required when specialised knowledge and skills are distributed among different persons
(Gibbons et al., 1994; Ziman, 1994). And given the increasing level of specialisation within
disciplines and sub-disciplines resulting from an increasing division-of-labour among
scientists, research would benefit from international rather than national recruitment of
scientists to participate in particular projects.
The growing internationalisation in scientific research may appear, at first sight, contradictory
to the recent literature on the geography of innovation. This literature tends to emphasise the
increasing localised nature of knowledge production at sub-national scale (Arthur, 1990;
Boschma and Lambooy, 1999; Caniëls, 2000; Feldman, 1999; Van Oort, 2002). A number of
theoretical arguments support this thesis. Most importantly, localisation economies of various
sorts arise when people and firms engaged in knowledge production are geographically
concentrated. These externalities range from labour market pooling, informal networking and
knowledge spillovers. In particular, tacit components of knowledge production typically
develop and diffuse through close interaction with suppliers and clients. Furthermore, tacit
knowledge is often reproduced in spin off firms that typically locate in the region of the
parent company. Another important reason for localised innovation concerns the difficulties
of governance of collaborative industrial R&D. As the modalities of collaboration are hard to
encode in contracts, collaboration often relies on informal contacts, reciprocity and trust
between partners, which is facilitated when participants share local ties and a similar
institutional environment. If one accepts that the economy has become rapidly more
knowledge-intensive, economic activity can be expected to have become more localised in
recent times.
From these theoretical rationales for localised industrial innovation, however, one can not
conclude that one should expect scientific research to develop into a more localised activity as
well. Scientific research is qualitatively different from industrial innovation. Though in some
disciplines the distinction between science and innovation has become less relevant such as in
biotechnology and informatics, scientific knowledge production generally differs from
industrial knowledge production in a number of ways. First, the tacit component is expected
to be much smaller in scientific knowledge production, which renders communication and
collaboration at a distance much easier. Second, the specificity of knowledge (‘appliedness’)
is expected to be much smaller in scientific research compared to industrial R&D.
Consequently, problem definitions are to a lesser determined by the local context, but emerge
from a global discours . Third, the incentive structure in scientific knowledge production is
explicitly oriented towards (international) diffusion, while investors in industrial R&D have
an incentive to appropriate the results (whatever the mechanism used to achieve this). For