Geography Reference
In-Depth Information
new and research-intensive productions, shifting to models closer to 'new social net-
works' (Belussi et al., 2007).
There are also examples of rising industrial clusters in some newly emerging and
rapidly changing industrial sectors such as biotechnology and multimedia (e.g. Baptista
and Swann, 1998; Fuchs and Koch, 2005; Swann and Prevezer, 1996). Many of the inno-
vations within these sectors take place in large MNEs whose locational criteria primarily
rel ect those of the 'industrial complex'. However, in situations where activities in these
sectors are geographically concentrated among small and medium i rms, they appear to
correspond most closely to the 'new social network' type of system, where inter-industry
spillovers emerge from the integration of dif erent types of network, and where the local
embeddedness of MNE competence-creating subsidiaries varies greatly among locations
(Cantwell and Piscitello, 2005). However, the recent emergence of these geographically
clustered industries means that as yet it is too early to point to a particular evolutionary
path.
At this stage, however, it must be made clear that dif erences in cluster types and also
in cluster evolutionary paths, where they exist, are not necessarily related to industrial
sectors. In particular, the technology content dichotomy mentioned in section 2 clearly
turns out to be mostly insignii cant: high-skill, high-technology sectors do not exhibit
particular cluster characteristics or evolutionary trajectories. In the case of the global
semiconductor and electronics industry, for example, much of the industry initially
emerged from oligopolistic i rms in other sectors, such as defence contracting, lighting
engineering, or radio- and telecommunications (Hall, 1998). The majority of the global
semiconductor industry, involving wafer fabrication and assembly activities, is still
largely dominated by large MNEs both in the USA, and in Europe and Asia. The loca-
tion behaviour of these i rms generally rel ects rather traditional location criteria involv-
ing a consideration of location-specii c factor costs and the transaction costs involved in
coordinating business activities over space (McCann, 1995). As such, in situations where
we observe i rms from these industries to be clustered together in space, their location-
organisation characteristics rel ect primarily those of the 'industrial complex' model
(Arita and McCann, 2002a, 2002b; McCann et al., 2002). These location-specii c sectors
emerged initially as an industrial complex, and have remained so for over i fty years. As
such, no real cluster-evolutionary path is discernible in this case.
On the other hand, there are some sub-components of the global semiconductor and
electronics industry that have emerged in quite dif erent ways, for example, the Silicon
Valley elements of this industry, which have tended to focus on semiconductor design
activities. Although the early post-war features of the Silicon Valley semiconductor
industry were mainly typical of the 'old social network' model (Hall, 1998; Saxenian,
1994), this industry initially developed during the 1970s along the lines of a 'new social
network', and has now emerged into something that is akin to a 'pure agglomeration'
model (Arita and McCann, 2000, 2004), exhibiting the supplier-dominated characteris-
tics of Pavitt's classii cation. The majority of the design innovations developed in Silicon
Valley are made possible essentially because of the miniaturisation innovations gener-
ated in the wafer processing and wafer assembly parts of the industry that are primarily
located elsewhere, as components of global value chains with large MNE l agships. As
such, the evolutionary transition of the Silicon Valley cluster has been from old social
network to new social network to pure agglomeration.
