BEFORE THE COMMITTEE ON COMMERCE SUBCOMMITTEE ON TECHNOLOGY, INNOVATION AND COMPETITIVENESS UNITED STATES SENATE WASHINGTON, D.C. HEARING ON MANUFACTURING COMPETITIVENESS Testimony of Thomas R. Howell Dewey Ballantine LLP June 8, 2005
Mr. Chairman and members of the Subcommittee, my name is Thomas R. Howell. I am a partner in the Washington D.C. law office of Dewey Ballantine LLP, where I specialize in international trade matters. Over the past 20 years I have represented a number of organizations representing U.S. semiconductor manufacturers, and in the course of that work I have prepared a series of studies of foreign industrial and R&D policies and their effects on international competition in microelectronics. The most recent of these, which I have provided to the Subcommittee, addresses China’s emerging semiconductor industry. I am also a contributing author to a study recently published by the National Academy of Sciences, Securing the Future: Regional and National Programs to Support the Semiconductor Industry. My testimony today is my own and not presented on behalf of any client or organization. I appreciate the opportunity to appear before you today.
The semiconductor industry plays a vital role in the U.S. economy and national defense. In terms of value-added it may be the largest U.S. manufacturing industry, and semiconductors are a key enabling technology for a broad range of other industries, including computers, consumer electronics, motor vehicles, telecommunications, and aviation. The U.S. semiconductor industry is currently the world leader both in terms of level of technology and market share, with about 50 percent of world sales. However, it faces significant challenges to its leadership which arise out of foreign government policies that are designed to alter the terms of competition. These policies represent promotional strategies that fall into two broad categories, “leadership” and “close followership.”
Leadership strategies. Japan and the European Union, the longstanding rivals of the U.S. in microelectronics, are pursuing promotional strategies designed to capture the leadership position from the United States with respect to market share and level of technology.
· Japan and the EU are implementing large scale, long range, industry-government R&D projects aimed at developing leading edge commercial technologies and state-of-of-the art manufacturing facilities. Commonly these projects involve hundreds of millions of dollars in government funding, more than anything we currently see in the United States.
· The strategy in both Japan and Europe is to build on a perceived leadership position in cell-phone technologies and develop leading edge semiconductors with cell phone applications, as opposed to PC-based chips in which the U.S. holds the lead. The Japanese and European strategy is based on the belief that in the 21st century, most people, particularly in the developing world, will access the Internet through cell phones and similar hand-held devices, not desktop PCs.
It is unclear that these foreign efforts will result in a loss of U.S. market or technological leadership -- in the past many large-scale government-funded R&D projects in microelectronics have fallen short of their goals or failed completely. But others have significantly affected the competitive balance. The EU’s JESSI project, for example (1988-96), is widely credited with contributing substantially to Europe’s current strong position in cell phone technology. Japan’s joint R&D projects have played a major role in establishing the Japanese industry’s strong competitive position in microelectronics. And while Japan and the EU have substantially increased the level of government spending on microelectronics R&D, in pursuit of this strategy, the U.S. is moving in the opposite direction. U.S. government funding of microelectronics R&D has been declining for a number of years and is projected to decline further in the coming decade. But the most complex challenge confronting the U.S. in microelectronics is not coming from Japan or the EU, but from China/Taiwan, who are pursuing a “close followership” strategy.
“Close followership” strategies. Under “close followership” strategies governments do not seek to achieve market or technological leadership but rather to integrate the operations of their own industries with those of U.S. companies, and, by so doing, not only remain one step behind the leaders but also capture high value-added technology-intensive industrial and research functions for their own economies. Taiwan has been the most successful practitioner of this strategy but it is now being emulated in countries such as Malaysia, Singapore, Thailand, Israel, and most significantly, China. The “close followership” strategy actually enhances the competitiveness of individual U.S. companies by providing low cost, high quality production and design services to them. But it may pose a greater challenge to U.S. leadership over the long run because it is drawing offshore important parts of the U.S. microelectronics infrastructure, particularly in the area of semiconductor manufacturing. The danger is that over the longer term other key functions associated with semiconductor production, such as R&D and design, will follow the manufacturing functions to East Asia. At some point a substantial part of the education infrastructure that supports the industry could migrate there as well. At present, roughly 77 percent of U.S.-owned semiconductor manufacturing is still located here in the United States. But much of this capacity is or will become obsolete over the next several years, and the trend is toward establishment of a larger proportion of the next generation of fabs outside the U.S. Earlier this year an executive at Applied Materials, one of the most important producers of semiconductor manufacturing equipment, indicated that 30 new fabs will be built in China in the next 3 years. During the same time frame, the same executive stated that there will be 6 built in the United States. In part this trend reflects the fact that China is the fastest-growing market for semiconductors in the world, with an estimated compound annual growth rate of 20-27 percent in 2002-2008, versus about 7 percent for the U.S. But relative regional market growth does not explain investment trends.
Nor do comparative costs explain current investment trends. The migration of some types of high tech manufacturing to Asia, such as assembly of electronics products incorporating semiconductors, reflects comparative cost advantages attainable by manufacturing in certain Asian countries. But the movement of semiconductor manufacturing to Asia is not being driven by comparative costs -- that is, if government measures taken to modify those costs are removed from the equation. The same equipment and processes are used everywhere to make semiconductors. Materials and other costs do not vary greatly from region to region. Direct and indirect labor costs are much lower in China and Taiwan than in the U.S., but because labor costs are such a small proportion of manufacturing cost, the total cost differentials are not that great. If the manufacturing costs for a 90nm, 300mm wafer fab in the U.S. is given a factor of 100, the comparable cost in Taiwan would be 93 and in China, 90. But the picture changes when the impact of government policy measures is factored in.
To begin with, consider the size of the investment required to establish a single state-of-the-art wafer fab -- currently between $2 and $3 billion for a facility that may be obsolete in 3-4 years. Only a handful of companies are in a position to undertake such investments, and given the volatility of the industry, an increasing number of companies understandably have reached the conclusion that risks associated with such large investments outweigh any potential for gain. How do governments affect this equation? In some countries governments have put up a substantial part of the total investment cost to establish a state-of-the-art fab. The world’s first 300mm fab, for example, was built in Dresden, Germany with substantial funding from regional governments. But other forms of government support are probably more important than direct funding.
One of the most important forms of government measure has been support for the establishment of semiconductor foundries, a phenomenon that occurred first in Taiwan but has spread to Singapore, Malaysia, Israel, and, most importantly, China. Under the foundry model foreign producers, usually with substantial government backing, in effect say “we’ll assume the costs and risks of building a fab. Give us your designs, and we’ll make them for you, in return for a service fee.” This is a very attractive proposition for a company trying to decide whether or not it can make a $3 billion investment to manufacture its designs. An increasing number of U.S. semiconductor firms are “fabless” and outsource all of their designs to foundries, while others are “fab-lite,” outsourcing a significant part of their total production. In other words, the chip is designed here in the U.S., manufactured in China or Taiwan, and in many cases incorporated into an end product somewhere in Asia. The U.S. “fabless” company does not take any of the risks normally associated with building a $2-3 billion facility. But the facilities themselves, and the skills to run them, increasingly reside elsewhere.
The first pure play foundry in the world, TSMC, was established on the basis of an equity investment by a special fund administered by the government of Taiwan. The investment would not have been attempted by the private sector because it was seen as too risky. Today I am not aware of a foundry anywhere in Asia that does not enjoy significant government support. In a number of cases governments have taken equity shares in foundries. Because the number of purely private, unsubsidized companies in the U.S. or anywhere else that are willing to invest $2-3 billion in a fab is declining, government-supported foundries are accounting for an increasing share of global semiconductor production. Most of the new fabs being built in China will operate as foundries. Tax policy is another particularly important form of government support. The world’s most successful foundries are TSMC and UMC, both located in Taiwan. They control nearly two-thirds of world semiconductor foundry manufacturing. The government of Taiwan has implemented policies which ensure that these and other similar Taiwan-based semiconductor enterprises pay no taxes, year after year. In fact, in a number of recent years, TSMC’s after-tax income has been higher than its pre-tax income, reflecting the application of accumulated tax credits. China has now replicated Taiwan’s tax holidays. Paying taxes, in jurisdictions like the United States, and paying no taxes in China and Taiwan, can have an enormous bottom-line impact and may constitute be a very significant decisional factor in determining where to open a new fab. Then there is infrastructure. The Silicon Valley phenomenon has been intensively studied abroad, and foreign governments have created their own versions of the Valley in many countries. These seek to integrate research universities, high tech manufacturing, and venture capitalists into a dynamic relationship that promotes innovation and entrepreneurialism. Perhaps the most successful version has been Taiwan’s Hsinchu Science-Based Industrial Park, which has become a magnet for foreign and domestic semiconductor investment. In addition to tax-free status. soft loans, grants and other forms of financial support, enterprises located in the Park enjoy extensive infrastructural support, nearby research universities, and superb institutes of applied industrial research. China is now creating its own versions of Hsinchu, and in some of the Chinese parks, semiconductor producers are reportedly receiving free land and free structures from regional and municipal governments. They also receive preferential rates on electricity, water, and specialty gases, all of which lower their operating costs.
Then there are government incentives to individuals. One of the key advantages enjoyed by TSMC and UMC has been their ability to attract and hold many of the highest quality managers and engineers in the industry -- it said that “they get the best people.” A key factor in the competition for such talent is Taiwan’s tax treatment of company stock and stock options given as compensation to individuals. Shares are taxed on their par value rather than on their actual market value at the time received, which may be many times par value. In addition, when the shares are sold, there is no tax on the income received (apart from a nominal transaction tax) because Taiwan has no capital gains tax. As a result, Taiwanese companies have been able to offer highly talented Taiwanese and foreign engineers the prospect of rapid accrual of substantial personal wealth. Taiwan has become a “talent magnet.” Chinese tax policy, while not identical, seeks to replicate such incentives to individuals.
Finally the location of new investments can be driven by government investment incentives such as China’s preferential value-added tax (VAT), which was revoked in April of this year after strong objections from the U.S. government, Japan, the EU and Mexico. In 2000, the Chinese government established a preferential rate of value-added taxation (VAT) for domestically based semiconductor design and manufacture. While all imported devices are subject to a 17 percent VAT, under the new policy domestic designers and manufacturers of semiconductors received a rebate, resulting in an effective VAT rate of 3 percent. The preferential VAT policy effectively enabled China to “capture” a portion of Taiwan’s semiconductor capability. Foreign investors, predominantly Taiwanese, rushed to the mainland and established new wafer fabs in order to benefit from the VAT preference. A talent rush to the mainland of experienced Taiwanese managers and engineers occurred. By 2003 roughly 20 new Taiwanese-owned fabs had begun operations on the mainland, were under construction, or were planned to become operational by 2008, all of them foundries. Executives at these new foundries cited the VAT preference, which gave them an “unbeatable” edge over imported devices, as the principal factor underlying their new operations. While China’s preferential VAT has been revoked, it has arguably already achieved its objective of a massive drawing in of capital, technology and talent, enabling China to establish a modern semiconductor industry.
It has been suggested by some that the migration of semiconductor manufacturing to Asia represents a natural division of labor with more advanced countries, and that the high-end functions -- R&D and design -- will remain in the United States, Europe and Japan. But over the long term the design functions are likely to migrate to where the action is, which is where the manufacturing is located. This is happening already in Taiwan, in particular, which is now using its strength in manufacturing to build a strong design industry, with extensive government support. China, too, is following this path, although it is at an earlier stage of development. The long run danger is that so large a proportion of leading edge semiconductor manufacturing and design functions come to reside outside the United States that the top graduates from engineering schools see their future not in the U.S., but in China and Taiwan and other parts of the world. They will seek to build their careers there, not here. At that point, it would be very difficult to reestablish U.S. leadership.
It is not in our national interest to see the entire infrastructure for the design and manufacture of semiconductors to migrate outside of the United States. A recent report by a Defense Science Board task force concluded that the migration of U.S. capabilities in semiconductors outside the U.S. posed “long term national economic concerns.” Given that semiconductors are at the core of virtually all critical defense systems, the national security concerns are obvious. The problem we confront is that the commercial realities of the semiconductor business are leading to a relocation of design and manufacturing functions outside of the United States.
Identifying a comprehensive set of recommendations for addressing this problem effectively would take a sustained industry-government dialogue of the kind we saw in the 1980s in connection with the challenge from Japan. I would like to offer several preliminary suggestions: First, it should be recognized that the present offshore movement of semiconductor production is being driven by deliberate government measures as well as by commercial imperatives. Therefore, the U.S. government should continue to place a priority on the elimination of trade and investment distorting measures like China’s preferential value added tax that violate international rules. China’s use of a WTO-inconsistent measure to attract inward investment that would not have otherwise occurred was a serious market distortion in a strategic industry. The U.S. acted properly in placing a priority on the elimination of this measure. At the same time it should be recognized that many of the incentives used by governments to attract high technology investment do not clearly violate any WTO or other international rules, so there is a limit to what can be achieved by invoking existing rules. Over the longer term it will be necessary to negotiate the establishment of international norms on the use of government incentives for high tech investment. Second, the U.S. government needs to examine domestic tax polices that affect U.S.-based manufacturing in light of foreign tax policies that are functioning like a magnet for manufacturing investment. While I do not recommend any particular tax measure, the fact is that U.S. measures are needed to offset the effects of foreign tax holidays in some way.
Finally, we must recognize that competition in this industry is increasingly a competition for a limited pool of talented people, whether U.S. or foreign born. The U.S. has the lead in this area, and we shouldn’t allow ourselves to lose it. This means above all maintaining our excellent system of research universities and ensuring that the world’s leading edge R&D continues to take place here in the United States. Specifically we should increase, not curtail federal spending on university-based, leading-edge R&D and other forms of support for U.S. research universities.
|