Innovation Process and Performance: Findings and lessons from inventors surveys in Japan, the U.S., and Europe

Until now companies' creative processes, which play a key role in innovation, have been invisible to outside observers. A recent survey of Japanese inventors shed light on these processes and will enable research that may help to enhance innovation performance. Survey responses were obtained from 5,300 inventors, and we would like to thank all the respondents. Building on the results of an earlier European survey, RIETI collaborated with the Georgia Institute of Technology to conduct an inventor survey in both Japan and the United States. We anticipate that international comparative analysis will clarify the structural characteristics of Japanese-style innovation. Honored by the participation of world-renowned experts Professor Richard R. Nelson and Professor Bronwyn H. Hall, we hope that this symposium will make a significant contribution to innovation research.

Based on the results of inventor surveys conducted in Japan, Europe, and the U.S., this session focused on both the backgrounds for and the facts discovered about innovation processes and differences among the three regions.

Keynote Speech 1 interpreted innovation from an economic perspective, while Presentations 1 and 2 outlined aspects of the Japan and U.S. inventor surveys. Keynote Speech 2 offered suggestions on policy initiatives based on the results of the European inventor survey, while Presentation 3 used data gleaned from the Japan inventor survey to discuss inventor incentives.

Overall, the following key points emerged from the session.

• Results from Japan, the U.S., and Europe were surprisingly very similar in many aspects. Competition in research and development (R&D) is globalizing, and at the same time a global inventors' community is developing. It was also suggested that all countries are facing common issues with respect to innovation.
• The inventor surveys offered insights into existing misconceptions relating to the nature of innovation policy, and pointers for the future.
• It is important that corporate innovation management and national innovation policy should be complementary. Stimulating a "taste for science" throughout society lends impetus to innovation.

Professor Hall's speech offered insights into the determinants of innovation performance from an economic perspective.

(1) Innovation as an economic phenomenon
Many kinds of indices such as growth rates, profitability, and productivity at the firm, sector, and country levels are used to measure the performance of innovation. How far economic explanations can be applied to inventions, particularly to radical inventions such as the Internet, is a moot point. But money and markets are required to turn invention into innovation, and economic analysis can play an important role. Nelson and Arrow characterize innovation as the production of new information. New information causes increasing returns within a firm but also holds the potential for imitation by competitors, due to the non-rival and non-excludable nature of innovation. Thus methods of handling the special nature of such information are an important issue at both the corporate and the policy level.

(2) Determinants of innovation: demand, supply, and environment
Economists often think in terms of demand and supply. Innovation supply includes factors such as the cost of capital including interest and tax rates, venture capital availability, and scientist or engineer capabilities and numbers. Public research institutes also accumulate scientific knowledge and are involved in supply of technology development opportunities. Research to date indicates that:

• R&D tax credits are effective in increasing R&D. However, this does not necessarily increase the innovation success rate.
• Venture capital accounts for an extremely small proportion of investment. However, it plays a major role in the selection of startup companies.
• All countries recognize the difficulty of creating effective links between public research institutes and industry.

Innovation demand includes factors such as firm size, market structure, competitive pressure, consumer tastes and demand from downstream firms, and appropriability. In extremely competitive market environments, corporations do not profit from innovation. On the other hand, in monopolistic market environments corporations only have weak incentive for innovation. Most markets fall between these two extremes. Companies use factors including lead times, first mover advantage, corporate secrets, complementary assets, and patents to secure returns through innovation. Over recent years patents, in particular, have grown in importance.

The environment for innovation consists of the macroeconomic environment, government regulations, education systems, public-private research interaction, and standard-setting. Together, these form a "national innovation system."

Summary

Based on research to date, we can say the following about how the above factors affect innovation.

• Large corporations and established companies innovate more than new entrants and startups. Continuous R&D matters for innovation.
• Demand is more decisive than supply in determining the success of innovation.
• Innovation output statistics are much more variable than corporate R&D input statistics, showing that major uncertainty is a characteristic of R&D.
• There is additionality of government support, but there are no consistent results on the relationship between government support for innovation and innovation performance.
• Results show that corporate product innovation and process innovation are complementary, suggesting that overall strategy is important for companies. Government innovation policies also need to be complementary.
• Effective venture capital requires thick financial markets to provide exit.
• Innovation does not progress if most of the scientists and engineers trained in a good educational environment are channeled into secure government lab jobs.
• Rapid increases in R&D funding tend to raise the salaries of scientists and engineers, but may reduce the effectiveness of R&D activities.

The first two presentations reported the main findings from the inventor surveys conducted in Japan and the U.S., and comparisons between the two countries.

(1) Purpose of inventor surveys

Until now there has been no systematic body of information on innovation processes in Japan and the U.S. at the R&D project level. Inventor surveys serve to resolve this dearth of information and to identify and elucidate various phenomena relating to innovation processes, such as industry-academia collaborations, the relative importance of product innovation and process innovation, and knowledge spillover.

(2) Inventions targeted by the survey

Responses were obtained from 5,300 inventors in Japan and 1,900 in the U.S. Triadic patents (patents filed in Japan and Europe, and granted in the U.S.) accounted for 70% of the Japanese responses, and all of the U.S. responses. Given that just 8% of all patents granted by the Japanese Patent Office to Japanese applicants are triadic (and just 23% of the patent grants in the U.S. to applicants from all countries are triadic), the surveys dealt primarily with high-quality patents. The patents came from a wide range of technology fields including chemistry, biotechnology, semiconductors, IT, and metal processing.

Characteristics of Japanese and U.S. inventors revealed by survey results

• Average age
  The average age at the time of patent application in the U.S. (47) was higher than that in Japan (around 39).
• Academic background
  Approximately 90% of inventors in both Japan and the U.S. were university graduates. In Japan the proportion of those with Ph.Ds was a little over 10%, while in the U.S. it was a little under 50%.
• Females
  In both countries few of the inventors were women (1.5% in Japan and 5.4% in the U.S.). These figures are low even in comparison to the proportion of women researchers (10% in Japan and 25% in the U.S.).
• Organizational affiliation
  In both Japan and the U.S., 80%-90% of inventors worked for large corporations of 250 or more employees (the proportion was higher in Japan than in the U.S.). In both countries very few of the inventors were employed by universities.
• Mobility
  Inventor mobility in the U.S. was twice as high as in Japan, even including temporary assignments and secondments for Japan (12% in Japan including temporary assignments and secondments, 26% in the U.S. within the five years preceding the survey). The mobility gap was especially large between the two countries in the high-tech fields of biotechnology, medical instruments, semiconductor devices, and computer software. Corporations in both countries tended to recruit their inventors from universities and non-competitors within the same industry.
• Functional affiliation
  In both Japan and the U.S., some 20% of patents originate from non-R&D departments.
• Motivation
  Inventors in both countries stated that their primary motivation was the satisfaction gained from solving technical problems or contributing to scientific progress, as opposed to monetary reward.

Japan-U.S. similarities and differences in invention processes revealed by survey results

• Goal of research projects
  In both Japan and the U.S., there is a strong orientation toward research projects aimed at product innovation rather than process innovation.
• Targeted research versus serendipity
  Fifty percent of inventions were the targeted achievement of a research project in both countries. However, in the U.S. more than 10% of inventions were unexpected by-products unrelated to the initial main goals of the research project. Such serendipitous outcomes were three times higher than in Japan (3.5%).
• Man-months required for invention
  The distribution of man-months put toward research in Japan and the U.S. was very similar, with 50% of patents requiring 12 or fewer man-months and 80% requiring 24 or fewer man-months.
• Internal and external collaboration
  In both countries more than 10% of inventions were co-inventions with external collaborators. Co-inventions resulting from vertical collaboration between suppliers and customers were more common than those resulting from horizontal collaboration between competitors. Twenty to 30% of collaborative R&D projects involve external collaborators with the exception of co-inventions. Collaboration with universities, including co-inventions, was very low in both countries.
• Sources of information leading to invention
  The main sources of information useful for invention were (1) scientific and technical literature, (2) patent literature, (3) customers or product users, and (4) knowledge existing within one's own firm in both countries. External information in the form of scientific and technical literature was more important in suggesting new inventions, whereas customers and product users were important in contributing to project completion. In Japan, patent literature was more important than scientific and technical literature, unlike in the U.S.
• Business objectives of research
  Enhancement of existing business was the most common goal in both Japan and the U.S., but more Japanese respondents cited this goal (Japan 67%, U.S. 49%). However, more U.S. respondents cited the goal of long-term cultivation of technology seeds not associated with current business (U.S. 24%, Japan 8%), and the disparity between the two countries was the greatest in semiconductors, information storage, computer software, and optics. Looking at the relationship between patent value and man-months expended, for core businesses the value tended to be greater the higher the number of man-months expended, whereas for discovery of new technology seeds the relationship was weaker, suggesting that other factors are more important.
• Use of inventions
  In both countries 50% of patents were used within the company where the invention was made. With the exception of semiconductors and biotechnology, Japan showed a higher proportion of licenses. U.S. companies tended to use patents exclusively in-house, whereas Japanese companies tended to license them. More patents were used for startups in the U.S. than in Japan (U.S. 7%, Japan 4%), and this difference was especially marked in biotechnology (U.S. 20%, Japan 6%). Only 20 percent of the inventions can be implemented as a stand alone, meaning that in most cases multiple patents were required for commercialization. An important reason for not using an invention was that the patent had originally been filed for blocking purposes. In both countries 20% of patents were filed for purely defensive purposes.
• Appropriation strategies
  In both countries, first-mover advantage was the most important strategy, but in the U.S. comparatively more emphasis was placed on first-mover advantage in market entry, while Japanese respondents stressed that in follow-up development of complementary technologies and the patent portfolio. The tendency for U.S. inventors to place importance on first-mover advantage in market entry was very pronounced in semiconductors and information storage. While U.S. companies placed greater importance on patent enforcement, this was less important than first-mover advantage.
• Sources of funding
  Internal funding (including borrowing) accounted for 90% of financing in both Japan and the U.S., while the proportion of funding from venture capital or government programs was low in both countries. However, the U.S. showed a higher level of venture capital funding in biotechnology (U.S. 8.6%, Japan 1.2%). Government programs represented a higher share of funding in the U.S. than in Japan (U.S. 5.7%, Japan 4.5%). Financial constraints were greater at the commercialization stage than at the invention stage, and even for R&D related to core business financial constraints were experienced in 8% of cases.

Conclusions

• Despite institutional differences, invention and innovation processes in Japan and the U.S. are remarkably similar. In particular, very similar results emerged in areas including the small proportion of university inventors and inventions resulting from industry-university collaboration; inventor motivations; time input for inventions; level of use of patents; and the roles played by suppliers, customers and product users.
• Differences between Japan and the U.S. included high inventor mobility in the U.S. and more exploratory R&D with a quantity of serendipitous inventions by U.S. firms. U.S. companies placed higher importance on first mover's advantage and patent enforcement, whereas Japanese firms tended to emphasize licenses.

General Comments by Professor Richard R. Nelson

Commending presenters on their outstanding research, Professor Nelson made some general comments including the following points.

• Some 35 years ago, Project SAPPHO had already identified customers and product users as one of the most important sources of knowledge for corporate innovation. At that time it was thought that innovation supply (technology push) was constant over the short term, but now circumstances are different. However, as today's presentations have shown, suppliers, customers and product users remain the key sources of information for companies, just as they were 35 years ago.
• Innovation is categorized into product innovation (invention of "things") and process innovation (invention of "methods"), but we should bear in mind that these categorizations are relative concepts. A process innovation in a downstream company may be a product innovation for an upstream company.

Questions and Answers

The following questions were received from the floor.

Q. Has the comparison between Japanese and U.S. results been statistically tested?
A. Statistical assessment has not yet been completed. We plan to release statistically tested results in the final report.

Q. The presentation mentioned that university professors account for a small proportion of patents and that there is only limited collaboration with universities, but doesn't this undervalue the role of universities?
A. The most important role of universities is training the people who will become inventors. The survey results also show that universities are a key source of human resources. From the perspective of patent filings, university professors certainly represent a small share, but this does not negate the most important fundamental role played by universities. We would like to further clarify this role in future studies.