Harnessing Science and Technology for America's Economic Future: National and Regional Priorities (1999)

David C. Mowery

Professor, Haas School of Business University of California, Berkeley³⁴

No assessment of future policy priorities and instruments in U.S. science and technology can ignore the international environment. The remarkable success of economic reconstruction and growth in the 50 years following the end of World War II means that the U.S. research and development (R&D) system, which is still by far the largest (measured in terms of annual investment) in the world, accounts for a smaller share of global R&D activity than was true in the 1960s and 1970s. As such, U.S. firms and citizens can benefit from expanding their monitoring and exploitation of R&D performed offshore. In addition, like other elements of modern capitalist economies, the R&D systems of the industrial economies (and, increasingly, those of the industrializing economies) are closely intertwined with one another.

Higher levels of global technological and economic interdependence, however, do not mean that the nation-state is dead. National government policies still are important sources of support for domestic R&D infrastructures that may or may not be conducive to economic growth and competitiveness. But closer links among these economies, which differ from one another in important structural

elements, have introduced a closer interdependence between trade policy and other areas of policy, such as regulation, technology policy, and competition policy. Although the innovative activities of private firms now appear to be more internationalized than at any time in the past 75 years, the "home base" of many large, multinational corporations continues to count for a disproportionate share of their inventive activities. At the same time, however, the operation of national government policies, especially those seeking to support the development of "strategic technologies," may be enhanced or frustrated by the actions of the private firms that operate global networks for innovation and the commercialization of new technologies.

The roles of national and regional governments in science and technology policy also have changed, creating another source of interdependence and conflict. With the end of the Cold War and intensified global competition, governments face greater demands to deliver tangible economic returns from their investments of public funds in R&D projects and infrastructure. Such political demands can assume a nationalistic tone and have been associated with efforts by governments in Western Europe, Japan, and the United States to restrict access to publicly funded domestic technology development projects. Resolving these conflicts, which are often heightened by other cross-national contrasts in the structure of domestic R&D systems, requires a review, and perhaps a revision, of the conceptual framework that underpins many of these publicly funded projects. In the contemporary global economy, efforts to restrict the international movement or exploitation of the results of such projects are likely to prove futile, if not counterproductive.

The U.S. policy posture toward these changing circumstances needs to proceed from the premise that the rapid and efficient adoption by U.S. firms of new technologies from foreign or domestic sources, rather than their creation, is the primary source of economic benefit (see OECD [1996b], for a recent analysis of this issue). Such a shift will confront a tension between the interests of U.S. firms seeking a nationalistic technology policy and those of U.S. citizens (and in many cases, other U.S. firms) who benefit from expanded access by foreign firms to the U.S. R&D infrastructure and from more rapid inward transfer and application of technologies developed offshore.

The issues raised by global interdependence for science and technology policy are too numerous and complex to be addressed comprehensively in this paper. Instead, I survey developments in three broad areas: (1) the "globalization" of the U.S. R&D system, (2) trends in public R&D investments among major industrial and industrializing economies, and (3) some implications and challenges created by such interdependence for the formulation and implementation of U.S. science and technology policies.

The terms of the domestic debate over the desirability of international R&D have undergone at least three broad shifts in the past 25 years. U.S. firms' increased investments in offshore R&D during the 1960s sparked expressions of concern over the loss of employment and other technological opportunities associated with the domestic performance of R&D. This discussion was part of a broader debate over the benefits and costs to U.S. citizens of the expanding international activities of U.S.-headquartered multinationals, in which some participants argued that the private interests of U.S. multinationals no longer coincided with those of U.S. citizens. Beginning in the early 1980s, as foreign investment into the United States grew rapidly and this economy became an important host nation for foreign-owned enterprises, critics argued that such investments created employment opportunities only in low-wage, low-skill assembly operations, and did not bring with them the "high-value-added" activities of R&D and innovation. Most recently, R&D investment within the United States has been criticized as a means of "cherry-picking" the fruits of U.S. R&D, especially publicly financed basic research in U.S. universities, and that such foreign investments are a conduit for the export of technological advances and economic opportunities from the United States to foreign economies (see OTA [1994]). The most recent debate cites differences among the "national innovation systems" of the industrial and newly industrializing economies, suggesting that asymmetries in U.S. and foreign firms' access to the technologies developed in one another's home economies creates disadvantages for U.S. firms.

To shed more light on the significance and implications of increased internationalization of R&D, some disaggregation is necessary. The pattern and trends in international R&D investment seem to differ significantly among industries and among different activities within the innovation process. Extending the work of Archibugi and Michie (1995), one can distinguish among the creation of new technologies (often identified with invention), the development of these inventions into commercially attractive products, and the production and marketing of these new products. None of these activities is well measured within industrial economies, and our measures of their international dimensions are even less reliable. R&D investment, for example, includes both the creation and the development of new technologies, and in many cases is associated as well with the exploitation of these technologies (as in the case of "localization" of new products for specific offshore markets). The available evidence on trends in each of these three activities suggests that the most significant increases in "internationalization" have taken place in the exploitation of new technologies, largely as a byproduct of increased cross-border investment in production activities. Other evidence indicates that much of the technology creation activities of large firms remains concentrated in their home economies.

Existing measures of internationalization of innovative activity are flawed for a number of reasons. They do not distinguish among the different stages of the innovation process, as was noted above. Public data on international flows of R&D investment do not cover manufacturing industries very well; their longitudinal coverage is imperfect and their coverage of R&D investment outside of manufacturing is limited. Much of the relevant activity in industrial innovation, especially in smaller firms, is not captured by conventional measures of R&D investment. Finally, some important mechanisms for internationalization of innovative activity (e.g., strategic alliances) are not captured in R&D investment data. Other indicators, discussed below, share many of these defects.

Table M-1 reproduces data from the 1996 edition of Science and Engineering Indicators (National Science Board, 1996) on trends during 1980-1993 in U.S. outward R&D investment, measured as a share of industry-financed R&D spending overall and in each of 11 industrial sectors. One of the most interesting points to emerge from this table is the minimal growth in the share of total industry-financed R&D in the United States that is invested in offshore R&D. Rather than a steady increase, this share declined during 1981-1985, increased from 1985 through 1992, and shrank from 1992 to 1993. Across the entire time period, the share of industry-financed R&D devoted to foreign R&D decreased by 2 percent. The table also underscores the intersectoral differences in these trends—electrical equipment, petroleum extraction and refining, pharmaceuticals, and non-electrical machinery display declines during this period in foreign R&D investment, while scientific instruments increases slightly. The chemicals industry and nonmanufacturing industry (for which the time-series coverage is especially imperfect) both display significant increases in the share of their R&D spending devoted to foreign sites. Table M-2, which reproduces other data from Science and Engineering Indicators, shows that Western Europe was the primary destination for outward flows of U.S. R&D investment during 1983-1993, although Japan has increased its share of U.S. firms' offshore R&D.

Although outward R&D investment by U.S. firms has scarcely grown relative to overall industry-financed R&D, R&D investment by foreign firms in the U.S. economy has grown since the early 1980s. Table M-3, drawn from Science and Engineering Indicators, compares the shares of industry-performed R&D in the United States and other industrial economies financed by foreign sources. Within the United States, this share more than doubled during 1980-1993 (from slightly more than 3 percent of industry-performed R&D in 1980 to 9.8 percent in 1993); but as of 1993, industrial R&D in the United States was less dependent on foreign sources of funding than that in the United Kingdom or Canada, while exceeding levels of foreign-financed R&D in Japan and Germany.

R&D investment measures inputs into the innovation process, rather than

outputs. The only reliable data on technology creation are patenting statistics, which have important drawbacks (e.g., the widely remarked differences among industries in their propensity to patent), but nevertheless capture an important input into the innovation process that is "downstream" from R&D investment. In

addition, patents contain information on the site of the invention that is assigned to a corporate entity, revealing the geographic location of the inventive activities of U.S. and other multinational corporations.

The patent data compiled by Patel (1995) suggest that the technology creation activities of large firms, measured by the site of the inventions underlying the U.S. patent applications of U.S. and foreign corporations, are less internationalized than their manufacturing operations, sales, or (in many cases) their R&D investment. Patel analyzed the patents obtained by 569 of the world's largest firms during 1985-1990 (Patel's 1995 work extends work by Patel and Pavitt [1991]). Table M-4, from Patel's study, shows that the U.S. patenting activity of large firms from the United States, Japan, France, Italy, and Germany is dominated by domestic inventive activity, based on the reported site of the patented invention—more than 85 percent of these firms' U.S. patents are based on ''home-country'' inventive activity. For U.S. and Japanese firms, these shares exceed 90 percent. Large firms from Great Britain, Canada, Sweden, the Netherlands, Switzerland, and Belgium are less domestically focused in their inventive activity but, with the exception of Dutch and Belgian firms, all of these firms report that more than 50 percent of their patents are based on inventions from their home

countries. Equally interesting is the geographic distribution of inventions made outside of their home countries by these firms. Their foreign inventive activities are sited primarily in Europe and the United States.

If this measure suggests that technology creation remains "localized," other evidence suggests that the activities further downstream in the innovation process, such as development and exploitation of technology, are more international in scope. Foreign firms seek U.S. patents in order to exploit their technologies in this market. The size and importance of the U.S. economy are such that most foreign firms are likely to apply for U.S. patents on only their most important inventions—in other words, foreign patents are likely to be somewhat higher in quality than patents assigned to U.S. firms and domestic inventors. Figure M-1 plots trends in the share of all U.S. patents granted during 1973-1993 that were obtained by foreign inventors. Foreign inventors' share of U.S. patents grew from less than 38 percent in 1978 to roughly 45 percent by 1993. Interestingly, increases in the share of non-U.S. inventors receiving patents did not result from growth in the share of patents granted to inventors from Japan, Germany, or other members of the G-7. Instead, this growth appears to reflect increased U.S. patenting by individuals and firms from a more diverse array of foreign nations. Although the share of U.S. patents received by foreign inventors has declined from its 1988 peak, it remains well above the 1973 level.

Figure M-1

Share of US. patents granted to foreign inventors (1973–1993)

Source: U.S. Patent and Trademark Office, Fiscal 1993 Annual Report.

Another indicator of the international reach of technology development and exploitation activities is the formation of international strategic alliances, which typically focus on the development, manufacture, and marketing of new products, rather than invention or basic research. The number of such alliances has grown since 1980, and they now appear in industries (e.g., commercial aircraft) that historically have not been major sources of direct foreign investment. During the 1980-1989 period, nearly 600 such alliances were formed between U.S. and Japanese firms, and more than 900 between U.S. and European firms (National Science Board, 1993). Thus far, very few of these alliances link U.S. firms with those from newly industrializing economies such as Taiwan or South Korea (Mowery et al., 1996), although such links are likely to increase in the future.

These alliances focus on the commercialization and exploitation of new technologies rather than the basic research underpinning their creation. Growth in alliance activity is attributable in part to the increased importance of foreign sources of technology for U.S. high-technology firms, but this trend also reflects the economic importance of foreign markets for these U.S. firms. Foreign markets for such high-technology industries as commercial aircraft are projected to grow more rapidly than the U.S. domestic market during the next 20 years. Faster growth in these large foreign markets combined with the need to recover the escalating costs of new product development have increased the economic importance to U.S. high-technology firms of penetration of foreign markets. The growth in alliances in at least some high-technology industries also reflects the response of U.S. and foreign firms to nontariff barriers to trade and investment, as well as government policies that seek to restrict access to domestic strategic technology programs, such as SEMATECH in the United States and JESSI in Western Europe (Mowery, 1997). These and other government efforts to restrict the international mobility of technology-based "created assets" paradoxically contribute to the formation of interfirm alliances that support such mobility.

Still another measure of the global scope of technology exploitation concerns the licensing of technologies. Here too, the balance between outflows from and inflows into the U.S. economy appears to have shifted somewhat in the 1990s. U.S. receipts of royalty and licensing income grew at an average annual rate of 20 percent during 1986-1990, but slowed to 7 percent per year during 1990-1993, and grew by only 2 percent during 1992-1993 (Figure M-2). The data in Figure M-2 portray trends in U.S. imports and exports of technology, as measured by licensing and royalty income and payments. These data reflect receipts and payments covered by all extant agreements; a far more revealing measure, for which the U.S. government does not collect data, concerns the "balance of trade" on only the contracts and agreements signed during the previous year. Japanese data on new contracts and agreements reveal a dramatic improvement in that nation's importance as an exporter of industrial technology (Mowery and Teece, 1993). The 1992-1993 slowdown in nominal growth may reflect the cyclical downturns in Western Europe and Japan (receipts from Western Europe declined

Figure M-2

U.S. royalty and licensing receipts and payments (1986–1993)

Source: Survey of Current Business, September 1994.

during 1992-1993). But these measured trends are dominated by intrafirm flows of technology—in 1992, 80 percent of U.S. receipts were accounted for by technology exports to foreign affiliates, a modest increase from 75 percent in 1986, and the average annual growth rate of intrafirm receipts during 1986-1992 exceeded that for exports to nonaffiliates.

U.S. payments of royalties and licensing fees grew even more rapidly than U.S. receipts during 1986-1993, registering an average annual growth rate of 23 percent during 1986-1990, and 16 percent per year during 1990-1993, although these payments declined by 3 percent during 1992-1993. These payments also are dominated by intrafirm transactions, although the share of affiliates is lower. In both 1986 and 1992, intrafirm technology flows accounted for approximately 65 percent of total U.S. technology imports. The importance of intrafirm transactions within both technology exports and imports makes it difficult to infer much about the technological competitiveness of the U.S. economy from these trends. The apparent increase in the role of foreign affiliates of U.S. multinationals as sources of licensing fees and royalties, for example, could reflect a tendency for these offshore affiliates to receive more advanced technologies from their U.S. parents, as offshore sites have become more attractive locations for advanced production operations than the United States. These trends might also result from stronger international intellectual property rights, which facilitate the arms-length transfer of technology between parent and affiliate.

In summary, these indicators reveal increases in foreign-financed R&D within the U.S. economy to levels that approach those observed for some time in several European nations. By contrast, the outward flow of U.S. R&D investment has remained nearly constant (as a share of total U.S. industry-financed R&D spending) since the early 1980s, although this flat overall trend conceals significant interindustry variation. Still other evidence suggests that the earliest stages of the innovation process are the least internationalized, by comparison with the exploitation or commercialization of technologies. The data on foreign firms' patenting activity in the United States, licensing and royalty income, and international strategic alliances also suggest that the channels through which R&D "internationalization" takes place are expanding in number and changing in structure. Finally, as was noted above, it is important to keep in mind the imperfections in these measures. They focus on inputs to the process of industrial innovation, their coverage of nonmanufacturing industry is poor, and they do not cover important phases of the overall process of industrial innovation.

The most straightforward explanation for the trends in international innovation is based on work by Cantwell (1991, 1995), who emphasizes the use by multinational firms of international R&D strategies to create interfirm and intrafirm networks for the creation and strengthening of firm-specific knowledge and technological capabilities. This view contrasts with the previous view (articulated in Vernon's [1966] celebrated product-cycle model) of multinational firms' R&D strategies as motivated primarily by efforts to exploit products developed to serve the market of their high-income home economies in foreign markets. Both motives influence international R&D strategies, but there is some basis to suspect that the first will become more important in the future.

Cantwell and others argue that the acquisition or maintenance of firm-specific technological capabilities relies on extensive contacts with external sources of expertise in both the home and foreign economies. These contacts require either a physical presence or some other complex organizational form, because of the difficulties of transferring technologies through conventional contracts or market channels. The local infrastructure supporting the creation of these competencies may be very concentrated in a specific region, such as the Silicon Valley in California or the biotechnology complex around Boston. As a result, specific sites become centers for specific technological competences and attract considerable investments by multinational firms in R&D and often production (because of the need for close links between this activity and R&D):

Such motives for offshore R&D investment also tend to direct such outward flows to other industrial economies, consistent with the data in Table M-2.

These influences on cross-national R&D investment and other forms of international interaction in the innovation process resemble the factors that have given rise to high levels of intraindustry trade—the growing returns to specialization in specific technological activities or competences, some apparent decline in "scope economies" among specific competences (reflected as well in the recent efforts of U.S. firms to restructure, divest unrelated lines of business, and focus on "core competences"), and the increased international dispersion of these competences. As a result, we find firms in industries such as pharmaceuticals seeking to establish R&D centers in ''centers of excellence" around the world, even as these R&D centers specialize in certain products or drug therapies. Strategic alliances among firms for the development or manufacture of new products often are based on the effort of participants to combine their complementary technological and other skills. Indeed, recent work by Mowery et al. (1996) suggests that a substantial fraction of recent strategic alliances tends to enhance the dissimilarity of participants' technological capabilities. Other strategic alliances, however, result in high levels of interfirm learning and transfer of such capabilities, producing greater similarity among participants' technological capabilities. This observation underscores the broader point that both technology transfer and knowledge accumulation are aided by cross-national R&D investments and other international linkages in the innovation process.

International flows of R&D investment thus are attracted to national or regional economies that can nurture specific technology-based capabilities, just as other types of international investment flows tend to reward policies favoring economic stability, property rights, and human capital. This argument has two implications. Even as the nationality of investors in R&D activity within a given locality may become more and more blurred, the importance of "national innovation systems" in supporting the infrastructure and other local capabilities to attract these investments remains important. Increased cross-national R&D investment thus may not reduce international or even intranational differences in such localized capabilities and infrastructure. (Within the United States, for example, California's Silicon Valley remains a dominant center for R&D in the electronics industry, although it now has scarcely any semiconductor manufacturing capacity.) Cantwell (1995) asserts that "The globalization of technological innovation in MNCs, in the sense here of an international integration of geographically dispersed and locally specialized activities, tends to reinforce and not to dis-

mantle nationally distinctive patterns of development or national systems of innovation. . . . Contrary to what is sometimes alleged, globalization and national specialization are complementary parts of a common process, and not conflicting trends" (p. 171). Secondly, these localized capabilities are developed through path-dependent processes in which both supply and demand factors, as well as history, matter a great deal.

This section discusses recent trends in R&D investment, primarily government-funded R&D, in the United States and other industrial and industrializing economies since 1980. Especially among the member states of the Organization for Economic Cooperation and Development (OECD), restructuring of domestic R&D systems has followed broadly similar lines since the early 1980s, largely as a result of the end of the Cold War. But many of the features that distinguished the U.S. federal R&D budget from those of other industrial economies, including its size and emphasis on defense-related and health-related objectives, remain salient. This cross-national comparison of industrial-economy public R&D spending also includes the European Commission, whose civil R&D spending priorities contrast with those of both Western European member states and other industrial economies. What is lacking in this comparison, however, are measures of effectiveness, be these defined in terms of "R&D productivity" or some other measure of economic or social returns from public R&D expenditures. The only reliable data for cross-national comparisons involve input measures, but the real concern for policy is the relationship between inputs and outputs. Nonetheless, although measures of performance are lacking, there is abundant anecdotal and descriptive evidence that suggests that performance is affected at least as much by the structure of government R&D programs and supporting policies as by the scale of these budgets.

The most dramatic shift in spending trends within the U.S. R&D system during the past 15 years is the decline in R&D spending by the federal government. Having grown at an average rate of 6 percent per year in real terms during 1980-1985, inflation-adjusted federal R&D spending declined at an average rate of roughly 1 percent per year during 1985-1995. R&D spending from "Other nonfederal sources" (R&D funded by state and local governments, as well as universities and colleges) grew by 2 percent in real terms during 1994-1995. The National Science Foundation (NSF) data currently available from the updated version of National Patterns of R&D Resources: 1996 provide only estimated levels of R&D spending for 1996 and 1997 (NSF, 1996), and these are less

reliable, particularly for industry-funded R&D investment, than the actual spending levels reported (with a lag) by the NSF. In addition, revisions in NSF data collection procedures mean that the data on industry-funded R&D before and after 1991 are not strictly comparable with one another, especially for the disaggregated components of R&D spending and for individual industries (see NSF [1996]). Accordingly, our discussion of spending trends covers only the period through 1995, and we confine the analysis of trends in the components of industry-funded R&D investment to the 1991-1995 period. The other major source of R&D spending within the U.S. system is industry, which accounted for 59 percent of total R&D spending in 1995. Industry-financed R&D scarcely grew at all in real terms during the early 1990s, but this trend was reversed in 1993, and the NSF data for 1993-1995 reveal that real industry-funded R&D spending grew at an average annual rate of nearly 10 percent during that period (NSF, 1998).

These shifting growth trends in industry- and government-funded R&D have produced wide swings in the rate of growth in overall U.S. R&D spending since 1980. Total national R&D spending grew by nearly 7 percent annually in constant-dollar terms during 1980-1985, but during 1985-1993, the average annual rate of growth in total constant-dollar R&D spending declined to 1 percent. More recently, however, total U.S. R&D spending has grown in real terms at an average annual rate of almost 3 percent between 1993 and 1995.

Declines in federal R&D spending are largely due to reductions in defense-related R&D spending, which increased from 50 percent of federal R&D spending in 1980 to almost 70 percent by 1986, a level from which it has declined once again to approximately 52 percent of total federal R&D spending. NSF measures of the share of defense-related spending in the U.S. federal R&D budget are somewhat lower than those from the OECD, which estimates the 1996 defense-related share of U.S. R&D spending to be closer to 55 percent. Both sets of data, however, highlight a decline in this share since 1980. The economic consequences of this reduction in defense-related R&D spending are difficult to project. Technological "spillovers" from defense to civilian applications of this spending now are less significant than was true of the 1950s and early 1960s, as the requirements for military and civilian applications in such technologies as aerospace and electronics have diverged. In addition, a considerable portion of federal defense-related R&D spending was directed to applied research, such as weapons testing, that generated few civilian economic benefits. Nevertheless, the enormous defense-related R&D budget contained a substantial basic research component, and defense-related R&D accounted for a considerable share of federally funded research in U.S. universities in such areas as electronics (see below). Reductions in spending in these areas could have negative consequences for civilian innovative performance. Further reductions below this share of the overall federal R&D budget appear to be unlikely, although pressure for increased procurement spending may increase the share of development activities within the defense-related R&D budget.

The outlook for growth in federal civilian R&D spending is uncertain. Legislative actions by the Senate and the House of Representatives in 1997 increased the fiscal 1998 federal R&D budget by more than 4 percent above its prior-year levels, and more recent forecasts of budgetary surpluses may result in further increases in federal R&D spending, especially in biomedical research. Longer-term trends, however, are less favorable for civilian R&D spending. In the absence of political agreement on reductions in entitlement spending for the elderly and health care, growth in these items will constrain growth in federal R&D spending. Even in the context of a balanced overall federal budget, a state of grace that is likely to be temporary at best, it is unlikely that future federal R&D spending will increase significantly above its 34 percent share of total U.S. R&D spending for 1995.

Another important shift in the profile of U.S. R&D spending growth during the 1990s is the reduction in the share of "research" within overall "R&D." During 1991-1995, total U.S. spending on basic research (measured in 1992 dollars) declined at an average rate of almost 1 percent per year. Industry-funded basic research dropped from $7.4 billion in 1991 to $6.2 billion in 1995 (in 1992 dollars)—real federal spending on basic research increased slightly during this period, from $15.5 to almost $15.7 billion. Industry-funded investments in applied research scarcely grew during this period, while federal spending on applied research declined at an annual rate of nearly 4 percent. In other words, the upturn in real R&D spending that has resulted from more rapid growth in industry-funded R&D investment is almost entirely attributable to increased spending by U.S. industry on development, rather than research. Indeed, the NSF reports that industry-funded real spending on ''development" grew by more than 14 percent during 1991-1995, from $65 billion to $74.2 billion (federal development spending declined during this period, reflecting the cutbacks in defense-related R&D spending).

Extrapolation of future trends from recent data that cover only four years is hazardous. Nevertheless, if the trends of the early 1990s continue unabated, U.S. R&D spending could change its profile and pattern of growth significantly. The reduction in the federal government's share of overall R&D means that increased federal R&D spending will do less to offset the effects of any future reductions in the rate of growth in industry-financed R&D spending on overall U.S. R&D spending levels. Since industry-funded R&D investment tends to move procyclically, future trends in total U.S. R&D spending are likely to be more sensitive to the domestic business cycle. In addition, the reduction in the federal government share of total R&D spending and the apparent shift in the profile of industry-funded R&D spending to favor development more heavily than "upstream" research activities (basic and applied research) could shorten the time horizon of overall U.S. R&D investment, with important consequences for both national and international scientific and technological advance.

How do U.S. R&D spending trends compare with those of other major industrial economies? One of the most important and dramatic trends is the declining share of global R&D spending accounted for by the United States. Figure M-3 displays trends during 1960-1994 in the U.S. share of total G-7 R&D spending, along with trends in the U.S. share of total R&D spending within the OECD economies since 1990. Within the G-7, U.S. R&D spending has declined from almost 70 percent of the total in 1960 to slightly more than 48 percent in 1994. The bulk of this decline occurred during 1960-1980, and the U.S. share of G-7 R&D spending has been nearly constant since 1990. Interestingly, the U.S. share of OECD R&D spending has increased slightly since 1990, from 42.8 percent in 1990 to 43 percent in 1994.

The overall decline in the U.S. share of G-7 R&D is by no means undesirable. First, it reflects the successful reconstruction and growth of the European and Japanese economies since 1945, developments that have contributed to international political and economic stability. Second, the growth of non-U.S. R&D spending creates opportunities for U.S. taxpayers to benefit from the public expenditures of foreign governments, just as foreign citizens have benefited from U.S. public financing of R&D. But it is critically important that U.S. firms and nationals have access to these foreign R&D programs, an issue that has sparked controversy in the past.

Figure M-3

U.S. share of G-7, OECD, and World R&D

Source: U.S. Department of Commerce.

Beyond its declining relative size, how do recent changes in the structure of the U.S. R&D system compare with those in other OECD economies? The most recent comparative data suggest that structural change in the pattern of U.S. R&D spending and performance since 1980 parallel trends in other OECD economies. Table M-5 contains data on trends during 1971-1993 in the distribution of R&D performance and funding among government, academia, and industry in the five OECD member states with the largest R&D budgets. These indicators suggest that the post-1981 restructuring of the sources of funding in the U.S. R&D system resembles similar processes in the United Kingdom, France, Germany, and Japan—the share of public funding of R&D is declining and industry R&D funding is growing (although the Japanese government made a public commitment in 1996 to significantly increase public R&D spending, slow economic growth may constrain growth in public R&D spending). The sharpest decline in public-sector R&D funding among these five nations during 1981-1993 occurred in the United Kingdom, where the share of national R&D spending funded by public sources dropped by roughly one-third. In both the United States and France, public funding declined by approximately 10 percent of R&D spending, a decline in the public share of roughly one-fifth. The data for Germany and Japan reveal smaller declines in the public share through 1993.

The shifts during 1981-1993 among universities, industry, and government in the performance of R&D within these five economies are less significant (again, with the exception of the United Kingdom, where a number of public research laboratories have been privatized). The data for the United States reveal very small increases (a shift of less than 1 percent in the share of each) in the share of R&D performed by industry and universities, and a slightly larger decline (of nearly 2 percent) in the share of R&D performed in government laboratories. The share of publicly performed R&D in both Japan and France declined by comparable or slightly larger amounts, while the German data (which include the effects of unification) rise modestly. The United Kingdom data, however, reveal a sharp decline of nearly 7 percent in the share of R&D performed in government facilities.

If there is an ''outlier" in these measures of structural change since the early 1980s in national R&D systems, it is the United Kingdom, rather than the United States. In general, the shifts in funding sources in the United States and these other economies are slightly larger than the shifts in performance. Indeed, the contrasting magnitude of the shifts in funding sources, as opposed to R&D performance, reflects the difficulties of undertaking radical structural changes in national R&D systems of the sort observed in the United Kingdom. The political costs of closing or privatizing large public research establishments often outweigh those associated with the gradual shrinkage of such facilities through incremental shifts in the shares of public and private R&D spending.

Although the patterns of structural change in these large industrial-economy R&D systems display considerable similarity, substantial differences remain in the objectives of their public R&D spending. These contrasts are heightened when the R&D budget of the European Commission, which accounts for roughly 4 percent of total government R&D spending in Western Europe, is added to a comparison of civil and defense-related R&D spending in 1991 and 1996 (Tables M-6 and M-7).

Despite the sharp cutbacks in defense spending and defense-related R&D, for example, the United States continues to spend substantially more on defense as a share of its central government R&D budget than any of these other indus-

trial economies. Reflecting its lack of responsibility for national security matters, the entirety of the European Commission's R&D budget is devoted to civil R&D. In 1995, the United States spent a smaller fraction of gross domestic product on nondefense R&D than any nation besides the United Kingdom among the five nation-states in this comparison (NSB, 1998).

There are also significant differences in priorities within the civilian R&D budgets of these five nations and the European Commission (Table M-7), contrasts that are stable across the 1991–1996 time period. The United States devotes a larger share of its civilian R&D budget to environmental and health objectives than any other entity in Table M-6—more than 6 times the environmental and health R&D share of the Japanese civil R&D budget in 1995, and more than 3 times this share in the German or French civil R&D budgets. The vast majority of the U.S. "environment and health" civil R&D budget is devoted to biomedical research. Interestingly, the share of the U.S. civil R&D budget devoted to "economic development" objectives (the OECD survey from which these data are taken defines ''economic development" to include "promotion of agriculture, fisheries, and forestry; promotion of industry; infrastructure; energy'' [OECD, 1997, p. 27]) does not differ greatly from those of the four other nation-states in Table M-7, as the United States ranks ahead of France and the United Kingdom, but behind Japan and Germany, in this share. The European Commission, however, allocated more than 60 percent of its R&D budget (more than $2.5 billion) to economic development, almost twice as high a share as that in Japan's civil R&D budget. The United States also devoted a much higher share of its civil R&D budget than any of these other nations to space exploration in both 1991 and 1996. Reflecting the longstanding dominance of the U.S. R&D budget by mission-oriented agency spending, the share of U.S. civil R&D allocated to undirected basic research was lower in 1996 than that of any other nation-state in Table M-7.

Both the European Commission and Japan also devote significant resources to electronics-related R&D within their civil R&D budgets. More than $370 million ECU (approximately $450 million to $500 million) were allocated by the European Commission in 1995 to support R&D in electronics. The Japanese government announced a new initiative in semiconductor-related R&D in 1996, involving public contributions of roughly $100 million to $110 million to a set of public-private collaborations whose total annual budget is nearly $200 million (Flamm, 1996). In contrast to many previous government-sponsored collaborative R&D projects, Japanese universities are involved in this initiative.

Nevertheless, these programs in electronics and information technology R&D are dwarfed by recent U.S. government-funded initiatives. The federal High-Performance Computing and Communications (HPCC) program spent more than $700 million in fiscal 1993 alone, and more than $1 billion annually in fiscal 1997 and 1998 (AAAS, 1997). The HPCC program includes the bulk of NSF spending on information technology R&D, but the Department of Defense (DoD)

contributes additional funds to support R&D in both academia and industry in this area. In fiscal 1995, the federal government spent $766 million on R&D in electrical engineering, $793 million on R&D in metallurgy and materials science, and $982 million on R&D in computer science. This total includes funds allocated to the HPCC. DoD funds accounted for more than 62 percent of this total budget of more than $2.5 billion (NSF, 1997). U.S. federal R&D spending in fields supporting advances in electronics and computer technology thus appears to be substantial.

Governments in the industrializing Asian economies, including South Korea, Indonesia, Taiwan, and Singapore, also laid plans for higher public spending on R&D in the 1980s and 1990s. By the late 1980s, industrial technology projects accounted for almost 20 percent of Taiwan's government R&D budget (Schive, 1995). In both Taiwan and South Korea, the Asian economies with the strongest electronics and semiconductor industries, the role of government shifted by the mid-1990s. In the 1970s and 1980s, governments encouraged inward technology transfer and supported applied R&D in industrial and government laboratories. As domestic firms developed their capabilities, government-performed R&D lost much of its importance and effectiveness. Beginning in the 1990s, both South Korea and Taiwan sought to develop domestic R&D infrastructures capable of supporting R&D at the frontiers of technology, rather than the sorts of "catch-up" activities involving the inward transfer, adoption, and improvement of technologies developed elsewhere. Among other mechanisms, increased funding of academic R&D, the reform of higher education, and the development of "science parks" have played important roles in these recent efforts.

Despite increased public R&D spending, however, the even more rapid growth of privately funded R&D spending in both economies reduced the publicly financed share of total R&D investment during the 1980s (Dahlman and Kim, 1992; Schive, 1995). In South Korea, as well as Malaysia and Indonesia, two other Asian economies seeking to strengthen their domestic R&D capabilities, the financial crisis now roiling the region is likely to further reduce the government share of total R&D investment. A few others, however, such as Taiwan, are maintaining programs of public-private collaboration in industrial and academic R&D.

Public R&D spending priorities in the United States contrast with those of other major industrial economies, contrasts that are heightened when the significant R&D programs of the European Commission are added to the comparison. The United States continues to devote a larger share of total public R&D spending to defense (a significant portion of which goes to support R&D in information technology and electronics, some of which in turn yields civilian technological "spin-offs"). In addition, within its civilian R&D budget, the U.S. government spends a much larger fraction of total resources on health-related and space R&D.

Any assessment of the likely effects of these contrasting priorities must recognize that the scale of the overall U.S. federal R&D budget exceeds those of

other nations and the EU by such a wide margin that even R&D priorities (such as information technology) that account for relatively small shares of the total federal R&D budget still receive a level of investment that compares favorably with those of other governments. Moreover, the economic effects of public R&D programs are heavily affected by the structure of these programs and the R&D systems within which they operate. Previous large-scale regional European programs of "strategic-technology" R&D in information technology (e.g., ESPRIT, JESSI) have failed to prevent the decline of large segments of the European information technology industry. Recent Japanese initiatives, such as the Fifth Generation computer technology program that sparked a hysterical reaction in the United States, as well as other collaborative efforts in software technology (see Baba et al. [1996]), have had little effect on the competitive fortunes of Japanese electronics and computer firms. Many European programs have been hampered by cumbersome and inflexible administrative structures, as well as continuing pressure to distribute R&D funds among EU members states in some equitable fashion. In addition, regulatory, trade, and competition policies within EU member states often have insulated domestic firms from competition, reducing pressure to adopt and implement the results of these R&D programs more rapidly. Japanese collaborative programs have suffered from the inability of program designers to develop a sufficiently robust and reliable "vision" of future technology developments to coordinate the R&D efforts of firms and universities effectively in ''frontier" areas of science and technology that are subject to severe uncertainties.

The pluralistic institutional and programmatic environment of the United States as well as the large-scale and highly competitive nature of the U.S. domestic market have in recent years produced high rates of product innovation that have yielded high economic returns. But U.S. firms arguably remain weaker in the "cyclical innovation" highlighted by Gomory (1989) as critical to long-term competition in more mature markets. In addition, Japanese and European policy makers are aware of the structural weaknesses of their innovation systems, and future programs may prove to be more effective. Although the recent performance of the U.S. R&D system seems to compare favorably with those of many nations, U.S. managers and policy makers cannot be complacent. As international competition is based more and more on knowledge, the assets and capabilities that produce national competitive advantage become more and more mobile across international boundaries. Competitive and technological challenges are likely to appear from unexpected quarters and will emerge more rapidly.

Although overall U.S. foreign R&D investment has been growing slowly during the past 15 years, cross-national R&D investment, especially inward R&D investment in the U.S. economy, and other forms of interaction between U.S. and

foreign firms in the technological innovation process are virtually certain to grow in the future. The forces giving rise to these trends are both pervasive and deeply rooted in the economic reconstruction and global growth that have characterized the post-1945 era. What does this imply for the future evolution of "systems frictions" in the areas of technology and trade policy that have previously been discussed by Ostry (1990)? In this section, I briefly consider possibilities for conflict flowing from the unusual structure of the U.S. "national innovation system" within the global economy, and then discuss some implications for U.S. technology policy.

The concept of a "national innovation system" emerged from earlier work by Freeman (1987) and Nelson (1993), among others. A "national innovation system" refers to the collection of institutions and policies that affect the creation, development, commercialization, and adoption of new technologies within an economy. As such, the U.S. national innovation system includes not just the institutions performing R&D and the level and sources of funding for such R&D, but policies—such as antitrust policy, intellectual property rights, and regulatory policy—that affect investments in technology development, training, and technology adoption. But government policies by no means determine all elements of the structure of national innovation systems, which are themselves the result of complex historical processes of institutional development. Moreover, the performance of these systems within most industrial economies depends on the actions and decisions of private enterprises, and these decisions can reinforce or offset the effects of public policies.

Much of the current controversy over foreign firms' exploitation of U.S. technological assets through their R&D investments in this economy (OTA, 1994) rests on a set of assertions about the contrasting structures of the U.S. and other national innovation systems, such as those of Japan and Germany. Access by foreign enterprises to locally developed inventions or technologies within an economy is heavily influenced by the structure of that economy's national innovation system. The U.S. system probably is "leakier" than other systems, because of (1) the prominent role of relatively open institutions, especially universities, as performers of world-class R&D; (2) the highly developed market for corporate control, which facilitates acquisitions of U.S. firms by other U.S. or foreign firms; and (3) relatively liberal U.S. government policy toward direct foreign investment. But the "openness" of the U.S. national innovation system is only partly a function of government policy—this condition also reflects historical evolution and other factors. The prominent postwar role of high-technology startup firms in the U.S. national innovation system, for example, is partly a result of government defense procurement and antitrust policies. But, this unusual structure was also influenced by the development of financial institutions and a finan-

cial system that are regulated, but hardly controlled, by government, as well as a university research infrastructure that mixes public and private funding and institutions.

The national innovation systems of other industrial economies are the outcome of similar combinations of government policy, historical evolution, and private decision making. As such, the possibilities for intergovernmental negotiations over access to industrial technologies to produce meaningful results may be limited. How, for example, should one measure the extent of openness of one nation's "innovation system," relative to that of another? What does reciprocity imply? Since a far greater proportion of Japan's R&D is carried out in industry (see Table M-5), does an agreement on reciprocal access imply that both U.S. and Japanese firms must open their sensitive technology development activities to visitors from firms in each nation? Such an agreement would not be welcomed by U.S. firms. In addition, as this example suggests, government policies may have little near-term effect on access—the differences between the U.S. and Japanese systems of corporate governance will not be eliminated by government initiatives alone. Concerns over the access by one nation's firms to another's industrial technology base may be well founded, but their resolution is likely to be slow.

The influence of government policies on the "openness" of national innovation systems also is a result of both public policies and private firms' reaction to these policies. Indeed, as was suggested earlier, a portion of the recent growth in international strategic alliances reflects the actions of individual firms, often in reaction to state policies. For example, the "technonationalist" R&D policies of the European Union and the United States in semiconductors have provided a motive for the formation of strategic alliances among firms from these economies; so have managed trade policies in industries such as automobiles. Both the static characteristics and the dynamic evolution of these national innovation systems thus depend critically on the behavior of private firms.

The Clinton administration, which came to power in a flurry of commitments to a "new approach" in U.S. technology policy, has in fact displayed considerable consistency with many of the programmatic precedents established by its immediate predecessors, reflecting the fact that many of the technology policies of the Reagan and Bush presidencies were the result of pressure from Democratic Congresses. These similarities extend to two dilemmas that also confronted the Reagan and Bush administrations: (1) the problems imposed by political requirements to capture the bulk of the economic returns from technology policies whose results may benefit foreign firms; and (2) the enduring tension between programs that support technology development and those supporting technology adoption. Portions of the following paragraphs draw on Ham and Mowery (1995).

The political justification for many U.S. technology development programs

(including those supported with DoD funds) now rests on the ability of U.S. firms and citizens to capture the economic benefits of these programs. Such justifications also apply to more and more federal science programs and funding. Unfortunately, given the characteristics of the outputs of many of these programs, the structure of the U.S. firms participating in them, and the structure of the markets for the goods into which the results of these programs are incorporated, capturing the entirety or perhaps even a majority of the economic benefits from some of these programs is infeasible. The constraints imposed on program design by these political realities exacerbate tensions between U.S. trade and technology policies and, paradoxically, may reduce the economic returns to U.S. firms and taxpayers from these programs.

Reconciling the political requirements for such a distribution of benefits with the economic and technological realities of the late twentieth century has proven difficult. Many of the technology policies of the Reagan, Bush, and Clinton administrations have attempted to restrict foreign firms' access to domestic programs or have attempted to limit the international diffusion of the results of such programs. The White House restricted foreign access to public discussions of research in high-temperature superconductivity in 1987, and foreign firms' access to the results of federally funded research in the national laboratories has been restricted in several cases.

The National Center for Manufacturing Sciences (NCMS) and the U.S. Consortium for Automotive Research (USCAR), announced by the Clinton administration in 1993, exclude foreign firms from formal membership. SEMATECH also excluded foreign firms from participation while it was receiving federal funds. This consortium now has enlisted electronics manufacturers from Taiwan, South Korea, and Western Europe in a new, parallel collaborative R&D organization (see Appleyard et al. [1998]). Foreign participation in the Commerce Department's Advanced Technology Program (ATP) is subject to various restrictions, which include determinations by U.S. policy makers that the home-country governments of these firms provide nondiscriminatory access to similar technology development programs, that they provide significant protection for intellectual property, and other conditions that have little bearing on the benefits to the U.S. economy of foreign participation (these conditions have resulted in the denial of funding thus far for only one ATP project, which included a Japanese firm among its participants). Transfer of NCMS-developed technologies by member firms to their foreign subsidiaries is selectively restricted. Cooperative research and development agreements between federal agencies (including the National Institutes of Health or the Department of Energy laboratories) must include provisions to ensure "substantial domestic manufacture" of the resulting technologies or products. Many of the current restrictions on foreign participation, which differ among U.S. technology programs, base the determination of foreign-firm eligibility on assessments of home-country government policies, on the assumption that denial of access to foreign firms will increase pressure for

change in the policies of their governments. The bases for such assessments of home-government policies are relatively subjective, and are surprisingly "nontransparent" (i.e., they are not based on any single or comprehensive published assessment, and there is no well-developed process for review of these determinations). These statutory requirements for the fulfillment of a lengthy, inconsistent, and complex set of conditions across programs also mean that foreign-firm participation that is deemed by policy makers to be economically beneficial for U.S. firms and taxpayers may be prohibited for one or another reason that has little to do with the specific merits of an individual proposal.

Many of these U.S. government restrictions on foreign access to U.S. technology programs are a response to similar restrictions on U.S. firms' access to the strategic "technology" programs supported by other industrial economies. Japan's cooperative R&D programs long excluded foreign firms, although many of these restrictions have been relaxed in recent years. In addition, many of the programs of the European Union and its member states have restricted participation by non-European firms, although partial exceptions have been made in the case of such firms as IBM.

U.S. restrictions on foreign participation or international dissemination of results are not likely to affect the distribution of the economic returns to these programs. For example, the automobile firms participating in USCAR maintain extensive manufacturing and product development links with foreign firms, as did the U.S. semiconductor firms participating in SEMATECH when the consortium prohibited foreign participation. The "U.S. discovery" of high-temperature superconductivity that led to the 1987 White House symposium was in fact accomplished by two German scientists working in a Swiss industrial R&D laboratory owned by a large U.S. multinational firm, IBM. Establishing the "national ownership" of this scientific accomplishment is futile and counterproductive. Such restrictions also create some risk that U.S. firms will continue to be excluded from foreign nations' current and future technology programs.

The focus of many Clinton administration policy makers, as well as those of previous administrations, on technology creation and development as the key source of economic benefits overlooks the benefits from technology adoption in a U.S. economy that now is "first among equals," rather than technologically preeminent, and that is open to international trade and technology flows. Like its immediate predecessors, technology policy in the Clinton administration supports the creation of knowledge-based competitive assets that are internationally mobile, while placing less weight on improvements in the ability of U.S. firms and workers to absorb and apply technological advances from external or foreign sources.

Government actions continue to matter a great deal in the modern global economy of mobile capital, goods, and technology. But the mobility of the technological assets created with federal funds means that the consequences of many government policies may differ from their intended goals. Moreover, the

economic interests of U.S. citizens may not always coincide with those of U.S. firms that seek restrictions on foreign firms' access to U.S. technology development programs. As foreign firms play a larger role in this economy in production, technology development, and, potentially, in the support of research in small firms and universities, the economic benefits from any restrictions on foreign-firm access are likely to flow primarily to the shareholders and managers of U.S. firms competing with these foreign entities, rather than benefiting the larger public. Similarly, U.S. government programs to support the development of "strategic technologies" such as flat-panel displays create considerable risk that imports of cheaper versions of these important intermediate goods may be restricted, imposing severe costs on U.S. firms seeking to compete in the production of systems incorporating such components.

The global environment within which future science and technology policies will be formulated and implemented will be characterized by a broader distribution of scientific and technological "centers of excellence," and by greater cross-border flows of R&D investment and technology. The U.S. R&D system and the federal R&D budget will remain by far the largest in the world, but the share of global R&D accounted for by R&D activity within the United States has declined significantly from its level of the 1960s and is unlikely to increase. Policies for the future thus must recognize the greater mobility of intellectual property, technological capabilities, and R&D investment. The United States must remain an attractive "platform" for R&D and related investments by U.S. and foreign corporations. Because much of the infrastructure that has contributed to the excellence of the U.S. R&D system and its attractiveness for U.S. and foreign corporations is public, federal investments in R&D remain essential to the future well-being of the U.S. economy.

But federal policies for science and technology also must be predicated on a more realistic view of the relationship among the U.S. and foreign nations' R&D systems and on a more realistic conceptualization of the sources of the economic benefits associated with innovation. Rather than restricting foreign access to the results of publicly funded R&D in the United States, results that themselves are internationally mobile, policy makers should focus on strategies to improve the domestic adoption and implementation of new technologies from domestic and international sources. Among other things, such an approach will require that both policy makers and U.S. industrial managers redouble their efforts to improve access to the growing R&D systems of other industrial and industrializing nations. The current environment is a legacy of enlightened U.S. and foreign policies of support for liberalized trade and economic development throughout the global economy. Future science and technology policies should be designed to exploit these legacies of past policy successes.

AAAS (American Association for the Advancement of Science). 1997. AAAS Report XXII: Research and Development FY1998. Washington, D.C.: AAAS.

Appleyard, M. M., R. M. Ham, and G. Linden. 1998. The Transition to 300-mm Wafers in the Semiconductor Industry: A New Era of Interfirm Cooperation. unpublished manuscript. University of California, Berkeley.

Archibugi, D., and J. Michie. 1995. The globalization of technology: A new taxonomy. Cambridge Journal of Economics 19:121-140.

Baba, Y., S. Takai, and Y. Mizuta. 1996. The User-Driven Evolution of the Japanese Software Industry: The Case of Customized Software for Mainframes. 104-130 in D. Mowery, ed., The International Computer Software Industry. New York: Oxford University Press.

Cantwell, J. 1991. Technological Innovation and Multinational Corporations. Oxford: Blackwell.

Cantwell, J. 1995. The globalization of technology: What remains of the product cycle model? Cambridge Journal of Economics 19:155-174.

Dahlman, C., and L. Kim. 1992. Technology policy for industrialization: An integrative framework and Korea's experience. Research Policy. 21/5.

Flamm, K. 1996. Japan's New Semiconductor Technology Programs. Unpublished report for ATIP. Tokyo.

Freeman, C. 1987. Technology Policy and Economic Performance: Lessons from Japan. London: Pinter.

Gomory, R. 1989. From the "ladder of science" to the product development cycle. Harvard Business Review 67: 99-105.

Ham, R. M., and D. C. Mowery. 1995. Enduring dilemmas in U.S. technology policy. California Management Review 37/4: 89-107.

Ham, R. M., and D. C. Mowery. 1997. The U.S. response to globalization: Searching for the keys under the lamp post. In J. Dunning, ed., Governments, Globalization, and Competitiveness. New York: Oxford University Press.

Mowery, D. C. 1997. Technological Innovation in a Multipolar System: Analysis and Implications for U.S. Policy. Prepared for the Study Group on the Globalization of Industrial R&D. New York: Council on Foreign Relations.

Mowery, D. C., and D. J Teece. 1993. Japan's Growing Capabilities in Industrial Technology. California Management Review 35/2 (Winter 1993): 9-34.

Mowery, D. C., J. Oxley, and B. Silverman. 1996. Strategic alliances and interfirm knowledge transfer. Strategic Management Journal 17 (S2).

National Academy of Engineering. 1996. Foreign Participation in U.S. Research and Development: Asset or Liability? Washington, D.C.: National Academy Press.

National Science Board. 1993. Science and Engineering Indicators 1993. Arlington, Va.: National Science Foundation.

National Science Board. 1996. Science and Engineering Indicators—1996. Arlington, Va.: National Science Foundation.

National Science Board. 1998. Science and Engineering Indicators. Arlington, Va.: National Science Foundation.

Nelson, R. R., ed. 1993. National Innovation Systems. New York: Oxford University Press.

Nelson, R. R. 1996. The Sources of Economic Growth. Cambridge, Mass.: Harvard University Press.

NSF (National Science Foundation). 1996. National Patterns of R&D Resources: 1996. Washington, D.C.: U.S. Government Printing Office.

NSF (National Science Foundation). 1997. Federal Funds for Research and Development: Fiscal Years 1995, 1996, and 1997. Washington, D.C: NSF.

NSF (National Science Foundation). 1998. Update of National Patterns of R&D Resources: 1996. Unpublished data.

OECD (Organization for Economic Cooperation and Development). 1996a. Science, Technology, and Industry Outlook. Paris: OECD.

OECD (Organization for Economic Cooperation and Development). 1996b. Technology and Industrial Performance. Paris: OECD.

OECD (Organization for Economic Cooperation and Development). 1997. Scoreboard of Science and Technology Indicators. Paris: OECD.

OECD (Organization for Economic Cooperation and Development). 1998. Main Science and Technology Indicators: 1997, #2 Paris: OECD.

Ostry, S. 1990. Governments and Corporations in a Shrinking World. New York: Council on Foreign Relations.

OTA (Office of Technology Assessment, U.S. Congress). 1994. Multinationals and the U.S. Technology Base. Washington, D.C.: U.S. Government Printing Office.

Patel, P. 1995. Localized production of technology for global markets. Cambridge Journal of Economics 19:141-153.

Patel, P., and K.L.R. Pavitt, 1991. Technological strategies of the world's largest companies. Science and Public Policy. 18/6 (December).

Schive, C. 1995. Industrial policies in a maturing Taiwan economy. Industry and Innovation 2/1.

Vernon, R. 1966. International investment and international trade in the product cycle. Quarterly Journal of Economics 80:190-207.