Geography Reference
In-Depth Information
particularly at a time when the inducements derived from global market competition and
changes in demand, and from technological change, arise at an incredibly high speed.
The original Marshallian-Arrow-Romer (MAR) stream of literature plus the variety
literature on urbanisation and diversii cation economies have thus been broadened by
including technological indices as endogenous variables in explaining local economic
growth (e.g. Acs, 2002; Jaf e et al., 1993).
In the main, however, while these neoclassical interpretations have focused on local-
ised spillover ef ects as the major analytical framework to explain cluster existence and
growth, they have largely disregarded other possible mechanisms underlying spatial
agglomerations (Autant-Bernard et al., 2003; Breschi and Lissoni, 2001; and Breschi et
al., Chapter 16 in this topic). In particular, production function-based models do not
give a full account of the diversity of possible spatial coni gurations, nor do they tackle
issues such as the evolution of clusters and regions and the disruptive changes imposed
by technological progress and globalisation processes on agglomeration economies
(Belussi and Sammarra, 2010; Hilpest, 2003).
As is mentioned in section 2, various factors are at work in shaping the relationships
between MNEs and local environments, and therefore in determining the nature and
extent of localised knowledge spillovers possibly stemming from such relationships. In
this respect, it appears that we have two sets of fundamental questions to be addressed.
First, we need to determine
whether
and
where
localised knowledge spillovers exist, and
then second, we need to understand exactly
how
they do occur and change over time.
It has to be highlighted that the dii cult analytical problems relating to the diverse
features of spatial agglomerations are compounded by severe problems of identii cation
and dei nition.
1
Here we provide a classii cation of spatial types that is independent of
either the sector or the location, but instead is based on the microeconomic behaviour
and objectives of the co-located actors, and on the transactions existing in the cluster or
region.
A transactions costs approach was adopted elsewhere to present three dif erent styl-
ised sets of geography-i rm-industry organisational relationships that exist in the litera-
ture (Gordon and McCann, 2000; McCann, 2001; McCann and Shefer, 2004; McCann
and Sheppard, 2003; Simmie and Sennet, 1999). This classii cation is based on the (often
implicit) assumptions underlying most of the existing literature on agglomeration phe-
nomena. As such, the categories are not meant to be interpreted as representing any
particular geographical place. These stylised characterisations of spatial clustering are
distinguished in terms of the nature of i rms in the cluster or region, and the nature of
their relations and transactions undertaken within the specii c local environment. They
can be termed the
pure agglomeration
, the
industrial
complex
, and the
social network
. The
main characteristics of each of the spatial types, all quite dif erent, are listed in Table
8.1.
2
This classii cation is useful to illustrate that knowledge spillovers do not stem as an
automatic mechanism from the location of MNEs in any particular spatial agglomera-
tion, but, rather, they i rst depend on the nature and characteristics of the transactions
taking place locally. In other words, this classii cation can help determine whether and
where knowledge spillovers may occur.
From a transaction cost perspective, the i rm's view of the net benei ts of knowledge
spillovers hinges on its assessment of the relative importance to itself of knowledge out-
