Geography Reference
In-Depth Information
of communities of practises that do not necessarily require a spatial
dimension, and where instead a strong role is played by relational and
cognitive proximity (Boschma 2005).
Some empirical analyses have shown that inter-industry knowledge
spillovers are likely to become more intense in centres of technological
excellence where spillovers seem to operate mainly through exchanges in
and around core technological systems (see also Crozet et al. 2004). This is
the case where spillovers are primarily rooted in 'general purpose technol-
ogies' such as, for instance, background engineering, mechanical methods,
new materials, electronics and ICTs, and the additional spillovers which
these create between actors in quite separate alternative fields of speciali-
zation. These types of centres of excellence - which correspond to the 'first
order' regions discussed in Chapter 4 - exhibit inter-industry technology
spillovers, and are most likely to be classified either as pure agglom-
erations or competence-based social networks. However, as we have seen,
these two different types of clusters tend to offer different possibilities with
regard to the local role of MNEs, and the contribution that they make to
local spillovers. Yet, it is these two types of clusters which tend to experi-
ence a faster process of convergence between old and new technologies,
and potentially therefore also a greater degree of competitiveness.
In the light of the arguments presented so far, it becomes clear that
all industrial clusters can be characterized both in terms of the transac-
tions costs and relations characteristics described in Table 5.1, and also
in terms of technological regimes and knowledge characteristics sum-
marized in Table 5.2. As in all taxonometric attempts to classify units
of analysis which are based on a stylized reduction of the complexity of
the whole population, differences between the categories are necessarily
maximized. However, as Pavitt himself said about his own taxonomy, the
main weakness of our attempt 'is the high degree of variance still found
in each category' (Pavitt 2000, p.xi). This is also true here in our case,
although there is a subtle difference between the Pavitt schema and ours.
Pavitt's approach was inductive and based on detailed empirical observa-
tions of individual units of analysis such as firms (Archibugi 2001), and
even though his classification scheme was the most parsimonious yet
developed, he still observed enormous variation even within his individual
taxonometric categories. In contrast, our taxonomy is deductive, and
based on the different theoretical streams of literature which exist regard-
ing the economics and geography of innovation, and the various attempts
within this literature to classify composite units of analysis such as clusters
and regional systems. However, fundamentally these various theoretical
approaches are all aiming to provide the most parsimonious and empiri-
cally and theoretically defensible classification system for explaining
