turn in economic geography as the successful development of the programme of institutional economists,
which have little success within the boundaries of the economics profession hitherto.
As depicted in Figure 2, we posit that a trend of theorising at interfaces is taking place in economic
geography. The main debate so far has been taking place between what we would describe as neoclassical
and institutionalist economic geographers (Amin and Thrift 2000; Martin and Sunley 2001), and we agree
with Martin (2003) that it has led to little fruitful exchange of ideas so far. In Figure 2, we have made this
clear by observing a clash between neoclassical and institutionalist economic geography. Where
geographers apply an inductive, and often, case-study research approach, signalling out the local
specificity of ‘real places’, economists approach the matter deductively and mathematically starting from
‘first principles’ (equilibrium analysis) and core assumptions (utility maximization). By starting from
‘neutral spaces’, geographical economists abstract from local specificity and particular spatial level of
aggregation. Progressively, they complexify their models to incorporate more relevant mechanisms, yet
hold on to spatially neutral spaces, and to the first principles and core assumptions characterising
neoclassical economics.
Going back to the main debates discussed in the previous section, we can understand the clash of
geographical economists and economic geographers as reflecting two incommensurabilities. First, the two
approaches differ in their methodology: institutional economists dismiss a priori the use of mathematical
modelling and econometric specification derived from these. By contrast, neoclassical economics do not
value the method of case study highlighting locally specific patterns of economic life (e.g., Overman,
2003). Second, the two approaches differ in their core assumptions underlying their explanations of
economic phenomena. Institutions being embedded in geographically localised practices, they form the
prime subject of economic geography, as institutional specificity largely determines the local economic
patterns. By contrast, institutions play no role in neoclassical models, or only in a loose and implicit sense
(e.g., relating to particular parameters in the model). Local institutional and cultural factors are left out of
the analysis are not regarded as essential to an economic explanation, and are ‘best left to the sociologists’,
as Krugman once put it.1 What is more, the recent wave of mathematical models develop by neoclassical
economic geographers does not even require differences between regions to exist (be it institutional
differences of different in factor prices). Rather, the models start from a ‘neutral space’, and aim to
explain how agglomeration can occur from this, i.e. how uneven patterns can emerge from an initially
uniform world. This has been called ‘putty-clay geography’ by Fujita and Thisse (1996): “there is a priori
considerable uncertainty and flexibility in where particular activities locate, but once spatial differences
take shape they become quite rigid” (Martin, 1999, p. 70).
We would argue that evolutionary economics is in a way linking the two approaches, as the
methodological and theoretical differences that underlie the clash between institutional and neoclassical
economics, are precisely the aspects that unite evolutionary economics to the two approaches. As
explained earlier, the use of formal modelling is supported by evolutionary economics as a way to verify
intuition and to derive precise hypotheses to be put to the test. In the context of economic geography,
evolutionary economics meets the neoclassical approach is in starting out with a uniform world and in
1 Note that the exclusion of institutions in geographical economics does not render this theory instrumentalist per se.
Where institutional economic geography and geographical economics differ is on the essential mechanisms
underlying economic phenomena. As argued by Marchionni (2004), Krugman is best regarded as a realist that uses
models as a research strategy to come closer to unravelling the complexity of the economy rather than an
instrumentalist that judge models primarily on the basis of its predictive value.
14