Geography Reference
In-Depth Information
(McCann and Mudambi, 2004, 2005). This will lead the MNE to decide not to locate
their technological operations in clusters characterised by pure agglomeration, although,
depending on transportation and other types of cost, they may consider locating in com-
plexes characterised by stable planned and long-term inter-i rm relationships, and by a
geographical coni guration that is often local or regional but rarely urban.
This argument provides a powerful counter-argument to the Porter logic in favour
of industrial clustering (Porter, 1990). This is because if we apply Akerlof's (1970)
market-for-lemons model, regions and industrial clusters that include large oligopolis-
tic competitors will be plagued by adverse selection and should either fail to form, or
become concentrations of mediocrity, unless the clustering of i rms exhibits the charac-
teristics of an industrial complex in Table 8.1. This counter-argument to the Porter logic
appears to explain the empirical i nding that many of the largest MNEs do not co-locate
their knowledge creation activities with those of their competitive rivals (Cantwell and
Santangelo, 1999; Simmie, 1998).
It follows that the only possible reason why MNEs would locate their innovative activi-
ties in an environment characterised primarily by a pure agglomeration, would simply be
as a means of facilitating the hiring of specialist labour. However, MNEs are often able
to hire appropriate specialist labour or to tap into local technological expertise by simply
locating within the same broad regional innovation system (Cantwell and Iammarino,
2003), rather than within the same urban location. According to this perspective, the
inter- i rm spillover arguments implicit in the pure agglomeration model are therefore not
always applicable to oligopolistic MNEs (Arita and McCann, 2002a, 2002b).
Similar arguments are often also pertinent in the case of the social network model of
Table 8.1, associated primarily with Granovetter's work (1973, 1985) and developed
as a response to the hierarchical model of Williamson (1975). Such social networks are
assumed to operate on the basis of trust relations. However, the notion itself of trust may
vary greatly between social networks characterised by substantial relational symmetry,
and those characterised by asymmetries among the clustered i rms and centred on large
leader- i rms, either locally owned or foreign ai liates. This latter case can be assimilated
to the 'hub-and-spoke' type of cluster identii ed by Markusen, where one or several
i rms/facilities in one or more sectors, usually large and vertically integrated (e.g. Boeing
in Seattle or Toyota in Toyota City), act as anchors or hubs to the regional economy,
with smaller suppliers and related activities spread around them like the spokes of a
wheel, with dif erent degrees of strength (Markusen, 1996). Interestingly, in no case will
there be perfect symmetry among the various agents operating in the cluster, but one
agent (or more of them) may play a distinct role, leading the cluster in terms of organisa-
tion, innovation, and/or i nance (Guerrieri and Pietrobelli, 2004).
Thus, on the basis of inward and outward knowledge l ows and appropriability argu-
ments, the conditions under which such trust relationships will emerge naturally are
largely opaque. If the i rms in a cluster are to be small, then the arguments for pure
agglomeration may be largely applicable in the case of the social network, although dis-
tinguishing between the two becomes problematic. However, if some of the i rms in the
cluster are to be large and/or MNEs, then it is not clear how such a social network would
evolve, and much depends on the nature of transactions they establish with (large and
small) local i rms. The structure of vertical and horizontal linkages between MNEs and
local i rms and, within the i rst type of connections, the extent of backward and forward
