Panel IV
The Taiwanese Approach

INTRODUCTION

Patrick Windham

Windham Consulting

Mr. Windham called the short history of the Taiwan semiconductor industry “a rare success story.” The principal Taiwanese companies in semiconductor production are already world leaders in their specialties: Taiwan Semiconductor Manufacturing Co. (TSMC),27 established in 1987, and UMC, originally founded in 1979.28 He called Taiwan’s journey to becoming the fourth-largest producer in the world a “remarkable” one and voiced the hope that today’s discussion would reveal some of the reasons behind the country’s success.

Clearly, he said, Taiwanese entrepreneurs deserve and receive credit for the lion’s share of that progress. The government as well deserves credit for pursuing a unique set of policies at the right time. Policies pursued by the government not only helped to reform the capital markets in the direction of equities but also contributed substantial R&D support through ITRI (the Industrial Technology Research Institute) and its chief R&D facility, ERSO (the Electronic Research

27  

TSMC changed the semiconductor market by specializing in the manufacture of custom wafers under contract to chip designers. This frees the designers to concentrate on making and marketing the integrated circuits formed on the wafers to form microchips; it helped spark an explosion of fabless microchip companies, such as those that populate Silicon Valley.

28  

United Microelectronics Corp. began as a designer and producer of integrated circuits. In the mid-1990s founder Robert Tsao changed it into a contract manufacturer like TSMC. Recently it has expanded into high-end chips, a segment in which TSMC is not dominant.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 149
Panel IV The Taiwanese Approach INTRODUCTION Patrick Windham Windham Consulting Mr. Windham called the short history of the Taiwan semiconductor industry “a rare success story.” The principal Taiwanese companies in semiconductor production are already world leaders in their specialties: Taiwan Semiconductor Manufacturing Co. (TSMC),27 established in 1987, and UMC, originally founded in 1979.28 He called Taiwan’s journey to becoming the fourth-largest producer in the world a “remarkable” one and voiced the hope that today’s discussion would reveal some of the reasons behind the country’s success. Clearly, he said, Taiwanese entrepreneurs deserve and receive credit for the lion’s share of that progress. The government as well deserves credit for pursuing a unique set of policies at the right time. Policies pursued by the government not only helped to reform the capital markets in the direction of equities but also contributed substantial R&D support through ITRI (the Industrial Technology Research Institute) and its chief R&D facility, ERSO (the Electronic Research 27   TSMC changed the semiconductor market by specializing in the manufacture of custom wafers under contract to chip designers. This frees the designers to concentrate on making and marketing the integrated circuits formed on the wafers to form microchips; it helped spark an explosion of fabless microchip companies, such as those that populate Silicon Valley. 28   United Microelectronics Corp. began as a designer and producer of integrated circuits. In the mid-1990s founder Robert Tsao changed it into a contract manufacturer like TSMC. Recently it has expanded into high-end chips, a segment in which TSMC is not dominant.

OCR for page 149
and Service Organization). The government also helped to set up industrial parks in an effort to build industrial clusters where they did not previously exist. He then introduced the first speaker, Genda Hu, vice-president for advanced technology development at the Taiwan Semiconductor Manufacturing Co. GOVERNMENT-INDUSTRY PARTNERSHIPS IN TAIWAN Genda J. Hu Taiwan Semiconductor Manufacturing Company (TSMC) Dr. Hu, who worked at IBM, Xerox PARC, Cypress Semiconductor, and other U.S. companies before moving back to Taiwan, played a major role in setting up Taiwan’s partnership programs. He said that many people have wondered what the secret was behind Taiwan’s rapid progress in the semiconductor industry. He dismissed the idea of a secret, saying that it was certainly not a miracle and that it had taken Taiwan 25 years of hard work to reach its current position. Nevertheless, he did consider a number of factors to be critical to the successful story in Taiwan. He said he could not cover the financial aspects, which were complex, but would confine his remarks to the story of the technology R&D. An Overview of the Industry He began with an overview for those not familiar with Taiwan’s industry. (See Figure 8.) The total revenue generated by the industry in Taiwan in 1999 reached about U.S. $14.3 billion; for the year 2000, revenues were expected to reach about U.S. $22 billion, a 57 percent increase, compared with a world growth rate projected at about 37 percent. He said that Taiwan divides its industry into four sectors: design, which consists primarily of fabless semiconductor or design houses; fabrication, which includes foundries and also IDMs (integrated device manufacturers), such as Winbond and some DRAM companies with fabs and other products; packaging; and testing. He said the Taiwanese industry is a “vertically dis-integrated” infrastructure where many companies focus on a particular expertise rather than trying to do everything. The definition of revenue is different from the revenue quoted by the WSTS (World Semiconductor Trade Statistics). The WSTS figure for revenue is based on the final integrated circuit as a product. In Taiwan revenue may relate only to a service value, such as packaging, which is considered a value added. Likewise, a foundry produces wafers but not integrated circuits per se. For example, Taiwan has 127 design houses without fabs, 21 companies with fab facilities, 42 packaging companies, and 33 dedicated testing houses. To many people, he said, this is a “really amazing” number of companies. In addition, there are supporting companies, such as those that provide starting substrates, chemicals, leadframe, and substrates for the packaging industry. Most companies

OCR for page 149
are of small to medium size. Only a few are very large, including TSMC, a foundry with 1999 revenues of about U.S. $2.2 billion, forecast to be in excess of U.S. $5 billion in 2000. UMC is the second-largest company, with revenues of U.S. $1.6 billion, and is also growing rapidly. The rest, by international standards, are small. R&D Activities Funding and Execution He divided R&D activities into two categories: funding and execution. Basic funding comes from the National Science Council (NSC) and the Ministry of Education, which sponsors primarily basic and applied research. The institutions or organizations that execute the activities are universities, research institutes, and industries. Among the more prominent labs are the Nano Device Laboratory under the NSC and the Chip Implementation Center, which is supported by NSC but managed by ITRI. Most of the R&D activity in Taiwan in the past has been concentrated around ITRI and most of the funding has come from the Ministry of Economic Affairs. FIGURE 8 Revenue of Taiwan integrated circuit industry. SOURCE: IEK ITIS project/ITRI, October 2000.

OCR for page 149
A small part of the activity takes place at the Hsinchu Science Park. Its activities cover a wide range but the funding is quite low. On the industry side most R&D work is in technology development and commercialization of the technology. Low R&D Funding for Universities He then focused on the university program, in which the number of projects, the budgets, and the workforce have increased steadily since 1996. Most of the researchers are students and professors. The National Science Council and the Ministry of Education fund most of the university research, at quite low levels, despite the large number of projects (236 in 2000). In the year 2000 the total budget is only about U.S. $5 million, which he called a “minuscule amount of research money by any standard,” but it was very important because of the number of people involved and because of the students trained in the universities who eventually support the industry. Funding levels have increased toward the applied research and development end of the spectrum. The NSC supports two main laboratories. The first is the Nano Device Lab, located in Chiao Tung University but totally financed by NSC. This lab’s R&D is mainly associated with silicon-based semiconductor devices and material, with a special focus on deep sub-micron MOS devices. This lab concentrates on process-technology development. An important mission is to support silicon-related research in universities, providing much of the equipment used by the professors and students. The Nano Device Lab’s annual budget runs roughly NT $300 million, with extra amounts sometimes needed for equipment. This amount has compensated, in some sense, for the low direct funding for universities. Help for Students and Professors The second lab, the Chip Implementation Center (CIC), concentrates on design. It functions much like the MOSES program in the United States in that students and professors execute design projects and then take them to the CIC to put them on a “multiproject wafer.” This process, pioneered at Xerox PARC many years ago, makes it possible to assemble 20 to 30 different designs on a single mask. For 0.13- and 0.15-micron technology, it costs close to half a million dollars for tape-out, which professors cannot afford. Instead, they use resources at the CIC, where the average budget is about NT $140 million. That program, managed by ITRI, has been very successful. The majority of the development work sponsored by the government has been financed through MOEA, the Ministry of Economic Affairs. The level of funding was very high before 1994. Since then it has been trending down, prima

OCR for page 149
rily because industry has shown sufficient vitality to take over more of the funding and perform the R&D itself. He noted that the amount of MOEA spending has been far larger than spending by the Ministry of Education, National Science Council, or other agencies. The MOEA has allocated between hundreds of millions to over a billion new Taiwan dollars each year for R&D. The MOEA is the funding agency and the work is done through ITRI by ERSO, which is the primary laboratory doing semiconductor research under ITRI. A History of Taiwan’s Rapid Rise Dr. Hu illustrated how this funding has been used to help the industry during its brief history. He said it has not been a matter of taking the government’s money to do the R&D and then hoping for technology transfer to industry. It began in 1974, when the Taiwanese government decided to focus on the semiconductor industry as a key industry. At that point Taiwan’s economy was primarily based on agriculture—“nothing but paddy fields, sugar cane, and pineapples.” The Ministry of Economic Affairs chose semiconductors as the industry to work on. In the early 1970s the government established ITRI and, under it, about 10 different laboratories, including ERSO, which focused on semiconductors. ITRI was not part of the government and its employees were not government employees. The organization was more like ASET (described earlier in the proceedings) in that it was an independent entity doing contract work for the government. Initially there was some grant money from the government, but once ITRI was established the government money stopped, and ITRI survived by obtaining contracts. The strategy was to make it more efficient. Creating a Company With ITRI as the interface, ERSO contacted RCA, and the government paid RCA several million dollars for its 7-micron metal-gate CMOS process. RCA transferred that technology to Taiwan and helped ERSO build Taiwan’s first 3-inch-wafer fab, in 1975, which started semiconductor activity in Taiwan. After ERSO had worked for six years on developing its own technology, the government decided to create an industry to use the technology. It needed a commercial company, so it created UMC (United Microelectronics Corporation). At that point, no one in the private sector wanted to invest. Therefore, the government supplied the initial funds and did some arm-twisting to get the banks to put up some money as well. This was before there was any venture capital in Taiwan. The government did not try to run the company and later sold all its shares, so the company became purely private. Once the company was formed, ERSO transferred technology to UMC, along with skilled people, and helped it build its 4-inch-wafer fab. UMC, established in 1980, was Taiwan’s first commercial semiconductor company.

OCR for page 149
Creating an Industry The government came to realize that it did not yet have an industry—only one company. The private sector was still fearful of risk and could not raise enough money to start new firms. Once again the government put up money, this time to start TSMC. And this time it applied a condition: The government’s share would be less than 50 percent. Lacking enough funds from the private sector, the government made an offer to Philips, which put in almost 35 percent of the initial investment; this pushed the private sector’s share above 50 percent, and TSMC was born. Again, the technology and all the people came from ERSO, including about 130 engineers, the 2-micron CMOS developed at ERSO, and its latest 6-inch-wafer fab. TSMC had no product, only the fab technology, which matched well with the idea of a pure-play foundry. That same year another group felt that it could do the same thing without government money, so nine months later a third company, Winbond, was formed. Again, most of the people came from ERSO, along with licensed technology and some old integrated-circuit (IC) products, such as wristwatch chips and other low-end consumer items. At first no one wanted to put money in, but once TSMC was formed some companies thought it probably was doable and put enough money in to create the third company. After that a new company started almost every year, and many were spinoffs from ERSO. ERSO continued to be the source of human resources for the industry. The Power of Spinoffs He skipped ahead to 1994, when a similar company formation gave rise to Taiwan’s DRAM industry. The story began in 1990, when the government awarded a major project contract to ERSO to make the half-micron CMOS, including the 8-inch wafer and two kinds of products. One was a 4M SRAM and the other a 16M DRAM. With that technology and product a new company called Vanguard was formed. This time the government did not have to invest additional money; rather, it owned part of Vanguard when Vanguard was spun off. The government had contributed about $100 million to the DRAM project, which was converted to stock in Vanguard, and the stock was worth about $400 million two years later. “No one had ever heard of a government making money on R&D,” said Dr. Hu, “but this one did.” Because of that success the private sector was able to create four or five companies to make DRAMs. He said the important message here was how much could be done by government to benefit industry from this kind of R&D work, primarily through spinoffs.

OCR for page 149
Government and Industry Funding The Strength of Small Grants Dr. Hu next introduced the other, smaller R&D projects sponsored by the government through the Industrial Development Board (IDB), an agency under MOEA. The IDB made R&D grants to companies with very specific projects that were close to commercialization. Companies proposed projects, but the funding amounts were not large. The Hsinchu Science-based Industrial Park (HSIP) administration also allocated R&D funding each year to support the dozen or so companies inside the park. The amounts of funding were small but they helped to create many new products. The industry itself has been growing since its inception, starting with UMC in 1980. In the past 20 years the industry has grown quickly, and in 2000 was estimated to reach about $22 billion in total revenues. The industry has been spending about 6 percent of revenues on R&D. Industry Assumes the Primary Funding Role Beginning in about 1990 the government was underwriting over 44 percent of total spending on R&D to the benefit of the private sector. Since then the government contribution has remained relatively constant, but its percentage has dropped rapidly. By 1999 the government’s share had fallen to only 6.5 percent. This was a clear indication that industry had moved in to take over the primary role in funding semiconductor R&D in Taiwan’s industry. Dr. Hu summarized by saying that the government has played a proactive and pivotal role in establishing the IC industry over the past 20 years. The turning point came around 1994, when Vanguard was spun off and industry started to play a leading role in R&D. The government received criticism for making so TABLE 1 Government Versus Industry R&D Investment Year $ from Gov’t (MNTD) $ from Industry (MNTD) Gov’t/Ind 1994 2,160.0 4,832.0 44.7% 1995 1,413.0 8,936.9 15.8% 1996 1,296.0 9,669.4 13.4% 1997 1,439.0 16,158.8 8.9% 1998 2,079.0 23,537.6 8.8% 1999 1,750.0 26,834.8 6.5%   SOURCES: NCS, HSIP, IEK/ITRI.

OCR for page 149
much money on its investment, and there have been no more spinoffs from ITRI since 1994. For its part industry does not want more spinoffs, which now would create more competition in the market place. As a result the government’s role in R&D has gradually evolved into a more traditional activity and is expected to continue to decrease in the future. DISCUSSION A Consortium in Taiwan? Glen Fong of Thunderbird, the American Graduate School of Management, asked whether Taiwan’s spectacular startup story would evolve to include more collaborative R&D, like in the United States, Europe, and Japan. He asked about the “so-called Taiwan SEMATECH” called ASTRO. Dr. Hu said there has indeed been an attempt to establish collaborative research in Taiwan, because the benefits of collaborative activity are well understood. As in other countries government funding levels are inadequate to support an effective R&D effort, simply because the cost of manufacturing equipment has increased so fast. Taiwan has studied the models of SEMATECH, Selete, ASET, IMEC, and other consortia, and it plans to combine industry money with government money to do R&D for the common good. For 2 years ERSO has been chartered to create that kind of consortium, which was moderately successful until the beginning of 2000. With a new political party in power, industry policy has become less clear, and all further action is on hold until plans are firm and new funds are committed for the consortium. Another questioner asked whether the Taiwan semiconductor companies are doing their own research. Dr. Hu said they are, and more noticeably at the international level. TSMC is a member of International SEMATECH, and UMC has formed an alliance with IBM and Infineon. The plans of smaller companies are not as well known. THE SCIENCE PARK APPROACH IN TAIWAN Chien-Yuan Lin National Taiwan University Providing Good Soil Dr. Lin of the Institute of Building and Planning at National Taiwan University said that he would introduce another approach to R&D, the science park. He pointed out that even though Taiwan is small, the government has promoted economic development very heartily, allowing the country to rise from “almost nothing after the Second World War” to a point of considerable accomplishment. He

OCR for page 149
said that one could think about promoting R&D or technology in the same way we might try to design a kind of tree that could produce good fruit. The science-park approach is designed to provide good soil that will grow healthy trees so that we can enjoy the fruit. He traced the beginning of the IC industry in Taiwan to the packaging industry that started in 1976. By the end of 1999 Taiwan had 237 IC firms, including IC design and manufacturing. The industry is still growing at a rapid annual rate of about 50 percent and has moved into third or fourth position globally in market share, depending on the segment. Most of the jobs created have been in the fabrication and packaging segments. Both the IC industry as a whole and the science parks have created many job opportunities as well. The Government’s Active Role He compared the government’s role in Taiwan and the United States, saying that the U.S. government maintained a primarily free market in which the industry had to grow and compete and survive. He contrasted this with Asian countries, notably Japan and Taiwan, where the government has been very active in promoting economic development. One of the ways the Taiwanese government has done that, beginning in the late 1970s, has been to identify the most promising industries and then develop an attractive environment in which high-tech companies in those industries could become established and grow. This succeeded even though the country had virtually no technology to build on—only a labor force. The Concept of the Science Park In 1980 this government policy was augmented by the concept of the science park. The government provided major venture capital as well as some tax deductions or exemptions for companies that moved to the park. In addition, it provided the infrastructure, one-stop business service as well as other services such as R&D and education. This complete package was considered to be a partnership between the government and the semiconductor industry. Incubating Factories Dr. Lin showed a map of Taiwan and the location of the Hsinchu Science Park plays. (See Figure 9.) He pointed out that most of the country’s IC factories are distributed in the northern part of Taiwan—either within the science park itself or nearby. He said that the Hsinchu Science Park played a key role in incubating these factories. The park was planned in the 1970s and began operation in 1980 under the administration of the National Science Council. That the park is managed by the central government directly rather than by local government is a measure of its

OCR for page 149
FIGURE 9 Spatial distribution of integrated-circuit factories. importance. Park land is leased to tenants—not sold—thereby allowing the government to maintain control over the use of the park. The park is located close to the Chang Kai Shek (CKS) International Airport and near Taipei City, so it enjoys locational benefits of transportation and human resources. Another important feature is that it is located close to two universities, the Tsing-Hua University and the Chia-Tung University. These are the leading universities in IC industries. It is also near ITRI, the Industrial Technology Research Institute, which has the important function of helping to incubate the industries in Hsinchu Science Park. Like a New Town In a sense, he said, the Hsinchu Science Park is like a new town. More than just a place for manufacturing, it also includes services such as restaurants, clin-

OCR for page 149
ics, banks, housing, and bilingual schools. He said that Taiwan does not have sufficient human resources, especially high-tech people, who are the key to a successful R&D operation. The park has had to recruit talented people from the United States. When it does manage to bring “those high-tech families” to Hsinchu, it has to provide the education programs for their children. In addition to those services the park also has support businesses, like the one-stop services, on-the-job training, and a range of domestic and international information services. Because this park is manufacturing-oriented it requires automated customs services for shipping. Rising Employment and Capital Employment growth at the park has been strong, rising from 19,000 workers in 1989 to 33,000 in 1994 to 83,000 in 1999. In the beginning most of the companies came from the United States, but by the end of 1999 more than 80 percent of them were local or other domestic companies. Total paid-in capital in the park has surged from U.S. $3.5 billion in 1994 to U.S. $20.4 billion in 1999. In the beginning most of this capital came from the government, the United States, and other countries; now 92 percent of it is domestic money, and only 4 percent is from the government. The park has a total of 291 units, including PC manufacturing, IC design, and other IC-related industries. The most important trading partners are the United States, Japan, and Europe. Dr. Lin noted that many in the labor force do not have a baccalaureate degree, so there are many opportunities for high-school graduates. This is because the park is very manufacturing-oriented. The production growth rate has slowed somewhat from the early years but was still 43.1 percent in 1999. Because of the success of Hsinchu Science Park, both TSMC and UMC have requested additional space for expansion. The government has planned new sites in Chunan, 30 km south, to focus on biotechnology, telecommunications, and opto-electronics, and other sites in Tonglou, a short distance farther south, which would focus on telecommunications and opto-electronics. In the southern part of Taiwan the new Tainan Science Park is developing 638 hectares for four new TSMC factories and other companies. Tainan Science Park had 15 IC factories committed to operations in 2000. The Strains of Rapid Growth On the downside Dr. Lin acknowledged worsening environmental problems caused by the sudden growth of the science park. Hsinchu Science Park has a “terrible” traffic situation, he said, and wastewater has become a critical and irritating issue for the local government. This is a politically difficult situation because the local government does not experience benefits from the park, which is

OCR for page 149
managed by the central government. These issues will become more acute as other areas, seeing the success of HSIP, ask for science parks of their own. In addition, competition is becoming a problem as other regions, including Mainland China, build their own parks. Dr. Lin concluded by saying that competition in general will have to be considered more closely in the future, now that the idea of science parks has become popular. DISCUSSION Dr. Flamm asked if Dr. Lin could expand on the nature of the tax deductions and exemptions for semiconductor producers. Dr. Lin said there are two kinds of exemptions. First, for the first five years of operation IC companies can deduct the cost of all equipment or investment, which is “a nice deal for them.” Second, an investment incentive allows the creation of a tax shelter for any money invested in the IC industry. Dr. Hu added that many Southeast Asian countries, unlike the United States, tend to have these tax incentives. Discussant Michael Luger University of North Carolina at Chapel Hill Dr. Luger began with a comment about tax incentives. The U.S. federal government, he said, provides general tax incentives, such as R&D tax credits, while state governments and some local governments have location-specific credits. He then congratulated the STEP Board for putting together such a rich program and, like Dr. Mowery, noted that he would limit his comments to what others had talked about during the program. He noted that most of the presentations had focused on the importance of national semiconductor consortia to national and global competitiveness. Three Conditions Favoring Consortia He noted three conditions favoring consortia: First, new science is increasingly expensive to develop and requires multiple partners to pay the bills; second, R&D has spillovers that invite free-ridership that consortia can internalize; and third, the absence of cooperation and coordination can lead to duplication of effort and what Schumpeter called “destructive competition.” He said he would expand on two points made in many of the presentations. One, national consortia like SEMATECH and SIRIJ are one of several models along a continuum that are intended to enhance competition. And second, all of these models have important geographic dimensions that make them variably important as economic development strategies, not just as science and technology strategies. Here, he noted that as a regional economist, he was especially interested in this aspect of the topic.

OCR for page 149
A Taxonomy of Collaborative Vehicles First, there are the national industrial consortia in advanced economies— SEMATECH, SIRIJ in Japan, perhaps MEDEA in Europe—whose goals are advances in fundamental science that can be used by industry. Second, there are national-government-funded labs and demonstration centers, such as the U.S. national laboratories. In transitional economies, such as Thailand, these would be science and technology centers that are used to develop new industries with strategic national importance. The Nano Devices Institute would fall into this category as a focused, leading technology center. An example at the state level in the United States would be the North Carolina Biotechnology Center. Third, there are laboratories with direct government support for industrial research. In the United States these are funded by the Department of Defense, National Science Foundation, and other agencies, including through the SBIR program. These are supported in the belief that public funds, if strategically spent, can unleash the inventiveness of the private sector. Fourth, there are what the National Science Foundation calls “virtual colaboratories,” vehicles to foster collaboration among experts around the world who share research interests. These vehicles may use mail, teleconferences, and other information-sharing mechanisms like membership and trade organizations. Fifth, there are university-based R&D centers such as those in Japan discussed by Dr. Morino. Many of the engineering schools in the United States have targeted centers for research in semiconductors, materials science, or information technology hardware. In general these centers have three common characteristics. They serve as recipients of federal, state, and foundation funding. They reach out to industry for active and passive partnerships, joint ventures, and funding. They create a focus within a university to attract faculty, students, and internal resources. With dozens or hundreds of universities doing the same thing, they tend to compete rather than cooperate for resources, for personnel, and for industrial interest. Recognizing this, states have responded by developing multi-university or regional centers as a way to reduce competition and to create a critical mass. That is an activity of the EPSCoR29 programs in some states. Local industries are important focal points for all of these centers. 29   EPSCoR, the Experimental Program to Stimulate Competitive Research, is a joint program of the National Science Foundation and several U.S. states and territories. The program promotes the development of the states’ science and technology resources through partnerships involving academia, industry, and state and federal government.

OCR for page 149
A Continuum of Consortia Dr. Luger placed these programs on a continuum. On one end are the large, national consortia that emphasize basic research; they also have a high degree of spillover and consequently fewer direct local applications. Next along the continuum are programs supported directly by federal funds, which also emphasize basic research, including university R&D centers. Beyond them are state-funded R&D centers, which feature more applied research, fewer spillovers, and more concentrated spatial effects. The Importance of Clusters Dr. Luger then elaborated on the importance of spillovers in the context of government-industry partnerships. The kind of research that economists call appropriable science has fewer spillovers and leads to more localized geographic effects. Businesses want to locate near this research or to do it themselves so that they gain the rights of first refusal of intellectual property. The more appropriable the science, the more important physical proximity becomes. As a new technology develops commercially in a particular place, spatial agglomeration occurs, bringing what are called localization economies. As in the case of Taiwan, these economies lead to clusters of firms—not only from single industries but also from industries that are related through input-output linkages and other relationships. These clusters stimulate growth in the local economy. This has been the logic behind the development of the Hsinchu Science Park and other science parks around the world. The Case of Research Triangle Park Dr. Luger mentioned Research Triangle Park in his own state of North Carolina, which he has studied extensively. In addition, he discussed his trip to Taiwan several years ago to visit Dr. Lin and HSIP. In North Carolina in the 1970s and early 1980s there was virtually no semiconductor research or industrial presence outside the engineering school of North Carolina State University. Research Triangle Park was built in 1959, with IBM the first corporate client, and in the 1980s the state invested $6 million dollars—a good deal of money at the time—in the Microelectronics Center (MCNC), expecting to attract semiconductor firms. Like in the case of Vanguard in Taiwan, the investment paid off in several ways. Recently a company called Kronos spun out of MCNC. Though the deal was private, the investors gave the state of North Carolina $30 million of the proceeds to build infrastructure in rural areas. MCNC has also served as an anchor for the recruitment of major IT companies to the park and region, including Nortel, Ericsson, and Cisco Systems.

OCR for page 149
The Importance of Being Adaptable The focus on information technology was not initially envisioned in Research Triangle Park (RTP). After IBM located there and the Microelectronics Center was built, the state expected to develop a large presence of semiconductor companies. During the 1980s RTP attracted Harris Semiconductors, Mitsubishi, and a few smaller microelectronics companies. The industry then began to restructure and RTP failed to attract additional investment in the semiconductor area. It changed its recruitment strategy. It renamed the Microelectronics Center simply MCNC and added programming for information technology firms. As noted above, the state was successful in attracting major information technology companies. “Adaptability,” said Dr. Luger, “was important.” By most standards RTP is judged to be highly successful. He showed a cluster diagram that indicated a 20 percent greater presence of information technology, communications, and software in the region than one would expect from the national averages. The location quotient has been growing at the same time these sectors have been growing nationally. On the applied side, semiconductor research is becoming the foundation of other industries, a point Dr. Knorr made earlier. So the earlier development of infrastructure to attract the semiconductor industry became useful later for attracting the two sectors that have come. He concluded with the speculation that the work of SEMATECH in helping to enhance the semiconductor industry also had impacts in related industries for many areas of the country. DISCUSSION Questions of Human Resources A questioner returned to Dr. Hu to ask about the supply of human resources in S&E, and he replied that there is a shortage of engineering talent. From TSMC’s point of view, he said, it will be very difficult to sustain the company’s growth rate as projected by the government unless the supply increases. The company is supporting university programs and encouraging professors to generate more graduates, but he affirmed that, “on a large scale, there is a problem.” Dr. Hu was also asked about the H-1B visa: Could young Chinese engineers still come to the United States, go to graduate school, gain work experience, and then return to Taiwan? Dr. Hu did not address this directly but said that most of the people TSMC attracts back from the United States are very seasoned people, many of whom are U.S. citizens. He said that the people who had visa problems working in Taiwan were the Mainland Chinese. Dr. Luger added that some parts of the United States are trying to duplicate the repatriation efforts of Taiwan. He said that Pennsylvania and Georgia both offer college loan payment exemptions to young, college-educated people to return.