Geographic localization of knowledge spillovers as evidenced by patent citations
Jaffe et al. present an economic analysis that purports to show support for geographic localization of knowledge and the existence of spillovers. The article looks at patent citations and finds that citations to patents are more likely to be domestic than foreign and more likely to be from the state or even from the same metropolitan area. They find that this localization fades over time but does so slowly.
The paper begins by pointing out that there has been a large literature on knowledge spillovers but that there is very little evidence as to where this knowledge goes. The authors suggest that spillovers will be geographically correlated.
Most of the paper is an in-depth description of the empirical methods that used the authors employ and to overcome challenges in, for example, assigning a geographic location to a patent or creating a group of control patents that can control for the fact that innovation in particular areas or fields may tend to cluster geographically and that this should not necessary be taken as evidence of spillovers. The authors create a group of control patents which are on the same topic and filed for at the same time and use these as a means of controlling for existing geographic concentration of research.
The results that the authors find are strong: citations are 5-10 times as likely to come from the same metropolitan area as control patents. Although this effect fades over time, it is robust to the addition of time. Also of interest, the authors find little difference between citations of university and corporate patents.
Theoretical and practical relevance:
This article has been cited more than 3,000 times and is a core citation in the literature on patents in practice, spillovers, and geographic localization of knowledge which, over time, has been a field of growing importance in economics, in innovation studies, and in strategy. It also provides a solid set of methodological contributions to the literature on innovation and intellectual property using patent citation data.