Informational networks and innovation in mature industrial clusters

Depósito de Investigación de la Universidad de Sevilla https://idus.us.es/ This is an Accepted Manuscript of an article published by Elsevier in Journal of Business Research, Volume 66, Issue 5, May 2013, available at: https://doi.org/10.1016/j.jbusres.2012.02.043. Copyright 2013 Elsevier. En idUS Licencia Creative Commons CC BY-NC-ND Informational Networks and Innovation in Mature Industrial Clusters CRISTÓBAL CASANUEVA IGNACIO CASTRO JOSÉ LUIS GALÁN Submission: February 2009 Revision: January 2010 Acceptance: December 2011 The authors would like to offer their sincere thanks to three anonymous reviewers for so many valuable and constructive criticisms and comments. This research was supported by the Ministerio de Ciencia e Innovación, Spain (ECO2009-12742). Corresponding author: Ignacio Castro, University of Seville, Department of Business Management and Marketing, Av. Ramón y Cajal 1, 41018 Seville (Spain), Phone:+ 34 954554432, Fax:+ 34 954556989 (Email: [email protected]). 1 Abstract The position of a firm in inter-organizational networks that occur in geographic clusters can affect innovation. This study analyzes a wide range of ties and how they impact on the transmission of tacit and explicit knowledge. Thus, a firm’s position influences its innovative performance in different ways, according to the type of network -tacit or explicit- and the type of innovation -process or productunder analysis. Few empirical studies have analyzed the relation between social networks and innovation in mature geographic clusters; an appropriate context bearing in mind the contingent nature of the relations under study. The results support the idea that a central position in both tacit and explicit knowledge networks is especially significant in the case of product innovation, but that the influence of structural holes is weaker. These findings contribute to the literature on clusters and social networks by explaining how strategic management of positions in knowledge networks can improve the innovative performance of a firm. Keywords: interorganizational networks; social network analysis; geographic cluster; mature sector; explicit and tacit knowledge; product and process innovation. 2 Introduction Over the last two decades, the literature on clusters in the field of management has snowballed (Bell, 2005; Gilbert, McDougall, and Audretsch, 2008; Krätke, 2002; Maskell, 2001a; Rocha and Sternberg, 2005; Tallman, Jenkins, Henry and Pinch, 2004; Uzzi, 1997). The nature of the cluster, which generates many different formal and informal relations, has become a reference point in the study of firm networks. Many recent studies have centered on knowledge transfer processes and on the generation of innovations within the cluster (Bell, 2005; Capello, 1999; Capello and Faggian, 2005; Chiu, 2009; Cowan and Jonard, 2006; Mu, Love and Peng, 2008; Pouder and St John, 1996). However, these studies have yet to establish differences between the results of innovation and the distinct types of knowledge that support these findings. Various studies have shown that the position of a firm in a network influences performance in general, and innovative performance, in particular (Ahuja, 2000b; Bell, 2005). The significance of this influence is contingent on various factors, including the industry in which the firm operates, the type of network in the industry, the type of information transfer, and the particular performance under analysis (Ahuja, 2000b; Inkpen and Tsang, 2005; Rowley, Baum, Shipilov, Greve and Rao, 2004; Zaheer and Bell, 2005). In this paper, two research questions arise from an in-depth examination of the influence that information and knowledge flows have on the innovation processes of firms in mature geographical clusters: the question concerns the extent to which different organizational, social and personal relations, used in earlier works as a proxy for tacit and explicit knowledge transfer within a cluster (Bell and Zaheer, 2007), are linked to these information flows. The second question concerns the influence that different network positions -centrality and structural holeshave on the innovative 3 performance of firms, by considering two different types (tacit and explicit) of knowledge networks and two different (product and process) innovation results . The results of this research contribute both to the literature on clusters and on social networks. On the one hand, the study finds support for the idea that the informational content of the various types of relations between the firms in the clusters are differ, with respect to both tacit and explicit knowledge. These differences suggest a need to determine which type of network is the most appropriate for the analysis of a specific business phenomenon. On the other hand, the relation between network positions and the innovative performance of firms is contingent on the type of network and the type of innovation under study; factors that appear in very few studies, although the literature repeatedly stresses their importance. After this introduction, the following two sections present a brief review of the main points in the literature, a more specific conceptual framework for the paper and certain hypotheses concerning that framework. The paper then provides details of the methodology and the results of a quantitative empirical study of a network cluster of firms in the Spanish footwear industry. The discussion of the results leads into a summary of the main conclusions of the study and the paper ends with an acknowledgment of its limitations and suggestions for future research. Clusters, information flows and innovation Firms may be embedded in a variety of different types of networks and structures —strategic alliances, subcontracting arrangements, joint ventures, interlocking directorates, associations, cross-ownership, and so on. Each of these networks facilitates access to various different forms of information and knowledge and affect performance, especially with regard to innovation (Ahuja, 2000a; Ahuja, 2000b; Bell and Zaheer, 4 2007; Koka and Prescott, 2002; Powell, Koput and Smith-Doerr, 1996; Shan, Walker and Kogut, 1994; Stuart, 2000; Zaheer and Bell, 2005). Clusters, which form networks of firms and other institutions, are receiving growing attention in the literature on management (Bell, 2005; Gilbert et al., 2008; Huggins and Johnston, 2010; Krätke, 2002; Rocha and Sternberg, 2005; Tallman et al., 2004). A cluster is defined as a set of firms from similar or related sectors that are geographically close to each other (Harrison and Kelley, 1996; Pouder and St John, 1996); generating certain externalities that stem from economic, social and historic factors (Becattini, 1990; Marshall, 1979; Porter, 1990; Rocha, 2004). A further definition states that the cluster is a knowledge production center (Tallman et al., 2004), that is characterized by the transference of knowledge and information between its members (MacKinnon, Cumbers and Chapman, 2002). One explanation for the geographical concentration of innovative activities is that knowledge that develops in a cluster circulates more easily within that industrial district, but its transfer outside the district is much more difficult (Dahl and Pedersen, 2004). Geographic proximity facilitates the exchange of knowledge, especially tacit knowledge between firms and their employees (Bell and Zaheer, 2007). Knowledge transfer is equally important for clusters in dynamic, high-tech sectors (Capello, 1999; Chiu, 2009), as well as in traditional, mature, low technology intensity sectors (Bell, 2005; Krätke, 2002). The difference between these transfers resides in the benefits that firms can extract from that knowledge: radical product innovations, in the high-tech sectors; gradual product and process innovations, in the mature sectors (Capello, 1999; Krätke, 2002). Damanpour (1991) considered that product innovations are new products or services, introduced to meet the needs of an external user or a market, and process 5 innovations are new elements introduced into an organization’s production or service operations. Although many studies analyze the relations between innovation management and inter-firm networks, few studies have closely examined how ties of a strictly noninformational nature (such as social networks of friendship or cooperative trust) facilitate the transfer of innovative knowledge (Tsai and Ghoshal, 1998; Tsai, 2001; Yli-Renko, Autio and Sapienza, 2001). Several studies have argued that social ties are channels for the flow of information and resources, so they therefore have a significant effect on innovation capability (Batjargal, 2003; Moran, 2005; Rodan and Galunic, 2004; Tsai and Ghoshal, 1998; Zaheer and Bell, 2005). The first research question centers on knowing how these non-informative links affect the exchange of tacit and explicit information. The second research question begins with the widely accepted idea that the knowledge circulating within the cluster network is not of equal benefit to all firms (Tallman et al, 2004). The literature has shown that the central position of firms in the network and the structure of the ego network of each firm can affect its performance and more specifically, its innovative performance (Bell, 2005; Chiu, 2009). Thus, on the one hand, the literature has shown how firms with more central network positions benefit to a greater extent from knowledge flowing through the network and from improved innovation performance (Bell, 2005; Chiu, 2009; Takeda Kajikawa, Sakata and Matsushima, 2008). Nevertheless, the research also points out that this result is contingent upon certain variables, such as the nature of the network under study and the type of innovative performance (Rowley, Behren and Krackhardt, 2000; Walker, Kogut and Shan, 1997). 6 Moreover, ego-net structure of each firm also influences its performance (Gulati, 1998; Shipilov and Li, 2008). This structure reflects the density of the reciprocal ties between one firm and all the other firms in a cluster. The so-called structural holes appear, if the ego network of a firm is not very dense. Burt (1992) contends that the absence of direct ties between some firms in a network (structural holes) implies an asymmetrical access to information within the network that actually provides some firms with greater innovative capacity. Supporters of this view argue that the existence of structural holes facilitates access (for some firms) to innovative information in remote parts of the network (Burt, 1992; Shipilov, 2006; Ven, 1986; Walter, Lechner and Kellermanns, 2007). This study explores the way in which the centrality of a firm in the network influences innovation performance in different types of networks (tacit and explicit knowledge networks). Different types of innovation (product and process) are therefore of great interest. However, as in the earlier case, the explanation of these divergent views may relate more to the contingent nature of the context under analysis. Contingent factors include: (i) the type of industry and its stage of development (Rowley et al., 2000; Walker et al., 1997); (ii) the internal capabilities of the firms (Zaheer and Bell, 2005); (iii) the type of network that is under study (Inkpen and Tsang, 2005); (iv) the amount of knowledge existing in a network (Cowan and Jonard, 2006); (v) the nature and content of the relations within the network (Ahuja, 2000b); and (vi) the architecture of the network or structural embeddedness (Rowley, 2000; Rost, 2011). Hypotheses The informative content of social and market ties Studies of networks in clusters often consider different types of relations between firms (commercial, directive, subcontracting,…) and assume that these links imply a 7 transference of knowledge, even though the informational content of the links that firms establish within the cluster have not been analyzed. Thus, Bell (2005) distinguishes between managerial networks (characterized by the transmission of tacit knowledge) and institutional networks (characterized by the transference of explicit knowledge); a very important distinction, in order to understand business performance, especially in terms of innovation (Nonaka, 1991; Nonaka and Takeuchi, 1995; Polanyi, 1967). A distinction therefore emerges between explicit and tacit knowledge networks within clusters to explore the first research question: the extent to which network types (commercial, friendship, cooperation…) imply either tacit or explicit knowledge transfer. These networks may be distinguished by the strength of their ties. Uzzi (1996, 1997) refers on the one hand, to a market-based or arm’s-length relationship and, on the other, to special or close relationships, such as social and embedded ties. He refers to the different force of the ties, which should be understood as a dimension (Granovetter, 1985; Hansen, 1999). The two extremes of each dimension would be the market or commercial ties as the weakest links (arm’s length ties) and the purely social ties (friendship and family) as the strongest ties. Trust-based relationships are basically social and not economic ones, such that they would be seen as strong ties from that perspective. Subcontracting and cooperative relations are situated at an intermediary position between social and economic ties. Following the ideas of Nohria and Garcia- Pont (1991) and Contractor and Lorange (1988), it may be understood that cooperative relations are stronger than subcontracting ones. The different types of ties affect the transfer of different types of knowledge. Uzzi (1996:678) points out that “informative exchange in embedded ties is more proprietary and more tacit than information exchange at arm’s-length”. It may therefore be affirmed that strong social ties are more appropriate for the transfer of tacit knowledge and weak 14 explicit knowledge networks belonging to mature clusters, where they develop a traditional activity, of little technological complexity in terms of process-related activities (Porter, 1990; Pouder and St John, 1996). Therefore, differentiation in this type of cluster between firms takes place through product innovation. In relation to process innovation, the cluster will try to achieve rapid improvements in its processes through central positions in the explicit knowledge networks, the tacit networks acquiring greater relevance in those innovations that require more complex knowledge (Pérez-Luño, Cabello, Carmona and Cuevas, 2011). Hypothesis 2b: The more central the position in the tacit and explicit knowledge networks within a mature regional cluster, the greater its process innovation, nevertheless that influence should be higher in the tacit knowledge network. The influence of structural holes on product and process innovation Coleman (1988) points out that the greater the cohesion of the firm’s network, the greater the potential benefits for the firm. In contrast, Burt (1992) argues that favorable brokering or entrepreneurial positions in a network can secure important advantages for firms, in positions that exploit their privileged relationships with colleagues that have no direct ties with one another. According to this view, the absence of direct ties between some firms in a network (so-called structural holes) implies an asymmetry of access to information within the network. Only some firms in the network will be able to access innovative information in remote parts of the network, and those that do, will be able to exploit this information and so generate greater innovation (Burt, 1992; Houghton, Smith and Hood, 2009; Koka and Prescott, 2008; Sapsed, Grantham and DeFillippi, 2007; Shipilov, 2006; Ven, 1986). 15 A corollary view is that the firms that play an intermediary role in the network will be in a better position to access novel information and knowledge and will therefore achieve better innovation results (Zaheer and Bell, 2005). In this regard, although the results of some empirical studies are contradictory (Ahuja, 2000a), the majority of investigations have identified a positive relationship between the innovative performance of a focal firm and its status as an intermediary in a network (Finlay and Coverdill, 2000; Hargadon and Sutton, 1997; McEvily and Zaheer, 1999; Zaheer and Bell, 2005). An interesting question is whether a favorable brokerage position is more important in an explicit rather than a tacit knowledge network, in the context of a geographical cluster, where firms are very close to each other. Different groups or components are identifiable within a cluster or inter-organizational network; as the tacit knowledge within each group becomes more complex, the group will transfer such knowledge to other groups within the cluster with greater difficulty (Tallman et al, 2004). The literature suggests that tacit knowledge is much less abundant than explicit knowledge in a cluster. Thus, tacit knowledge makes information sticky which tears open holes (structural holes) in the flow of information between groups (Burt, 2008). However, codified knowledge spreads more easily in a mature regional cluster formed of smaller firms (Pinch, Henry, Jenkins and Tallman, 2003) and, therefore, explicit knowledge in a geographic cluster can quickly become both a local and a collective resource and it is therefore very difficult for structural holes to emerge (Capello, 1999). By definition, exploratory learning has something to do with the intermediary or brokerage positions of the firms within a network (Burt, 1992); these structural holes provide access to a novel knowledge (Adler and Kwon, 2002). Process innovation requires the exploitation of learning, which feeds on exchanges of internal knowledge 16 through narrower or more intense channels. Accordingly, the assumption is that firms that occupy structural holes within a knowledge network will certainly lean more towards product rather than process innovation (Lewin, Long and Carroll, 1999; Rothaermel and Deeds, 2004). Hypothesis 3: In a mature regional cluster, the structural holes spanned by a firm in the tacit knowledge network will have a positive impact on product innovation. Embeddedness in a specific environment context: exploitation versus exploration Finally, we move to our general empirical context: a regional mature cluster made up of small and medium-sized firms dedicated to footwear manufacturer. As we uphold in the previous hypotheses and as many other researchers have postulated: on the one hand, centrality is better for performance, when high levels of exploitative behavior pervade the environment, such as in the steel industry or footwear manufacture; and, on the other hand, structural holes tightly linked to weak ties improve performance, when the environment demands high levels of explorative behavior, as in the semiconductor industry (Rost, 2011; Rowley et al., 2000). In this regard, Rowley et al. (2000) argued that, on the one hand, firms need to occupy structural holes in exploration, which means that they can become intermediaries between alternative partners, that represent the opportunities that will allow the firm to become permanently adapted to its environment; on the other hand, exploitation, as we have affirmed in the earlier hypotheses, is linked to process innovation, which requires deeper specific knowledge of the firm’s current direction. This type of knowledge is acquired to a great extent by those firms that occupy more central positions, since these ties involve rich exchanges of internal and customized knowledge (Granovetter, 1992; Kogut and Singh, 1988). Therefore, we can conclude that centrality is more important than structural holes 17 for the purpose of exploitation (Rost, 2011; Rowley et al., 2000). In this paper, a stable environment characterizes the cluster, in which there is less uncertainty about future directions and disturbance in the environment and where, a priori, process innovation is the dominant innovation in the mature sectors (Porter, 1990). Therefore, most firms in this context will invest more in exploitation than in exploration. Hypothesis 4: Centrality is better than structural holes for both product and process innovation when the environment demands high levels of exploitation behavior, such as in a mature regional cluster. The following table (1) details all the hypotheses in this empirical study. Table 1 here Empirical study Research setting The present study examines a traditionally-segmented industrial cluster a description of which can be made with a reasonable degree of confidence: the geographically localized footwear cluster in the region of Valverde (Southern Spain) (Santamaría, Giner and Fuster, 2004; Schmitz, 1999). The network consists of 52 small and medium-sized firms with similar structural characteristics such as employee numbers (Street and Cameron, 2007), of which 33 are manufacturers and 19 are auxiliary firms. This context is an appropriate one in which to study the research question, insofar as many researchers highlight the importance of innovation in mature industrial clusters (Geroski, 1995; Zaheer and Bell, 2005). Data collection 18 The data collection method consisted of personal interviews of varying lengths with representatives from the 52 firms in the sample. The interviews lasted between 30 minutes and three hours depending on the length of the conversations. Owing to the size of each firm (number of employees: Mean=6.92; SD= 8.891), only the chief executive officer or the owner was present at the interview, as these two positions are more likely than any others to be able to supply basic data and information on the personal and organizational ties, simultaneously. The data comprises three classes of information: (i) the characteristics of the firm; (ii) the types of relationships (friendship, trust, tacit and explicit information exchange, cooperation, commercial ties and subcontracting) in which the firms were engaged in the network; and (iii) the opinions held by some firms with regard to others in the network. Each firm received a questionnaire on its links with the other firms in the same cluster, in order to study these relationships. Measures a) Matrices The study required the construction of the following matrices: ∙ EXPLICIT matrix: the transmission of explicit knowledge—such as information on product prices, production inputs, new market models, machinery, suppliers and clients, and so on; ∙ TACIT matrix: the exchange of tacit knowledge—such as ‘know-how’, new ways of working, new models that are under development, the firm's productive processes, internal production costs and so on; ∙ COMMERCIAL matrix: standard trading relations that develop between the firms; ∙ TRUST matrix: whether one firm trusts another; 19 ∙ COOPERAT matrix: relationships of cooperation within the sector—including cooperation to conduct purchases, carry out joint production, access distribution channels, share tools or machinery, etc.; ∙ SUBCONTR matrix: vertical subcontracting relationships between supporting businesses; ∙ FRIENDS matrix: ties of friendship between the owner and/or chief executive officer of the firms (established on the basis of the opinions of two interviewees, including only those relationships both interviewees reciprocally describe as ‘friendship’ - a symmetrical matrix); and ∙ KINSHIP matrix: family ties between the owners and the main directors of the firms. Each matrix has 52 rows and columns. The cells take a value of 1 if they show evidence of these interactions, and a value of 0, if otherwise. A social network analysis software package UCINET VI analyzed the relational data (Borgatti, Everett and Freeman, 2002). b) Dependent Variable Two indicators measure the dependent variable in the study: (i) product innovation (INNOVPRODUT); and (ii) process innovation (INNOVPROC) (Damanpour, 1991). Obtaining objective assessments of innovation levels is difficult when dealing with a wide range of firms, products, and productive systems -as is the case in this cluster of footwear companies. Under these circumstances, participant firms are likely to provide highly subjective assessments of their own levels of innovation. Bell (2005) recommends a reference to external informants, to avoid this problem of subjectivity. Thus, two groups of external assessors received invitations to measure degrees of innovation for the current study: 20 ∙ Members of the network: The question for the interviewees concerned innovation in all firms except their own. Two indicators (one for product and another for process innovation) reflect the number of times the interviewees marked each firm as innovative. ∙ Industry experts: Following the recommendation of Zaheer and Bell (2005), three experts assessed the innovativeness of each firm on a 5-point Likert-type scale with a view to checking the validity of the interviewees’ (members of the network) innovation evaluations. Their mean scores represent the levels of product innovation and process innovation for each firm. Finally, a principal components factor analysis with varimax rotation developed a composite measure for each of the innovativeness performance constructs: product and process innovation (INNOVPRODUT and INNOVPROC). c) Independent Variables An analysis of the principal interaction requires the study of the exchange of tacit and explicit knowledge between firms. Thus, the matrices of explicit (EXPLICIT matrix) and tacit knowledge (TACIT matrix) serve to define the indicators of centrality and structural holes. The number of relationships provides a measure for the centrality of network positions. This indicator aims to capture the number of direct relationships that a firm has with other firms either in the network (Koka and Prescott, 2002) or in a geographical cluster (Krätke, 2002a). In theory, if the relationships are unidirectional, a distinction is possible between in-degree centrality (the number of ties going into a node) and out-degree centrality (the number of ties going out of a node). In the present study, the use of in-degree centrality (INDEGREE) is because the flow of information that enters an organization is more relevant in an assessment of innovation. 21 Burt (1992) develops the concept of constraint (CONSTR) for the measurement of structural holes. A high level of constraint indicates that colleagues in the firm have a dense connection to each other, whereas a lower value of constraint indicates that the connection of the firm’s own network is sparse (Bell and Zaheer, 2007; Borgatti et al., 2002; Burt, 1992; Koka and Prescott, 2002). d) Control Variables The control variables in this study are age, size, and internationalization of the firm, using measures such as the number of years since the creation of the firm, the number of employees and the presence in or absence from international markets. With respect to the age of a firm, some researchers note that older firms are likely to be less innovative than younger firms (Ahuja, 2000b; Zaheer and Bell, 2005). Others establish a positive relationship between size of the firm and its innovativeness (Bell, 2005; Shipilov, 2006). Finally, other researchers analyze how the internationalization of firms affects the extent of their innovation (Kafouros, Buckley, Sharp and Wang, 2008). Data analysis Tichy, Tushman, and Fombrun (1979) recommend that Social Network Analysis (SNA) should examine the elements in the network and the various types of relationships that exist between them. The UCINET VI program performed all necessary calculations and data processing (Borgatti et al., 2002). Firstly, following the recommendations of (Krackardt, 1987), a Quadratic Assignment Procedure (QAP) was applied to compare a matrix as the dependent variable (relational data) with one or more matrices as independent variables, in order to analyze the information content of different types of organizational and personal networks. 22 An Ordinary Least Squares (OLS) regression estimated the significance of network positions in relation to innovation. Four models separately analyzed the influence of the main effect variables. Model 0 examined the influence of the control variables, size, age and internationalization on innovative performance. The main effect variables (Centrality and Structural Holes) were added to Model 1 from the network of tacit knowledge. The main effect variables (Centrality and Structural Holes) from the network of explicit knowledge were added to Model 2. Finally, the main effect variables from both networks (i.e. the full model) were jointly analyzed in Model 3. Results The informative content of social and market ties Table 2 sets out the principal indicators of the various relational structures among the 52 firms in the cluster for the analysis of the first research question. Its contents show that some types of networks have similar structures, whereas others are quite different. Table 2 here In response to the first research question, a Multiple Regression Quadratic Assignment Procedure (MRQAP) analysis of the data in the matrices gave the results shown in Tables 3 and 4, on the exchange of tacit and explicit knowledge, respectively. The sequential introduction of each of the matrices (commercial, cooperation, trust...) into consecutive models tested the degree to which the explained variance increased in each new model. Table 3 here Table 3 shows that R2 in the models used to test hypothesis H1a varied from 0.318 (for the commercial relationship as the only independent variable) to 0.402 (with the incorporation of all the six matrices). The final model shows a clear association between 23 the flow of explicit knowledge and commercial relationships, trust relationships, cooperation relationships, and friendship relationships. The greatest influence on the flow of explicit knowledge has an association with commercial relationships. Table 4 here Table 4 shows significant associations between the exchange of tacit knowledge and the relationships of trust, cooperation, and friendship. In contrast, commercial relationships, kinship links, and subcontracting relationships do not adequately explain the exchange of tacit knowledge. Cooperative relationships have the strongest association with the exchange of tacit knowledge. These results support hypothesis H1b. Network positions Considering the hypotheses, the following points arise in any analysis of the second research question. The results of a correlation matrix and the descriptive statistics, in Table 5, were as expected. On the one hand, extreme values are evident in the case of innovations (Koka and Prescott, 2002): firms were either very innovative or hardly innovative at all. On the other hand, a negative correlation is evident between measurements of centrality and structural holes, owing to differences in the nature of the skills and capabilities that firms need in order to benefit from such positions. Hence, a firm is unlikely to benefit simultaneously from both positions (Koka and Prescott, 2008). Table 5 here Finally, a high correlation is evident between the two dependent variables (INNOVAPRODUT and INNOVAPROC). Nevertheless, two reasons led us to analyze them separately: in the first place, as analyzed in the theoretical section, because they require a completely different type of learning knowledge; in second place, a Chow test 30 networks can be a relevant factor when considering improvements to the innovative performance of the firm (Ahuja, 2000b; Bell, 2005). Limitations and Future Research As with all research, the present study has certain limitations. Nevertheless, acknowledgment of these limitations does point the way towards future avenues of productive research. The first limitation of the study is its focus on a single cluster of firms, which restricts any generalization of the findings to other empirical contexts. Therefore, similar studies in a variety of contexts are advisable to complement the findings of this study. Nevertheless, the results underline the contingent relation between network positions and innovation; studies in other contexts would help to define the nature and meaning of that contingent relation. Secondly, the present study assesses the relationships of the cluster at a particular point in time. Future lines of research could propose longitudinal studies of the ways in which different relationships evolve, which would imply a study of the relations of causality between the variables. Finally, the interactions between the variables in the present study could serve as the independent variables in further studies. Future research could consider other variables, as well as other types of inter-organizational relations, for a fuller explanation of the relations under investigation in the present study. 31 Table 1 Hypothesis associated with four combinations of knowledge complexity, type of innovation and network positions (based on (Hansen, 1999)). NETWORK POSITIONS Centrality Structural Holes KNOWLEDGE NETWORKS EXPLICIT AND TACIT Type of innovation Product innovation H2a H3 Process innovation H2b Mature regional cluster H4 Table 2: Structural characteristics of the networks under analysis Density Neighborhood average size Degree (outdegree/ indegree) Betweenness Isolated nodes Components The size of main component Clustering Coefficient COMMERCIAL 22.9 11.65 70.66/22.68 7.26 1 1 51 0.33 TRUST 9.8 4.98 48.02/16.03 22.99 1 1 51 0.26 COOPERAT* 3.1 1.57 13.10 7.28 24 2 26 0.29 FRIENDS* 11.1 5.65 37.41 27.99 2 1 50 0.31 KINSHIP* 2.4 1.23 5.65 1.44 16 7 10 0.23 SUBCONTR 4.9 2.5 22.99/38.98 0.73 7 1 45 0.09 EXPLICIT 29.4 14.98 54.01/30.02 6.71 0 1 52 0.36 TACIT 4.3 2.21 13.57/13.57 15.96 16 1 36 0.16 * symmetrical matrix VARIABLES: Density measurements: number of real ties divided by the number of possible ties Average neighborhood size: average number of firms with which each firm in the cluster establishes relations Degree centralization: the percentage to which the structure of the network has a totally centralized form, differentiating between in-degree and out-degree centrality in the directional relations Betweeness centralization: shows the level at which the possibility exists of exploiting the broker position due to the elements linked to each other along the shortest paths Number of isolated firms: elements not connected to any others Number of components: groupings of two or more connected elements Size of the main component: the most numerous component Clustering coefficient: shows the propensity of the network to form ties around cohesive groups Table 3: MRQAP results for explicit knowledge network EXPLICIT Full model Sequential model Beta Std. Sign. Mod. 1 Mod. 2 Mod. 3 Mod. 4 Mod. 5 COMMERCIAL 0.51 0.00 0.56*** 0.51*** 0.51*** 0.51*** 0.51*** TRUST 0.13 0.00 0.17*** 0.14*** 0.13*** 0.13*** COOPERAT 0.00 32 0.17 0.21*** 0.18*** 0.17*** FRIENDS 0.11 0.00 0.10*** 0.11*** KINSHIP 0.03 0.06 0.03 SUBCONTR 0.00 0.47 R2 0.40 0.31 0.34 0.39 0.40 0.40 ∆ R2 0.31 0.02 0.04 0.01 0.00 Table 4: MRQAP results for tacit knowledge network TACIT Full model Sequential model Beta Std. Sign. Mod. 1 Mod. 2 Mod. 3 Mod. 4 Mod. 5 COMMERCIAL 0.02 0.14 0.08*** 0.02 0.02 0.02 0.02 TRUST 0.12 0.00 0.20*** 0.13*** 0.12*** 0.12*** COOPERAT 0.43 0.00 0.47*** 0.43*** 0.44*** FRIENDS 0.13 0.00 0.14*** 0.14*** KINSHIP - 0.00 0.48 -0.00 SUBCONTR 0.01 0.19 R2 0.28 0.00 0.04 0.26 0.28 0.28 ∆ R2 0.00 0.04 0.22 0.01 0.00 33 Table 5 Means, standard deviation, and correlations Variable Mean Standard deviation 1. 2. 3. 4. 5. 6. 7. 8. 1. Innovprodut .00 1.00 2. Innovproc .00 1.00 .59** 3. Indegreetacit 2.2 2.26 .47** .30* 4. Indegreeexplic 14.9 6.98 .54** .48** .58** 5. Constrainttacit .28 .39 -.32 -.23 -.64** -.49** 6. Constraintexplic .04 .016 -.033 .048 -.10 -.17 .34* 7. Size 6.9 8.97 .57** .62** .14 .31* -.22 .02 8. Age 39.6 23.86 .12 -.08 .25 .22 -.06 -.25 .14 9. International .3 .48 .51** .43** .21 .39** -.03 .15 .35* -.08 34 Table 6: Chow test centrality and structural holes TACIT NETWORK EXPLICIT NETWORK VARIABLES F. Sig. F Sig. CENTRALITY 22.80 .000 132.58 .000 STRUCTURAL HOLES 12.78 .001 211.38 .000 Table 7: Analysis of network positions (product innovation) DEPENDENT VARIABLE PRODUCT INNOVATION Variables Model 0 Controls only Model 1 Adding main effects Model 2 Adding main effects Model 3 Full model B B B Size .43** .44** .38** .41** Age .13 .07 .05 -.02 International .34** .22 .24* .04 NETWORK OF TACIT KNOWLEDGE Centrality .52** .47** Structural Holes .13 .30 NETWORK OF EXPLICIT KNOWLEDGE Centrality .31* .47* Structural Holes -.02 .03 Statistics Value Sign Value Sign Value Sign Value Sign R2 .40 0.00 .54 0.00 .52 0.00 .66 0.00 N 52 52 52 52 p<0.10; * p<0.05; ** p<0.01;*** p<0.001 35 Table 8: Analysis of network positions (process innovation) DEPENDENT VARIABLE PROCESS INNOVATION Variables Model 0 Controls only Model 1 Adding main effects Model 2 Adding main effects Model 3 Full model B B B Size .56*** .62** .51*** .59*** Age -.09 -.18 -.17 -.29 International .20* .07 .07 -.11 NETWORK OF TACIT KNOWLEDGE Centrality .29 .24 Structural Holes .10 .29 NETWORK OF EXPLICIT KNOWLEDGE Centrality .34** .50** Structural Holes -.02 .02 Statistics Value Sign Value Sign Value Sign Value Sign R2 .40 0.00 .44 0.00 .52 0.00 .58 0.01 N 52 52 52 52 36 References Adler, P. S., & Kwon, S. (2002). Social capital: Prospects for a new concept. Acad Manage Rev, 27, 17. Ahuja, G. (2000a). Collaboration networks, structural holes, and innovation: A longitudinal study. Adm Sci Q, 45, 425. Ahuja, G. (2000b). The duality of collaboration: Inducements and opportunities in the formation of interfirm linkages. Strateg Manage J, 21, 317. Arikan, A. T. (2009). Interfirm knowledge exchanges and the knowledge creation capability of clusters. Acad Manage Rev, 34, 658-676. Bathelt, H., Malmberg, A., & Maskell, P. (2004). Clusters and knowledge: Local buzz, global pipelines and the process of knowledge creation. Prog Hum Geog, 28, 31-56. Batjargal, B. (2003). Social capital and entrepreneurial performance in Russia: A longitudinal study. Organ Stud, 24, 535. Becattini, G. (1990). The Marshallian industrial district as a socio-economic notion. In: F. Pyke, G. Becattini& W. Sengenberger (Eds.), Industrial Districts and Inter-Firm Co-Operation in Italy (pp. 37- 51). Geneva: International Institute for Labor Studies. Bell, G. G. (2005). Clusters, networks, and firm innovativeness. Strateg Manage J, 26, 287. Bell, G. G., & Zaheer, A. (2007). Geography, Networks, and Knowledge Flow. Organ Sci, 18, 955. Borgatti, S., Everett, M., & Freeman, L. C. (2002). Ucinet 6 for Windows: Software for Social Network Analysis. Analytic Technologies: Natick. . Burt, R. S. (2008). Information and structural holes: comment on Reagans and Zuckerman. Ind Corp Change, 17, 953-969. Burt, R. S. (1992). Structural Holes: The Social Structure of Competition. Cambridge, MA: Harvard University Press. Cantner, U., & Graf, H. (2006). The network of innovators in Jena: An application of social network analysis. Res Policy, 35, 463-480. Capello, R. (1999). Spatial transfer of knowledge in high technology Milieux: Learning versus collective learning processes. Reg Stud, 33, 353. Capello, R., & Faggian, A. (2005). Collective Learning and Relational Capital in Local Innovation Processes. Reg Stud, 39, 75. Casanueva, C., & Galan, J.L. (2004). Social and Information Relations in Networks of Small and Medium-Sized Firms, M@n@gement, 7: 215-238. Chiu, Y. T. H. (2009). How network competence and network location influence innovation performance. J Bus Ind Mark, 24, 46. Cohen, W. M., & Levinthal, D. A. (1990). Absorptive Capacity: A New Perspective On Learning And Inno. Adm Sci Q, 35, 128. Coleman, J. S. (1988). Social Capital in the Creation of Human Capital. Am J Sociol, 94, S95. 37 Contractor, F.J., & Lorange, P. (1988). Cooperative strategies in international business. Lexington: Lexington Books. Cowan, R., Jonard, N. (2006). Innovation networks and the distribution of knowledge. Dahl, M. S., & Pedersen, C. Ø. R. (2004). Knowledge flows through informal contacts in industrial clusters: myth or reality? Res Policy, 33, 1673-1686. Damanpour, F. (1991). Organizational Innovation: A Meta-Analysis of Effects of Determinants and Moderators. Acad Manage J, 34, 555. Dyer, J. H., & Nobeoka, K. (2000). Creating and managing a high-performance knowledge-sharing network: The Toyota case. Strateg Manage J, 21, 345. Ettlie, J. E., & Reza, E. M. (1992). Organizational Integration and Process Innovation. Acad Manage J, 35, 795-827. Finlay, W., & Coverdill, J. E. (2000). Risk, opportunism, and structural holes: How headhunters manage clients and earn fees. Work Occupation, 27, 377. Freel, M. S., & Harrison, R. T. (2006). Innovation and cooperation in the small firm sector: Evidence from ‘Northern Britain’. Reg Stud, 40, 289-305. Geroski, P. (1995). Markets for technology: knowledge, innovation and appropriability. In: Stoneman P. (Ed.), Handbook of the Economies of Innovation and Technological Change (pp. 90-131). Oxford U.K.: Basil Blackwell. Gilbert, B. A., McDougall, P. P., & Audretsch, D. B. (2008). Clusters, knowledge spillovers and new venture performance: An empirical examination. J Bus Venturing, 23, 405-422. Gopalakrishnan, S., & Damanpour, F. (1997). A review of innovation research in economics, sociology and technology management. Omega, 25, 15-28. Gopalakrishnan, S., Bierly, P., & Kessler, E. H. (1999). A reexamination of product and process innovations using a knowledge-based view. J High Technol Manage Res, 10, 147-166. Granovetter, M. (1985). Economic Action and Social Structure: The Problem of Embeddedness. Am J Sociol, 91, 481-510. Granovetter, M. (1992). Problems of explanation in economic sociology. In: N. Nohria& R. G. Eccles (Eds.), Networks and Organizations: Structure, Form, and Action (pp. 25-56). Boston: Harvard Business School Press. Gulati, R. (1998). Alliances and networks. Strateg Manage J, 19, 293. Gulati, R., Nohria, N., & Zaheer, A. (2000). Strategic networks. Strateg Manage J, 21, 203. Hansen, M. T. (1999). The search-transfer problem: The role of weak ties in sharing knowledge across organization subunits. Adm Sci Q, 44, 82. Hargadon, A., & Sutton, R. I. (1997). Technology brokering and innovation in a product development firm. Adm Sci Q, 42, 716. Harrison, B., & Kelley, M. R. (1996). Innovative firm behavior and local milieu: Exploring the intersection of agglomeration, firm.. Econ Geogr, 72, 233. 38 Henderson, R. M., & Clark, K. B. (1990). Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of Established Firms. Adm Sci Q, 35, 9-30. Houghton, S. M., Smith, A. D., & Hood, J. N. (2009). The influence of social capital on strategic choice: An examination of the effects of external and internal network relationships on strategic complexity. J Bus Res, 62, 1255-1261. Huggins, R., & Johnston, A. (2010). Knowledge flow and inter-firm networks: The influence of network resources, spatial proximity and firm size. Entrep Region Dev, 22, 457-484. Inkpen, A. C., & Tsang, E. W. K. (2005). Social Capital, Networks, and Knowledge Transfer. Acad Manage Rev, 30, 146. Jaffe, A. B., Trajtenberg, M., & Henderson, R. (1993). Geographic localization of knowledge spillovers as evidenced by patent citations. Q J Econ, 108, 577. Kafouros, M. I., Buckley, P. J., Sharp, J. A., & Wang, C. (2008). The role of internationalization in explaining innovation performance. Technovation, 28, 63-74. Katila, R., & Ahuja, G. (2002). Something Old, Something New: a Longitudinal Study of Search Behavior and New Product Introduction. Acad Manage J, 45, 1183-1194. Kogut, B., & Singh, H. (1988). The Effect Of National Culture On The Choice Of Entry Mode. J Int Bus Stud, 19, 411. Kogut, B. (2000). The network as knowledge: Generative rules and the emergence of structure. Strateg Manage J, 21, 405. Kogut, B., & Zander, U. (1992). Knowledge of the Firm, Combinative Capabilities, and the Replication of Technology. Organ Sci, 3, 383-397. Koka, B. R., & Prescott, J. E. (2008). Designing alliance networks: the influence of network position, environmental change, and strategy on firm performance. Strateg Manage J, 29, 639. Koka, B. R., & Prescott, J. E. (2002). Strategic alliances as social capital: A multidimensional view. Strateg Manage J, 23, 795. Kraatz, M. S. (1998). Learning by association? Interorganizational networks and adaptation to environmental change. Acad Manage J, 41, 621. Krackardt, D. (1987). QAP partialling as a test of spuriousness. Soc Networks, 9, 171-186. Krätke, S. (2002). Network Analysis of Production Clusters: The Potsdam/Babelsberg Film Industry as an Example. Eur Plan Stud, 10, 27-54. Levinthal, D. A., & March, J. G. (1993). The Myopia of Learning. Strateg Manage J, 14, 95-112. Lewin, A. Y., Long, C. P., & Carroll, T. N. (1999). The Coevolution of New Organizational Forms. Organ Sci, 10, 535-550. Liu, J., & Chaminade, C. (2010). Dynamics of a Technological Innovator Network and Its Impact on Technological Performance. Innov-Manag Policy P, 12, 53-74. MacKinnon, D., Cumbers, A., & Chapman, K. (2002). Learning, innovation and regional development: a critical appraisal of recent debates. Prog Hum Geog, 26, 293-311. 39 March, J. G. (1991). Exploration and Exploitation in Organizational Learning. Organ Sci, 2, 71-87. Marshall, A. (1979). Principles of Economics. London: Macmillan. Maskell, P. (2001a). Knowledge Creation and Diffusion in Geographic Clusters. Int J Innovation Manage, 5, 213. Maskell, P. (2001b). Towards a knowledge-based theory of the geographical cluster. Ind Corp Change, 10, 921-943. McEvily, B., & Zaheer, A. (1999). Bridging ties: A source of firm heterogeneity in competitive capabilities. Strateg Manage J, 20, 1133. Moran, P. (2005). Structural vs. relational embeddedness: social capital and managerial performance. Strateg Manage J, 26, 1129. Mu, J., Love, E., & Peng, G. (2008). Interfirm networks, social capital, and knowledge flow. Journal of Knowledge Management, 12, 86. Nohria, N, & Garcia-Pont, C. (1991). Global Strategic Linkages and Industry Structure. Strateg Manage J, 12, 105-124. Nonaka, I. (1991). The Knowledge-Creating Company. Harvard Bus Rev, 69, 96. Nonaka, I., & Takeuchi, H. (1995). The Knowledge-Creating Company. New York: Oxford Universtiy Press. Pérez-Luño, A., Cabello Medina, C., Carmona Lavado, A., & Cuevas Rodríguez, G. (2011). How social capital and knowledge affect innovation. J Bus Res, 64, 1369-1376. Pinch, S., Henry, N., Jenkins, M., & Tallman, S. (2003). From 'industrial districts' to 'knowledge clusters': a model of knowledge dissemination and competitive advantage in industrial agglomerations. Journal of Econ Geogr, 3, 373. Polanyi, M. (1967). The Tacit Dimension. New York: Double Day. Porter, M. E. (1990). The Competitive Advantage of Nations. New York: Free Press. Pouder, R., & St John, C. H. (1996). Hot spots and blind spots: Geographical clusters of firms and innovation. Acad Manage Rev, 21, 1192. Powell, W. W. (1990). Neither Market nor Hierarchy: Network Forms of Organization. Res Organ Behav, 12, 295. 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. Rocha, H. O. (2004). Entrepreneurship and Development: The Role of Clusters. Small Bus Econ, 23, 363. Rocha, H. O., & Sternberg, R. (2005). Entrepreneurship: The Role of Clusters Theoretical Perspectives and Empirical Evidence from Germany. Small Bus Econ, 24, 267. Rodan, S., & Galunic, C. (2004). More than Network Structure: how Knowledge Heterogeneity Influences Managerial Performance and Innovativeness. Strateg Manage J, 25, 541. Rost, K. (2011). The strength of strong ties in the creation of innovation. Res Policy, 40, 588-604.