Information Storage Industry Center

Thailand's Hard Disk Drive Industry

Richard F. Doner
Department of Political Science
Emory University

Peter Brimble
The Brooker Group (Bangkok)
Report 98-02 October 1998

Abstract
      Thailand has become one of the world's largest production sites for hard disk drives and related components. The disk drive industry is now one of Thailand's major exporters. The country contains clusters of firms producing large volumes of drive assemblies, head-related products, and spindle motors. The industry's growth is the result of two factors: Thailand's generic factor costs and policy incentives, (e.g. inexpensive but experienced labor, stable macroeconomic conditions, tariff reductions for exporters); and proximate clustering of firms both within Thailand and between Thailand and the rest of the region. Thailand is effectively part of a regional production network including Singapore, Malaysia, the Philippines and China. But the disk drive industry in Thailand exhibits significant weaknesses. In addition to lacking indigenous technicians and engineers, the industry's component base is almost totally foreign owned, and its capacity for production-related services such as precision metalwork is very thin. Disk drive production is effectively isolated from Thai firms. These problems reflect protectionist tariffs, tax policies that inhibit subcontracting, weak or non-existent support for training and technology diffusion, and a lack of bureaucratic expertise and cohesion.

      This report is part of the Information Storage Industry Center (ISIC) at the University of California, San Diego. ISIC is funded by the Alfred P. Sloan Foundation. The authors are grateful for research support from Ms. Pattanan Woodikarn-Timm (Brooker Group), Bryan Ritchie (Emory University) and Allen Hicken (University of California San Diego) and for logistical support from the Brooker Group. This is a working paper with the usual errors and omissions. Any questions and comments are welcome and can be directed to Rick Doner. This paper will become a chapter in an edited volume.

Table of Contents
Introduction and Executive Summary
I. Basic Features

II. Explanatory Factors

Tables
Bibliography
Endnotes

 Introduction and Executive Summary¹

Since Seagate's initial 1983 investment in head-stack assembly, hard disk drive (HDD) production in Thailand has grown to become one of Thailand's major export industries and a major component of global production. This report first reviews the industry's basic features: its size, scope and depth, and evolution. The second section explores the factors accounting for the industry's growth and evolution. The paper's principal arguments are as follows:

1. Structure: The Thai HDD structure contains three principal clusters, each of which involves many of the world's most significant disk drive firms: head-stack and finished drive assembly (Seagate, IBM and Fujitsu); head-related components such as sliders and suspensions (Read-Rite, KR Precision, and Magnecomp); and spindle motors and related components (Nidec and NMB/Minibea).



2. Position in Global Production Network: Thai disk drive operations began largely as "hand-off" sites for products whose volume manufacture was perfected in Singapore. Subsequently, Thai operations themselves ramp-up volume production directly from Japan and the U.S. Thailand's product focus is becoming drives for mobile and desktop computers.



3. Reasons for Location: Disk drive producers have come to Thailand due to a combination of generic factor costs, proximity, managerial networks, and follow-the-leader behavior. Typically, firms want to locate somewhere near clients and/or suppliers. But proximity can mean locating elsewhere in Southeast Asia. The choice of Thailand generally reflects a) an appreciation of the country's inexpensive and experienced labor and or its export-oriented tariff, taxes and infrastructure, and b) low search costs due to implicit signals from established producers that Thailand is a relatively low-cost, hassle free place to do business. Once established, firms value the existence of experienced personnel and some firms deepen their activities to include various kinds of process-related research. As wage rates have risen, there has been a departure of more clearly assembly-type operations (e.g. HGA) to cheaper labor areas such as the Philippines and China.
4. Weakness of Indigenous Supply Base: Thailand's indigenous HDD supplier base is very shallow. Outside the foreign HDD and components producers, there are no firms providing the kinds of components or basic services, repair, clean rooms, or metal finishing, that one sees in Singapore and, to a lesser extent, Penang. To the degree that such goods and services, along with failure analysis and other types of production-related such as tooling, equipment research, are provided in Thailand, they are part of foreign firms' in-house operations. The Thai industry exhibits features of a "high technology enclave." A key question is whether locally based suppliers can develop the capacity for volume production integrating new technologies such as MR heads and TSA suspensions.
5. Mixed Governmental Performance: The growth and the weaknesses of disk drive production in Thailand are in large part a reflection of the country's public policies and institutions. On the one hand, the government has facilitated ease of entry with a range of financial incentives, provided relatively good infrastructure and (until recently) a stable macroeconomic context. The Board of Investments (BoI) has not only supplied export-oriented incentives; it has also helped to resolve particular problems in areas such as labor relations and infrastructure shortages. On the other hand, public policies have not promoted an indigenous supplier base. The government's trade and investment regime has encouraged a dualistic structure in which the country's large number of local firms have failed to develop the kinds of capacities necessary for participation in HDD sourcing networks. Moreover, government efforts at vendor development , technical training, and collective technology development and diffusion within the industry have been either absent or ineffective. A combination of bureaucratic fragmentation, the dynamism of low-wage exports, and private sector ambivalence over any government involvement in technology development have undermined such efforts.
6. Impact of East Asian Crisis: The impact of the East Asian crisis has been mixed. On the plus side, the Thai devaluation has reduced the cost (in dollar terms at least) of locally-sourced goods and services. But since the HDD industry uses primarily imported components, it benefits only from lower labor and operations costs. This has resulted in some relocation of labor-intensive operations from China to Thailand (by Seagate) and some expansion of process development in Thailand (Fujitsu). But for many firms, wages are not a significant percentage of total costs. Inflation has not yet risen to any great extent and wages are kept down by growing unemployment. The loosening of the labor market has significantly expanded the supply of technicians and engineers. The crisis has created what one firm called a three-year reprieve — a time to get one's house in order for the future. Finally, there is some evidence that the crisis is stimulating the Thai government towards more coherent and industry-specific policies designed specifically for upgrading. On the negative side, crisis may seriously affect demand for HDD products at the regional and global level (Fujitsu reports that one third of its disk drive sales are to Asia). It is believed that this factor is partly behind some recent closure HDD operations in Thailand, rather than the Thai crisis per se.

I. Basic Features

A. Size: The hard-disk drive industry in Thailand began with Seagate's 1983 investment in head-stack assembly, employing some 50 workers. Some fifteen years later the industry has expanded to include around 25 firms engaged in most of the industry's assembly and manufacturing activities. Although there are no official data for HDD and related production or exports, we estimate that disk drive production and exports grew from roughly $2.6 million in 1985, to $500 million in 1988, to $1.5 billion in 1990, to a little over $2 billion in 1992, to $5-6 billion in 1996, to $6-7 billion in 1997. These figures suggest that the Thai industry is around half the size of Singapore's 1994 totals.²

Hard drive production is a significant part of the country's overall electronics industry, itself key to the Thai economy. Computers and parts, which are dominated by HDDs, accounted for somewhere between 30% and 40% of the electronics industry in the early 1990s, with the percentage probably higher in 1997.³ Based on figures cited above, disk drives constituted a significant portion of Thailand's 1996 electronics exports, which amounted to $13.2 billion. Indeed, "computers and parts" constituted the largest category of Thai manufactured exports from 1992 to 1996.⁴ Seagate alone, according to one source, accounts for 4% of total Thai exports.⁵ HDD and related firms are also some of the largest employers in Thailand. Seagate, having expanded to over 44,000 workers in 1997, is the country's largest single employer (although 1998 layoffs reduced this number). At the end of 1997, Seagate's Thailand plants accounted for around half of the firm's global workforce.⁶ Minibea, with 25,000, is the second largest employer overall and the largest Japanese employer in the country.

B. Scope/Depth: The Thai disk drive industry emphasizes drive assembly, mostly for mobile and desktop computers; production of HGAs and related components such as suspensions; and spindle motors and related parts. These production clusters are owned and run almost totally by foreign multinationals. The absence of indigenous firm participation extends to key support services such as basic metalworking and finishing

1. Assembly: As seen in Table 1, three of the HDD majors - Seagate, IBM, and Fujitsu were operating in Thailand in 1998. Hyundai closed its Maxtor operation in 1996 and Micropolis closed in 1997. Neither Quantum/MKE nor WD has ever had operations in Thailand. Through these firms' operations, Thailand is becoming a more self-contained site for the assembly of drives used in desktop and mobile computers. This has two important implications. First, and most important, Thailand is not a hand-off site for manufacture of products whose volume production has already been ramped up in Singapore. Seagate's links to Singapore seem to be the most extensive, but these involve testing and final assembly. Linkages involving design and development seem to be closest to Japan (for IBM and Fujitsu) and the U.S. (for Seagate, at least in the future). Second, because desktop and mobile products often have shorter life cycles than network drives, Thai facilities may have to be even more flexible than those in Singapore.⁷ These trends are reflected in the following firm-specific descriptions:

TABLE 1 (about here).

In 1987, Seagate was producing 5 1/4" head-disk assemblies, the firm's mature product lines. As of 1991, Seagate's Thailand operations emphasized 3.5" drives, accounting for 40% Seagate's total production of these drives.⁸ By 1997, the firm's Thai facilities were producing all of Seagate's subassemblies for mobile and desktop drives, with final assembly and back-end testing in Singapore. This seems to conform to a regional division of labor in which Thailand assembles several mid-range products based on model "kits" from Singapore, while Singapore performs ramp-up and testing. However, firm officials stressed that Seagate was moving toward a more tripartite, "focused factory" division of labor in the region. Production of drives for mobile (notebook) computers will occur in Thailand, for desktops in Malaysia, Shenzhen and Wuxi, and for server drives in Singapore. Under this arrangement, Seagate's Chockchai facility will expand from HDA to HGA and HSA, as well as back-end testing for the 2.5" mobile drive. This is essentially a merging of production and assembly in order to reduce time-to-market. Thus, Thailand will do both ramp-up and mass production for mobile products, Perai for desktop products. There are a couple of implications of this shift. One is that the Thai operations, rather than working through Singapore, will be more directly linked with Seagate's design center for mobile products in San Jose (just as Perai will be directly linked to the desktop design facility in Colorado). Second, it implies that tooling and other process changes will occur in Thailand itself. To some degree this has already happened. Tooling used to be done in Scotts Valley and Omaha. It's now done largely in Asia, although largely in-house, at least in Thailand. But, as discussed below, this will not sever Thailand's head- and spindle motor supply links with other parts of the region.⁹

IBM now has two facilities operating in Thailand. Both are linked to Japan, not Singapore. The first IBM plant is actually a contract manufacturing operation under the management of a diversified Thai conglomerate, Saha-Union. This plant originally divided HDD production into low-end products for Japan and mid-high range for San Jose. When IBM opened its facilities in Singapore for mid-high range drives, the Saha-Union plant focused on 3.5" desktop products for Japan. The plant currently produces 70% of the hard drives manufactured in Thailand. Product life cycles for the Saha Union plant, as for the new Prachinburi plant, range between 6 months and 1 year; IBM introduces an average of two new models per line per year. Pilot production, which involves refining the product itself, occurs in Japan and takes roughly two months.¹⁰ De-bugging of the production process itself (i.e. ramp up) occurs in Thailand with volume production.

IBM's new facility in Prachinburi produces 2.5" drives for mobile computers. These drives come in three thicknesses: 9.5, 12.5, and 17 mm with a maximum capacity of 10 GB.< COLOR="#ff0000"> One principal difference between the Saha-Union and Prachinburi facilities is the latter's focus on managing rapid product change and new technology. The IBM plant at Prachinburi has successfully implemented cutting edge technologies designed to improve areal densities, like GMR and TSA, into the production process.¹¹ IBM's strategy seems to be to segregate production by market segment and then< COLOR="#ff0000"> to rely on changes in the electronics technology or microcode rather than platform changes. Models are differentiated through the number of disks, which changes the thickness of the drive. Each thickness requires jig modification. To manage these changes, the Prachinburi plant is designed with: 1) a special floor capable of accommodating a class 100 clean room in ways that allows easier movement, 2) more flexible assembly lines (in contrast with the more parallel lines of the Saha-Union facility), and 3) greater flexibility in the HSA process.

Fujitsu's Thai operation is described by its managers as a low-end facility, producing mostly 3.5" drives with capacities up to 10 gigabytes, although it used to produce equal volumes of 2.5" drives. Prior to 1990, Fujitsu had focused on high-end drives with low volumes but high margins produced in Nagano. Fujitsu's Thai facility represents a decision by the firm to expand into higher volume, lower-end drives (high-< COLOR="#ff0000">-capacity products are produced in Japan and the Philippines). Low-end products are designed in Japan with input from Fujitsu's tech center in Singapore. Ramp up occurs in Fujitsu's Yamagata plant, a process taking less than one month, after which the product is handed off to the Thai facilities.< COLOR="#ff0000"> Roughly one third of production goes to Japan, the U.S. and Europe respectively.

Fujitsu plans to expand their Thai facilities and make Thailand the regional focal point for HDD production in Southeast Asia.¹² However, most of this expansion is only in terms of volume capacity, not R&D. They are hoping to transfer some processes that are currently done in Japan, like media and suspensions to Thailand. Also, they hope to create a technical support center. The devaluation of the Baht has strengthened the decision to create this center in Thailand.

2. Head-Related Products: Seagate has an integrated head system in Thailand. This consists of an HSA facility at Wellgrow and two HGA facilities at Korat and Teparuk.¹³ We would presume that the firm's HGA facilities in Korat and Teparuk would supply Wellgrow. In addition, it is probable that the Lat Krabang facility's production of PCCs and coils is devoted to the HSAs/E-Blocs produced at Wellgrow, while Rangsit's poles and motors go directly into HDAs assembled at Chokchai.¹⁴ Seagate's HSA operations are well integrated into the firm's regional networks. The Wellgrow facility will reportedly supply all HSA's for the region, according to interviews. Wellgrow also interfaces with head groups in Malaysia as well as in Teparuk (Thailand).

The Thai industry also includes one of the world's major independent head producers - Read-Rite, two of the world's four producers of suspensions, and the world's largest supplier of thin film lead wire assemblies used in suspension assemblies. Read-Rite fabricates sliders (heads) and assembles HGAs in Thailand. In fact, Thailand is the firm's only mass production slider fab facility. Read-Rite also assembled HSAs in the early 1990s but moved this operation to Penang (for proximity to Singapore) and, to a lesser extent, to the Philippines. Most HGAs are thus shipped to Penang for further assembly. Read-Rite's largest customer is Western Digital (50%), then Quantum (30-40%), Micropolis (under 10%) and Maxtor and Fujitsu. Almost half of Read-Rite's sales are to Thai-based customers, 50% to Malaysian-based customers, and 5% to the Philippines. The firm reportedly intends to shift HGA out of Thailand to the Philippines, leaving the Thai facility with slider fab and failure analysis/testing (see below).

Until recently, a smaller HGA assembler— Magtric — also operated in Thailand. This firm belongs to SAE, a Hong Kong HSA producer which is in turn controlled by Japan's TDK. Magtric sold all of its product to SAE, whose production facilities are in China and design facilities in Hong Kong. The firm closed Thai operations in 1998 due to declining demand.

KR Precision and Magnecomp are the second and fourth largest producers of suspensions in the world (the first is Hutchinson Technologies, the third is Japan's Nippon Hatsujo Kogyo/NHK). KRP's only production site is in Thailand.¹⁵ Its major clients are Seagate (60%) and Read-Rite (40%), both presumably in Thailand. In addition to its Thai facilities, Magnecomp has higher volume operations in China. Magnecomp's Thai operations sell mainly to Read-Rite in Thailand. Its China facilities supply Seagate in Thailand. Remaining Thai production is sold to hard disk manufacturers in India. Finally, Innovex is the largest supplier of thin film lead wire used in suspension assemblies. Innovex supplies Boron, which has its factory in Korat, who then prepares the lead wires (number of wires, twists, connectors, etc.) to be shipped directly to the disk drive manufacturers< COLOR="#ff0000">.

3. Spindle Motors: Three major producers of spindle motors operate in Thailand.

Seagate's Rangsit facility reportedly supplies roughly one quarter of the firm's total spindle motors. Seagate also sources some motors from Nidec, the world's largest producer, and Minibea, a firm whose motor production Seagate actually helped to promote (see below). Nidec established Thai operations in 1990 to produce motors for low-end drives. High-end production has remained in Japan, but some of this production will shift to Singapore for reasons of proximity to high-end drive producers. Nidec's Thai operations consist of three facilities, each dedicated to a particular group of customers which include IBM, Seagate, Fujitsu, Maxtor, WD and Toshiba. Half of Nidec's Thai production goes to Thai-based drive producers, 40% to Singapore, and the rest to Japan and Europe. The firm also has low-end production in China and the Philippines (to serve Toshiba's new facility), but its Thai factories constitute Nidec's major low-end operation.

According to interviews, Nidec plans a significant shift in responsibility from Japan and Singapore to Thailand. For example, the Thai operations have expanded their responsibility for debugging. Until two years ago, ramp-up to volume production took place in Japan. Because this took too much time to get product to market, Japan now produces only around 1,000 pieces, after which debugging for high volume production occurs in Thailand. This saves an estimated three months in the process of getting the product from design to market. It is also consistent with the already-mentioned tendency for Thailand to specialize in low-mid range products through direct links to Japan or the U.S. In addition, the President of the Thai operations will become the regional chief executive and will make production decisions and give general direction to the other locations in the region including Singapore, China, the Philippines, and Malaysia. Japan felt that it was out of touch with the customer and that product cycles and technology demanded that the key decisions be made on at least a regional, if not local, basis. This is a major initiative to decentralize operations and give autonomy and decision making power to each of the locations. This change has begun in the last year and will include changes in process, procurement and, in the future, some product development.

The third large motor producer is Minibea (NMB). Thailand is Minibea's principal spindle motor production facility, although the firm also produces motors in Shanghai (check). It sells to most if not all the HDD majors. Finally, at least three Japanese firms - Habiro, Nippon Super, and Hsin AE - have set up operations in Thailand at Nidec's urging to supply base plates, hubs, and brackets for spindle motors. Another firm, TPM, reportedly supplies baseplates to Seagate.

4. Weak Indigenous Supply Base: If the Thai HDD industry is relatively broad, it is also very shallow. Outside the foreign HDD and components producers, there are no firms providing the kinds of components or basic services such as tooling, equipment repair, clean rooms, or metal finishing one sees in Singapore and, to a lesser extent, Penang. To the extent that such goods and services, along with failure analysis and other types of production-related research, are provided in Thailand, this occurs only through the foreign firms' in-house operations. This shallow production base reflects an almost total absence of locally owned and controlled firms in the HDD industry. Hana and KR Precision are both Thai based and listed on the Thai exchange. But both are run by expatriates and seem to have very little Thai input into basic managerial decisions. Avatar is also Thai owned but run by an American.

The only significant local presence is in PCBs, but even here the picture is relatively bleak. KCE is reportedly one of the 20 largest PCB firms in the world. But 90% of its product is exported and seems to have relatively weak ties with the HDD industry. Two contract manufacturers, SCI and GSS, are also significant PCBA producers. Although both were active suppliers in the early stages of Thailand's disk drive industry, neither has much to do with the industry at present. From the perspective of local PCB firms, low margins and demanding schedules made production for disk drive firms less attractive than other alternatives. The need for rework also encouraged the HDD assemblers to develop their own, in-house capacity.

C. Evolution: Two trends stand out with regard to the industry's overall evolution: a spurt of assembly growth in the late 1980s, and a general tendency for foreign suppliers of HGA-related products and spindle motors to follow the assemblers.

Although Seagate began a small HSA operation in 1983 (Table 2), its main HDA assembly operations began at Chokchai in 1987. This was followed by three drive producers in 1988: Micropolis, Fujitsu, and IBM (Saha Union). The only recently established assembly facility, IBM (Prachinburi) began operations in 1997, and this was in large part due to the availability of experienced personnel from IBM's Saha Union operation. The growth of components production has largely accompanied or followed assembler growth. Seagate's initial HSA facility stimulated Minibea's expansion into spindle motor production in 1985 (Minibea had been producing ball bearings in Thailand since 1980). GSS Arrays, a contract electronics manufacturer, set up PCBA and HGA operations in 1985.

Seagate's establishment of a full-scale HDA facility at Chokchai in 1987 resulted in further supplier development: KRP and TPW were established to produce suspensions in 1988 and housings in 1989 respectively. Seagate began a move into components in 1989 by opening the Teparuk facility to produce HGAs. This was followed by two other head-related firms: Magtric with an HGA facility in 1990; Read-Rite, which established slider/HGA/HSA operations in 1991 (initially through the use of GSS facilities and subsequently in its own facilities within an EPZ in 1993); and Magnecomp to produce suspensions in 1992. Nidec began spindle motor production in Thailand in 1990, although it had supplied Seagate during the 1980s. Seagate itself established a facility to produce spindle motors in 1994 at Rangsit, as well as two other components facilities that same year: - Lat Krabang for coils/PCCs and Wellgrow for HSAs. Seagate's most recent plant is its 1996 facility in Korat to produce HGAs. The growth of spindle motor production, especially by Nidec, led to the entrance of Japanese firms supplying hubs and parts for motors (Habiro, Nippon Super) in 1995-96. In the mid-1990s, HGA and HSA production stimulated the entrance of one of the world's largest supplier of thin film lead wire assemblies, Innovex, through one of its subcontractors, Boron.

Thailand has some limited media production, the most important of which seems to be Asahi Komag's 1996 facility to produce plated, polished substrates for use in final manufacturing at Komag's other facilities, as well as Hoya Opto's plant in Lamphun which produces 2.5 inch media for IBM.

II. Explanatory Factors

The growth of disk drive production in Thailand, what we have termed "dispersion," is a result of several factors. These include an inexpensive, accessible, committed and experienced work force; proximity to clients and suppliers; and a policy environment that has allowed drive producers to operate without constraints while providing macroeconomic stability and export-oriented incentives and infrastructure. At the same time, the shortage of skilled personnel (at least prior to the 1997 downturn), the weakness of the local supply base, and the lack of industry-specific infrastructure and services have inhibited the industry's deepening in Thailand.

In assessing the importance of different motivations, one point merits emphasis: Firms made investment decisions on the basis of several factors, with the particular hierarchy or combination of considerations varying over time. For example, Read-Rite seems to have made location decisions concerning HGA and slider on the basis of labor cost and experience consideration, whereas head-stack assembly was located in Penang for reasons of proximity (to Singapore). Similarly, whereas low labor costs and the general business may have been primary at the point of entry for Seagate, for example, the experience of the labor force and links to clients and suppliers subsequently gained in importance.

A. Labor: The labor market has been a key factor in the growth of HDD production in Thailand. Wage rates were and continue to be important, but labor market fluidity (i.e. ease of recruitment) and the general availability of experienced personnel have been also been key, especially when it comes to explaining decisions by firms to stay and/or expand operations in Thailand. Indeed, based on the availability of experienced personnel, Thailand seems to be occupying a mid-range niche above China and the Philippines, more or less even with Malaysia, but significantly below Singapore.

      1. Labor Costs and Availability: Thailand's relatively low wage rates clearly played a key role in Seagate's decision to shift HSA production from Singapore to Thailand in 1983 and to expand into assembly in 1987. During the first half of the 1980s, labor costs remained low in part due to the fact that this was a recessionary period and in part because the government kept minimum wage changes below the inflation rate.¹⁶ As Table 2 illustrates, Thai labor rates in 1985 were almost $5,000 less than those of Singapore, and over $8,000 less in 1990. And despite the increasing importance of accessibility, experience and skill, wage rates continue to influence investment decisions.

IBM decided to set up its new plant in Thailand in part because Singapore was too expensive, Malaysia had sharp labor shortages, and Thailand offered better incentives than the Philippines for firms establishing facilities outside Bangkok. Where operations are both labor intensive and require little skill or experience, the trend is to shift production out of Thailand. This is reflected in Read-Rite's decision to move HGA production to the Philippines (noted earlier), and the new IBM plant's sourcing of HGAs from China and, to a lesser degree, Mexico.¹⁷ The importance of wage rates may also vary within firms producing a similar product, albeit with different core competencies.¹⁸

But low wages are not sufficient to account for 1) Seagate's choice of Thailand over the Philippines or Malaysia, especially in the late 1980s (since Malaysian rates were somewhat lower than Thailand's by 1990), or 2) subsequent industry expansion. While part of the explanation involves the general investment climate (see below), the rapidity of start-up in Thailand was equally if not more important. This was in turn due to the easy recruitment of mature workers and trained personnel. This is reflected in Seagate's capacity to begin its initial operations in 1983, and its subsequent establishment of the Chokchai (1987) and Korat (1997) operations. According to the Seagate official in charge, the Chokchai plant was producing as many drives within six months of startup as it took Singapore to do from 1983-87. The Korat plant was producing in volume six months after construction was begun. The commitment and flexibility of the workforce was also important, especially in the early days. One Seagate official noted that Thai labor allowed Seagate to "go beyond the normal;" the official could call his workers in the middle of the night and get them to work extra hours to meet a special order.¹⁹

TABLE 2 (about here)

Of course, Thai labor markets tightened considerably in the early 1990s. Seagate experienced a strike in the Teparuk plant in the late summer of 1991. The strike itself resulted from a combination of workers' concerns with high exposure to lead and, according to Seagate, support for strikers from a party faction backing the military government then in power.²⁰ The strike also reflected Seagate's difficulties with high staff turnover at the time. Indeed, Seagate's move to Korat in the Northeast reflects the difficulty in recruiting workers in Bangkok and Seagate's efforts to set up operations closer to the regional origins of most Thai manufacturing workers (as well as attractive incentives for investments outside of Bangkok).²¹

These difficulties do not seem to have been much of an obstacle to the industry's expansion, however. Seagate established four of its six Thai operations after 1990, the period during which wages increased at an annual rate of almost 10% compared to a rate of around 2% during the 1982-90 period (Table 3); and workers in the electronics industry were getting annual pay raises of 25% prior to the recent slump.²²

TABLE 3 (about here)

      2. Labor Skills and Experience:²³ New investments following the tightening of the labor market were facilitated by higher salaries, greater attention to human resource issues, and recruitment of engineers from the rest of the region by firms such as Seagate and Minibea. But continued investments also reflected Thailand's pool of trained engineers and supervisors, and experienced workers. Until 1988, when Read-Rite and Micropolis came in, Seagate had the Thai labor market to itself. Thailand offered "a large pool of unemployed but trainable graduates of secondary, vocational and commercial schools..." as well as experienced supervisors and engineers educated in an expanding vocational system and trained by the electronics (IC) companies established earlier in the 1970s (see below).²⁵ And during the early-mid 1980s, when neither the overall economy nor the labor market for engineers was very active, trained personnel tended to remain with their firms. This proved critical for reducing the costs of on-the-job training and consequently reducing defect rates throughout electronics. This was especially true for IC, PCB and disk drive firms such as Seagate, NMB, GSS and SCI.²⁶ According to one source, Seagate valued Thailand for its greater availability of workers with a 9th grade education than was the case for Malaysia.²⁷ Indeed, a Singapore-based Seagate official argued that the pools of trained managers in Thailand constituted an important attraction for Seagate. The availability of experienced personnel has meant that prior to its 1998 layoffs, Seagate had less than 20 foreign executives out of its 44,000 employees in its Thai operations.²⁸

It is, however, important not to exaggerate the strengths of Thailand's technical personnel. Lack of high-end competence helps to explain why Seagate retreated from initial plans to do complete assembly and back-end testing in Chokchai soon after the plant's opening in 1987-88. Yields in Thailand could not compete with those in Singapore where workers knew how to "tweak," and tweaking is critical to keeping yields high.²⁹ In addition, Read-Rite's recent difficulties with regard to volume production of MR heads seems to reflect weaknesses in technical personnel.

In sum, at least until the crisis, the costs, accessibility, and experience of Thailand's labor force encouraged an expansion of drive and related production. At the same time, the increasing tightness of the labor market and the fundamental lack of engineering education (about which more below) imposed certain limits on the potential for higher value added production.

The 1997 economic downturn may, however, result in a significant expansion in the pool of technical personnel. The crisis has encouraged a consolidation of the industry in Thailand. Several companies have closed down and others have scaled back operations. This has resulted in a surplus of experienced, unemployed engineers. At the same time, graduates of new engineering programs from several universities, notably KMITT, have begun to enter the job market in increasing numbers. The availability of engineering talent is allowing several companies to expand their presence in Thailand and position themselves to take market share when the market rebounds. Further, the expanding pool of technical personnel, when combined with Thailand's lower costs, may facilitate an expansion of higher value added process research and development activities to Thailand by firms such as Nidec and Magnecomp.

B. Clustering and Proximity: The particular evolution of the Thai HDD from assembly to major component producers owes much to a pattern of clustering of foreign firms. This section first traces the origins of these supplier clusters to three factors: follow-the-leader behavior, positive externalities (trained/experienced personnel), and decisions by large firms to leverage their own capabilities and strengthen the local supplier base. The section then reviews the benefits of clusters and proximity. Section C explores the limits to those benefits. Thai-based firms do obtain some benefits from proximity to each other, and in that sense, Thailand's disk drive producers constitute an agglomeration economy in certain respects. But Thailand differ in at least two key ways from the more stylized agglomeration story (and the patterns seen in Singapore and, to a lesser degree, Penang): more geographically extensive sourcing patterns, and low levels of pooling of productive inputs.

      1. Follow-the-Leader: Successful operations by already-established firms reduce search costs and risks to new entrants by providing valuable information as to the ease of doing business in a particular locale. This kind of dynamic has been critical throughout the expansion of disk drive production in Thailand. Read-Rite and Micropolis both began operations in Thailand in 1988 without much extensive research on other countries. With wage rates still reasonable, with labor shortages in Singapore and Penang, and with the Philippines uncertain, they moved to Thailand in response to Seagate's demonstrable success and expansion in the country.³⁰ Seagate's success (and the fact that Fujitsu already had a production facility in the country) was also a key factor in Fujitsu's decision to set up production of low-end drives in Thailand, although yen appreciation, the BoI's incentive program, and the availability of experienced operators also played a role. Similarly, IBM's decision to set up a new assembly operation in Prachinburi is due in part to government incentives, but equally if not more importantly to increasing capacity of the Saha Union plant in dealing with rapid product cycles and rapid ramp up to volume production (see above).

      2. Labor-Market and Facility Externalities: The ability to launch new operations was greatly facilitated by operators and technical personnel, managerial networks, and facilities generated by other disk drive and related firms. For IBM, the "accumulated manufacturing knowledge" available in the IBM/Saha Union plant facilitated the establishment of a new operation in Prachinburi for two reasons. First, it demonstrated the capacity of Thai personnel to deal with the volume and product mix shifts common to disk drive production. Second, the success of the Saha Union plant will allow the new plant to train its new hires with an experienced workforce in Thailand rather than sending them back to Japan for training. This represents a significant savings in time and money. The importance of an experienced workforce reportedly also weighed heavily in Read-Rite's decision to shift more labor-intensive operations to the Philippines while using Thai operations for mass production of sliders, for the introduction of MR heads, and for failure analysis and other performance testing functions.³¹ Finally, the value of an experienced labor and managerial force is reflected in the rapid start-up of Seagate's operations in Korat and in Seagate's decision to have its Thai operations manage the firm's new HGA operations in the Philippines (Cebu).³²

Also important was the emergence of higher-level management networks. The contribution of the earliest electronics multinationals in Thailand to the provision of experienced personnel for the disk drive industry was important. These IC firms, especially NS and Data General, became "the main training ground" for the country's electronics industry, including Seagate. Interviewees all noted that, at higher management levels, all the actors know each other. Several notable cases emerged in interviews.³³

Finally, the availability of existing facilities combined with a trained workforce to encourage investments. With relatively few funds, Read-Rite avoided the delays of obtaining licenses, hookups, and workers by renting GSS facilities and providing some engineers, specifications and inputs. When Read-Rite eventually set up its own operations, it hired some 70% of GSS personnel. Micropolis, for its part, began with land purchased from GSS. Intending to produce HSAs for the high-end market, Micropolis encountered problems in recruiting experienced workers. The firm's capacity to outsource HGAs from Read-Rite and another firm (Desktech?) helped to offset this problem. Most recently, in 1996 ADFlex decided to move to Asia to escape the high costs of producing in the UK and for proximity to customers. It decided to produce in Thailand, rather than in China or Malaysia, because of the availability of Hana's experience and its available facilities in northern Thailand.

      3. Leveraging Capabilities and Proximate Supplier Development: As noted earlier, Thailand lacks the kinds of metalworking, heat treatment and related suppliers seen in Singapore and Penang. A 1992 report on the availability of local equipment for information industries states, for example, that the country's machine tool industry had actually shrunk despite growing demand and a long period of protection. In general, Thai industry is distinctly dualistic, with technologically advanced firms producing for export alongside weaker Thai firms importing CKD parts for products to be sold in the domestic market.³⁵

We address the reasons for this gap below. Suffice it to stress here that the several large IC firms such as National Semiconductor operating in Thailand since the early 1970s did not generate indigenous spin-offs. They were, as noted above, important sources of skilled personnel for other foreign firms, but they did not spawn indigenous entrepreneurs.³⁶ There was, as a result, essentially no supplier base when Seagate arrived in Thailand in the early 1980s. Since then, Seagate and almost all of the foreign producers have made systematic attempts to identify and/or nurture local suppliers. To our knowledge, none have succeeded.³⁷

Conversely, a foreign-dominated proximate supplier base has emerged as a result of initiatives first by Seagate and subsequently by other firms. Ironically, the most critical component of this process has been Seagate's efforts to develop internal capacities in spindle motors and head-related components, especially suspensions.³⁸

     Spindle Motors: In the early-mid 1980s, Seagate sourced motors from Nidec (presumably in Japan), but wanted to produce its own motors as leverage to drive down Nidec prices. Seagate designed its own drives in Scotts Valley but, lacking manufacturing space, transferred production to Minibea (NMB in Thailand). The Seagate/NMB linkage in and of itself reflects the benefits of proximity. NMB had been in Thailand since 1980 producing ball bearings, some of which it sold to Seagate. The relationship between the firms had grown when the head of NMB visited Seagate's initial plant to learn how the firm could manufacture in such small facilities. All of this led Seagate to encourage NMB to begin producing spindle motors with Seagate specs. NMB became a motor supplier for Seagate. And although Seagate itself took most of the motor production back in house (with the 1994 Rangsit facility), NMB has remained an important motor producer.

Subsequently, in 1990 Nidec itself established its largest overseas production facility in Thailand, despite the fact that in the 1980s a high Nidec official pledged that the firm would never go offshore. It did so, according to a non-Nidec source, under pressure from Seagate. Given the industry's volatility, Seagate (and other majors) cannot afford to have a geographically long supply chain.³⁹ A Nidec official noted that while Philippines and China would become more attractive if the Thai minimum wage increased further, Nidec has to be here as long as the customer is here. Nidec's first plant is devoted to Seagate and IBM products; the second supplies IBM and Fujitsu; and the third supplies Maxtor, WD and Toshiba.

Nidec has in turn brought other firms to Thailand. Habiro, one of the world's three firms producing spindle motor hubs, came to Southeast Asia at the "encouragement" of Nidec, which takes 60% of Habiro's product (NMB takes another 20% and JVC 10%). In essence, Habiro is part of a Nidec-organized, mini supply chain involving the diecasting of baseplates by a Japanese firm (Shin AE) in Thailand's northeast province of Korat, the machining of these baseplates and motor parts, and the assembly of the baseplates and motors by Nidec.⁴⁰

     Suspensions: A similar dynamic occurred with regard to suspensions, a component that looks simple but is both critical and technically demanding.⁴¹ When Seagate set up its initial HSA operations in Thailand in 1983, the firm was getting HGAs from Korean firms at prices it considered too high. Seagate began sourcing HGAs from a Philippine-based firm but subsequently duplicated this firm's assembly arrangement in Thailand. This led to further efforts at cost reduction and quality improvement, part of which involved pressure on the world's largest and most advanced suspension producer, Hutchinson Technology (HTI), to set up facilities in Thailand in the interests of providing better service, engineering support and pricing. HTI refused to establish overseas operations. In response, Seagate initiated what turned out to be an unsuccessful effort to develop in-house capacity in suspension manufacturing. The firm then looked to create a new suspension producer. It found a precision etching firm in Japan - Kyosei, and, through the Thai Board of Investments, linked it up with a Thai real estate development company, Ramon. It is worth emphasizing that Seagate had actually attempted but failed to find a Thai mechanical production firm as a partner for Kyosei. Ramon's functions involved land procurement, government relations, and the hiring of local engineers. The resulting firm was KR Precision, established in 1988 and operational a year later. Highly dependent on Seagate initially, KRP has subsequently become an independent supplier. This growth nicely illustrates the interaction between Seagate's supplier development initiatives and the existence of other KRP clients operating in Thailand due to follow-the-leader dynamics.⁴²

What specifically are the benefits of supplier proximity?

     Facilitates rapid, JIT deliveries of supplies in small volumes: Nidec delivers to customers in Thailand once a day compared to once every week and a half to customers outside of Thailand. By virtue of its location in an industrial estate 30 minutes from Nidec's facilities, Nippon Super is able to make six deliveries per day of around 100,000 items to Nidec. Shipping from Japan would take 3-4 days by air. This can be especially important for firms supplying diverse customers with different specifications. Changes in such specifications can be very time consuming without local sources.⁴³

     Promotes rapid and frequent information exchange between client and supplier: Such exchanges can alert and prepare suppliers for upcoming programs, an issue emphasized by Nidec and KRP in their relations with Read-Rite and Seagate. Given that suppliers are often involved in attempting to qualify for several programs at any one time, such exchanges are important. KRP has been able to win some programs because it is able to develop prototypes more quickly than the industry's technology leader, Hutchinson Technologies. By facilitating information exchanges with clients, proximity has facilitated this quickness. A Magtric manager noted the advantages of proximity for facilitating frequent changes in materials and specs required by customers.

     Facilitates rapid resolution of quality problems: With increasingly short time frames for ramp-up and short product runs, the need to quickly respond to any problems experienced is necessary. Read-Rite has established relatively sophisticated production-oriented technical capacity in Thailand specifically to deal with ramp-up and production problems at the source. Read-Rite has also set up a failure analysis lab for sliders. This facility is one in a hierarchy of labs ranging from those focusing on HSA and HGA problems in Penang and the Philippines, to the Thai lab's focus on problems inherent in the slider fabrication process, to facilities in Japan and the U.S. capable of testing for wafer deficiencies. Nippon Super technicians are at Nidec facilities every other day to deal with scratched or out-of-standard parts. NMB set up what it calls an "R/D" facility in August 1995 to test products for dust, outgassing, etc. An NMB official stressed that five years ago such problems could be dealt with by sending products back to Japan. Problem solving is especially important for Magnecomp in its relations with Read-Rite. Completed suspensions are checked twice before assembly into HGAs: once by the supplier - Magnecomp, and once by the client - Read-Rite. The real challenge involves "correlation problems," i.e. situations where the product checks out OK at the supplier plant but measures substandard when checked at Read-Rite's facilities. To see whether the product is in fact faulty or rather a function of a poorly calibrated measuring device, Read-Rite has begun to implement a "self-source inspection" process: Read-Rite trains Magnecomp inspectors who remain Magnecomp employees but are stationed at the Read-Rite plant.⁴⁴

In sum, there seems to be a growth of on-site testing facilities and an intensification of supplier/client relations under pressure for both speed to market and reliability. However, these kinds of proximate linkages are not consistent across or even within firms. There are limits to the attractions of proximity, as illustrated by a particularly striking example: KRP was located literally next door to an HGA producer, Magtric (recently closed); yet, for reasons discussed below, KRP did not supply suspensions to Magtric.

C. Limits to the Agglomeration Economy Explanation: For all its benefits, proximity-based clusters in Thailand differ from a classic agglomeration economy story in at least two ways. First, the value chain of which Thai-based firms are a part is more regional and global than national. Second, Thai-based firms do not engage in much pooling of productive inputs. What pooling does occur has less to do with public or collective arrangements than with large firm efforts to develop a supply base and general infrastructure.

TABLE 4 (about here)

1. Extra-National Relations: Although incomplete, Table 4 provides a sense of the geographical range of interfirm relations in which Thai-based HDD firms are involved. Following is a list of factors that can help to explain interfirm relations of a more regional and global nature.⁴⁵

2. Limited Resource Pooling: The "benefits of proximity" story is also

weakened by the fact that conscious resource pooling among HDD firms in Thailand is quite limited. A Read-Rite manager with several years experience in Thailand explicitly contrasted Thailand with Singapore where, he noted, firms share information and, at times, machines and raw materials. In Thailand, on the other hand, "we don't share anything, not even epoxy."⁵⁰ The one area where some pooling occurs involves labor market information. There are three associations in which HDD and related firms participated: the Ayuthya Industrial Association, the Electronics Components Employers Association, and AJ Khai. The most active seems to be the AIA, a group comprising some 17 firms in the Ayuthaya area. The group meets regularly 1) to exchange views and develop a consensus on how extensively to apply minimum wage increases, i.e. whether the increase should be applied evenly to all workers; 2) to develop a no-poaching agreement; 3) and to exchange information on wage levels, bonuses and job descriptions. This group thus facilitates monitoring and reduces the costs of acquiring information in what was, until the summer of 1997, a very tight labor market, especially among a geographically concentrated group of firms.⁵¹ To our knowledge, there are no linkages between these foreign-dominated associations and the more Thai-influenced electronics associations operating within the Federation of Thai Industries.

Despite some private sector initiatives, there is almost no pooling of equipment or production-related knowledge. In one effort, a Seagate official attempted to promote a center for testing and failure analysis by donating equipment to the Asian Institute of Technology. The effort never got off the ground - indeed the equipment was never unpacked - because, according to one source, AIT lacked the resources and desire to promote the effort.⁵² In another, KRP attempted to establish a tool and die institute. As a collective effort, this failed. But KRP persevered and established a totally in-house training program. This is a four-year program that was established not by a Thai but by an Indian graduate of an Indian tool and die institute. Most of the more experienced toolmakers in the firm are from India. In a clear illustration of the externality problems involved, Thai trainees have tended to depart for the auto sector after two years of training.⁵³

D. Country-Specific Features and Public Policies: Disk drive producers have made decisions within a context heavily influenced by public policies. In Thailand this context is highly laissez-faire, at least in comparison with Singapore and Malaysia. The Thai government has earned relatively high marks for its generic policies in areas such as macroeconomic stability, labor costs, overall infrastructure, and export incentives. Conversely, the government has been strikingly weak in policies designed to promote a more industry-specific supply infrastructure of indigenous suppliers, skilled labor and industry-wide technology diffusion. A Thai quality assurance director at Micropolis summed up the impact of public policies in stating that Thai government incentives "are for investors, not technology developers." Along the same line, an official at Magnecomp said that when he worked in Hong Kong, his main focus was on engineering issues. In Thailand, he spent most of his time on personnel and regulatory issues.

      1. Non-Industry Specific Supply Infrastructure: When combined with easy access to an experienced labor force, Thailand's reputation for macroeconomic and political stability constituted an important attraction for HDD production into the mid-1990s. With its currency devalued in 1984 and anchored to the dollar, and with state macroeconomic agencies known for their aversion to inflationary policies, the factor and product market costs of doing business in Thailand were reasonable and clear. As one Read-Rite executive noted, Thailand was also attractive in terms of non-economic costs, especially compared with Malaysia and the Philippines. Politically, Thailand was viewed as more attractive than the Philippines. Whereas martial law under Marcos was devastating the Philippines, Thailand was ruled by an efficient, semi-democratic government during the 1980-88 period. And although the country subsequently underwent several government changes, one involving widespread demonstrations against the military, the bureaucracy has been relatively stable.

The key bureaucratic actor has been the Thai Board of Investment, the agency responsible for investment screening and incentive provision. The Board's principal instruments have been tax and tariff exemptions. Traditionally, it has not targeted specific industries.⁵⁴ But, beginning in the early 1970s and continuing into the present, the BoI has been one of the key initiators of export promotion policies, including bonded warehouses and EPZs, streamlined customs procedures, tax and tariff exemptions and rebates, and cuts in electricity costs.⁵⁵ The government also expanded sea and air transport facilities, a feature mentioned by officials at Fujitsu and IBM as an important attraction. When combined with labor market policies and financial incentives (see below), these measures helped to stimulate the early investments by Minibea and Seagate that led to subsequent investments. Moreover, HDD firms have viewed the BoI as highly flexible and generally supportive, if not all that well informed about the technical aspects of the HDD industry. Several firms mentioned the BoI's willingness to act as a problem solver in areas such as labor relations (strikes against Seagate and Magnecomp), and infrastructural bottlenecks. In addition, the Board has pushed for continued across-the-board tariff reforms and has resisted Cabinet tendencies to restrict incentives for foreign investors.

There are, however, important limits even regarding these kinds of generic policies.

     a) Infrastructure: Despite the intention of the 1972 Export Promotion Act, only one EPZ had been established as of the mid 1980s.⁵⁶ Several zones have subsequently been built, and firms such as Read-Rite, Seagate, and Nidec occupy them. But in several cases, a lack of government capacity has pushed the private sector into an important administrative role in these zones. With increased constraints on government expenditures imposed by the World Bank and IMF rescue package, the private sector will have to take an even greater role in economic zone development and in the provision of roads, energy, water and other kinds of infrastructural services.⁵⁷ Problems in these areas have already been noted for Seagate's new plant in Korat, and an IBM official noted that power typically goes out a couple times per year, creating difficulties for testing. On the other hand, a Read-Rite official saw Thailand as attractive compared to the danger of power outages in the Philippines due to hurricanes and volcanoes.

     b) Incentive Implementation: Indirect exporters, especially smaller firms, face problems in duty rebates.⁵⁸ VAT rebates, part of export incentive packages, typically take 6 months, a significant burden for small firms. Indeed, Thailand's trade administration, especially Customs, is generally inefficient.⁵⁹ And established firms desiring to qualify a new project for incentives must state their plans several months in advance of implementation. This can be difficult due to the frequent changes in product mixes and processes typical of the HDD industry. The BoI takes an especially long time to approve projects without local content. And while the brunt of this administrative weakness seems to fall on small firms, officials can be rigid and picky even for Seagate.⁶⁰

     c) Bureaucratic Weaknesses: The preceding problems are in part a function of bureaucratic fragmentation and conflicting constituencies. The Ministry of Industry, with its ties to more protected, local firms often conflicting with the BoI. We shall note other examples of intra-bureaucratic rivalry. But the problem is exacerbated by the BoI's lack of expertise about particular industries, including disk drives. For example, several firms mentioned temporary delays in obtaining BoI permission for tariff exemptions on imported equipment due to the Board's ignorance regarding the function of particular equipment. This lack of expertise is in part responsible for what some observers see as a total lack of discrimination in the BoI's approach to investment promotion. And while this has had the virtue of encouraging competitive market structures, it has not contributed to the growth of an industry-specific supply base, about which more below.

The overall impact of Thailand's macroeconomic shifts, especially the Thai baht's 40% depreciation since its flotation on July 2, seems mixed. The downside of the depreciation is of course higher import costs, and since HDD production is highly import dependent, this may be considerable. To some degree the problem is attenuated by the fact that many HDD firms' transactions are dollar denominated. But even Seagate, a highly vertically integrated firm, has some 20 locally based suppliers, and these firms expect to raise their costs. If they do so by the full amount of the depreciation, Seagate's own costs would increase 5%.⁶¹ On the upside, the depreciation makes HDD exports cheaper. KRP is taking advantage of lower export costs to open a third plant and increase staffing by 500.⁶²

     2. Financial Incentives and the Weakness of Indigenous Supply Base: Public financial incentives, administered by the BoI, have had decidedly mixed results. On the one hand, they have been quite effective in attracting HDD firms to Thailand. These incentives were part of a series of export-promoting measures that began in the early 1970s in response to the balance of payments problems caused by import substitution industrialization policies and declining prices for raw material exports.⁶³ All HDD firms interviewed stressed the benefits of duty reduction for machinery, corporate tax exemption, and duty exemption on raw materials for firms exporting over 30%. While the BoI does not officially tailor incentives for specific firms, it does have the capacity for discretionary incentives.⁶⁴ Also attractive have been BoI decisions to lower minimum required investments and to relax preferences for local majority joint ventures. Perhaps most important, drive firms have also taken advantage of an expanded incentive package designed to encourage investment outside of Bangkok in what is known as Zone 3.⁶⁵ The manager of IBM's new plant, for example, cited Zone 3 incentives as one of two major factors in his firm's decision to locate IBM's new plant in Thailand as opposed to the Philippines (the other factor was the accumulated manufacturing knowledge of Thailand's workforce). In sum, public financial incentives encouraged the growth of clusters in heads and spindle motors by reinforcing efforts of larger firms such as Seagate and Nidec to develop proximate suppliers.

But these incentives are weaker than those found in neighboring countries. According to a recent Minister of Industry, Malaysia's incentives are about 10 years ahead of Thailand's in terms of low-cost funds, tax privileges, factories and equipment and government participation.⁶⁶ Even more serious, Thai incentives have contributed to the (earlier noted) problem of a local supply base largely isolated from export-oriented drive producers. This "de-linked dualism" is ironic: by the late 1980s, Thailand was acknowledged to have a relatively deep base of supporting industries in metalworking, plastic parts, tool and die production, and electrical and electronics components.⁶⁷

How have financial incentives led to this dualism? The key is a trade and investment regime best described "insulated export promotion."⁶⁸ In the early 1970s, the 1980s, and the 1990s Thai officials responded to balance of payments problems with a series of measures designed to attract export-oriented foreign investment. These include tariff and tax exemptions, customs reforms, EPZ development, and relaxed ownership requirements noted above. However, until the early 1990s, these export-oriented reforms were grafted on to protection for local firms producing for the domestic market. Several aspects of this regime help to explain the growth of a foreign-dominated drive industry de-linked from local firms.

     a) Discretionary Protection: The 1977 revisions to the Export Promotion Act granted the BoI not only the right to provide tariff and tax exemptions to exporters, but also the power to impose import bans and surcharges. The revisions also expanded the BoI's membership to include more cabinet ministries, including those whose clients would benefit from such protection.⁶⁹

     b) Tariff Escalation: Overall tariff levels actually increased during the 1970s and 1980s, even as the government proclaimed its shift to export promotion. Nominal tariffs rose from 35% to 30-55% during the 1970s, with the highest rates on consumer goods, and vehicles and textiles having the highest effective rates. By 1985, the effective rate of protection for Thai manufacturing was 52%, nearly twice that of the Philippines, Malaysia and Korea. Efforts to restructure tariffs during the latter 1980s foundered on the opposition of private firms in import-competing industries, their allies in the Ministries of Industry and Commerce, and a Ministry of Finance reluctant to lose revenues generated by tariffs.⁷⁰ What emerged, as Michael Rock has noted, was a "combination of rising effective rates of protection and countervailing export subsidies..."⁷¹ Significant tariff liberalization began to occur in the 1990s, with a wide range of cuts targeting the electronics industry in 1995. But the implementation of this reform has taken time due to inter-industry conflicts of interests.⁷²

     c) Cascading Tariff Structure: A tariff structure in which the highest rates were on final goods, encouraged local firms to concentrate in final assembly since local suppliers had to pay duties on their own imported inputs. And in most industries, including electronics, knocked-down assembly kits paid lower tariffs than individual components, further discouraging local assembly firms from unpackaging their kits and obtaining parts locally. To remedy this problem, the government extended duty offsets to indirect exporters in the late 1980s. But there remain obstacles to local sourcing within the tariff structure: A BoI official recently noted that local production of electronics components is impeded by the fact that imported parts are taxed at 1% compared to 20% for materials. Moreover, the government designed its 1995 cuts in import duties on electronics items "in order to encourage foreign investment" rather than to promote local electronics development.⁷³ And finally, despite considerable progress in speeding up customs procedures during the 1980s, indirect exporters, who paid import duties up front and later reclaimed them from the Customs Department, still had to submit extensive paperwork to prove that the imported goods had in fact been processed and sold to a final exporter. These weaknesses in trade administration have constituted a significant cost for multinationals and SMEs alike, as reflected in the lead wire producer noted earlier.

     d) Business Tax and Vertical Integration: The business tax undermined links between local suppliers and foreign producers because it was levied not on the value added but on the complete value of inter-firm sales. Any product manufactured through inter-firm subcontracting suffered a higher tax burden than one produced in-house. A VAT tax replaced the business tax in 1992, but at least one result was a growth of fraudulent export claims in order to obtain VAT refunds.⁷⁴

     e) Lack of Technology Incentives: BoI promotion has provided little if any incentive for firms to utilize technology transfers. Indeed, in the midst of the country's financial crisis, the government established a committee to help the BoI shift from provision of promotional incentives, which must be phased out under WTO rules over the next five to eight years, to enhancing industrial competitiveness.⁷⁵

     3. Weakness of Industry-Specific Development Policies:. We have noted the general lack of industry-specific policies.⁷⁶ This is especially important with regard to electronics. As recently as September 1997, the head of the principal domestic electronics association claimed to be "confused by the government's policies." Differences existed between industry and commerce ministries as to which industry to promote and between the cabinet and electronics association as to which electronics products merited specific support.⁷⁷ This confusion is reflected more concretely in the areas of vendor development, training/skill development, and collective technology institutions.

      a) Vendor Development: By the early 1990s it was clear that very few indigenous suppliers could meet the quality required by foreign electronics firms. A 1990 report found that Sharp and Seagate had tried but failed to find even simple metal and plastic injection molded components from local firms. IBM, Fujitsu, ADFlex, and Nidec also failed to find suppliers for more complex tools. The lack of a supplier base also become a concern for the Finance Ministry since the influx of FDI in the late 1980s had expanded the current account deficit. In response, the Ministry began to advocate the substitution of imported intermediate goods in several manufacturing industries and pressured the BoI to impose stronger local content criteria in new foreign investment projects.⁷⁸ With a couple of exceptions in autos and consumer electronics (condensers and cathode ray tubes), the Board never did use these criteria for awarding incentives. But in 1991 officials within the BoI initiated a non-incentive-based vendor development program to encourage technology diffusion and subcontracting linkages between MNCs and local suppliers. The BoI Unit for Industrial Linkage Development program (BUILD) first developed a database of local firms in areas such as metal working, tool and die making, PCBA and plastic injection molding, and then attempted to match these firms up with the specific needs of MNC assemblers. Despite positive responses from long-established firms such as Minibea, the program basically faded, even after it was upgraded to a "National Supplier Development Program" involving all major agencies responsible for small and medium-enterprise development. Unlike Malaysia's Vendor Development Program, the Thai effort lacked solid political and administrative support. Indeed, most if the initiative and resources for the program came not from the BoI but from a consultant under contract to the Board. Indeed, when IBM asked the BoI to supply names of indigenous companies that might be potential suppliers the BoI was unable to produce a single name.⁷⁹ By 1997 the program was basically dormant, although the recent crisis has stimulated new efforts to promote local suppliers

     b) Training and Skill Development: The government's approach to training and skill development has been equally weak, especially in the 1990s. Earlier policies were actually quite effective. The government kept the minimum wage below the inflation rate during the 1980s. Equally important, the democratic movement of the mid-1970s led to educational reforms resulting in near universal primary education and a significant expansion of vocational training. By the time Seagate entered the country in the early 1980s, these reforms had resulted in serious unemployment among vocational and university graduates and, as noted earlier, facilitated the firm's capacity for in-house training.⁸⁰ But by the mid-1990s, the labor market had constricted significantly. Prior to the country's financial crisis (summer 1997), most HDD firms had difficulty recruiting and/or keeping engineers and skilled workers. And despite years of warnings, the government had only begun efforts to bolster the general education system and expand training for engineers. According to one study, Thailand's rapid growth since the mid-late 1980s had resulted in "tremendous quantitative shortages of technical human resources in the productive sector."⁸¹ And government efforts were undermined, at least until the 1997 economic crisis, by a de facto policy of support for the hiring of low-wage, illegal workers.⁸²

The private sector responded to the lack of publicly-supported training efforts with several collective programs, none of which have had major impacts.⁸³ HDD firms have responded to the shortage of skilled personnel with firm-specific, in-house initiatives, several of which have already been noted (increased salaries and benefits for Thai engineers by Micropolis, Minibea, Seagate; in-house training programs by Seagate, KRP; recruitment of skilled personnel from China, the Philippines and South Asia by KRP; and sending engineers back to Japan for in-depth training by IBM, Nidec, Fujitsu).< COLOR="#ff0000"> The 1997 crisis may also be helpful. With large numbers of business failures, firms are finding it much easier to recruit good staff at all levels, but especially in the engineering field. But virtually all of these human resource initiatives have come from the private sector. Interviews with HDD firms yielded no cases of firms benefiting from public sector or collective training facilities. The only partial exception, noted earlier, were industry associations focused on compensation issues.

      c) Technology-related Institutions: The problems of the vendor and skill development reflect broader institutional weaknesses in the areas of technology development and diffusion. A 1996 report concluded that Thailand's science and technology environment "was characterized by mismatches in S&T human resource supply, very low levels of R&D activity, a private sector with strong manufacturing capability but weak research, development and engineering capability, and an information technology that is growing but remains insufficient."⁸⁴ The government is weak both directly - in terms of government-provided services, and indirectly — in terms of government incentives for collective efforts within the private sector. Indeed, the most disturbing aspect of this situation is the government's failure to respond to private sector, market-driven initiatives. Two specific areas of concern are university-industry linkages and technology-related diffusion.

The lack of linkages between universities and industry has been a concern for several years.⁸⁵ In response, several firms, most notably Fujitsu, have begun to establish links to local universities. In the case of Fujitsu, the company donates old equipment to the school when it comes time to upgrade to newer equipment. In most cases the equipment is still useable for most types of training as well as research and development. More importantly, engineers graduate with a higher level of familiarity with Fujitsu equipment and processes. These linkages have been formed primarily with those universities, like KMITT, that have received autonomous status and function with block grants outside the civil service bureaucracy. However, for the most part the universities are resisting the change to autonomy because of the inherent risk to the institution.⁸⁶ A Read-Rite engineer in Bangkok contrasted the Thai situation with a case in Singapore involving the development of head cleaning technology. Read-Rite uses a water-based process which obviously can introduce impurities. While in Singapore, this engineer learned of laser-based cleaning equipment developed by NUS researchers. But the researchers were not able to make the shift from equipment that would clean a few heads to those cleaning millions. With support from Japanese investors, a spin-off resulted that did develop higher volume equipment. Fujitsu is reportedly using the equipment and RR is considering it.⁸⁷

Thailand has also been weak in both direct and indirect provision of diffusion-oriented technical services. The country's metrological and industrial standards capacity remained fragmented until and weak, despite efforts at consolidation since the mid-1980s.⁸⁸ Nor did this agency move quickly to establish mutual recognition agreements with standards authorities in Thailand's major export markets. There have also been problems with regard to testing laboratories and services, despite requests for an expansion of such services from both local and foreign firms. In 1987, Thailand's peak business association urged the government's major lab to accredit private laboratories to perform testing services, but only 25 firms were accredited by 1994. Based on a 1987 study in response to the influx of Japanese firms, the Japanese aid agency JICA proposed the strengthening of a single organization for testing and standards, the Thailand Industrial Standards Institute (TISI) under the Ministry of Industry. Despite $30 million of Japanese loans and technical assistance, TISI has been undermined by government budget limitations and bureaucratic rivalries. The government's failure to staff TISI at agreed-upon levels, at least up to 1995, resulted in Japanese protests. Unable to make use of all the equipment donated from Japan, TISI had only limited testing capacities. According to Felker, these capacities were "were virtually monopolized by a handful of multi-national companies able to pay higher fees, what amounted to a quasi-legal retainer, forcing other firms to endure long service delays."⁸⁹

Interviews yielded no examples of HDD firms making use of TISI or any other government or collective testing services. An engineer at Read-Rite did, however, identify at least two industry initiatives to provide collective services that failed due at least in part to a lack of public sector response. In the late 1980s, a Seagate official attempted to establish a center for testing and failure analysis. Seagate actually donated equipment to the Asian Institute of Technology, but the equipment was never unpacked, much less used. Another case involved an unsuccessful Micropolis initiative to set up something like a Center for Magnetic Testing at the King Mongkut's Institute of Technology.

These weaknesses are not for lack of interest. Thailand's Five-Year Plans and policy statements are full of well-intentioned commitments to efforts such as such as national surveys of technology needs, subsidies for private-sector technology acquisition, a public Technology Transfer Corporation, university-industry linkages, and industry-wide technical services. The lack of achievement is due largely to factors noted in Chapter 6: a general predisposition against industry-specific or even cluster-based development policies on the part of the country's central bank and Finance Ministry; extensive bureaucratic turf battles; opposition to government intervention by local firms enjoying a protected domestic market; and a lack of support for technology development on the part of competitive and rural-based political parties. This effort was blocked by inter-ministerial rivalry and by the Thai private sector's opposition to any government interference in its relations with foreign technology suppliers.

We conclude with one recent illustration.⁹⁰ In August 1997 one of the local technology universities proposed the creation of a new national institute to promote electrical, electronics and communications industries, with a particular focus on hard drives as well as several other products such as IC packaging, display monitors, home appliances and solar cell panels.⁹¹ The fate of this effort is doubtful due to continued bureaucratic fragmentation and the skepticism of the foreign private sector with regard to government capacity. One Read-Rite executive even argued that, given the government's lack of expertise and cohesion and the risk of politicians taking advantage of large firms, the best thing to do was for companies like Read-Rite to keep a low profile.

Sources: BoI; firm interviews; corporate information.
Years in parentheses are year the operation began.

Sources: UNIDO, Industrial Development Global Report 1996. Compiled by Brooker Group.

Note: Wages include supplements

Source: Thailand Development Research Institute, cited in Warr (1997: 329). Data are based on an index of nominal wages for the manufacturing sector deflated by the consumer price index.