INTERNATIONAL ECONOMIC REVIEW August 1999 Vol. 40, No. 3 INTERNATIONAL TECHNOLOGY DIFFUSION: THEORY AND MEASUREMENT* By Jonathan Eaton and Samuel Kortum 1 Boston University and NBER, U.S.A. We model the invention of new technologies and their diffusion across coun- tries. In our model all countries grow at the same steady-state rate, with each country’s productivity ranking determined by how rapidly it adopts ideas. Re- search effort is determined by how much ideas earn at home and abroad. Patents affect the return to ideas. We relate the decision to patent an invention interna- tionally to the cost of patenting in a country and to the expected value of patent protection in that country. We can thus infer the direction and magnitude of the international diffusion of technology from data on international patenting, productivity, and research. We fit the model to data from the five leading re- search economies. A rough summary of our findings is that the world lies about two-thirds of the way from an extreme of technological autarky to an extreme of free trade in ideas. Research performed abroad is about two-thirds as po- tent as domestic research. Together the United States and Japan drive at least two-thirds of the growth in each of the countries in our sample. 1. introduction What is the geographic scope of technical progress? One camp holds that by its very nature technology is freely available everywhere. A questionable implication is that countries enjoy no relative advantage from being innovative. 2 At the other extreme, the new growth theory typically relates a country’s technical advances to only its own innovations. A troubling implication here is that innovative countries leave everyone else behind. 3 In contrast to either polar position, economic historians describe world growth in terms of the gradual diffusion of advances from a small set of innovators. * Manuscript received September 1997; revised April 1998. An earlier version appeared as “Inter- national Patenting and Technology Diffusion,” NBER Working Paper No. 4931, 1994. 1 Akiko Tamura and Deepak Agrawal provided excellent research assistance. We take responsibility for any errors. We gratefully acknowledge the support of the National Science Foundation under Grant No. SBR 9309935-001. 2 A corollary is that differences in worker productivity across countries must result from differences in capital per worker, where capital may be construed very broadly to include human capital. Mankiw, Romer, and Weil (1992) take this stand. 3 See, for example, Romer (1990) and Agrion and Howitt (1992). 537