Restoring Japan’s Leadership In Innovation, Part II
Japan Ranks Very Low In International Collaboration in Science & Commercial Inventions
Source: All the charts in this post come from OECD Science, Technology, and Industry Scoreboard 2017 at https://read.oecd.org/10.1787/9789264268821-en?format=pdf
It’s remarkable how many of the new generation of successful Japanese entrepreneurs have had significant international experience. Some have studied or worked overseas while others have worked for a foreign firm in Japan. Yes, the kind of people likely to become entrepreneurs are also more open to a variety of experiences. Beyond that, however, overseas experience opens the mind. It teaches people that there are many ways to skin a cat, leading them to ask the kind of questions and see the sort of options that those without such experience do not.
CEOs who have had managerial experience overseas generate better firm performance than those with no such experience because having to work in a different business culture makes their thinking more fluid. That’s why CEOs at nearly 90% of leading companies in Europe and North America have had such experiences. In Japan, the number is only 17%, according to a Booz Allen survey. For similar reasons, when foreign-owned firms in Japan conduct R&D, they get produce seven times as much spillover productivity benefits for other firms than do domestic firms spending the same amount. Says one scholar, that is “probably because knowledge of foreign firms is often new to domestic firms.”
This is why Japan’s dearth of open innovation with startups (as discussed in our recent blog) is so damaging to innovation. It’s not simply that today’s products incorporate too many different technologies for one company to master, e.g., 10-15% of a car’s value these days comes from electronics hardware and software. It exposes older firms to fresh ideas.
International collaboration is a vital part of open innovation. Different countries excel in different kinds of ideas. It was the presence of Japanese auto transplants on US soil that finally got the Detroit Three to realize the competitive importance of stopping defects before they get incorporated into vehicles rather than trying to find and fix them later. Similarly, Japanese corporate giants had no idea how much they lagged in digital trends until the Silicon Valley Japan Platform started bringing top CEOs to Silicon Valley almost a decade ago. Unfortunately, across almost the entire range of international collaboration, Japan ranks the lowest, or near the lowest.
Japan’s Scientists Insulated From Global Trends
Japan has no problem in producing scientific and technological advances. In the last couple of decades, more than 3,400 research organizations and companies, of which about 40% are universities, have produced more than 1.8 million papers and articles.
Unfortunately, only a thin slice of this work involves international collaboration. That’s a problem because the OECD has found that, when scientists in any country work with foreign counterparts, the quality of their work improves. Both Japanese scientists who have worked abroad and foreign scientists who come to work in Japan produce “more important” papers, as measured by how many other scientists cite their work. Only in the most influential journals, do papers in which Japanese scientists collaborate with foreign counterparts outnumber those involving just domestic authors.
Looking at all journals, Japan comes in dead last among 30 OECD countries on the share of scientific articles on which they collaborate with scientists from other countries, just behind Korea.
Note: Foreign-led means a paper in which a foreign scientist was the lead author, and a Japanese scientist was a co-author, while domestic-led means the other way around.
By contrast, in a somewhat narrower index of papers, Australian, British, Canadian, Dutch, French, and German scientists now produce more papers involving international collaboration than ones with only domestic authors.
There is one positive exception to Japan’s overall trend. Among Japan’s top 12 research institutions involved in international collaboration, that collaboration has increased. In 2009, 21% of scientific papers produced at the University of Tokyo involved international cooperation; that increased to 28% by 2017. While this is welcome news, it’s an exception to the general rule.
Only 2.7% of Japanese scientists have spent time working overseas. Among the rest of the OECD, the average is 6.5%. Only 1.3% of Japan’s scientists are foreigners, compared to an OECD average of 4.4%. On this measure, Japan comes in second-to-last, again just behind Korea.
. This isolation starts early. Not only is there a drop in the number of Japanese university students studying abroad, but only 2.1% of such students are science majors, the 3rd lowest in the OECD. By contrast, 70% of those studying abroad are in fields of education, the arts, journalism, social science, business, or law. The problem is not that there are too many non-STEM (Science, Technology, Engineering, Math) students studying abroad, but that there are too few STEM students
Things are very different at the doctoral level. Among Japanese Ph.D. candidates who are internationally mobile, 29% of them are studying the natural sciences. That’s not so far from the OECD average of 37%.
It’s valid to ask how much of Japan’s low ranking on most of these measures stems from the fact that fewer Japanese are proficient in English, the global language of science. While I cannot find comparable data for past years on co-authorship and the like, I do have data on the number of Japanese students studying abroad. That number almost quadrupled from 18,000 in 1986 to 65,000 in 2003. Then the numbers fell back to only 30,000 in 2016 (UNESCO 2022). Moreover, only a couple thousand spent at least a year abroad. Language does not explain either the rise or the subsequent plunge. Among 27 OECD countries, Japan comes in second-to-last in the share of students who get a college degree from a foreign university, just ahead of the US. I suspect the time trend on international mobility of scientists and their co-author with foreign scientists would show similar trends. Language may explain some of Japan’s gap, but it does not seem to be the primary problem.
Firms Engage In Little Cross-Border Open Innovation
Companies fail as badly as scientists in collaborating across borders. Only 1.3% of all Japanese patents include a foreign company or researcher as a co-inventor, the lowest rate in the OECD. Second to last is Korea at 3.6%. The average is 14%. Japan also came in last in international collaboration on Information and Communications Technology (ICT) patents.
. Among large companies, only 8% engage in any collaborative innovation with foreign companies, the second lowest share in the OECD. The OECD average is 40%. Korea’s record is even more dismal.
. It’s important to note that there is a big correlation between Japan’s low rate of international collaboration among scientists and the low rate of co-invention by companies. Partly this stems from a common cause: Japan’s low rate of globalization across many dimensions. Beyond that, collaboration among scientists and joint inventions among companies mutually promote each other. The more countries do one, the more likely they are to do the other (see chart at the top).
Procurement From Outsiders Is Not Open Innovation
All too often, when leading Japanese companies talk of their plans for open innovation, they mostly mean licensing of someone else’s patents or making more purchases of complex equipment from companies outside the corporate group, e.g., outside the hundreds of companies in the Toyota group. That’s good, but it’s not open innovation. The latter means genuine two-way collaboration on developing new products or processes. It requires training technical staff in how to collaborate because it requires a different skill set than just choosing what part to buy or what patent to license.
Still, the resistance to open innovation stems from many of the same attitudes that have traditionally hindered purchases of core equipment from outside the corporate group. Consider Honda. It tried to develop sensors in its vehicles that could avoid collisions. When it tested the sensors by using a dummy child, Honda’s SUV mowed it down. It scored only 0.2 points out of 25, the worst of all tested vehicles. It finally gave in and bought off-the-shelf sensor technology from Germany’s Bosch. The score soared to an almost perfect 24.4. While Honda’s CEO has said the company is now more open to buying from others, many of its R&D veterans resist on the grounds that using only home-grown parts is “Honda’s soul.” Across many leading companies, the complexity of modern products has led to more procurement from outsiders. One can hope that this trend will promote more receptivity to open innovation.
Re-engineering the Firm
For many years, the Ministry of Economy, Trade, and Industry (METI ) has been trying to induce firms to embrace open innovation. So far, it has not worked, in part because desire does not by itself translate into capability. Henry Chesbrough, the professor who invented the term “open innovation,” worked with Japanese companies that had sent people to Silicon Valley to partner with firms there. In a report to a METI affiliate, Chesbrough concluded that, “Many internal changes will be required to Japanese firms’ organizational processes in order to use open innovation to work more effectively with Silicon Valley.” And with others as well—at home and abroad.
Just a couple of biases overlooked here:
1. Most OECD countries are contiguous with another OECD country. Traveling abroad isn’t a big deal for scientists in Europe or North America, e.g.: you just need a car. So their travel stats will be higher.
2. If you base international collaboration on co-patentee status, that suggests you are looking at assignees, not inventorship. Any decent IP lawyer will tell you that co-ownership of a patent is a pain in the rear. When I was working in the patent section of Applied Materials, we avoided such situations as a matter of policy. That is not the same thing as co-inventors coming from different companies (which would be a more relevant indicator of collaboration, but is not necessarily evident from the patent). Language in a joint development agreement can (and should) allocate ownership on a different basis than inventorship. As a result, looking for co-patentees will hide instances of co-inventorship.
3. Some of what you’re seeing reflects a difference between, esp. US R&D philosophy and Japanese. At least in the industries I worked in (semiconductors, electronics), US companies trying to develop a new widget would work on several projects in parallel, and see which one would reach the spec first. Some of these projects might be entirely in-house, but the rest would be collaborations.
Japanese style, OTOH, was to try to guess which is the one “best“ technology, gamble on that, fail and then repeat. Because each effort was one-at-a-time, engineering groups with a strong NIH bias (quite common when I was at Sony) were a real choke point for reducing the scope for collaboration. This doesn’t contradict your point, but helps to explain why there are fewer collaborations.
This is not to negate entirely the notion that Japanese academia and industrial R&D can be very insular. But quantitatively your stats might not give a clear picture.