Information Storage Industry Center

Stephan Haggard, University of California, San Diego
Lim Pao Li, DCT Consultancy Services Sdn. Bhd.
Anna Ong, DCT Consultancy Services Sdn. Bhd.

Report 98-04
December, 1998

Stephan Haggard, Lim Pao Li and Anna Ong

Abstract
          Despite a relatively long history as a site for foreign direct investment in electronics, the first hard disk drive (HDD) investment came to the Northern Region of Malaysia¹ only in 1988, seven years after the industry first came to Singapore and five years after Seagate first invested in Thailand. The pioneers included both drive assemblers and component suppliers, but for a number of fortuitous reasons, the region has never became a major center for hard drive assembly (HDA). Seagate maintains one of its four major overseas assembly facilities in Perai², but except for production of cartridge drives by Syquest, Iomega and the newly launched Castlewood, no other drive assembly takes place in the region.

     Nonetheless, Northern Malaysia has come to occupy an important niche in the regional and global division of labor as a producer of two key subassemblies: the "heads complex" (heads, head gimbal assembly [HGA] and head stack assembly [HSA]) and an emerging media cluster housed in part in the Kulim Hi-tech Park in the state of Kedah, less than an hour's drive from Penang.³

          This study provides an overview of the development of the industry and attempts to explain why the industry entered, how it evolved, and why Penang has maintained and deepened its position in the regional division of labor despite (through 1997) rapidly rising land and labor costs. We begin with an analysis of the entry of the assemblers and the "major suppliers": the American and Japanese multinationals in the heads and media business. The entry of Maxtor and Conner marked an expansion of what we call a "regional production system": HDD firms with an initial production base in Singapore experimented with a lengthening of their cross-border production networks to exploit cost and other advantages at a proximate site.

          In undertaking these moves, labor costs certainly played a role; Northern Malaysia had particular cost advantages compared to Singapore. But two important qualifications are in order. First, a number of policy factors also mattered, including free trade in the EPZ's, quality infrastructure, and government incentives that increasingly approached those offered by Singapore. Second, Malaysia's advantage with respect to labor markets rested not simply on the cost advantages of unskilled labor; indeed, that advantage eroded over time. Rather, the availability of an increasingly deep pool of managerial, engineering and technical talent, including production workers with experience in electronics assembly, played a role. These features of the labor force resulted in part from Penang's long history in the electronics industry, but also from concerted efforts at educational upgrading.

     Given that there is relatively little drive assembly in Northern Malaysia, and that firms housed there are integrated into global as well as regional networks, proximity to local suppliers and customers would not appear to be a major factor in the development of the industry in Penang. But this conclusion is too hasty. First, Penang's general reputation as a site for foreign investment in electronics had been established through the early entry and success of a number of major semiconductor firms. The success in attracting earlier entrants had a spillover effect in attracting subsequent investment in the hard drive industry, just as early entrants in the hard drive industry attracted others. Several major heads firms located to supply Maxtor, Conner, and then Seagate's, assembly operations. Komag's major investment in media was not motivated by local proximity, but it induced other investments in aluminum substrates at Kulim.

     Second, we find that proximity to local suppliers and the development of an integrated industry cluster does play an increasingly important role in the industry over time. We call this the "second tier" of the supply chain: the firms supplying other components to assemblers (such as printed circuit boards [PCB], flexcircuits [FCB] and actuators [sometimes referred to as e-block]) and firms supplying and servicing the heads and media business (such as actuators and process engineering services). American, Japanese and Singaporean firms in this second-tier of the supply chain entered to supply local customers, demonstrating that the regional production system is not simply a creation of the major firms but a dense network of firms of many different sizes. A number of Malaysian firms also grew by supplying the industry, including with engineering services and specialized machining capabilities (such as tooling, jigs, and dies). This "local" supply base--consisting of both foreign and Malaysian firms--has become more important over time both in attracting new entrants and in rooting the HDD industry cluster in Penang.

     To explore in more detail the nature of relations between the assemblers and major suppliers on the one hand and their suppliers on the other, we conducted a survey of 16 suppliers in Penang. The survey shows a level of interaction between customers and suppliers which encompasses a variety of functions, a long-term view of relationships, and extensive technology transfer to both firms and workers.

     With respect to government policy and the lessons to be drawn, Penang has been an innovator within Malaysia in facilitating entry for foreign investors and developing and supporting export processing zones. Financial and other policy incentives have also undoubtedly played a role, but these did not substantially distinguish Malaysia from its neighbors; if anything, Malaysia played a game of catch up in this regard. The HDD industry began to internationalize in the early 1980s, but the entry of the industry into Penang came only after important policy reforms beginning in the mid-1980s that gradually brought Malaysia into closer alignment with incentives offered in Singapore.

     To what extent did more industry-specific policy and institutions serve to attract and maintain the drive industry in Penang; did industrial policy and institutions matter? It is important to underscore that Penang lacks policy autonomy to a substantial degree; major development initiatives are the responsibility of Kuala Lumpur. Moreover, we do not find the level of sophistication within the industrial policy agencies that is visible in Singapore nor evidence of the innovative vendor development schemes or university-industry linkages visible there.

     However, we find that officials in Penang exploited a certain degree of freedom that they were granted by unusual political circumstances to develop a more sophisticated industry promotion structure than in other Malaysian states or in Thailand. Over time, the state government played a more active role in discouraging labor-intensive investment, matching foreign firms with local suppliers and developing innovative public-private partnerships with respect to labor market needs. We conclude by suggesting ways that these efforts can be continued and deepened in the wake of the region's economic crisis.

1. Majors and Major Suppliers: Regional and Local Proximity

     Neither the government of Malaysia nor the government of Penang maintain consistent data on output and performance of the disk drive industry. However, we do have the benefit of a survey conducted by the Penang Development Corporation that provides information on employment, sales, investment, and fixed assets of the major foreign invested firms in the industry (Table 1), including assemblers and major suppliers. In 1990, the survey picked up the activities of three drive-related firms. For 1996, the survey covered eight firms, including the major investments in HDA, heads and media. The survey does not cover either Seagate's investment in Ipoh nor the new entrants to the Kulim Hi-tech Park in Kedah; nor does the survey capture the activities of local suppliers, which we take up in the next section. Even with these downward biases, it is evident from Table 1 that the drive segment has grown extremely rapidly and has come to constitute an important industry cluster in Penang, accounting for over 25 percent of total employment in the electronics sector as a whole.

Sources: Annual Survey of Manufacturing Industries in PDC Industrial Areas, Corporate and Investment Division, Penang Development Corporation, various issues; Economic Report 1997/98, Ministry of Finance, Malaysia.      Appendix 1 outlines the history of investment in the industry through mid-1997, focusing on hard drive assemblers and what we will call "major suppliers," defined as firms making heads and media. Although investment has been more or less continuous, there do appear to be two somewhat distinct phases: a period of early entry from 1988 through the early 1990s, and a new wave of investment beginning in the mid-1990s. This continued investment in the industry raises some questions for an explanation based on labor costs alone, since the early 1990s was a period of extremely rapid employment and wage growth and tightening labor markets. We return to these labor market issues below, but begin by outlining the pattern of investment in the industry, dividing the discussion into three segments: HDA, heads and media.

     Hard Drive Assembly (HDA)

     With respect to hard-drive assembly (HDA), fortuitous events play an important role. DEC, Hewlett-Packard, and Quantum initiated drive manufacturing in Penang, but failed to sustain it. Maxtor considered relocating assembly lines from Singapore to Penang, but never did so and ultimately sold its HSA operations there to Read-Rite. Conner did prove successful in assembling drives in Penang, but was subsequently acquired by Seagate. As of early 1998, Seagate's assembly operation at the Conner facility in Perai (mainland Penang) is the only HDA operation in the Northern Region.⁴

     The important role played by acquisitions makes it somewhat difficult to make firm judgments about why the majors entered. Takeovers had wider strategic motivations, and so entry into Malaysia arose as a by-product. Several of these acquisitions succeeded (Seagate's absorption of Imprimis and Conner) while others didn't (Quantum's purchase of DEC's storage division). However, none of the failure cases had anything to do with characteristics of Penang as an investment location; to the contrary, the H-P investment in particular showed confidence in Penang as a location for the production of the most advanced, high-performance, high-capacity drives. Moreover, both Conner and Seagate continuously expanded their assembly operations even in the face of rapidly rising costs and labor shortages. By the mid-1990s, Seagate was ramping new product directly into Penang from the U.S. design center, and Penang even served as a "transfer station" for the extension of Seagate's production networks into China. In the mid-1990s the two major manufacturers of cartridge drives--Iomega and Syquest, established major assembly operations as well. In the late 1990s, Castlewood also entered the cartridge drive market, using a 100 percent locally-owned firm with a 5 percent stake in Castlewood, Trans Capital, to manufacture its first product.

     As in Singapore, if cost advantages played a role for the initial entrants in the late 1980s, they can hardly be cited as a motivation either for subsequent entrants or for why Seagate continued to expand its assembly operations until early 1998. In outlining the history of the majors' investments in Penang, we see two additional factors at work: the perceived advantages of regional proximity to suppliers, customers and other operations in Southeast Asia, in sum, the advantages of a regional production network; and for a number of major suppliers, the willingness to follow major customers into Penang, and thus the advantages of local proximity, agglomeration economies and clustering.

     The early entrants into Penang in the disk drive industry came in 1988-89, and included Maxtor, Conner, Hitachi Metals, Control Data and Applied Magnetics; of these, Conner and Maxtor were HDD assemblers. Although Conner was the larger and more successful operation, Maxtor was the pioneer investment in 1988 that generated a demonstration effect within the industry as a whole, spurring the entry of a number of suppliers in the head segment.

     At the time of Maxtor's entry into Penang, Singapore had already become the company's major offshore production site, but rising costs were deemed a problem for further expansion. Maxtor Singapore took the decision to expand into Penang. In considering moving offshore from Singapore, they looked at Penang and Thailand. Several firms in Penang, including particularly Eng Hardware, were already supplying Maxtor's Singapore operations. Joe diBene, a US industrial consultant had been hired to do the work for Maxtor and he preferred Penang. Personal connections between a Maxtor executive and the Penang Development Corporation (PDC) through previous employment in Penang paved the way for a meeting between Maxtor's board and both Penang officials and, at PDC's urging, Prime Minister Mahathir himself.

     Both the island and the mainland were considered, and issues of land drove the decision to locate on the mainland at Perai⁵. The plant started doing HGA, while HSA remained in Singapore, but high-end HSA, FCBA and spindle motor production were quickly moved to Penang as well. Within six months of the plant coming on line, a glut in heads and falling prices lead to a decision get out of HGA altogether and subcontract it to Sunward (Philippines) and Applied Magnetics and Dastek, both of which had entered Penang by this point. A debate ensued within Maxtor over whether they should also abandon HSA and move drive production to Penang. Penang's management argued that they were capable of doing it--Conner had entered by this point and was proving it--while Singapore expressed reservations, including concern over the costs of recertifying the entire Penang facility for its major customers. Maxtor began to experience financial difficulties just as these discussions were taking place, and they had also just acquired Miniscribe, which gave them substantial capacity in Singapore; the decision was therefore taken to sell off the entire HSA business. Bids were made by several firms, including Applied Magnetics, which had entered Penang by this time, but the sale went to Read-Rite in December 1991. The smaller spindle motor operation was sold to CAM Technologies, a Singapore-based firm acquired in 1998 by Tektronix.

     As a result, Maxtor never produced drives in Penang. However, as this history demonstrates, the location of a major did contribute substantially to the development of a local cluster of heads/HGA/HSA operations, which in turn influenced subsequent investment decisions.

     The story of Conner's entry in 1989 bears important similarities to Maxtor's. The industry as a whole was in an expansion phase, and Conner's sales were particularly strong. Singapore had become Conner's main production base, and the decision to go into Malaysia was taken by the managing director of Conner Singapore; no other sites were considered. Conner Singapore began the experiment in the new location with HSA under a Singaporean managing director. The initial headcount was 300 workers who had "trained" by working for 2-3 months on the lines in the Singapore facility; regional proximity thus contributed to lowering training costs. But the managing director believed that drives could ultimately be made there as well, and within a year, they started to transfer lines from Singapore into the Perai facility, beginning with the Stubby, a low-end 20 Mb drive. By 1994, however, the strategy of using Singapore as an intermediary had been rejected in favor of direct ramp-up from the Colorado design center to Perai for the Hopi (3.5 inch, 80 and 120 Mb), Rambo (3.5 inch, 212 Mb) and Kato (2.5 inch) drives, a strategy also pursued by Seagate after its acquisition of the facility.⁶

     Until it was acquired by Seagate, Conner retained a regional division of labor in which Singapore got the most technically demanding products (capacity over 1 gigabyte in late 1996), while Penang produced drives with lower capacities. By the mid-1990s, however, Penang was playing the role that Singapore once had as a transfer point for further offshore operations. In 1992, Conner opened a facility in Shenzhen; rising labor costs and high turnover in Penang and the inability to secure technical people were cited as reasons. By 1994-95, Conner Penang was itself serving as a "staging point" for further regionalization: new product was ramped up in Penang, where production was stabilized before organizing the transfer of lines to China.

     Because Conner was engaged in HDA, and because its corporate strategy emphasized the benefits of extensive outsourcing, its entry had an even wider spillover effect on the development of the local supply base than Maxtor's. Naturally, key components were sourced from Japanese and American firms within the region. However, Conner's entry also resulted in a parallel regionalization of existing suppliers based in Singapore and Thailand as well as the development of several local firms. A reconstruction of Conner's early supply chains from Conner managers shows the following sourcing patterns:

• HSA (MIG) from Kaifa (China & Hong Kong), SAE (China & Hong Kong), Sunward (Philippines) and Applied Magnetics;
• HGA from Read Rite (Thailand);
• Motors from NMB (Thailand) and Nidec (Japan);
• Media from Conner's media division in San Jose;
• PCBA from Trans Capital (Malaysian-owned, Penang), Tongkah (Malaysia-Singapore joint venture, Penang), Natsteel (Singapore-owned, Perai), TriM (Singapore-owned, Perai)
• Baseplates from MMI (Singapore-owned, Perai, also supplying VCM) and CAM (Singapore-owned at Perai);
• Other metal parts from Leksun (at Sungai Petani, Kedah, Malaysia)

     In September 1995, Seagate acquired Conner. The motives for the merger included the acquisition of media technology, but it also gave Seagate production capability in mid-range drives in Malaysia which it decided to keep. Of the 11,000 people working for Conner in Asia only one--the managing director of the Perai facility--was let go, replaced by the manager of Seagate's Wuxi facility. Prior to the acquisition, Seagate's Asian HDA was concentrated in Singapore (high-end) and Thailand (desktop and mobile; subsequently consolidated to China), but the acquisition corresponded with a rapid growth in production capabilities beyond these two traditional poles in the form of new facilities in China at Wuxi and Shenzhen. Penang's niche in this highly differentiated production strategy was an intermediate one. For example, Singapore and Perai served both as stand-alone production facilities for certain product and as "transfer points" for China; drives were initially ramped from the United States to these two facilities, and then passed on to lower-cost Chinese operations. By 1997, however, Asian management in Singapore was working toward a "focused factory" model in which drives were ramped directly from the US into each facility, with a rough division of labor continuing in which high end-drives and arrays were assembled in Singapore and desktop drives in Penang and China.

     Seagate inherited high-quality test and assembly equipment layout from Conner in Penang, and focused on improved processes, cycle times and material flow, resulting in substantial improvements in productivity.⁷ The merger also involved substantial adjustments for suppliers, because of Seagate's higher level of vertical integration and more demanding materials management system. For example, Conner had outsourced PCBs, while Seagate met its needs through major new investments in Senai in Johore; we return to those adjustments below in discussing suppliers. 100 percent of Perai's heads (Thailand) and media (California) came from Seagate.

     Mention should also be made of three other HDA efforts which did not succeed, since they shed light on firms' perceptions of Penang's capabilities and their entry strategies. Hewlett-Packard's components group had been one of the first foreign electronics firms to locate in Penang in the early 1970s. In 1993, the Disk Memory division was making high-end SCSI drives for workstations and servers, with all production located in Boise, Idaho. Boise had begun as a vertically-integrated operation, leveraging off of H-P's capabilities in boards, chips and heads, but they had gradually moved toward outsourcing in the Asia-Pacific region, including HSA's from Seagate in Penang. When considering offshore production, they assessed over ten sites, ranking each on ten characteristics ranging from costs and "quality focus" to infrastructure, taxes and supporting industries. Each site was compared to Boise on each parameter and Penang and Singapore were ranked at the top; Penang was chosen both because of costs and because of the ability to leverage technological capabilities at H-P's existing Penang facility.

     H-P's plan was to begin with manufacture of a one gigabyte, and then two gigabyte drive, but to leapfrog the competition within three years by producing a 10 platter, MR-head 9 gigabyte drive. The first two products were successfully ramped up to full production, with both local and regional sourcing (media from Komag, baseplates from Singapore, motors from NIDEC Thailand, PCBA from SCI); only heads were coming from the United States, through a joint venture with Headway in Newark. However, the one and two gigabyte products faced competition from thinner drives (H-P's design was five-platter; others were achieving similar capacities with three) and technical problems remained with the 9 gigabyte model. In April 1996, H-P management made the decision to end the project and to concentrate efforts in tape and optical storage and sold production equipment to SCI.⁸ Nonetheless, the effort shows that by the early-1990s, a major firm believed that Penang was capable of ramping up production of a drive which, had it succeeded, would have been at or near the technological frontier at the time.

     The last firm to enter Penang in HDA was DEC, which approached Penang Development Corporation concerning two investments in 1994, one in heads and the other in high-end drive assembly. By this time, PDC was concerned about the effects of labor-intensive investments on the labor market, and discouraged the heads investment, which went to Batam, Indonesia. PDC was interested in the HDA investment, however, and brokered the application process, but DEC headquarters made a strategic decision to get out of storage altogether, selling the division to Quantum.

     At this point, Quantum had established its relationship with MKE in the production of desktop product, but was still producing high-end drives in the US using a highly automated process. Some effort was made to integrate DEC's Penang operations into a single product trajectory, but they were unsuccessful in part for technological reasons; high-end production was ultimately turned over to MKE as well, which had no use for DEC's Penang facility. Again, however, the case shows that a major firm believed Penang capable of ramping up production in a high-end product.

     To summarize a complex picture, we can see three modes of entry into HDA in Penang. The two pioneering firms--Maxtor and Conner--entered through an extension of Singaporean operations, beginning initially with component production. The storage divisions of DEC and H-P did not have Singapore operations, but H-P's decision was influenced by long experience in Penang itself. Seagate, finally, entered through the acquisition of Conner, although the HDA operations in Perai reported to Singapore as well. Thus in four of the five cases, entry into Penang must be seen as an extension of existing production networks in the region.

     The failure of Penang to host a larger concentration of HDD assemblers seems largely fortuitous; there are no good reasons why Penang couldn't host more drive assembly in the future. As the cases of DEC, H-P and Quantum showed, firms were willing to locate drive assembly in Penang, in H-P's case, high-end drives. The outcome seems primarily a function of timing and resultant first-mover advantages. Drive investment in Asia in the early 1980s was concentrated overwhelmingly in Singapore, with some mobile and lower-end drives assembled in Thailand. Singapore's position as a "first mover" allowed it to hold on to low-end drive assembly for some time by shifting subcomponent assembly offshore, including to Malaysia. By the time this became not feasible as a strategy, the labor-intensive nature of low-end assembly appeared to make Penang too costly as a site and it was passed over in favor of Indonesia, the Philippines and China. Singapore could hold on to its position in the industry as a whole by graduating to the production of higher-end drives and serving as a regional HQ, again with Malaysia playing a supplier role with respect to heads, media and electronics.

     However, Conner and Seagate Perai demonstrate that Penang is capable of retaining HDA despite rising costs. Seagate Perai continues to be an important production location for the company; until the crisis of 1997-8, it has continued to grow by moving into the production of increasingly sophisticated product. Moreover, there are clearly benefits from having an increasingly well-developed local supply base consisting of both Malaysian and foreign companies.

     It is thus quite possible that Penang could become a host for high-end drive assembly in the future as costs in Singapore continue to rise, particularly when compared with the rest of the region. The crisis of 1997-98 has substantially lowered unit labor costs in Malaysia while not substantially affecting the agglomeration economies that have been built up over nearly three decades of extensive electronics investment. It is to a more detailed examination of the suppliers that we now turn.

Heads, HGA and HSA

     The "heads complex" encompasses four steps. Wafer fabrication is a highly capital- and technology intensive activity, and remains concentrated overwhelmingly in the U.S. and Japan. Slider (head) fabrication--whether older ferrite metal-in-gap (MIG) technology, thin-film inductive heads, or the now-dominant thin-film MR heads--has both capital- (ion milling, photolithography and vacuum technology) and labor-intensive steps, but the two typically are located together. Head-gimbal assembly (HGA; attaching a single head to a flexure or suspension arm and a wire/tubing assembly) and head stack assembly (HSA; attaching the HGA to an actuator arm and increasingly to flexcircuit subassemblies) and testing, by contrast, have traditionally been labor intensive activities. However, HSA involves componentry, including the actuator that holds the heads and flexcircuit assembly, which are increasingly demanding in terms of both basic technology and production capabilities. There is a growing tendency for HSA to be a stand-alone subassembly, and headstack assemblers are increasingly working with suppliers to add value, for example, by integrating flexcircuit and actuator design. At present, only Quantum and Seagate are fully vertically integrated in HSA.

     The process of regionalization that we have traced for HDA was also replicated in important measure in the heads complex; heads makers such as Applied Magnetics invested into Penang out of their Singapore operations largely for reasons of cost and regional proximity. However, there is also evidence that both forward linkages to customers and backward linkages to suppliers--i.e., "local proximity" and agglomeration economies--were significant. Entry into Malaysia was influenced by the presence of major customers including not only Maxtor and Conner but also the larger integrated heads manufacturers like Applied Magnetics, who themselves sourced some heads externally. Initially, the local supply base did not appear to matter substantially, and most components and subassemblies--even the most simple--were sourced elsewhere. Over time, however, the heads cluster spawned important subassembly operations in actuators and flexcircuits, as well as machine tooling capabilities. This supply base, in turn, has allowed Penang to remain competitive in the production of heads and the increasingly complex, stand-alone HSA business, even as Penang operations assisted firms in the heads complex in moving further offshore.

     The investment in the heads complex shows a marked clustering in time, and there is evidence that proximity to customers mattered. The Applied Magnetics investment in Penang was roughly contemporaneous with Maxtor's entry, but was not associated with it. Although Applied Magnetics did end up supplying Maxtor's HGA and HSA operations in Penang, its initial role was supplying Maxtor's assembly operation in Singapore for their 200 megabyte LXT series drive (single MIG ferrite head). As with the entry of Maxtor and Conner, the decision concerning Applied Magnetic's entry was made by the Singapore office and the main grounds included labor costs, proximity to Singapore, but also technological expertise from the electronics sector, including in tooling. Applied Magnetics was quickly followed by both independent and captive heads suppliers for whom "local proximity" appeared to matter more. Control Data entered Penang to make HGA for Maxtor, and was supposed to make HGA for Applied Magnetics as well before a contract dispute led to their disqualification. In 1988, Control Data established its data storage operations as a subsidiary, Imprimis, with a presence in higher-end drives and strong integrated circuit and head technology. Imprimis was acquired by Seagate in 1989, giving the company Imprimis' HGA operation in Penang. Shortly after Maxtor's entry, Dastek and Hitachi Metals also came to Penang. Dastek's main customer was Maxtor. They began with HGA and then went into HSA, trying to capture some of Conner's business. They failed in this effort, and built HGA and HSA for Micropolis (Singapore) before going out of business. Hitachi Metals entered in part to supply Applied Magnetics, which in turn was supplying Maxtor.

     The evolution of Applied Magnetics operation demonstrates the complexity of the heads portion of the value chain and the gradual upgrading of capabilities in Penang. AM's original operation focused only on HGA, with virtually nothing sourced locally; sliders were manufactured in Goleta, suspensions were sourced both in-house and from Hutchison and NHK, and even wire and wire-tubing subassemblies came from the U.S. and elsewhere in the region. Full slider fabrication did not begin until 1990, still based on wafers from Goleta but with ion milling and photolithography done in Penang. Applied Magnetics designs the lapping machines and photolithographic equipment itself, but subcontracts ion milling equipment or buys it off the shelf. However, substantial process engineering was increasingly done in Penang. For example, equipment for the slider fab and automated lines was produced there, and while "high end failure analysis" goes back to the United States Penang operations performs failure analysis with locally-made testing equipment.

     HSA was added in 1991 as a service to customers, although AM continued to see itself self primarily in the head business. Customer specifications drove the sourcing of many HSA components, with much still sourced from the US and Japan, but also from elsewhere in the region (bearings, crash top, voice coil). But Applied Magnetics had a vendor development program and started to source some components locally, including actuators (Xolox) and crash tops (Eng). As the head business evolved into a headstack business, proximity to two key subassembly operations--actuators and flexcircuit assembly--became more important. The development of the vendor base in HSA still involved the drive manufacturer, which would bring in and qualify vendors when they started new programs. However, manufacturing benefited from "local" as well as "regional" proximity. For actuators, Applied Magnetics purchased from Belton (HK/China) but primarily from firms in Singapore (BJ and Polimicro) or Penang (Xolox and Eng). A similar pattern is visible in flexcircuits: Asian Micro (Singapore), CAM (Singapore with Penang operations), Solectron (US with Penang operations) and Leksun and Trans Capital (Malaysian).

     Applied Magnetics and Hitachi entered via greenfield investments, but the other two major operations in the heads complex--Seagate and Read-Rite--entered through acquisitions. Both are closely integrated into regional production networks, although they differ because Read-Rite is specialized in heads, HGA and HSA and Seagate is vertically-integrated assembler.

     In 1991, Conner decided to spin off its HSA business; the company's general strategic approach was exactly the opposite of Seagate's: to outsource extensively, usually from multiple and competing suppliers. Conner had only developed internal HSA capabilities in response to problems with suppliers, including Read-Rite. At the time, Read-Rite was moving in the opposite direction, integrating forward into HSA. Under the conditions of the deal, Read-Rite took over the Conner HSA facility, all equipment and 1100 employees and Conner committed to absorb all of Read-Rite's Penang output. Proximity to a major customer was thus clearly a motive in Read-Rite's investment. However, Read-Rite couldn't keep up with cost pressures from Conner, which was continuing to squeeze life out of metal heads by drawing on low-cost supply from SAE and Kaifa (in China) and AMC. Read-Rite's Penang operation gradually diversified to serve Western Digital (WD) (since 1993, in both Kuala Lumpur and two Singapore facilities across the whole spectrum of drives), Iomega (since early1996) and Hyundai-Maxtor (since 1996), before finally closing in 1998.

     Partly because it is specialized in the head business, partly because of business strategy, the evolution of Read-Rite's production networks in Southeast Asia look somewhat different from those of the principle drive assemblers; nonetheless, they show the same pattern of regionalization. Rather than moving first into Singapore, Read-Rite's first Southeast Asian operation was in Thailand, where both Read-Rite and Read-Rite SMI, a joint-venture with Sumitomo Metal Industries, operate subsidiaries doing slider fabrication and labor-intensive HGA assembly. As with the extension of production networks in drive assembly, however, Thailand oversaw the development of HSA assembly in Penang, and later in Manila as well. ⁹ Entry into Malaysia was an extension of an existing production network, but rather than subcontracting "down" the value-chain, Read-Rite Thailand in effect subcontracted "up," moving existing HSA out of Thailand and integrating forward to a greater extent. In doing so, it was assisted by the existence of local technical capabilities, both through suppliers and through the workforce. In a pattern we have already noted in other segments, suppliers were located throughout the region (HGA from Read-Rite Thailand, shipped daily, actuators from Min Aik in Taiwan), but local Malaysian and foreign companies are strongly represented, including FCBA from Trans Capital and Asian Micro and actuators from Xolox (for WD) and Eng (for Iomega). The workforce was also more specialized than in Thailand. The company employed 4000 direct labor and 800 indirect labor; of the 800 indirect labor about 150 are engineers and assistant engineers and more than 200 are technicians. The company also maintained an Advanced Manufacturing Technology Group in Penang which carried out process development, although less than one percent of total revenue is spent on R and D. However, the labor-intensive nature of its operations ultimately ran up against labor market conditios; just as the crisis was breaking in Malaysia and labor market conditions were set to change dramatically, Read-Rite took a decision to consolidate its operations and close the Penang facility because of labor shortages!

     Seagate is the behemoth of the drive industry in Northern Malaysia, and its slider-HGA-HSA operations are substantially larger than its assembly ones in terms of employment although not sales; at the end of 1996, Seagate had 10,000 workers in slider fabrication and HGA-HSA in Penang and another 3,500 at new facilities in Ipoh in the neighboring state of Perak which are managed by the Penang facility; these numbers compare to 4,500 workers at its assembly operation in Perai on mainland Penang. The history of Seagate's entry into Penang begins with the acquisition of Imprimis in 1989, and with it the Penang HGA facility (known as STPG-A) that was supplying assemblies for virtually the full range of Imprimis drives. The next stage in Seagate's expansion came with a $7.5 million investment in a slider fab in 1992 (STPG-S) to perform machining, grinding and finishing of heads.

     However, less than six months after the first facility was operational in June 1992, Seagate followed in with the announcement of a third facility, a $50 million "expansion" of the slider fab that in fact constituted a wholly new investment. The facility housed the company's most advanced slider fabrication processes, including high vacuum technology, photo lithography and ion milling, processes explicitly deemed too technically demanding for location in Thailand. ¹⁰ Out of space in Penang, and with labor markets tightening dramatically in the early 1990s, Seagate Penang oversaw a greenfield expansion into the neighboring state of Perak at Ipoh that followed the same pattern of Seagate's entry into Penang: the opening of HGA operations in 1995 followed by a second proximate slider fab in 1996.

     In the case of both expansions, Seagate was responding to periods of increased demand for drives, but also a decision to aggressively pursue original equipment manufacturers (OEMs) as customers for Seagate heads. ¹¹ With respect to its own networks, sliders from the two fabs were first shipped to the Teprok and Kurat facilities in Thailand (HGA) as well as to its own STPG-A HGA operation, which was subsequently integrated into HSA as well. HGA and HSA were then shipped to Singapore (Tuas and Kallang), which prior to the Conner acquisition were the company's only assembly facilities in Southeast Asia. Interestingly, Penang did not initially supply the Perai facility on the mainland, which secured its heads from Thailand. In the wake of the crisis, however, Seagate made some strategic changes in its slider operations in Asia. HGA-HSA operations were consolidated into Thailand, reflecting as with Read-Rite the more purely labor-intensive nature of this operation; this consolidation resulted in a retrenchment of approximately 1,000 workers. However, many of these were reabsorbed into the company's slider operations, which now constitutes the entirety of Seagate's mass production of sliders.

A New Media Cluster

     Like the production of heads-HGA-HSA, media manufacturing is a multi-staged production process that is increasingly spread across multiple locations, both globally and within Southeast Asia. ¹² As with heads, this is partly due to differing skill and natural resource requirements at different stages; for example, polishing, texturing and cleaning of substrates is relatively labor intensive, while other stages in the process are highly intensive in the use of water. As with heads, and with HDA itself, firms differ in their degree of vertical integration (see Table 2).

     In the 1990s, Northern Malaysia witnessed a strong clustering of media production, beginning with a crucial investment by Komag that became operational in 1992 and sparked the interest of other producers (a demonstration effect) and suppliers (linkage effects). The opening of a science-based industrial park at Kulim in the neighboring state of Kedah, roughly 40 minutes from Penang, attracted several key investments in aluminum substrates in the mid-1990s, although there is some question about whether any research and development is occurring in the park. Nonetheless, by the year 2000, the Northern Region of Malaysia (along with Singapore) will house one of the largest concentrations of media manufacturers in the world. ¹³

  * Sky Aluminium is a subsidiary of Showa
  + Akashic Kubota was acquired by Stormedia and its substrate section is renamed Strates
 

     Before choosing Malaysia, Komag did not have any operations outside of the US and Japan; in this, it was somewhat unusual. The company conducted an extensive search of possible sites before narrowing the decision to two: Penang and Singapore, both of which made strong bids to attract the company. With respect to proximity to customers, Singapore appeared to have the edge. But given that some shipments were also going to Thailand, this advantage was not decisive; products could be trucked easily in either direction, and flown in if necessary. Neither site had an advantage with respect to key inputs (primarily chemicals and substrates) or capital goods, which would be supplied from Japan and the United States. According to Komag's managing director, the main consideration was personnel. Costs were rising in Singapore and studies conducted by Komag indicated that there were adequate personnel with requisite technical skills in Penang.
 
 

     Komag's Malaysian facilities were expansions out of US operations, and have manufactured similar product; the Penang lines are fully integrated from plating forward. Initially, new product was ramped up in the US, then transferred to Penang. By 1997, however, new product was being ramped up in Penang, and although tooling and new lines are designed by Komag and imported from the United States, local firms (LKT, Eng, Mate and Excel) have played an increasing role over time in line modification, tooling and automation. When Komag launched a major restructuring of its manufacturing operations in 1997, Malaysia was a big gainer because " these facilities [were] closer to customers' disk drive assembly plants in Southeast Asia and enjoy[ed] certain cost and tax advantages over ...U.S. manufacturing plants." ¹⁴ As a result of the restructuring, the share of Komag's total output produced in Malaysia was planned to rise from 50 percent to over 75 percent.

     As in other cases we have already noted, the initial point of entry into Southeast Asia became a regional headquarters managing further expansions. Penang not only managed a substantial local expansion in Bayan Lepas ("Campus 3"), with twelve more back-end texturing and sputtering lines, but also a major investment in Sarawak in plating, polishing and grinding. Within five years, Komag had developed an integrated network within Malaysia that runs from finishing of substrates through production and final test of media.

     In addition to process related products, such as chemicals (plating solution, slurries) and texturing supplies (buffing tapes), the main input into media manufacture is blanks and/or substrates. When Komag came into Penang, Kobe Precision Technology, a wholly-owned subsidiary of Kobe Steel followed at Komag's request, commencing operations in April 1993. ¹⁵ Although Kobe supplies other customers from Penang, including in the U.S. (IBM) and Singapore (Stormedia), 70 percent of sales are to Komag Penang with another 5 percent going to AKT, a Komag subsidiary in Thailand. Kobe managers cited labor costs and proximity to HDD firms in Malaysia and Singapore as motives for investing. But given that their output is virtually dedicated to Komag Penang, local proximity was clearly an important factor, and in late 1997, the firm began consideration of an investment in Sarawak to support Komag there.

     The second major company to enter Malaysia in the media part of the value chain was Akashic Kubota. Akashic began operations in 1997 producing polished aluminium substrates, with plans to go into the manufacture of media, but was sold less than a year later to Stormedia when Kubota decided to exit media production altogether. Akashic considered six sites, including Bintan (Indonesia), Singapore, and four sites in Malaysia before choosing Kulim. Among the company's considerations in choosing Kedah both labor costs and availability were important; at the time the decision was taken, manufacturers in Penang were bussing in workers from the mainland. But tax benefits and proximity to customers also figured into the decision; Western Digital, with manufacturing operations in Kuala Lumpur and Singapore was a major customer of the new facility.

     1997 saw a rapid succession of investment announcements in media-related production in Kulim, announcements which appeared unaffected by the currency crisis which struck Malaysia by late summer. In early 1997, Fuji Electric Co. announced 77 billion yen global expansion plans aimed at doubling its capacity to 118 million 3.5-inch disks to supply hard disk drive manufacturers in Southeast Asia. ¹⁶ . This target was to be met in part by expansion in two Japanese facilities (68 million of target), but the remainder was to be met by a 60 billion yen greenfield investment in Kulim producing high-end disks, scheduled for shipments by April 1998. In July 1997, Hyundai Electronics America and Hyundai Electronics Industry Co. Ltd. announced their entry into Kulim as well through a joint-venture subsidiary (MMK Technology Sdn Bhd). The $200 million factory was designed to produce 22-28 million media, mostly for export to Maxtor operations in the US and Singapore, but construction of the facility was suspended in 1998. ¹⁷

     As with the entry of Komag, the increased density of media manufacturers created opportunities for suppliers. As a result, Malaysia's backward integration into both the basic materials and chemicals elements of the production chain increased substantially, with a marked increase in Japanese investments in particular. In 1997, Ohara, a specialized Japanese glass firm, announced a major investment to produce crystalline glass for substrates, a technology in which Fuji has been a leader, and Toyo Kohan (through its joint venture with Marubeni, Toyo Memory Technology) and Showa Aluminum announced even larger investment in substrates. ¹⁸ In 1997, Uemura & Co. also announced its intention to form a wholly-owned Malaysian subsidiary to produce non-electrolytic nickel plating chemicals for local area disc makers; a local Uemura subsidiary in Singapore will conduct marketing activities there.

Summary

     The foregoing summary shows that entry into Penang was part of a regional production strategy on the part of major assemblers and suppliers. This regional production strategy contributed to the development of linked suppliers of major components--heads and media--which in turn further reinforced the advantages of locating in Penang; in sum, both regional and local agglomeration effects were operating. Central to subsequent developments were the entry of two majors with operations in Singapore, Maxtor and Conner, beginning with the move of certain subassembly operations. The stated reason for this move was primarily to reduce costs, and familiarity on the part of Singapore managers' with Malaysia in general, and Penang in particular made it in easy choice. However, there were also other advantages of what we have called "regional proximity." First, proximity to Singapore allowed for close managerial oversight and training. We were surprised to find that training extended below the middle-management ranks to production workers, which in several instances were sent to Singapore for "training" by working on lines there. Second, there are advantages in transport costs and speed of delivery. Subassemblies--the initial activity for the HDD majors in Penang--could be shipped to final assembly operations in Thailand or Singapore in less than a day by ground transport, and even more rapidly by air. When full HDA was established, the ease of transferring lines and the regional proximity to other subassembly and component suppliers were pluses.

     The entry of the majors induced investments by Japanese and American firms supplying assembly operations with heads, HGA and HSA. As with the majors, these investments were often managed out of pre-existing Southeast Asian operations, but were motivated by advantages of being locally proximate. Komag's critical investment in media is more mixed in this regard; regional proximity between customers in Singapore and Thailand appeared as important local proximity. However, the entry of heads and media producers had a further ripple effect on what we call the "second tier" of the supply chain: suppliers of other subassemblies and components (baseplates, PCBA, turned parts); and suppliers to the major suppliers (actuators, flexcircuit). These firms included three distinct types: Japanese and American firms selling to the major suppliers in heads and media; Singapore-based suppliers who followed their customers to Penang; and local firms that had either been supplying Singapore and Thailand assembly operations or who entered the industry as it grew. It is to these investments that we now turn.

2. The Second Tier of the Supply Chain

     An examination of the second tier of the supply chain demonstrates both the increasing integration of the regional production system in Southeast Asia and the development of significant agglomeration economies and clustering (See Appendix 2). Conner was particularly intent in bringing a number of its Singapore-based suppliers to Penang when it relocated some production there in 1989. CAM Technologies invested in Penang to supply baseplates to Conner, and later acquired Maxtor's spindle motor facilities in its effort to integrate baseplates and motors into a single subassembly. MMI also entered to supply Conner, initially with VCM but subsequently with baseplates (50 percent of total output) as well. Several Singapore-based contract manufacturers also followed Conner to provide PCBA, first Natsteel Electronics and TriM, followed by Tongkah Electronics, a joint venture between Tongkah Holdings (a publicly-listed Malaysian group, 51%) and Goldtron Electronics (a Singaporean firm, 49%). These firms faced somewhat more difficulties following Conner's acquisition by Seagate because of the latter firms vertical integration in PCBA. However, all successfully diversified, Natsteel by finding other customers in the drive industry (in Singapore), TriM by servicing customers in the office automation business (copiers, printers, fax machines), and Tongkah both by developing new lines of business and expanding their HDD customer base to include JTS (India) and Micropolis (Singapore) as well as Syquest (Penang) and Seagate (both Penang and Singapore).

     A similar pattern is visible in the major American contract-manufacturing firms that entered Penang in the early 1990s. SCI Manufacturing, the largest contract-manufacturer in the world, was initially drawn to Penang to service HP, but decided to enter even after HP's HDA efforts faltered; they currently supply three customers, only one in the drive industry and that one in Singapore. Solectron was already supporting Conner from its Milpitas, California headquarters; when Conner came to Penang, Solectron followed to supply them with PCBA and also sold FCBA to Maxtor, and subsequently to Read-Rite. Despite the fact that they relocated to follow customers, all sales are currently for export. Finally Xolox Malaysia was incorporated in October 1994 and began operation in April 1995. Before investing in Penang, it already had customers in Penang, including Applied Magnetics and Read-Rite (80% of its business) as well as SAE (China) and Tandon (India). All of these customers in turn supplied to Western Digital in Kuala Lumpur and Singapore.

     Tracing the effect of the development of the industry on local suppliers is somewhat more complex, since most firms had a start in some other line of activity, often with other foreign electronics firms playing a role. ¹⁹ However, there are some firms, including particularly Trans Capital, whose growth is tied almost entirely to the HDD industry. Trans Capital began making racks for Conner, but made a leap into PCBA with assistance from Conner. As with other electronics firms, Trans Capital faced major adjustments following the Seagate acquisition, and had to both find new customers and diversify into new product lines. The main diversification efforts were into "complete build" subcontracting in areas like CD-ROM drives. However, a second major thrust was to leverage experience in the drive industry by diversifying into FCBA. FCBA now accounts for 50% of their business, and of that, 80% (or 30-40 percent of their total business) is with Read-Rite, primarily for Western Digital. Trans Capital's strategy is to go into complete HSA on a subcontract basis for firms such as Read-Rite; as with CAM's efforts in baseplates, the strategy is driven by an effort to vertically integrate in subcomponent manufacturing, in this case by automating HGA and combining FCBA and HSA into an integrated product. In 1997, Trans Capital acquired 10% of Xolox¹s share in order to acquire actuator capabilities, and in 1998, it entered into an even more ambitious venture as a manufacturer of cartridge drives for American start-up Castlewood Systems. ²⁰

     The more typical pattern, however, appears to be for local firms to begin with some engineering and machining capabilities, and to leverage those into long-term relations with HDD customers, which in turn has the effect of upgrading their technological capabilities. Among the more prominent local spinoffs from Maxtor's presence was Eng Teknologi Holdings Bhd. Eng Hardware, the parent firm, began as an engineering services company to the MNCs, particularly in semiconductors; it was not a components supplier and its initial work for the HDD industry was in specialized components and automation systems design and production. However, PDC brokered Eng's initial contact with Maxtor, and Eng exploited its capabilities in machining to become a supplier to Maxtor of actuators for high-end drives; the company has maintained that market niche at the high-end, explicitly rejecting invitations to enter higher-volume business, and has gradually diversified to supply a number of other firms in the industry including Quantum, Seagate, Conner, and Fujitsu. Among the other local firms supplying the industry with components, Microcut Precision and Newtecho Precision Manufacturing produce turned parts (spacer rings, motor hubs and top clamps) primarily for Seagate and Jabil Circuit briefly supplied Maxtor with PCBA.

     A final pattern is local firms which produce automation equipment, tools, jigs and fixtures and provide specialized engineering services; these include Excel Precision, LKT Precision Engineering, and Newtecho Engineering and Newtecho Tooling. These firms got their start servicing other electronics segments, and used their design as well as specialized production capabilities to diversify into the drive industry. Employment tends to be substantially smaller than in the suppliers engaged in high-volume production, but the number of engineers much higher. For example, Excel Precision has 90 employees, eleven of which are engineers, and LKT Precision has 70 employees with 10-15 categorized as engineers. Both companies consider themselves design houses.

     Our survey also asked local firms about the extent of their internationalization, and whether it was motivated by an interest in staying close to HDD customers. Of the seven Malaysian-owned suppliers, five had established international operations, all with one exception in Asia (the Philippines, Thailand and China). Of those five, three cited that the primary reason for doing so was "to be closer to HDD customers who have redistributed operations from Malaysia" while another cited the need to support customers who had established new overseas operations. Interestingly, none stated that reducing costs were the primary motive. Thus the cycle of regionalization that began with the majors moving assembly operations into Penang, followed by the entry of their suppliers, has now been extended by the movement of Malaysian firms into international waters as well.

3. Supplier-Customer Relations in the HDD Industry in Penang

     The previous two sections have suggested that suppliers entered Penang with the purpose of servicing customers who were there, whether assemblers or "major suppliers." We have also shown, however, that the majors and major suppliers have highly developed global as well as regional supply networks that extend far beyond Penang, and that while both local and regional proximity to suppliers may be desirable neither appears necessary. To get a better picture of the significance of regional proximity and more local agglomeration economies, we conducted a survey of 16 suppliers that included both the "major suppliers" and the second-tier of the supply chain. If proximity appears to be important from the perspective of the suppliers, then we would have more confidence in making three claims: that suppliers did in fact enter Penang in order to be proximate, even if their customer base subsequently widened beyond Penang; that assemblers and major suppliers might be inclined to stay because of relationships they have developed; and that even if assemblers and major suppliers relocated, they would be inclined to take their local suppliers with them, assisting in the internationalization of Malaysian firms.

     We are also interested in finding out the extent to which customers in the HDD industry--again, whether assemblers or their suppliers--have contributed to the development of the capabilities of their suppliers. If so, it would suggest that the advantages of such clusters go beyond the wage bill or sales to technological development that will endure even if the industry relocates.

     One indicator of the significance of proximity is the sheer frequency of interaction between customers and suppliers; if proximity were an advantage, we might expect frequent meetings with customers. We asked each supplier about the frequency of interaction with their customers in five areas: to pre-qualify for a job order; to improve product design for manufacturability; to respond to product specification modifications and engineering changes; to qualify, assure, improve, or debug the production process in the ramp-up stage for new products; and to conduct joint R&D or product development. Suppliers were asked whether such interaction took place more than weekly, weekly, monthly, or less than monthly.

     Except for the conduct of joint R and D and product development, which only 7 of the 16 firms reported, all suppliers reported that they had some interaction with their suppliers with respect to the these activities. However, that contact was not particularly frequent. One supplier--a major manufacturer of heads--reported meetings more than once a week across all categories of activities, and a locally-owned actuator supplier reported meeting weekly with their customers. However, the rest reported overwhelmingly that they met customers less than once a month, with no differences across firms based on nationality of ownership.

     However, the sheer frequency of contact may not capture the depth of a relationship. We also asked for firms' own evaluations of the extent to which their HDD customers involved them in six different activities listed in Table 3. Such involvement might be significant even if the customer is not proximate, but we can plausibly assume that proximity should enhance the ability of firms to form close working relationships even if face-to-face interaction is not particularly frequent. One firm reported no involvement across all activities, and on several activities some firms reported no involvement. However on most activities, the involvement was rated as fairly significant. Interestingly, the self-reported level of involvement of local firms was higher than for foreign firms, perhaps because larger customers exercise substantial control over smaller suppliers through vendor development and rating programs.

     We began the study with the expectation that proximity might be particularly important for suppliers that were engaged in new product ramp-up and planning and as the figures in Table 3 show, such involvement seemed reasonably high. We also asked firms if they provided parts and components for routine production needs, for ramp-up of new products or both. ²¹ Despite the claims that such interaction did not involve particularly frequent contacts, all firms in the sample which provided parts and components for routine production reported that they also provided parts and components for ramp-up.

     One presumed benefit of proximity is the ability to develop closer, iterated relationships that rest not solely on performance but on trust and loyalty. We asked each supplier the extent to which their relationship was based solely on price and delivery requirements, with an expectation that they would switch if a better supplier is found; mainly on such requirements, with an attempt to foster a longer-term relationship; or whether the relationship was based on long-term mutual gain and some degree of loyalty. ²² None of the respondents said that the relationship was based solely on performance, but nine of fifteen said that it was based mainly on such expectations. Only six claimed that longer-term relationships and loyalty were important. Interestingly, five of these six were foreign firms, suggesting that the development of closer working relationships may be more likely among firms with more equal technological capabilities, or where the customer is more dependent on supplier capabilities. Only one of seven Malaysian firms claimed that relationship.

     A final measure of the closeness of relations is the extent to which output is customized. We asked the question in several ways, two of which proved relevant. For those suppliers with customers outside of the HDD business (nine) we asked the extent of the customization of the production process to HDD customers vs. non-HDD customers; respondents could answer on a scale from one ("little") to five ("substantial"). With the exception of one local firm, all firms answered "four," suggesting fairly substantial customization. We also asked the extent of customization from one order to the other by the same HDD customer. Of the eleven firms responding, the average answer was exactly 4.0, with no difference between foreign and local firms.

     If the ability to develop close supplier-customer relations is one benefit of proximity, we also considered that suppliers might benefit from "horizontal" connections among themselves; such horizontal connections have been a staple of the literature on industrial districts and agglomeration economies. We asked the supplier sample whether they cooperated with other firms in the HDD supporting industry in one of five ways: joint training of workers; sub-contracting workers to one another when facing supply constraints; referring work to one another to complement each other's capabilities; joint bidding for jobs; and joint R and D. The answers were completely uniform; of the 16 firms in the sample, not a single one, local or foreign, signaled any cooperation in any of these areas. As we will explore in more detail below, we believe that there are "horizontal" connections in the industry that work through social networks, government institutions and programs, and the labor market. However, the findings certainly do not confirm the importance of such horizontal ties.

     A second cluster of questions has to do with effect of serving the hard drive industry on firm capabilities; here, we are particularly interested in the local firms. We asked firms how much they think their HDD customers helped them upgrade technologically, either directly through transfer of product and process technology, or indirectly by providing the opportunity to learn and observe. The results are outlined in Table 4. Local firms believed that they benefited more from both direct and indirect technology transfers than did foreign firms, in part because the larger foreign suppliers were more likely to have their own R & D capabilities. We also asked firms to pick their three top sources of technology. Five of seven local firms, but only three of eight foreign ones, checked "technology transfer from customers." Similarly, four of seven local firms, but only three of eight foreign ones, checked "learning from customer specs."

     In sum, the relatively low frequency of interactions seems to call into question the significance of proximity and thus agglomeration economies. However, frequent, significant contacts can develop over longer distances as can relationships of trust based on high levels of customization. We know that all of the suppliers surveyed had at least some business with HDD customers in Northern Malaysia, and that all had extensive relationships with suppliers and customers in Southeast Asia. These findings suggest that drive industry has benefited from "regional proximity," which marries the advantages of physical propinquity with differences in factor costs.

4. Government Policy: National Incentives; Local Institutions

     Malaysian industrial policy has been a complex mix of offsetting incentives. As in Thailand, the government's policy stance has included both import-substituting policies and other forms of industrial policy aimed at fostering local firms as well as policies aimed at attracting export-oriented foreign investment. After a severe economic downturn in the mid-1980s, there has been a marked shift in the direction of the latter and increasing convergence on what might be called "Singapore standards" with respect to incentives. The electronics sector generally, and the disk drive segment in particular, were beneficiaries of this change, and there is an apparent coincidence between the timing of policy reform and the entry of the HDD industry; as we noted, firms entered Northern Malaysia relatively late when compared to their entry elsewhere in Southeast Asia, only after industrial policy had changed.

     Given that Malaysia is a federal system, it is also important to understand not only the initiatives emanating from Kuala Lumpur but the efforts on the part of the state government of Penang to attract and sustain investment as well. It was at this level that government policy has mattered, but not in the usual sense in which the term "industrial policy" is used in East Asia. Lacking direct financial or tax incentives, the Penang Development Corporation was forced to use other instruments to assist the development of the industry. Three phases can be distinguished. Beginning in 1971, the Penang government relied on the Free Trade Zone mechanism to attract offshore investment in labor-intensive manufacturing; electronics played a central role in this first phase, giving Penang important advantages with respect to the quality of managers, technicians and production workers.

     In the aftermath of the downturn in the mid-1980s, PDC invested heavily in attracting investments in new activities; the entry of Maxtor and Applied Magnetics reflected in part this new thrust as did a new wave of Japanese investment in other sectors. With increasing tightening in the labor markets, however, PDC's emphasis gradually shifted toward greater selectivity and the provision of more sophisticated "network" objectives, including forging relations with local suppliers and developing training programs that matched evolving labor market needs.

     The Course of National Policy

     Malaysia's first experimentation with an outward-oriented strategy for the manufacturing sector came with generous tax exemptions for export-oriented firms in the Industrial Incentives Act of 1968 and amendments to labor laws in 1969 which allowed for three work shifts. But it was the enactment of the Free Trade Zone (FTZ) Act (1971) and the proliferation of Licensed Manufacturing Warehouses (LMW's) that drew the first major wave of export-oriented electronics manufacturing, concentrated initially in components. ²³ A key feature of the FTZ legislation was that it allowed 100% foreign equity ownership at a time when the general thrust of Malaysian industrial policy, as outlined in the New Economic Policy (1970) and the Industrial Coordination Act (1975), was moving to support indigenous Malay (bumiputera) interests by requiring their involvement in new investments.

     These investments were also supported by perhaps the most important financial incentive, the granting of Pioneer Status. Pioneer Status granted a partial exemption from payment of income tax for firms meeting changing criteria; companies granted PS status pay tax on only 30 percent of their statutory income with five years of tax exemption commencing from the production date. The HDD majors were routinely able to meet the criteria; all of the HDA assemblers that entered Penang received Pioneer Status and all that survived longer than the initial five-year period were able to renew it by claiming an upgrading of their product mix. Our survey of suppliers found that six of eight foreign-owned suppliers received Pioneer Status and three of seven local ones did, although it had either lapsed or was set to expire for all of the local firms, a policy which deserves reconsideration.

     The first disk drive investment came to Malaysia during the Fifth Malaysia Plan (5MP, 1986-91) period, and it is possible that this relatively late entry was a result of the direction of government policy in the early 1980s. The manufacturing sector, which accounted for about one-fifth of the GDP in 1985, had played a relatively modest role in realizing the growth, employment creation and regional development objectives of the Fourth Master Plan (1980-1985), and the growth of foreign investment in the electronics industry slowed. The nature of incentives naturally came under scrutiny at the time of the economic downturn of the mid-1980s. During the early 1980s, Prime Minister Mahathir had experimented with his "Look East" industrial policy that implied a much more rigid and static policy course.

     In drafting the 5MP, however, the manufacturing sector was targeted to play a much more important role, and the development of more efficient and dynamic export-oriented industries--both foreign and domestic owned--was seen as key to that strategy. Beginning with the 5MP, the government undertook a number of important changes in industrial incentives; the most important were:

• Amendments to the 1975 Industrial Co-ordination Act (ICA)
• Promotion of Investments Act 1986 (PIA)
• The Industrial Master Plan (IMP) (1986-1995) and
• The Second Industrial Master Plan (IMP2) (1996-2005).

     The Investment Incentives Act (IIA) of 1968 provided relief from payment of various forms of taxes. Incentives given included: Pioneer Status; Labour Utilization Relief; Locational Incentives; Investment Tax Credit; Export Incentives; and Increased Capital Allowances. Today, under the 1986 PIA, incentives given to manufacturing companies have expanded to include:

• Investment Tax Allowance;
• Reinvestment Allowance;
• Incentives for Industrial Adjustment;
• Incentives for High Technology Industries;
• Incentives for Strategic Projects;
• Incentives for Small Scale Industries;
• Incentives for Operational Headquarters;
• Incentives for Research & Development;
• Incentives for Training; and
• Incentives for Exports.

     The reform of the IIA in the PIA was paralleled by the new Industrial Master Plan 1986-1995 (IMP), which was characterized by an outward-oriented industrialization approach, in which electronic components were considered strategic. The IMP identified three policy instruments for increasing technology capability, namely, research manpower; institutional arrangements, such as industrial parks; and incentives for R&D. The Second Industrial Master Plan 1996-2005 (IMP2) superseded the IMP 1986-1995 and extended its approach. Beyond the focus on manufacturing operations, the IMP2 included R&D and design capability, development of integrated supporting industries, packaging, distribution and marketing activities. Emphasis was given to deepening integration of manufacturing operations along the value chain to enhance industrial linkages, increase productivity and competitiveness.

     Borrowing a page out of Singapore's play book, the IMP2 was based on the cluster approach towards industrial development that emphasizes the growth not only of the manufacturing sector but of the supporting industries, to include the services sector. Electronics was identified as one type of cluster--and internationally-linked one--and within electronics, computers and peripherals were identified as one of six sub-clusters.

     Again, however, we did not find evidence among either the majors or from our census of suppliers that these new incentives mattered, again, perhaps because of the greater significance of Pioneer Status. The convergence toward "Singapore standards" may have had a general effect on perceptions of the investment climate in Malaysia, but with a few exceptions noted by PDC staff, such as the inability to offer outright grants, as Singapore has on occasion, we found little evidence that national-level incentives differentiated Malaysia from its neighbors. ²⁴ Firms enjoyed free trade and an effective exemption from taxation, although with greater restrictions on the granting of pioneer status, the other incentives, especially the reinvestment allowance, became more significant over time. Many firms were upset by additional qualifications (PER) introduced in 1997, but these were suspended in response to the crisis.

     Industrial Policy: the State Level

     Penang was well-positioned to exploit the policy changes of the early 1970s, and had in fact pioneered them. Following a change in Penang state government following the elections of May 10, 1969, a developmentally-oriented administration under Dr. Lim Chong Eu, from the opposition Gerakan Party came to office. He quickly reached a political modus vivendi with Kuala Lumpur (ultimately aligning with the National Front), and established the Penang Development Corporation as the state's development agency. Penang did not enjoy policy independence; basic financial and fiscal incentives were set at the Federal level in Kuala Lumpur, and were addressed increasingly to bumiputera concerns. The PDC and state government thus concentrated their efforts in two basic areas: promotion and marketing of Penang overseas and coordination, facilitation and expedition of investor proposals with Kuala Lumpur; and the provision of basic infrastructure, including not only the administration and development of FTZs and industrial states, but complementary land and new township development.

     This, in turn, required an organizational structure that had at least some independence from the national level. Interestingly, the PDC was formed at roughly the same time as the Malaysian Industrial Development Authority (then call the Federal Industrial Development Authority, FIDA) itself. The first Director General of MIDA was a German, but with a prominent civil servant from Penang as General Manager. Malaysia's relatively centralized federal structure meant that while there were separate levels of government, the civil service at the state level was made up overwhelming of federal civil servants; the state governments only had control over appointments to municipal councils. Thus the creation of PDC was motivated in part by the desire to gain independence from the center in hiring. PDC initially had to work with the state government departments, but over time they developed their own in-house capabilities in engineering, land development, and housing, growing to a staff of over 400 which developed a strong and close relationship with MIDA.

     The manufacturing sector in Penang was not strong before the 1970s. Commerce and trade dominated Penang's economy, with manufacturing concentrated in the traditional consumer goods sector: food, beverage and tobacco; paper and printing; and apparel, leather and footwear. In 1969, the Penang Master Plan Study was undertaken by a US-based consultant company. The central thrust of that report was to revitalize the regional economy through greater foreign linkages and ties with multinationals. This objective was advanced by the establishment of the first Free Trade Zone (FTZ) in Malaysia in Penang in1971, and the rapid entry of a number of electronics firms, particularly in semiconductor packaging. By 1973, the electronics & electrical industry provided one-fifth of all manufacturing jobs in the State and contributed 28 percent of Penang's manufacturing value-added. There can be little question that the success of other manufacturing operations in the electronics sector played an important role in attracting the HDD industry.

     While there is some evidence that the state government, through PDC, also sought to develop linkages between foreign-invested firms and local suppliers where possible, our survey finds that only one of seven firms attributes its connections with an HDD customer to the intervention of the PDC; business associations and direct contacts were more frequently cited. But these numbers underestimate the indirect effects of the PDC through its efforts to attract the majors. One important source of local linkages was the purchasing agents of foreign firms, many of whom were Malaysian nationals and given a relatively free hand in sourcing locally where feasible; interviews with Conner officials and suppliers suggest that they were particularly open to developing new suppliers.

     By the early 1980s, the dominance of the electronics sector, and semiconductors in particular, created what were seen as a number of vulnerabilities for Penang. These included the narrowness of the industrial base, the dominance of multinationals, and particular exposure to the highly volatile semiconductor segment. The shake-out in the electronics industry during the downturn of 1985 created a surplus of semi-skilled labor in Penang. As a result, the State Government of Penang directed the Penang Development Corporation in 1987 to identify new industries for priority promotion, search for new markets, accelerate their efforts with respect to training, improve administrative efficiency and pay greater attention to opportunities for local linkages.

     By its own admission, PDC did not have a clear map of the disk drive industry as a whole in the late 1980s, but it nonetheless played a crucial role in brokering the entry of Maxtor, which in turn had important demonstration effects, as well as Applied Magnetics and later Komag. The entry of the pioneer firms in the industry corresponded with an unprecedented boom in Penang, succinctly captured by an increase in the manufacturing sector's share of state product from 15% in 1980 to 46% in 1990. The boom, in turn, generated a new set of problems, the most significant of which was rising labor costs and shortages of both skilled and even unskilled workers.

     In the light of these changes, the PDC drew up new objectives to deepen its economic base as well as to promote inter-and intra-sectoral linkages in the Penang Strategic Development Plan (PSDP) 1991-2000. The objectives of the new plan included:

• to transform the structure of manufacturing towards increasing capital- skills- and technology-intensive systems of production;
• to diversify the manufacturing base, increase inter-industry linkages and encourage greater local participation through joint-ventures;
• to develop a highly skilled workforce;
• to develop Penang into a regional manufacturing and research center, for the northern region as well as the entire Asia Pacific; and
• to enhance the participation of Bumiputera and other economically disadvantaged groups in the manufacturing sector.

     Deregulation, liberalization and provision of infrastructure were seen as necessary but not sufficient conditions for upgrading industrial capabilities. Rather, policy recommendations identified under the PSDP 1991-2000, and reiterated in later policy documents in the mid-1990s and following the crisis of 1997-98, included a more fundamental restructuring of the manufacturing sector to develop Penang into an Integrated Manufacturing Centre (IMC) that would have design and R & D capabilities and house not only manufacturing but marketing and support services as well.

     Tight labor markets, rising wages and increasing dependence on foreign labor provided both the impetus, and one of the main instruments, for transforming the regional economy towards more capital- and skills-intensive production. The PSDP called for an explicit reorientation of the government's investment promotion strategy away from labor-intensive manufacturing. PDC officials relate increasing focus on the headcount and cite specific examples in which HDD-related investments were discouraged because of their high demand for low-skilled labor. Outside of Seagate's expansion in the early 1990s, which occurred partly as a result of acquisitions and were controversial in Penang, the last heads-related investment in Penang took place in 1991, and we saw that both Read-Rite and Seagate's HGA-HSA operations were consolidated out of Penang in 1998. Rather, the government favored firms that supported the large companies in providing high value-added intermediate inputs, capital equipment and engineering support, such as the metal-based, iron and steel, engineering, plastics and rubber products industries.

     However, the government could not control the expansion plans of existing enterprises, some of which continued to grow despite rising real wages. Moreover, the crisis has changed labor market conditions and provided a breathing space for the government to exploit. With labor costs falling, but technical capabilities of managers, engineers and workers essentially unchanged, Penang is positioned to gain relative to other locations where labor costs have not dropped as sharply, such as Singapore.

     A related set of policies centered on education: skills development, the public training system, university training, import of skilled technical workers (short-term), and manpower planning. Central to this effort was the launching of the Penang Skills Development Centre in 1989. The central feature of the PSDC's success has been its demand-oriented structure. Firms communicate their training needs to PSDC management, and donate equipment and even trainers, while funding comes from mandatory concentration.

     A second objective was to encourage the development of supporting industries through a variety of measures ranging from financial assistance, development of a SMI Centre and technical support system and providing common-user R and D and design facilities. Through the crisis of 1997-98, there were efforts in this direction, but apart from small successes, it is not clear that these efforts were beneficial to the industries. The 1998 PC and Peripherals Task Force report reiterated the importance of these efforts, however, and included calls for more attention to vendor development, including through the creation of a Cooperative Research Centre.

     By Way of Conclusion: Policy Implications for Penang

     Our study of the industry in Penang suggests a number of conclusions for the analysis of the industry and for public policy in Malaysia and Penang. First, free trade, liberal investment incentives, and quality infrastructure, have been important in allowing the industry to flourish, but in part because they contributed to the location of linked industries, follow-the-leader behavior and resultant agglomeration economies. These agglomeration economies have continued to act as a pull on foreign investors, most recently in the development of a media-substrates cluster.

     Despite the international furor surrounding Malaysia's capital controls, the crisis has in fact given the government strong incentives to continue and even deepen its open stance towards multinationals. Although the capital controls have imposed some additional uncertainties and paperwork on firms, we found little evidence that they have impeded the investments of existing firms. However, the crisis could affect the ability of the government to continue to provide high-quality infrastructure. For example, air cargo handling facilities at the Penang International Airport are already operating at capacity, and despite the economic slowdown, air cargo traffic has been increasing during the first six months of 1998. Although plans to expand the facilities are going ahead, they have been cautious and slow.

     A crucial component of Penang's advantage is the availability of a pool of managerial, engineering and skilled labor; our interviews with companies found that they were able to easily enter Penang and ramp up production because they could draw on labor with industry-specific skills. Given labor shortages and rising labor costs, it has also been important to continually upgrade labor at all levels, and we have seen that Penang has assisted firms in that effort through the PSDC and other programs. However, the crisis creates the temptation to slacken these efforts by relying on lower costs to sustain competitiveness. Our analysis of Singapore suggests that this would be a serious error; even more attention has to be paid to the types of education and training that are relevant to the HDD industry, which includes not only the kind of technical training available through PSDC, but the engineering skills that need the engagement of the local university USM. Discussion with industry indicates that while many of them are satisfied and comfortable with the graduates from the local universities, the number of graduates are still too small to meet the needs of the industry. The location of the USM engineering school in Perak was costly for Penang, since it prohibits the development of close relationships between the university and industry that is a hallmark of Singapore's industrial policy efforts in the sector.

     If we believe that the linkages between MNCs and local firms have played an important role in the development of technological capabilities in Penang, it raises the question of how to foster those linkages more effectively, including through direct supports, or whether such linkages are best left to the market and provision of information to foreign investors. While vendor development in the Penang region appears to have been private-sector driven, PDC has had an important role in the process. In the 1980s, PDC tried to encourage the MNCs to increase their local inputs, but without the ability to offer grants or even incentives, PDC's role was restricted to development of serviced industrial areas adjacent to the FTZs and to organizing vendor/procurement exhibitions and meetings. PDC compiled and produced the first directories of supporting industries to make the capabilities of local industry known to foreign investors. PDC also continually submitted proposals to the Federal level, although it is difficult to claim that the programs and incentives that are now in place were attributed to Penang's initiatives.

     The small and medium scale industry centre (PIKS) was established under the auspices of the Penang Industrial Council (PIC) in 1992, and suggests how the development of agglomeration economies in the market can be augmented through more formal organizational efforts. The PIKS provides secretariat support to the Penang State Industrial Transformation and SMI Development Committee; as of December 31, 1997, the PIKS has 162 members.

     The objectives of PIKS include acting as a means for the government to plan, coordinate, and implement programs and projects for the development of SMIs in Penang. The PIKS helps to overcome many of the informational problems that small entrants might face in seeking to supply the drive industry, including knowledge of relevant regulations, incentives, financial packages and assistance related to the SMIs as well as efforts to establish linkages with MNCs. Since its inception the PIKS has organised 23 programmes on finance, market & vendor development programs, technology enhancement, grants, human resources development, quality management and occupational safety. These activities reflect a new stage in Penang's industrial policy efforts, and seem congruent with the foundations of the industry's development that we have traced in this report.

Endnotes

• ¹ The Northern Region of Malaysia encompasses the four states of Penang, Kedah, Northern Perak and Perlis. Although the bulk of investments are in Penang, Seagate has facilities in Ipoh, Perak and the Kulim Industrial Park in Kedah has attracted a cluster of media investments. Investments in Southern Malaysia are nest understood as an outgrowth of Singapore operations. This is true of Western Digital's assembly in Kuala Lumpur - the only HDD investment of significance in the Klang Valley - and the rapid growth in the 1990s of both major and supplier investments in Johore.
• ² The other facilities are in Singapore and in Wuxi and Shenzhen in China
• ³ Local capabilities have also developed in printed circuit board assembly (PCBA), although not to the extent visible in Johore in the southern part of the country proximate to Singapore.
• ⁴ Seagate's other investments in Penang and Perak, which are substantially larger operations in terms of both capital investment and employment though not sales, are concentrated in heads, HGA and HSA.
• ⁵ There was a lack of industrial land on the island, the reclamation of land there had not started, but there was discussion of several other facilities. H-P had recently moved into a new building just as the recession hit in 1985; Atlas - Intek had five buildings and when they closed in 1985, Applied Magnetics took one of them
• ⁶ This began with the Cabo 217, a 275 Mb drive.
• ⁷ "Seagate's Secret," Electronic Business Asia, Feb. 1997, pp. 42-3
• ⁸ There was a reluctance to sell the entire operation because of fear that associated staff would leave with the product.
• ⁹ Sales and marketing are managed from the United States.
• ¹⁰ Read-Rite's slider fab uses a less demanding grinding process for shaping the underside of the slider.
• ¹¹ Business Wire, May 5 1994
• ¹² See David McKendrick, "Profile of Suppliers to the HDD Industry: Media and Heads," UCSD, June 21, 1996.
• ¹³ In addition, Komag has substantial investment in the Sama Jaya Free Industrial Zone in Sarawak
• ¹⁴ Stephen C. Johnson, president and chief executive officer of Komag, Incorporated, PR Newswire, August 20, 1997.
• ¹⁵ When Komag entered Japan in 1987, they formed a number of joint ventures, including one with Kobe steel, Komag Material Technologies, to provide blanks; Komag thus had a prior connection with Kobe.
• ¹⁶ Nikkei English News, March 3, 1997.
• ¹⁷ MMK also announced plans to double capacity during a second phase of the project after 2000. Asia Pulse, July 4, 1997.
• ¹⁸ Comline Daily News: Computers, June 11, 1997; Asia Pulse, December 17, 1997; Agence France Press, February 11, 1998
• ¹⁹ See Rajah Rasiah, Foreign Capital and Industrialization in Malaysia, New York: St. Martins, 1995.
• ²⁰ Far Eastern Economic Review, August 27, 1998, p. 55-6
• ²¹ Suppliers of engineering services were asked if they provided continuos process improvement and maintenance, process engineering for new products/process lines, or both.
• ²² See questionnaire for precise wording
• ²³ FTZs involve the provision of infrastructure by the government, and have the advantage for some industries of housing suppliers and customers in close proximity. LMWs provide a mechanism for the government to extend the trade provisions of the FTZs to individual firms by essentially bonding them. The only drawbacks to the latter are that it does not have the proximity benefits of the FTZ, and LMW status must be renewed annually. However, the case of securing LMW status, the development of other industrial parks, and the general lowering of tariffs have gradually eroded the particular advantages of being housed in the FTZ. In Penang, most of the majors and major suppliers are located in an FTZ, while those which are not in FTZs, enjoy LMW status.
• ²⁴ Allen Hicken, "FDI, Tax Incentives, and Hard Disk Drives," UCSD Globalization Project report