Seite - (000178) - in Knowledge and Networks

172 processes have been identified as generative of cluster topologies from a bottom-up perspective. These processes suggest that a core–periphery structure is the most likely one to emerge in innovation networks. Owen-Smith and Powell (2004) identified geographical proximity and owner- ship as the main explanation for cluster configurations. In their view, geographical proximity facilitates informal knowledge transfers by providing trust-based chan- nels of communication. In the absence of face-to-face interactions that are perceived as secure, information needs to be formally protected. Distant knowledge transfers, therefore, are more likely to occur through codified conduits of knowledge regu- lated by property rights. These underlying processes have been shown to affect the resulting structure of the network providing different positional advantages to actors. Owen-Smith and Powell (2004), for example, estimated a model using data on the Boston biotechnol- ogy industry and showed how geographical proximity and ownership have an impact on centrality. Geographical proximity favored a process of embedding by which groups became cohesive over time. Cohesiveness benefited all members, regardless of their centrality. Conversely, in geographically sparse communities, being central is essential to success. The study by Owner-Smith and Powell showed that ownership also mattered in terms of configuration. When clusters were driven by private commercial firms, networks ties likely spread across the globe to reach key partners. Policy-anchored districts, by contrast, tended to take root in the com- munity to which they belonged, thus creating dense local networks. The dyad of local–tacit versus global–formal knowledge transfers was discussed by Bathelt, Malmberg, and Maskell (2004), who argued that both these types of interactions need to coexist. Cognitive proximity is considered in their study as the baseline mechanism of cluster topology. Based on the consideration that collabora- tion requires a certain degree of cognitive distance, local buzz is seen as a way to reduce distance and acquire familiarity with different knowledge bases. In order to maintain creative diversity within the cluster, it is simultaneously essential to have intakes of fresh information via pipelines connecting the buzz to the rest of the world. Should the buzz prevail, the cluster risks technological lock-in. Should pipelines overmultiply, the cluster risks disintegration. Finding the balance between inward- and outward- looking ties is, therefore, key to sustainable cluster development. Zooming out from individual ties, we note that the configuration resulting from the process of cognitive recombination is a core–periphery model in which the buzz represents the cohesive nucleus of the topology and the pipelines its periphery. In a study on the wine districts in Italy, Giuliani (2007) also underlined the importance of knowledge bases in producing a core–periphery network configuration. She con- sidered two distinct relations: a knowledge network built on survey data tracing technical advice, and a business network recording any type of business relations among cluster firms. A model was then estimated to understand the effect of knowl- edge bases and firms’ characteristics on degree centrality. Estimation results were reinforced by calculating a core index to locate core actors within the networks. Results showed that the two networks had very different configurations. Although the business network was complete, representing a collective and pervasive L. Prota et al.