00

1240

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

Table 5 Firms’ process and product innovation affected by networking.

Networking Institutional support Proprietorship: sole Owners (log) Firm-age (log) Firm-size (log) Motive: Necessity Motive: Opportunity and necessity Motive: Opportunity while in job Motive: Other reason Self-efficacy Opportunity-alertness Risk-propensity Education Income: Highest third Income: Lowest third Age Gender: Male Intercept

Process innovation

Product innovation

0.079*** −0.013 0.013 0.013 −0.077*** 0.003 −0.004 −0.016 −0.032 −0.013 0.005 0.020* 0.006 0.004*** −0.032*** 0.008 −0.001** −0.013 1.312***

0.275*** 0.073 0.076 *** 0.015 −0.042*** 0.029*** −0.057*** −0.083*** −0.024 −0.062** 0.056*** 0.018 0.011 0.003** −0.019 0.006 −0.001 −0.014 1.519***

For motives, the reference category is motivation by opportunity. For income, the reference category is the middle third income. The estimation is based on 68 countries with 18,880 firms. * P-value < 0.05. ** P-value < 0.01. *** P-value < 0.001.

measured as a dichotomy, and coded 1 if self-efficacious and 0 if not. Opportunity-alertness denotes the owner’s manager’s perception of local opportunities for business, and is measured as a dichotomy, 1 if opportunity-alert and 0 if not. Risk-propensity refers to not fearing failure if trying to start a business, and is measured as a dichotomy, 1 if risk-willing and 0 if risk-averse. Education of the responding owner-manager is initially measured as highest completed level of education, and then recoded into an approximate number of years. Income denotes household-income before taxes and is initially measured in brackets in the national currency, and then coded as low, mid or upper third of household incomes in the sample in the country, and treated as categories in analyses. Age of the owner-manager, is measured in years. Gender of the owner-manager, is coded 0 for female and 1 for male. These characteristics of the firm and its responding ownermanager will all be controlled for in the analyses. The measurements are crude, each concept is measured with much random error. Furthermore, numerous relevant variables are not measured. For example, the firm’s investment in innovation is not measured, characteristics of the firm’s workers other than the respondent are not measured, and characteristics of countries other than institutional support are not measured. Furthermore, differences among countries in effects of characteristics of firms and owner-managers on innovation (other than effect of networking) are not modeled. Finally, using models with linear and additive effects also entails imprecision. These many kinds of imperfections in measurement and modeling increase the disturbance terms and the residuals. Therefore, the modeling expectedly explains only a small proportion of the variance in innovation. Despite this, effects may still be estimated as strong. 3.6. Analyzing effects: two-level linear modeling The data on firm-level behavior affected by country-level institutional support are analyzed by two-level linear models, that are extensions of multiple regression, where hypothesized effects are

Table 6 Quantity of networking affected by institutional support. Institutional support Proprietorship: sole Owners (log) Firm-age (log) Firm-size (log) Motive: Necessity Motive: Opportunity and necessity Motive: Opportunity while in job Motive: Other reason Self-efficacy Opportunity-alertness Risk-propensity Education Income: Highest third Income: Lowest third Age Gender: Male Intercept

0.006 −0.011 0.014 −0.003 0.064*** −0.031*** −0.005 0.015 −0.011 0.026*** 0.026*** 0.002 0.004*** 0.016*** −0.017*** −0.001** 0.018*** 0.272***

For motives, the reference category is motivation by opportunity. For income, the reference category is the middle third income. The estimation is based on 68 countries with 18,880 firms. * P-value < 0.05. ** P-value < 0.01. *** P-value < 0.001.

tested by coefficients (Snijders and Bosker, 2012). The modeling is linear (except for a check on robustness where we dichotomize innovation, and analyze this dependent variable by a two-level model, that is similar to a logistic regression). Analyzing ordinal variables such as innovation by a linear model is not only common practice but has some statistical justifications (Angrist and Pischke, 2008). In the ‘population’ of firms and countries, the effects of characteristics of firms and institutions are population parameters, called fixed effects. They contrast the effects of countries, that are more or less randomly sampled and without interest, which are called random effects. The countries differ in levels of networking and innovation, with country effects that are modeled as intercepts, and thus called random intercepts. The country-level condition is centered, and each firm-level variable is centered within each country (Snijders and Bosker, 2012). 4. Results Variables have distributions and correlations listed in Table 4. The strongest correlation is between sole ownership and log of owners (-0.87), so their Variance Inflation Factors are about 4 in linear regressions corresponding to the models in Tables 5–7, and thus far below the rule-of-thumb value 10. All other correlations are weak (less than 0.39) and all other VIFs are less than 2 in linear regressions like in the tables. This shows that there is no problem of multicollinearity. The hypotheses to be tested concern, first, effects of the quantity of networking upon innovation, second, effect of institutional support upon networking, and, third, effects of institutional supTable 7 Firms’ process and product innovation affected by the combination of networking and support.

Networking Support Networking * Support Control variables as in Table 5

Process innovation

Product innovation

0.079a −0.013a 0.100*

0.277a 0.073a 0.195***

The estimation is based on 68 countries with 18,880 firms. **P-value < 0.01 a Significance not tested (main effect was tested in Table 5). * P-value < 0.05. *** P-value < 0.001.

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

port upon the quality of networking in terms of its benefits for innovation. 4.1. Effects of networking upon process and product innovation Hypotheses 1a and b, that firms’ networking benefits their process and product innovation, are tested by two-level linear modeling, as described above. The independent variables are networking and control variables and the dependent variable is process innovation in one model and product innovation in another model, Table 5. Networking has a positive effect on process innovation and also a positive effect on product innovation, thus supporting Hypotheses 1a and b. Networking benefits product innovation much more than it benefits process innovation. The effect upon process innovation from networking is about as strong as the effect from education of the owner-manager, and the only stronger effect upon process innovation is from firm-age (judged by comparing the coefficients for the standardized variables; not reported). The network effect upon product innovation is far stronger than any other effect upon product innovation (as judged by standardized variables). The variance in process innovation explained by the model, or, more precisely, the proportion reduction in variance in process innovation (similar to R2 ) is 0.020 (low partly because of much random error, as was discussed in Section 3.5) (Snijders and Bosker 2012, Ch. 7). The variance in product innovation explained by the model, or, more precisely, the proportion reduction in variance in product innovation (similar to R2 ) is 0.022 (low partly because of much random error, as was discussed in the measurement section) (op cit). The effects of networking upon process and product innovation were also estimated by hierarchical models for ordinal data, considering innovation measured on the three-point ordinal scale. This analysis also shows that networking promotes process and product innovation. The effects of networking upon process and product innovation were also estimated by hierarchical models for binary outcomes, considering innovation dichotomously (collapsing medium and high innovation into one category). This analysis also shows that networking promotes process and product innovation. In short, networking benefits both process and product innovation considerably, as hypothesized. Later we shall see that the benefits are moderated by institutional support. 4.2. Effect of institutional support on quantity of networking Hypothesis 2 that institutional support for networking in a country actually has a positive effect on networking of firms, states a cross-level effect, from the macro-level of country to the microlevel of the individual firm. A first estimate of the effect is by the correlation between support and networking of the firms. This correlation is 0.06, computed across the firms, as listed in Table 4. A second estimate is the correlation between support and the national mean of networking. This correlation is 0.04, computed across the countries. This correlation is not statistically significant; the one-sided p-value is 0.36 based on only 68 countries, which is so small a sample that such a correlation should not be expected to be significant. A third and more appropriate way of estimating the effect of support on networking is by a two-level linear model, Table 6. The coefficient for the effect of support is positive as expected. But the coefficient is not statistically significant, being based on a sample of only 68 countries. So there is only very modest evidence favoring Hypothesis 2, that institutional support for networking actually promotes the extent of networking of firms.

1241

The variance in networking explained by the model, or, more precisely, the proportion reduction in variance in networking (similar to R2 ) is 0.075 (rather low partly because of much random error, as discussed in Section 3.5) (Snijders and Bosker 2012, Ch 7). Furthermore, comparison of the models without and with the variable for institutional support shows that adding this variable does not add discernibly to explaining variance in networking, so also this suggests that there is hardly any effect of institutional support upon the quantity of networking. The quantity of networking may thus be unaffected by institutional support or the sample may be too small to discern an effect. However, the quality of networking may be affected by institutional support, as examined next. 4.3. Institutional support moderating benefit of networking for innovation Hypotheses 3a and b state that institutional support for networking moderates the effect of networking upon innovation, namely in the way that great support amplifies the effect and lack of support reduces the effect. This moderating influence is modeled as an interaction effect; we just include the interaction or product of support and networking in the linear model of effects on innovation, Table 7. The combination of networking and support has a positive effect on process innovation, and also a positive effect on product innovation. This corroborates Hypotheses 3a and b. Support thus enhances the benefit of networking for both process and product innovation. In other words, the institutional support provided in a country for networking enhances the quality of networking in the sense that it enhances the performance benefit from networking. The interaction effects of support with networking upon process and product innovation were also estimated by two-level models for ordinal data, considering innovation measured on the three-point ordinal scale. This analysis also indicates positive interaction effects on process and product innovation. The interaction effects upon process and product innovation were also estimated by two-level models for binary outcomes, considering innovation dichotomously (collapsing medium and high innovation into one category). This analysis also indicates positive interaction effects on process and product innovation. The model enables us to estimate the importance of support for networking in terms of enhancing the benefit of networking for innovation. Network effect in a place with low support can be contrasted network effect in a place with high support. For process innovation in a place with low support (where support is one standard deviation below the mean of 0, i.e. −0.30; Table 1), the network effect is estimated to be 0.079–.30*.100 or 0.049. By contrast, in a place with high support (a standard deviation of 0.30 above average), the network effect is estimated to be 0.079 + 0.30*.100 or 0.109. So for process innovation, the network effect with a high support is estimated as more than twice the network effect with low support, a considerable enhancement. To get a sense of how precise this estimate of the network effect on process innovation is in a low support society, we use the confidence intervals around the coefficients. The 95% confidence interval around the main effect of networking is [0.050; 0.108], and the 95% confidence interval around the interaction effect is [0.071; 0.129]. Thus, a low value for the network effect is 0.050 + 0.129*(−0.30) or 0.011. A high value for the network effect is 0.108 + .071*(-0.30) or 0.087. So the interval [0.011; 0.087] expectedly includes the true value for the network effect on process innovation in a low support society. Recalling (from Table 1) that networking has standard deviation 0.28 and process innovation has standard deviation 0.57, this interval indicates that we can be fairly confident that networking benefits process innovation in a low support society notably.

1242

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

In a high support society, of course, the benefit of networking for process innovation is much higher. For product innovation in a place with low support, the network effect is estimated to be 0.277–0.30*0.195 or 0.218. By contrast, in a place with high support, the network effect is estimated to be 0.277 + .30*0.195 or 0.336. So for product innovation, the network effect with high support is estimated to be one and a half times the network effect with low support. To also obtain an impression of how precise this estimate of the network effect on product innovation is in a low support society, we use the confidence intervals around the coefficients. The 95% confidence interval around the main effect of networking is [2.420; 3.120], and the 95% confidence interval around the interaction effect is [0.081; 0.309]. A low value for the network effect is 2.420 + 0.309*(−0.30) or 2.327. A high value for the network effect is 3.120 + 0.081*(−0.30) or 3.096. Thus the interval [2.327; 3.096] expectedly includes the true value for the network effect on process innovation in a low support society. Bearing in mind that networking has standard deviation 0.28 and product innovation has standard deviation 0.69, this interval indicates that we can be very confident that networking benefits product innovation in a low support society considerably. In a high support society, of course, the benefit of networking for product innovation is far higher. The variance in process and product innovation explained by including the interaction in the modeling in Table 7 is slightly higher than the variance explained without the interaction (Table 5). In short, institutional support in a country for firms’ networking enhances the quality of networking by enhancing the benefits of networking for both process and product innovation, where both enhancements are considerable.

5. Conclusions and discussion The core dynamic of a national system of innovation is the interactive learning pursued by the innovating actors, the firms (Lundvall, 1992). Therefore our starting point was the old proposition that innovation benefits from networking. This proposition was refined as several hypotheses. Distinguishing between process and product innovation, the first hypotheses were that both process and product innovation benefit from networking, collaboration about operations of the firm. These firm-level activities are contextualized by considering them embedded in national institutional contexts with varying generalized perceptions of supports for networking. The institutional support for networking expectedly has macroto-micro effects that are both quantitative and qualitative. The quantitative effect is in the second hypothesis that institutional support for networking increases the quantity or intensity of networking. The qualitative effect is in the hypotheses that institutional support enhances the benefit of networking for both process and product innovation. Two-level modeling showed, as expected, that firms’ networking benefits both process and product innovation. Institutional support has no discernible effect on the quantity of networking, but affects the quality of networking, in the way that institutional support enhances the benefits of networking for both process and product innovation. The finding of a positive impact of firm networking on firms’ product and process innovation is well in line with findings from prior research. Although, as we have outlined in this paper, there is some ambiguity in the results concerning the impact of particular network compositions and specific structural characteristics of the network surrounding a firm, reviews in general are conclusive that networks contribute positively to firms’ innovation.

Compared to the meta-analyses by Rosenbusch et al. (2011), which found no performance impacts from firms’ collaboration in formal innovation projects, the conceptualization of firms’ collaboration networks in this study is not restricted to formal innovation projects but includes various other forms of collaboration. Comparing our results with those of Rosenbusch et al. (2011) may thus be an indication that important parts of firms’ inspiration for innovation are acquired in less formal collaborative arrangements and from collaborative relationships in which joint innovation is not the primary purpose of collaboration. Surprisingly, we did not discern a significant relationship between a country’s institutional support for networking and the extent of networking activity of firms. It would be going too far to induce from this non-finding, that institutional support for networking has no influence on the quantity of inter-firm networking. However, the result can be seen as an indication that firms may still be quite keen networkers even when conditions are not supportive. This reasoning corresponds well with the finding by Staber (2001) that organizing for clusters often does not increase networking among firms within clusters. One explanation could be that networking had already been established before clusters were more formally recognized. The findings by Staber (2001) are especially interesting as they are similarly indicative of a contextual impact on the association between networking and innovation. Staber (2001) found that even if clusters did not increase networking activity, clusters increased the level of innovation. Seeing networks as a significant driver of innovation within clusters, one interpretation is that the nature, or the quality, of inter-firm relations changed to become more effective for innovation as clusters were molded. While our analyses are at a more macro-level with a focus on the impact of institutions, we see a comparable contextual impact: the more supportive institutions are of networking, the higher innovation benefits firms attain from their networking. Prior research reveals only scant empirical evidence of such moderating impact of institutions on the association between networking and innovation. However, quite extensive research has examined how different institutional structures, ranging from regulative arrangement over normative beliefs to taken-for-granted assumptions embedded in national cultures, associate with different networking traditions and practices and with different modes of inter-firm collaboration (Welter et al., 2005; Vasudeva et al., 2013a; Ács et al., 2014). These studies commonly invoke mechanisms such as generalized trust, norms of reciprocity, and legally prompted governance mechanisms, to explain national differences in firms’ networking. For example, the tendency of firms in collectivist countries to form close relational ties with in-group members and to be skeptical of cooperation with out-group members is associated with lower levels of generalized trust (Chen and Li, 2005). In a similar way, research has associated national levels of innovation with much the same institutional elements and has invoked much the same underlying mechanisms to explain this association. Perhaps most illustrative is the research on the relationship between collectivism and innovation. Collectivism has been associated with lower innovativeness, which has in part been reasoned by a lack of outward orientation and a stronger value orientation toward coherence (Shane, 1993). Thus, what we have seen as the innovation effects from these cultural orientations are likely in part an effect of different collaborative forms pursued in different cultures. Differences in innovation thereby arise in part because not all collaborative forms are equally conducive to innovation. Following this reasoning, it is to be expected that the network − innovation relationship is moderated by institutions.

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

5.1. Implications for policy and research This study is among the first large scale studies to explain firm-level innovation as a result of an interaction between microlevel and macro-level factors. It is also among the first large scale studies to consider institutional contexts as a contingency for the functioning of networks. In most studies, networks effects are considered independent of the environment surrounding the network. By explaining variations in network impacts, such as on innovation, by embeddedness in the institutional environment, this study provides a more nuanced understanding to help policy makers design national institutional arrangements with the potential to reap innovation benefits from inter-firm networking. The contribution of this study is thus, concretely, to account for innovation as benefitting from networks around firms and the support for networking from institutions in the country. More abstractly, the contribution is to understanding the successive embeddedness of firm operations, the relational embeddedness of operations in a network, as an embeddedness at the micro-level, and the structural embeddedness of networking in institutions in society, as an embeddedness at the macro-level (Zukin and DiMaggio, 1990). Thereby the study also contributes to neoinstitutional theory, specifically theory about how institutions shape organizational behavior and inter-organizational dynamics. From an innovation perspective we contribute to the increasing number of studies, which from different theoretical perspectives argue for a contingent effect of networking on innovation. Most of such studies have so far focused on contingencies internal to the firm, such as absorptive capacity (Tsai, 2009) and intellectual capital (Zheng, 2010). The contingent effect from institutions, as found in this study, implies that there are also factors outside the firm and outside the agency of managers, which are important to firms’ considerations of their networking for innovation. Likely, many of such considerations will be guided by prevalent norms and more or less unconscious beliefs and taken for granted assumptions of which network strategies to pursue. But, considerations of the desirability and appropriateness of different forms of collaboration may also occasionally be challenged. Such challenges may arise for instance when new ideas and logics are infused through changes in government policies or regional policy initiatives. Policy-makers are keenly aware of networking as a social capital that yields competitive advantages in innovation for firms; advantages that differ across clusters, regions and countries (OECD, 2010). With this awareness, inter-firm networking takes a prominent position in policy development (Magro and Wilson, 2013). The findings in this study are well in line with such development. However, there are also some words of caution on how policy makers develop and implement such policies, and on the expectations that such initiatives will lead to immediate changes in collaborative dynamics followed by increasing levels of innovation. Our findings are suggestive of a tight coupling between the regulative, normative and cultural-cognitive institutional pillars. One interpretation of such tight coupling is that institutions coevolve across the three institutional pillars. In that sense, high levels of institutional support for networking will be present when support for networking is induced not only by instrumentally oriented perception of incentives and constraints, but also by norms and taken-for-granted assumptions. From this reasoning, abrupt changes in the regulative structures sparked by policy initiatives cannot by themselves suffice to reach a high level of institutional support for networking. To really reap greater innovation benefits from institutional support for networking, support for networking needs to be further institutionalized on a normative basis, and even better, operated on a more unconscious level guided by taken-forgranted assumptions.

1243

Prior research has shown that while policy makers are important agents in the evolution of institutional fields, the transference of logics from the regulative elements of institutions to the normative, and in particularly to the cultural-cognitive elements, is typically a long process (Powell and Colyvas, 2008; Scott, 2003). In this way, the immediacy of the impact from networking policies on firms’ networking behavior and innovation will likely depend on the correspondence of the logics behind these policies and the logics that are currently dominating in the normative and cultural-cognitive institutional elements. In addition to these challenges of time of reaching support across all institutional pillars, there are some additional challenges of congruence. We know from institutional theory that policy in part diffuses across national boundaries based on isomorphic mechanisms. To the extent that this is the case, there is a risk that new policy initiatives are only loosely coupled with the prevalent normative and cultural-cognitive elements of institutions. Prior studies have found that such loose couplings are likely to occur for example when developing countries model their entrepreneurship policies with inspiration from developed countries, and that such loose coupling is associated with poorer entrepreneurial performance. This means that for instance countries, which culturally have a preference for less openness and flexibility in collaborative structures, are unlikely to benefit from implementing networking policy inspired by more individualistic oriented countries. Problems of misalignment in institutional elements have caused lower levels of trust and less close inter-firm collaboration (Welter et al., 2005), and thus seem to reduce the quality of networking. 5.2. Limitations of the study This study has attended to the effect of networking activity in terms the number and strength of firms’ collaborative relationships. Our explanation of the effects of institutions on networking quality is based on relational rather than compositional or structural arguments. Thus, we have not examined how differential institutional contexts may similarly impact the innovation potential from collaborating with different kinds of alters, nor from occupying particular positions in the social structures surrounding the focal firm. What may also be seen as a limitation in this study is the onedimensional construct of institutional support for networking. In contrast to several recent studies, which have examined the impact of specific institutional arrangements on entrepreneurial behavior and performance (see for example Stenholm et al., 2013; Ács et al., 2014), this study has conceptualized institutions as generalized perceptions of support for networking from the regulative, normative, and cultural cognitive pillars of institutions. This conceptualization is consistent with original definitions of institutions in new institutional theory (Powell and DiMaggio, 1991; Jepperson, 1991). What we measure is the actual perceptions of an element of the institutional order in a given country. We measure this for a specific activity of interest for this study, namely networking. The advantage of this approach is that it yields an overall picture of a specific institutional order. The limitations, however, are also noteworthy. As a first limitation, the uni-dimensionality that we find for institutional support for networking, even though we consider three institutional pillars, likely arise from this conceptualization of institutions as generalized perceptions. In this way, our conceptualization may fail to disclose variation in how institutions have manifested in institutional arrangements. Such variation is important since different institutional arrangement may have differential impacts on both the quantity and quality of networking. Identifying such variations is essential from a policy perspective since institutional arrangements to a wide extent can be more

1244

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

directly influenced by human design. Future studies that attend to such differential impacts of various institutional arrangements on the quantity and quality of firms’ collaborative networking are of particular interest for policy makers engaged in shaping national systems of innovation and attempting to influence national framework conditions to increase innovation. However, there also seem to be an importance in devising future research that more explicitly deals with the complexity and interconnectedness of institutional elements. In particular, there is a need for studies that consider the impact on networking for innovation from the degree of consistency among institutional elements. Based on results from this study in combination with observations from previous studies (Welter et al., 2005) there seem to be a potential for examining how different institutional elements may either complement or undermine the effect of one another on the qualities of inter-firm networking and the further effect on innovation. As a final limitation, this study showed that institutional support for networking increases the quality of networking by providing for higher levels of general trust and better access to resources. These considerations are grounded in theory and empirical observations from other studies, and not based on specific observations from this study. Conducting large scale cross-national studies of how these and other competing mechanisms impact the quality of networking would serve to improve our understanding of how the institutional context affects innovation in inter-firm networks. Acknowledgements Data were collected by the Global Entrepreneurship Monitor. Responsibility for analysis and interpretation rests with the authors. We acknowledge hospitality at Tsinghua University’s School of Economics and Management, beneficial comments at a key-note presentation to Cicalics (China Innovation and Circles Academy—Learning, Innovation and Competence Systems) hosted by the Sino-Danish Center for Education and Research, and insightful suggestions by the anonymous reviewers. References Ács, Z.J., Autio, E., Szerb, L., 2014. National systems of entrepreneurship: measurement issues and policy implications. Res. Policy 43, 476–494. Ahuja, G., Lampert, C.M., Tandon, V., 2008. Moving beyond Schumpeter: management research on the determinants of technological innovation. Acad. Manage. Ann. 2, 1–98. Ahuja, G., 2000. Collaboration networks, structural holes, and innovation: a longitudinal study. Adm. Sci. Q. 45, 425–455. Aidis, R., Estrin, S., Mickiewicz, T., 2008. Institutions and entrepreneurship development in Russia: a comparative perspective. J. Bus. Venturing 23, 656–672. Amara, N., Landry, R., 2005. Sources of information as determinants of novelty of innovation in manufacturing firms: evidence from the 1999 Statistics Canada Innovation Survey. Technovation 25, 245–259. Andersen, K.V., 2012. The Problem of embeddedness revisited: collaboration and market types. Research Policy 41, 139–148. Angrist, J.D., Pischke, J.S., 2008. Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton University Press, Princeton. Anokhin, S., Schulze, W.S., 2009. Entrepreneurship, innovation, and corruption. J. Bus. Venturing 24, 465–476. Arundel, A., Casali, L., Hollanders, H., 2015. How European public sector agencies innovate: the use of bottom-up, policy-dependent and knowledge-scanning innovation methods. Res. Policy 44, 1271–1282. Autio, E., Ács, Z., 2010. Institutional influences on strategic entrepreneurial behavior. Strateg. Entrepreneur. J. 4, 234–251. Autio, E., Kenney, M., Mustar, P., Siegel, D., Wright, M., 2014. Entrepreneurial innovation: the importance of context. Res. Policy 43, 1097–1108. Bartholomew, S., 1997. National systems of biotechnology innovation: complex interdependence in the global system. J. Int. Bus. Stud. 28, 241–266. Baumol, W.J., Litan, R.E., Schramm, C.J., 2007. Good Capitalism, Bad Capitalism. Yale University Press, New Haven. Bell, G.G., 2005. Clusters, networks, and firm innovativeness. Strateg. Manage. J. 26, 287–295. Bodas-Freitas, I.M., Clausen, T., Fontana, R., Verspagen, B., 2011. Formal and informal external linkages and firms innovative strategies: a cross-country comparison. In: Pyka, A., da Grac¸a Derengowski Fonseca, M. (Eds.), Catching

Up, Spillovers and Innovation Networks in a Schumpeterian Perspective. Springer, Berlin, pp. 119–146. Bosma, N., 2013. The Global Entrepreneurship Monitor (GEM) and its impact on entrepreneurship research. Found. Trends Entrepreneur. 9 (2), 143–248. Burt, R.S., 2000. The network structure of social capital. Res. Organ. Behav. 22, 345–423. Chaminade, C., Intarakumnerd, P., Sapprasert, K., 2012. Measuring systemic problems in National Innovation Systems: an application to Thailand. Res. Policy 41, 1476–1488. Chen, X.P., Li, S., 2005. Cross-national differences in cooperative decision-making in mixed-motive decision contexts: the mediating effect of vertical and horizontal individualism. J. Int. Bus. Stud. 36, 622–636. Chesbrough, H.W., 2003. Open Innovation: The New Imperative for Creating and Profiting from Technology. Harvard Business Press, Boston. Chua, R.Y.J., Morris, M.W., Ingram, P., 2009. Guanxi vs networking: distinctive configurations of affect- and cognition-based trust in the networks of Chinese vs American managers. J. Int. Bus. Stud. 40, 490–508. Das, T.K., Teng, B.-S., 1998. Between trust and control: developing confidence in partner cooperation in alliances. Acad. Manage. Rev. 23, 491–512. De Clercq, D., Lim, D.S.K., Hoon Oh, C., 2011. Individual-level resources and new business activity: the contingent role of institutional context. Entrepreneur. Theory and Practice 37, 303–330. Dhanaraj, C., Parkhe, A., 2006. Orchestrating innovation networks. Acad. Manage. Rev. 31, 659–669. DiMaggio, P.J., Powell, W.W., 1983. The iron cage revisited: institutional isomorphism and collective rationality in organizational fields. Am. Sociol. Rev. 48, 147–160. Dyer, J., Singh, H., 1998. The relational view: cooperative strategies and sources of interorganizational competitive advantage. Acad. Manage. Rev. 23, 660–679. Eisenhardt, K.M., 1988. Agency- and institutional-theory explanations: the case of retail sales compensation. Acad. Manage. J. 31, 488–511. Feldman, M.S., March, J.G., 1981. Information as signal and symbol. Adm. Sci. Q. 26, 171–186. Fitjar, R.D., Rodríguez-Pose, A., 2013. Firm collaboration and modes of innovation in Norway. Res. Policy 42, 128–138. Freeman, C., 2002. Continental, national and sub-national innovation systems—complementarity and economic growth. Res. Policy 31, 191–211. Guan, J., Chen, K., 2012. Modeling the relative efficiency of national innovation systems. Res. Policy 41, 102–115. Guan, J., Zuo, K., Chen, K., Yam, R.C.M., 2016. Does country-level R&D efficiency benefit from the collaboration network structure? Res. Policy 45, 770–784. Guler, I., Guillén, M.F., Macpherson, J.M., 2002. Global competition, institutions, and the diffusion of organizational practices: the international spread of ISO 9000 quality certificates. Adm. Sci. Q. 47, 207–232. Hagen, J.M., Choe, S., 1998. Trust in Japanese interfirm relations: institutional sanctioning matters. Acad. Manage. Rev. 23, 589–600. Helfat, C.E., Peteraf, M.A., 2009. Understanding dynamic capabilities: progress along a development path. Strateg. Organ. 7, 91–102. Herstad, S.J., Aslesen, H.W., Ebersberger, B., 2014. On industrial knowledge bases, commercial opportunities and global innovation network linkages. Res. Policy 43, 495–504. Hirsch, P., 1997. Sociology without social structure: neo-institutional theory meets the brave. Am. J. Sociol. 102, 1702–1723. Hitt, M.A., Ahlstrom, D., Dacin, M.T., Levitas, E., Svobodina, L., 2004. The institutional effects on strategic alliance partner selection in transition economies: china vs. Russia. Organ. Sci. 15 (2), 173–185. Hoffman, A.J., 1999. Institutional evolution and change: environmentalism and the U.S. chemical industry. Acad. Manage. J. 42, 351–371. Hofstede, G., 1980. Culture’s Consequences: International Differences in Work-related Values. Sage, Thousand Oaks. Hovne, A.S., Hovne, B.S., Schøtt, T., 2014. Entrepreneursø innovation benefitting from their education and training and from national policy and culture: a global study. Int. J. Entrepreneur. Small Bus. 23 (1/2), 127–144, http://dx.doi. org/10.1504/IJESB.2014.065312. Huber, G.P., 2004. The Necessary Nature of Future Firms: Attributes of Survivors in a Changing World. Sage, Thousand Oaks. Huff, L., Kelley, L., 2003. Levels of organizational trust in individualist versus collectivist societies: a seven nation study. Organ. Sci. 14, 81–90. Hullova, D., Trott, P., Simms, C.D., 2016. Uncovering the reciprocal complementarity between product and process innovation. Res. Policy 45, 929–940. Jensen, K.W., Schøtt, T., 2014. Firmsø innovation embedded in their networks of collaboration: China compared to the World. J. Chin. Econ. Bus. Stud. 12 (3), 273–292, http://dx.doi.org/10.1080/14765284.2014.931427. Jensen, K.W., Schøtt, T., 2015. Start-up firms’ networks for innovation and export: facilitated and constrained by entrepreneurs’ networking in private and public spheres. J. Soc. Netw. Anal. Min. 5 (1), 1–17, http://dx.doi.org/10.1007/s13278015-0287-8 (Article 48). Jensen, K.W., Rezaei, S., Schøtt, T., Ashourizadeh, S., Li, J., 2016. Chinese entrepreneurs human and social capital benefiting innovation: in China and in the Chinese diaspora. Int. J. Bus. Glob. 16 (3), 350–377. Jepperson, R.L., 1991. Institutions, institutional effects, and institutionalism. In: Powell, W.W., DiMaggio, P.J. (Eds.), The New Institutionalism in Organizational Analysis. University of Chicago Press, Chicago, pp. 143–163.

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246 Kshetri, N., 2009. Entrepreneurship in post-socialist economies: a typology and institutional contexts for market entrepreneurship. J. Int. Entrepreneur. 7, 236–259. Larédo, P., 1998. The networks promoted by the framework programme and the questions they raise about its formulation and implementation. Res. Policy 27, 589–598. Levie, J., Autio, E., Ács, Z., Hart, M., 2014. Global entrepreneurship and institutions: an introduction. Small Bus. Econ. 42, 437–444, http://dx.doi.org/10.1007/ s11187-013-9516-6. Leyden, D.P., Link, A.N., Siegel, D.S., 2014. A theoretical analysis of the role of social networks in entrepreneurship. Res. Policy 43, 1157–1163. Lhuillery, S., Pfister, E., 2009. R&D cooperation and failures in innovation projects: empirical evidence from French CIS data. Res. Policy 38 (1), 45–57. Liefner, I., Hennemann, S., Xin, L., 2006. Cooperation in the innovation process in developing countries: empirical evidence from Zhongguancun, Beijing. Environ. Plan. A 38, 111–130. Lipparini, A., Sobrero, M., 1994. The glue and the pieces: entrepreneurship and innovation in small-firm networks. J. Bus. Venturing 9, 125–140. Lokshin, B., Hagedoorn, J., Letterie, W., 2011. The bumpy road of technology partnerships: understanding causes and consequences of partnership mal-functioning. Res. Policy 40, 297–308. Lorenz, E.H., 1999. Trust, contract and economic cooperation. Camb. J. Econ. 23, 301–315. Love, J.H., Roper, S., Bryson, J.R., 2011. Openness, knowledge, innovation and growth in U.K. business services. Res. Policy 40, 1438–1452. Lundvall, B.-Å., 1992. National Systems of Innovation: Toward a Theory of Innovation and Interactive Learning. Francis Pinter, London. Magro, E., Wilson, J.R., 2013. Complex innovation policy systems: towards an evaluation mix. Res. Policy 42 (9), 1647–1656. March, J.G., Olsen, J.P., 2006. Elaborating the new institutionalism. In: Rhodes, R.A.W., Binder, S., Rockman, B. (Eds.), The Oxford Handbook of Political Institutions. Oxford University Press, Oxford, pp. 3–20. Mezias, S., 1990. An institutional model of organizational practice: financial reporting at the Fortune 200. Adm. Sci. Q. 35, 431–457. Molina-Morales, F.X., Martinez-Fernandez, M.T., Torlo, V.J., 2011. The dark side of trust: the benefits, costs and optimal levels of trust for innovation performance. Long Range Plan. 44, 118–133. Nieto, M., Santamaria, L., 2007. The importance of diverse collaborative networks for the novelty of product innovation. Technovation 27, 367–377. Nishimura, J., Okamuro, H., 2011. Subsidy and networking: the effects of direct and indirect support programs of the cluster policy. Res. Policy 40, 714–727. North, D.C., 1990. Institutions. In: Institutional Change and Economic Performance. Cambridge University Press, Cambridge. OECD, 2010. The OECD Innovation Strategy: Getting a Head Start on Tomorrow. OECD, Paris. Obstfeld, D., 2005. Social networks, the tertius lungens orientation, and involvement in innovation. Adm. Sci. Q. 50, 100–130. Owen-Smith, J., Powell, W.W., 2004. Knowledge networks as channels and conduits: the effects of spillovers in the Boston biotechnology community. Organ. Sci. 15, 5–21. Owen-Smith, J., Riccaboni, M., Pammolli, F., Powell, W.W., 2002. A comparison of US and European university—industry relations in the life sciences. Manage. Sci. 48, 24–43. Ozman, M., 2009. Inter-firm networks and innovation: a survey of literature. Econ. Innov. New Technol. 18, 39–67. Parmigiani, A., Rivera-Santos, M., 2011. Clearing a path through the forest: a meta-review of interorganizational relationships. J. Manage. 37, 1108–1136. Partanen, J., Chetty, S.K., Rajala, A., 2011. Innovation types and network relationships. Entrepreneur. Theory Pract. 35 (3), 1–29. Phelps, C.C., 2010. A longitudinal study of the influence of alliance network structure and composition on firm exploratory innovation. Acad. Manage. J. 53, 890–913. Pittaway, L., Robertson, M., Munir, K., Denyer, D., Neely, A., 2004. Networking and innovation: a systematic review of the evidence. Int. J. Manage. Rev. 56, 137–168. Powell, W.W., Colyvas, J.A., 2008. Microfoundations of institutional theory. In: Greenwood, R., Oliver, C., Suddaby, R., Sahlin, K. (Eds.), The Sage Handbook of Organizational Institutionalism. Sage, Thousand Oaks, pp. 276–298, http://dx. doi.org/10.4135/9781849200387. Powell, W.W., Grodal, S., 2005. Networks of innovators. In: Fagerberg, J., Mowery, D.C. (Eds.), The Oxford Handbook on Innovation. Oxford University Press, Oxford. Powell, W.W., Koput, K.W., Smith-Doerr, L., 1996. Interorganizational collaboration and the locus of innovation: networks of learning in biotechnology. Adm. Sci. Q. 41, 116–145. Putnam, R.D., 2000. Bowling Alone: The Collapse and Revival of American Community. Simon & Schuster, New York. Reynolds, P., Bosma, N.S., Autio, E., Hunt, S., De Bono, N., Servais, I., 2005. Global entrepreneurship monitor: data collection design and implementation 1998–2003. Small Business Economics 24, 205–231. Rodríguez-Pose, A., Di-Cataldo, M., 2014. Quality of government and innovative performance in the regions of Europe. Papers in Evolutionary Economic Geography # 14.06. Rogers, M., 2004. Networks, firm size and innovation. Small Bus. Econ. 22, 141–153.

1245

Rosenbusch, N., Brinckmann, J., Bausch, A., 2011. Is innovation always beneficial? A meta-analysis of the relationship between innovation and performance in SMEs. J. Bus. Venturing 26, 441–457. Rost, K., 2011. The strength of strong ties in the creation of innovation. Res. Policy 40, 588–604. Rowley, T., Behrens, D., Krackhardt, D., 2000. Redundant governance structures: an analysis of structural and relational embeddedness in the steel and semiconductor industries. Strateg. Manage. Journal 21 (3), 369–386. Schøtt, T., Cheraghi, M., 2015. Gendering pursuits of innovation: embeddedness in networks and culture. Int. J. Entrepreneur. Small Bus. 24 (1), 83–116. Schøtt, T., Sedaghat, M., 2014. Innovation embedded in entrepreneurs networks and national educational systems: a global study. Small Bus. Econ. 43 (2), 463–476, http://dx.doi.org/10.1007/s11187-014-9546-8. Schøtt, T., Cheraghi, M., Rezaei, S., Vang, J., 2014. Innovation embedded in entrepreneurs networks in private and public spheres: a global study focusing on China and Denmark. Int. J. Entrepreneur. Small Bus. 231/2, 145–167, http:// dx.doi.org/10.1504/IJESB.2014.065302. Scott, W.R., 2003. Institutional carriers: reviewing modes of transporting ideas over time and space and considering their consequences. Ind. Corp. Change 12, 879–894. Scott, W.R., 2014. Institutions and Organizations: Ideas, Interests, and Identities. Sage, Los Angeles. Shane, S., 1993. Cultural influences on national rates of innovation. J. Bus. Venturing 8 (1), 59–73. Shenhav, Y., 1995. From chaos to systems: the engineering foundations of organization theory. Adm. Sci. Q. 40, 557–585. Smith, K., 2006. Measuring innovation. In: Fagerberg, J., Mowery, D.C. (Eds.), The Oxford Handbook on Innovation. Oxford University Press, Oxford, http://dx. doi.org/10.1093/oxfordhb/9780199286805.003.0006. Snijders, T.A.B., Bosker, R.J., 2012. Multilevel Analysis: An Introduction to Basic and Applied Multilevel Modeling, 2nd ed. Sage, Los Angeles. Staber, U., 2001. Spatial proximity and firm survival in a declining industrial district: the case of knitwear firms in Baden-Wurttemberg. Reg. Stud. 35, 329–341. Stam, W., Arzlanian, S., Elfring, T., 2014. Social capital of entrepreneurs and small firm performance: a meta-analysis of contextual and methodological moderators. J. Bus. Venturing 29, 152–173. Steensma, H.K., Marino, L., Weaver, K.M., Dickson, P.H., 2000. The influence of national culture on the formation of technology alliances by entrepreneurial firms. Acad. Manage. J. 43, 951–973. Stenholm, P., Acs, Z.J., Wuebker, R., 2013. Exploring country-level institutional arrangements on the rate and type of entrepreneurial activity. J. Bus. Venturing 28, 176–193. Stephan, U., Uhlaner, L.M., 2010. Performance-based vs socially supportive culture: a cross-national study of descriptive norms and entrepreneurship. J. Int. Bus. Stud. 41, 1347–1364. Suchman, M.C., 1995. 1995. Managing legitimacy: strategic and institutional approaches. Acad. Manage. Rev. 3, 571–610. Tödtling, F., Trippl, M., 2005. One size fits all? Towards a differentiated innovation policy approach. Res. Policy 34, 1203–1219. Tavassoli, S., Karlsson, C., 2015. Persistence of various types of innovation analyzed and explained. Res. Policy 44, 1887–1901. Tether, B., 2002. Who co-operates for innovation, and why: an empirical analysis. Res. Policy 31, 947–967. Tiessen, J.H., 1997. Individualism, collectivism, and entrepreneurship: a framework for international comparative research. J. Bus. Venturing 12, 367–384. Tortoriello, M., Krackhardt, D., 2010. Activating cross-boundary knowledge: the role of Simmelian ties in the generation of innovations. Acad. Manage. J. 53, 167–181. Tsai, W., Ghoshal, S., 1998. Social capital and value creation: the role of intrafirm networks. Acad. Manage. J. 41, 464–474. Tsai, K., 2009. Collaborative networks and product innovation performance: toward a contingency perspective. Res. Policy 38, 765–778. Vasudeva, G., Spencer, J.W., Teegen, H.J., 2013a. Bringing the institutional context back in: a cross-national comparison of alliance partner selection and knowledge acquisition. Organ. Sci. 24 (2), 319–338. Vasudeva, G., Zaheer, A., Hernandez, E., 2013b. The embeddedness of networks: institutions, structural holes, and innovativeness in the fuel cell industry. Organ. Sci. 24 (3), 645–663. Wang, J., 2016. Knowledge creation in collaboration networks: effects of tie configuration. Res. Policy 45, 68–80. Welter, F., Kautonen, T., Chepurenko, A., Malieva, E., Venesaar, U., 2005. Trust environments and entrepreneurial behavior: exploratory evidence from Estonia, Germany and Russia. J. Enterp. Cult. 12, 327–349. World Economic Forum, Global Entrepreneurship Research Association, 2015. Leveraging Entrepreneurial Ambition and Innovation: A Global Perspective on Entrepreneurship, Competitiveness and Development. World Economic Forum, Geneva (Accessed March 1 2016) www.gemconsortium.org. Yamakawa, Y., Yang, H., Lin, Z., 2011. Exploration versus exploitation in alliance portfolio: performance implications of organizational, strategic, and environmental fit. Res. Policy 40, 287–296. Zaheer, A., Gulati, R., Nohria, N., 2000. Strategic networks. Strateg. Manage. J. 21 (3), 203-. Zeng, S.X., Xie, X.M., Tam, C.M., 2010. Relationship between cooperation networks and innovation performance of SMEs. Technovation 30, 181–194.

1246

T. Schøtt, K.W. Jensen / Research Policy 45 (2016) 1233–1246

Zheng, W., 2010. A social capital perspective of innovation from individuals to nations: where is empirical literature directing us? Int. J. Manage. Rev. 12 (2), 151–183. Zucker, L.G., 1986. Production of trust: institutional sources of economic structure, 1840–1920. In: Staw, B.M., Cummings, L.L. (Eds.), Research in Organizational Behavior, vol. 8. JAI Press, Greenwich, CT, pp. 53–112.

Zukin, S., DiMaggio, P. (Eds.), 1990. Structures of Capital: The Social Organization of the Economy. Cambridge University Press, Cambridge.