problem of policy by imitation concerns the high degree of tacitness and (often subtle) interdependencies
that exist between the different factors contributing to a successful model. This implies that imitation of a
subset of factors is likely to fail, because of the mismatch between the new subset and the existing
structures and routines.5 The historical trajectory of a territory sets serious limits on copying an external
model that owed its success to its deep roots in an alien environment (Zysman 1994; Rivkin 2000).
Even though comparative analysis has its critics within evolutionary economics, the innovation
system approach has its roots in evolutionary economics (Lundvall, 1988; Nelson, 1993; Edquist 1997).
Perhaps ironically, the innovation systems approach has many characteristics of a truly institutional
economic geography approach outlined above. It aims to uncover the institutional setting that more or less
determines what kind of organizations and interaction patterns in a territory are involved in the innovation
process. For this reason, it takes the role of institutions for granted, and tries to link differential
performance to different institutional settings. It is quite common for this body of literature to associate an
innovation system with a particular spatial level a priori, as is illustrated by notions like national and
regional innovation systems. However, evolutionary economists reason from routines being embedded in
firms and its relations. Then, the region as a unit of analysis is problematic, though not without meaning,
as there is no strong reason to assume that routines are place-specific (Boschma 2004). Rather, some
regions may be characterised by a strong degree of homogeneity in routines while other may not. This is
why a more evolutionary approach to innovation systems is to adopt a sectoral innovation system
perspective. Breschi and Malerba (1997) and Breschi (2000), for instance, stress the specificity of sectoral
systems and their similarities across different regions. Such an approach also suggests that the history of
innovation systems, in specific places, should be understood from a dynamic sectoral analysis, by
analysing how institutions have co-evolved with the emergence of a new sector (cf. Galli and Teubal,
1997). Since the implementation and diffusion of novelty often requires the restructuring of old
institutions and the establishment of new institutions (Freeman and Perez, 1988), the main issue is to what
extent institutions are flexible and responsive to changes in different places. A well-known example is the
rise of the synthetic dye industry in the second half of the nineteenth century, which required many
institutional changes (mainly patent laws and new scientific and educational organisations), which
Germany succeeded to implement, but the UK and the US, did not (Van den Belt and Rip, 1987;
Murmann, 2003).6 Institutional differences between regions or nations, in this view, are part of the
explananda, as institutions co-evolve with processes of technological innovation and industrial dynamics.
Put differently, ‘real places’ emerge from innovative actions of economic agents, rather than fully
determining actions of economic agents.
5 In biology, interdependency between a system constituting elements is better known as epistasis (Frenken 2004).
6 Another example is the case of old industrial regions, which may become locked into rigid trajectories because their
sectoral legacy of the past (in terms of resources, competences and socio-institutional structures) has weakened their
ability to develop new promising activities. Although the concept of lock-in at the territorial level is still largely
underdeveloped, it sheds light on the way restructuring may succeed or fail in particular regional settings. In this
respect, it is fruitful to distinguish between different forms of lock-in on the cognitive, organisational and political
level. Evolutionary economists stress cognitive lock-in as routines are based on particular technological paradigm
(Dosi 1982). Geographers (Camagni 1991), who showed that also stress organisational lock-in, and argue that a
higher number of relations with other organizations reduces the danger of lock-in, including non-local linkages
granting access to knowledge and resources in the outside world. Finally, political inertia stems from vested interests
in the political-economic realm actively opposing change (Grabher 1993). Depending on the strengths of these
different forms of lock-in, old industrial regions may face distinct problems and different options for transition.
22