Patent citations and the geography of knowledge spillovers: A reassessment
Citation: Peter Thompson, Melanie Fox-Kean (2005) Patent citations and the geography of knowledge spillovers: A reassessment. The American Economic Review (RSS)
Internet Archive Scholar (fulltext): Patent citations and the geography of knowledge spillovers: A reassessment
Tagged: Economics (RSS) Spillovers (RSS), Geography (RSS)
Thompson and Fox-Kean's article revisits the work of Geographic localization of knowledge spillovers as evidenced by patent citations by Jaffe, Trajtenberg, and Henderson (1993) (called JTH in their paper and in this summary). JTH tests for the presence of knowledge spillovers by looking at citation patterns of patents. Essentially, they find that patents are much more frequently cited by patents from geographically close areas as measured by the metropolitan region, state, and country. They take this as evidence of knowledge spillovers.
The real innovation in JTH was creating a method to control for the fact that industries tend to be geographically concentrated. For example, if there are lots of semi-conductor producers in Silicon Valley, you would expect to see more citations from that area and this would not necessary be evidence of geographic concentration. JTH do this by using the patent classifications and creating a control set of patents in the same fields or classification and seeing how much more often the geographically close patents are cited than the otherwise similar but geographically far away patents are cited.
Thompson and Fox-Kean's basic observation is that the patent system classification systems used in far from perfect and that it is including many matched patents which are not actually in the same industry. As a result, the evidence of geographic concentration ends up capturing industry concentration and the numbers of inflated. Thompson and Fox-Kean go through and use a more detailed form of classification which is also more labor intensive on a subset and find that the effects at the metropolitan region and state disappear, but that country-level spillovers seem to be robust.
Theoretical and Practical Relevance
In Patent statistics as economic indicators: a survey, Zvi Griliches (1990) makes the basic observation on the problems associated with the mismatch between patent classification schemes and industry. The importance of this basic observation is driven home closely by Thompson and Fox-Kean. The article has been cited more than 130 times since it was published 5 years ago in the economics literature on knowledge spillovers. It shows a much reduced effect and an important qualification on JTH's influential article.