Industry and state partnership: The historical role of the engineering research associations in Japan
In 1986, Professor Jon Sigurdson, then Director of the Research Policy Institute in Lund, Sweden, published a monograph entitled Industry and State Partnership in Japan: The Very Large Scale Integrated Circuits (VLSI) Project. Amazingly enough, given that the VLSI project had been formed ten years earlier, and had completed its work by 1980, and given the universally acknowledged significance of the project in raising the Japanese semiconductor industry to world competitive heights, this small monograph was one of the first accounts, in English, by a non-Japanese, of the structure and organization of the project and its effects.1 Sigurdson’s monograph had the special characteristic of placing the VLSI project in the context of Japan’s long experience with collaborative research, utilizing the institutional framework of the Engineering Research Association. It was also informed by lengthy interviews with the project’s instigators and leaders, and provides case studies not only of the VLSI project itself, but also of several of the preceding ERA projects, and of the firms that were drawn into the VLSI project and benefited from the experience. Sigurdson utilizes the theoretical notion of “ultrastructure” to describe this Japanese institutional innovation of the ERA, in an attempt to capture the sense of its providing a template or framework for the encouragement of collaborative R&D. This too was one of the first efforts to create a theoretical framework for understanding the success of Japanese technological catch-up efforts.
Professor Sigurdson’s text has long been out of print, and is available only to a handful of scholars who received copies and in a handful of libraries. What follows is an edited version of the original text, leaving out some of the institutional details, particularly of the VLSI project itself, which are now available from numerous sources, but retaining the organizational analysis of the ERA framework and the lively presentation of the people involved in the projects and their perspectives.
John A. Mathews
Australia-Asia Management Centre, ANU, Canberra
PRESENTATION BY JON SIGURDSON (1998)
My research project on the Engineering Research Associations (ERAs) was conceived in the early 1980s, just after Professor Chalmers Johnson had published his pathbreaking work, MITI and the Japanese Miracle. Japan’s efforts to catch up with the West were coming to fruition, and the country was emerging as a technological superpower and was in all aspects the dominant technological and industrial power in Asia Pacific. The ERAs appeared to be an intriguing but under-studied aspect of Japan’s developmental state institutional arrangements.
Already in Sweden I had realized the significance of technological partnerships between Japanese companies and government agencies. During an exploratory trip in 1982, I met with some of the officials of some projects organized under the auspices of the Next Generation Basic Technologies Project (Jisedai). Then in the summer of 1983 I became a guest researcher at the Graduate School of Saitama University where Professor Toru Yoshimura had attracted a number of prominent scholars in political science and technology management. The latter included Professors Fumio Kodama and Taizo Yakushiji.
The Graduate School maintained close relations with MITI; mid-career officers were on long-term assignment in classes and one MITI official was a member of the faculty. This greatly facilitated access to MITI bureaux and laboratories. The breakthrough to developing a research project on the ERAs came from an introduction by Professor Kodama to one of his earlier informants within the Toshiba Corporation. This man introduced me to the extensive network that had developed among companies arising from the VLSI project, which had run from 1976 to 1979.
The mapping of such networks for the VLSI Project and other ERAs became the main focus of my project, aside from tracing the early development of ERAs and their evolution, in terms of organization and outcomes. Tarui, leader of the VLSI project, was at the time overexposed but many others were very eager to give their time and insights, although my faulty knowledge of Japanese set certain limits on my understanding. They exposed me to the informal structures in industrial Japan and substantially modified my view on the role of MITI and its ability to guide Japan’s technological development.
Japan at the time had basically completed its technological catch-up phase. The consequences for government policy makers and for managers of corporate strategies were not altogether clear although companies were reacting more quickly to market developments. With the knowledge of hindsight it would have been more rewarding and more challenging to try to grasp the essence of this shift. The need for a transformation of technology policies is apparent in my attempt to assess the success and failures of ERAs. When looking back on my research in Japan in 1983-85 I feel content to have discovered important aspects of industry-state partnerships and the creation of networks. I was fortunate in my timing in that these networks subsequently were taken up as a model around the world.
1. The institutional innovation of Engineering Research Association: its origins and its adaptation by Japan
2. The emergence of research associations in Japan
3. The VLSI project
4. Organizational evaluation of the VLSI project
5. The Engineering Research Association as a policy instrument Appendix: Engineering Research Associations in Japan, 1961-1983
THE INSTITUTIONAL INNOVATION OF ENGINEERING RESEARCH ASSOCIATION: ITS ORIGINS AND ITS ADAPTATION BY JAPAN
Engineering Research Associations are an institutional means of promoting collaborative R&D work between companies, either on their own initiative or in partnership with a public agency. They have been used to great effect by the Japanese in their technological catch-up efforts, particularly during the 1960s and 1970s. But the Japanese did not invent the institution of the ERA. Rather, as with most of their innovations, it was an adaptation of a previously existing institution, which traces its origins to Europe and initially to British research during the First World War.
In 1917 the British government started to give financial aid to industries which organized research and technical activities on a co-operative basis. The scheme was introduced by the then newly established Department of Scientific and Industrial Research. The objective was to meet the acute technological needs of British industry, the shortcomings of which had been revealed at the outbreak of the First World War. The scheme provided for government grants matching those of member companies in each of the research associations (RAs). It had originally been intended that the grants should be discontinued once a research association had been fully established. Had this policy been followed, most of the associations would have disappeared. The policy was formally changed in 1945 to provide permanent grants. The scheme has since then been introduced in a large number of European countries, in principle patterned on the British model. The RAs continued to be important in Britain right up to the 1960s.2
The British RAs were successful in promoting certain industrial areas, particularly in enabling consortia of small- and medium-sized companies to match the scale and resources of large firms. As a rule the research associations in UK as well as in most other industrialized countries have catered more for small and medium enterprises which have been relatively weak as far as their own R&D efforts are concerned. A major OECD review also claims that those industries, which have a large number of members whose overall research efforts are not highly developed, are most suitable for the research association concept.3
The British RAs served as inspiration and in certain aspects as a model for a similar policy instrument in Japan-the engineering research associations (ERAs). However, there are also major differences and, it should be noted, the concept of research association has undergone a dynamic development in Japan-which is not the case in most other countries where the research associations or their functional equivalents have remained rather static. In order to understand the development of the ERAs in Japan as a dynamic policy instrument it is necessary to trace its origins in the period shortly after the Second World War.
One person can be regarded as the father of engineering research associations in Japan. This is Dr Masao Sugimoto, who is now serving in an advisory capacity at Hitachi Ltd, his first employer after graduation. His influence on the introduction of this particular policy instrument and its adaptation to the country’s needs can be traced to the following three factors. First, he was in an important decision making position as director, in the late 1950s, of the Mechanical Engineering Laboratory (MEL), affiliated to MITI. Second, he became aware of the British research associations at an opportune time. Third, he had a keen sense of the needs and possibilities in the Japanese companies in the early post-war period.
Dr Sugimoto became director of MEL in 1953, after having joined the laboratory around 1947.4 While discussing his responsibilities at MEL in the early stage after the war, he points out that almost all the Japanese industry had been destroyed. At the time all main designers were making simple things like kettles. Japan was forbidden to make trucks, bulldozers and other heavy machinery as well as aircraft. So the objective was to export small and light machines like bicycles, sewing machines, cameras, ball bearings, valves and so on. “Our research laboratory had much work to do to develop such products in order to build up the export-oriented mechanical industry”, Sugimoto says. He adds: “My task was to develop mechanical industry in Japan.”
Almost all manufacturers of cameras, sewing machines, bicycles were at that time comparatively small- or medium-sized companies. The major financial company groups like Mitsui, Mitsubishi and Sumitomo, known as zaibatsu, had been dismantled by the Allied Powers and the big companies like Mitsubishi and Hitachi barely managed to continue. For example the Mitsubishi group had been divided into three parts. The Hitachi group was not divided but almost all its factories had been completely destroyed.
After the Second World War Japan was not allowed to engage in jet engine research and manufacture and many engineers like Sugimoto went to work in other industries. The ban was eventually lifted and in this connection Dr Sugimoto, who was a specialist in the field, was sent on a study tour for six months in 1953 which took him to Italy (Fiat), Switzerland (Oerlikon), France and UK and finally to USA where he visited, amongst other places, the predecessor of NASA. As he had an interest in the British research associations, Sugimoto wanted to establish contact with the Department of Scientific and Industrial Research in the UK in order to arrange an additional study programme. This was refused because the research associations were private and served British industry and Dr Sugimoto had been sent by the Japanese government. It was understood that the engineering research associations were guarding the secrets of private companies.
However, Dr Sugimoto found that he “could get printed material” and he was able to learn about various results of the associations through published booklets. In those days the booklets on associations’ activities contained minute details, Sugimoto notes, which is no longer the case. As a consequence it was possible to obtain a fairly accurate picture of the engineering research associations in UK. Dr Sugimoto mentions that he did not learn about similar organizations in other countries but he says, “perhaps I was the first (Japanese) person to know” about such associations in UK. Later on it was possible for a few members of the Mechanical Engineering Laboratory to go and work in British engineering associations.
When Sugimoto returned to Japan he wrote several articles on the subject, which were published in the Journal of the Japan Society of Mechanical Engineers as well as in various other publications. He also lectured for the Chamber of Commerce in Nagoya and other places. At the time, Sugimoto notes, there was an urgent need to “level up the technology of small firms” which is well illustrated by the need of the makers of optical glass for eye-pieces. This sector at the time had many small manufacturers with very poor technology.
When introducing the concept of engineering research associations in Japan they came to be established for a very specific purpose at a very specific time. Dr Sugimoto recapitulates the early introduction in the following way. When he was director of the Mechanical Engineering Laboratory a president of a car radiator company contacted him (around 1955) and told him: “We have to establish a research association.” Sugimoto asked: “What is the purpose?” and the president said that he could not answer. The following evening Dr Sugimoto was asked to come to a restaurant in Akasaka in central Tokyo where five or six heads of car components companies had gathered. He was then told that “frankly speaking, they had no purpose” in setting up a research association. They only wanted to do research, they said. He was then told that “the purpose must be considered by you”. It was further explained to him that the radiators were only produced on the basis of trial and error, without scientific knowledge. They took their products to car companies like Nissan and Toyota where engineers examined the radiators. They were often told that their radiators were no good and were never given any explanation as to why the products were rejected. They realized that research was needed but frankly admitted, “we have no knowledge about how to organise research. You, Dr Sugimoto must show us how”.
This episode may be taken as the starting point for the government’s involvement in engineering research associations in Japan. A first association established to improve radiator technology, according to Dr Sugimoto’s recollections, was created in 1956. Then followed a second research association for air cleaners and several more for other car components. The decisions for setting up the associations were generally made by the vice-minister of MITI and Dr Sugimoto, director of the Mechanical Engineering Laboratory.5
The objective of the Mechanical Engineering Laboratory was at the time to support small- and medium-sized companies. The laboratory’s importance was greater immediately after the war than it is today. Now, Sugimoto notes, research can be done in many places-in universities and in company laboratories. At that time there was only one mechanical laboratory in Japan. However, some of the MEL researchers did not like to assist small- and medium-sized companies as this would never lead to “Doktor-Arbeit” or excellent papers which could be published in foreign journals. Such troublesome attitudes remain in government laboratories to this day, Sugimoto notes. He argues that government laboratories-even today-should exist for solving practical engineering problems and not for academic work.
So the foremost purpose of MEL and the engineering research associations (ERAs), at this time, was to help small- and medium-sized companies to develop small and light machinery. After some five or six years reliability and quality improved. Consequently, Japanese products could be exported to many countries and there was little need for further support from ERAs. So it is to be expected that no new ERAs were established in the second half of the 1960s.
In discussing the introduction of the engineering research associations in Japan it has already been revealed that they existed informally several years before MITI enacted a special law in 1961.6 The Mechanical Engineering Laboratory (MEL) under the directorship of Dr Sugimoto was the key actor on the side of the government and his story about the initial contacts from the auto parts companies has already been recounted.
In the following paragraphs I will present the view of Mr Suzumoto who was the counterpart for the auto parts makers. Mr Suzumoto, who is now director of the technical department of the Japan Auto Parts Association (JAPIA), mentions that this organization was established in 1957. According to him the first research association, for filters, was established in 1955 while the “radiator association” was established two years later.7 The third association, for air suspension equipment was established in 1960 and the fourth one for engine parts followed in 1961. The latter was to do research on pistons, piston rings, crankshaft bearings, connecting rods and camshaft bearings. All four associations, which were established before 1961 were under the direct control of JAPIA, which received a subsidy and then distributed it to each of the associations. It should also be mentioned that the government had urged the auto components makers to establish JAPIA. The contribution from the government was quite modest. JAPIA received 0.1 billion yen for the buildings which consisted of central facilities and eventually a small research building for each of the four associations. The central facilities were established to handle data collection and provide a photo-laboratory. There was also a conference room and accommodation for 20 people, to be used for engineers and researchers from the participating companies.
Suzumoto was executive chief engineer for the facilities, which were to remain in the same place until 1970 when they were moved to Higashi Kurume and placed under the leadership of the Engineering Research Institute of the Machinery Promotion Association (Kikai Shinko Kyokai Gijutsu Kenkyu Jo). The role of the centre has changed over time; more recently it has focused on pollution problems.8 Aside from the government subsidy to the buildings the associations received very modest support, as shown in Table 1.
What was the benefit of the early associations? According to Suzumoto: “Each centre had participating companies. They could not do their own research. JAPIA provided equipment and companies could do research. If there was only one company the costs for failures would have been 100 per cent. With 5 participating companies the costs were reduced to 20 per cent which were further reduced through the government subsidy.”
At the time Japan did not make passenger cars (only 3-wheelers) and so the automotive components associations concentrated initially on problems to do with the manufacture of buses and trucks. The companies involved were often competitors and had their own specific requirements for product development, which could not be disclosed. Testing therefore provided a common ground for co-operation within the associations. Aside from the practical results of testing components and finding improved production methods the car component research associations also played a very important educational role. Each of the participating companies sent their engineers to the joint facilities where they stayed for shorter or longer periods. Several of these young people later attained key positions in their companies.
The optical technology research association
The story of the camera association established under the special ERA law is quite similar.9 In this case many of the organizational ideas were borrowed from the Scientific Instrument Research Association (SIRA) in the UK. Mr Namera who was with the association from the very beginning mentions that the establishment took place in 1962, one year after the Engineering Research Association Law was enacted. Initially there were 44 member companies but these were eventually reduced to 36. By the time the association was disbanded, in 1981, the member companies had grown in size and sophistication, so that they were able to sustain their own independent technology development. A trade association (kyoka:) was formed immediately after the research association closed.
Mr Namera claims that Japan’s camera technology had in 1955 already reached the level of the manufacturers in West Germany, and it was necessary to move into new technological fields. There was a need for the camera makers to introduce the use of electronics and plastics in order to compete more efficiently in the world market. Thus the timing of the camera association was very good.
At first the association could not afford its own research laboratory and so the research was done in universities and in member companies. The research themes were selected by the member companies, which included Nikon, Canon, Olympus, Konishiroku (Konica), Fuji Film and others. However, the research leadership was often provided by university professors who had been selected by the company members in various research committees. Thus professors at Tokyo University, Osaka University, Waseda University and Yamamachi University, and others, played at that time a very important role. The camera association had two special committees, for lens quality and for lens calculations. Computers were introduced for lens calculations as early as 1960.10 A few years later, in 1963, research started on the design of automatic lenses, using computers, and Namera stresses that the quality of Japanese lenses today owes its existence to that particular initiative. Another important project also started by the association was research on automatic exposure control.
In summing up this early experience, Namera says that the most important function of the association was that it accelerated research, which was carried out by the camera makers themselves. However, the companies needed leadership at the time, and the association was able to provide that leadership in directing the research. The budget allocations from the government were relatively limited, with MITI providing approximately 30 per cent of the overall budget.11
THE EMERGENCE OF RESEARCH ASSOCIATIONS IN JAPAN
It has already been shown that the interests and initiative of a key individual in the Mechanical Engineering Laboratory triggered the establishment of the Engineering Research Associations as a technology policy instrument which was subsequently strongly supported and further developed by the Ministry of International Trade and Industry. The research associations were seen by MITI as a powerful tool to increase the country’s competitive posture in certain industrial sectors, with a strong focus on computers and related information technologies. MITI officially saw the benefits from ERAs as including: risk-sharing and cost sharing between participating units; pooling of resources to speed up the research process and eliminate overlaps; and ensuring that research would cover all aspects of product development. A further benefit was seen as the exchange of technical information which would raise the technological capabilities of an industry across the board.12
The apparent success of the research association concept for the auto component manufacturers prompted the formalization of the ERAs, in a law passed in 1961. Early on, the Japanese government provided grants that usually matched the contributions made by the private companies. There is little doubt that the contribution made by the government was much more important in its indirect catalytic effect-although this is difficult to quantify. However, it is important to realize that, although based on a British concept, the ERAs at the outset became different from its UK model and, even more important, have constantly changed in character, reflecting the technological needs in Japanese society, or rather the needs of certain parts of the industrial spectrum in Japan.
The original purpose of the engineering research associations in Japan was to carry out research for industrial companies to meet their immediate and specific needs. The initiative for establishing an ERA came in the earlier period from industry, to serve their basic needs, and not from the government. This is also reflected in the longer duration of the associations during the 1960s. In more recent years the initiative for establishing an Engineering Research Association has often come from MITI or its Agency for Industrial Science and Technology (AIST) where the laboratories often have served as a breeding ground for new technological developments. This is particularly so in the case of the ElectroTechnical Laboratory (ETL), which spawned several ERAs.
The ERAs were established as non-profit making organizations which are established according to certain rules contained in a law which was promulgated in 1961, when the first ERA-on high polymers-was established. Since then, up to the end of 1983, a total of 71 ERAs have been established, involving more than 500 companies. Details of these ERAs, including their topic of development, their time span and their sponsoring agency, are provided in the Appendix. There was an initial burst in creating ERAs in the early 1960s (as discussed), followed by a lull, and then a renewed period of formation of ERAs in the 1970s-as shown in Table 2. A major contributing factor for this was a need to find new organizational forms for the large-scale national projects, which had been initiated by the Agency for Industrial Technology (AIST) in 1966.
From the beginning the Japanese approach to ERAs differed in important respects from its UK model. These differences increased with time, particularly recently. Firstly, only a few of the ERAs, established during the period 1961-1965, had a large number of participants. The largest one was the camera association, which was formally called the optical technology research association, with 44 members. This provided the technological basis for continuous improvement for the manufacture of cameras and is seen as an outstanding success. The average membership has been about 12 companies per ERA since 1971 against 17 in the 1960s, as shown in Table 3.
None of the ERAs in Japan was a permanent organization. The average length of existence was 11.5 years, for the ERAs established during the 1960s. Using similar statistics, available only for those ERAs that have been wound up, it appears that the length of existence has been drastically shortened to around six years. As for the size of ERAs, the modal membership was between five and nine companies, as shown in Table 4.
Major companies constituted the dominant force within the ERAs. The role of the major companies becomes apparent from the fact that some 30 major companies account for 258 memberships (out of a total of 878) in the 71 ERAs. These companies fall into eight industrial sectors-fibres, chemicals, glass, steel, non-ferrous metals, machinery, shipbuilding and electrical machinery. The last could more appropriately be called information technology industries and will be discussed below in the VI.SI project. In the early period there was a substantial membership of smaller companies but with a few exceptions this is no longer the case. Thus in contrast to their British and other European counterparts, the Japanese engineering associations are at present chiefly concerned with technology development as such rather than with technical assistance to small- and medium-sized companies.
In all sectors with the exception of electrical machinery industries, the ERAs have in the main been geared to singular technology projects, which have served the diversification interests of the companies. In most of the eight sectors (again with the exception of steel and again electrical machinery) the participating companies can be seen as having common interests. Five major steel companies have participated jointly in seven ERAs and it can be assumed that a network has been established through this process. Most of these ERA projects are basically serving the diversification interests of the companies, which of course is a sound indication of the restructuring of the Japanese economy.
Establishing an ERA
The ERAs are established under the guidance of MITI. They have the objective of organizing the major industries concerned to solve technical problems common to the sector or to a smaller group of companies. The initiative for establishing an ERA often comes from MITI and is based on ideas from their government laboratories, university professors and, naturally, opinions from the industries concerned. On the basis of reports and plans, which are usually carefully worked out, an ERA is established as a legal entity made up of the companies who were keen to join or coerced to do so by MITI.
The formal proposals for a new ERA generally come from one of two different sources. First, private companies are asked individually to submit proposals. They are asked to consider total projects and indicate in which particular areas they are able to make a contribution. Second, MITI bureaucrats are also asked to submit proposals. Such proposals may reflect ongoing discussions between MITI and the trade associations of private industry, for example Japan Electronic Industry Development Association, which has close and frequent contacts with the Industrial Electronics Division of MITI.
The formal procedure for establishing an engineering research association usually includes the following steps:
1. Tenders (research proposals) are invited by MITI (through the relevant bureau);
2. Companies are evaluated on the basis of their technological and financial capabilities;
3. The companies (often around 10-12) are selected to form an ERA;
4. MM suggests that an association be established;
5. A core company is (usually) designated which is given coordination responsibilities;
6. The ERA is formally established as a legal entity;
7. A contract is negotiated between MITI and the ERA, and signed.
A government official explained the choice of organization for technological development in the following way. The government, or rather MITI sees four main organizational alternatives in channelling funds to support company-based industrial research:
Joint contracts-where the government and several companies enter into a contractual agreement to carry out research with a certain objective;
Non-profit foundation (e.g. the Fifth Generation Computer Project);
Engineering Research Association;
Individual contracts with each separate company.
There is little doubt that MITI officials prefer the ERA to the other alternatives with the possible exception of the “foundation”. From MITI’s point of view the ERA is much more efficient in terms of resources required for its initiation and supervision. Furthermore, it is usually quite easy to identify responsibilities within an ERA. Responsibility for making the ERA an institutional vehicle of choice lies principally with the Agency of Science and Technology (AIST). Originally the agency signed joint contracts with a number of private companies for specific research projects. However, the responsible AIST officer had quite a lot of trouble obtaining the necessary consensus among the participating companies. Then, a senior officer, Mr Shimada, promoted the idea of using the ERA as an instrument for AIST policies in high technology sectors and this was first tried out for the direct reduction of steel using nuclear energy in the early 1970s.13 In recent years all Large-Scale Projects organized by AIST have generally included an engineering research association.’4
By the late 1960s the original policy instrument for promoting large-scale projects had become obsolete. Co-operative research, conducted on a voluntary basis, was still being organized but the government provided its support in indirect ways by offering benefits such as tax reduction and depreciation allowances. Shimada says about the organization of large-scale projects (LSPs) as ERAs: “a) changed the interpretation of the law-but not the law itself, when I was assigned to a nuclear steel making project as ‘kaihatsukan’-i.e. development officer for a big project. This happened around April 1972 … I was looking for the proper arrangement. Maybe it was necessary to have a new law. Then I found the old (ERA) law within bureaux. I asked for permission (to use the law). There was a lot of discussion … The purpose was only to use the law for Large Scale Projects (LSP). Since then the ERA concept has been used for all LSPs.”‘5 However, ERAs have also been used as a vehicle for joint research by other agencies as well, as revealed in the Appendix.
The rapid increase in the rate of formation of engineering research associations in the 1970s is attributable to three factors. First, the “oil shocks” in 1973 and 1979 brought home to Japan its precarious energy situation; thus the Agency for Energy Resources started a number of ERAs. Second, a number of associations have recently been established to revitalize raw material industries such as paper and aluminium. Third, the Jisedai (Future Generation Basic Technologies) Project, which was initiated in 1981, includes three ERAs. The shift in the focus of engineering research associations since the early 1960s has also meant that the membership has come to be dominated by the large companies listed on the stock exchange. The underlying reason is that the large companies are more capable of undertaking research in Large-Scale Projects. Research results are then generally transferred from the central research laboratories of the big companies to their subsidiaries and their subsubsidiaries.
Contrary to the notion widely entertained abroad, most ERAs do not contain any joint facilities. Where they do so, it is for clear and compelling reasons. First, a joint laboratory may be established in order to save research funds in projects where large amounts of expensive equipment are needed. This avoids unnecessary duplication in several company research laboratories and/or government laboratories. The VLSI and the Optoelectronics projects exemplify such a situation. Second, the number of qualified researchers may be very limited and the required research can only be conducted efficiently if such scarce resources are pooled into a joint laboratory. This is true in the case of Gallium Arsenide (GaAs) researchers in the Joint Optoelectronics Laboratory. Third, joint facilities are called for where the projects are exceedingly complex. Examples include the jet engine (turbine) project, the Flexible Manufacturing System (FMS) project and the Sub-sea Oil Production System project. In these projects the system concept is very important and the necessary integration of research on various sub-systems, carried out by individual companies, can only be done efficiently by establishing a central design centre. This was first tried out in the very important, Very Large-Scale Integrated Circuits project-the most famous of Japan’s ERAs.
THE VLSI PROJECT
The VLSI Project was sponsored for four years by MITI to the sum of 30 billion yen, with the participating companies providing an equal amount. Many foreign observers, often highly qualified in technology or economics or both, consider this project to have provided the underpinning of the market penetration that major Japanese companies were able to make for large-scale integrated circuits at the very beginning of the 1980s.
The co-operative VLSI Project was established by five rival companies which were all producing general-purpose computers-except one, Toshiba, which had already stopped manufacturing mainframe computers. These companies pooled their researchers and worked for four years in a joint project started in 1976 which was the first of its kind. In this context it is interesting to note that the VLSI Project was actually thought of as a response to a rumour about development within IBM. Some influential people in Japan, at least for a certain period of time, thought that the American computer giant was far ahead in developing a one-chip-computer as an integral part of its Future Systems.
The background of the project, according to its leader Professor Tarui, can be traced to new requirements for higher packing densities of integrated circuits in the mid-1970s. This in turn called for new sophisticated manufacturing technology. Thus a proposal was made by Professor Shoji Tanaka, from the Department of Applied Physics at Tokyo University and Japan Electronics Industry Development Association JEIDA) to establish joint research in order to develop the required new technologies. A decision was taken to analyse the research requirements for a VLSI Project with the five companies as participants and with financial aid from MITI. This was not altogether a new idea as the companies had been involved in technology partnerships in the preceding years. Accordingly a small committee was formed in October 1975, chaired by Professor Shoji Tanaka. Progress was slow, and it was not until well into 1976 that the specifics of the project were clarified.
Thus it was on 20 May 1976 that three senior scientists-Tarui, Komiya and Izuka-started the detailed planning of the now famous VLSI Project in a room in Kasumigaseki in central Tokyo, near the MITI headquarters. All three were scientists from the ElectroTechnical Laboratory (ETL) of MITI. The role of the ETL was changing, from serving the more simple needs of the manufacturers of electric power equipment in the 1960s to becoming an incubator of new information technologies in the 1970s. The laboratories of Nippon Telegraph and Telephone Public Corporation (NTI) were coming to play a similar role.
The VLSI Project started with two rows of desks on the 29th floor of a high-rise building in Kasumigaseki. The recruitment of the Project staff started in May and by July the same year there were already 20 members. Everyone was concerned that time must not be lost. Thus a decision was taken to quickly establish clean rooms, which are essential for all work on integrated circuits, and to order all the necessary equipment. It was also agreed that an electron beam machine (delineator), which was needed for etching the ultra-fine circuit lines, should not be built in one stage but should be preceded by a first trial. The first stage would last two years. Many ideas could be introduced and tried out. The best ones would be followed up during the second stage. In August 1976 the Project moved to the new location south-east of Tokyo and clean rooms were established. The tempo was not lost and new young staff were quickly recruited.
The significance of the industrial involvement can be understood by looking at the composition of the committees that performed the project analysis.l6 There were four groups altogether, covering the topics VLSI, micro-manufacturing, magnetic bubble elements and Josephson junctions. The first group was chaired by the then still relatively unknown Professor Shoji Tanaka. The other three groups were all chaired by scientists from the ElectroTechnical Laboratory (now in Tsukuba Science City). It is noteworthy that two of the groups, which were studying bubble memories and Josephson junctions, did not have any impact on the orientation of the VLSI Project. The inclusion of these topics is explained by the fact that IBM at that time paid considerable attention to these new technologies, the development efforts of which have since been drastically reduced.
In all four groups, with one exception, there were representatives from the five companies, which were later to form the VLSI Engineering Research Association. In each of the groups there was one representative from the Musashino Electrical Communications Laboratory of NTT. The limited role played by the universities and their electronics departments is obvious from the fact that apart from Professor Tanaka the only representation was from the Research Institute of Electrical Communication at Tohoku University in Sendai, considered to be one of the major institutions in the field of information technologies.
The VLSI joint laboratories
Professor Tarui initially thought that it would be desirable to bring all the researchers together in one location and thereby keep the facilities of the five companies as an important backup. This was a novel idea for both MITI bureaucrats and the participating companies. (The same approach has later been followed in two major national projects-The Optoelectronics Project and The Fifth Generation Computer Project.) There were many views as to where to locate the Institute and considerations were strongly influenced by the availability of buildings and time. Finally, a new empty building beside the NEC Central Research Laboratory in Kawasaki, south-west of Tokyo, was made available. Clean rooms were built and installed as early as the end of 1976.7 There is no doubt that renting premises, rather than building new ones, proved to be very useful as the project got underway very quickly. It was also possible to make use of common facilities such as air conditioning, gas, pure water, water treatment, canteen and library. In general there was very little problem in sharing such facilities with NEC.
In selecting the themes of the research project it soon became evident that it would only be possible to carry out research, within the VLSI Project or the Joint Laboratory, which was fundamental in nature and which was of great common interest. Thus it was necessary to find out the common interests of the participating companies in order to make co-operation possible. This would avoid the necessity of companies having to divulge their own proprietary secrets. Thus a basis for collaboration was found. Joint work focused on six different technological fields:
1. Fine processing technology
2. Crystal (silicon) technology
3. Synthesis technology-e.g. circuitry
4. Process technology
5. Testing technology
6. Device technology.
The first two topics were by definition both fundamental and common. The third one did not have these characteristics at all while the remaining three had minor elements, which were both fundamental and common. The still uncertain attitude that the companies had about working closely together is reflected in the laboratory set-up. Only one of the laboratories, the fourth one working on crystals, was actually a mixed one. All of the others were each dominated by members from one of the five participating companies. Furthermore, each company had also been asked to name a laboratory head for each of the laboratories with the remaining laboratory to be headed by a scientist from ETI. The choice of Professor Tarui from ETL as scientific leader and the naming of Mr Nebashi from MITI as overall director was done in order to guarantee neutrality in the operation of the joint laboratories as well as the research project.
Results of the joint research effort
In a short period of time, by the early 1980s, Japanese suppliers captured a major proportion of the market for 64K memory circuits. However, it is very doubtful that the results flowed from this particular project-if it is seen as limited to the research that was jointly financed by government and company funds. Certainly, there is little doubt that the coincidence of the VLSI Project and the Japanese market success for memory circuits may have served as an eye opener for almost everyone concerned in Europe and the USA about the future role of Japan in information technologies. As a consequence, when Japan a few years later launched projects such as the High Speed Computer and the Fifth Generation Computer there have been immediate policy reactions both in Europe and in the USA.
It is desirable but difficult to make an assessment of the extent to which the various research themes of the VLSI project have been successful. The scientific leader of the Project, Professor Tarui thinks that the results have not yet become fully evident. He argues that it is important to realize that the establishment of joint laboratories was an important achievement in itself. This has had the effect of making similar projects possible-basing themselves on the experience of the VLSI Project-such as the Optoelectronics Project and the Fifth Generation Computer Project. Looking for more specific results it is possible to mention the development of fine processing equipment and their subsequent production in Japan. Three technologies in particular were successful: the electron-beam pattern delineator, the optical projection system and silicon (crystal) technology. The development of these technologies may in fact be the biggest achievements of the VLSI project both for the supplier industries and for the major companies (confining our interest to the achievements of the joint laboratories).
Before the VLSI Project started in 1976 the chip manufacturers in Japan were dependent on US equipment suppliers. Thus in the past, Japan could only develop its industry along the lines of US industry and could only trail the development of US competitors. But the Japanese semiconductor industry realized that in order to develop a new field it is also necessary to develop and manufacture its production equipment domestically.
The direction of research remained basically unchanged throughout the four years. However, the ambitions changed as the level of integration to be attained was raised on several occasions. Originally, the technology to be developed was to be utilized for 256K DRAMs (dynamic random access memories) and SRAMs (static RAMs).18 There was a clear division of labour between the Joint Laboratories and the participating companies. Fundamental research was the responsibility of the joint laboratories while applications were carried in the companies. The Ministry of Finance originally checked the early planning reports in order to weed out what its officials would consider unnecessary duplication.
There is an important element of economizing in the joint development of VLSI technology. By dividing costs over five companies the actual costs became relatively low for the development of the electron beam pattern delineator as well as for other equipment. Furthermore, the instruments were developed in the same place that facilitated coordination and selection of the best parts and the best designs. In addition, the concentration of expertise from the five companies made it possible to identify design specifications much more accurately which was a great bonus to the equipment and silicon materials suppliers. There was also healthy competition among the staff. After completing the development and design the results could be brought to the different companies. This was probably done in a very efficient way and thus contributed to establishing a basis of fundamental VLSI technology in the Japanese companies.
There are comments to the effect that the project has only shown limited creativity. Professor Tarui wants to refute them and set the record straight by pointing out the following three factors. First, the project ran for only a limited period (only four years) and so major original results could not be expected. In fact, the project was fully staffed for only two years. Investigations that produce creative results should be done in universities and national institutions, where the researchers are not usually constrained by time, Tarui says. A project like the VLSI programme must be properly timed. Tarui says that in the VLSI Project “we tried to grow the buds which can be developed in a short period of time”. Most ideas had originated with the staff already before they joined the VLSI Project. This is true for the electron beam pattern machine for which Dr Goto of the Riken Laboratory had developed the idea.l9 Second, almost all fundamental ideas require five to ten years to develop and come to full fruition. Many of the ideas contained in the more than 400 patents yielded by the joint research laboratories will eventually be realized and have fundamental effects in the future, Tarui says. Third, the choice of research theme(s) was very significant for the future of the electronics industry in Japan. The impact of VLSI on the modern electronics has been, and continues to be, profound. This was not quite so obvious in 1975, but all the five participating firms came to realize its importance and made strenuous efforts to work together. Achieving consensus on the importance of the research themes should also be seen as part of the success story because very few people initially realized how important VLSI technology was going to become.
In the laboratories Professor Tarui persuaded people that certain things were of common interest to all companies. The area of joint activities was characterized by this common interest and fundamental technology (research). Thus the companies were not expected to share the company specific know-how and could instead concentrate on themes of common interest. The development of design concepts for specific integrated circuits eventually proved to have little common ground and was consequently difficult to organize within the joint laboratories. Mathematical research of relevance for such practical applications was of course possible. In the end, circuit design development was not included and Tarui also confirms that each company established its own pilot production line. No other joint research projects, under MITI sponsorship, were carried out in connection with the VLSI Project. This meant that the other “participating” companies such as Canon and Nikon, were only paid for delivering equipment and services and did not share the research results, nor did they participate in discussions organized within the VLSI Project. This was a conscious policy desired by this ERA and other ERAs as well in order that non-participants should not be able to obtain “unjustified gains”.20
A common source of surprise in the organization of the VLSI project is that only five companies were involved as core participants. Tarui and others, both inside and outside MITI, have pointed out that there were only five companies making general– purpose computers when the project started-although in the case of Toshiba the company had already suspended production of computers. The VLSI project was conceived out of necessity to match IBM’s next generation computer. Since then VLSI chips have become general-purpose commodity items for the electronics industry and many related industries. At the time computers demanded high reliability and high performance-in contrast to consumer electronics, which could do with less sophisticated integrated circuits. The central purpose of the VLSI Project was to support the development of mainframe computers in Japan. This was part of the country’s national policy-in a context in which computers were considered to be important for national security, as broadly defined. Consumer electronics, at least in the mid-1970s, did not warrant similar attention. Tarui stresses that the situation has been somewhat similar in the USA where the development of ICs has been supported by the government because of their importance for space and military applications– initially obtaining high prices as special devices. The development of common technology, achieved through the VLSI Project, was fairly quickly diffused within the industrial structure in Japan. Several companies like Oki, Matsushita and Sharp have been able to catch up. Furthermore, companies which were excluded from the VLSI Project have later on joined other MITI-sponsored projects/programmes such as Future Generation Basic Technologies (Jisedat) and the Fifth Generation Computer Project.
It is worth noting that NTT had already started its own VLSI Project one year later with the co-operation of three of its major suppliers of telecommunications equipment-NEC, Fujitsu and Hitachi. In many ways this was a parallel VLSI project with a focus on developing technology of relevance for circuits to be used in applications for the telecommunications sector. A major difference between the two projects was that the objective for NTT’s efforts was to develop integrated circuits that would be useful for the continued digitalization of the telecommunication networks in Japan. The MM-sponsored project had in a sense a much broader, less application oriented objective, and clearly aimed at building up an advanced industrial competence within the major companies-bearing in mind the overall aim of supporting the Japanese computer industry.
In comparing co-operation among the companies in the VLSI Project with that of the NTT’s VLSI efforts it is possible to identify two distinct characteristics. First, although it was not yet in electron beam technology, NT was able to provide greater technical support. In the VLSI Project it was possible to obtain technical specifications, as all the participants were active in VLSI manufacture although not making the instruments. In light of the existence of two “competing” projects it would seem natural that some sort of exchange of information would have been established. This was not really the case although formal relations between VLSI Project and NTT laboratories were initiated in 1976 or 1977 and from then on monthly meetings were held. However, informally very close contacts were maintained between engineers of the Joint Laboratories and NTT laboratories-not least in the area of crystal technology. This fact clearly indicates the homogeneous nature of the industrial structure in Japan and its strong inter-linking networks.
There were initially several ideas-hopes or expectations-that the VLSI Project should be followed by a second stage. Already at the start many scientists and company researchers had their eyes set on considerably higher levels of integration and realized that new technologies would have to be developed if these goals were to be achieved. But a projected continuation of joint VLSI research ran into problems for two very different reasons. First, the success of the VLSI manufacturers in Japan created serious concern in the USA, and the US trade representatives at various levels repeatedly criticized the Japanese government for unfair trade practices-by criticizing the support given to private companies under the VLSI Project. Second, the technological expectation that drove the VLSI project, namely that electron-beam technology would become necessary in the 1980s as levels of circuit integration increased, was not in fact borne out. As it happened, optical lithography continued to be refined, so that finer line widths could be achieved with optical steppers and aligners. These developments postponed the necessity for a radical breakthrough in process technology, and thus diminished the need for a continuation of the VLSI programme.
THE VLSI PROJECT-ORGANIZATIONAL EVALUATION
There were some critical features of the VLSI Project which help to account for its success. Firstly, the timing was good. Initially there were great doubts about how the computer companies would fare in the 1980s. So, presidents of the companies were concerned about technology for computers and made great efforts to support the VLSI Project-once they had clarified their own thinking for long-term development. Second, the preparations made, before the VLSI Project actually started, were very good. The MITI bureaucracy had developed methods which were sound-including operation, research plan and the arrangements for the staff from the participating member companies. In addition the company staff joining the Project worked enthusiastically which naturally was essential. In sum, there was good co-operation and good staff at the joint research laboratories. Third, the objectives of the Project were clearly spelled out, as was its duration. Thus the plan that had been agreed to by all parties could be executed effectively. Fourth, the actual operational plan was quite flexible, enabling the research teams to devise new ways of tackling problems as they were identified.
A major benefit of the VLSI Project is the continued attention that the participating companies has given to providing financial and other necessary resources for further development of VLSI technology. In this sense the VLSI project has had a strong catalytic effect on the major Japanese companies in this technological field. The VLSI project made companies and almost everyone concerned realize that R&D in VLSI technology required much money and that fundamental research was needed in order to stay competitive. Such benefits may in particular accrue to the heads of companies who went through a learning process-of considerable significance.
A key factor in the success of the project was the initial identification and promotion of a technological field of future relevance to Japan. This is actually a key element of the Japanese success. In the mid-1970s Japan had with great effort established a relatively viable mainframe computer industry, which in the Japanese market was holding its own against IBM. No other country has been able to devise national policies to achieve that aim-although many have tried. The increasing level of integration of integrated circuits and the need to have reliable devices for computers clearly indicated to the Japanese manufacturers that they needed to have access to advanced VLSI technology. The rumour that IBM was on its way to launch a one-chip-computer became a rallying force, inviting references to the earlier experience of 1853 when Commodore Perry’s “black ships” forced Japan to completely change its relations with the outside world.
There is little doubt that timing was important and that the amount of money was substantial. But similar conditions have existed for several similar attempts in Europe, which have had nothing like the impact of the Japanese project. The rallying cry to fight against IBM made it possible to create “VLSI fever” in Japan, which made almost everyone concerned aware that action was necessary, if Japanese companies were to survive in advanced electronics. This had three important effects. First, it created a necessary enlightened atmosphere among company researchers, scientists in universities and government laboratories and government bureaucrats who then had common ground for discussion. Second, once the decision was taken to carry out certain planned activities these had the backing of almost everyone concerned. Third, in this atmosphere it was possible to find the organizational solutions, which would be most efficient for attaining the objectives. Finally, the most important effect, although indirect, was the decision by somewhat reluctant top executives and heads of research departments that their companies should move full steam ahead into the new era of VLSI technology.
Today there exists a consensus that the VLSI Project in Japan, established in 1976 as an engineering research association, had an exceptional success in promoting technological development. The focus of the project was clearly on micro-fabrication technology, which in essence meant improving and developing new lithography methods. Silicon crystal quality and an improved understanding of physical and mechanical properties were a second focus. The project resulted in firmly raising the level of VLSI manufacturing technology of the five participating companies-Toshiba, Hitachi, NEC, Fujitsu and Mitsubishi Electric. This is a contributing factor in these companies winning an expanded share of the world market for memory circuits. However, the VLSI Project also had a profound effect on companies outside the group of five participants. For example, Matsushita has become a major actor in the field, as have Sharp, Seiko-Epson and others.
Furthermore, the project raised the technical level of two major groups of supporting companies. First, the major silicon vendors, although not formally participating in the project, were drawn into a close interaction with the project which rapidly increased understanding about how silicon crystals could be considerably improved-at a time when the global as well as domestic demand was rapidly increasing. Second, a substantial number of equipment manufacturers, companies attached to the five participants as well as independent companies, in a similar fashion considerably improved their understanding as to what types of equipment would be technically and economically most suitable for the VLSI circuits to be manufactured in a few years time. So not only have Japanese VLSI companies established themselves among the world leaders but its crystal and VLSI equipment manufacturers have also established themselves in the top league.
There can be no doubt that the high level of funding was a very significant factor as it meant that the VLSI research funds in 1976 and 1977 increased through this project by approximately 20 billion yen. Thus, the VLSI Project actually contributed research funds of the order of 3-4 per cent of the semiconductor sales of Japanese companies during the early years. However, Sakakibara, a Japanese management researcher, mentions that it has been estimated that between a quarter and a third of the total project funds were spent in the USA to purchase the most sophisticated equipment available at the time.21 Thus procurement of sophisticated equipment was not in the first stage seen as a policy instrument for promoting the Japanese equipment fabricating industry. Sakakibara mentions a number of other favourable conditions that were extrinsic to the Project. First, everyone concerned knew that the Project was to counter the possible threat of IBM developing a super-chip which-if this were true and if it were successful-would probably erode the basis of the computer industry in Japan. This awareness was, in the words of Sakakibara, “favorable for the integration, motivation, and concentration of research efforts of many members”-in spite of the fact that this assertion was found to be incorrect already in the early planning phase. A second favourable condition was that the five participating electrical machinery companies had already participated in joint projects under MITI guidance and had accumulated a considerable experience on how to handle matters like personnel, budget, and filing of patents. Third, there existed in Japan a substantial scientific basis for high quality development work on crystal technology and microscopy, which made it possible to make a leap forward.
The idea of sharing a joint laboratory for a fixed period of time did not come naturally to the participants in the VLSI Project. However, NEC and Toshiba had already operated a joint laboratory or rather joint research, in the form of the NEC-TOSHIBA Information System (NTIS), initiated in 1972. Not to be outdone, Fujitsu, Mitsubishi Electric and Hitachi operated a third joint research organization, the CDL. In fact the VLSI ERA was formally made up of seven participants, where two of these were the NTIS and CDL joint research organizations. However, there was no precedent of a research association having a joint laboratory. Furthermore, having research members in the joint laboratory coming from competing companies was a completely novel idea. This naturally posed a number of problems, which have already been touched upon.
In view of the existence of a technological threshold to be climbed, the project may also have been seen to need larger group-oriented research because of the complexity and interdisciplinary character of many of the problems encountered in VLSI manufacture. This in turn favoured the idea of establishing a joint laboratory, which may have been a major instrument in integrating the various development capabilities of the participating companies.
More recently the idea of joint laboratories has been incorporated in several MM-sponsored projects aiming at the development of information technologies. This is the case for the optoelectronics engineering research association started in 1981, which with its six laboratories and almost exactly the same membership as in the VLSI group appears to be a true copy. A very similar approach has been followed for the Fifth Generation Computer Project initiated in the same year. The same approach can be identified in the R&D Institute of Metals and Composites for Future Industries, although it is not formally organized as an engineering research association. All these institutes have been established for a limited period, which is not likely to be extended, with the implicit expectation that they will repeat the success of the VLSI Project.
Sakakibara makes the point that “it was neither MITI nor the five private companies, but the ruling Liberal Democratic Party (LDP), that initiated the idea of setting up a co-operative laboratory in 1975”.22 This was prompted by the fear that the computer manufacturers in Japan would not be able to stand up to IBM. Apparently, the manufacturers were first opposed to such an idea and were much more in favour of government policies or subsidies that would allow them to act by themselves. However, the political interest in supporting the computer industry dates from a much earlier period. Similarly to the situation in the car industry, the attempts to restructure the computer sector had not been very successful. The new approach tried in 1975 was to influence and support the computer industry more indirectly– through technology support programmes. Thus, under political pressure according to Sakakibara, the major companies were forced to accept the two joint laboratories already described which also formed the basis for the much wider project-the VLSI ERA-with its joint laboratory.
Sakakibara further argues that the VLSI Project benefited greatly from factors which were intrinsic to it; in particular, he stresses communication and institutionalization. First, the joint laboratories, including the planning section, had their leaders drawn from each of the five companies and the ElectroTechnical Laboratory (ETL), which is directly controlled by MITI. All the six laboratory teams included members from several companies. It had been indicated to the companies which members it would be desirable to have in the Project but it was naturally left to the companies to make the final decision. Then it was left to the teams so selected and the overall research director, Professor Tarui from the ElectroTechnical Laboratory, to work out the research themes with the necessary details. This was a very lengthy process. However, the considerable time spent in discussions may have been well spent as it narrowed down an otherwise possibly unwieldy project and achieved a consensus among the researchers. This is likely to have formed a very strong basis for the continuing interchange of ideas by the researchers throughout the project period.
THE ENGINEERING RESEARCH ASSOCIATION AS POLICY INSTRUMENT
An attempt will now be made to provide an overall assessment of the efficiency of the engineering research association as a technology policy instrument. There is little doubt that many industrial technologies are becoming increasingly science based and the industrial companies in principle have three options open to them. First, they can develop the scientific capability inside their organizations. Second, they can pool their resources together with other companies. Third, they can draw on general scientific capability, which exists within the university system and national research institutes. Finally, the three approaches can be combined together in various ways. There is no doubt that Japanese companies have in many areas scaled the technological peaks and no longer require government support in order to catch up with their counterparts in other countries.
The first approach may not be possible except for very large companies. The second approach may have severe limitations when it comes to companies that are competing in the same markets. If it is possible to draw a clear demarcation line between basic research and the applied research and development, to which a company will normally allocate a majority of its resources, then such co-operative arrangements are possible. But if the basic research is seen as having very close links with applied research in areas where the companies compete then it will be difficult to find a basis for co-operative research. This has also been clearly demonstrated in many of the ERAs not least in the Joint Laboratories of the VLSI project where it was only possible to join forces on research themes which were “fully basic” and of common interest to the participants.
Such restrictions, from the economic perspective of the companies, may be modified under several circumstances. First, companies within a sector may co-operate on research, which has a direct relevance to the commercialization of products if this is done within the same sector to gain an advantage over competing sectors, at home or abroad, which was partly the case in the VLSI Project. Second, if there are substantial economies of scale in carrying out certain categories of research, a basis for co-operation may similarly exist. Finally, in market-sharing arrangements there also exist conditions for co-operative research, modifying the restrictions, which were mentioned earlier. Thus it is possible to state that the point beyond which an individual firm would not be prepared to participate in a joint research programme depends on the intensity of competition and the level of indivisibility in R&D. The decision point will naturally vary from industry to industry and from project to project.
It can be argued that the large number of ERAs established in Japan have not been true co-operative research projects in the sense of sharing the research results. Many of the ERAs should rather be seen as research consortia where the research tasks and the allocation of government grants among companies were clearly decided upon from the beginning by the participating companies. Thus, there was no expectation that new research results would be freely transferred between all participants-although patented results would become government property if a project were mainly financed by government funds. It is only in exceptional cases that an ERA has established a joint laboratory where very special conditions have existed for sharing research results.
The ERAs were launched with two major objectives, namely to obtain industrially relevant research results, and to trigger or stimulate industrial companies to initiate or accelerate industrial research. In the early period of using ERAs, in the late 1950s and early 1960s, the first motive was wholly dominant and it has remained so for certain categories of ERAs until today. In the second ERA period, since the early 1970s, the second motive has become increasingly important and this has also prompted a re-orientation of ERAs in promoting joint research. Thus the following three factors are important to consider.
Form of organization. A major aspect is the number of participants, which must be kept rather low for ERAs, which are involved, in advanced research. Otherwise it is difficult to maintain a sufficiently strong leadership and direction of research. Second, the members can be also selected to include users or suppliers. This is important as it opens up the possibility of establishing efficient linkages. The research association can include various types of joint facilities such as joint laboratories-exemplified by the VLSI and the Optoelectronics Projects-or joint design centres when the careful coordination of a number of projects are necessary. The jet engine and flexible manufacturing system exemplify the latter situation.
Level of funding. The government funds can cover more or less total costs, which has been the case in all national projects sponsored by AIST. In many other ERAs the government and the members of the ERA share costs which has basically been the principle for the ERAs and is also the principle followed for collective research in most countries in Western Europe.
Timing and direction of research. This may in fact be the most important variable in initiating research. Naturally it depends on a keen understanding of the situation both at home and abroad-so “intelligence operations” are critical. Given the above mentioned variables it may be asked under what conditions does joint research come into existence and under what conditions is it viable. First, it should be noted that joint research is found in both industrially mature sectors and in rapidly developing sectors. The prevalence of the second category has in the past been particular for Japan but now occurs in many other industrialized countries. Second, the modern industrial enterprise accepts joint research when the benefits of (government or outside) coordination (and subsidy) clearly reduce risks and costs for subsequent product development. Third, the advantages of internalizing the research stage jointly with other companies does not arise until the risk of unforeseen technological events and the magnitude of required resources pose problems which are more than offset by the risks associated with carrying out research jointly with competitors.
Joint research in rapidly developing technologies will mainly be organized in big projects and is likely to favour large enterprises with a good foundation in R&D with the consequence that they increasingly dominate the industrial scene. Finally, it should be noted that projects, which are completely financed by the government, might not truly be joint research-unless members also make contributions of scarce research staff or exchange information with one another. Thus, it turns out that several ERAs are not organizations for joint research but only an efficient way for the financial arrangements of subsidies.
The successes and failures of research associations in Japan The engineering research associations underwent a distinct change in Japan as a policy instrument when they were reintroduced in the early 1970s after a lapse of more than five years when no new ERAs were established (see Table 2). The new phase saw a new focus on more advanced research and much wider product technologies, the average number of companies per ERA became smaller and the duration of ERAs shorter. In more recent years an element of basic research has become very prominent in several ERAs although others, aiming at the revitalization of basic industries, are still run on more traditional lines.
It may be of interest to attempt an assessment of the relative success of the new style engineering research associations, which have been established since 1971. Many of these ERAs are still in operation and it may be premature to try to assess the relative success or failure of projects, which have not yet been completed. The ERAs established in the period 1971-78 number altogether 29. If the first two in that period-geared to environmental aspects in the testing and development of car components-are excluded, we are left with 27 ERAs. In an attempt to consolidate personal judgements from various sources on the performance of these associations, it has been possible to collect information for 18. Ten of these were generally considered as failures while the remaining eight were classified as successes.
Basing myself on such judgements it has been possible tentatively to identify certain characteristics, which contribute to the success of an engineering research association. Five such characteristics are as follows. First, it is important to have a clearly defined objective. Second, it is useful to have an association made up of a fairly small number of companies. These points are borne out in projects 32 and 39-considered to be only moderately successful. (Information on the various ERAs is given in the Appendix.) These projects had a large number of member companies and their R&D activities have been described by one observer as “committee research”. Third, it is important that the members do not have any strong competing interests or that such conflicts of interest are overshadowed by a perceived common threat. ERAs 37 and 38 did not quite meet these demands. Fourth, it is important that the external economic and technical conditions do not change so drastically that the basis of project is undermined. This happened in ERAs 31 and 41 as the expectations of continued high oil prices did not materialize. In other words, the period of the project must be commensurate with the period for which it is feasible to predict the technological and economic conditions. Finally a few remarks about the remaining projects are in order. The software modules associations probably suffered from a lack of orientation. The traffic control system was based on the assumption that the cars would include expensive electronic devices and that the city authorities were willing and in the position to rebuild the streets and signals for the system.23
When all the ERAs have been evaluated and the assessment data examined more carefully, the picture presented above may have to be modified. When looking at the seven Large-Scale Projects (LSP) which were established during the period, three fall in the category of successes while four are failures. Now, it must be remembered that every research project cannot be successful and that several of the energy projects in Japan and in other energy-starved countries should be seen as part of a National Insurance scheme in case the global energy situation had developed in a different way. However, in general terms many insightful people both within the government agencies and in private companies have confirmed the overall assessment. A somewhat more positive evaluation is given by government officers.
It appears, when looking into the establishment of ERAs, that they have in most sectors been introduced on an ad hoc basis without a clear frame of reference-with the clear exception of the ERAs established to promote information technologies. Finally, it should also be considered that several of the more recent ERAs have a strong element of basic research, which naturally modifies some of the requirements listed above. A high-ranking official in MITI had the following comments to make about the efficiency of ERA as a technology policy instrument. He said that the engineering research association is efficient if it carries out the day-to-day operations at a distance from MITI, is geared to operational activities and is focused on system development. In terms of management it is important to have one main contractor which should ideally be one of the participating companies even if the association is legally the formal contractor. He mentioned that one or two in ten of the ERAs might turn out to be failures, which is usually due to changes in the economic and technological environment in which new products and systems are to be used. He agreed that examples of failures might include the olefin project, nuclear energy for steel making and the desalination project. However, the export of such systems, in the latter case-although not intended at the time-may revise the judgement. So far only one association (deep sea oil drilling) has been suspended. Another one (the electric car) has not led to a product which is being used.
The same official pointed out that the engineering research association could only be used for the development phase, after which it must be dissolved. Consequently, the utilization of results is not always secured. Furthermore, it is important to put MITI’s financial allocations for industrial technology development in a proper perspective. The Science and Technology Agency (STA) allocates approximately 200 billion yen to nuclear technology and another 100 billion yen to space research. MITI altogether uses 60 billion yen for industrial technology development of which 20 billion yen goes to a coal gasification project. So the annual funds going to joint research may only be in the order of 20 billion yen.
It is the linkage effects that are possibly the most significant benefits of ERAs. Again the VLSI project provides the outstanding example. While its membership was limited to five major companies-NEC, Toshiba, Hitachi, Mitsubishi Electric and Fujitsu-and no doubt they benefited from the project, the linkage effects on other actors of the industrial system may in fact be the most important result of the VLSI Project. First of all that there existed two distinct technology-industry systems to promote the development of integrated circuits-one led by NTT and the other by MITI-we will focus on the latter one. The companies, through the association, established the joint laboratories of which the three for lithography technology were the most important. These three laboratories worked in two different directions-back towards the companies which provided the majority of researchers in each of the laboratories but also towards the equipment suppliers which meant in particular Canon and Nikon which were making what could be termed front-end equipment. The back-end equipment manufacturers-e.g. Shinkawa for bonding and Disco for slicing were already well established. Thus specifications based on the experience of the five companies and also on research in the joint laboratories were channelled to Nikon and Canon together with development contracts for specific machines. These contracts and linkages in fact triggered an almost completely new development at Nikon-wafer-stepper printers-and served as a stimulus for Canon for its proximity aligner. A further effect of the development contracts given to the optical companies was that they in their turn gave a substantial development contract to Ushio Denki to develop the very special lamps needed in the IC printers. The Ushio company, which already had a strong base in plain copier lamps where it maintains a healthy 70 per cent share of the world market, has rapidly captured 40 per cent of the global market for lamps used in IC-printing equipment. Canon is today the leading producer of proximity printers and Nikon is well on its way to establish a similar position for wafer steppers.
Similar linkages were also created between the crystal laboratory and the silicon crystal vendors with a frequent exchange of information. However, the producers of silicon wafers were at the time already fairly well established which is also true for the major maker of photo-resist and other chemicals needed in the manufacturing process. The printing companies, DaiNippon Printing and Toppan Printing, also benefited from close contacts with the joint laboratories, which in a way served as a clearinghouse for the various groups within the IC manufacture industrial system. It may also be mentioned as an indication of the strength of the material suppliers that the makers of ceramic packages for ICs, already strong at the time, today control 90 per cent of the markets for such packages.
Testing was not considered as a functional area of research within the VLSI Project sponsored by MITI. However, NTT paid close attention to the requirements of VLSI testing. In a complementary way NTT supported the very ambitious project at Takeda Riken to develop a high-speed tester.24 There are many other cases of joint research between NTT and electronics companies but it has not been my intention to discuss them in this context.
Thus it is possible to see a number of ripple effects flowing from the VLSI Project. First, the material suppliers who were already fairly well established were further strengthened. Second, the equipment makers who were already strong in back-end equipment established an almost equally strong position in front-end equipment– exemplified by IC printers. Third, the sub-suppliers of which there are likely to be more examples than Ushio Denki climbed into a new very strong position. Fourth, there are today several indications that the most important long-term effect on the IC industrial system is the emergence of industrial groups around Toshiba and Hitachi and possibly also NEC which aim at delivering complete sets of process equipment for making very large-scale integrated circuits. So, Japan may in the future have two or even three of maybe less than ten companies worldwide which will have this capability. Such an effect could hardly have been foreseen as a specific objective at the time of planning the VLSI Project, but seems to have its roots in the project in any case.
New policy instruments
MITI in its vision for the 1980s argued that Japan should increase its R&D spending from slightly more than 2 per cent (the level at the time) to 3 per cent of GNP and that the government should shoulder a larger share of the total. It was at the time suggested that the government ought to have a share of 40 per cent against the “paltry” 25 per cent at the time. One underlying reason was that the change in the funding pattern would lessen the criticism that Japan is enjoying a “free ride” by utilizing scientific and technological research results from abroad while giving little in exchange. The concern at the time was also centred on the fact that Japan had completed its catch-up phase of technological development and would have to start generating the results of basic research at home instead of scouting around abroad or waiting for results to appear abroad. Obviously this created a new situation.
There are two major elements that are now forcing the policy planners in MITI and in other agencies too to reconsider their technology support programmes. First, the huge trade imbalance with the US but also with most countries in Western Europe is, outside Japan, often taken as an indication that the Japanese are set on conquering the world through trade, and that technology support at home provides an important underpinning for that “trade war”. Second, the companies-the earlier “disciples” of MITI and NTT-have now grown up and are well staffed and often have very big and modern research laboratories. So the government agencies are, when formulating technology support programmes, constrained by the international situation and have to look in new directions if they are going to make any contributions that are useful for the big private companies. During the last few years a number of major companies have established new research institutes which are geared to basic research, serving the long-term interests of the companies. Such research institutes typically have a few hundred researchers and investments in the order of some ten billion yen.
The new situation has also prompted MITI to consider the industrial scene more broadly and focus on research areas that are normally outside the Ministry’s sector of influence. For example, they include various aspects of biotechnology, which would normally fall under the jurisdiction of the Ministry of Agriculture or the Ministry of Health. MITI has not previously competed with these ministries although it has competed for some time with the Ministry of Education and Ministry of Post and Telecommunications. In the struggle for formulating and implementing new policies and achieving influence, there are two major organizational elements to be considered. First, it is probably desirable to use an organizational concept, which will make it easier to promote collaboration between private companies and research departments, be they in government laboratories or in universities. This will require new laws but also a very different attitude from all agencies concerned, in particular the Ministry of Education and by the faculty in the university departments concerned. Second, long-term planning will have to be organized in other ministries similar to that within MITI which has its Agency for Industrial Science and Technology (AIST).
The juridical basis for much of MITI’s successful support programmes in information technologies lies in the provisional law on the promotion of specific machinery and information industries which expired in June 1985. This law which was only the last one in a succession of such laws, initiated in 1956, has apparently fulfilled its useful role. A new more general law, which will extend to private firms tax incentives, and financial support, which will make it easier for the companies to embark on costly and risky research, will now take its place. The technologies to be eligible under the new scheme are new industrial materials, microelectronics and biotechnology, as reported in mid-1984. The legal framework for the future high technology industries in Japan also includes special provisions for small companies in high-tech industries, which will be given access to tax-free loans. An additional element of the new MITI approach is that AIST will change its rules so that joint government-industry research, and not only contract work, can take place inside government laboratories. Thus the conditions which encouraged the formation of ERAs are changing, as Japan matures and as external circumstances shift.
CONCLUDING REMARKS: THE “ULTRASTRUCTURE” THAT UNDERPINS JAPANESE TECHNOLOGICAL LEARNING
Japan has continued to build up its infrastructure for research and development, most of which has in fact increasingly come to be funded by the private sector. In similar situations in other mature industrialized countries, the government and its agencies have limited possibilities directly to influence the use of resources within the infrastructure, once it has been established. However, it appears that the Japanese case illustrates the efficiency by which changes can be affected if ability exists to establish and dismantle R&D ultrastructures.
An ultrastructure is a network for communication and collaborative relations, which facilitates transactions and exchange of information.25 It links and integrates various competencies, which are diffused through the infrastructure without visibly changing or extending the latter. Its existence and efficient functioning usually depends on external factors which act as catalysts. Its main function is to identify and handle major strategic R&D issues by an optimal use of competence contained in the infrastructure. The ultrastructures are often ad hoc organizations although some of them may eventually be made permanent. The degree to which they are formalized may vary considerably. Many of the ERAs and other similar organizations established in Japan should in fact be seen as ultrastructures. Although the level of funding cannot be neglected, the efficiency of many projects owes much more to new or expanded relations for communication and collaboration, which continue to exist long after the projects have been terminated.
Contemporary works, such as Japan’s High Technology Industries, edited by Hugh Patrick, and published in 1987, contained references to the VLSI project, but not in-depth case studies. An earlier account of the project appeared in the well-known 1984 text edited by Daniel Okimoto, Takuo Sugano and Franklin Weinstein, Competitive Edge: The Semiconductor Industry in the US and Japan, while an account in English by the Japanese scholar, Kiyonori Sakakibara, “From imitation to innovation: The VLSI Semiconductor project” appeared in 1983 as a Working Paper published by the MIT Sloan School of Management.
2 Based on a study, which was carried out in the late 19(Os, P.S. Johnson argues that although the RAs are only a small part of the country’s overall effort, they are often very important in individual industries especially those, which are traditionally based’. See Johnson, P.S. 1971/72: The role of cooperative research in British industry, Research Policy, No. 1: 332-350.
OECD 1965: The Structure of Industrial Research Associations. Paris. Dr Sugimoto had originally been responsible for an Aerodynamic Laboratory. He remained as director of MEL until 1961 when he was named director of Science and Technology Agency (STA) for one year-an assignment which eventually lasted more than two years. In 1964 Sugimoto again joined Hitachi, his original employer before World War II, as assistant director for a research institute. He was then given the responsibility to establish the Hitachi Mechanical Research Laboratory, was appointed its director and remained for two years in this position. Later on he was transferred to the Headquarters where he remained for five years.
Many senior people had been purged by the Allied Command in Japan. As a consequence, Dr Sugimoto mentions, he was very young to be a director in the Mechanical Engineering Laboratory, having been appointed in August 1953. As a consequence it was not necessary to carry out “nemawashi”-rootbinding-the traditional Japanese way of forming consensus, because of the young age of the top decision-makers. The name of the law is ‘Kokogyo Gijutsu Kenkyu Kumiai Ho” in Kokogyo Gijutsu Kenrkyu Kumiai Ho’ in Kokogyo Gijutsu Kenkyu Kumiai Seido no Kaisetsu. Tokyo: MITI, Agency for Industrial Science and Technology, August 1981, p. 232. There are some discrepancies about the timing and order of establishment for the engineering research associations established before the law was enacted.
8 The central facilities were closed down in 1975 and together with the staff, were handed over to Japan Vehicle Inspection Association (Nibon Sboriya Kenza Kyokai) which now uses the facilities.
9 The name of the ERA was Kogaku Kogyo Gijutsu Kenkyu Kumiai. This ERA was established in June 1962 and terminated in January 1981.
10 The computer used was Okitac, manufactured by Oki Electric Co.
I 1 The budget in the first year was 30 million yen of which MITI provided 6 million. Altogether the contributions from MITI have been in the region of 150-180 million yen. The total budget in the final year was 100 million yen.
12 Source: “On the Situation of Engineering Research Associations in Mining and Industry”, Industrial Council, Ministry of International Trade and Industry, Tokyo, April 1982 (in Japanese).
13 Mr Shimada, presently councillor in the Minister’s Secretariat of The Environment Agency is the person to whom it has been attributed the achievement of having rediscovered the ERA as a policy instrument to be used for the National Large-Scale Projects.
14 The Large-Scale Programme began in 1966 with three projects-development of a super-high performance electronic computer, research into desulphurization technology and into Magnetohydrodynamic (MHD) power generation-and a budget of 1 billion yen.
15 The concept of LSP was introduced in 1966; no fewer than nine were established in the first six years. Source: Gif utsa kaihatsu sokushin no joken choga waga kuni nf okeru san gaku kanrenkei no genjo oyobi mondai ten (An investigation on the conditions for the promotion of technical development-Report-The current state in the collaboration between industry, research and the government, and the points at issue), Japan Efficiency Society Coordinated Research Centre, Tokyo, 1985.
16 Source: Sogo Keikaku Sogyo Gurupu Chukan Hokoku (Interim Report of the Operational Group on Comprehensive Planning), The Japan Electronics Industry Development Association (JEIDA), Tokyo, 1975. This document provides a detailed analysis of the technical problems, which would have to be overcome in order to handle large-scale integrated circuits. Subsequently the report became the basis for the decision to launch the VLSI Project.
17 The top of the building had the NEC logo as an advertisement, which was rejected by the staff from all other companies. Naturally, such feelings could be expected-in view of the strong attachment that employees in Japan feel towards their own organization-and as a consequence it was decided that the joint laboratories should have a separate entrance from NEC. Tarui mentions as a curiosity that there was a rumour that an executive from NEC had not been able to enter because he lacked the necessary identification card.
18 An interviewee notes that the VLSI Project ran into “trouble” when one of the non-participating VLSI manufacturers, Matsushita, announced that it had already developed a 64 K bit static RAM chip which in terms of integration was basically identical to the target of VLSI Project. It was then suggested by MITI that the project should be able to do still better and the integration target was subsequently revised upwards to I Mbit.
19 Riken is a major research Institute in Japan. Riken is short for Rikagaku Kenkyusho: the Institute of Physical and Chemical Research. It was originally established in 1917, at Komagome, Tokyo, with the financial support of the Imperial Household, the Government and many private organizations. It was in fact Japan’s first comprehensive research organization for studies related to chemistry and physics in general.
20 The impartiality may in fact have meant that NEC suffered although the joint laboratories were located on their premises. At least there is no evidence that the company received any special benefits from renting its premises to the VLSI Project. A special entrance and dining room were established in order that the geographical proximity should not favour NEC. Similar arrangements have later been made for the Optoelectronics Joint Laboratory which is now located on Fujitsu premises in Kawasaki.
21 Sakakibara, Kiyonori 1983: From imitation to innovation: The Very Large Scale Integrated (VLSI) semiconductor project in Japan, WP #1490-83. Cambridge, Massachusetts: Sloan School of Management, MIT, 32 pp.
22 In the interviews carried out by this author there is little support for this claim. However, it appears that key LDP personalities argued that co-operation between major companies involved in the manufacture of mainframe computers was necessary.
23 These requirements were not considered when planning the project, which furthermore was based on the use of a centralized computer power and not distributed computers.
24 Takeda Riken changed its name to Advantest, and in 1998 it was one of the world’s leading manufacturers of testing equipment. (Note added in 1998.)
25 Granberg, Anders and Stankiewicz, Rikard 1985: Langsiktig teknisk forskning och nationella FoU-system (Long-range industrial research and national systems for R&D). Lund: RPI.
Copyright Journal of Industry Studies Dec 1998
Provided by ProQuest Information and Learning Company. All rights Reserved