Geography Reference
In-Depth Information
(Owen-Smith and Powell, 2004). The dense patches may rel ect i rms, for instance, or
geographic regions. Actors within each patch sit socially proximate to one another but
relatively distant from actors in other patches. A question of great practical importance
is: When does knowledge dif use within the patch where it originated but not across
the thin areas into other patches? When will knowledge dif use within a i rm but not to
competitors, or within a region but not to other locales? When is inequality of knowledge
sharpest across social borders? Our results suggest that the nature of the knowledge,
specii cally its degree of complexity, plays a critical role. One might initially suspect that
highly complex knowledge, the most dii cult to reproduce, would create the greatest
inequality across boundaries. Yet this intuition ignores the fact that inequality in its
sharpest form requires
some
dif usion: to create the most inequity across social bounda-
ries, knowledge must creep up to the edge of the thick patch of connections in which it
originated but not beyond. This phenomenon, we have argued, most likely occurs for
moderately complex knowledge.
Accordingly, the results suggest a resolution to the replication/imitation dilemma that
has puzzled evolutionary economists and strategy scholars. To achieve a competitive
advantage from knowledge, a i rm must typically leverage that knowledge across mul-
tiple applications, for example, across all its production facilities (Winter, 1995). Yet
any would-be replicator with a valuable piece of knowledge faces a dilemma: the proi ts
produced by its original knowledge attract the envious attention of imitators. Valuable
knowledge provides a source of
sustained
advantage only to the extent that it lends itself
to replication yet dei es imitation. Unfortunately for the innovator, replication and imita-
tion typically go hand in hand (Nelson and Winter, 1982). Our results suggest, however,
that replication-without-imitation is especially likely when the target knowledge entails
moderate complexity. This micro-level phenomenon may manifest itself in outcomes at
the industry level. One might expect that,
ceteris paribus
, industries based on moderately
complex knowledge will display especially wide intra-industry dispersion in long-run
i nancial returns. We leave this promising hypothesis for future research.
The results also speak interestingly to the literature on the geographic agglomeration
of industries. Researchers frequently cite knowledge spillovers as a prominent reason
that i rms within an industry cluster together (Krugman, 1991; Marshall, 1890) and
congregate near universities (Zucker et al., 1997). Our results certainly support this
point of view: dense social networks, which tend to localize geographically, give i rms
and individuals close to the source of knowledge an important advantage in reproducing
and building on the knowledge. This begs the further question, why do some industries
cluster while others do not? Though research on economic geography points out that
knowledge spillovers can contribute to agglomeration, it does not identify
what type of
knowledge
most likely engenders these clusters. Our i ndings suggest that industries that
rely on moderately complex knowledge more commonly form industrial districts (cf.
Sorenson, 2004). Simple knowledge can dif use far and wide because incremental search
ef orts can substitute for high-i delity communication. As the complexity of knowledge
increases, a gap emerges between local dif usion and distant dif usion; thus, the potential
return to locating near to innovators rises.
In addition to inl uencing geographic agglomeration and industry structure, the
nature of the underlying knowledge used by a i rm may have implications for organiza-
tional design. Firms have both formal and informal structures that inl uence the degree
