to unravel fundamental mechanisms from mechanisms or context that is only of secondary importance. In
this vain, Nelson and Winter (1982), Winter (1984), Klepper (1996) and others have turned to models as a
means to verify inductively generated ideas, and to derive propositions from them that can be tested in a
relative unambiguous manner. Typically, econometric tests in evolutionary economics make use of
historical time series (Klepper 2002a,b), rather than in a predictive manner as advocated by Friedman. The
emergence of what is called applied evolutionary economics (Saviotti, 2003) marks the importance
attached to the further development of evolutionary econometric techniques.
Neoclassical and evolutionary approaches thus agree on the importance of formal theorising, yet apply
these models in slightly different manners. What is more, evolutionary economics also makes use of other
modes of theorising summarised by Nelson and Winter (1982) as appreciative styles of theorising (p. 46).
They make a strong argument in favour of finding a sort of balance between formal and appreciative
theorizing, the latter being more inductive making use of stylised facts. In practice, one can observe that,
broadly speaking, evolutionary economics is evolving into three main methodological directions. The first
is focussed on abstract, formal modelling. In this respect, it is rather ironic to see that the evolutionary
approach runs the risk of falling prey to the same weakness of the neo-classical theory defined by Nelson
and Winter (1982) more than twenty years ago, that is, building ‘a rococo logical palace on loose
empirical sand’ (p. 33). The second is focussed on more descriptive, empirical studies, as illustrated, for
example, by the ‘Innovation System’ literature (Freeman, 1987; Lundvall, 1988; Nelson, 1993). These
kind of studies make an attempt to analyze the innovation process in a territory in terms of what kind of
organizations are involved, what kind of interaction patterns exist, and how these are embedded in a
particular institutional setting (Edquist, 1997; Cooke et al. 1998). The third tries to find a sort of balance
between these two extreme methodological positions.
3.3 Static versus dynamics
Characteristic for evolutionary theory, be it as a theory of natural history in biology, as a theory of
learning in artificial intelligence, or as a theory of economic development in economics, holds that it
explains processes of qualitative change. Essentially, evolutionary theory asks the question how a current
state of affairs can be explained from history, i.e., as an interplay between accumulating structure (e.g.,
genes, routines) and selection conditions prevailing during history. The current state of affairs cannot be
derived from current conditions only, since the current state of affairs has emerged from, and has been
constrained by, previous state of affairs. This is why history matters (David, 1985).
For example, consider two regions that are equal in terms of factor prices and in terms of institutions.
Still, it is perfectly possible that in each region a different technology is used, because at the time of
adoption selective conditions were different. Being used hereafter for a sustained period of time, learning
has been localised in this technology only (Atkinson and Stiglitz, 1969). The technology has become
entrenched in the cognition and routines of agents, and has co-evolved with interrelated technology and
institutions. Jointly, technological and institutional structures thus develop in a path dependent manner.
Similarly, at a micro level, two firms that are subject to the same selection environment can adopt very
different technologies if selective conditions have been different at the time of adoption. Hereafter,
learning and innovation have been localised in their specific technology, and only partially spills over to
other firms. And, technology and organisational structures mutually adapt in the course of time in firm-
specific manner, explaining why firm heterogeneity is persistent. These are only two examples of
11