The R&D Process in the U.S. and Japan: Major findings from the RIETI-Georgia Tech inventor survey

執筆者 長岡 貞男  (研究主幹) /John P. WALSH  (Georgia Institute of Technology)
発行日/NO. 2009年3月  09-E-010


This paper analyzes and compares the objective, the nature and the performance of R&D projects in the US and Japan, based on the first large scale systematic survey of inventors, focusing on the R&D projects yielding triadic patents. Major findings are the following. First, the projects for enhancing the existing business line of a firm account for a large share of R&D projects in both countries, confirming the view that the R&D investment is significantly conditioned by the existing complementary asset of a firm. In both countries, the inventions from R&D for existing business have the highest in-house utilization rate but use least the scientific and technical literature for their conceptions, while the reverse is the case for the inventions from R&D for new technology base (or for cultivating seeds). R&D projects for enhancing the technology base are much more common in the US. This difference can be partly accounted for by US inventors being more likely to have a PhD, but not by the differences in the structure of finance. US government financial support is relatively more targeted to projects for existing business and US venture capital provides support mainly projects for creating new business (6% of them), but not for more upstream projects. Only about 20-30% of the projects are for process innovation in both countries, providing direct evidence for the earlier findings that were based on US patent information. Product innovation generates process patents more often in Japan than in the US (25% vs. 10%), while product innovation projects are relatively more numerous in Japan. In both countries a significant share of inventions (more than 20%) were not the result of an R&D project, and a substantial proportion of such inventions are valued among the top 10% of patents, suggesting that R&D expenditure significantly underestimates inventive activities. A US invention is more often an unexpected by-product of an R&D project (11%) than in Japan (3.4%). The two countries have surprisingly similar distributions of R&D projects in man month and the average team size. In both countries, smaller firms tend to have relatively more high-value patents. In the US, inventors from very small firms (with less than 100 employees) and universities jointly account for more than one quarter of the top 10% inventions, even though they account for only 14% of all inventions. Man-months expended for an invention has a significant correlation with the performance of the R&D projects for existing business, less so for new business and not at all for those enhancing the technology base, suggesting substantial heterogeneity by project types in the determinants of the performance and in the uncertainty. A PhD has a significant correlation with R&D project performance especially for new business.