networks, which suggests that only a few central firms profit from knowledge spillovers. This
hypothesis has been put to the test in a follow-up study presented in this volume (Giuliani,
forthcoming), in which it is shown that a firm’s centrality in knowledge networks is indeed
positively affecting innovative performance, even after controlling for heterogeneity in internal
competencies. A recent study by Boschma and Ter Wal (2006) on a footwear district in Southern
Italy tends to suggest that the absorptive capacity of firms is indirectly related to their innovative
performance, through having non-local instead of local relationships. That is, the higher the
absorptive capacity of a district firm, the better it is connected to organisations outside the
district, which, in turn, impacts positively on their innovative performance. These studies show
that social network analysis is a powerful tool in analysing the geography and structure of
knowledge networks and the effect of a firm’s network position in these networks on its
performance. In a similar fashion, the concept of regional innovation systems (Cooke et al., 1998)
can be operationalised empirically more systematically by mapping the various network relations
of actors that are part of the regional system with other actors within and outside the regional
system.
Evolutionary theorising has also argued that, due to bounded rationality, consumers rely on
personal networks. As a result, certain decisions by central actors can propagate through the
network leading many consumers to opt for the same product (Cowan et al., 1997; Plouraboue et
al., 1998; Solomon et al., 2000). The strength of these networks effects, and the geographical
nature of such personal networks, can also be explored empirically using social network analysis.
A nice example of such an approach is the study by Birke (forthcoming) who conducted a survey
among students asking them about their personal networks and their choice of mobile telephone
operator so as to analyse the effect of personal networks on the choice of operator.
Hitherto, the use of social network analysis in evolutionary economics has been almost
exclusively static. A future challenge is to understand the spatial evolution of networks. This
requires longitudinal data and methods to analyse the dynamics of networks over time. An
influential theoretical model of network dynamics is the model by Barabasi and Albert (1999). In
this model, a network grows as new nodes connect to a network. Nodes are assumed to attach
themselves to other nodes with a probability proportional to the latter’s connectivity. This
principle is known as ‘preferential attachment’ which means that a new node prefers to link with
a well-connected node so as to profit from its connectivity. Well-connected nodes will then tend
12