location choice such that their joint actions are in equilibrium. As noted by Krugman (1996) and Brakman
and Garretsen (2003, footnote 14), the concept of equilibrium underlying the models of new economic
geography is close to the concept of Nash-equilibrium in an evolutionary game: an equilibrium is
characterised by a set of actions such that, given this set, no individual has an incentive to do otherwise.
This means that a model is analysed by looking at Nash-equilibria (the stable ones usually being all firms
being all located in city 1 or all in city 2). A change in equilibria is ‘caused’ solely by a change in the
exogenous parameters, i.e. outside the model setting. This differs from evolutionary models, in which not
only the set of actions of firms is analysed, but also endogenous changes in the possible set of actions
occurring over time (due to innovation). In this view, economic dynamics may only show temporary
convergence to equilibria, which are continuously being ‘upset’ by deviant innovative behaviour by some
firms in the population. What is more, the disequilibrium tendency caused by deviant behaviour is not
regarded as ‘noise’ but as the fundamental drive underlying economic development. Firms have an
incentive to deviate, especially when a tendency towards equilibria is present and profits are shrinking.
Following Schumpeter (1934, 1939, 1942), evolutionary economists view the search for supra-normal
profits by innovation caused by technological competition is seen as the primary dynamic in the economy
(moving away from equilibria), and the erosion of profits due to price competition as a secondary dynamic
(converging to equilibria).
4.2 The interface between institutional economic geography and evolutionary economic geography
As stated earlier, it is quite common that institutional and evolutionary approaches are treated as one
and the same by geographers. This may have something to do with the rejection of the neoclassical
paradigm by both approaches, which is widely supported by economic geographers. This may, for
example, be illustrated by the fact that neoclassically based institutional thinking a al Williamson, which is
an established and influential field in economics, has hardly found any supporters in economic geography
(one exception being Scott 1993). However, we claim it would be confusing, wrong and misleading to
equate institutionalist and evolutionary approaches in economic geography, and we argue that the
approaches can be considered as being rather different. We make use again of the three debates outlined in
Figure 2 in order to explain why. With respect to the methodological debate, quite substantial differences
remain between the institutional and evolutionary approach, because the latter approach tends to employ
more quantitative and analytical tools. Interestingly though, there is increasing opposition from within the
institutional economic geography itself, pleading for more rigour analyses (Martin and Sunley, 2000;
Taylor, 2004). Institutional and evolutionary approaches, however, agree on geographical specificity and
the important of context, and both criticise the neoclassical view for ignoring this. With respect to the
‘static versus dynamics’ debate, institutional approaches rely heavily on comparative analyses while
evolutionary approaches favour longitudinal analyses. Nevertheless, although both approaches recognise
the importance of path-dependent processes, they do so in different ways, especially with regard to the
type of change involved.
As explained earlier, institutional approaches take a critical stand toward abstract, formal theorising,
as employed by neoclassical economics, and, to a lesser extent, by evolutionary economics. In institutional
economic geography, descriptive analyses are the rule rather than the exception. Regional development
being regarded as a complex and multi-faceted phenomenon, it can only be understood using anti-
reductionist methodologies. In particular case studies research based on a variety of qualitative techniques
18