Geography Reference
In-Depth Information
linkages, may be crucial in determining the occurrence and the extent of spillovers. Even
in the same industry, clusters may be characterised by highly vertically integrated i rms,
or by stage production with signii cant sub-contracting linkages: the trust relationships
and the role of reputation will dif er considerably in such dif erent spatial and industrial
agglomerations (Guy, 2009).
Yet, the classii cation reported in Table 8.1 can only represent a partial and static
picture. The actual detailed interplay between industries and geographical locations,
or the specii c evolutionary features of dif erent regions and clusters, have been largely
neglected by most current stylised economic geography approaches. This has caused
hectic ef orts to identify localised knowledge spillovers between local i rms, MNEs
and other organisations, irrespective of the characteristics of any functional dimension
of knowledge processes depending on the type of industry and spatial agglomeration
model.
4. The dynamics of spatial coni gurations: knowledge, and technological and structural
change
One of the advantages of the transactions costs classii cation scheme reported in section
3 is that it provides an organising framework capable of dealing with the diversity of
spatial concentrations we observe, of ering possibilities for the inclusion of a technol-
ogy and innovation component as an additional explanation for such a diversity. At
the same time, a weakness of this framework is that, in the transaction cost perspective
itself, hierarchies, and particularly, but not exclusively, i rm structures, are reduced to a
consequence of changes in transaction costs, whereas dynamic factors such as learning,
accumulation, and knowledge creation are largely ignored. More specii cally, the trans-
action cost approach underlying Table 8.1 has some limitations because of its essentially
static nature, its narrow dei nition of knowledge and technology, and its discounting of
the relationship between innovation processes and industry structures.
With respect to the i rst point, the classii cation presented in the previous section is
mainly static, and rel ects the bulk of the existing literature on the cluster concept that
it summarises. Underlying most of such a literature there are once-and-for-all
ei ciency
gains (in terms of economies of scale and scope, transaction and transport costs and
input-output linkages). However, the relationship between i rm location and technol-
ogy is mainly dynamic (see among others Audretsch, 1997; Dosi, 1988; Dosi et al., 1997;
Nelson, 1991; see also Bottazzi and Dindo, Chapter 24 in this topic), both in terms of
industrial demography (i rm entry, exit, growth, relocation, etc.) and in terms of cluster
life cycle (cluster birth, growth, decline, openness, attractiveness, etc.). Thus, clusters and
regional systems are not necessarily static in that they may evolve over time, possibly
blending the features of the three typologies into dif erent combinations, and possibly
shifting from one main type to another according to the relative stage in their life cycle.
Dynamic agglomeration economies and the central role of learning and new knowledge
creation therefore lead to distinctive paths of regional evolution.
Regarding the second point, the logic of the contractual transaction cost arguments
about appropriability of the returns on innovation, widely applied to the perception of
knowledge inl ows and outl ows by MNEs, depends to a large extent on a narrow dei ni-
tion of technology. The traditional approach in economics has regarded knowledge as a
public good, therefore assumed to be 'non-excludable', 'non-depletable', and free to be
