Geography Reference
In-Depth Information
way, becoming a sort of local 'elite', and widening the gap with more isolated or marginal-
ised i rms. Through the selective development of knowledge networks, thus, the heteroge-
neity in the distribution of i rm knowledge bases may breed persistent unevenness among
cluster i rms, leading their performance to diverge over time. In this way, moreover, the
process of economic development of a cluster may become vulnerable and dependent on
the behaviour of individual i rms, rather than on any other meso-level force.
5. Conclusions
Geography has come to play an increasingly central role in the analysis of economic
growth and competitiveness of countries (Krugman, 1991; Porter, 1998). This is con-
sidered to be because of the fact that, in the context of increased trade liberalisation,
processes of Marshallian industrial localisation fuel the emergence of 'growth poles' or
successful clusters. This is often associated with the fact that i rms operating in clusters
are likely to generate a social environment, characterised by dense inter-i rm networks,
which enhances their likelihood to innovate. In this chapter I have argued that two types
of question are still open for investigation in this domain of study. The i rst one has to
do with the understanding of the causal relationship between the behaviour of individual
actors - such as i rms - and the structural properties of the networks that they are capable
of generating at the local level. The second one relates to the relationship between the
characteristics of localised networks - in terms of their content and structure - and the
economic development of regions.
This chapter deals with the second question. In particular, it has analysed the ef ect
of two types of network - the business and the knowledge network - on the similarity in
the performance of i rms in three wine clusters. The analysis has been carried out using
methods of
social network analysis
(Wasserman and Faust, 1994). Multiple regression
quadratic assignment procedure (Dekker et al., 2005) is applied to test the impact of the
knowledge and business networks on the dependent variable represented by a similarity
squared matrix of i rm performance.
The results of the QAP multiple regression show that in all three clusters the knowl-
edge network has a positive and signii cant impact on the dependent variable, while no
signii cant evidence is found for the business network. This implies that the existence of
a knowledge linkage between any two i rms increases the likelihood that both i rms are
good performers - which is not the case for the business linkages. These results raise
two types of consideration. First, that the
content
of the network ties may be extremely
important for the economic performance of i rms and that it is not networking per se that
enhances performance in clusters, but the existence of valuable, knowledge-rich linkages.
Second, that the
structure
of knowledge-rich networks may indeed af ect the
quality
of
regional economic development: where the knowledge network was slightly less selective
- as in Valle de Colchagua - the number of good performers was higher. The key issue
is to understand what determines the formation of valuable or knowledge-rich linkages
and what is the relative weight of i rm-level variables and territory-specii c factors.
My contention is that this is very much af ected by the characteristics of i rms in clus-
ters (see also Lazerson and Lorenzoni, 1999; Martin and Sunley, 2003; Maskell, 2001)
- an issue that is in line with an evolutionary economics view of the i rm. As I am able to
show in Giuliani (2007a), in fact, the structure of knowledge-rich networks is shaped by
the distribution of i rms' knowledge bases. According to this, it is possible to suggest that
