Given the methodological similarities between the evolutionary and the neoclassical approach, and the
theoretical similarities between the evolutionary and institutional approach, one can expect the exchange
of ideas along these two interfaces to be most fruitful in economic geography. We will therefore explore
in more detail some of the recent contributions along these interfaces.
4.1 The interface between neoclassical economic geography and evolutionary economic geography
In recent years, the main contribution of neoclassical economics to economic geography has been the
development of a new family of models, based on Krugman’s (1991) core model, commonly headed under
the label of ‘new economic geography’. As these models are better understood as economic models
treating only some aspects of geography (in particular transportation costs), some prefer the label of
‘geographical economics’ (Brakman et al. 2001) instead of ‘new economic geography’. Though the new
economic geography has been attacked on various occasions by economic geographers and others (Martin
and Sunley 1996; David 1999; Martin 1999; Amin and Thrift 2000; Nijkamp 2001), it can be considered
an important contribution to our theoretical understanding of possible mechanisms creating uneven spatial
development. Moreover, as the new economic geography is firmly based in neoclassical economics, using
the same core assumptions, it also provide a way to connect issues of geography to other areas covered by
neoclassical economics. Our aim is not to discuss neoclassical economic geography in any depth. In the
following, we explain that, despite different assumptions and micro-foundations, the new economic
geography often comes to similar conclusions as evolutionary theory. At the same time, we make clear it
would be wrong to assume that convergence between the two approaches will necessarily occur. We go
into the similarities and differences using the three debates outlined in Figure 2: evolutionary and
neoclassical approaches share a common methodology of modelling, yet differ in assumptions and in the
treatment of static and dynamics.
The new economic geography can be considered a recent extension of neoclassical economic
geography, relaxing, for instance, the standard assumptions of perfect competition and constant returns to
scale. It is basically a ‘micro-economic theory of spatial agglomeration’ (Martin, 2003). In doing so, it
reintroduces agglomeration economies as a self-reinforcing process of cumulative causation that results in
spatial concentration of economic activities. As evolutionary approaches, the new economic geography
differs in important respects from the traditional neoclassical approaches that typically involve models of
a-historical and reversible processes with an unique spatial equilibrium. In this sense, it shares a number
of features with evolutionary modelling, such as the possibility of multiple outcomes of spatial
concentration of economic activity, path dependence in the process leading to one of possible outcomes,
irreversibility of outcomes leading the economic system to lock-in, and the possibility of sub-optimal
outcomes as end result. Moreover, both approaches are keen on explaining how uneven spatial patterns
emerge from neutral uniform spaces. For example, the Polya urn model underlying some evolutionary
models (Arthur 1994), applied to the spatial evolution of a new industry, assumes that initially, firms can
emerge in any region with equal probability. However, new firms being modelled as spin-offs of existing
firms, uneven distributions of firms will emerge automatically. This fundamental mechanism driving both
the birth of an industry and its spatial distribution, also underlies a recent paper by Klepper (2002a)
explaining how Detroit became the capital of the U.S. car industry from spin-off dynamics alone. Using a
recent evolutionary approach based on graph theory (Barabasi and Albert 1999), one can also take
16