How do the Innovation Systems of the U.S. and Japan Differ? What are the Potential Implications? Evidence from the RIETI-Georgia Tech Inventor Surveys

John P. Walsh
1. Research objective and method
The system of global innovation is changing, and due to its importance to economic growth, these changes warrant further analysis. Bibliometric indicators are overemphasized in innovation policy analysis; thus other aspects of innovation need to be explored.

Our analysis surveyed inventors rather than research and development (R&D) managers. The sampling frame we used was the Organisation for Economic Co-operation and Development (OECD) Triadic Patent Families. This sampling frame avoids a home country bias and focuses on economically important patents. However, it may be biased toward commercialized inventions and may be biased against non-profit, small and/or independent inventors.

The data was collected through mail and the Internet. The Japanese survey was undertaken between the winter and spring of 2007, while the U.S. survey was done in the summer of 2007. Over 3,600 responses were collected in Japan, with priority years being 1995-2001, and around 1,900 responses were collected in the U.S., with priority years being 2000-2003.

2. Who invents? (career, background and motivations)
The vast majority of inventors had a college degree, while 45% of U.S. and 12% of Japanese inventors held doctorates. Female inventors were few and the median age of all inventors was 47 for the U.S. and 39 for Japan. Although American inventors are, on average, seven years older than their Japanese counterparts, American inventors begin and end their careers later than Japanese inventors. Around 80% of both U.S. and Japanese inventors worked for large firms. An important difference, though, is that 12% of U.S. inventors were from very small firms (firms with fewer than 100 employees) versus only 5.1% in Japan. In both countries 2.3% of inventions were developed by inventors who worked at a university, regardless of whether the university owned the patent or not. Additionally, roughly 30% of patents in the U.S. were invented by people born overseas.

In Japan, job mobility is relatively low, with 10% of inventors having worked in different job locations in the last five years, versus 25% in the U.S. In Japan, half of job mobility experiences were secondments rather than changes of employer. Only 3% of inventors at large firms in Japan have changed companies, thus the vast majority of inbound inventors at Japanese large firms are from secondments. In both countries the rate of all kinds of inbound mobility increases as firm size decreases. In the U.S., inventors who move between companies are more likely to use external information in R&D, including university research, scientific literature, customer feedback, and competitor information, while they are less likely to use information from within the firm. While not uniform across all categories, Japanese inventors with secondments are more likely to use external information than those who stay with one firm or those who have changed companies.

In terms of job motivation the results were, in general, similar across the two countries. The top answers were "satisfaction from solving technical problems," and "satisfaction from contributing to science," while the bottom answers were "monetary rewards" and "beneficial working environment," respectively. However, U.S. inventors were more likely to respond favorably to "generating value for my firm" than Japanese inventors. More specifically, inventors at very small firms in Japan are much less likely to be motivated by generating value for their firm as compared to their U.S. counterparts.

Sadao Nagaoka
3. Business purpose of R&D and invention process
On average, Japanese research is more focused on strengthening existing business than is the case in U.S. firms. In terms of capability requirements, the probability that Japanese Ph.D. holders are involved in the research is the highest for the research to "enhance the technology base of the firm," then for the research to create new business lines, and then the lowest for that to strengthen the current business line. In the U.S., the distribution of Ph.D. holders is more even through the three categories of research, but follows a similar pattern. The use of scientific and technical literature is also most common for the research in enhancing the technology base of the firm in both countries. Research projects to create new business are highly important for very small firms in the U.S., while the very small firms in Japan are more likely to have projects aimed at enhancing the competitiveness of the existing business lines. As for government versus venture capital funding of the R&D projects by business objective, Japanese and U.S. firms have similar patterns for what kind of research they are more likely to receive the government support. The U.S. is more focused on existing business and less focused on enhancing the technological base. In Japan, venture capital support of research is near negligible, while the level of the funding of research to create new business lines by venture capital in the U.S. exceeds such funding by the government. However, venture capital plays little role in exploratory (enhancing the technology base) research or research to enhance current business even in the U.S.

Comparing product innovations and process innovations also exposed differences in each country. Japanese firms were more likely to be focused on creating new products than on improving current products, while improvement of current products was more important in the U.S. On the process side, both countries focused more on improvement of existing processes than on creation of new ones.

In both countries, 50% of patented inventions were the targeted outcomes of their research projects. Both countries also have similar amounts of inventions that come from non-R&D activities and inventions that do not involve R&D at all. However, 11% of the U.S. patents in the survey were an unexpected outcome of a research project; it is only 3.4% for the same category in Japan. As Japanese firms are more focused on improving existing business, unexpected by-products, or serendipity patents, are less common. Also, expected by-products were more common in Japan, accounting for 12%.

The distribution of patent values varied for both countries by firm size. The analysis was focused on inventions with values in the top 10% as those inventions count for 80% of the entire economic value of all inventions. In both Japan and the U.S., very small firms are more likely than other firms to produce the patented inventions in the top 10% of value as a share of the inventions generated. University patents are not significantly different from most other types of firms in terms of value in Japan. However, U.S. universities' patents are much more likely than any other type of firms to be valued in the top 10%.

John P. Walsh
4. How "open" is innovation? Collaboration/sources of information
The survey then measures how open innovation is in the U.S. and Japan. Primary types of collaboration include: co-assignment, whereby two or more organizations are named on the patent; cross-organizational co-inventors, where the inventors come from different organizations, but the patent may only go to one of the organizations; and collaboration partners, where people from different angles (competitors, suppliers, customers, universities) all take part in the inventing process.

Collaboration with competitors and non-competitors in the same industry is not common. University collaborations account for 2.5% and co-inventors from any other organization account for about 13% of collaborations in both countries. As such, the distribution of collaboration partners is similar among the U.S. and Japan.

Vertical collaboration (with suppliers, customers, etc.) is more common in the U.S. and Japan. The distribution of firms engaging in vertical collaboration is relatively even in the U.S. across firm sizes, whereas there is great difference in collaboration among Japanese firms. Smaller Japanese firms are much more likely to engage in vertical collaboration. This difference is likely due to dissimilarities in firm structure between the U.S. and Japan.

For both countries, the figure for co-inventors for patents is roughly 10%. However, the number of co-assignees on patents is less than 2% in the U.S. Smaller firms in both countries are more likely to have co-inventors hail from a university. However, Japanese large firms are much more likely than U.S. firms across the board to have collaboration partners from universities. From this data, it can be seen that universities and firms in Japan have very different relationships than those in the U.S. In both countries the fields of drugs and biotechnology are the most common areas of co-invention between universities and private firms.

Medium-sized firms in the U.S. are much more likely than their Japanese counterparts to receive government funding. Also, U.S. venture capital funds go mainly to very small firms. Japanese universities are more likely to receive funding from private firms than are U.S. universities. However, this figure includes university professors working for a company on a certain project, so it does not necessarily constitute direct investment in university R&D projects.

Patent literature, potential users, knowledge within a firm, and scientific and technical literature are the most important sources of knowledge for researchers. Knowledge from competitors scores relatively high in this regard in Japan but not in the U.S.

In general, projects that had different kinds of co-inventors tend to have higher technical significance.

Sadao Nagaoka
5. Commercialization
In both countries, around 40% of all patents do not proceed to any commercialization. In the U.S., 40% of inventions are used exclusively internally versus 34% in Japan. The Japanese firms license 23% of their patents and U.S. firms only license 13% (Note, however, that U.S license cover only a license to non-related party). Also, patents are more likely to be used for business startups in the U.S. than in Japan. Japanese firms are, on average, willing to license 40% of their inventions but only 20% of inventions ever come to licensing agreements. The same is generally true for U.S. firms. Willingness to license does not necessarily change with the size of the firm, but actual licensing increases as firm size declines. Thus, large firms have many inventions they find difficult to license.

As for the use of inventions for startups, 35% of U.S. universities' patents are used for startups, a huge number compared to the ratio of the use of inventions for startups by the other types of firms in the U.S. or Japan.

Commercialization rate in Japan tend to low for the patents from R&D meant to expand the technological base, as these ventures can be more risky. On the other hand, those patents from R&D meant to expand existing business are commercialized at a rate of 60%.

The size of the group of the complementary patents necessary for commercialization is very similar between the U.S. and Japan. Only 20% of U.S. and Japanese patents can be commercialized as standalone products. All other inventions require complementary patents, the most common for both countries being bundles of two to five patents in one commercialized product, accounting for almost 50% of commercialized patents. In these markets where patent bundling is more important, firms are forced to license. As the size of the necessary patent bundle increases, the frequency of licensing and cross-licensing increases dramatically.

The "first-mover advantage" in commercialization is more highly regarded by U.S. inventors. Also, U.S. inventors appreciate internal exploitation of new technologies based on patents protection more than Japanese inventors.

6. Tentative conclusions
In conclusion, despite institutional differences, surprisingly, there is quite a bit of similarity across the board between the two countries in terms of their innovation systems.

Among the differences, first, the U.S. possesses a higher rate of inventor mobility, thus facilitating greater information flows. Second, the share of foreign-born researchers in the U.S. is significant. Third, very small firms in the U.S. have unique innovative capability, compared to those in Japan and are very important to the U.S. economy. Fourth, using R&D to develop new business lines is just as likely in the U.S. as it is in Japan. Fifth, the U.S. has the advantage in exploratory research with more unexpected, or serendipity, outcomes. Sixth, while the level of university participation in patented inventions is similar between the two countries, U.S. universities play a larger role and develop higher value patents. Additionally, the U.S. has a much higher number of university-based startup companies. Lastly, Japanese firms put more emphasis on licensing while U.S. firms put more emphasis on "first-mover advantage" and patent enforcement.

As for potential policy implications, our findings are as follows. Encouraging mobility across firms and across borders may contribute to increased technological progress. University entrepreneurship requires ideas and flexibility in terms of employment and funding. Exploratory research requires high internal R&D capabilities on the part of the firm. The significance of scientific literature for an invention highlights the importance of supporting "open science." Considering a large gap between the willingness to license patents and the actual licenses, an innovation policy could encourage the development of the markets for technology.

DOI Ryoji (Consulting Fellow, RIETI / Director, Research and Development Division, Industrial Science and Technology Policy and Environment Bureau, METI)
Do you think that the high number of unexpected by-products in U.S. R&D is due to the high number of Ph.D. holders engaged in R&D? The exploratory research collaboration between universities and companies may also lead to this. Also, is the difference between the U.S. and Japan in terms of collaboration with suppliers and competitors significant or negligible?

Sadao Nagaoka
We have not fully analyzed the causes of the U.S. having a higher rate of unexpected by-products, but I can tell you that the R&D projects which produce serendipity involve researchers with Ph.D.s significantly more often than the other projects in Japan

John P. Walsh
As our error rate is between 1% and 2%, the difference between collaborations in the U.S. and Japan are within that range for both suppliers and competitors. Not much emphasis should be put on the differences in either of these categories.

Sadao Nagaoka
We are planning to do statistical tests to validate whether our major results are really as significant as they seem.

Q: As for the unexpected by-products in the U.S., do you think that since Japanese researchers tend to work in groups, in contrast to American researchers who generally work alone, the Japanese are more likely to follow through to the expected outcome rather than exploring other possibilities?

Sadao Nagaoka
A: It seems that independent thinking is important for undertaking risky, uncertain research which may result in unexpected discoveries. We have the information on team sizes and can check whether single-person research tends to come up with unexpected outcomes more frequently or not. Thanks a lot for your very useful suggestion.

Q: Do you think the preponderance of university startups in the U.S. is related to the importance of very small firms? Also, how do you respond to the findings from METI's recent small- and medium-sized enterprises initiative that such firms are not significant in overall R&D value?

Sadao Nagaoka
A: I agree with the view that university startups are a major source for the high-tech startups in the U.S. The priority years of the Japanese firms surveyed are 1995-2001, which was before this initiative. In order to gauge the overall economic importance of the startups, we would need to have more data.

Q: In terms of the U.S. having 30% foreign-born inventors, what is the major source of heterogeneity among U.S. co-inventors?

John P. Walsh
A: Slide 43 refers to heterogeneity across organization types. This chart does not deal with heterogeneity in terms of country of origin, rather, it talks about inventors coming from different types of firms, including universities, suppliers, customers, and others. Basically, it is the number of different organizations that are represented in the research group.

Q: What do you think about heterogeneity of inventions in terms of types of patents made?

Sadao Nagaoka
A: Vertical collaborations are beneficial for research meant to strengthen current business. An exploratory research gets benefitted most from university collaboration. These are the preliminary results, and we will look at this question in more detail later on.

Q: What is your overall view of the differences between U.S. and Japanese R&D? Do you find that Japan is more or less productive than the U.S. in terms of R&D?

John P. Walsh
A: The results tend to be quite similar, but there are fundamental differences as to how those results have come about. For example, the relative role of large versus small firms, the relative role of licensing, etc.

Sadao Nagaoka
A: While Japan seems to have a functioning R&D system, the small differences in the systems of the U.S. and Japan may have long-term implications for the two economies. The cultivation of certain features of the U.S. system that the Japanese system currently lacks, like startups and venture capital, will take time, but could have a large effect on the sustained growth of the Japanese economy and the Japanese R&D.

Q: Do you see the large number of foreign-born researchers in the U.S. as compared to the almost complete absence of such persons in Japan as a significant difference between the two countries?

John P. Walsh
A: The U.S. clearly benefits from this situation. Foreign-born inventors generally have more valuable inventions, having been educated either in the U.S. or abroad. However, after the September 11 attacks, there is evidence to show that the situation for foreign-born inventors in the U.S. has become more difficult in terms of entering the country and staying in the country for an extended period of time. Due to the contributions of foreign-born inventors, further restrictions on immigration will have a negative effect on U.S. R&D.