conclusion presents some final thoughts and suggestions for enhancement of the system of data
and information on science, technology and innovation.
1. The geography of innovation: spillovers, spin-offs, networks, and regional systems
The correlation between geography and innovation has been demonstrated empirically by
several authors. Feldman (1993; 1994) and Audretsch & Feldman (1996), for example, show that
there is a clear relationship between the localization of innovative activities, measured in terms of
the number of patent citations, and the geographical concentration of innovative inputs such as
R&D in universities, industrial R&D, the presence of related industries, and the presence of firms
that provide specialized business services, demonstrating the importance of “geographically
mediated spillovers”. They also show that there is an important correlation between the location
of innovation production and the location of industry value added, but that it is the presence of
related industries that is most relevant to innovation activities, indicating the significance of
regional innovation networks.
There are in fact several schools of thought with differing approaches to the theoretical
and empirical explanation of the relations between geography and innovation, including the
formation of geographically concentrated clusters of firms in many economic activities, but
above all in technology-based industries. This is not the place for a detailed discussion of all
these approaches.1 Given the scope of this paper, it will suffice to summarize the key points that
are common to several approaches and substantiate the findings presented here.
The foundation shared by all the approaches discussed here is the perception that
geographical proximity facilitates the transmission of new knowledge characterized as complex,
tacit, and specific to certain production and innovation systems and activities. This may seem
paradoxical in the age of information and communication technology but, as noted by Audretsch
& Thurik (2001), it is important not to confuse knowledge with information. While the marginal
cost of transmitting information is not proportional to distance, the cost of transmitting
knowledge, especially tacit knowledge, increases with distance. This type of knowledge is best
transmitted through interpersonal contacts, frequent interaction, and mobility of workers from
one firm to another. Hence the advantage of geographically concentrated configurations of
production such as clusters.
However, although they have this common foundation, two groups of approaches can be
distinguished by their emphasis on differing mechanisms of knowledge transmission. One group,
comprising the innovation economics and innovative systems approaches,2 attributes a key role to
spillovers in the transmission of knowledge among neighboring firms. These spillovers are
triggered by innovative firms or institutions that generate new knowledge. The other group,
comprising approaches based on regional economics, seeks to explain what makes firms localized
in clusters more innovative than isolated firms. In doing so these authors emphasize a different
set of key factors in knowledge transmission. According to Breschi & Malerba (2001: 819-820),
the main points of these approaches are as follows: (1) learning through networking and
interacting, including user-producer relationships, formal and informal collaborations, interfirm
mobility of skilled workers, and spin-offs of new firms from existing firms, universities and
research centers; (2) the high-level embeddedness of local firms in a very thick network of
knowledge sharing, supported by close social interactions and facilitated by shared norms,
conventions and codes, and in institutions that build trust and encourage informal relations
1 See Breschi & Malerba (2001) for such a discussion.
2 The systems approach considers national, regional, sectoral and local innovation systems and technological
systems.