Patenting in different countries differs in ways that affect observed citation frequencies. Some indications of these differences are presented in Table 1, which summarizes the patent data, from both the cited and citing perspectives, for the five different countries. We also classify patents into five broad technological fields, based on the main patent class assigned to the patent by the patent examiner. These fields are: Drugs and Medical Technology; Chemicals, excluding Drugs; Electronics, Optical and Nuclear Technology; Mechanical Technology; and All Other. Table 1 gives some totals for patents and citations for each of the five countries and five technology fields. Overall, 6% of the cited patents are in Drugs and Medical, 28% in Chemicals, 22% in Electronics, etc., 35% in Mechanical and 9% All Other. There are, however, significant variations across the countries in the field composition of their patents. In particular, Japan has a larger share of electronic patents and Germany a larger share of chemical patents than the U.S. Since citation intensities vary by field within countries, raw differences between countries as in Figure 1 are a mixture of country effects, field effects, and field-country interaction effects. We discuss below how to sort out these effects.

Table 1 also shows that a significant fraction of citations from each country are “selfcitations.” Self-citations are defined as those for which the citing and cited patent are both assigned to the same corporate organization. Self-citations are more common in the U.S. than in other countries. It also turns out that self-citation come more quickly on average, and are more geographically localized. In order to get measures that more closely correspond to knowledge “spillovers,” most of the analysis below is carried out excluding these self-citations.

Finally, the number of patents taken out in the U.S. has grown at dramatically different rates for different countries. In particular, while the number of U.S. invented patents in 1993 was essentially equal to the number in 1969, the number of Japanese-invented patents increased by 1194% over that same period, and the number of Great Britain-invented patents declined by about 26%. Thus when we compare overall citation frequencies for the different countries, we are looking at averages which are tilted towards different citing cohorts.