Economic Geography

A perspective of economic geography
63

---

Equally, the model diverts attention away from the fact that for any kind of
regional development to occur, productive assets need to be physically mobilized
and integrated with one another on the ground in specific regions (Hirschman
1958). In fact, the model, as such, has virtually nothing to say about the endoge-
nous intra-regional organization and dynamics of production, and almost as little
about the region as a motor (as opposed to a receptacle) of economic activity (cf.
Scott 2002b). In more specific terms, and despite the fact that Krugman and his
co-workers make frequent reference to Marshall, the model actually gives short
shrift to any meaningfully Marshallian approach to regional development and
agglomeration.
4
Four specific lacunae of the core model merit further attention in this connection:
First, the model identifies productive activity only in terms of monopolistically-
competitive firms with fixed and variable costs. In its initial formulation it makes
no reference whatever to the dynamics of the social division of labor and the
networks of transactional relations that flow from this process. In later formula-
tions (e.g. Krugman and Venables 1996; Venables 1996) an intermediate goods
industry is assumed by fiat to exist in the model. However, the model is silent on
the endogenous relations that exist in reality between the vertical structure of
production and spatially dependent transactions costs. These relations tend to 
be of special interest and importance in clustered economic systems where intra-
and inter-firm transactional structures are usually extremely complex (e.g. Scott
1983). Accordingly, the model pays inadequate attention to the wider logic 
of locational convergence/divergence, and, in particular, it is deficient in its 
grasp of the individual regional economy as a source of competitive advantage 
(cf. Porter 2001).
Second, these failings are compounded by the model’s neglect of local labor
market processes, such as information flows, job search patterns, labor-force
training, and so on (Peck 1996). True enough, Krugman pays lip service to the
existence of processes like these, but makes no effort to incorporate them into
the workings of the core model.
Third, region-based learning and innovation processes are conspicuous 
by their absence from the core model. A consequence of this absence is that the 
core model pays little or no attention to patterns of temporal change in the qualita-
tive attributes and competitive advantages of regional production systems.
The rich parallel literature by economists such as Jaffe et al. (1993), Audretsch and
Feldman (1996), or Acs (2002) on regional innovation systems compensates in
some degree for this omission, but the model itself remains more or less impervious
to conceptions of technology-led growth (Acs and Varga 2002).
Fourth, given its resolute commitment to microeconomic forms of analysis, the
model actively suppresses the possibility that collective region-based strategies of
economic adjustment might play a role in the construction of localized competitive
advantages (Neary 2001). In practice, such strategies are often highly developed 
in regions with active production systems, both in the private sphere (e.g. inter-
firm collaboration), and in the public sphere (e.g. local economic development and
training programs under the aegis of regional agencies). Numerous researchers
64
Allen J. Scott

---

have shown time and again that strategies like these are critical to the creation of
regional competitive advantages and an important tool in the search for
improved rates of local economic growth (Bianchi 1992; Cooke 1999; Saxenian
1994; Storper and Scott 1995).
Some of the lacunae pointed out here can no doubt be dealt with in part by
appropriate reformulations of the model (such as the introduction of commut-
ing costs to reflect the spatial organization of local labor markets, or explicit
reference to coalition formation processes), but at the cost of enormous increases
of algebraic complexity. The Krugman model is for the most part a black box
that occludes what by many accounts must be seen as some of the most impor-
tant aspects of regional economic growth and development. As such, it casts only
a very limited light on the full play of externalities, competitive advantage, and
locational agglomeration in economic geography. Needless to say, the model is
silent on wider social and political issues of relevance to the analysis of agglom-
eration, such as, for example, region-specific forms of worker socialization and
habituation, the emergence of local governance structures, or the historical shifts
that occur periodically in technical-organizational structures of accumulation,
and that greatly impact regional trajectories of development.
By its elevation of atomistic exchange relations to an exclusive ontology of 
the economic and the geographic (albeit in a Chamberlinian context), the 
core model provides only a very partial account of the genesis and logic of the
economic landscape. In the end, the model can be seen more as an effort to
codify atomized market processes in simple spatial frameworks than it is an
attempt to understand spatial relations in any thorough-going sense of the term.
The strong point of the model is its description of pecuniary externalities in
multi-region systems; the weak point is its account of locational adjustment in
which units of capital and labor move like billiard balls (except for the ones that
have been nailed to the table) from one equilibrium to another, and then simply
fall magically into a fully functional economic system as they accumulate in
receiving regions. Nevertheless – and to be fair to the wider body of urban and
regional economics generally – there is an obvious and encouraging trend in
much of the current literature to move beyond the limitations of the core model
as expounded by Krugman and to deal in a more flexible and open-ended
manner with many of the issues where it is most vulnerable to criticism (see, for
example, the essays collected together in Cheshire and Mills 1999; and
Henderson and Thisse 2004).