Pathways of Technological Diffusion in Japan

American policymakers, during the Cold War, treated technology and economic strength as means to achieve military and political ends. The Japanese, in contrast, never subordinated economic interests to defense objectives and, indeed, rejected arguments to that effect as naive. Instead, the Japanese saw technology and territory alike as vital national interests that could and had to be defended. Two different ways of approaching civilian and military technology development thus grew from the divergent ideas that the United States and Japan had about national interests. For years, the Cold War obscured the consequences of that divergence and made it possible for Americans to believe that they had found the single right way of defining and organizing to meet their defense needs.

In the United States during the Cold War, private research and development grew considerably smaller relative to public R&D. In Japan, the opposite was true. Military R&D became the single largest source of R&D spending in the United States, whereas, in Japan, it virtually disappeared. Since both military and trade relationships depend increasingly on technology, and since the end of the Cold War is stimulating considerable readjustment, the relationship between civilian and military activities within each of our economies has become central to the U.S.-Japan relationship. Americans have lessons to learn from the way the Japanese have subordinated defense production yet have emerged as one of the most technologically sophisticated nations in the world. The Japanese may have demonstrated, like the Venetians and the Dutch before them, that butter is as likely as guns to make a nation strong and, further, that nations cannot be strong without advanced technology.

Undergirding each nation’s policy was a coherent, consistent view of technology as the source of national security. I call this view “technonationalism” and differentiate it from what often is excoriated as “technoprotectionism.” The embrace of technology for national security in Japanese practice was a very positive force. Japanese technonationalism never isolated defense production from the general commercial economy. To the contrary, industrial, technological, and national security concerns have long been fused in the institutions of the Japanese economy. Japanese technonationalism is derived directly from ideas about adversaries; the “imaginary wars” of Japanese leaders are interpreted in technological and economic, as well as military, terms. In the Japanese case, adherence to three tenets both precluded the pathology of spin-away (in which the military and commercial economies move so far apart that technologies are not shared) and stimulated the positive benefits of spin-on (in which technologies developed in the commercial sector are picked up by the military sector) as part of the more comprehensive interdiffusion of technology. The three core elements of Japanese technonationalism are autonomy, nurturance, and diffusion. In this paper, I focus on the third element.1

Diffusion — From Technology Highways to Technology Jetways

The Japanese have achieved a remarkable degree of technology diffusion throughout the economy along three dimensions: (1) horizontally, between and among prime contractors; (2) vertically, among primes, subcontractors, and suppliers; (3) across military and commercial applications, internally within highly diversified firms at the prime contractor level, and within highly flexible firms at the supplier level.

Japanese analysts value the aircraft industry for its capacity to stimulate and diffuse technology; as a result of this capacity and its perceived usefulness, the industry provides a good window into how diffusion works and how it is supported.

“Separate but Together” — Horizontal Diffusion

Most technology diffuses vertically in Japan as elsewhere, through the interactions of “makers” and “users.” But Japanese prime contractors repeatedly cooperate in major aerospace programs. Each of Japan’s airframe makers has played a role in every major postwar aerospace project, and, in engines, all major propulsion projects have been carefully coordinated. While firms compete to become prime contractors, they do so in the knowledge that the competition is not “winner take all.” Failed bidders routinely become subcontractors, receive a fixed workshare, and participate in the design or licensing process.2 The commercial side is little different. The airframe manufacturers who were partners in the domestic YS-11, and in every military project, cooperated as risk-sharing subcontractors on the Boeing 767 and 777.

A stable division of labor has been achieved in which firms target areas of technological competence in order to continue to participate. Occasionally a firm will vie for expanded responsibility, but limitations of capital and skills usually result in “traditional” assignments.

Insiders describe the Japanese aircraft business as a cozy “friendship club” in which each participant has, over decades of cooperation, become intimately familiar with the others and their particular engineering capabilities. One defense contractor from the more competitive electronics sector suggests that “in aircraft, like in construction, it’s all rigged.” Close personal ties help. For example, the plant managers of Japan’s three largest aerospace works — MHI-Nagoya, FHI-Utsunomiya, and KHI-Gifu — worked together on collaborative projects in both military and civilian sectors, both in Japan and in Seattle. Similarly, the chief engineer of MHI’s F-15, T-2 CCV, and other major projects graduated from Tokyo University in 1957, the same year as the chief engineer of KHI’s T-4 program. Although employed by different firms, they worked together on the YS-11 and the C-1.3 Much horizontal diffusion of technology is abetted by such interpersonal relationships. The deputy chief engine designer at MHI, for example, spent two years at the Japan Aero Engines Consortium (JAEC), where, by his own account, he visited Ishikawajima-Haruma Heavy Industries’ (IHI) engine facilities some “ten to twenty times.” He added that the men who later became the chief engine designers at KHI and IHI were with him at JAEC.

Collaboration is closest during the initial design stage of a project, and ultimate responsibility falls to the chief designer. After tasks are assigned, the work is done at the home plants. As a director of the Japan Aircraft Development Corporation put it, “At that point, firms do not give each other technological information, but individuals certainly do.”

Some observers attribute Japanese collaboration to the rising costs of aircraft projects and the decline in their number, but, in other economies, the number of firms has been reduced in response to the same pressures.4 In Japan, partners are stable even if they simultaneously compete. Shared tasks and complementary technological skills, not industry consolidation, are consistent with the diffusion aspect of Japan’s technology and security ideology.

Keidanren, the Federation of Economic Organizations, Japan’s peak business interest group, has been a leader in exhorting horizontal collaboration. In 1965, it acknowledged that large-scale projects required the integration of enormously complex technologies from disparate fields and urged that interfirm, interdisciplinary teams of engineers be created to undertake national projects:

While it is valuable that each firm in the aircraft industry undertakes its own research and development, it is even more important that each specialized firm come together in a comprehensive body in a spirit of fellowship, and that government-business cooperation be achieved.5

Private firms are certainly competitive, yet their behavior closely tracks these sentiments. Companies argue for inclusion in major projects, for example, on the ground that technology diffusion will help them, and the economy as a whole, compete against the rest of the world. In 1986, the Japanese government subsidized the country’s small Advanced Turbo Prop (ATP) engine project. Corporate participants, many of whom duplicated each other’s skills and capabilities, variously justified their roles on the basis of (1) how much the project would contribute to their ability to “confront Western makers” (2) the capacity to expand Japanese global market share; and (3) the need “to be able to compete with Western firms.” No firm sought to justify participation on the basis of individual advantage, much less return to shareholder equity, and none mentioned domestic competition. Each of the leading participants linked the ATP project to their other commercial activities.6

A Key Technology Center Aerospace Project

The ATP consortium deserves closer analysis, for it reveals much about Japanese ideas on the ways in which firms and the state conceive of the technological enterprise. In June 1985, after considerable bureaucratic infighting, the Ministry of International Trade and Industry (MITI) and the Ministry of Posts and Telecommunications (MPT) established the Key Technologies Research Promotion Center (KTC) to support joint research in basic high-technology areas. The KTC thus became the latest institution for the promotion of Japanese cooperative research.7 The KTC supports three categories of activities by providing: (1) equity for research and development companies comprised of private firms engaged in joint research (up to 70 percent of total capitalization); (2) loans to private joint venture research firms (interest repayment is “success conditional”); and (3) basic infrastructure to collect and diffuse scientific and technical information, to promote international research cooperation, and to facilitate other forms of joint research, including access for private researchers to national research facilities. By 1991, KTC support for capital investments in joint R&D firms had risen to ¥21.7 billion, nearly eleven times the original budget, and seventy-seven separate joint R&D firms had been capitalized.

“Key Tech” projects reflect clearly the political compromise that created KTC. There are two kinds: the MPT-supported “teletopias,” which provide regional information systems to local governments, and the MITI-supported “R&D projects,” which look very much like other MITI consortia but give the state an equity stake. There is considerable incentive for firms to participate in KTC projects, technology diffusion being of central interest. Declares the chief executive officer of one joint R&D firm:

This support is critical, as research firms have no income, only expenses. We need this money, and by the time this effort is transformed into products, we will have fully depreciated all our equipment and the benefits of the technology will have been diffused broadly throughout our parent firm.

State funding is important, but private firms are the primary actors in all phases, from conceptualization through research to exploitation of results. Research consortia are initiated by private firms, not by MITI or MPT. Participant firms retain all patent rights, which the research partners divide among themselves. The first four years of KTC investments generated nearly 1,000 patent applications. Technology is diffused and nurtured, while competition is maintained.

In March 1986, the Key Technology Center created the Frontier Aircraft Basic Research Center Co., Ltd. (FARC), in which the KTC held 70 percent equity and thirty-four private firms shared the remaining 30 percent of the initial $5 million capitalization. Only seven of the thirty-four firms are engine or airframe manufacturers. Ishikawajima-Harima Heavy Industries (IHI) holds more than 10 percent of FARC equity, followed by Kawasaki Heavy Industries (KHI), Sumitomo Precision, Mitsubishi Heavy Industries (MHI), Fuji Heavy Industries (FHI), Kobe Steel, and NEC. The remaining shareholders — three automobile manufacturers, materials fabricators, and machinery makers — purchased less than 1 percent of FARC equity. The FARC president was from IHI, while its board of directors represented each of the major private shareholders and included a former Air Self-Defense Force general, then a managing director of the Society of Japanese Aerospace Companies (SJAC), the industry association. FARC would receive ¥4 billion in annual capitalization for its seven-year lifetime and would undertake research to generate intellectual property for subsequent licensing of an ATP aircraft engine. No manufactured products were planned for this R&D company, which ended operations in March 1993.

ATP technology — especially the highly fuel-efficient and low-polluting unducted version — is considered by some the “crown jewel of commercial aeroengines in the next century.” U.S. research was actively pursued in the early 1980s under NASA contract, and an ATP prototype was mounted and flown successfully at the Paris Air Show later in the decade. Although FARC principals deny it, the initial inspiration for the Japanese ATP project may have been the Boeing 7J7 midsized commercial transport.8 In June 1985, a U.S. consulting firm, Hill & Knowlton, surveyed the potential U.S. military market for the ATP and unducted fan aeroengine. Hill & Knowlton reported that, due to “military sensitivity to releasing information,” they had “met some resistance in the course of interviewing.”9 They concluded there was no immediate U.S. military demand but expected a military market to emerge. A more comprehensive follow-up report, covering the commercial airlines, NASA, and the Federal Aviation Agency, concluded that U.S. airlines did not recognize the need to convert to ATP because energy prices had declined and because airline price competition made investment in advanced technologies unattractive. SJAC and MITI commissioned both reports. As one former SJAC official notes, the ATP (like the supersonic transport of the future and the YS-11 of the past) is “seeds, rather than needs driven. We are doing it because we have the opportunity to position the industry.”

If there is no immediate market, why proceed with the project? What is the industry being positioned for? The reasons are closely related to the Japanese security and technology ideology. First, the advocates stressed autonomy — the chance to free Japanese manufacturers from dependence on Western firms. The ATP project enables Japanese manufacturers to do fundamental design and prototype work.

In particular, the ATP project involves independent computer-aided design and development of the sophisticated “hot parts” and digital controls of a technologically advanced engine. Licensing of foreign technologies and subcontracts to foreign firms are permitted, but such collaboration must “provide special technological advantages” to the Japanese participants and are considered undesirable:

We are trying to establish our own capabilities so that we can collaborate with foreign firms in the future as equals, not as subcontractors. We want to catch up so that we will be able to do more than just give money to the participants in joint projects. We want to participate technologically. It is not really collaboration when all we are is subcontractors.

Second, the official “Project Outline” stresses how testing and research for the ATP enhances prospects for a wide range of industrial applications. A stated central purpose is “to create stunning developments in broad areas of commercial industry such as information systems, mechantronics, transportation, machinery, energy, and materials.”10 Diffusion across business applications is “designed in” to FARC. Eight main subsystem projects are each allocated approximately ¥15 million to ¥25 million from FARC and are undertaken at the home laboratory of the participating firms. There is no central laboratory, and, as in most Japanese consortia, this limits the diffusion of technology. For example, the participants are required to share their research results, but they are not required to share the process technologies that enabled them to achieve those results. Indeed, the distribution of research activities and the containment of these activities in the home laboratories ensure that participants can acquire competitive advantages even while they collaborate.

The organizing principle that balances diffusion and competition is referred to as “together but separate.”11 In the case of the ATP, each major research group is the responsibility of a different major private shareholder, and each subsystem is assigned to competing groups of materials fabricators and systems integrators. For example, Sumitomo Precision was assigned design and final evaluation of the composite structures for the turbine blades, but Toray, Toho Rayon, and Bridgestone each pursued separate materials processing technologies. FARC pays varying portions of the salaries of researchers participating in FARC projects and pays for new testing equipment. The equipment will be sold to participants at the end of the project at a highly depreciated price. By 1990, ATP subsystem models had been designed and tested. FARC officially claimed ten patent applications by 1990, eight for power plant technologies and two for the propeller system. Each participant enjoys the right to license the intellectual property of the entire project.

Third, the project nurtures the capabilities of a large number of manufacturing firms that otherwise could not engage in advanced engine design. A FARC senior manager pointed out:

FARC is not just an organization to build an advanced turbo prop engine. It is designed for basic research with very broad application in almost any kind of future engine, gas turbine, space, automobile, military, or commercial project. The ATP per se is less important than learning how to do concurrent design and engineering, and how to integrate a system.

Many participants are now using materials and tools they have never used before. The FARC manager noted that this was deliberate: “Including the parts and materials firms was the result of MITI guidance.”

The ATP engine project is one of dozens of consortia in aircraft propulsion, materials, and components —consortia that receive public support. Each case ensures virtually all major industry players a substantial role. Japanese firms compete vigorously, but this vigor has its limits, and competition is rarely allowed to hamper the search for technological advantage for domestic firms or the nation.

A crucial institutional nexus for technology cooperation and diffusion is the SJAC. Airframe and engine makers join component and equipment manufacturers in dozens of collaborative research activities conducted under the auspices of the SJAC and its Advanced Aerospace Technology Development Center. The AATDC was established in June 1980 to promote R&D in propulsion, control systems, airframe manufacturing, and other basic aerospace technologies, and it — like SJAC itself — serves as a conduit for MITI and other government funds.

Fewer than one-quarter of SJAC collaborative projects combine the resources of competing firms. Although results are shared widely through reports and briefings to all SJAC members, most collaborative research under SJAC auspices diffuses technology vertically (or diagonally) among firms at different stages of the value-added chain. These collaborative relationships are often designed specifically to test competing technological approaches.

“All in the Family” — Vertical Diffusion

The vast majority of Japanese joint research is conducted among noncompeting firms rather than direct rivals. Fully four-fifths of interfirm research in Japan is “vertical,” involving firms at different phases of production. Common examples are “maker-user” combinations, such as steel firms and automakers or ceramics and semiconductor manufacturers. In aerospace, the usual combination is electronics and airframe integrators. Often, these combinations are part of larger, integrated families of firms, keiretsu connected by financial and historical ties: Mitsubishi Electric and MHI, for instance, or IHI and Toshiba. However, vertical relationships between makers and users, between assemblers and suppliers, in different keiretsu had become very common by the 1980s.12 It is likely that increased specialization stimulated managers to look beyond their conventional partners.

As elsewhere, technology also diffuses through vertical links outside the formal arrangements afforded by research consortia. Subcontracting is vital to aircraft manufacturing, and roughly 70 percent of Japanese aerospace work is subcontracted by the leading primes. Each prime contractor maintains 300 to 500 direct relationships with domestic vendors of materials, components, and parts.13 As the primes develop their networks of suppliers and affiliated firms, which, in turn, re-subcontract, thousands of firms participate. These networks are geographically concentrated. In airframes, MHI and KHI are located near Nagoya in central Japan, and, in the early 1990s, FHI began to construct its own facility in the same area. Nearly three-quarters of airframe production takes place in Aichi Prefecture. A significant number of vendors and components suppliers in that region serve airframe and final assemblers, but more than half of Japan’s aerospace electronics, engines, and materials manufacturing is undertaken near Tokyo in the Kanto region.14 While such geographic concentration is not unknown in the United States, relationships among the Japanese firms are often exceptionally durable and diversified; subcontractors and suppliers in turn ensure access to technology and skills from the primes in a fashion that does not favor or exclude particular firms but diffuses knowledge as widely as possible.

Unlike most sectors in Japan, however, aircraft industry subcontractors have not yet assumed primary responsibility for product design and integration. Military and commercial licensing or subcontracting has generated a one-way flow of knowledge from the primes to lower-tier producers. In most cases, technology or manufacturing know-how is transmitted at the start of each project, when the subcontractors dispatch a team of engineers to the primes for weeks or months of detailing training. After both sides are satisfied that the engineers can meet production objectives, the team members return to the subcontractor and apply what they have learned.

There is some evidence, however, that aerospace and defense subcontractors have already taught the prime contractors much about certain specialized technologies. According to one former chief engine designer for IHI:

Our bearing suppliers, such as NSK, really knew their stuff. They taught us about bearing life, wear, and maintenance. But we also took good care of them. I personally took two of our subcontractors engineers to Cleveland so TRW, our licensor, would teach them how to make turbine blades. After all, prime contractors cannot do it all alone.

Some evidence suggests that the flow of technology from supplier to assembler is increasing. Most of the subcontractors to KHI, for example, are no longer members of the Kawasaki regional industry association, partly because the suppliers no longer need the technological guidance that accompanies membership. An MHI plant manager explained:

The amount of design work undertaken by our subcontractors is increasing along with overall outsourcing. Sometimes we give them blueprints, sometimes just the specifications. The latter is increasing.

Overall, however, most technology flows from prime to subcontractors. Production problems are solved in several different ways. Engineers and technical staff interact with their counterparts at the primes on almost a daily basis. At the MHI Sagamihara factory, where the Japanese Defense Agency’s (JDA) Model 90 tank is assembled, slightly fewer than 10 percent of the 1,200 outside contract workers come directly from the supplier firms. For the J-3 engine, IHI’s control bureau chief had a map of Japan in his office showing the site of each subcontractor (down to nuts, bolts, and pins) and where his “troubleshooters” had gone “to threaten and cajole” delivery and quality. A former chief of the JDA’s Technical Research and Development Institute (TRDI), now a senior director of SJAC, connects this supervisory function to the diffusion of technology:

It is perfectly natural in Japan for primes to dispatch engineers and managers to their subcontractors to make sure that products are built to specification. Of course, it is a natural path for technology transfer when primes send their personnel to these out-of-the-way facilities.

An MHI factory production department manager confirms that “part of our long-term strategy is to distribute skills and capabilities to other firms — without sacrificing cost or quality.” In the case of solid rocket propellent, MHI dispatches four or five of its own engineers to the vendor. Each subcontractor also gets an annual or semiannual inspection that generates a detailed report card, which “grades” the subcontractor on an A through D basis in a variety of categories. The subcontractor often uses the report as an action plan to upgrade its performance.

Many subcontractors hire retired technical staff from the primes to obtain production knowledge and, in effect, to “buy” direct access through personal contacts. MHI sends about ten retirees each year to subcontractors as “directors” and “counselors.” When active cooperation is required, “joint engineering teams” are formed. Machinery purchases and autoclaves for metal bonding and composites manufacturing are made in close consultation with the prime, to complement or supplement the prime’s internal capabilities.

Although systematic data about vertical flows of technology are not available, two things seem clear. First, prime contractors and vendors alike benefit technologically from their mutual dependence. Two-thirds of the subcontractors in the Gifu region alone were established by former employees of the larger automobile and aerospace manufacturers, for example. Second, this mutual dependence tends, over time, to attenuate. Prime contractors rely increasingly on their vendors, while the vendors diversify their dependence on single prime contractors. Of the twenty-five KHI subcontractors in the Gifu association in 1985, only one overwhelmingly relied on KHI for orders. The most competitive firms in the components sector are those which find multiple uses for the technologies they developed for aerospace.15 Prime contractors look for a diverse supplier base; subcontractors seek a diverse customer base. As a consequence, highly flexible prime contractors have nurtured an equally diverse and flexible supplier base.

“The Glass Wall” — Intrafirm Technology Diffusion

In Japan, where large and small manufacturers are increasingly diverse and flexible, just as much technology flows within firms as outside. Commercial and military work is performed by the same shop personnel using the same test and production equipment, usually in the same plant or facility. Separation occurs usually only at final assembly, and there are few secrecy restrictions. At Mitsubishi Electric, for example, access to military research results is formally proscribed for unrelated employees, but as a matter of common practice, supervisors never decline requests for access from within the corporation. Military researchers openly acknowledge this free flow of nominally restricted information. One TRDI official notes:

Although from the perspective of secrecy it would be better if civilian and military activities were separated, in Japan we have to be concerned about efficiency and about learning how to make products that are low cost as well as high performance.

In 1985, SJAC studied technology diffusion and identified mechanisms by which technologies are transferred.16 Even though the report was produced by the industry it claimed to evaluate, the study showed that the extent of aerospace and defense technology interdiffusion is dramatic. In the case of submersible craft, for example, marine engineers were dispatched to the aircraft divisions of their parent firms for training, and they also received “technical leadership” from competing submersible manufacturers. In the case of the space industry, engineers and designers were transferred in-house to take advantage of their experience in aircraft materials and testing.17

Technologies were transferred for a variety of reasons, including more efficient product development (e.g., to shorten development cycles in automobile and shipbuilding and to streamline the assembly of buses and rolling stock) and improvement of product performance (e.g., for energy efficiency of automobiles, for safety of engines and underwater craft, for improved durability of commercial equipment). The study documented more than 500 discrete cases of technology transfer across sectors, largely within individual firms.

As 80 percent of all aircraft manufactured in Japan have been for the military, the value to the commercial economy of the military sector — despite nominal restrictions on the use of technology — is obvious. An MHI manager confirms this:

We have used government-funded R&D results for our commercial applications; but officially, we do not use military R&D for civilian production. But the know-how we gain via experience in building, shaping, forming, and designing tools and equipment is used. We cannot measure the spillover, but it is there. The key is not to ask permission. We just do it.

In Japan, where defense production is firmly embedded in the commercial economy, there are exceptionally low barriers to the interdiffusion of military and commercial technologies. A senior manager of Mitsubishi Electric explained:

We have no defense specialists because we do not distinguish civilian from military projects. We have no ‘walls’ separating civilian from military projects.

This is accurate as a general characterization of Japanese production, but it is overstated. A MITI official notes that government policy requires at least some kind of barrier:

Just as in Keidanren [where the Science and Technology and the Defense Production Committees share offices and personnel but are formally separated], in MITI we have to build a glass wall between space and aircraft and ordnance. Industry, however, builds no such wall, and the management for rockets is the same as for missiles and jet fighters.

An unpublished JDA memorandum begins to substantiate the extent of this interdiffusion, although it does not provide details about specific firms (see Table 1). An accompanying memorandum provides examples of spin-on within firms from their commercial to their military businesses (see Table 2).

The capacity to spin on or off commercial and military aircraft technologies to other industries varies with the scale and organization of the firms involved. The process is least impressive at the prime level, yet even there, the extent of technology interdiffusion is remarkable. Japanese primes usually house their aircraft facilities in factories geographically separated from the rest of their commercial activities. In electronics and other components, however, there is little distinction made, even in larger firms.

There is only limited evidence that Japanese prime contractors systematically transfer engineers across product lines, from ships to aircraft or from automobiles to tanks.18 The engineers who designed the Model 90 tank also helped design the Model 74, and were assigned responsibility for the Model ’04. The same aeronautical engineers who worked on the FS-X participated in the F-15, F-104, C-1, and YS-11. Instead, the mechanism used in large Japanese firms to diffuse expertise might be called “multifunctional specialization,” and it varies across firms.19 Where applied, it nurtures application-specific expertise (e.g., radars, tanks, aircraft, automobiles) through the systematic rotation of engineers across functional fields. Morino and Kodama may be correct in suggesting that “spin-off effects are more in fundamental design procedures, development management, and the improvement of an engineer’s potential than in the products themselves.”20

There is surprisingly little transfer of personnel — engineering or production — across from one plant to another unless whole divisions are in financial trouble, like the MHI shipbuilding division in the late 1980s or Toshiba in the early 1990s. Most senior engineers at Japanese prime contractors will have spent their entire careers working on aircraft, but they will have had experience in all aspects of aeronautical engineering, including hydraulics, design, materials, and electronics. For example, personnel records show that the three deputy directors of the MHI Nagoya Works in 1971 and in 1981 had spent their entire careers at Nagoya specializing in aircraft. But each of the six had experience in different functional specializations: materials, manufacturing, fuselage design, personnel, fighter design, aeronautical testing.21

The depth and breadth of the engineer’s functional expertise seems to matter far more than the application of that expertise to civilian or military projects. In 1981, for example, the top officials of the MHI Nagoya Works Second Technology Department (former designers of the T-2 trainer) supervised eleven divisions and projects that included all structure and design for civilian and military projects. Often the result is direct cross-fertilization across applications, as cases from the Toshiba defense systems group and from Nissan Motors suggest (see Tables 3 and 4).

To overcome — or perhaps to maximize — the benefits of “multifunctionally specialized” career paths, prime contractors create elaborate networks of research committees to consolidate their knowledge in specific functional areas. At Shin Meiwa, Mitsubishi Electric Corporation (MELCO), and MHI, for example, the Technology Headquarters sponsors firmwide study teams that coordinate at both the plant and the corporate level on functional topics such as electrical machinery, heat treatment, and inspection. Each study team meets quarterly to share know-how.22 Within Mitsubishi Electric, for example, an in-house “engineering institute” has more than 10,000 participants (including more than 500 PhDs) and sponsors in-house working papers produced by and circulated to every business group within the firm. An engineering “cram school” selects twenty promising engineers each year from across the firm’s business areas and subjects them to more than 600 hours of special lectures and requires submission of research papers in twelve separate fields. In addition, MELCO factories consign research projects to central laboratories and across business groups. This consignment often entails the secondment of engineers as well.

In the case of MHI, the Technology Headquarters supervises six separate R&D centers, most attached directly to MHI factories. While each focuses primarily on the products specific to the plant to which it is attached, central headquarters overlays the local organizations with twelve functional responsibilities. As a result, a matrix of functions and applications is created that forces researchers from across the firm to interact regularly. It was this system of corporatewide, centrally coordinated, functionally defined, multiple-application study teams that introduced image processing, stealth technology, and a range of process technologies across MHI. In addition, the top research managers of each firm of the Mitsubishi Group meet on a regular basis to discuss new technological developments.

Although aircraft and defense operations are separate from other divisions at the prime level and despite career paths that direct engineers into application-specific channels, there is considerable technology interdiffusion within firms. These transfers occur through at least three distinct mechanisms: (1) across sectors through combined project teams, (2) through common machinery on the shop floor, and (3) through common engineers in the same sector who are not expected to distinguish military and civilian tasks.

MELCO’s active phased array radar (APAR) illustrates how commercial technology can be “spun on” to military applications across sectors within a single Japanese firm.23 APAR technology was first developed in the United States, and, by the 1970s, U.S. engineers had learned how to realize all transmission and reception through solid-state integrated circuits first developed by Texas Instruments. This development made a reliable airborne APAR possible for the first time.

MELCO learned of the Texas Instruments project in the 1960s and raised the prospect of an indigenous APAR with the TRDI. The TRDI first funded indigenous APAR development in the 1968 budget. MELCO carried more than half the development costs, as well as 30 percent of the costs of the prototype. Each application needed a separate module, and MELCO relied on gallium arsenide (GaAs) and silicon transistors developed originally for commercial use. But these integrated circuits were not available in the MELCO radar division, and MELCO’s semiconductor division thought the quantity too small to bother with. Several engineers from the radar group briefly transferred to MELCO’s electronic devices group, where they received training in GaAs chip manufacturing technology. Leveraging MELCO’s GaAs commercial memory technology, MELCO produced, with a fraction of the R&D support American firms received, an APAR prototype that many regard as fairly close to leading-edge U.S. capabilities.24 Commercial modules were being sold to U.S. firms as early as 1990, and, in the early 1990s, the JDA and private industry were actively studying nonmilitary, nonaircraft applications of the technology.

Superconductivity research at MELCO exhibits a different pattern in the use of engineers to achieve inter-diffusion. The same researchers have worked simultaneously on superconductivity applications for magnetic levitation, electric power generation, and submarines. The result, what a MELCO senior manager calls “concurrent interdiffusion,” was obtained by overlapping researchers and projects across nearly two decades. Although MELCO’s first major superconductivity research was funded by the military, researchers made no distinction between military and commercial applications. The JDA paid for 3,000 hours of the MELCO researchers’ time (375 person-days) while they worked concurrently on other superconductivity research projects between 1972 and 1974. Thus there was no shifting in and out of specific military and civilian projects, and there were no classification and secrecy requirements.

The MELCO researchers involved in the MITI Large-Scale Project on Superconducting Magnets were also supported by the JDA. Their career paths reflect the ease with which the results of these projects diffused across civil and military applications (see Table 5).

Research engineers were formally assigned to other projects as they participated in the TRDI submarine project. In 1992, MELCO was awarded a ¥500 million TRDI contract to develop superconductive magnetic sensors to research and prototype fabrication of an advanced magnetic anomaly detector (MAD) to be installed on submarine hunting aircraft. Once again, the Number Four Research Group at MELCO’s Central Research Laboratory — including some of the same researchers listed in the table — was assigned the project.

Interdiffusion has long been stimulated during production as well as during research. In 1966, nearly 90 percent of the capital equipment used for military production (92,000 machines) in Japan’s defense plants was general-purpose equipment, designed, built, and employed simultaneously for commercial use.25 Defense industrialists did not embrace this flexibility, suggesting they did not yet understand the benefits of dual-use production technology. Indeed, the Defense Production Committee of Keidanren bemoaned the easy “diversion” of these machines to commercial projects as a “sacrifice.”26

Neither is it clear that contemporary defense production is the product of grand strategy. A senior manager of MHI’s aircraft division offers a compelling alternative in the particular case of engines:

We have debates here over separating commercial from military [production]. From the business side, it may make sense to separate, but since our aircraft business is so small to begin with, if we divide it further, we lose economies of scale. It is not that it is MHI policy to combine civil and military [production] for technological purposes, however. More important is the “loading” problem. We have so few projects and have stable employment responsibilities, so we need to flatten the peaks and valleys associated with aircraft projects. So, while we get technological benefit from mixing, the loading issue is more important.

Whatever the reason, the same story can be told about each of Japan’s prime defense contractors, where the same work groups, on the same machines, on the same day, produce parts for jet fighters, missiles, and for Airbus or Boeing. Scattered around factories are pallets of titanium fighter components, hardened missile cases, and aluminum 767, MD-11, or A-321 fuselage parts. Indigenous trainers are wired and equipped by teams that can and do shift with ease to commercial subcontracts. Blueprints for military and commercial aircraft are stacked next to, if not on top of, one another in even the largest factory. And in assembly areas, military aircraft take shape next to sub-assemblies for commercial transports.

This story involves more than shared buildings, however. Like some U.S. counterparts, especially at the subcontractor level, Japanese producers routinely continue to use common production equipment. Kamata Satoshi describes a Yokohama plant of Japan Avionics in which military and civilian production was indistinguishable except for the replacement of aluminum with stainless steel casing on the electronics gear. At Shin Meiwa Industries’ Konan factory in Kobe, commercial and military production is completely integrated. Operations are divided by function (design, final assembly, structure, production planning, etc.) but not by final product or component. The same managers supervise production without regard to end use: the machining center, bonding equipment, welding, and surface treatment activities for the JDA, Boeing, and McDonnell Douglas are all combined.27 At the MHI Nagoya Aerospace Systems Works, the autoclave installed to cure the composite wings of the FS-X in 1989 as well as the plant’s chemical etching equipment, machine shops, inspection facilities, and flexible manufacturing systems are all used for commercial and military projects without distinction. The dual-use fabrication and processing equipment bought by the firm is “charged off,” as appropriate, on a percentage basis to the JDA. There is no quarantine in the job shop for military products.

In the MHI Sagamihara plant, Japan’s only active tank factory, a thin, unguarded canvas screen separates the final assembly area for the Model 90 tanks from the rest of the plant, where forklift trucks (18 percent of the plant total), diesel engines (33 percent), and heavy construction equipment such as bulldozers (12 percent) are also manufactured. While final assembly of tanks is segregated from the rest of the plant, they share production equipment — including the machinery that makes crank shafts, camshafts, gear cases, revolving engine parts, turbochargers, and fuel pumps and gears; the equipment that provides heat treatment and surface treatment; and the same tools, including jigs, and test and measuring equipment. Plant officials reported that the segregation of tank assembly operations is due not to security concerns, but simply to different volumes of production. Tanks are assembled in very low quantity (twenty per year) at a stationary site, with design changes only every decade and a half; forklifts and engines are mass-produced on a conveyer line, and designs change every five years. Moreover, the tanks contain a considerably larger proportion of ready to assemble, procured goods, and, as a result, they need less access to production equipment and more access to loading docks. The proximity of tank and equipment manufacturing has resulted in spin-offs of measuring and test equipment and materials processing, and in spin-ons of components and production process control. In the meantime, the combination has enabled plant managers to engage in fully integrated engine design work, testing, and other activities that enhance the interdiffusion of technology.

MELCO’s development of three-dimensional (3-D) radar provides an opportunity to understand intrafirm spin-on and spin-off in the broader context of indigenizing technology and nurturing Japan’s technological capabilities.28 In the early 1960s, the JDA wanted to incorporate altitude data and improve the data processing capabilities of the (then new) BADGE air defense system. The TRDI announced its 3-D radar requirement in 1961, and in 1962, it commissioned NEC, Toshiba, and MELCO proposals. Only MELCO pursued 3-D radar independently, and in 1963, the JDA selected the MELCO design “in recognition of its efforts to develop indigenous technology.”29 During the next two years, MELCO and TRDI jointly developed the prototype, which was first tested in June 1966. After deployment, Japan had the world’s only permanent ground-based 3-D radar for air surveillance. The JDA notes two central benefits: (1) it replaced much imported hardware in Japan’s air control systems, and (2) it led to new civil air control applications.30

Just as prototype production began, development started on a mobile 3-D radar of different design. After this technology was adopted for civil aviation and improved, it was reintroduced to the JDA. Thus the inter-diffusion of 3-D radars is not a simple, idealized process. Tamama Tetsuo, supervisor of MELCO’s radar development, charts the flow, shown in Figure 1.

Of the five top MELCO engineers responsible for fixed 3-D radar development, only two had previous experience with a JDA radar project and only one subsequently moved to another field — commercial digital process controllers and then the presidency of a commercial software subsidiary of MELCO. Engineers become “multifunctional specialists” and tend to be assigned to limited applications areas, but firms are willing to circulate their engineering talent in order to enhance the interdiffusion of commercial and military technologies. At Shin Meiwa, Kikuhara Shizuo, chief designer of the PS-1, was on the design team of the wartime Model II seaplane and the US-1, a failed attempt to develop a towed array sonar system (pulled by seaplane). Each of his deputies on the US-1 had followed him from the PS-1 but subsequently went to other projects — some even outside the sector (see Table 6).

This seems to be the case in other firms as well. To-shiba makes military and civilian radars with the same design teams in the same facilities. A former IHI engine controls designer reports:

I never paid much attention to whether the control systems I was building were for military or civilian uses; I just built them to the specification, and, when I became a manager, I supervised their production simultaneously.

Japanese subcontractors and suppliers achieve even more systematic interdiffusion of aerospace and nonaerospace technologies. Unlike their U.S. counterparts, Japanese lower-tier producers are primarily not suppliers to aircraft manufacturers. Typically, 80 percent to 90 percent of their production is in nonaircraft industries. The combination of aircraft and nonaircraft production facilitates an enormous cross-fertilization of technologies and skills. Here are two examples of this process.

In one case, a well-established aerospace machine-shop subcontractor with 100 employees discovered that chip removal for the sophisticated numerically controlled (NC) machine tools involved in aircraft production was difficult. It experimented with conveyer systems and telescopic covers for NC equipment, forming a joint venture with a German firm to import technology. Now the company designs and produces, under its own name, world-renowned conveyers and covers. It also produces the specialty machines required to make the conveyers. While remaining an integral part of the Japanese aircraft industry’s subcontracting network, the firm relies on aerospace work for only 15 percent of its revenues; machine tool accessories now account for 85 percent of its business and nearly all of its profits.

A second example illustrates the spin-on capabilities of supplier firms. One of Japan’s most successful textile firms is also a highly sophisticated aircraft component supplier, specializing in fuel injectors and flight control equipment. Approximately 25 percent of the company’s sales are in aerospace; the remainder are in textile equipment, robotics, and industrial machinery. The production of robotic transfer gear systems, the firm discovered, actually involved tolerances more acute than aircraft parts specifications. The firm began to adapt its nonaircraft quality control and process techniques to its aerospace operations, dramatically increasing the quality and reducing the cost of its products, achieving a commanding presence in segments of the world aircraft market. The company is now a sole source of flight control equipment for at least one major overseas commercial aircraft program, and, for many others, it is one of two or three remaining sources worldwide.

American subcontractors and suppliers can and do mix aerospace technologies and machinery, but they have generally stayed in the aircraft business.31 The Japanese have found very little difference in customer needs in the aircraft and in other industries, but diversification has eluded their U.S. counterparts. One first-tier U.S. supplier admits that his firm lacks confidence that it can make a successful foray into sectors “where standards are lower.”32 An American defense consultant came to a similar conclusion:

The forging, casting (foundry), and fastener industries share several important characteristics. In each of these industries, firms that manufacture products for the defense industry do so almost exclusively for defense and aerospace customers. The products they sell are manufactured in very small quantities and are of high quality relative to products sold in . . . commercial markets. . . . As a consequence . . . these firms are generally unable to compete in commercial markets for high-volume, low-technology products. Although they are technically capable of making commercial products, they are usually unable to do so in an economic fashion. At the same time, firms that manufacture in large volume for commercial markets are usually unable to compete in defense and aerospace markets because they lack the necessary skills and equipment. In those instances where it may be possible to manufacture a product, it generally cannot be done economically, again because of the inappropriateness of the equipment, the people, and the organization to do the job.33

It is most striking that virtually all of the conclusions explaining why aircraft and nonaerospace production are incompatible apparently do not apply in Japan.