Who is Us?--The Nationality of Production in the Hard Disk Drive Industry

Peter Gourevitch
Roger E. Bohn
David McKendrick      
Report 97-01
March, 1997

The Information Storage Industry Center
Graduate School of International Relations and Pacific Studies
University of California
9500 Gilman Drive
La Jolla, CA 92093-0519
http://isic.ucsd.edu/

Copyright © 1997, University of California

University of California, San Diego

Funding for the Information Storage Industry Center is provided by the Alfred P. Sloan Foundation. To receive printed copy of this document, send an e-mail with your address to the Publications Coordinator at isic@ucsd.edu.

Table of Contents

  1. Executive Summary
  2. Who is Us?
    1. Where Do Disk Drives Come From? The Nationality of Production
      1. Domination by American Firms
      2. Location of Final Assembly
    2. The Value Chain
      1. Major Steps in the Value Chain
    3. Jobs and Wages: The Data
      1. Geographic Distribution of Employment
      2. Where are the Wages Paid?
        1. Geographic Distribution of Wages Paid
        2. U.S. Firms
        3. Summary of the Employment/Wage Discussion
      3. Labor Costs, Labor Skills, and Location
      4. Recent Movements in the Value Chain
    4. Drivers of Location Decisions: Factors Costs, Agglomeration, Regulation, Networks/Search Costs
      1. Factor Costs
        1. Labor costs
        2. Yields
        3. Land, energy, transportation, communications
        4. Capital costs.
      2. Agglomeration Effects
      3. Regulation and Public Policy
      4. Networks
    5. Further Trends
      1. Recent Trends
      2. Technology, Manufacturing, and Design
      3. The Nationality of Firms
      4. Public Policy
  3. Appendix
    1. Sources for 1995 Employment and Wages
      1. Hard Disk Drive Employment
      2. Media and Heads Employment
      3. Integrated Circuit Employment
      4. Equipment
    2. Wages and Salaries
      1. Hourly wages
      2. High-skilled salaries
    3. Calculating the Global Distribution of Wages Paid
American firms dominate the hard disk drive industry. Most drives are made by firms based in the US, but most of those drives are assembled outside the U.S. American firms have moved most of their jobs, in assembly and parts production, offshore over the last ten years. While many of the jobs have gone overseas, the bulk of wages paid by US firms remain in the U.S.

This paper investigates the geography of activity in the hard disk drive industry. In order to measure it, we constructed a database of plants and employment at all steps of the industry¹s value chain in our base year, 1995. We find that by several measures, most activity in the industry is overseas:
  • Only five percent of final assembly was in the U.S. Southeast Asia dominated with 64 percent of world production, based on units shipped. By the end of 1996, the final assembly figure fell to about 1%.
  • Only 20 percent of employment (headcount) was in the U.S. Although this higher than any other single country, 67 percent of employment was in Asia, including 41 percent in Southeast Asia.
And yet, these numbers by themselves severely understate the contribution of the industry to the American economy. By other measures, the industry remains strongly American:
  • American firms have about an 85 percent market share despite steady competition from technologically sophisticated Japanese firms.
  • We estimate that about 42 percent of all wages in the industry are paid in the U.S. and American owned firms pay about 62 percent of their wages in the U.S. Thus, while employment by American firms has shifted heavily overseas, a substantial portion of their total wage payments and value added remain here.
  • 66 percent of all workers in the industry are employed directly by American firms. This contrasts with many firms in other branches of electronics, which manufacture overseas by subcontracting to non-U.S. firms.
What shapes firms¹ decisions about the location of their activities? One important factor is the costs of labor at various skill levels. Lower skilled activities have moved to low wage countries, while higher skilled steps in the value chain such as R&D, remain in higher wage areas. However, the wage/skill equation cannot explain many aspects of firms¹ location decisions, such as the overwhelming tendency to concentrate in a few countries rather than in low cost areas around the world. Other factors are at work, which we classify into several categories: non-wage costs, agglomeration effects, government activities, and networking relationships.

We are continuing research on all of these issues in an attempt to better understand how firms in the industry deal with location decisions.

This report is part of the Information Storage Industry Center funded by the Alfred P. Sloan Foundation of New York. The project is being conducted at the Graduate School of International Relations and Pacific Studies (IR/PS), of the University of California at San Diego (UCSD). Roger Bohn and Peter Gourevitch are co-Principal Investigators; David McKendrick is Project Director. Other faculty researchers involved with the project include Stephan Haggard and Mike Bailey (UCSD); Richard Doner (Emory University); Wong Poh Kam (National University of Singapore); Dieter Ernst (Copenhagen Business School and UC Berkeley). Gigo Alampay, Kyle Eischen, Allen Hicken, Colin McIff, John Richards are graduate students at UCSD working on this project. Nikki Alexander (nlalexander@ucsd.edu) is the project's staff person.

02/28/98

Copyright © 1997 The Regents of the University of California


      What drives the geographic location of economic activity? As liberalization of the world's economy proceeds apace, this question grows in importance in the debates over the impact of trade upon countries and the various populations within them -- skilled and unskilled, rich and poor, high tech and low tech, commodity production and specialized production, the owners of capital, the sellers of labor power, etc. The location of economic activity drives employment, trade, capital flows, migration, and profits. Cities, states, countries and regions compete to attract business. Managers confront the challenge of balancing the varying claims of location for production, markets, taxation, regulation, transportation, and design. Understanding the political economy of location is thus increasingly central to contemporary issues of international relations, public policy, and management.2

      Four "families of interpretation" offer alternative explanations for the location decisions of firms: factor prices, agglomeration, regulation, and networking.3 The oldest of these, the factor price approach, predicts that firms seek the geographical location that provides the least expensive costs relevant to its production activity. The classic components of factor costs are land, labor, and capital. Labor tends to dominate the rhetoric about factor cost location discussions, but it is by no means the only variable, nor as we shall explore below, a transparently simple one.4 Notable examples of location that reflect this approach are the rapid movement of textile and toy manufacturing to China, the flight of shoes a few decades back from New England to the U.S. South, and the recent migration of consumer electronics production to Asia.5

      The agglomeration approach notes the benefits of location complementarities: firms are better being near other firms undertaking the same or related activities, even if these firms are competitors. Clustering around a geographical center concentrates information, markets, entrepreneurs, workers, and other resources. Going to a new location means paying "first-mover" costs.6 Thus, firms locate near suppliers upstream and customers downstream. Striking examples of this phenomenon are the proximity of car companies in Detroit, electronics in Silicon Valley, and hard disk drive manufacturing in Singapore. Path dependence plays a prominent role in discussions of agglomeration. In Krugman¹s recent book on geography and trade, for example, he attributes the modern rug industry around Dalton, Georgia, to the quilt made a century ago by a local resident as a gift to a friend; Michael Porter uses ceramics in Northern Italy to make the same point.7

      The regulation approach notes the importance of rules in structuring the incentives that bear on location: firms seek to overcome tariff walls, to escape costly regulations on environment, health, land use, labor and competition, or to take advantage of inducements in the form of taxes and subsidies.8 The large investment in Europe in the years preceding 1992 is one notable example of how local content rules influence investment choices. Quotas for textiles are another well known example; aviation services, telecommunications, and financial systems provide other cases.9

      The network approach stresses problems of information and coordination in economic activity, problems that can be reduced by informal linkages such as ethnicity, religion, and long-standing relationships. These linkages are particularly useful in managing trade and production across diverse cultural, political and regulatory frontiers where informal attachments help solve problems of trust, the weakness of courts or regulatory systems, and weak information about markets and producers. They may also be helpful in managing problems of trust and search costs within the U.S. or another country with strong courts. A noted example of this approach is the work on the overseas Chinese in East Asia.10 Social capital -- knowledge of people from ethnic or cultural or individual contacts -- provides an important foundation for economic interaction, overcoming problems caused by regulation and politics.11

      These alternative theories are not mutually exclusive. Moreover, it is unlikely that only one will explain all aspects of each location decision in an industry. One country may have cheap labor but unstable policies on discharging workers, enforcement of contracts and transfer of profits. In one region, cultural networks may overcome the limits of instability, while in another region their absence will fail to do so. The study of industrial policy could be seen as linking the first three explanatory approaches: government regulations structure the factory costs parameters that induce concentration leading to the agglomeration affect, locked in over time by path.12 A full account of location issues will have to explore all of these factors in one way or another, carefully noting where and in what ways each contributes to a full explanation of location decisions.

      This paper provides data relevant to an exploration of these issues. It looks at a specific industry: the hard disk drive (HDD) component of the data storage industry. The HDD industry has worldwide revenues of about $25 billion per year. It is one of the world's most dynamic industries, with product life cycles of less than 18 months and prices falling about 30% per year for more than a decade. At its inception, all aspects of the storage industry were located in the United States and the industry has historically been dominated by American firms. Today, virtually no hard disks drives are assembled here, and the industry has become one of the most globalized among all industrial sectors.

      Globalization of manufacturing is a well-known phenomenon. American and foreign firms have broken the process of production into a number of stages and located these stages around the world. Any manufactured product we buy is likely to have a complex provenance: it may say "Made in USA" (or Japan or Singapore), but much of what is inside the box will have been made, designed, or invented somewhere else. While we know globalization is occurring, we often don't know the precise details of what is done where. As a result, it becomes difficult to be sure just where the activities of globalization are being distributed and just what are the most effective managerial and policy strategies for taking advantage of the globalization process.

      Understanding location requires information from many sources and research techniques: publicly available company reports, studies by market analysts, economic journalism, interviews with top executives, field trips to factories, talks with civil servants, engineers, scientists and academic experts, official data, and so on. We have been doing all of these things and all of these sources have influenced our understanding of the industry and its dynamics. This paper relies substantially on preliminary data we have collected from publicly available sources and market-analysis companies. We have been very careful to respect any concerns about data which might conceivably be seen as proprietary, a normal concern in any highly competitive industry. As the project develops, we plan to write reports dealing with other aspects of the data storage industry and globalization that highlight other kinds of information and concepts.

      In this report, we use 1995 as our base year of measurement. Methodological issues behind the numbers are discussed in Appendix 1. We have focused on gathering data relevant to a discussion of the various interpretations of location discussed above. We have sought to establish a detailed plant-by-plant portrait of the industry around the world. This allows us to establish the number of employees in each country, the number of employees in each step of the value chain also by country, the size of the wage bill in each country, the step of the value chain in each country, and the quantity of production in each country.


      This information allows us to clarify the variety of meanings that exist in defining location. Debates about "globalization " turn on nationality: in what country does the economic activity occur? Our findings show that the geographic distribution of activity in the HDD industry varies substantially depending on the measure being used. Our findings can be summarized in six numbers, each providing a different way of measuring the geographic pattern of the industry, and specifically the "share of each nation" in the industry.

1. Nationality of the firm that does the final assembly of HDDs. By this measure, the U.S. has 85% of world production, Japan about 15%.13

2. The geographic location of final assembly activity by country. By this measure , the U.S. share in 1995 was 5% (and nearly 0% in 1996). Southeast Asia dominates final assembly with 64% of world production.

3. The location of employment in the industry, including R&D and the various stages of production. By this indicator, the U.S. has about 20% of world employment, larger than any other country but decidedly not the bulk of world employment. Singapore and Malaysia have about 12% each while Thailand has some 13%.

4. The domestic-overseas split of employment by U.S. firms. Looking only at people who work for American firms in the industry, almost 30% of them are in the U.S. The vast majority of the remainder are in Southeast Asia.

5. The location of wages paid in each country. This reflects the actual amount of wages paid, reflecting the difference in wages that go to different skill levels in different countries. The U.S. share of total wages is about 42%, substantially larger than its share of total world employment.

6. Location of wages paid by U.S. firms. The percentage of the total wage bill of U.S. firms which is paid to employees in the U. S. is 62%. Thus, while employment by U.S. firms has shifted heavily overseas, a substantial portion of their total wage payments remain here.

7. Fraction of industry employment controlled by U.S. firms. 66 percent of all workers in the industry are employed directly by American firms. In some other branches of electronics, U.S. firms globalized their production by subcontracting manufacturing to overseas firms, but in HDDs firms globalized by setting up fully owned and controlled subsidiaries overseas.

      Thus in a business sense the U.S. is still the dominant nation in the industry, and U.S. residents are earning most of the money paid by U.S. HDD firms. But by other measures the U.S. is less than half the industry, and by some measures it is very small.

      There are certainly other measures that could be relevant:

8. Location of R&D. One pattern noted by all observers is the concentration of R&D in the U.S. and Japan. Data on this would clarify the extent of this observation and would also provide a benchmark for measuring future shifts in the industry. For example, Singapore and other countries seek to move up the value chain into R&D. Although opinions vary on their capacity to do so, the outcome is of very great importance: much of the benefit of this industry which accrues to the U.S. in employment terms is closely linked to R&D. Thus, if R&D goes offshore, so will many of the highest paid jobs. We intend to explore data that could help us understand this relationship.

9. Location of ownership by stockholder shares, dividends, profits. A major benefit in any industry is of course the value of share prices and dividends that flow from profits. We are not able to track ownership in detail however interesting and important it might be. We suspect that what is true of many industries is also true of this one: ownership of publicly traded American companies is generally highly concentrated in the U.S.

      Each of these figures tells us something different about the meaning of globalization, and its implications for managers, investors, employees and governments. We see this information as useful in the analysis of a variety important issues including:

  • Evaluating the location strategies of firms in the industry, such as their conformance to industry norms.
  • Predicting future siting trends by firms.
  • Evaluating manufacturing and management practices.
  • Developing public policies to assist firms in the industry and attract investments to the industry.
  • Analyzing employment effects of industry, not only the direct job losses to activities moving overseas, but the indirect effects of jobs gained and lost from overseas activities.
      This document is a preliminary report on these issues from a project funded by the Alfred P. Sloan Foundation of New York on "Information Storage Industry Center . There is much room for improvement of our data set before we can be confident about all the findings. Many readers of this report will know far more than we do about various aspects of the industry. We hope they will let us know about errors and contribute to improving our data and analysis.

      The paper represents but one portion of a larger project on the industry. Other aspects of the project include: a study of the evolution of the industry, looking at historical patterns of globalization and company success; an exploration of the coordination issues between R&D and location involved in globalization; an examination of the "conversation" about the industry, that is the understandings and assumptions about its dynamics that prevails among specialists in various aspects of the industry; a study of regional dynamics in Southeast Asia linking up various aspects of the value chain between final assemblers and suppliers; and an exploration of the role of public policy in influencing location decisions.

I. Where Do Disk Drives Come From? The Nationality of Production

A. Domination by American Firms
      The nationality of firms can be defined as the country of their incorporation. By that definition, the HDD industry is dominated by a small number of American firms.14 Japanese firms, although they have been in the industry for three decades, continue to have a small market share, whether measured by units shipped or revenues (Table 1).


      Measuring the industry by nationality of the firm which sells the product tells us something important. American firms overwhelmingly dominate this global industry to a degree not matched by much else in the modern economy, from consumer electronics to automobiles. For example, among segments of the computer industry, only microprocessors rival HDDs in domination by U.S. firms.

B. Location of Final Assembly

      While the largest producers of HDDs are U.S. owned and managed, a substantial amount of the work done in the industry lies outside the U.S. This is most conspicuously the case if we look at final assembly. It is this step which gets the "Made In Country X" label which is commonly taken to describe the location of production. By that measure, there is very little American presence in the HDD industry. The firms are American, but they do not assemble their products in the United States (Table 2).



      Southeast Asia, notably Singapore, is the primary location of final assembly. Slightly less than half of the world¹s disk drives come from Singapore, with most of the rest of Southeast Asia production being done in Malaysia (principally the island of Penang) and Thailand. Somewhat surprisingly, two-thirds as much production takes place in Europe, principally Ireland, Germany, and Hungary, as in Japan. The U.S. produced less than half the European level, and less than a fifteenth of the Southeast Asian volume.

      By this measure the U.S. accounted for a very small share of the industry in 1995. And in late 1995 and early 1996, the U.S. final assemblers shut down their remaining final assembly activities in the U.S. IBM moved its final assembly from San Jose to Singapore and from Germany to Hungary; Quantum shut down its remaining final assembly in California and Massachusetts and subcontracted the business to MKE, mostly in Japan, and Hewlett-Packard closed its Idaho and Malaysian plants and withdrew from the business. Firms do continue to produce in the U.S., largely, though not entirely, to sustain pilot lines; Seagate appears to support modest volumes of production in Oklahoma City. The 1996, U.S. production seems to be about 1% of the world total.

II. The Value Chain
      We now have two diametrically opposed measures of national origin of HDDs. According to one, firm ownership, the industry is U.S. dominated. According to the other, location of final assembly, the U.S. is negligible and Southeast Asia is dominant. In fact we claim that although both are useful, neither measure captures the essence of the national origin issue. Instead, we believe that the location of employment and of value added are more fundamental measures for many purposes. Location of employment measures the number of jobs. Value added captures where economic activity takes place, i.e. the transformation of inputs into outputs. For methodological reasons, in the discussion below we use wages paid (employment times wage rate) rather than value added.15

      In measuring employment and wages, it is not enough to look just at final assembly. Many other activities go into making a product: research, development, parts manufacturing and subassembly, tooling, repair, service, marketing, and so on. Knowing where final assembly takes place does not capture the full range of activities for the creation of a complete product and getting it to market. Linking together all elements of an industry requires describing its "value chain." This is the sequence of economic and physical steps from raw materials to final products in the hands of consumers.

      If firms had complete vertical integration of all steps of the value chain, then employment in the value chain would equal employment in the firms that assemble disk drives. This is not the case. The largest American HDD firms have very different degrees of vertical integration. The least vertically integrated is Quantum, which for the most part handles only design and marketing, leaving all assembly to its Japanese partner MKE. At the other end of the continuum lies IBM, which makes and assembles most of its key components.16 The other major American firms, as well as other firms in the industry, have intermediate levels of vertical integration, assembling their own drives, but buying some or most of the key components from other firms.

A. Major Steps in the Value Chain
      The next step in understanding the industry is to identify the major components of the value chain. For example, each drive includes a printed circuit board, which is populated with integrated circuits and other components. Meanwhile, there is always a question of how far back to take value chain analysis. Integrated circuits, for example, are made from bare silicon wafers, which are made from silicon ingots, which are made from sand. Similarly, there is a choice of how far forward to go: disk drives are just a component in computers, which in turn are a component in computer systems.

      The value chain of the HDD industry is shown schematically in Figure 1. The product is built in four main sub-chains: electronics, heads (devices that read and write the data), media (platters that hold the data), and motors. These four chains come together at final assembly. In addition, there are other activities for processing knowledge and information that do not directly handle physical material.

  • Electronics. This includes semiconductors that are used in the disk drive, discrete components, printed circuit boards (PCBs), and their assembly. It also includes the flex circuits that connect the PCB to the drive and to the heads.
  • Heads. Heads are the critical devices that read and write information. Manufacture of the head subsystem is done in stages, beginning with very highly automated and technically complex wafer fabrication and ending with the quite labor intensive activities of head-gimbal assembly (HGAs) and head-stack assembly (HSA). Critical mechanical components feeding into head assembly include the actuator and the suspension. One U.S. company, Hutchinson Technology, has around 70% worldwide share of suspensions.
  • Media. Media constitute the material on which the information is stored, the "platter," and can be made of aluminum or glass. This is also a very high technology aspect of the process. Typically, aluminum blank substrates are nickel-plated and polished before the platters are sputtered and finished.
  • Motor. The spindle motor spins the media with extreme precision. One Japanese company, Nippon Densan, has about a 75% worldwide market share.
      These constitute the major subassemblies of disk drives. The heads and media are then brought together, along with a spindle motor, in a head-disk assembly (HDA). During final assembly the HDA is brought together with the electronics, and the HDD is tested.17 In addition to these activities the value chain includes steps that are outside the core flow of materials but are extremely important:

  • Tools and equipment. An important component of the process involves the creation of the tools and equipment that make components, assist in assembly, or test outputs at various stages of assembly. Different firms use different degrees of automation, and different degrees of purchase versus in-house manufacturing of tooling. But almost all test instruments are made by specialist suppliers.

  • Research and Development. In an industry with breathtakingly short product life cycles (less than a year between successive product generations, and less than two-year product life cycles), R&D is of great importance. R&D continues to be overwhelmingly located in the U.S. and Japan. We include technology development, product design, process development, and pilot production in this category.
  • Sales, service and management. These are critical functions for an industry characterized by intense competition and rapid technological change. After-sales service is becoming a particularly important competitive tool, involving such value-added activities as repair, operator training, and engineering support.
      The best measure of employment in the HDD industry is the sum of employment in all the steps of the production chain noted above. This captures employment within disk drive companies and within a wide range of the suppliers that are intimately involved with the value chain.18 However, because of unavailability of data, we could not include them all in this project. Our analysis thus only covers companies that make disk drives, media, heads, semiconductors and equipment.

III.Jobs and Wages: The Data

      Armed with this model of the value chain in the HDD industry, we can now begin to fill in the boxes about where activity in each step of the chain actually takes place. We can also develop data about the wages paid to different skill levels in each of these steps.

A. Geographic Distribution of Employment

      We estimated 1995 employment by location and activity for each firm in the HDD industry value chain. This entailed some approximations and gaps. The main assumptions and missing data are described in Appendix 1.19

      The geographic distribution of these 287,000 people is quite distinct (Table 3). The U.S., with 20% of employees, has the highest headcount of any single country. Singapore and Malaysia have about 12% each, while Thailand has some 13%, making Southeast Asia the dominant region. (A country-by-country listing is in Appendix 3.) Japan¹s 7% of employment puts it in sixth place, behind China¹s 10% share.

(Table 3)

      The geographic distribution of employment is also distinct in terms of stage of the value chain. While only 9% of employment in heads and 7% of HDD final assembly are in the U.S., more than 50% of media employees and 32% of those in semiconductors work in the U.S. A large portion of global employment in production and test equipment is located in the U.S. Our figures for employment in Japanese equipment companies are rough approximations, and we have no numbers for European equipment suppliers, which are important in some types of equipment.

      Putting aside these caveats, we find that the industry employed 287,000 worldwide in 1995. By stage of production, almost 44% worked in heads and head subassembly, making it by far the most labor intensive. Final assembly was second with just under 22%, while the rest of employment is widely distributed.

B. Where are the Wages Paid?

      The major goal of the previous section was to measure employment. It focused entirely on the number of jobs in each location and step of the value chain. This tells us a lot about globalization, but it does not answer a very important question: What is the actual wage bill that is paid in each country? Looking only at the number of jobs does not measure the value of the jobs, and these figures are thus deeply misleading about the geographic distribution of the return to employment. Some stages of this industry, such as R&D, require highly skilled workers, with extensive training and education who are paid at a high wage. Other stages of production, notably head subassembly, use large numbers of low wage, low skill workers.

i. Geographic Distribution of Wages Paid

      To evaluate the geographic distribution of the value of employment, we estimated the size of the wage bill, taking into account the varying level of wages according to skill level. This can be done by ascertaining the wages paid to workers of different qualifications, how many of each are located in each country, and then multiplying the two together. Presently we are using national average wage data, rather than actual wage levels in the HDD industry. We divide workers into two skill levels: high skilled, for whom we use engineering salaries, and regular workers, for whom we use national average factory worker wages. For now we use gross estimates of the ratios between skilled and unskilled workers in each country¹s portion of the HDD industry. The procedure is described in greater detail in Appendix 1.

      Table 4 shows the percentage of worldwide wages that is paid in each country. Clearly, the shares of American and Japanese employment are larger on a wage basis than on a headcount basis, with the U.S. first at 42% of worldwide wages and Japan second at almost 24%. No other country has a share larger than 10%, and the entire Southeast Asian region has slightly less than 13% of the world¹s wages.


      Thus, using wages as a proxy for economic activity, we estimate that more than 40% of the industry¹s worldwide economic activity took place in the U.S. Figure 2 summarizes the same information and compares headcount with wage-based measures. The disproportionate role of Southeast Asia in headcount, and of the U.S. in wages, is clearly visible.


ii. U.S. Firms

      Looking at the economic activity of American firms in isolation is also interesting. Much of the controversy in the U.S. about globalization has to do with the impact of that process on wages and employment. Much is made of the flow of jobs to Southeast Asia, and other aspects of globalization revolve around the distribution of different pieces of the value chain (R&D, assembly, parts manufacturing, etc.).

      Table 5 presents information on employment and estimated wages for U.S. firms only.20 Clearly, these two alternative measures of globalization--the distribution of headcount and wages paid--are drastically different. While just under 30% of the headcount of these firms is in the U.S., about 62% of their wages are still paid in the U.S. Appendix 4 shows the location of employees in U.S. firms.

      The data appear to show that U.S. companies spend a higher percentage of their wage bill in the U.S. than do foreign firms. This would seem to confirm the view held by analysts who argue that the nationality of firm influences the geographical distribution of benefits in an industry.


iii. Summary of the Employment/Wage Discussion

      The hard disk drive industry has become one of the most globalized in modern industry. American firms dominate the industry, and globalization has been a key strategy for them. Over the past 15 years, these firms have sent final assembly outside the U.S. while keeping important elements of the value chain, notably product development and equipment manufacturing, in the U.S. The data in the previous pages can be summarized as follows:

      The numbers show quite strikingly that the meaning of globalization varies substantially according to the indicator. American firms dominate the production of HDD quite massively, but final assembly is being phased out in the United States. Although only a fifth of global employment in the HDD industry is located in the United States, two-fifths of the global wage bill is paid by U.S. firms and three-fifths of the wages paid by U.S. firms are paid in the United States. These figures convey a complex reality. They provide the foundations for some probing of cause: What drives some aspects of the value chain to go outside the United States? What causes some aspects of the value chain to remain in the United States? As we probe answers to these questions, we can learn things that help managers and policy makers.


C. Labor Costs, Labor Skills, and Location

      The data above show that the production of hard disk drives has been thoroughly globalized. Firms have learned how to locate different steps of the value chain in different places. IBM makes media and wafers in Germany and the U.S., heads in California, Mexico, and Singapore, and assembles drives in Singapore, Thailand, Hungary, and the U.K. Read-Rite makes wafers in the U.S., sliders and heads in Thailand, and head stack assemblies in Malaysia. Japanese HDD firms moving production offshore have predominately been going to the Philippines. Why do companies select one location over another?

      One element is surely the cost of labor: low skilled labor is cheap in some places, and relatively expensive in others. Another element is the availability of expertise, especially of engineers with skills needed for that stage of production. A fundamental relationship in world labor markets is that cost of unskilled labor and availability of skilled labor are positively correlated. So, we would imagine that companies seek to place their least complex activities in those locations that have the greatest abundance of inexpensive low skilled labor, and their most sophisticated activities where highly skilled labor is the most plentiful. Intermediate activities, those requiring a mix of labor skills, should be placed in countries with intermediate levels of wages, expertise and skilled labor.

      We can show this graphically by plotting along the horizontal axis steps in the value chain according to their degree of technical complexity, and along the vertical axis the cost of labor by country (Figure 3). To the extent that labor cost and skill are the main drivers of location decisions, this graph expresses the connection among wages, the skills required for various steps in the production network, and the location of production. Low skill items tend to be labor intensive, and will go where unskilled labor is the least expensive. For highly automated or complex tasks, little unskilled labor is needed but availability of skilled engineers is crucial. Hence production processes are done in the high wage, high skill availability countries of the U.S. and Japan.

Figure 3 Stylized Availability - Wage-Skill Relationship



      This relationship helps explain movement within the value chain over time. Tasks that went to Singapore when that country had low wages have moved elsewhere in Asia as Singapore¹s wages have risen. Singapore moves up the value chain in skill levels of the activities there, and in automation of production.

      To test the hypothesis represented by Figure 3, we plot the actual locations of employment in different production stages (Figure 4; various adjustments and consolidations were done to the data to display more clearly). Each "bubble" in Figure 4 shows the workers in that stage in that country. The area of the bubble shows the number of employees, its vertical location shows the average factory wage in that country, and the horizontal location shows the approximate skill level. Ordinal rankings were used for skill levels, with no attempt to subdivide the workers in a given process step into different skill levels.

Figure 4 Wage level versus technology level: empirical data


Key:

1 Head subassembly

2 Final assembly

5 Disk media

6 Head fabrication

7 Semiconductor fabrication

8 Equipment design and manufacture

9 Research and development

      Figure 4 shows some support for the hypothesized relationship between skills needed and national wage levels in countries involved in the industry. From the lowest wage countries (China) to higher wage ones, the mix of activities shifts toward higher technology processes. China, with the lowest wages at $.39 per hour, is used almost entirely for head subassembly, the lowest skilled and most labor-intensive activity. Thailand and Malaysia, with average wages $1.61 and $1.53, do some disk assembly as well. Singapore (average wage $6.29) has employment predominantly in disk drive assembly.

      Moving to the high wage countries, the key role of the U.S. is clearly visible, with sizable employment levels in all activities except head subassembly. The highest technology steps of the value chain -- head fabrication, semiconductor fabrication, equipment, and R&D -- are done only in the high-wage countries, especially in the U.S. but also in Japan.21

      There are also a number of anomalies in the location patterns revealed by Figure 4, i.e. decisions not explained by the wage/skill hypothesis. In part this is due to taking a snapshot of changing patterns, but it also shows the importance of other factors, discussed in Section IV.

  • Korea has about the same wage levels as Singapore, yet activity there is shifted toward lower skills than Singapore. This reflects industry presence by Korean headquartered firms.
  • There is almost as much media manufacturing employment in Thailand as in Singapore, despite the 4:1 ratio of wages.
  • Since Thailand and Malaysia have almost the same wage rates, the hypothesis gives no information about their relative roles.
  • Many other countries where the industry does not appear also have low wages.
  • Western Europe and Japan shows activity in both high and low tech parts of the value chain, while the hypothesis predicts they should only do the high tech. The U.S. has some disk drive assembly, although that practically vanished in 1996, as discussed below.
D. Recent Movements in the Value Chain

While the 1995 numbers show that considerable value added remains in the United States, the industry continues to globalize production. Among U.S. HDD firms, Seagate has been by far the most active in continuing to globalize its operations. Since the beginning of 1996, it began to construct a new facility in the Philippines, opened two media plants in Singapore, added two factories in Thailand, and will begin production of disk substrates at a new plant in Ireland. IBM ceased assembly but continued other operations in San Jose California. IBM has also expanded its international reach by funding a joint venture to produce HGAs in China, while StorMedia has announced plans to set up its third media facility in Singapore.

      During this period, Japanese firms have been particularly active in expanding global production. Mitsubishi Chemical is adding production in Singapore, Ohara began to produce crystalline glass for use in substrates in Malaysia, TDK invested in heads production in China, and Asahi Komag began shipping from two new plants in Malaysia. This year Fuji began factory construction in Malaysia, Hoya Magnetics began production at a new glass disk media plant in Singapore, and TDK announced a major five-year investment to make heads in the Philippines.

IV. Drivers of Location Decisions: Factors Costs, Agglomeration, Regulation, Networks/Search Costs

      In the previous sections we described the value chain of the HDD industry and showed the location of employment and wages paid by country and place in the value chain. This section seeks to analyze the data through the frameworks outlined at the beginning of the paper: factor costs, agglomeration, regulation, and networks.

A. Factor Costs

Several elements arise in examining factor costs.

i. Labor costs

      Labor cost is the most commonly cited variable in discussions of globalization by members of the industry, academic specialists and other observers of the industry. "Companies go overseas to save on labor costs." This simple sentence summarizes the most commonly held explanation of the industry. It is certainly the prevailing wisdom among many economists and students of trade. Wages have a powerful impact on location decisions: wage rates vary greatly from country to country, especially wage rates for unskilled labor. Other things being equal, firms locate to save whatever they can on labor. The data above shows significant empirical support for the attention paid to labor cost. By and large, the low skilled end of production in the value chain has gone to countries where low skilled wages are significantly lower than can be found in the advanced industrial countries. High wage-high skill activities associated with R&D and capital intensive processes, on the other hand, are located in the more technologically advanced countries where these skills are more readily available.

      Yet it is clear that wage rates and skill requirements are by no means the whole story. While wages may be the prevailing explanation for location, their role deserves some questioning. Several issues arise:

a. Since wages are a small percentage of total cost, can they really explain location decisions? A common response to that question points to the competitive nature of the HDD industry. So intense is competition that companies look for ways to save even small amounts of money. Thus even if wages are only five percent of total cost for a step in the value chain, they are not zero and lower wages will reduce costs, all else equal. Still, if wages fall as a component, other factors become more relevant and a single minded emphasis on labor cost becomes harder to justify.

b. The mix of workers and other elements of production are a matter of strategic judgment by management, not a given determined exogenously by technology. Managers choose how they combine elements of a production system: employees, technology, research, logistics, etc. In other words, labor and other factor costs are components, not drivers. Quantum through MKE, for example, has sought to automate assembly, reducing labor content. This complements MKE¹s strategy of putting plants in higher wage, higher skill areas. Its strategy focuses on automation of manufacturing. Seagate, by contrast, uses more labor intensive techniques, and is a leader in moving to low wage areas. The importance of labor is thus a managerial decision, in which other elements are quite important.

c. The wage share of costs is not easily measured. Specialists who have examined the evaluation of labor costs by firms raise questions about the measurement of costs and the savings to be found in moving. It has been common, for example, for manufacturers to allocate fixed overhead expenses proportional to wages; doing so makes workers seem very costly, so that savings on labor seem to provide an immense benefit. The best accounting practices have shifted to better measure the role of wages, but this is another area where research is needed.

ii. Yields

      Many stages of the value chain are yield driven, i.e. suffer from scrap and rework. Since material costs are higher than other variable costs, especially in the later stages of the value chain, low yields can dramatically raise cost. Anecdotal evidence conflicts, but we believe that different locations and labor forces have different yields. Yield loss that arises from small glitches in moving from design to volume production, aggravated by long distances or unskilled workforces, can swamp savings from direct labor cost.22 Separating design from production geographically therefore requires tremendous confidence in processes for long distance ramp-up. We will be continuing research on this topic.

iii. Land, energy, transportation, communications

Wages are not the only element of factor costs. Land and building costs, energy and water, transportation and communications infrastructure -- these can be significant elements of production. Countries vary substantially in the costs and quality of these items. For much of the industry, manufactured products and their components are light in weight, so that air freight is the normal form of transport. Many items cross the ocean more than once before their final appearance in a computer. In the early days of data storage, hard disk drives weighed substantially more, and there thus may have been a greater reward to being near final assembly or computer manufacturers. Some observers note the high costs of land and buildings as barriers to further growth in Silicon Valley. While true, other parts of the U.S. have much cheaper land, so this can be only one aspect of overseas location.

iv. Capital costs.

In a capital intensive business, this is certainly an important variable. Do capital costs vary substantially across countries? Does the geography of where capital is raised shape location decisions in manufacturing? We suspect not. Do subsidies to firms for investment purposes affect location? We suspect they do, but this belongs under the rubric "regulation" below.

B. Agglomeration Effects

      The HDD industry is both global and concentrated. Its production chain spreads around the world, but at the same time, certain regions stand out for the concentration of activity in them. In the U.S., Silicon Valley continues to hold a substantial share of research, design, development, marketing and management. High end manufacturing continues to take place there: there are plenty of clean rooms, robots, testing facilities, and component manufacturing that takes place in the region, even if there is virtually no final assembly. Other parts of the U.S. find important clusters: Minnesota, southern California, Colorado.

      The major counterpart to the U.S. outside North America is Japan. While Japan does not dominate final HDD assembly the way it does other elements of the electronics industry, it is the other major location of the high-end research, design, and development in the HDD industry, and it manufactures some critical components in the system.

      Singapore and Southeast Asia comprise another substantial cluster of activity. This region dominates final assembly, most of the labor intensive sub-assemblies, and low tech components such as base-plates. Europe consists of a third cluster, with significant activity in Ireland, Germany and Hungary.

      Clearly, some kind of concentration occurs in key regions, and within regions, around a major urban pole inside it. The key regions are the U.S., Japan, Southeast Asia and Europe. The key poles are Silicon Valley, Tokyo, and Singapore; for Europe it is not clear what the urban pole is.

      While region and urban poles appear to be developing, it is not clear how best to understand the relations among them. How do decisions, knowledge, and management flow among Silicon Valley and Tokyo, Singapore and Ireland, rural Thailand and China? Does Singapore "manage" the lower value added activities in S.E. Asia, or are these done directly by the company headquarters? Does Singapore specialize in activities for the region? Does Ireland do this for Europe? These are important questions with implications for globalization, and they require further research.

      The concentration of key activities in certain cities and regions thus raises some important research issues for grasping the agglomeration effect and its future dynamic.

€ The global management of high tech R&D, face to face contact, and long distance communication. Modern technology shrinks distances. Information flows electronically conveying valuable content about design, markets, and technique. At the same time, face to face relations matter a lot. Managers fly long distances to monitor operations. Certain R&D activities remain strongly rooted in concentrated areas. Is this duality likely to persist and in what areas?

      While low wage labor is obviously widely dispersed, the high end skills in research, development and design remain quite concentrated. The U.S. has the highest concentration of skilled people. That concentration acts as a magnet or gravity core to hold related activities nearby and attract new ones. Singapore would like to move up the value chain into R&D; can it do so, and for what activities? What is likely to remain in Silicon Valley or elsewhere in the U.S.? How do firms manage the coordination of design with manufacturing over great distances? Some firms regard their skills in this regard as among the most valuable of their assets, proprietary knowledge not to be shared. As technological development pushes the knowledge boundaries, manufacturing tolerances grow ever tighter. These may increase the importance of manufacturing skills, and their relationship to design. Just how will this influence location?

€ Proximity to markets. Some observers think that being near the purchasers of final product is important, be it because of specialized information, or the service demands of consumers. By this logic, firms upstream in a chain of production move towards those who are downstream. By this logic, the HDD business globalized to be near their customers. The same point has been made about Ireland: HDD manufacturers went there because big players in computers are located in Ireland, Scotland, and parts of England. Other observers doubt that proximity to markets matters very much in location; computer assembly is located in Japan the U.S. and Taiwan, but there is very little final assembly of disk drives in these locations.

€ Proximity to suppliers and service activities. As the value chain spreads out (R&D in one region, final assembly in another, component supply in yet a third), the coordination among these elements becomes increasingly important. The capacity to supply components and services (repair, installation) gains in importance. It may be important for steps down the value chain to be near items above it.

C. Regulation and Public Policy

      A wide range of policies, rules, and regulations influence location decisions. Some of them do so directly by shaping the firm's bottom line calculations: tax breaks and subsidies for land, training and buildings. Others do so indirectly by influencing factor costs (labor skills, capital, land, environment regulation -- the costs of doing business).

€ Taxation: Many informants report that tax subsidies are a fundamental reason behind specific location decisions. These subsidies take the form of tax holidays, grants for training and research, rent, and other inducements. Some observers think that companies use transfer pricing to enhance the value of tax holidays.

€ Trade barriers and local content rules. The liberalization of international trade has been a major force in enabling the globalization of production. At the same time, existing trade regulations induce location in order to get around tariff barriers and to comply with local content legislation. This seems particularly salient in Europe.

€ Speed and ease of regulatory processes. Countries vary substantial in the speed and complexity of regulatory approval for building permits, shipping, labor, tax and a wide range of matters which are particularly important to an industry acutely aware of the shortness of its product life cycles. Singapore is cited as particularly quick, while China is quite slow. For the U.S., observers complain about delays with building permits, environmental inspection reports, and the like.

€ Public agency efforts. In some cases, public agencies vigorously recruit companies to locate in their jurisdictions. Observers comment that the Economic Development Board of Singapore is especially effective as a "recruiter." Some of their influence may have to do with taxes and infrastructure inducements already noted. Another factor may be the ability to provide quality information to busy executives about cost structures in alternative locations, thereby enabling companies to lower their "search costs."

€ Education and research programs. The skill levels required for various stages in the production process differ greatly. As noted above, the U.S. has a substantially higher concentration of the relevant talent needed for product development and equipment manufacturing, and is thus preferred for those elements of the productive system. The existence of this talent is influenced by larger systems of education and the funding of research programs.

€ Labor cultures and industrial relations systems. Managers often contrast the "work ethic" of different countries and the regulatory systems that influence the ability to hire or fire various kinds of employees.

€ Legal infrastructure. Some locations may be avoided even if labor is skilled because of concerns about adequate protection of intellectual property. Similarly, in some locations commercial law is underdeveloped which deters investment.

D. Networks

      A fourth approach to location examines the influence of relationships, networks, social ties, coordination, and search costs in shaping managerial decisions about location.

€ Ethnicity, religion, nationality, occupation, and culture. It is well known that most jobs are found via a personal contact. It is increasingly understood that informal or ascriptive criteria influence many other forms of economic activity as well. While these ascriptive influences were once seen as "irrational" or "non-meritocratic," they are increasingly seen as providing some real content to solving economic problems of coordination, search costs, and trust building. Networks among overseas Chinese, among Singaporeans, Taiwanese, Americans, university alumni, and other criteria seem to play a large role in linking together the tremendously spread out system of the electronics industry in general, hard disk drives in particular. The ability of businessmen from Singapore, Hong Kong, and Taiwan to operate effectively in China is one example of this phenomenon.

€ Coordination, vertical integration, and globalization. HDD firms differ in their managerial strategy on a number of dimensions, ranging from technology and labor force to globalization and R&D strategies, and the degree of vertical integration within the storage sector and between storage and the computer electronics business more generally. Each mixture poses special challenges to coordination The fully integrated firm manages by hierarchy internal to the firm; the fully disintegrated network relies on markets for some functions of coordination, but also upon a complex package of social ties that build trust and reduce coordination costs without formal contracts. Each managerial strategy for coordination may have different implications for localization.

€ Executive preference. Some observers suggest that executives have some range of discretion about their choice of location and that personal preferences play a role in the final decision. Managers may have special contacts with important figures from engineering, regulatory agencies, investors, and managers. These ties can be based on personal contact or ethnicity. "We went there because the boss knew a group of engineers with management experience," said one observer about a location choice. Managers may also vary in their choice of decision-making formulae and benchmarking criteria for evaluating location issues. Thus some managers may pick a location because they like to live there, or they think employees will like living there, or because they feel comfortable with the culture of the country.

V. Further Trends

      The data in this report focuses on a single year. The information on the geography of jobs and wages tells us a lot about globalization in this industry. However, a portrait of a single year cannot by itself demonstrate dynamics needed to understand the future. To do that will require more research on the past, and an exploration of the dynamics that drive the future. We have some suspicions about what might be important to an understanding.

A. Recent Trends

      By our measures it is clear the HDD industry is quite extensively globalized, perhaps among the most globalized production systems in the world. Since our study began in 1995 several important developments have taken place which indicate certain directions:

€ Mergers and consolidations: Intense competition has led to mergers and consolidations. Seagate acquired Conner Peripherals, Applied Magnetics made a short-lived move to acquire Read Rite, and Phase Metrics acquired several companies to become a larger player in test equipment. Hewlett-Packard and DEC have left the disk drive industry altogether.

€ Relocation of manufacturing away from the U.S.: IBM closed its disk drive assembly plant in San Jose, California to locate in Singapore and has announced a new plant in Ireland.

€ Research and development: Is it moving? Singapore claims to be moving up the value chain in R&D. Micropolis, once an American company, is now owned by a Singapore concern that has announced its intention to conduct more R&D in Singapore. Many plant officials in countries outside the U.S. report that R&D takes place at their sites. Are these signs that this core activity is also at play in the competition for location?

      The mergers and consolidation do not by themselves predict location decisions, but they may have bearing on the process. Historically, intense competition has put very great pressure on lowering factor costs. Firms seek every possible way of reducing what they pay for everything involved in production. As a result, firms are driven overseas to globalize outsourcing and production. This familiar story is often told with respect to the competition U.S. firms experience from foreign producers; in the HDD case, it is competition among American firms themselves that drives them overseas, not competition from foreign producers. The industry is now rather highly concentrated, with four U.S. firms producing over 70% of the final product. As mergers continue to concentrate the industry further, some observers have begun to wonder whether competitive pressures will decrease. If that happens, the pressure to globalize in search of lower factor costs may slacken.

      The relocation moves do hint at possibilities of a continuation of an earlier trend: twenty years ago most of this industry was indeed located in the U.S. Much of the value added still is, but, we suspect, a declining share.

B. Technology, Manufacturing, and Design

      As technology moves rapidly forward, it appears that issues of manufacturing are growing in importance. As technological requirements push the frontiers of knowledge, tolerances grow smaller. This increases the importance of manufacturing and tightens the link between R&D and manufacturing. This will certainly pose substantial challenges for management. It may require closer proximity between manufacturing and product development, which could pull manufacturing back to the U.S. Conversely, it could drive R&D abroad to existing production sites. Or no change in location may happen because modern communication methods will allow fluid communication between geographically distant manufacturing and research.

      Interaction between companies in the design of products and components may also influence the decision about whether R&D among different stages of the value chain need to be located close to one another. Close relationships between component makers and disk drive firms appear to be important factors in "designing in" heads or media for particular disk drives. To the extent that face-to-face interaction at the engineering level is critical in coordinating design, it may constrain a company from putting more R&D abroad.

      Technology issues may also impact vertical integration. Firms will have complex choices to make about which stages of the value chain they must retain in house and which they can or should buy from other companies. It is not clear whether vertical integration affects location decisions.

C. The Nationality of Firms

      Does nationality influence firm strategy and the distribution of benefits? Some observers believe that the market drives all firms toward a convergence of best practice, independent of the nationality of the firm¹s incorporation. Other analysts think that national differences of various kinds will continue to sustain different combinations of firm structure and strategy. Since there is more than one model of success, national differences can coexist.

      This debate appears quite salient in discussions of the role of Japanese firms in the HDD industry. Firms based in Japan have been active participants in many other branches of the electronics industry. Why have they been relatively small players in this industry? Although some important components in the value chain are dominated by Japanese firms, such as spindle motors and DRAM, American firms dominate the industry as a whole. Is this because of the volatility of technology?

      Some observers credit the intensity of competition and the rapidity of the product cycle change as the explanation for U.S. dominance over Japanese manufacturers. The latter, it is alleged, have distinctive skills in manufacturing, skills that come into play when the technology has stabilized. So long as the technology is unstable and rapidly evolving, the more free wheeling style of American firms is advantaged. Were that to stop, relationships would shift. Do American firms do best in free wheeling market conditions, while Japanese firms do better when the technology stabilizes? Will the situation continue, or will Japanese firms move through various aspects of the value chain to increase their share of the whole? Do Japanese firms behave in noticeably different ways? It appears that they do: while Japanese firms are globalizing, they are mostly going to the Philippines rather than to Singapore-Malaysia-Thailand, they do less technology transfer, and they involve fewer local managers and employees.23 What effects will these differences have on the evolution of the industry and on allocation of benefits to each country? If the largest returns in wages go to the people at the top research, design and management positions, are these jobs connected to the nationality of the major manufacturer?

D. Public Policy

      How does public policy influence firm location and the development of the industry? Public policy obviously affects taxes, education and research, and infrastructure. Should it be targeted to specific industries and companies, as Singapore does? Singaporean policy and practice are frequently cited by people in the industry as a major influence on firm location and behavior. Should the U.S. government, or state governments, undertake policies to compete with Singapore? Although actions of this kind are common practice in other countries, they are controversial in the U.S. Can or should the U.S. do more in the way of basic research, research consortia, or public purchasing such as defense procurement? Can U.S. cities and states do more by targeting HDD companies and working with them?

      By all evidence, the data storage industry can only continue to grow. It is a key element of the computer business and is appearing in more applications beyond the home and office workstation. It is also a dynamic industry, with constant improvements in technology and falling costs. All forecasts project growth and continued change. This report is a preliminary effort to take a snapshot of that industry in ways which allows us to think about the future.

Appendix


I. Sources for 1995 Employment and Wages

      The employment figures generated for this study come from a wide range of public and private sources. Financial reports filed with the SEC were very helpful in getting total employment for listed companies. TrendFocus was the major source of employment figures for media and heads. The 1995 -1996 IDEMA Directory served as the initial source for employment in equipment companies. Because these sources typically don¹t report employment at specific locations, we conducted broad searches of the popular and business press and Rumors & Raw Data to find statements mentioning a company¹s employment at a particular facility. The public sources used include Asian Economic News, The Arizona Republic, Business Journal (San Jose, ) Bangkok Post, Business Times (Singapore), Computer Business Review, Computer Dealer News, Chilton Electronic News, COMLINE Daily News Computers, Computer Weekly, Computer System News, Computerworld, Denver Business Journal, Daily Yomiuri, Electronic Business, Electronic News, Financial Times, Irish Times, InfoWorld, Jiji Press Ticker Service, Los Angeles Times, Nihon Keizai Shimbun, New Straits Times Malaysia, The Nikkei Weekly, New York Times, South China Morning Post, San Diego Union Tribune, The Straits Times (Singapore), and The Wall Street Journal.

      However, our coverage is far from complete. We had to make a number of assumptions to generate informed estimates about industry employment and its location. The estimation procedures are described below. An example of an important assumption we had to make is posed by firms that are diversified: How do we attribute an employment number to the part of their business that sells to the disk drive industry? Moreover, our figures understate worldwide employment in the industry, broadly defined as the value chain in Figure 1. In addition to all of the caveats mentioned below, our figures omit employment associated with the making of flexible circuits, bearings, actuators, magnets, lead-wire assemblies, consumables, and aluminum and other raw materials that feed into these products. Many of these products are made by non-U.S. firms. Our figures also understate HDD-related employment in Japanese companies, especially that country¹s equipment suppliers.

A. Hard Disk Drive Employment

      Employment figures for disk drive companies come from SEC reports, exhaustive Lexis-Nexis searches of the business press, and company web pages. Our employment data for U.S. disk drive companies is moderately better than for Japanese disk drive companies. One reason for this is that Hitachi, Fujitsu, NEC, etc., are diversified companies, and disk drives are not their primary line of business. This makes it difficult to trace employment in disk drive activities. Another is that their disk drive operations are not followed as closely by the English language press as are the operations of U.S. disk drive companies.

B. Media and Heads Employment

      As mentioned above, the primary source for employment in heads and media is TrendFocus, supplemented by articles from the business press.

C. Integrated Circuit Employment

      All employment figures refer to merchant producers and do not include any captive production of semiconductors by IBM or Seagate.

Basic Employment Estimation

      Owing to the lack of individual company information, integrated circuit employment for HDDs was estimated from global semiconductor employment figures. These global figures were based on a SGS-Thomson survey of regional and national semiconductor employment by 50 semiconductor companies as reported by CMP Media Inc., Aug. 12, 1996. The basic employment estimation procedure involved taking the estimated global HDD IC market share in 1995 and multiplying it by the employment figures to estimate HDD IC employment by region and company nationality.

Market Estimation Procedure

      The 1996 IC market for HDDs was estimated at $3.2 billion by Dataquest as reported in Investors Business Daily on 6/25/96. In order to measure the 1995 HDD IC market, it was assumed that HDD IC production matched the global movements in the semiconductor market. Since the semiconductor world market shrank by 11.8% in 1996, it was assumed that HDD IC production decreased by the same amount. Thus, the HDD IC market total was estimated to be $3.628 billion (3.2/.882) for 1995. This is equivalent to HDD IC products using 2.51% of the world semiconductor market.

Final Estimation

      This market share number (2.51%) was then multiplied by regional and world semiconductor employment figures to get the estimates for world and U.S. company HDD IC employment. Data on individual countries in Asia were not available.

D. Equipment

      Employment in equipment is based on a telephone survey of IDEMA members in August, 1996. We called only firms headquartered in the United States, so our estimates refer only to the distribution of employment in U.S. firms. (But we also used these numbers to derive an employment figure for Japanese equipment companies. We describe that procedure below.) In addition, we surveyed only IDEMA members that indicated in the directory that they made equipment. Thus our employment numbers exclude consumables, chemicals, etc.

      Of the 132 equipment companies telephoned by UCSD, 86 provided us with information. We asked them to confirm their sales, the number of employees, the percent of their sales related to the disk drive industry, and where they manufacture. Because there was virtually no difference between these 86 companies and the 46 companies that did not participate in the survey (as measured by number of employees reported in the IDEMA directory), we used these findings to estimate an HDD-related employment figure for the entire 132 equipment companies. The steps we followed are:

Adjusting for the percentage of employees involved in HDD-related activity.

      Of the 86 companies in the survey, only 64 were able to estimate either the percent of total sales derived from the disk drive industry or the amount of sales related to the disk drive industry (from which we could calculate a percentage based on their total sales). For these 64 companies, we then applied this percentage to that company¹s total employment to estimate its HDD-related employment. The number of HDD employees was then summed for these 64 companies, and the mean percentage of disk drive employment in these firms was determined to be 53%. The estimated HDD-related employment for these 64 companies is 5,078.

      The next step was to calculate the total employment for the remaining 68 equipment firms. Using employment figures reported in the 1995-96 IDEMA directory, the total employment for all of these firms summed to 14,270. This number was then multiplied by 53% (the estimated percent of HDD employment in the 64 firms). This calculation yielded 7,563 employees. Total HDD-related employment in the equipment business was thus estimated to be 12,641.

Where are these 12,641 located?

      Because companies could not break down location by HDD-related activity, we estimated it from the information obtained from the 86 companies participating in the telephone survey. These 86 firms had a total employment of 19,178. Of these, 15,526 (80.9%) were employed in the U.S. and 3,652 (19.1%) in unknown countries abroad. (Although a number of respondents were able to specify locations, they seldom could say how many employees worked in those locations. We therefore classified them merely as non-U.S. locations.) Thus, we estimated that 10,234 employees involved in HDD-related activity worked in the U.S. and 2,407 worked overseas. We then assumed that two-thirds of these 2,407 overseas workers were in Asia and one-third in Europe.

Estimating employment in Japanese companies

      We did not have any employment figures for Japanese equipment suppliers. But rather than exclude them, we thought it best to try to derive an estimate since they are an important part of the industry. Some industry executives we consulted thought that the Japanese share of the equipment market was roughly the same as their share of the HDD market. Their reasoning was that, with a few notable exceptions, Japanese HDD companies bought equipment from Japanese suppliers, while American HDD firms sourced equipment from American firms.

      We thus assumed that U.S. firms had 85.5% of the equipment market and the Japanese had 14.5%. We applied this reasoning to share of employment--that the 12,641 employed by American firms represented 85.5% of global employment in equipment. This resulted in an estimate of 2,144 employees working for Japanese equipment suppliers. We also assumed that all of them were employed in Japan.

II. Wages and Salaries

A. Hourly wages

      To calculate the wage bill for people involved in assembly or other non-managerial and non-engineering operations, we used 1995 average hourly wage in manufacturing, including benefits, for each country hosting HDD activity. These wage rates were calculated by Morgan Stanley as reported in The Economist, November 2, 1996, p. 77. Hourly wages for the Philippines, which were not reported in The Economist article, were taken from the Philippine Statistical Yearbook 1995 published by the National Economic Development Authority in Manila, 1996.

      From these sources we got the following hourly wage rates: Indonesia, $.30; Malaysia, $1.59; Philippines, $.71; Singapore, $7.28; Thailand, $.46; Japan, $23.66; South Korea, $7.40; China, $.25; and United States, $17.20. We also averaged wage rates across countries for regional or unknown locations: Other Asia and Unknown Asia, $5.21; Eastern Europe, $1.70; Western Europe, $18.83; and Other Country and Unknown Country, $10.31.

B. High-skilled salaries

      We adjusted for the higher-skilled jobs involved in design, development, process engineering, management, marketing, etc. We used 1994 engineering salaries for each country as a proxy for all higher-skilled jobs. These salaries are taken from The World Competitiveness Report 1995, published by IMD/World Economic Forum, Geneva: September 1995. (We are not yet in possession of the 1996 report.) The one exception is salaries for Chinese engineers. These were taken from the U.S.-China Business Council 1994 Survey, cited in the China Business Review, May-June 1995. Dollar values for these salaries were derived using an exchange rate of 8.62 RMB to $1, which was the average exchange rate for 1995 as reported by the EIU Country Report, 1996.

      From these sources we got the following annual salaries for high-skill workers: Indonesia, $31,615; Malaysia, $30,582; Philippines, $32,830; Singapore, $52,536; Thailand, $39,239; Japan, $123,154; South Korea, $31,607; China, $2,514; and United States, $62,219. We also averaged yearly salaries across countries for regional or unknown locations: Other Asia and Unknown Asia, $43,010; Eastern Europe, $21,165; Western Europe, $63,749; and Other Country and Unknown Country, $52,081. These are unweighted averages.

III. Calculating the Global Distribution of Wages Paid

      We made several assumptions in calculating the wage bill. First, we had to make an assumption about the percentage of high-skilled employees working in the industry in Japan and the U.S. We set the U.S. and Japanese percentages at 50% since these are the centers of product design, development and pilot production, and very little final assembly of disk drives and other labor intensive activities takes place at these locations. Next, we made assumptions about the ratio of higher-skilled to lower-skilled employees in facilities located in the rest of the world. Based on information acquired from visits to media, heads and disk drive factories in Southeast Asia, we set the percentage of high-skilled workers in disk drive employment in all other countries at 5%. (The percent of engineers and managers at these factories ranged from 3-10% of total employees.) Finally, we made an assumption about the number of hours worked by factory workers. We assumed a 40-hour week for 50 weeks, or 2000 hours.

      We also experimented with skilled worker percentages other than 5%. It turns out that even doubling this number to 10% raises total wages, but has very little effect on the percentage distribution of economic activity by country.

      In the future, we hope to obtain some actual payroll information by company and facility. There may be systematic differences between national average wages and salaries, and levels in the HDD industry. Such differences would affect the absolute level of wages estimated, but probably not the national shares.

Footnotes

  • 1Our thanks to Jim Porter of Disk/Trend, Inc. for advice and the use of the data from his Disk/Trend Reports. We also thank Mark Geenen of TrendFocus, Inc. for permitting us to use data from his various reports on heads, media and data storage; and Jack McLaughlin, Business Research Consultants, for access to his Rumors & Raw Data. We also are grateful to Denis Mee, Phil Devin, Dennis Waid, Stephan Haggard, Rick Doner, Dieter Ernst, John Zysman, Michael Borrus, and numerous anonymous executives for discussion on the industry. Gigo Alampay, Kyle Eischen, Allen Hicken, Colin McIff, and John Richards provided crucial research assistance.
  • 2The title of this paper "Who is Us" is adapted from a well -known article by Robert Reich, former Secretary of Labor for the Clinton Administration. See "Who is Us?" Harvard Business Review, January -February, 1991, p. 53-59. See criticism of Reich's analysis by Laura Tyson , "They Are Not Us: Why American Ownership Still Matters," The American Prospect, Winter 1991, p. 37-49.
  • 3This typology was suggested by James Rauch, Professor of Economics, University of California, San Diego, in oral conversation, December, 1996.
  • 4The factor approach goes back to Ricardo's two factor model of comparative advantage. See also Raymond Vernon, Sovereignty At Bay: The Multinational Spread of U.S. Enterprises (New York: Basic Books, 1971).
  • 5Michael Borrus, "Left for Dead: Asian Production Networks and the Revival of U.S. Electronics," in Eileen M. Doherty (ed.), Japanese Investment in Asia : International Production Strategies in a Rapidly Changing World (Berkeley: Berkeley Roundtable on the International Economy, 1994).
  • 6James Rauch, "Does history matter only when it matters little: the case of city-industry location?," The Quarterly Journal of Economics, (August, 1993), p. 843-867.
  • 7Economists' analysis of the agglomeration are traced back to Alfred Marshall, Principles of Economics (New York: The Macmillan Company, 1960 [1890]), Paul Krugman, Geography and Trade (Cambridge: MIT Press, 1991), and Michael Porter, The Competitive Advantage of Nations (New York: The Free Press, 1990).
  • 8On issues of regulation see Robert Gilpin, U.S. Power and the Multinational Corporation: The Political Economy of Foreign Direct Investment (New York: Basic Books, 1975), Chalmers Johnson, MITI and the Japanese Miracle (Stanford: Stanford University Press, 1982), Robert Wade, Governing the Market (Princeton: Princeton University Press, 1990), and Stephan Haggard, Pathways from the Periphery (Ithaca: Cornell University Press, 1990).
  • 9Gary Gereffi and M. Korzeniewicz, eds., Commodity Chains and Global Capitalism (Westport, Ct: Praeger, 1994), Peter Cowhey, "Domestic Institutions and the credibility of international commitments," International Organization 47:2 (Spring, 1993), p. 298-326, John Richards, "The Domestic Politics of International Regimes: The Regulatory Institutions for Trade in Aviation Services," dissertation in progress, UCSD, Department of Political Science, 1997, Roger Noll and Frances Rosenbluth, "Japanese and U.S. telecommunications policy" in Peter Cowhey and Mathew D. McCubbins, Structure and policy in Japan and the United States (New York: Cambridge University Press, 1995).
  • 10James Rauch, "Trade and Search: Social Capital, Sogo Shosha, and Spillovers," National Bureau of Economic Research, Working Paper 5618, June 1996, and "Networks versus markets in International Trade," NBER, Working Paper 5617, June , 1996, Linda Lim, "Social Capital and Foreign Direct Investment By Newly-Industrializing Countires: A Perspective from Southeast Asia," Workshop, NBER, "Social Capital, International Trade and Investment," Stanford, March 1-2, 1996. On the importance of networks and informal relationships, see AnnaLee Saxenian, Regional Advantage: Culture and Competition in Silicon Valley and Route 128 (Cambridge, Mass.: Harvard University Press, 1994).
  • 11See papers from NBER, "Social Capital, International Trade and Investment," Stanford, March 1-2, 1996; the papers are being edited for publication by James Rauch.
  • 12See Rauch, "Does History Matter," for a discussion of the ways government policy overcomes first mover disadvantages and induces concentration and the agglomeration affect.
  • 13We include Quantum, which subcontracts the manufacture of its drives to MKE of Japan, in the U.S. numbers; likewise, we consider Maxtor a U.S. company, although it was acquired by Hyundai of South Korea in November, 1995.
  • 14We use the conventional definition of firm nationality, namely that the parent is incorporated in the U.S.
  • 15For a further discussion of the methodology used to calculate these numbers, see Appendix 1.
  • 16IBM is considered more vertically integrated than Seagate because IBM makes its own semiconductors for use in its hard-disk drives while Seagate purchases semiconductors from outside vendors.
  • 17Most companies assemble the head-disk assembly, do final assembly, and do final testing in a single integrated process, although a few companies break them up more finely. Therefore we aggregate these steps together in our numbers.
  • 18A total input/output analysis would include additional elements that generate employment: transporting products, buying electricity and other services, basic research in universities, etc. In addition, one could extend the definition of the value chain downstream to disk drive array integrators (almost all of which are U.S. firms, with much of their employment in the U.S.), computer manufacturers, and distributors and resellers.
  • 19The major gaps include our coverage of Japanese firms, which is less complete than our coverage of American firms, and our coverage of indigenous supplier firms in Malaysia and Singapore. For about 15 percent of employment, we know employees work in the industry but we don't know what country they are in, what activity they work on, or both.
  • 20Since this data is for 1995, Maxtor is treated as U.S.-owned.
  • 21Our level of detail on semiconductors oversimplifies the distinction between semiconductor fabrication, done in the high wage countries, and semiconductor assembly, usually done in low wage countries but not shown in Figure 4. See Appendix 1.
  • 22For example in one case we analyzed, a ten percent change in yield has more impact on cost than a $20 per hour change in the wage rate. See Process Cost Analysis for Hard Disk Manufacturing by Scott M. Hampton, Report 96-02 of the Information Storage Industry Center, 1996.
  • 23A notable exception is IBM-Japan, which moved assembly to Thailand via a subcontractor.