Geography Reference
In-Depth Information
spread may help in explaining the spatial distribution of the knowledge l ows generated
by inventors who move both across organizations and geographical locations.
This research ef ort is relevant for evolutionary economic geography for at least two
reasons. First, it does not merely pay lip service to the conceptualization of knowledge as
tacit and situated, but it explores in depth the implications of such conceptualization. In
order to do so, it avoids taking the logical shortcut of assuming that tacitness necessarily
implies the localization of knowledge l ows, as often happens in the applied literature
based on both the institutionalist and mainstream economic approaches (Boschma and
Frenken, 2006; Breschi and Lissoni, 2001).
Second, it promotes the social network as the main unit of analysis for knowledge
dif usion, instead of the city or region. By doing so, it shares one of the distinctive features
of evolutionary economic geography outlined by Boschma and Frenken (2006), which
consists of studying whether the spatial relations between economic agents matter for tech-
nological change, and how those relations and their importance may change over time.
In section 2 we sum up the key details of the JTH experiment, and of the related
experiment by Agrawal et al. (2006). In section 3 we show that patents contain enough
information to measure social distance quite accurately, recall a few notions of social
network analysis, and apply them to a large set of patent applications in organic chemis-
try, pharmaceuticals, and biotechnology, signed by US inventors. In the same section, we
provide descriptive evidence on the extent of inventors' mobility across i rms and space,
and on the resulting shape and geographical features of the social network of inventors.
In section 4 we use our patent sample to reproduce Agrawal et al.'s version of the JTH
experiment and to show that inventors' mobility across i rms and social ties between
inventors largely explain both the original JTH and Agrawal et al.'s results.
In the Conclusions we emphasize that our evidence casts some doubts on the common
interpretation of citation-measured knowledge l ows as pure externalities, or spillovers,
and outline our future research plans.
2.
The JTH experiment: methodology and interpretation
Methodology
The JTH experiment starts with the selection of a sample of
originating
(cited) patents.
For each originating patent, all subsequent patents that cite it as prior art are then col-
lected, after previous exclusion of company self-citations, that is, pairs of citing-originat-
ing patents assigned to the same company.
1
The address of inventors recorded in patent
documents is then used to assign patents to a geographical area, in order to compare the
locations of citing and originating patents.
2
A
control
sample of patents is also built. Each citing patent is matched to a randomly
drawn patent, with the same technology class and application date, but no citation link
to the corresponding originating patent.
A test follows, which consists of comparing the frequency with which citing-
originating patent pairs match geographically (in our experiment, at the city level) to
the corresponding frequency for control-originating patent pairs. If the former turns
out to be signii cantly greater than the latter, this should be interpreted as evidence of
localization ef ects of spillovers
over and above
the agglomeration ef ects arising from
other sources.
3
