International, intra-national and inter-firm knowledge diffusion and technological catch-up: the USA, Japan, Korea and Taiwan in the memory chip industry

Abstract

Using data on US patent citations, this paper investigates the pattern of international, intranational and interfirm knowledge diffusion in the process of technological catch-up by Japanese, Korean and Taiwanese firms in the memory chip industry. First, regarding international diffusion, this paper finds that the ordering of citations is exactly the same as the order of entry into the industry: Taiwanese firms tend to cite Korean firms, Korean firms tend to cite Japanese firms and Japanese firms tend to cite US firms. Second, the degree of intranational knowledge diffusion is proportional to the level of technological capability or order of entry, although it is also affected by organisational differences among the firms. Third, the difference in patterns of interfirm knowledge reflects difference across organisations, such that big Korean group firms are less oriented toward interfirm knowledge diffusion compared with their Taiwanese small and medium enterprise (SME) counterparts. To explain such difference, the role of government research institutions has been highlighted, especially since the Industrial Technology Research Institute (ITRI) accounts for the lion's share of Taiwanese-held patents and in the spin-offs of many firms in the industry.

Keywords:

• knowledge diffusion
• catch-up
• semiconductor industry
• Korea
• Taiwan
• Japan

Acknowledgements

Earlier versions of this paper were presented at several meetings, such as the 2006 Globelics Conference, October 2006, held in Trivandrum, India; a seminar at the Innovation Research Center of the Hitotsubashi University, Japan, May 2006; and the international conference on ‘Economic Cooperation and Development in the North-East Asia’, organised by the Korea Development Economic Association in October 2004 in Incheon, KOREA. The authors acknowledge useful comments by two anonymous referees, as well as Franco Malerba, Hiro Odagiri, Jinyoung Kim, Sangchul Lee, Namhoon Kang and other participants at these conferences, and the research assistance provided by Ms Hyesung Kim and Hayoung Kim.

Notes

This paper further extends the earlier work by Lee and Yoon (2004) to discuss more countries with wider angles and more rigorous methodologies.

The USPTO does not have assignee codes and instead uses assignee names. Therefore, it can be that one assignee may have various assignee names. Cases such as these were checked and cleaned up for this study.

While it is plausible that Taiwanese firms hire Korean engineers, we do not have hard data on this. To check on this in terms of patent citations, we investigated the inventor data of all Taiwanese DRAM patents and found no reference to any Korean national. However, it is also possible that Taiwanese firms did not use the true inventors’ names during patent application since scouting for Korean engineers is apparently an internationally sensitive issue that is related to industrial espionage.

One might also consider the measure of pure intranational diffusion as the ratio of (A)/(B), rather than the gap, (A) – (B), as proposed herein. However, if we follow this ratio, problems may occur. For instance, for Japan (A=50% and B is 30%) and for Korea (A=4% and B=2% ), the result of (A/B) yields a higher value of 2 for Korea as compared with 5/3 for Japan. In contrast, taking the gap (A – B) yields a higher value of for Japan at 20% as compared with 2% for Korea. More importantly, subtraction should be conducted because we aim for a measurement on probability; if we make a division, units are lost (this should not be larger than 1) and thus cannot be interpreted as a probability.