In our view, the results in this paper demonstrate that there is much to be learned about international knowledge diffusion from patents and their citations. Despite the fact that we focus on patents granted by the U.S. patent office, rich patterns of interaction are revealed, including interesting findings about the diffusion of citations within and between countries other than the U.S. Some widely-held notions about differences in the inventive processes across countries were confirmed, such as the reliance of the Japanese on a relatively recent technological base. Others are less obvious, such as the strong symmetry between citing and cited intensities, and the greater proximity of Japan to the U.S. relative to Europe.

Overall, the results confirm our earlier findings that there is significant geographic localization of knowledge flows. We can now, however, tell a more complete story about the localization process, distinguishing the issue of speed from the issue of total intensity, and describing how citations diffuse over time to more distant locations. In future work, we intend to extend this in two directions. First, we will continue to look at more and finer geographic distinctions, including other countries and regions within the U.S. We conjecture, for example, that the West Coast of the U.S. is “closer” in technology space to the Pacific Rim, while the East Coast is closer to Europe, for both geographic and cultural reasons.

The second research avenue we are pursuing is to relate the knowledge flows implied by the citation patterns to the commercial impact of invention as measured by productivity improvements. If citations are a proxy for the pathways by which the cumulative impact of new technology is brought to bear, then they ought to play in a measurable intermediating role between the R&D series of various countries and the international productivity series. Thus our estimated citation flows can be used in place of trade flows to construct weighted stocks of foreign R&D to search for international R&D spillovers as in Park (1995) and Coe and Helpman (1995).