Taiwan is the world’s largest producer of motherboards, monitors and modems. In 1994, it turned out 7.4 million motherboards – the main printed circuit boards in computers – while producing 12.7 million monitors domestically and another 4 million in Taiwanese factories abroad. It holds a 55% global market share in modems. But it is the manner in which it has achieved this status that is as impressive as the numbers themselves.
Computers and consumer electronics items traditionally have been made by large, vertically integrated conglomerates that produce most of the component parts in-house. Samsung of Korea is a typical example. But such huge manufacturing structures as the Korean chaebol and the Japanese keiretsu fit uneasily with Taiwan’s Chinese society. The Taiwanese, have disaggregated the production process, permitting much greater flexibility in production. While in many areas of industry this can be a severe disadvantage, in electronics it has benefits.
The structure has reached its apogee in Taiwan in the manufacture of integrated circuits (ICs), commonly known as computer chips. This process can be split into four main stages: design, when the circuits for the chips are created; fabrication, when the circuits are etched with lasers onto wafers of silicon; packaging, when the individual dies are cut from the wafers and set into gold-leaf boards, attached to outside connecting pins and bound in a resin or ceramic; and testing, when the packaged chips are tested before being sent to the end-user.
Since the mid-1980s, Taiwanese companies have been offering these services separately. In particular, Taiwan Semiconductor Microelectronics Corporation (TSMC) became the first company in the world to offer a “foundry” service for the manufacture of wafers. In its fabrication plants (“fabs”) in Hsinchu (north-east Taiwan), TSMC can make to order wafers for other companies. Initially, many of these were integrated manufacturers needing extra capacity. But there were also increasing numbers of small “fabless” design houses, which in pre-foundry days, had to leave production to their potential rivals – the integrated manufacturers. With the invention of foundries, “fabless” houses boomed.
Taiwan’s leap into the new electronic world has created some of the fastest-growing and most profitable companies ever. In the past two years alone, the Taiwanese electronics sector has experienced huge earnings growth rates. Peregrine Securities estimates the industry had net profit growth of 111% in 1995. Individual companies have registered rates even higher. For example, Orient Semiconductor Electronics (OSE), one of Taiwan’s oldest electronics firms, is likely to have net profit growth of 266% in 1995 after 487% year-on-year growth in the first nine months.
The traditional mainstay of the Taiwanese electronics sector is the production of computer motherboards, with Taiwanese companies providing roughly 80% of world supplies. Imagine, then, the consternation when mighty Intel, which dominates production of the CPUs, set up its own motherboard assembly facility in 1995. Intel was aggrieved by the failure of Taiwanese manufacturers to upgrade to Pentium-compatible products.
Intel’s decision caused analysts to revise downwards Taiwan’s domestic motherboard production to 12 million units, an annual increase of only 5%. Longer term, the decision could imply that Intel may control the core technology and input prices. Signs are, however, that Intel has bitten off too much. Observers speculate that Intel undershot its 1995 sales target of 10 million units by as much as 30%. Most suggest that Intel’s move was an error – motherboard manufacturing is comparatively labour-intensive with a low profit margin. The company is trying to scale back its production, which should allow the Taiwanese to bounce back.
“Unlike CPUs, which are standardized, motherboard production requires close customer relationships as each is tailored to customer specifications,” says Peter Lin, an analyst at Masterlink Securities.
One major effect of Intel’s decision to go into motherboard production was to create a sudden surge in the prices of motherboard components, especially ICs.
However, when Intel failed to sell as many motherboards as planned, it dismantled its stockpile and the chips flooded on to the market. This sudden oversupply led to a rapid fall in prices. According to Eugene Ha, an analyst at Peregrine, prices of the currently standard 4-megabyte DRAM have fallen from $13 per megabyte in the second half of 1995 to $11 now. SRAM prices have crashed: 256k synchronous SRAM prices dropped from $4 to $1.50, while the newer 1-megabyte SRAMs have fallen from $25 as late as the third quarter of last year to $6 now.
Sheng-Yui Wang, supervisor of the finance department of United Microelectronics Corporation (UMC), says: “At $6, 1-megabyte SRAMs are below cost for some smaller companies or fabless design houses. Many have stopped producing already”. Opinions differ on whether there will be an oversupply. Makio Inui, electricals analyst with Dresdner Bank-Kleinwort Benson in Tokyo, believes that demand growth has exceeded expectations; and even though supply has beaten many forecasts, he is “not too worried” about oversupply into next year. “The growth rates might be peaking,” he concedes, “but, even so, the subsequent profit levels are so huge you have to invest in them as value stocks.”
However, because Taiwan is a relative newcomer to higher-yielding DRAM production, it will be more affected by the slump in SRAM prices. This will be partly offset by growth in higher-priced new generation SRAMs and strategic deals with major end-users; an example is the recent agreement between Winbond (a leading Taiwan electronics company) and Toshiba.
Nevertheless, SRAM suppliers are viewed cautiously. “SRAMs will experience a squeeze in 1996,” says Stan Shpetner at Crédit Lyonnais Securities Asia in Taipei. “Winbond, which derives 60% of its business from SRAMs, and which saw earnings growth in 1995 of 300%, is likely to see growth this year of under 20%.”
Even DRAMs may hit difficulties in the medium term. Despite the optimism of analysts such as Inui, many observers in the region consider an oversupply possible, beginning in 1997. According to SBC Warburg, there are around 100 new fabricators planned or under construction worldwide. In Taiwan, several are on the way. Most of these are DRAM facilities which will come on stream between 1997 and 1998.
In any oversupply, DRAM fabricators will also face a margin squeeze, meaning there are bound to be some losers. The Koreans, Taiwan’s biggest competitors, are confident that it will not be them. “Korea has a technological advantage over Taiwan and Japan in DRAMs,” says Jong Park, senior analyst at LG Securities in Seoul. “Korean companies will have finished their investment in 16-megabyte DRAMs by the end of 1995 and will be in full production by 1996, whereas Taiwan won’t start until 1997. By then, the Korean companies will have a yield ratio – meaning the number of viable chips per silicon wafer – of between 80% and 90%. The Taiwanese will achieve only 70%. They are four to five years behind.”
Shpetner agrees that the Koreans are a big threat. “If I were a Taiwanese manufacturer,” he comments, “I’d be scared to death of Samsung. It has aggressive accounting policies that allow it to write off costs over just one year, which gives it tremendous scope for price-cutting.” Samsung is reported to have signed five-year forward supply contracts with six major US computer manufacturers totalling US$65 billion. This means that it has tied up 15% of the world market up to the year 2000. More recently Samsung also has done a deal with Intel, thought to be an exclusive agreement to supply synchronous DRAMs for the next generation of CPUs, which will replace the Pentium series.
The Taiwanese may yet withstand the onslaught. “We are not worried by Samsung,” claims UMC’s Wang. “At each stage of production, Taiwanese companies are 100% more efficient than those in Korea. It is easier for us to adopt new equipment and technology and we have much quicker decision-making processes.”
Though this smacks of overstatement, the Taiwanese have so far enjoyed significant margin advantages over rivals, including the Koreans, in IC production. According to Capital Securities, operating profit margins for TSMC and UMC are generally over 40%, about double that of Samsung and 10 times that of the Japanese. Dixon Ho, analyst in Taipei for UBS Securities, expects margins to fall to more normal levels of between 25% and 30%.
Such a level of success reflects a fortuitous occurrence of events. In the 1980s the government focused resources on the electronics industry and set up the Hsinchu Science Park on a greenfield site south of the capital, Taipei. Generous tax-breaks were introduced for high-technology companies. Government institutes were established to carry out the research and development with which Taiwan’s relatively small, family-owned businesses were reluctant to burden themselves. At the same time, many American-educated Taiwanese working in the major US companies – Intel, Motorola, IBM – sensed new opportunities back home amid Taiwan’s increasing prosperity and growing computer industry.
The result was a dynamic mixture of interests. Government research institutes were corporatized, privatized and taken over by entrepreneurs, both home-grown and with experience in the US. UMC, for example, one of the leading local integrated-circuit producers, was spun off from the government’s Electronics Research and Service Organization. Its major shareholders are the state-run Chiao Tung Bank and the ministry of economic affairs, as well as two listed private companies, Teco Electric & Machinery and Yao Hua Glass.
However, the big problem facing Taiwanese stocks is Beijing’s refusal to to recognize Taiwan as an independent state and to regard the island as anything other than an integral part of the People’s Republic. Tensions have heightened greatly since Beijing began test-firing missiles from the mainland into waters provocatively close to Taiwanese shores last summer. Last month’s resumption of testing and subsequent naval and air manoeuvres, bringing the US navy rushing to the area, hardly helped further the cause of Taiwanese stocks. Price/earnings (p/e) ratios of the island’s technology stocks are in the range of 11 times earnings, lower than those of many counterparts in other markets.
Both investors and electronics companies are keen for a relaxation of tensions to allow the Taiwanese stock market to recover from its current slump. And the companies need to raise cash to finance new facilities in their highly capital-intensive industry. But apart from the political tensions, there would appear little downside. In its September 1995 review of the global semiconductors industry, Morgan Stanley forecast robust demand and slapped on the front page as a sub-title: Taiwan: explosive growth to continue.
Nevertheless, in recent months some caution has crept in among industry observers. Fundamentally, there are fears of lower PC sales and of an impending global oversupply in semiconductors. “In this type of industry, low p/e ratios are not necessarily attractive,” explains Terry Wou, assistant fund manager at Thornton Asset Management in Hong Kong. “They mean that the earnings are already huge. It indicates a time to sell.”
Longer term, no-one expects anything but growth for the electronics sector. But there will be a shift in emphasis from consumer-electronic products towards more computer-based products. “The main challenge to the consumer-electronics industry is that most of its mainstay products will eventually disappear,” reports SBC Warburg’s 1996 Technology Outlook. “The soon-to-be ubiquitous PC with its on-line connection and CD-ROM capability is likely to replace TVs…audios and VCRs…video-game machines…and ultimately even DVD (digital video-disk).”
There is ample room for expansion of PC sales worldwide. Global penetration of the market stands at only 4%. (The highest penetration by country is the 36%, measured by PC per person, registered in the US). Even many countries with a high per-capita GDP have low PC sales, especially in Asia. New systems that make Chinese and Japanese character inputting easier to manage should help to narrow this gap, making Asia the fastest growing market for the next few years.
Morgan Stanley’s review of semiconductors puts global PC sales growth at 20% in 1996 and 17% compounded to 2000. But many are questioning whether growth will be so vigorous and the industry had a very poor fourth quarter in 1995. This is attributed to the failure of Microsoft Corporation’s Windows 95 software to meet sales expectations. Christmas retail sales in the US were disappointing and many corporations have not yet upgraded to Pentium-based machines and instead seem to have been waiting for the just-launched Intel P6 microprocessor, which is designed to run specifically with the Windows NT operating system.
Yipin Lee, director of corporate finance and investment management at Acer Incorporated, Taiwan’s premier computer manufacturer, forecasts a 15% rise in global PC sales this year, while Lin at Masterlink Securities in Taipei says: “Global PC growth should be in the range of 15% to 20% in 1996, hitting a sales volume of 67 million units. Sales growth should stabilize thereafter, reaching 100 million units by 2000. This is much later than Intel’s original 1998 best-case sales estimate of 100 million units.”
This is clearly not the best news for computer manufacturers, but it is not very bad news either. Profit growth will slacken. Capital Securities Corporation, a Taipei-based brokerage house, estimates that 1995 revenues for Acer were NT$52.5 billion ($1.9 billion), up 58% year-on-year. Its forecast for 1996 revenue growth is a still-healthy 33%.
The reason for relatively robust profits is the falling price of microchips, one of the main cost components in computers. In the short term this should permit computer manufacturers to maintain margins. In the medium term, they will be able to pass on these lower component prices to consumers, which should give a great boost to sales.
“The advent of a $1,000 machine capable of running powerful software is not far away,” says Acer’s Lee. “This will give a tremendous lift to computer sales around the world.”At the higher end of the market, meanwhile, companies and individuals will upgrade machines to cope with the demands of ever-increasing memory-eating software.
In some ways, Taiwan’s computer makers are less likely than others to suffer in the event of lower sales. Taiwan accounts for only a little over 8% of world shipments of desktop computers but about 28% of portables. Growth in portables is forecast to outpace that of desktops significantly. Prices of thin-film transistor liquid crystal display screens, which represent about 40% of the cost of notebooks (portables), fell by 37% at the end of 1995. Technical advances mean that CD-ROMs can now be fitted. Dataquest, a specialist research firm, predicts that this will enhance portable sales by 35% in 1996.
If the performance of the computer, motherboard and IC manufacturers looks mixed, that of the packaging and testing sector is generally strong. The largest IC packager in the world is Anam of Korea, but Taiwan has a vibrant sector led by world number two Advanced Semiconductor Engineering (ASE), world number three Siliconware Precision Industries Limited and OSE.
Packaging services are in great demand worldwide. Order books are full and likely to remain so for some time. But it is a relatively labour-intensive activity and has lower profit margins than chip production. ASE assistant vice-president Jeffrey Chen estimates that each $100 of capital invested in fabrication produces around $100 in revenues and $50 of profit. Packaging, by contrast,
produces $200 in revenues but only $36 in profits.
As revenues come under pressure, integrated manufacturers will move to outsource packaging to a greater degree than now. Dataquest (a sector analyst) forecasts that independent IC packaging will increase from 17% of the current packaging market to 40% by 1998 – compound annual growth rate over the period of 43%. Unlike fabrication plants, which have long lead-times, packaging plants are quick to set up. Therefore they can respond to demand much more easily than fabs, leading to less volatile earnings. “Risk is lower than the upstream side but packaging still carries great upside potential,” says Chitung Liu, associate director of equitiy research at SBC Warburg in Taipei. “Since packagers are expanding slower than fabs, there will continue to be a capacity shortage.”
Regional fund managers and brokers recommend packaging companies as solid buys. “Ninety per cent of chips are packaged in the Far East,” says Chen of ASE, “but they are typically shipped to the US – where they were designed – for testing and then shipped back to the Far East for assembly.”
But price pressures are forcing change and testing is beginning to move out of the US and into Taiwan. Testing is capital-intensive, like fabrication, yet is even less profitable (measured by return on equity) than packaging. The same $100 that produced profits of $36 in packaging would produce profits of only $20 in testing, on revenues of $65.
In addition to savings, outsourcing testing to Taiwan should enhance computer manufacturers’ efficiency. “The product life-cycle is now so short, only six to nine months, that it makes no sense to ship to the US and back again,” says Chen. Companies such as ASE are therefore offering turnkey services where packaging and testing are carried out at the same site.
Chen estimates that testing will grow at a rate about 50% more than packaging. “Our unlisted testing subsidiary, ASE Test, has seen earnings growth in excess of 100% for the past three years,” he says. “It will continue to grow very fast and our plan is to seek an overseas listing – a sub-IPO [initial public offering] of ASE Inc.”
In looking to the next several years, Taiwan’s electronics industry appears to be set along a course of continuing expansion at a rate faster than the global industry. Morgan Stanley expects Taiwan to expand its market share of ICs from 5.6% in 1994 to 7% by 2000.
But there are challenges which could stifle some of the industry’s growth if not promptly addressed. One is cost. Lower-end production will continue to move to other parts of Asia. Acer already manufactures in the Philippines and Malaysia and ASE imports 20% of its workforce from the Philippines.
A more critical area is research and development (R&D). “There is a lack of commitment to R&D on the part of Taiwanese companies,” complains Elizabeth Sun, vice-president of the international department of Capital Securities in Taipei. “They typically budget less than 10%. In a fast-changing industry such as electronics, this means Taiwan will always be a follower.”
Sun’s colleague, senior associate Chris Hsieh, feels that the strategic alliances being forged by the major industry players, such as those between Samsung, NEC and Intel, also pose a threat. While Taiwanese companies have also been forging alliances, their lower technology levels could leave them with less bargaining power.
If Taiwan continues to lag in technology, its electronics industry will depend on squeezing even more efficiency out of the manufacturing process. Thus far, it has achieved wonders. But the economic miracle will be harder to sustain in the longer term.
Fab time at the foundry
Taiwan Semiconductor Microelectronics Corporation (TSMC) is the only make-to-order electronics “foundry” company in Taiwan – known as fabs – for silicon wafers used in the integrated circuits of computers. And business is booming.
With growing numbers of design houses without fabrication facilities (“fabless”) the reliance on foundry production is increasing. But, the high capital costs involved in the construction of fabrication (foundry) plants – about $1 billion – mean that integrated manufacturers are increasingly contracting out portions of their production to foundries. As a result, many companies are climbing on to the TSMC-led bandwagon.
As demand for DRAMs, which provide memory storage in a computer’s CPU, is expected to outpace general demand for semiconductors, foundry production is forecast to grow faster than IC production generally.
The response to declining margins of many companies which concentrated on the manufacture of SRAMs, which help to speed passage of information, has been to build DRAM foundry fabrication plants. United Microelectronics Corporation (UMC) has three joint-venture foundries planned and under construction, in conjunction with 10 – mainly US – partners. And the lucrative foundry business has led to a number of Taiwanese companies from very different areas of activity to get in on the act – Nan Ya Plastics being one of the most high-profile.
Newcomers will need to establish a reputation for consistent high-quality production. To achieve this they will have to attract sufficient numbers of engineers – and these professionals now come at a considerable premium in Taiwan. Engineers are attracted to companies through huge stock bonuses, but new companies – of necessity unlisted – cannot offer the same incentives. As a result, they may suffer from skill shortages or have to make huge cash payments that will hurt the bottom line.
Many investors are unaware that the stock bonuses given by firms may be affecting company profits. Under Taiwan’s accounting rules, such payments do not pass through the accounts. Fabs typically pay about 10% of net income in stock bonuses. If listed companies overall make NT$10 billion in one year, or NT$9 billion after legal reserves, this level of stock bonuses accounts for NT$900 million, which at a par value of NT$10 per share makes 90 million shares. At average actual share-price levels, this translates into NT$5.4 billion. This would reduce net profits to NT$4.6 billion, in turn doubling the average p/e ratio.
China factor threatens fundraising opportunities
The Taiwanese banking sector has warmed to high-tech companies and their strong cash-flows. The main players regularly raise funds through syndicated loans, at fine pricing. Last November, for example, Advanced Semiconductor Engineering (ASE) got a $100 million five-year loan at Libor plus 65 basis points (bp). United Microelectronic Corporation’s (UMC) first joint-venture fabrication plant, into which shareholders have injected $600 million in equity, has recently signed a two-tranche seven-year loan of $260 million priced at Sibor plus 75bp and NT$500 million at prime plus 50bp. However, financial avenues are opening as more companies in the industry begin to list on the Taipei exchange.
Rights issues by listed companies, such as the NT$6 billion issue from Winbond (one of Taiwan’s largest electronics companies) recently digested by the market, are becoming increasingly common. As well as ASE, most larger high-tech firms have plans to list subsidiaries, to diversify funding sources and to pay stock bonuses. Acer Incorporated, Taiwan’s leading computer producer, has an ambitious plan to list all of its 21 subsidiaries worldwide.
GDR (Global Depositary Receipt) and ADR (American Depositary Receipt) issues have occurred, but success has tempered by market tension. For example, Morgan Stanley, CS First Boston and SBC Warburg led a July 1995 issue of 8.6 million GDRs (representing 43 million shares) for ASE, which raised $131 million. But investors have not been smiling – the market is down due to tensions with China. “We launched the issue just two weeks before the [last summer’s] Chinese missile tests,” explains Jeffrey Chen, assistant vice president of ASE. “But our company has outperformed estimated revenues and net income; and while now below the issue price the stock is trading at a premium to the market.”
Meanwhile, UMC planned a $500 million GDR issue last year. But this has been postponed to later this year, and many believe that it will become a convertible bond (CB) instead. Indeed, this could be the new trend, with CBs becoming flavour of the year for Taiwanese electronics companies. “As interest rates decline around the world, CBs will become more attractive,” says UBS Securities analyst Dixon Ho.
Taiwanese CBs have been sensibly priced, with reasonable coupons and put options. The great snag has been that they have not been convertible. But this has now changed – nearly. Central bank regulations now permit conversion, with the only hurdle being the actual method of conversion. “We are now working with the SEC [Securities and Exchange Commission] and the central bank on the exact mechanism,” says Yipin Lee, director of corporate finance at Acer. “We expect this process to be finished shortly, which may mean that we will hold a bondholders’ meeting.” Acer issued a US$45 million 10-year Euro convertible in 1991. The bond carried a 4% coupon with a 7% put, but the CB has a conversion price of NT$20 against a current stock price of around NT$60. It is easy to see the potential for capital raising which the transformation of these instruments from debt to quasi-equity would unleash.
The first CB of the year, a proposed $250 million seven-year issue by Winbond, is about to be priced in New York. It may be a little unfortunate as a test-case. Although the company is Taiwan’s third largest IC maker, but is thought by analysts to be one of the most vulnerable to a margin squeeze. And worries over China persist. The bookrunners, SBC Warburg and Salomon Brothers, as well as local underwriter Capital Securities Company, will have to get the pricing absolutely right if the issue is to go smoothly.
