Geography Reference
In-Depth Information
Inventors who work for dif erent companies are responsible for a large number of
citations, but are scarcely mobile in space: they move or dif use their knowledge across
dif erent i rms, but not so much across dif erent localities. They also contribute to create
social networks, which also spread knowledge across i rms, but not in space. These
results are coni rmed by another exercise, based on the same methodology, on a sample
of Italian data (Breschi and Lissoni, 2009).
Those few inventors who move not only across companies, but also across locations
(the 'movers') maintain their ties with former co-inventors and their networks who are
still located in the cities they have left. It is those cities, not those where the movers
reside, that seem to enjoy an advantage over others in terms of access to the movers'
knowledge.
Our results also raise a few substantive issues that deserve to be further discussed and
investigated.
In the i rst place, our results qualify the original intuition of those economists and soci-
ologists that i rst emphasized the tacit content of technological knowledge: knowledge
always travels along with people who master it. If those people move away from where
they originally learnt, researched, and delivered their inventions, knowledge will dif use
in space. Otherwise, access to it will remain constrained in bounded locations. That is,
knowledge l ows are localized to the extent that cross-i rm activity and the resulting
social networks also are localized. Why US cross-i rm inventors exhibit the observed
(quite limited) mobility patterns is of course an important question, but one that goes
beyond the scope of the present study.
In addition, our results suggest that social ties derived from co-invention activity do
not disappear when inventors move across space; on the contrary they seem to convey
more knowledge than those built by movers in their new locations.
Networks of inventors, of course, capture only a tiny subset of all the relevant social
contacts enabling an individual to achieve an invention. However, most inventors listed
on EPO patents are professionals, and their population is much more than a tiny and
unchecked sample of all the individuals who can inl uence them; rather, it is the most
immediate and inl uential social environment from which inventors draw ideas and
information, at least for the technical contents of their patents.
The fact that by controlling for the role of the network of inventors we manage to
reduce the apparent role of spatial proximity, but not to eliminate it, suggests that other
social networks, dif erent from the professional one we considered here, may matter. In
the near future, we will move in this direction by considering both patents and scientii c
publications, so that our network includes both inventors and pure academic scientists.
The social network approach to knowledge dif usion proposed in this chapter may
be further extended to comparative analysis. Ideally, one could compare the extent of
knowledge localization in dif erent regions or nations, and explain it with the dif erent
degree of mobility and resulting network dispersion of inventors, scientists, and knowl-
edge producers in general.
Notes
1.
The exclusion of company self-citations is motivated by the fact that these citations do not represent
spillovers. Originating patents with no citations are similarly excluded from the analysis, because they are
supposed not to have generated any knowledge l ow.
