The Information Storage Industry Center
Graduate School of International Relations and Pacific Studies
University of California
9500 Gilman Drive
La Jolla, CA 92093-0519
Copyright © 1997, David McKendrick
University of California, San Diego
Funding for the Information Storage Industry Center is provided by the Alfred P. Sloan Foundation.
To receive hard copy of a document, send an e-mail with your address to the Publications Coordinator at firstname.lastname@example.org.
The hard disk drive (HDD) industry represents an interesting exception to received wisdom about American industrial competitiveness. Until recently, scholars have been pessimistic about the competitive prospects of much of U.S. industry, observing that "[i]t is too late for the United States to regain its position of the exemplar of best practice in the world" (Kogut (1993: 11). Florida and Kenney (1990) concluded that America may be good at generating new industries but is bad at sustaining them as they become more mature. American industry in general was said to have "attitudinal and organizational weaknesses" leading to "shortcomings in the quality and innovativeness of the nation's products" (Dertouzos et al., 1989). Yet, the experience of the disk drive industry suggests that these characterizations of American industry need not be its paradigmatic form. Like many industries that emerged in the 20th century, the disk drive industry was dominated by American firms during its early years. Unlike other industries, however, the United States never relinquished its leadership. U.S. companies hold more than 85% of the global market, an even greater share than in the late 1970s.
Why has the United States been so consistently successful in this industry? This paper argues that the industry's globalization was an important factor in sustaining its competitive advantage. This is not to say that other mid-level and micro-level factors often invoked to explain the success of certain nations in particular industries do not apply to the disk drive industry. Home market demand, form of industrial organization, innovative capabilities and the role of institutions--universities, government agencies, business associations and other regional and national entities--all play or have played some role in the industry's evolution. The principal point here is that foreign investment complements innovation, style of industrial organization, and the favors conferred by historical chance, and is critical to sustaining industrial performance.
Curiously, while considerable recent attention has been given to accounting for the competitive performance of national industry in international markets (Dertouzos et al. 1989; Porter, 1990; Hollingsworth et al., 1994; Nelson, 1996), global competition is seldom addressed directly except in reference to exporting from the home base (e.g., Porter, 1990). One result is that global investment has not come through as an important factor in sustaining or reinforcing the competitive performance of national industry.1
This is surprising since industries are producing an increasing portion of their value-added overseas. By the late 1980s an estimated 25-30% of global production was accounted for by the affiliates of multinational corporations (UNCTC, 1992). And while Japanese multinational corporations emerged later than their European and American counterparts, since the 1970s they have globalized production rapidly. Between 1982 and 1985 the value of Japan's outward investment actually exceeded that of the U.S. and was roughly equal to U.S. outward flows between 1990 and 1992 (Nunnenkamp et al., 1994). More recent evidence from the business press puts the overseas production by prominent Japanese companies even higher: Sony currently produces 47% of its output outside Japan with plans to increase this to 80%; Toshiba's production in Asia, excluding Japan, will make up over 50% its global output by early 1999; in 1995, Toyota's offshore plants made 58% of vehicles sold abroad, the first year that vehicles produced overseas by a Japanese automaker exceeded that company's exports (CL, 1997).
While only one industry, disk drives offer a fascinating context for charting globalization, as well as industrial evolution more broadly. American firms in particular have accumulated the organizational skills necessary for managing the geographic separation of R&D, production and distribution to achieve location economies. This kind of dispersion has been increasing as well in other industries, but since the HDD industry is farther ahead than most in globalizing its activities, its experiences may provide a glimpse of what is to come for other parts of the American economy and a touchstone for the maintenance of industrial leadership.
This paper is organized into 5 sections. The first three sections address the factors raised in other studies of industrial competitiveness. Section 1 considers the origins of the HDD industry and how they affected competitive advantage. Section 2 analyzes the innovative bases of industrial performance. Section 3 addresses the role played by national style of industrial organization. While I find that each of these factors has contributed to the success of U.S. disk drive firms, taken together they are insufficient in accounting for uninterrupted U.S. dominance in the industry. Rather, as Section 4 argues, American firms were able to sustain their leadership by being the first to move assembly offshore. This move gave U.S. firms not only a cost advantage, it gave them time to accumulate the skills necessary for managing the international separation of product development from volume assembly. The concluding section summarizes the paper's findings.
I. NATIONAL EMBEDDEDNESS AND PATH DEPENDENCE
The home market confers advantage upon national firms, and the success of customer industries confers success upon their suppliers. This national embeddedness is especially true for nascent industries. An industry emerges through the cumulative interactions of entrepreneurs and organizations. Interdependencies are established through the sharing of information and resources, and clusters of firms begin to form (Van de Ven and Garud, 1989). Clusters are largely national or regional phenomena, with firms serving national customers before growing through foreign trade and investment (Chandler, 1990). Ties among national firms persist as they expand into international markets. In his study of competitive advantage, Porter (1990: 138) finds that "a group of internationally successful domestic firms, selling worldwide, channel[ed] global demand to the domestic supplier industry."
The Origins of the Disk Drive Industry
One possible explanation for the success of the American HDD industry, therefore, is American success in the computer industry. This explanation seems reasonable on the face of it. At the time IBM shipped the first rigid disk drive in 1956, America was already the world's dominant computer producer and exporter. Though Europe contributed enormously to the technical development of the early computer industry, American firms led the world in computer installations, and many of these same computer companies developed their own hard disk drives. General Electric, Control Data, Burroughs and Digital Equipment followed IBM's entry into hard disk drives in the 1960s. Some independent companies, such as Bryant Computer Products and Data Products, also emerged in the early 1960s to develop disk drives for sale to computer manufacturers that had not yet made their own, notably Sylvania, RCA, Honeywell and Univac. In the late 1960s, after IBM secured its position as the clearly dominant mainframe maker, a new wave of independent companies emerged to make disk drives that were "plug compatible" with IBM systems: Memorex, Potter Instrument, Marshall Laboratories and Information Storage Systems. Without incurring IBM's R&D expenses, the plug compatible companies were able to offer disk drives identical to IBM's, or better than, at a much lower price. Plug compatibility was not limited to IBM systems but extended to systems made by other computer manufacturers as well.
A parallel trend was evident in Japan and Europe, although on a smaller scale. In Japan, the principal computer companies made their own disk drives: NEC, Fujitsu, Hitachi, Toshiba all entered in the mid-to-late 1960s. Only in the 1970s did Japanese companies attempt to market disk drives to non-Japanese customers in the U.S. market; until then the size of the market for Japanese computers limited the market for their disk drives. A smaller domestic market also meant fewer independent Japanese disk drive companies entered in the 1970s as alternative sources of supply, namely Mitsubishi and Hokushin Electric Works.
In Europe Siemens and Philips made disk drives for their own computer systems, while Data Recording Instruments (DRI) and BASF produced for the OEM market. (DRI was Europe's first firm to ship hard disk drives -- in 1968 to ICL.) Honeywell-Bull (later CII-Honeywell Bull and then Bull Peripherals) did both captive and OEM production. In Eastern Europe COMECON organized the computer industry in such a way that DZU of Bulgaria was designated as the principal disk drive supplier for all computers in the region and became the most vertically integrated producer in the world. Only in rare cases did European disk drives find their way into American or Japanese computer systems.
Thus, throughout the 1960s and 1970s, the relative positions of the U.S. Japanese and European disk drive industries could be explained by incorporation of their products into the systems manufactured by their respective national (or regional) computer industries.2 During the 1970s, captive production remained the largest channel for disk drives, though the relative importance of the original equipment market grew. Led by Control Data, Diablo Systems, CalComp and Memorex, the OEM segment reached $631 million in 1979, but still well below the $2.8 billion associated with captive production (Disk/Trend, 1980). In 1979, American firms had 81.1 percent of the global HDD market, Japan 14.3 percent, and Europe the remainder. Between them, IBM and Control Data controlled just short of 40 percent of the market. The Personal Computer and the Desktop Disk Drive
Up to this point the story conforms strongly to an explanation of competitive advantage through path dependence and increasing returns: the large U.S. market for mainframe, and later minicomputers, gave the American disk drive firms an unassailable long-term advantage. But it does not account for the divergence in the fortunes of the American disk drive and computer industries after 1980 when both came under greater global competitive pressures.
For the computer industry, a watershed event was the debut of the IBM PC in 1981. The PC defined the dominant design in the industry for many years (Langlois, 1992; Anderson, 1995). In addition to setting the standard for what a desktop computer should look like, it featured an open architecture that attracted the entry not only of some of IBM's established mainframe and minicomputer rivals but de novo start-ups that set out to manufacture IBM clones. Compaq and Dell became two of the most important American entrants, but more interesting for our purposes is that so many of the new clone makers emerged outside the U.S., especially Taiwan, Korea and Japan. Daewoo, Epson, Hyundai, Acer, and scores of other smaller companies collectively dispersed the production of computers. As a result, the global market share of U.S. computer makers steadily eroded over the course of the 1980s and early 1990s. Whereas U.S. firms held an estimated 88% of the worldwide computer market in 1983, their share had fallen to around 56% by 1992 (including mainframes). Over the same period Japanese market share in the computer industry increased from 8% to 30%.3
The same open architecture that attracted the new clone manufacturers also stimulated entry into peripheral equipment. Where mainframe and minicomputer manufacturers made many of their own peripherals and components, the assemblers of personal computers almost entirely outsourced their production. Even more dramatically than for the PC market, U.S. firms were displaced as market leaders in peripherals and components by Japanese, Korean and Taiwanese producers of keyboards, floppy disk drives, monitors, DRAMs, and motherboards.
Given these trends, and the development of national clusters of computer-related capabilities in these countries, one might have expected other Asian companies to erode America's position in hard disk drives. Much the same competitive dynamics faced the HDD industry as disk drives were adapted to fit into a PC. Drive sizes decreased from 14-inch and 8-inch diameters in the 1970s to 5.25-inches in 1980 and 3.5-inch in 1983. An explosion of some 100 new entrants, intense competition, and shakeouts occurred between 1980 and 1996. And by 1996, fewer firms made disk drives than at any time during the previous 20 years (Figure 1). The HDD landscape became littered with the graves of once prominent American companies. Priam, Prairietek, Conner Peripherals, MiniStor and Hewlett-Packard were all leaders at one time or another in certain market segments or technologies. There was also considerable entry by Japanese, Korean, and Taiwanese firms, as well as a few from Europe.
But whereas most of the rest of the American computer peripherals industry has largely vanished, its hard disk drive industry remained dominant in the face of competition from Asia and Europe.4 Though U.S. firms exited, so did those from other countries: Mitsubishi, Matsushita, Rodime (the first firm to introduce the 3.5-inch disk drive), Olivetti, BASF, Sony, Philips and Siemens are among the formidable foreign companies unable to remain in the industry. Asian and European PC makers bought HDDs from U.S. firms. South Korea, for example, depended almost entirely on American companies to meet the HDD requirements of its major PC exporters such as Samsung Electronics, Hyundai Electronics, Goldstar Co., Daewoo Telecom., and Trigem Computer (MR, 1991). In Europe, companies like PC companies such as Amstrad also purchased American disk drives. Compared with the computer industry, the American HDD industry held a roughly steady 75% of the global market throughout the 1980s, only to see it rise above 80% by 1992 (Figure 2). By 1995 U.S. global market share reached 85%, where it had been in the early 1970s.
In summary, though the market share for American computer manufacturers fell throughout the 1980s, the American floppy drive industry practically disappeared, and the world increasingly turned to non-American suppliers of other computer components and peripherals, American firms continued to be the overwhelming source for hard disk drives. It is clear that the disk drive industry owed its birth to the American computer industry. But given the fate of other computer peripherals, a path-dependence argument is incomplete. America has been the inventor of other promising technologies that it relinquished to the Japanese. Why could it hold on to hard disk drives but not floppy disk drives, monitors, or optical storage devices? One possible factor that animates many analyses of national competitiveness is innovation.
II. INNOVATIVE CAPACITY
Many argue that Japanese and American firms have distinctly different methods of innovation and innovative capabilities. Japanese firms are said to possess a number of organizational, incentive and communication advantages over their Western counterparts that are conducive to innovation (Aoki and Rosenberg, 1987; Aoki, 1990). More specifically, the organization of research and development is characterized by decentralization and strong horizontal and vertical communications channels. First, dispersed problem solving gives units greater autonomy, meaning there is "heavy reliance on the localized use of on-site information rather than hierarchical control over specialized knowledge along with an integrative approach to problem-solving and operating tasks" (Aoki and Rosenberg, 1987). Second, there is a lot of "semi-horizontal" communications, with superordinates acting as mediators rather than authority figures. Senior Japanese managers are said to be more experienced in and knowledgeable about the relationship between R&D and production, and are more favorably disposed towards investing in R&D (Odagiri and Goto, 1993). This gives them technical credibility. Third, the engineering department has relatively more status in Japanese than in American firms, and it is usually collocated with manufacturing. This organization results in a smoother product development process (flexible phases that are "intermeshed" and "overlapped" (Aoki, 1990)) and tighter design/manufacturing linkages, with greater useful feedback from manufacturing to development. Fourth, Japanese companies are strong in "analytic design"--modifications of designs already within the state of the art to accomplish new tasks. Fifth, closer coupling between engineering and "research" is accomplished through temporary personnel transfers; personnel involved in design stay with the product through manufacturing (Aoki, 1990). Finally, the close vertical linkages between Japanese suppliers and Japanese customers facilitate knowledge sharing and problem solving (Asanuma, 1989), which are necessary to innovation and commercialization.
This system has its strengths and weaknesses relative to the stylized facts about American innovative capabilities. The close contacts between R&D and other units in Japanese companies mean that "research undertaken will be more commercially relevant and the introduction of a new product into the production and marketing stages will be faster" (Odagiri and Goto, 1993: 107; Mansfield, 1988). The Japanese firm is said to be stronger in testing and redesign and better at small product modifications based on careful engineering (Imai et al., 1985), making it well adapted for innovations along a predictable technological trajectory. Relative to its American competitors, however, the links between scientific research and invention are weaker in Japan. The Japanese firm is less flexible in gathering specialists from different disciplines and so probably weaker in interdisciplinary research, and it is thought weaker in utilizing the long market feedback loops for implementation of radically new products. While it is adept at "the better known, closer-at-hand technologies," it is less suited to choose "bolder, riskier, and more visionary technologies" that lead to pivotal new products or process technologies (Okimoto and Nishi, 1994: 201).
Given these characteristics, we would expect that Japanese firms would perform better in situations where continuous incremental improvements, tight engineering tolerances, and manufacturing strength are the bases of competitive success, and less well in segments incorporating radical or unproven technologies that rely on more fundamental technical research. Innovations in Disk Drives
Rapid product or process innovation is a necessary condition for competitive success in all high technology industries. This is especially true for disk drives. Although technological advances in semiconductors have generally been credited for most of the price/performance improvements in computers, fewer people are aware that progress in disk drive speed and capacity kept pace. The amount of data that can be stored on a square inch of a disk grew almost 30% per year between 1957 and 1990, and since 1990 has increased about 60% a year. Data transfer rates have increased while average access times have fallen. Between 1980 and 1995, the price per megabyte of storage fell at an annual rate of 40% (CRN, 1997). All of these advances were accomplished on increasingly smaller disk drives. Since the 1970s, the disk drive's size, called form factor, has continued to decrease, from 14 inches in the 1970s to 5.25 and 3.5 inches in the 1980s. These "architectural innovations" (Henderson and Clark, 1990) challenged the competencies of incumbent disk drive companies, and the inability of firms to make the transition to smaller form factors has been cited as a central reason behind firm failure (Christensen and Bower, 1996). Can U.S. success in the industry be explained by the greater innovative capabilities of its firms?
It is true that one American company -- IBM -- served as the technological fountainhead for the industry, spawned many of its firms, and continues to demonstrate remarkable technological leadership.5 As Table 1 shows, IBM established the industry and introduced many key innovations: the first removable disk pack drive, the Winchester standard, the first drive with ferrite, thin film, and magneto-resistive (MR) heads, and the first 8-inch disk drive that proliferated with minicomputers. More than any other institution, IBM displayed engineering brilliance in overcoming critical technical constraints. The 1301 disk drive in particular pioneered in areas that led to follow-on improvements in storage density and access times. Nonetheless, the co-evolution of technology and competition in the HDD industry confounds the conventional wisdom about differences in Japanese and American innovative capability in two ways.
First, Japanese firms have been stronger than theory would predict in technologically advanced new products. One measure of this is the shift to different form factors. The movement from 14-inch to 8-inch to 5.25-inch disk drives and then to 3.5 and 2.5-inch drives represent architecturally distinct product generations. While IBM was first to introduce the 14-inch and 8-inch form factors, since then young entrepreneurial firms, rather than older incumbents, pioneered architectural innovations.
One might thus expect Japanese firms to lag behind their American competitors in the shift to new form factors. At first glance this seems to be the case. Table 2 lists the order of entry into 5.25-inch, 3.5-inch and 2.5-inch disk drives. Eight of the first ten companies to introduce 5.25-inch drives were American, led by Seagate in July, 1980. The other two were European: Rodime and Olivetti. Three additional firms shipped 5.25-inch drives by the end of 1981, but only one of those, Nippon Peripherals, was a Japanese firm. By the end of 1982, 13 more firms had begun shipping 5.25-inch drives, seven of these were Japanese firms, including Fujitsu and Hitachi. In 1983, 14 more firms made the shift to 5.25-inch drives, five of which were Japanese. Thus, from 1980-1993 only 13 of 41 HDD firms that shipped 5.25-inch drives were Japanese and these firms were a year or more behind in introducing 5.25-inch drives. However, among firms that still made disk drives through the end of 1996, the Japanese were quicker than most of their U.S. counterparts in making the move to 5.25-inch drives. Again Seagate was the first to introduce 5.25-inch drives, but Fujitsu, Hitachi and NEC shipped 5.25-inch drives before or concurrently with Quantum, Maxtor, Micropolis and IBM. (Western Digital, another leader in 1996, did not make disk drives until 1988 when it acquired Tandon's hard disk drive operations.)
(Table 2 omitted)
A similar story can be told regarding the shift to the 3.5-inch form factor. In this case the first HDD firm to ship 3.5-inch drives was Rodime, a European firm, in September, 1983. However, the next three firms to ship were U.S. firms: Microcomputer Memories, Microscience International and MiniScribe, all of which shipped in 1984. The first Japanese firms to ship a 3.5-inch drive was Nippon Peripherals in February, 1985. All four of these early innovators have since exited the industry. The first surviving HDD firm (until 1996) to ship 3.5-inch drives was Hewlett-Packard in March of 1985, but Hitachi, Fuji Electric, NEC, and Fujitsu followed close behind. More importantly, as occurred with the transition to 5.25-inch drives, these Japanese firms were quicker to make the shift to 3.5-inch drives than every U.S. firm that is a leader today. IBM did not introduce 3.5-inch drives until May, 1986. Seagate first shipped 3.5-inch drives during the third quarter of 1987, the same date the Japanese firm Toshiba began shipping and one year after the Japanese firm Seiko Epson had begun shipping 3.5-inch drives. Quantum and Maxtor did not make the move into 3.5-inch drives until 1988 and Micropolis until 1991. Thus, while the Japanese were behind the first movers into 3.5-inch and 5.25-inch drives, most of whom were from the U.S., they introduced these new form factors before what are today's most successful American HDD manufacturers. In the shift to the 2.5-inch form factor the Japanese firm JVC was among the first movers. Other Japanese firms were no slower at adopting the new form factor than surviving American firms.
Japanese firms have also been among the leaders in incorporating advanced technology in their disk drives, specifically the new, thin film magneto-resistive (MR) recording heads. MR heads are designed to read media with very high recording densities and are the reason that growth in areal density jumped up to a 60% annual rate since 1990. Unlike previous head technologies which function like small electro-magnets, MR heads use a different approach for reading data--a thin stripe of magneto-resistive material is deposited on the head which senses the strength of the magnetic patterns on the disk and creates corresponding electrical pulses. However, the MR stripe cannot write data and so a traditional thin film component must be placed on the head along side the MR strip (Quantum, 1997; EBN, 1996; EET, 1997). Because the switch to MR heads requires corresponding changes in media and electronics technologies, and because they are very difficult to make, many companies have been slow to commit resources to the new technology, choosing instead to try and increase capacity through conventional technologies.
Stylized notions of American and Japanese innovative capabilities suggest that U.S. firms would be more likely to move first into smaller market segments with more sophisticated technology while abandoning to firms from other countries the market segments dominated by older technology. In disk drives, we would thus expect American firms to lead the way into MR technology. Similarly, some would argue that Japanese drive designers would push technological improvements using the inductive thin-film technology with which they are familiar rather than make the complex shift to MR heads. In one sense this is true: IBM invented MR technology and entered the market with it almost three years before the nearest competitor. Yet, three of the next six companies to introduce disk drives with MR heads were Japanese companies (Table 3).
Moreover, Japanese technological strength is further revealed by looking at the areal density of a disk drive. Areal density (the amount of data that can be squeezed onto a given space of a disk) encapsulates in one picture a company's ability to bring together head, media and flying height technologies and is a major feature of the technology race in hard disk drives. As Table 4, shows, the Japanese are also among the leaders in areal density. The table ranks firms according to the disk drive with the highest areal density each offers as of May, 1997 (Disk/Trend, 1997). Once again, though IBM is clearly far ahead, three of the top five are Japanese. This ranking changes frequently, as the newest product to the market seems to embody the highest areal density, but the illustration nonetheless demonstrates Japanese innovativeness.6
A second exception to the conventional wisdom is that Japanese firms have also been weaker than theory would predict: within a given form factor, technology has evolved in ways that should have given the Japanese an advantage. All companies have technology roadmaps, technological progress has moved along well known paths, and firms have solved well known puzzles within form factors, especially in the technological development of the current generation of disk drives employing inductive thin film heads and disks (although MR heads are now displacing inductive). IBM was the first company to ship disk drives with thin film inductive heads in 1979; drives with thin film media appeared four years later. Innovations in areal density over the next decade involved improvements to these two increasingly understood technologies. Japanese firms would thus have been expected to advance more quickly along this technological trajectory while simultaneously obtaining cost advantages through more efficient manufacturing. But the reverse is in fact true. American firms have dominated this largest segment of the disk drive market and are making interesting adaptations to the basic technology.7 In this way, American firms have been most responsible for extending the life of inductive head technology, which innovation theory would not predict.
Overall, there is little evidence that the Japanese are less innovative than successful American companies according to these key measures. They have not been far behind their U.S. competitors on the technological frontier, and they even introduced advanced new products before leading U.S. companies. While innovation has been necessary for all companies to stay in the game, it has not been a sufficient condition.8
III. FORM OF INDUSTRIAL ORGANIZATION
Many scholars argue that the Japanese form of industrial organization, with its complex interfirm relations, may have distinct advantages (Aoki, 1988; Gerlach, 1992; Teece, 1992). Although the evidence comes almost entirely from the automobile industry, the general claim is that Japanese firms are less vertically integrated than their American counterparts and maintain closer relationships with suppliers, often through some equity holdings (Aoki, 1990; Hill, 1995; Dertouzos et al. 1989). Was there, ironically, something about the American form of industrial organization that sustained U.S. advantage in the HDD industry? Backward Integration: Components and HDD Assembly
The basic issue is whether Japanese and American disk drive firms practiced different methods of organizing production and delivery. I focus on four of the most important disk drive components: the recording heads that read and write the data; the disk to which data is written and stored; the motor used to rotate the disk; and the semiconductors that control the drive and manage the flow of information between it and the computer. I also consider the extent of contract assembly of disk drives.
In contrast to the microcomputer industry (e.g., Langlois, 1992), vertical integration has been an important, though not universally implemented, strategy for HDD firms. Table 5 compares the degree of backward integration into components for a sample of 28 firms. The sample includes both surviving firms and firms which exited the industry (of the 28 firms, 16 were still producing HDDs in 1995). Ten of the firms are Japanese, one is Canadian and the remainder are from the United States. (The Canadian firm, Northern Telecom, is listed with U.S. firms because its disk drive operations were in the U.S. as a result of acquisitions of two American companies.) The degree of backwards integration across these firms was compared at four different points: 1983, 1987, 1991, and 1995. The combined global market share of the 28 firms was 85% in 1983, 91% in 1987, 98% in 1991, and 99% in 1995. The data show that backwards integration has clearly been an important strategy in the industry and one that has become more prevalent over time. In 1983 75% of the HDD firms in our sample were vertically integrated in one or more key component. This number increased to 91% in 1991 and 94% in 1995.
(Table 5 omitted)
In-house assembly of HDDs has also been the dominant model in the industry, regardless of nationality. While contract assembly relationships have been common, they have not accounted for a large share of total production. Table 6 summarizes all known the contract assembly relationships since 1976. Of the more than 100 firms that shipped disk drives under their brand names during this period, only 20 used contract assemblers. The majority of firms that engaged contract assemblers did so because they were small and with limited resources, or competed in niche segments. The important exceptions to the general model are Quantum and IBM. Probably 30% of all disk drives shipped in 1996 were done on a contract basis by for these two firms. All of Quantum's disk drives are assembled by Matsushita-Kotobuki-Electronics. IBM has used its former English disk drive subsidiary, spun off in 1994 and now called Xyratex, to assemble drives. In 1997 Xyratex assembled about 2 million drives for IBM. IBM has also used a Thai subcontractor, Saha Union, to assemble 2.5-inch and 3.5-inch drives designed by its Japanese disk drive operation in Fujisawa.
However, when viewed historically contract assembly has played a relatively insignificant role quantitatively. With the exception of the Quantum-MKE connection and IBM's recent use of Saha Union and Xyratex, the vast majority of units shipped have come from HDD companies' own factories. All leading firms except Quantum have maintained a strong manufacturing capability. IBM, for example, no longer sources drives from Xyratex and is building a plant in Thailand to make its own. Moreover, most of those performing contract work have themselves been HDD firms rather than specialist assemblers. Several of those who contracted work to others also engaged in contract assembly themselves, including IBM. Interestingly, many American disk drive firms performed contract assembly. Of the companies that have contracted all assembly to others, only Quantum and tiny Nomai (France) survive.
Despite this dominant theme of in-house assembly and the general trend towards backwards integration into components, there is certainly interfirm variation in the levels of integration. However, there does not appear to be any systematic difference in backward integration between leading Japanese and American firms. There are highly integrated firms on both sides of the Pacific: Seagate, IBM, Fujitsu and Hitachi make virtually all of the key HDD components in-house (though they are not completely self-sufficient). There are also Japanese and American HDD firms that for a time only assembled HDDs and did not integrate upstream into ASICs, motors, heads or media, namely Kyocera and JTS. Those that never integrated into one of the four components were typically the smallest firms in the industry. Finally, the majority of firms in both countries lies somewhere between these two extremes, producing one or two key HDD components. If we focus only on integration into heads and media the story is much the same.
Moreover, the closest observable interfirm relationships appear to be between American and Japanese firms: Quantum and MKE, and Integral Peripherals and Fuji Electric, which owns a small share of Integral and assembled Integral drives intended for the Japanese market. Samsung also provided most of the initial $1 million in start-up financing for Comport (U.S.) and assembled all of Comport's drives. More recently, IBM has announced a contract assembly relationship with NEC, which will manufacture IBM drives later in their life cycles and use IBM components. At the component level, the top three independent media companies are Komag (U.S.), Fuji Electric (Japan) and Mitsubishi Chemical (Japan). Only Fuji Electric has an equity relationship with an HDD firm, holding a small percent of Fujitsu. Yet, Fuji Electric's largest customers are MKE/Quantum and Seagate Technology, not Fujitsu (TrendFocus, 1996a). For recording heads, Read-Rite (U.S.) TDK (Japan) and its Hong Kong subsidiary, SAE Magnetics, and Yamaha (Japan) are the three largest independent producers (TrendFocus, 1996b), and none has an equity relationship with any HDD producer. Forward Integration: Computer Systems
Is there a systematic difference between Japanese and American firms in forward integration into computer assembly? It is true that a view of the industry in 1997 reveals that virtually all of the surviving Japanese HDD firms make computers, and that none of the American firms save IBM do. However, this misses much of the evolution of the industry. Throughout the 1970s and 1980s many of the largest American HDD manufactures were computer makers: IBM, DEC, Burroughs (later Unisys), Control Data, and Hewlett-Packard. Conversely, there were numerous Japanese companies that began to make HDDs during the 1980s but did not make computers, such as Alps Electric, Otari Electric, TEAC, Tosoh and Fuji Electric.
Nonetheless, conventional wisdom in the industry is that the Japanese HDD industry has been at a competitive disadvantage due to its heavy reliance on captive sales. This dependence, it has been argued, tends to slow the speed and degree of innovation since captive drive makers may not be subject to the same competitive pressures as non-captive firms. An additional disadvantage captive makers face has been expressed by George Scalise, former Maxtor president and chief executive: "The history of computer systems manufacturers in most instances has been that building peripheral products like disk drives is not a core business that can generate volume and economies of scale to be cost competitive" (EN, 8/21/89). Finally, captive HDD manufacturers are said to find it difficult to sell drives to outside computer firms: "To enter into a large OEM relationship with a disk drive company means disclosing future computer plans. Most large OEM computer companies would prefer not to disclose those plans with a competitor" (EN, 8/21/89).
While the captive market has indeed been whittled away, the perception of Japanese companies as over-reliant on internal sales can be challenged on at least two fronts. First, captive sales have not made Japanese firms notably slower to innovate than successful American HDD firms. This was discussed in some detail in the preceding section. Second, the Japanese HDD industry has not been more reliant on captive sales than the Americans. Historically there has been little difference between the two nations; if anything the opposite has been true. Figure 3 compares the share of captive revenues in total revenue for American and Japanese industry. It indicates that between 1983 and 1993 captive sales accounted for a slightly higher percentage of total HDD revenue among American firms than Japanese. Only for the periods prior to 1983 and since 1994 is it possible to argue that the Japanese have been more reliant on captive sales than the Americans. In either case, the differences are too small to provide a compelling account of competitive advantage.
In summary, it is unlikely that variation in mode of organizing among U.S. and Japanese firms is responsible for the sustained U.S. dominance in the HDD industry. Until very recently, the industry has been characterized by a great amount of diversity in organizing, with specialists coexisting alongside vertically integrated firms. During the last decade, both Japanese and American firms have increasingly relied on backward vertical integration while reducing their reliance on captive sales to an in-house computer business. In fact, their integration strategies have been remarkably similar. IV. GLOBALIZATION OF ASSEMBLY9
While degree of vertical integration, innovation, and path dependence in the disk drive industry are useful for understanding a part of the competitive advantage of U.S. firms, a closer look at the data in the HDD industry revealed problems with these explanations. This section argues that an important ingredient in U.S. dominance in the disk drive industry has been the globalization of assembly. While innovation is of course critical, companies have to be equally effective at transferring new products quickly into volume production while keeping costs down in the face of rapid price erosion. The president of Seagate, the world's largest disk drive company, says that his company is happy to be a follower rather than an innovator, and to outproduce its competitors (SoS, 1996). The centerpiece of this production strategy has been overseas assembly.
In general, American firms have not been known for their manufacturing prowess. Yet U.S. disk drive companies have demonstrated that this generalization does not hold for all industries. American disk drive companies competed squarely in, and came to dominate, the low-margin, high-volume segments: the price and capacity points most in demand by users of personal computers. Judged by what scholars have had to say about the manufacturing failures of American firms in other industries, this is an extraordinary accomplishment. The primary way the American HDD industry achieved this was by being the first as a group to shift assembly offshore to lower-cost locations, where it quickly constituted an entire value chain of activities and learned to manage the physical separation of development from assembly. If Silicon Valley is a nonce term for innovation, non-Japan East Asia has come to signify low-cost assembly and logistics management. This section describes the movement of the industry overseas and how it furthered U.S. competitiveness. Home-Based Assembly: 1956-1982
In the 1960s and 1970s, before the introduction of the 5.25-inch disk drive, assembly of disk drives by American firms occurred primarily in Silicon Valley, the Los Angeles area, Minneapolis, Oklahoma City, and the region around Boston. Some HDD firms that were vertically integrated into computers also assembled disk drives in Europe. IBM manufactured drives in Germany, England and Italy; Control Data manufactured in England and had a joint venture in Romania; and Burroughs had operations in Scotland. Burroughs also had assembly operations outside the U.S. and Europe in Brazil, Mexico and Canada. Japanese and European companies, with the exception of Germany's BASF, which had an operation in Silicon Valley, all assembled in their home countries. Among the firms that located assembly outside their home market, the principal rationale was proximity to customers: placing assembly in those markets where governments, banks and insurance companies -- the primary customers for their computer systems -- were likely to look favorably upon firms committed to local assembly of systems and peripherals.
In 1982 and 1983 Seagate, Computer Memories, Ampex and Tandon, all independent producers, became the first companies to move HDD assembly to locations for reasons other than access to host country markets. These firms began to assemble drives in what they saw as the best location from a cost standpoint, selecting low-wage areas in Asia, particularly Singapore. By the end of 1983 assembly was scattered geographically, but still overwhelmingly in the countries where the firm had it headquarters. As Table 7 shows, virtually all of the production of hard disk drives in 1983 was concentrated in two countries, the U.S. (72.3% of shipments) and Japan (12% of shipments). With almost 5% of global shipments, Europe produced more disk drives than all of Asia outside of Japan. In 1983 U.S. firms produced some 93% of their drives in the U.S. while Japanese firms produced all of theirs in Japan.
A Shift in the Center of Gravity: 1983-1990
The experiences of Seagate, Tandon and Computer Memories in S.E. Asia began to influence other American HDD firms. The perceived success of Seagate's Singapore facility, in particular, spurred several other HDD producers to adopt a similar cost-based siting strategy. Table 8 shows the movement of overseas disk drive assembly among firms headquartered in America, Japan, and elsewhere. Many American firms followed Seagate's lead and chose Singapore as their first overseas manufacturing site. In addition to Computer Memories and Tandon, both Maxtor and Miniscribe began to ship drives made in Singapore plants in 1984, followed by Micropolis (1986), Conner Peripherals (1987), Cybernex Advanced Storage Technology (1987). American HDD companies also opened overseas facilities in other low cost Asian locations such as Taiwan (Microscience 1987, Priam 1987), in addition to Ampex's investment in Hong Kong.
(Table 8 omitted)
In the span of just seven years, a dramatic change in the locus of assembly occurred. By 1990 Singapore was the world's largest producer of hard disk drives, accounting for 55% of global output, measured in shipments, with the rest of S.E. Asia accounting for only a percentage point more (see Table 9). As more firms located in S.E. Asia the supporting industries emerged in the region so that by 1990 three-fourths of the parts needed to produce a disk drive could be purchased in Asia (LAT, 1990; BT, 1993). The revealed global strategies of American and Japanese firms could not have been more different. By 1990, eight years after the first HDD was produced in Singapore, American firms assembled two-thirds of their disk drives in Southeast Asia. What began as a variation from the norm became a collective phenomenon. In contrast, Japanese companies assembled almost none there, and only 2% in the rest of Asia. Japanese companies instead continued to manufacture predominantly in Japan, where they produced 95% of their disk drives.
As a group, Japanese firms were clearly hesitant to abandon a strategy that appeared to be working up until the mid-1980s: exporting from Japan. In 1984, for example, TEAC Corp. was shipping almost 60% of its output to the U.S. Even as late as 1989, both Matsushita and Hitachi invested in Japanese manufacturing capability for 3.5-inch drives, judging that applying more automation to drive assembly would enable them to overcome the otherwise higher costs of manufacture in Japan. However, as the yen strengthened against the dollar and they turned their attention abroad, the U.S., not Asia, was the site of their first overseas manufacturing investments. Fujitsu opened a U.S. plant in 1986, NEC followed in 1987, and Toshiba entered in 1992. At one point, Fujitsu reportedly intended to manufacture nearly all of its disk drives in the United States (CW, 1985). Toshiba explained that its strategy in HDDs was proximity to the market--to respond to market needs more effectively by designing and building products closer to the markets where they are sold (LAT, 1991). Nor was S.E. Asia the chosen strategy for new Japanese entrants. After they entered in 1985, Fuji Electric, JVC, Seiko Epson, and Alps Electric all confined their manufacturing to Japan; only in 1997 did Fuji Electric begin to make drives overseas. Strategic Convergence: 1990-1996
For high-volume, low-priced and low-to-medium capacity drives, where cutting costs was paramount, S.E. Asia was clearly the location of choice for American companies, and their strategy increasingly confined the Japanese to niches in the high-capacity segments. This was a surprising switch since high-volume, low-cost manufacturing is an area where the Japanese traditionally excel. Eventually the success of the American firms impelled the Japanese to follow with investments in S.E. Asia. Between 1991, when Fujitsu began production in Thailand, and 1996, all the principal Japanese HDD firms gradually shifted manufacturing to S.E. Asia, principally the Philippines.
By 1991 Fujitsu had reached the maximum capacity of its Yamagata, Japan HDD facility and was searching for ways to expand HDD production capacity (CI 1/3/92). In addition to expanding the Japanese facilities and investing in the U.S., as it had done in the past, Fujitsu decided to manufacture drives in Thailand and retooled an existing recording heads facility for production of low capacity 3.5-inch drives (IDC 2/28/91). Production stayed at low levels until 1993 when the appreciation of the yen forced Fujitsu to move a large share of its manufacturing to Thailand. By the end of 1995 Fujitsu was doing nearly all of its volume manufacturing at the Thailand facility and a new facility in the Philippines. The President and CEO of Fujitsu Computer Products of America cited the move to S.E. Asia as one of the prime factors behind the company's rapid growth in 1996. Fujitsu doubled its worldwide hard drive revenues for 1996 and experienced a 123 percent growth in shipments (compared with overall 1996 market growth of 17 percent) (BWI 6/17/97).
NEC, Hitachi, and Toshiba soon joined Fujitsu overseas. NEC completed its own HDD facility in the Philippines in 1995 and increased its off-shore production to 75% of total HDD output (COM 10/9/95). Hitachi also made its first HDD investment in the Philippines in 1995 and now has 90% of its 2.5-inch disk drive production in the Philippines, and will soon make all 3.5-inch drives there as well.
By 1995, over 64% of the world's disk drives were produced in S.E. Asia generating nearly 61% of the industry's revenue (Table 10). HDD production in the U.S. fell to below 5% of world shipments which generated less than 9% of world revenues, while production in Japan fell to 15.7% of shipments and 13.3% of revenue. By 1995 Japanese firms had greatly increased their presence in S.E. Asia, producing nearly 55% of their HDDs in the region. Virtually all of the remaining drive production for Japanese firms was still located in Japan--45% compared with 18% still located in the U.S. or Japan for U.S. firms.10 By the mid-1990s, then, the geographic distribution of Japanese assembly had begun to resemble that of their American competitors.
The Value Chain Follows
Through continued investment in the region, nearly every part of the HDD value chain is now produced in Southeast Asia in some quantity, reinforcing its preeminence as center of HDD and components production.11 Seagate offers a good illustration. In almost every year its since its initial investment in 1982, Seagate has reinvested in Singapore--upgrading existing facilities or building new facilities. The largest investments include a $56 million investment in 1988, a $100 million investment in 1992 and a $200 million investment in 1994. Seagate has also invested heavily in Thailand and Malaysia in upstream activities like motors, heads, and printed circuit board assemblies (PCBA), and has recently opened plants in Indonesia (PCBA), China (HDDs) and the Philippines (labor-intensive head assembly). Today it is the largest private employer in both Singapore and Thailand.
Independent manufacturers of media and heads have also moved into the region, further reinforcing it as the focus of the industry's global strategy. The first head-maker to invest in the region was Applied Magnetics which opened a plant in Singapore in 1983. Read-Rite, another head company, opened or acquired facilities in Thailand and Malaysia in 1991. The first media maker to locate production outside the U.S. or Japan was Domain Technology, Inc., which began volume production in Singapore in 1988. Komag and StorMedia invested in Malaysia and Singapore, respectively, in 1993 and 1995. The first investment in Southeast Asia by a Japanese media company was Hoya Media's Singapore plant in 1996. As of 1995 nearly 70 percent of the firms which make heads or head assemblies had plants in Asia (excluding Japan) while 36 percent of the firms had plants in Southeast Asia. Among media producers, 81 percent had plants in Asia in 1995 while 38 percent were producing in Southeast Asia.
American HDD assemblers initiated the move to S.E. Asia and much of the value chain followed. By 1995, more than 60% of global employment in the HDD industry, including upstream activities, was in Asia outside of Japan (Gourevitch et al., 1997).12 After a decade of investment by both multinationals and local supplier firms, low-cost Asia has become the region of choice for the HDD industry. The technical and economic imperatives of the industry ultimately led to a convergence of American and Japanese strategic posture. Global Strategy and National Advantage
Despite recent Japanese movements into S.E. Asia, American industry was able to sustain its advantage by being the first to implement this global strategy. Although Japanese industry increased its proportion of total assembly in S..E Asia to nearly 55% by 1995, it was still below the almost 67% produced there by U.S. industry.
How, exactly, did this strategy confer an advantage on American firms? According to industry participants, American industry's early move into S.E. Asia gave it the time to establish regional manufacturing, secure complementary assets and move down the learning curve at a time the competitive logic of the industry dramatically shifted to low-cost, high-volume manufacturing. An industry consultant who set up a number of overseas production facilities explained the benefits of making disk drives in Southeast Asia in this way: "While loaded labor cost is typically one-quarter of the U.S. equivalent, material cost can be 30 to 45 per cent less than in the U.S. But since the cost of material can account for as much as 80 per cent for some peripheral products, the real savings are achieved through local sourcing for materials rather than in savings of labor costs. Overall, cost savings of 30 to 40 per cent can be achieved in making peripherals in Southeast Asia versus the U.S." (EN, October, 1984). Other benefits of assembly in Southeast Asia included lower overhead costs, government incentives, faster investment approvals, and a less costly but developed infrastructure such as precision machinery, die casting and a pool of skilled personnel in process engineering. Disk drive firms reportedly cut costs by 30% by moving production to Singapore (FW, February 24, 1987).
These cost savings coincided with the emergence of the 3.5-inch disk drive to generate enormous advantages in that critical high-volume market. As is shown in Figure 4, the demand for 3.5-inch disk drives exploded in the late 1980s, and, as Table 11 indicates, the largest part of the 3.5-inch market was for non-captive sales. It was America's success in this market that extended its advantage. During this period, America's dominance was led by its independent HDD firms--Seagate, Conner, Quantum, Maxtor and Western Digital--and was dependent on their ability to ramp-up low-cost, high-volume production in S.E. Asia where, with the exception of Quantum, they assembled the overwhelming majority of their 3.5-inch drives.
It took three years for the U.S. industry to claim more than 50% of this market. Although each of these firms trailed Hitachi, Fujitsu, and NEC in the introduction of this form factor, this strategy allowed American industry to claim 90% of the non-captive market by 1991. Operating from Southeast Asia, U.S. companies put tremendous pressure on higher cost Japanese manufacturers even in the Japanese market. While the retail price in Japan of a 20mb drive made by Japanese producers cost about 200,000 yen, drives made by Singapore-based U.S. firms were beginning to be sold at a retail price of under 70,000 yen (COM, April 1988). According to the business press, the cost structure of Japanese manufacturing was simply not competitive with that of American industry in the product segments most in demand.
This early move overseas also gave U.S. firms time to learn how to transfer technology internationally. In the course of the industry's evolution, pockets of technical sophistication developed in the United States (the Los Angeles area, Silicon Valley, Minnesota, Colorado, and, to a lesser extent, the Boston region), Japan, and Europe (the U.K., the Netherlands, France and Germany). Through industry consolidations, surviving firms have found themselves in possession of R&D assets in more than one location. Table 12 lists all HDD firms in operation as of mid-1997 and the location of product development and volume manufacturing for HDDs. (Locations for component development and manufacturing in vertically integrated firms are omitted.) One of the most dramatic aspects of the HDD industry is that, with very few exceptions, product development is geographically separated from volume manufacturing and, among the leaders, by great distances.13
Typically, companies conduct pilot production proximate to product development because of the greater risk with design during initial assembly. Companies also form product transfer teams consisting of product developers and process engineers from both the U.S. and the volume manufacturing facility. Today, a product transfer team might be as big as 40 people and stay with the product from pilot through ramp-up overseas. Then the manufacturing team takes over responsibility for volume assembly.
Seagate's Malaysian facility, for example, has had a good experience with product transfer and can ramp up quickly.14 Six to eight engineers in the U.S. write the code, do mechanical design and testing, and work with the process people in the U.S. to stabilize yields during pilot production, which might involve as many as 10,000 drives if the product were especially advanced. A few weeks before the transfer to Malaysia, quality, operations, and lead operators in Malaysia go the U.S. to prepare for transfer. Then 6-10 people from the U.S. team come for 3-5 weeks to ensure a good start. The product release date is typically met because, according to the plant manager, "transfer is almost routine now, very smooth." The transfer process involving its Thai facility is organized slightly differently but is also effective: it makes less sophisticated products and hosts fewer U.S. engineers during transfer, but sends more staff to the U.S. prior to transfer because its workforce is less skilled.
Western Digital uses a formal new product introduction (NPI) process that allows it to achieve 92% yields in overseas assembly.15 WD has a NPI group to manage the process and claims this gives the company an advantage in time to market, time to volume and time to high yields. In product design, WD uses a typical "gating" process common to well-managed high technology firms. A product concept, and then product, needs to pass certain gates on its way to manufacturing. These are milestones that have to be met at each step. Design, engineering-level build, test and tooling build up take place in the U.S. because there are still bugs that need fixing. Concurrent engineering is occurring in Asia where production level equipment is introduced and components are chosen. Transfer teams to and from Asia then ensure a smooth product transition. The transfer team from Asia (the Asia NPI organization) visits San Jose to work on the pilot line to "wring out" the process, while the U.S. team stays with the product all the way to ensure manufacturability. The transfer process involves 40 people for 30 days.
An important reason for the success of this model is that both the technology and the assembly process are better understood than they were 15 years ago, new products are increasingly designed so as not to disrupt existing manufacturing processes, and computer information systems lower the costs of long distance management. Although there is some variation among firms in their ability to minimize changes to products that might otherwise require substantial changes in tooling for the assembly process, companies try to maintain substantial commonality in components across products. Western Digital, generally thought of as the leader in this regard, has 70-80% commonality between products.
At the same time that companies have accumulated skills in design, transfer, and ramp-up, the quality of the infrastructure in Singapore and Malaysia has facilitated technology transfer and rapid ramp-up to volume manufacture. The disk drive industry has developed a large base of skilled professionals in the region with specialized industry knowledge, and the Malaysian and Singaporean governments have been aggressive in offering complementary services, such as rapid investment approvals, access to land, and labor training programs. As a consequence, there appears to be little lost in product yields or volume output, but considerable cost savings. As product cycles shorten, ramping up has become even faster in Asia. In 1995, Western Digital ramped up production from zero to 750,000 units within three months (CRN, 1995). In 1996, Quantum/MKE went from zero to 7 million disk drives in nine months (NST, 1996). Moreover, overseas assembly does not appear to cost companies in terms of yield. When IBM shifted production from California to Singapore, it not only ramped up quickly (from October to December), but did so with no loss of yield. Like other drive companies, IBM could find all the engineering and managerial skills it needed in Singapore, and the Singaporean government facilitated the move by approving the investment quickly and even leased IBM a plant that the government had specifically prepared for disk drive assembly. According to one IBM manager, the company could ramp to volume manufacturing in Singapore faster than anywhere else in the world, including the U.S.
This is a surprising twist: Japanese and European firms were early to market with an innovation but were ultimately squeezed by the price competition brought to bear by American firms who have excelled not only at volume manufacture but at the international transfer of technology. Coordinating technical activities with volume manufacturing across national boundaries has become standard practice for the industry. The model for American firms is design and pilot production in the United States, fast ramp-up in Singapore and Malaysia, and matured products and process transferred out to Thailand or China. The system has become routinized, and other than niche players, firms that have not adopt this organizational model, or have executed it poorly, have exited the industry. The timing, direction, and scope of globalization thus extended the leadership of America's disk drive industry by enabling it to move down the learning curve in overseas assembly and to accumulate the skills in managing the long-distance product development/assembly relationship.. V. CONCLUSION
Path dependence, industrial organization and innovation all contributed to American success in the disk drive industry. The American HDD industry was built by successful computer firms, which enabled the industry to achieve an early lead over European and Japanese drive manufacturers. In addition, product development capabilities and some degree of vertical integration have been necessary conditions for industrial performance. Yet, while each of these contributed to America's initial industrial advantage, taken together they are insufficient in explaining the ability of American firms to sustain their dominance. Differences among American and Japanese along these dimensions do not appear strong enough to explain the persistence of American leadership in the industry: Japanese companies have been innovative in important ways, and American and Japanese companies exhibit very similar levels of vertical integration. The principal contribution of this paper is to suggest that a potentially important yet overlooked variable in studies of national industrial advantage may be the scope, timing and direction of an industry's overseas manufacturing operations.
By being the first to shift assembly off-shore, American firms were able to learn the organizational technology of international coordination and production. While their activities were dispersed, they were at the same time concentrated in key regions: research and development concentrated in the U.S.; labor-intensive assembly in low-cost Asia; and somewhat more skilled assembly activities in Singapore (Gourevitch et al., 1997). American firms combined the benefits of low-cost high-volume assembly with sophisticated management of these value networks. It seems clear that HDD firms have learned how to manage international operations and to coordinate home-based product development with foreign assembly. Innovative firms that failed to shift assembly abroad exited the industry, or else claimed imperceptible shares of the market. Those that survived, including the largest and most diversified companies like IBM, Fujitsu and Hitachi, were ultimately pressured to do the same.
The history of the disk drive industry also differs from other high technology industries in additional ways. First, the American HDD industry excels at manufacturing. This is contrary to what researchers have observed in other industries, where Japanese firms are leaders in manufacturing. Early reports on the disk drive industry by the business press also embraced this prevailing view: "Once in production, a disk drive is basically a commodity product that must be assembled as quickly and as cheaply as possible -- something that the Japanese are expert at doing" (BW, 1984). Here we find that, if anything, they lagged behind American firms in their ability to ramp to volume manufacturing. Especially interesting is that the vast majority of assembly was conducted in-house. Though companies have frequently resorted to contract manufacturing, such arrangements have played a small role quantitatively.
Second, the industry was largely ignored by the federal government and university departments during its first two and a half decades. Certainly, disk drive programs in private firms benefitted at least indirectly from federal monies targeted for computers and semiconductors. Yet technical progress in disk drives went largely unnoticed by those outside the industry and was achieved through heroic mechanical and materials engineering efforts in firms, especially in IBM, rather than publicly funded research. Moreover, unlike software (Mowery, 1996), where the federal government played a prominent role in developing computer science as an academic field, in data storage the private sector initiated the establishment of academic programs specifically for magnetic recording, though the federal government stepped forward with critical funding. These programs also emerged much later than those targeted at the computer, semiconductor, and software industries. ACKNOWLEDGEMENTS
This research was supported by the Alfred P. Sloan Foundation, grant numbers 95-6-13 and 97-1-10. I am grateful to Allen Hicken, John Richards, Peter Gourevitch, Roger Bohn, Frank Mayadas and David Mowery for careful and insightful comments on an earlier draft. For assistance with data collection and compilation, I want to thank Allen Hicken. I also thank James Porter, President of Disk/Trend, Inc., who not only reviewed this paper but has generously shared his data, time and knowledge about the disk drive industry. Mark Geenen, President of TrendFOCUS, Inc. kindly provided data on the media and heads segments of the industry.
- 1Note that the focus of this paper is organizations in an industry. There have been many studies of foreign direct investment behavior. But these have usually been at the national or sectoral levels, or even at the organizational level. Few have examined international investment behavior of a population of firms in an industry.2A major exception to this general tendency was the success of Control Data in selling to European computer manufacturers, and it claimed the bulk of the world's shipments of "non captive" drives in the 1960s and still almost half by the late 1970s. IBM's disk drives were solely for IBM computers.3Global computer market shares were calculated from the Datamation 100 for various years.4An important exception is the printer industry. While the U.S. lost the impact printer market, it has a huge lead in laser printers.5For a technical history of IBM's first 25 years of innovation in the industry, see Harker, et al (1981) and Stevens (1981).6Firms also competed in the desktop market in terms of "volumetric" density -- how much capacity one could cram into the slot allotted to the disk drive. One trick in mechanical design was the introduction of "half-high" disk drives in which more disks were stacked closer together. A company might be a leader in areal density (data on a disk) but a laggard in volumetric density. Some say that IBM did not understand this distinction. Unfortunately, systematic data to test this notion are unavailable. I thank Frank Mayadas for bringing this to my attention.7Improvements to inductive technology include "proximity" or virtual-contact heads. These involve significant enhancements to etched air bearing and transducer technologies.8It is important to note that I have not addressed the ability of firms to introduce successive generations of products. When product cycles are so short, firms face intense pressures to stay competitive in terms of capacity, performance, and interfaces. Keeping design teams together in such a pressurized environment is difficult. It is possible that American firms have been better at this than those from Japan and Europe. I hope to explore this possibility in a later paper.9This section draws heavily on McKendrick and Hicken (1997).10American firms also extended the global assembly strategy to low-cost areas of Europe between 1990 and 1995 (Ireland and Hungary).11The story of Southeast Asia's emergence as a center for disk drives is being researched by Stephan Haggard, Rick Doner, and Wong Poh Kam, three colleagues from the Data Storage Industry Globalization Project.12This percentage is actually understated because it does not capture employment associated with a few of the least expensive components going into a disk drive: base-plates, condensers, capacitors, screws, etc. These are sourced almost entirely from vendors in Southeast Asia.13The exceptions are for the very smallest disk drive firms. The last ten companies listed in Table 14 had less than one half of one percent of the global market in 1996.14This information was provided by the plant manager during the author's visit.
- 15This information was provided by a Western Digital vice president during the author's visit to WD's Malaysian facility.
- Anderson, Philip (1995). Microcomputer manufacturers. In Glenn R. Carroll and Michael T. Hannan (eds.), Organizations in Industry: Strategy, Structure and Selection: 37-58. New York: Oxford University Press.Aoki, Masahiko (1988). Information, Incentives and Bargaining in the Japanese Economy. New York: Cambridge University Press.Aoki, Masahiko (1990). Toward an economic model of the Japanese firm. Journal of Economic Literature, 28 (March): 1-27.Aoki, Masahiko and Nathan Rosenberg (1987). The Japanese firm as an innovating institution. CEPR Discussion Paper, No 106, Stanford University.Asanuma, Banri (1989). Manufacturer-supplier relationships in Japan and the concept of relation-specific skill. Journal of the Japanese and International Economies, 3: 1-30. BT (Business Times) (1993). Plugging into and Asian gold mine. May 3.BW (Business Week) (1984). The disk-drive boom has suppliers spinning. February 6.BWI (Business Wire) (1997). Fujitsu ranked the fastest growing hard drive manufacturer. June 17.Chandler, Alfred (1990). Scale and Scope: The Dynamics of Industrial Capitalism. Boston: Belknap Press.Christensen, Clayton M. and Joseph L. Bower (1996). Customer power, strategic investment, and the failure of leading firms. Strategic Management Journal 17: 197-218.CI (Computergram International) (1992). Thailand: Fujitsu slashed local production of small hard drives as prices collapse. January 3.CL (Corporate Location) (1997). The oriental express. January/February.CNN (Cable News Network) (1993). Seagate introduces first 9-gigabyte disk drive. Transcript # 533-2. November 4.COM (COMLINE Daily News Computers) (1995). NEC hard disk drive plant in Luzon now operational. October 9.CRN (Computer Reseller News) (1995). Western Digital's Burger Speaks Out On State Of Storage Industry. December 4.CRN (Computer Reseller News) (1997). Will emerging new digital technologies ever stop? February 24.CW (Computerworld) (1985). 'Made in Japan' tag penetrating components market; U.S.-Japan pacts create added equipment sales. December 9.CW (Computerworld) (1988). The checkered past of Computer Memories. October 31.Dertouzos, Michael L., Richard K. Lester, and Robert M. Solow (1989). Made in America: Regaining the Productive Edge. Cambridge, MA: MIT Press.Disk/Trend (various years). Disk/Trend Report: Rigid Disk Drives. Mountain View, CA: Disk/Trend, Inc.EB (Electronic Business) (1987). October 1.EBN (Electronic Buyer's News) (1996). MR technology may finally have its year. April 29.EET (Electronic Engineering Times) (1997). MR heads inch toward mainstream. August 26.EN (Electronic News) (1984). Peripheral manufacturers making move to SE Asia; cheaper labor, materials draw U.S. firms seeking to be more price-competitive. October 22.EN (Electronic News) (1989). Seagate/Imprimis deal forces industry shift. August 21.FT (Financial Times) (1988). Seagate sets sail for Singapore. June 1.FW (Financial World) (1987). Disk drive technology is changing so quickly that only a few U.S. companies can keep up with it. But those survivors should prosper -- and soon. February 24.Florida, Richard and Martin Kenney (1990). The breakthrough illusion: Corporate America's failure to move from innovation to mass production. New York: Basic Books. Gerlach, Michael L. (1992). The Japanese corporate network: A blockmodel analysis. Administrative Science Quarterly, 37: 105-139.Gourevitch, Peter, Roger Bohn and David McKendrick (1997). Who is us? The nationality of production in the hard disk drive industry. Report 97-01, The Data Storage Industry Globalization Project, Graduate School of International Relations and Pacific Studies, U.C. San Diego, March.Harker, J.M., D.W. Brede, R.E. Pattison, G.R. Santana, and L.G. Taft (1981). A quarter century of disk file innovation. IBM Journal of Research and Development 25(5): 677-689.Henderson, Rebecca, and Kim Clark (1990). Architectural innovation: The reconfiguration of existing product technologies and the failure of established firms. Administrative Science Quarterly, 35: 9-30.Hill, Charles W.L. (1995). National institutional structures, transaction cost economizing and competitive advantage: The case of Japan. Organization Science, 6: 119-131.Hollingsworth, J. Rogers, Phillipe C. Schmitter, and Wolfgang Streeck (1994). Governing Capitalist Economies: Performance and Control of Economic Sectors. New York: Oxford University Press.IDC (IDC Japan Report) (1991). Fujitsu will produce small hard disk drives in Thailand starting this summer. February 28.Imai, Ken-Ichi, Ikujiro Nonaka, and Hirotaka Takeuchi (1985). Managing the product development process: How Japanese companies learn and unlearn. In Kim Clark, et al. (eds.), The Uneasy Alliance: Managing the Productivity-Technology Dilemma. Boston: Harvard Business School Press.Kogut, Bruce (ed.) (1993). Country Competitiveness: Technology and the Organizing of Work. New York: Oxford University Press.Langlois, Richard N. (1992). External economies and economic progress: The case of the microcomputer industry. Business History Review, 66: 1-50.LAT (Los Angeles Times) (1990). Why American high-tech firm recruits in Asian ricefields. June 25.LAT (Los Angeles Times) (1991). Toshiba develops disk drive of high capacity. August 6.Mansfield, Edwin (1988). The speed and cost of industrial innovation in Japan and the United States: External vs. internal technology. Management Science, 34: 1157-1168.McKendrick, David and Allen Hicken (1997). Global strategy and population level learning in the hard disk drive industry. Report 97-05, The Data Storage Industry Globalization Project, U.C. San Diego.Mowery, David (1996). Spinning off and spinning on (?): the federal government role in the development of the US computer software industry. Research Policy, 25: 947-966.MR (Market Reports) (1991). Korea - disk storage devices. July 19.Nelson, Richard R. (1996), The evolution of comparative or competitive advantage: A preliminary report on a study. Industrial and Corporate Change, 5: 597-617.NKS (Nihon Keizai Shimbun) (1985). Electronics companies inactive in locating plants in Asia, other areas. January 1.NST (New Straits Times) (1996). Quantum makes new move in drive business. April 25.Nunnenkamp, Peter, Erich Gundlach, and Jamuna P. Agarwal (1994). Globalization of Production and Markets. Tubingen: J.C.B. Mohr.OCBJ (Orange County Business Journal) (1994). Western Digital's dramatic resurgence. March 14.Odagiri, Horiyuki and Akira Goto, 1993. The Japanese system of innovation: Past, present and future. In Richard R. Nelson (ed.), National Innovation Systems: 76-114. New York: Oxford University Press.Okimoto, Daniel and Y. Nishi (1994). R&D organization in Japanese and American semiconductor firms. In M. Aoki and R. Dore (eds.), The Japanese Firm: Sources of Competitive Strength: 178-208. Oxford: Oxford University Press. Porter, Michael E. (1990). The Competitive Advantage of Nations. New York: Free Press.Quantum (1997). Storage resource center on the web: www.quantum.com/src/storage_basics.SoS (Scotland on Sunday) (1996). Patents pay-off pending. September 29: 1.Stevens, L.D. (1981). The evolution of magnetic storage. IBM Journal of Research and Development 25(5): 663-675.Teece, David J. (1992). Competition, cooperation, and innovation: Organizational arrangements for regimes of rapid technological progress. Journal of Economic Behavior and Organization, 18: 1-25.Teece, David J. (1996). Firm specialization, industrial structure, and technological innovation. Journal of Economic Behavior and Organization, 31: 193-224.TrendFocus (1996a). Rigid Media Information Service. Annual Study. March. Palo Alto, CA: TrendFocus, Inc.TrendFocus (1996b). HDD Recording Head Information Service. Annual Study. March. Palo Alto, CA: TrendFocus, Inc.UN (1993). World Investment Directory, Volumes 3, 4 and 5. New York: United Nations.UNCTC (1992). The Determinants of Foreign Direct Investment. New York: United Nations.Upside (1995). Al Shugart. February: 68-75.
- Van de Ven, Andrew and Raghu Garud (1989). A framework for understanding the emergence of new industries. In Richard Rosenbloom and Robert Burgelman (eds.), Research on Technological Innovation, Management and Policy: 195-225. Greenwich, CT: JAI Press.