National Academies Press: OpenBook
« Previous: Coping With Technological Change: U.S. Problems And Prospects
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 191
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 192
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 193
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 194
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 195
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 196
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 197
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 198
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 199
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 200
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 201
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 202
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 203
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 204
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 205
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 206
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 207
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 208
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 209
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 210
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 211
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 212
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 213
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 214
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 215
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 216
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 217
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 218
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 219
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 220
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 221
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 222
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 223
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 224
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 225
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 226
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 227
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 228
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 229
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 230
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 231
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 232
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 233
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 234
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 235
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 236
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 237
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 238
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 239
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 240
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 241
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 242
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 243
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 244
Suggested Citation:"Does Technology Policy Matter?." National Research Council. 1987. Technology and Global Industry: Companies and Nations in the World Economy. Washington, DC: The National Academies Press. doi: 10.17226/1671.
×
Page 245

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

DOES TECHNOLOGY POLICY MATTER? 191 Does Technology Policy Matter? HENRY ERGAS How do technology policies differ among nations? What impact do these differences have on innovation and, more generally, on industrial strutures? These questions are the central concerns of this chapter. Innovation is the use of human, technical, and financial resources to find a way of doing things. As an inherently uncertain process, it requires experimentation with alternative approaches, many of which may prove unsuccessful. Even fewer will survive the test of diffusion, where ultimate economic returns are determined. The historical success of the capitalist system as an engine of growth arises from its superiority at each of these levels: generating the resources required for innovation, allowing the freedom to experiment with alternative approaches, and providing the incentives to do so.1 Though relying primarily on market forces, the system has .interacted with government at two essential levels. The first relates to the harnessing of technological power for public purposes. Nation-states have long been major consumers of new products, particularly for military uses, and the need to compete against other nation-states provided an important early rationale for strengthening national technological capabilities. Whether this rationale persists as the primary motive for government action is a major factor shaping each country's technological policies (Earle, 1986). The second arises from the system's dependence on its social context. The development and diffusion of advanced technologies requires a system of education and training as a basis for supplying technology and skills, a legal framework for defining and enforcing property rights, and processes

DOES TECHNOLOGY POLICY MATTER? 192 such as standardization to reduce transactions costs and increase the transparency and efficiency of markets. These are, at least in part, public goods. The benefits of investment in education are appropriated by a multitude of economic actors, and those of the system of property fights are even more widely spread.2 The way these public goods are provided, and the role industry plays in this respect, differs greatly from country to country. This chapter examines the interactions between the technological system and government policy in seven industrialized countries: the United States, the United Kingdom, France, Germany, Switzerland, Sweden, and Japan. It pays particular attention to the relation between innovation policy and industrial structures. The countries examined are placed in three groups. Technology policy in the United States, the United Kingdom, and France remains intimately linked to objectives of national sovereignty. Best described as ''mission-oriented,'' the technology policies of these nations focus on radical innovations needed to achieve clearly set out goals of national importance. In these countries, the provision of innovation-related public goods is only a secondary concern of technology policy. In contrast, technology policy in Germany, Switzerland, and Sweden is primarily "diffusion-oriented." Closely bound up with the provision of public goods, the principal purpose of these policies is to diffuse technological capabilities throughout the industrial structure, thus facilitating the ongoing and mainly incremental adaptation to change. Finally, Japan is in a group of its own. Its technology policy is both mission-oriented and diffusion-oriented, and the form the policy takes differs in important respects from that in the other countries. Every taxonomy involves a loss of information, and the one proposed here does not escape this general rule. Thus, the United States has important policies —for example, in agriculture and in medical research—that are diffusion- oriented; equally, Germany and Sweden have major mission-oriented programs. But the focus of policy does differ in the three groups, and this allows a clearer examination of the relation between technology policy and innovation performance. These differences in policy stance—though not as sharp as they may at first appear to be—are important in shaping patterns of technological evolution, but the central hypothesis of this chapter is that technology policies are a facilitating rather than explanatory factor. The critical variables lie in how industry responds to the results and signals of efforts to upgrade national technological capabilities. In turn, this depends to a substantial extent on the environment in which industry operates. Technology policies cannot, in other words, be assessed independently from their broader economic and institutional context. A central feature of this context is a country's technological infrastruc

DOES TECHNOLOGY POLICY MATTER? 193 ture—its system of education and training, its public and private research laboratories, its network of scientific and technological associations. The effectiveness of this infrastructure depends not only on its internal functioning but also on the way a country's factor and product markets respond to innovation opportunities. Overall, this suggests that, even within the framework of a market economy, the process by which innovations are generated, selected, and imitated will differ according to the features of each country's institutional and economic structure. In exploring these features and their relation to countries' technology policies, this chapter follows the broad grouping set out above: The next three sections examine, respectively, the technology policies of the mission-oriented countries, namely, the United States, the United Kingdom, and France; the diffusion-oriented countries, namely, Germany, Switzerland, and Sweden; and Japan. The last two sections present, respectively, a synthesis of similarities and differences, with analyses of their broader implications for economic performance, and conclusions for policy formulation. THE MISSION-ORIENTED COUNTRIES Mission-oriented research can be described as big science deployed to meet big problems (Weinberg, 1967). It is of primary relevance to countries engaged in the search for international strategic leadership, and the countries in which it dominates are those where defense accounts for a high share of government expenditure on R&D (Table 1). Though it has also been used in these countries to meet perceived technological needs in civilian markets (for example, in nuclear energy or telecommunications), the link to national sovereignty provides its major rationale. The dominant feature of mission-oriented R&D is concentration. First and most visibly, this refers to the centralization of decision making. As TABLE 1 Share of Defense-Related R&D in Total Government Expenditure on R&D, 1981 Country Percent Defense-Related United States 54 United Kingdom 49 France 39 Sweden 15 Switzerland 12 Germany 9 Japan 2 SOURCE: Organization for Economic Cooperation and Development.

DOES TECHNOLOGY POLICY MATTER? 194 its name implies, the goals of mission-oriented R&D are centrally decided and clearly set out, generally in terms of complex systems meeting the needs of a particular government agency. Specifying these needs and supervising project implementation concentrates a considerable amount of discretionary power in the hands of the major funding agencies. Concentration also extends to the range of technologies covered. Virtually by its nature, mission-oriented research focuses on a small number of technologies of particular strategic importance—primarily in aerospace, electronics, and nuclear energy. As a result, government R&D funding in these countries is heavily biased toward a few industries that are generally considered to be in the early stages of the technology life cycle (Table 2). The scale of mission-oriented efforts also limits the number of projects and restricts the number of participants. At any particular time, only a small sham of each country's firms, likely among the larger ones, will have the technical and managerial resources required to participate in these programs. The concentration of government R&D subsidies on a small number of large firms is therefore also a feature of the countries in this group. Overall, mission-oriented programs concentrate decision making, implementation, and evaluation. A few bets are placed on a small number of races; but together, these bets are large enough to account for a high sham of each country's total technology development program. This concentration raises two obvious questions: First, how successful are the bets in relation to their own objectives? And second, do they have any effect on the efficiency with which the many other races are run—that is, are TABLE 2 Proportion of Total National Public R&D Funding by Type of Industry, 1980 Estimates Country Percentage of Total Public R&D Funding High-Intensity Medium-Intensity Low-Intensity United States 88 8 4 France 91 7 2 United Kingdom 95 3 2 Germany 67 23 10 Sweden 71 20 9 Japan 21 12 67 NOTE: High-, medium-, and low-intensity R&D industries are defined as firms whose ratios of R&D expenditures to sales are, respectively, more than twice, between twice and half, and less than half the manufacturing average. SOURCE: Organization for Economic Cooperation and Development.

DOES TECHNOLOGY POLICY MATTER? 195 technological capabilities more broadly diffused through the industrial structure? These questions will be considered in turn. Direct Effectiveness Attempting cost-benefit analyses of major mission-oriented programs involves enormous difficulties (Hitch and McKean, 1960). Three criteria for evaluating success can nevertheless be established: First, are stated product development goals being met? Second, is this being done within the original limits of time and cost? And third, are objectives for commercial markets being achieved? No country's programs perform extremely well when measured against these criteria. On balance, the effort in the United Kingdom has probably been the least successful, whereas that in France and the United States has generated a mixed record. Three factors seem to be critical in differentiating success from failure. First, do the agencies involved have the technical expertise, financial resources, and operating autonomy required to design and implement the program--and the incentives to ensure that it succeeds? Second, are relations with outside suppliers such as to provide appropriate incentives and penalties and do they allow for experimentation with alternative design approaches? Third, can agencies be prevented from expanding their "missions" indefinitely and, in particular, from moving into areas for which their capabilities and structures are inappropriate? The answers to these questions have differed in each of the three mission- oriented countries considered in this section. United Kingdom The United Kingdom's major difficulties arise from the pervasive lack of incentives in its system of mission-oriented R&D.3 The British system of public administration—with its emphasis on anonymity, committee decision making, and administrative secrecy—ensures that individual public servants have little interest in "rocking the boat." The emphasis on internal and procedural accountability also makes government reluctant to devolve major projects to reasonably autonomous entities, so that responsibilities are tangled, decision making is cumbersome, and the organizational and cultural context is inappropriate for developing new technologies. At the same time, the propensity of British agencies to form "clubs" with their suppliers—within which each supplier is treated on the basis of administrative equity rather than commercial efficiency—weakens whatever incentives suppliers may have to seek an early lead, while also ensuring that the resources available are so thinly spread as to be inef

DOES TECHNOLOGY POLICY MATTER? 196 fectual. Finally, the reluctance to build penalty clauses into development contracts, and to terminate unsuccessful projects (particularly when this would jeopardize the viability of an indigenous supplier), aggravates an inherent tendency to cost overruns. France France's relative success arises in considerable part from the great political legitimacy, operating autonomy, and technical expertise of its user agencies, combined with the strong incentives for success built into the highly personalized nature of power and careers in the French public administration.4 Particularly over the last decade, them has also been an effort to increase the competitive pressures bearing on suppliers, notably through tighter controls on costs, recourse to penalty clauses, and easing previous market-sharing arrangements. The effects of these moves have been heightened by improved financial and operating control within the agencies themselves. However, the French system has two major weaknesses. First, resource constraints have usually prevented experimentation with alternative design approaches, and the number of suppliers involved in each major project has typically been small.5 Second, though the French system has been compared favorably to that of the United Kingdom because there has been a reasonable willingness to run down (if not terminate) failures, the system has been highly vulnerable to goal displacement as a sequel to success. Agencies that have successfully accomplished a mission perpetuate themselves by designing new missions, frequently in areas unrelated to their original function. This "Frankenstein" effect is particularly noticeable in the energy and communications fields, where agencies have sought to expand their power base by diversifying their operations, generally into markets for which their technological capabilities and organizational structures are inappropriate. As a result, success in one period has in several cases been followed by failure in the next; and the system has had few mechanisms for reallocating resources smoothly.6 United States Considering only the efficiency with which projects are designed and implemented, the United States is intermediate between the United Kingdom and France; but it has over them the great advantage of scale.7 This advantage has three important dimensions. First, U.S. agencies draw on a much larger pool of external technological expertise both in selecting and implementing projects —and have much better mechanisms for doing

DOES TECHNOLOGY POLICY MATTER? 197 so, notably in university research. Second, funding for mission-oriented programs in the United States, particularly in defense, rarely falls short of the critical mass required to complete the development stage and usually has a higher continuity than program funding elsewhere. Third, the scale of funding is large, and the range of qualified suppliers is wide. Even the relatively small sums spent by the U.S. Department of Defense on programs of the Defense Advanced Research Projects Agency are large in relation to total defense R&D in the United Kingdom and France. The result is that experimentation almost invariably occurs with alternative design approaches and philosophies, even if only in the early steps of program conception. The United States may also benefit from the high degree of accountability inherent in its system of congressional scrutiny. This system has generated strong pressures for terminating unsuccessful projects, notably in the civilian sector (the supersonic transport plane and synfuels being prime examples), but seems to exercise much less control on the defense sector. Thus, an incidental effect of the system is that military programs may be allowed to continue too long, and some largely civilian programs are shut down too early. It has been argued that this places an excessive burden of financing projects with a high "public goods" content on the private sector. The safety and decommissioning of nuclear power plants may be cases in point (Brooks, 1983). Any overall assessment of the direct effectiveness of mission-oriented research must therefore be mixed; but the immediate returns on the research do appear to be higher in the United States and France than in the United Kingdom. However, even in the United States the products conceived directly by mission-oriented programs account for only a small share of the economy (Riche et al., 1983); the extent to which technology generated in these programs spreads to other areas of activity is therefore a major component of its overall impact. Secondary Effectiveness There are relatively few studies of the extent of secondary effects of mission-oriented technology policies or of the pace at which such effects occur. The few studies that do exist come to widely differing conclusions, frequently reflecting individual authors' views of the desirability of defense spending. None of the studies draws international comparisons. Two broad statements can nonetheless be advanced on the basis of the existing material: first, in every country, the direct spin-offs—in the sense of immediate commerical use of the results of mission-oriented research—are limited;8 second, the indirect spin-offs —arising mainly from improve

DOES TECHNOLOGY POLICY MATTER? 198 ments in skills and in technical knowledge transferable from the mission- oriented environment to that of commercial competition—appear to occur both in greater number and more rapidly in the United States than in the United Kingdom or France.9 It can be argued that the greater number and frequency of indirect spinoffs in the United States are partly due to differences in the way programs are designed and implemented. But the impact of these differences is compounded by differences in the countries' economic structures and scientific and technological environments. The Role of Program Design Four factors distinguish the design and implementation of mission-oriented programs in the United States from that of their counterparts in the United Kingdom and France. The first is the more limited direct role of the public sector in mission-oriented R&D in the United States. In general, the U.S. government performs a small share of its research in-house; the bulk of it is contracted to outside sources (Table 3). Even the management of national laboratories has been separated to a considerable extent from the public sector and devolved to universities or to private companies. Problems of technology transfer from the public to the private sector therefore concern a smaller share of government-funded R&D than is the case in France or the United Kingdom. Second, mission-oriented research in the United States involves a greater number and diversity of agents. It is true that within the private sector, most government research and procurement contracts go to a small number of suppliers. But the sums flowing to university research and to small and medium- size businesses are large in absolute terms.10 Thus, the number of small firms receiving 20 percent or more of their total R&D finance from government sources is nearly 10 times larger in the United States than in the United Kingdom or France. Moreover, insistence in defense TABLE 3 Share of Government-Financed R&D Performed in the Government Sector Country (Year) Share Performed by Government France (1983) 46.8 United Kingdom (1981) 38.9 Federal Republic of Germany (1981) 31.6 United States (1983) 25.7 Switzerland (1981) · 24.7 SOURCE: Organization for Economic Cooperation and Development.

DOES TECHNOLOGY POLICY MATTER? 199 procurement on "second-sourcing" of key components ensures a fairly broad diffusion of technological capabilities. The effects of this dispersion are compounded by a third factor, namely, greater U.S. willingness to disseminate the results of mission-oriented programs.11 Despite obvious security concerns, U.S. defense R&D programs have generally either made their results public or at least made them known to a wider circle than that immediately involved in the program. The information inherent in these results—such as measurement standards, properties of materials, or even identification of unsuccessful approaches to technical problems—is an important "public good." A greater U.S. willingness to disseminate results probably contains an element of bowing to the inevitable: Given the number and range of participants, results will be known sooner or later. However, other factors have also been at work. The widespread dissemination of results has been important in securing ongoing political approval for the programs. It has also been a way of preventing contractors from consolidating a "first-mover advantage" over competitors. At universities especially, dissemination has been facilitated by a research community that generally has not questioned the legitimacy of the program so long as their results could be fed into the system of "public or perish!" The dissemination of the results of mission-oriented programs in the United Kingdom and France differs from that in the United States in three respects. First, after programs are set up and running, there is little external political pressure to disseminate results. Second, the members of the program "club" themselves have little interest in seeing results publicized and tend to count more heavily in decisions about dissemination. Third, the external environment—notably that in the universities--has been perceived as probably hostile and possibly untrustworthy. As a result, the information generated by mission-oriented programs has tended to remain confined to a small circle of participants. Finally, the U.S. government moved somewhat earlier than its counterparts in France and the United Kingdom to encourage commercialization of the results of government-financed R&D. The National Aeronautics and Space Administration and a few other federal agencies have long had specified units concerned with technology transfer. Regarding government-financed R&D in the private sector, the 1980 Patent Law Amendments Act established a uniform policy allowing contractors—notably, small businesses, universities, and nonprofit laboratories—to own inventions resulting from federal R&D funding. The assurance this act provides of clear tide to government-funded inventions has greatly facilitated patent licensing by universities and other federal contractors to industry and has encouraged industrial participation in federally supported university research.

DOES TECHNOLOGY POLICY MATTER? 200 TABLE 4 Research Scientists and Engineers in the Labor Force, 1981 Country Number per 1,000 of Labor Force United States 6.2 Japan 4 Federal Republic of Germany 4.7 United Kingdom 3.9 Norway 3.8 France 3.6 SOURCE: Organization for Economic Cooperation and Development. Differences in the Environment Economic interests in the United States therefore have greater direct or indirect access to whatever may be transferable in the outcomes of mission- oriented programs. At the same time they are well placed to exploit these results for commercial purposes and have substantial incentives to do so. Lower Degree of Crowding Out The sheer size of the U.S. scientific and technological system means that mission-oriented programs probably "crowd out" other research efforts only to a limited extent. The size differential is particularly marked in terms of the stock and flow of research manpower. The share of R&D scientists and engineers in the U.S. labor force is one-third greater than that in the United Kingdom and France (Table 4). The share of secondary students going on to university training in the United States is about double that in France or the United Kingdom (Table 5), and the proportion of those students choosing scientific or engineering training is reasonably responsive to market circumstances.12 To this difference in endowment must be added the effect of inflows of scientists and engineers from overseas. In 1982, foreign-born scientists and engineers accounted for fully 17 percent of all scientists and engineers employed in the United States. Accessibility and Mobility of Scientific Know-how The U.S. stock of human and technological capital, in addition to being relatively abundant is also more easily accessible. It is, in the first place, accessible through contract research, both with private research firms and with universities. Though the share of university research financed by industry in the United States is not high, the links between universities and industry have traditionally been strong (Noble, 1977; Ben-David, 1968)—far stronger, at least, than in France or the United Kingdom, both these countries lagging even by

DOES TECHNOLOGY POLICY MATTER? 201 European standards in this respect (Ahlström, 1982; Organization for Economic Cooperation and Development, 1984a; Ben-David, 1968). These links take several forms: active efforts by U.S. universities to commercialize their technological skills, widespread consulting for industry by university scientists and engineers, frequent coauthorship of journal articles by researchers in industry and academia, and sizeable gifts of equipment by industry to university research facilities. The operation of the U.S. labor market also promotes the accessibility of its stock of human and technological capital. In general, the U.S. labor force is more mobile between employers and regions than the labor force in Europe: Average job tenure is about 20 percent lower in the United States than in France or the United Kingdom; the share of the labor force crossing regional boundaries each year is—at about 3 percent—double that in Europe. Moreover, U.S. scientists and engineers are almost as mobile as other segments of the labor force: Their average job tenure is only about 15 percent higher than the average. In contrast, mean tenure in France with a given employer is nearly 40 percent higher for highly qualified staff than for the labor force as a whole (Pham-Khac and Pigelet, 1979; Stevens, 1986). Differences in labor mobility are even greater regarding movement from university to industry. Some 2 to 3 percent of all U.S. scientists and engineers move from academia to industry or vice versa every year; the figure for Frantic can be estimated at well below 0.5 percent.13 The civil service status of public sector researchers in France makes movement difficult and eliminates incentives to move. TABLE 5 Diplomas Giving Access to Higher Education as Proportion of Age Group Country (Year) Percent Japan (1981) 87 Sweden (I 982) 82 United States (1980) 72 Federal Republic of Germany (1982) 26 Denmark (1980) 25 France (1983) 28 Italy (1981) 39 United Kingdom (1981) 26 Finland (1980) 38 Austria (1978) 13 Netherlands (1981) 44 SOURCE: Organization for Economic Cooperation and Development.

DOES TECHNOLOGY POLICY MATTER? 202 Competition in Factor and Product Markets High levels of mobility of scientists and engineers in the United States ensure that technological capabilities generated by mission-oriented research are rapidly diffused among firms but do not ensure that such capabilities will rapidly be exploited. This in turn hinges on the intensity of competition in product markets, which encourages rums to innovate. Three factors distinguish the United States in this respect: the receptiveness of capital markets to innovation efforts, the extent of the threat of new firm entry, and the incentives to innovation arising from a large and unified market. Capital markets in the United States are distinguished from those elsewhere largely by two features: the depth and breadth of equity markets and the availability of venture capital finance for start-up companies (Gönenç, 1986). It can be argued whether these institutions have proved appropriate for financing long-term market share strategies; but—perhaps because they provide a low-cost means for realizing capital gains—they appear to do reasonably well at providing concurrent finance for a broad range of innovation efforts. Certainly the balance of evidence indicates that they are effective mechanisms for the monitoring and diversification of innovation-related risks and opportunities. The functioning of capital markets reinforces the degree of competition in U.S. product markets in two important respects. First, the widespread availability of venture capital—together with a range of other environmental factors that reduce the costs of setting up and dissolving businesses-increases the threat of entry by new companies. This is reflected in rates of creation and disappearance of new manufacturing firms, which are nearly twice those in France (Arocena, 1983; Ergas, 1984b). Ideas not exploited by large companies are likely to be tried out quickly by an entrepreneur. This is of particular importance in the early stages of a new technology, when a large number of alternative design approaches are being explored (Clark, 1985; Freeman, 1974; Nelson and Winter, 1982). Second, an active market for corporate control provides an effective means of liquidating new firms that do poorly and incorporating into larger concerns the activities of those that do well. At the same time, the takeover market reduces the risks associated with entry by diversification. Large U.S. firms have tended to enter new markets by buying smaller firms already operating in those markets, knowing that if the venture failed, it could be disposed of (Scherer and Ravenscraft, 1984). The effects of potential competition are compounded by the far greater supply in the United States of potential entrants into advanced technology markets. More than 15,000 firms in the United States have R&D laboratories; this compares with about 1,500 in France and 800 in the United Kingdom. The number of firms with some technological capability in any given area is likely to reflect this differential. This provides the United

DOES TECHNOLOGY POLICY MATTER? 203 States with a large seedbed capable of responding quickly to the ''focusing'' effects of innovations and acting as an incubator for potential entrepreneurs. It also provides a large number of firms capable of acting as a "fast second"--- moving into a new market as its attractiveness is established and as the appropriate technological approach becomes clear. Size of the U.S. Market The nature of competition in the U.S. market also intensifies firms' interests in new product areas, notably as a technology approaches the stage where mass marketing becomes essential. Three factors are of particular relevance. First, because of the importance of economies of scale in a relatively homogeneous market, firms vie for leadership in the transition to mass production and marketing.14 Second, reliance on de facto or proprietary standards provides the firm whose product emerges as a dominant design with a considerable advantage. Third, the U.S. market appears to be highly sensitive to "perceptual" product differentiation, which tends to favor early entrants to the mass marketing and production stage.15 Each of these factors can create first-mover advantages, compounding the benefits the United States derives from having a greater number of potential first-movers. As a result, the two basic components of the "swarming" process— by which firms flock to an emerging market—tend to operate particularly rapidly in the United States: the experimentation stage, in which a range of alternative design approaches is explored, frequently by smaller firms; and the transition to mass commercialization, as the technology matures to the point of market acceptability. Preeminence in both of these stages increases the likelihood that U.S. firms will be well placed to spot an emerging dominant design. The Link to Performance The preceding discussion of mission-oriented countries can be summarized as follows. In the United Kingdom, mission-oriented research has tended to yield few direct benefits while possibly crowding out a substantial share of commercial R&D. The indirect spin-offs have been low, creating a "sheltered workshop" type of economy: a small number of more or less directly subsidized high-technology firms, heavily dependent on and oriented to public procurement, and a traditional sector that draws little benefit from the high overall level of expenditure on R&D.16 In France, mission-oriented research efforts have themselves been reasonably successful. This has created export markets for France, notably in the largest weapons-importing countries of the Third World and in other countries where state-to-state trading is important. However, the spin-offs from these efforts have been relatively limited, so that French industry

DOES TECHNOLOGY POLICY MATTER? 204 has become increasingly dualistic in its access to, and reliance on, advanced technology. This has been most visible in France's shifting pattern of international trade. Exports of products requiring a high intensity of skills, though rising, have concentrated to a growing extent on Third World markets, reflecting the predominance of state-to-state trading, whereas in trade with the advanced countries, the relative skill intensity of French exports has tended to diminish. The centralized and concentrated nature of mission-oriented research has therefore led to an increasingly polarized pattern of specialization.17 The situation of the United States is more complex. Although the direct effectiveness of mission-oriented programs is no higher than in France, the results of these programs tend to diffuse particularly rapidly through the U.S. economy. This rapid diffusion is a result of three features: the wide range of economic interests capable of exploiting these results for commercial purposes, the low level of the obstacles they encounter in seeking to do so, and the strength of the incentives for rapid exploitation. The mission-oriented stage of research in the United States remains highly centralized, but its results are more rapidly carried over into the decentralized experimentation of the commercial market. Particularly in recent years, the rapid carryover of the results of the U.S. programs has generated advantages that may be cumulative at the level of the firm but are not cumulative at the level of the product. More specifically, although U.S. firms appear to retain many of their established strengths, U.S. production sites have proved considerably better at the experimentation stage than the follow-on to mass production (Lipsey and Kravis, 1985).18 This partly reflects the macroeconomic circumstances associated with the overvaluation of the dollar, but more fundamental factors may also be at work. Historically, the United States has lacked a system for training craftsmen, while possessing an abundance of higher-skilled (white collar) and lower- skilled or unskilled workers (Floud, 1984; National Manpower Council, 1954; Floud, 1984). At the same time, the structure of blue-collar earnings in the unionized parts of U.S. industry (with low differentials between trainee wages and those of craftsmen) and high labor mobility have discouraged employer investment in transferable skills (Glover, 1974; Mitchell, 1977; Ryan, 1984). Combined with a large and unified national market, this pushed U.S. manufacturing firms in two directions: pioneering mass production techniques that made little use of craft labor and developing organizational innovations intensive in their use of managerial or supervisory staff--such as multiplant production, multidivisional management, and the multinational firm. The advantage that superior mass production techniques gave U.S.

DOES TECHNOLOGY POLICY MATTER? 205 production sites has tended to erode over time, for at least three reasons. First, in an increasingly integrated world economy, being located in the world's largest single market is of diminishing importance as a determinant of competitiveness. Second, the quality of the .U.S. labor force—and particularly that part with only a high school degree or less—has probably declined relative to that overseas, and notably relative to that in Japan (Murray, 1984, pp. 96-112). Third, classical mass production techniques along "Taylorist" lines may be of diminishing effectiveness as the variability and differentiation of products increases, as product workmanship becomes a more important factor in consumer choice, and as new technologies for "mid-scale" production become available Ergas, 1984a). These factors place the U.S. manufacturing industry at a clear disadvantage, but they have less impact, if any, on the service sector. As a result, U.S. firms tend to reap the advantages of innovative capabilities in manufacturing mainly at the early stages of the product life cycle (or, if the dollar is low enough, in products that are mature). In services, the gains from innovation have been consolidated further downstream as markets grow. Given a reasonably flexible and open economy, this pattern is reflected in the structure of trade. Thus, resources have tended to cluster around emerging or science- based industries. In this sense, the United States comes closest to the classical product cycle model, abandoning mature industries in favor of activities with better growth prospects.19 A system of mission-oriented research, which helps ensure that the frontiers of these activities are constantly being explored, may provide a useful source of ongoing stimulus to this process. It therefore has a certain degree of coherence relative to the U.S. economy. Whether this process would not occur of its own volition—that is, even in the absence of mission-oriented research— remains an open question. THE DIFFUSION-ORIENTED COUNTRIES Diffusion-oriented policies seek to provide a broadly based capacity for adjusting to technological change throughout the industrial structure. They axe characteristic of open economies where small and medium-size manufacturing enterprises remain an important economic and political force and where the state, bearing the interests of these firms in mind, aims at facilitating change rather than directing it.20 The primary feature of these policies is decentralization. Specific technological objectives are rarely set at a central level. Central government agencies play a limited role in implementation, preferring to delegate this stage either to industry associations or to cooperative research organizations dominated by industry. Whatever funds are disbursed tend to be fairly

DOES TECHNOLOGY POLICY MATTER? 206 widely spread across firms and industries, with the high-technology industry obtaining a far lower share than in the mission-oriented countries. Given this degree of decentralization, the precise boundaries of technology policy are often difficult to identify. Switzerland, for example, would certainly deny having a "technology policy" in the sense in which France has one. A more fruitful approach is to view technology policy in these countries as an intrinsic part of the provision of innovation-related public goods: notably education, product standardization, and cooperative research. These countries' distinguishing feature is the importance they attach to the organization and high quality of the provision of these goods and the decentralized mechanisms they have developed for supplying them. The Economic and Institutional Framework The priority accorded to the provision of public goods has its origin in process of industrialization in these countries. Two interrelated features distinguished this process: an emphasis on "education push," notably through innovations in higher education and in the training of engineers (Ahlström, 1982), and an early specialization in the chemical and electrical industries on the one hand and in mechanical engineering on the other.21 This early pattern of specialization fed back into the demand for innovation-related public goods. The chemicals and electrical industries were distinguished from the start by the closeness of their links to the science base (Beer, 1959; Freeman, 1974; Liebenau, 1985; Rosenberg, 1976; Rosenberg and Birdzell, 1986). They needed a high-quality university system, capable of training scientists for industry, of monitoring scientific developments worldwide, of providing external support to the emerging industrial research laboratories. Achieving this system in turn depended on developing an increasingly efficient and effective school system, which could prepare and select candidates for higher education. The Lutheran tradition of universal literacy and broadly based instruction provided an ideal basis for this evolution (Sandberg, 1979). The chemicals and electrical industries therefore acted as a politically powerful and well-organized lobby for education and for academic research. Being highly concentrated and largely cartellized, they were fully capable of mobilizing in their collective interest (Forman, 1974; Schröder-Gudehus, 1972). But the needs of the mechanical engineering industries were different. First, whereas chemicals and electricals were science-based, mechanical engineering relied on learning-by-doing and on the tacit, unformalized know-how of skilled craftsmen. Second, whereas chemicals and electricals tended to be concentrated, mechanical engi

DOES TECHNOLOGY POLICY MATTER? 207 neering was not, mainly because a high level of decentralization was more efficient in monitoring the type of team production required to maintain the quality of workmanship. For decentralization to persist, the engineering industry had to resolve three major problems. First, it had to be able to draw on an external pool of skilled labor, since no single small or medium-sized firm could efficiently rely on its internal labor market alone. Second, it had to reduce the transaction costs involved in the decentralized production of components that are close complements from an economic viewpoint— e.g., nuts and bolts. Third, it had to find ways of keeping firms up to date with technological developments, ensuring that the fruits of technical advance accumulated and were appropriated at the level of the industry as a whole, rather than primarily or solely at the level of the firm. Mechanical engineering was therefore an active lobby for three policies: comprehensive vocational education, product standardization, and cooperative research. It sought these policies mainly through provision by industry associations rather than by government; and, particularly in Germany and Switzerland, this coincided with a governmental practice of according quasi- public status and functions to private bodies, originally to regulate markets (Berger, 1981; Katzenstein, 1985a). As it has evolved in these countries, the overall system of public policy affecting technological capabilities has therefore had three key features. Vocational Education The most significant feature is probably the depth and breadth of investment in human capital, centering on the dual system of education. This involves comprehensive secondary education based on streaming into a high- quality university system that is paralleled by an extensive system of vocational education.22 A distinguishing characteristic of the educational component of the system in diffusion-oriented countries is high retention rates. More than 85 percent of 17-year-olds are in the education and training system in these countries; this compares with around 60 percent in the United Kingdom and 70 percent in France. The system is characterized further by a relatively high level of per capita expenditure on education at all levels. Over the last decade, the elasticity of total public educational expenditure with respect to gross domestic product (GDP) has been around 5 times higher in Switzerland than in the United States, starting from a base where Swiss expenditure per pupil was already a higher share of per capita GDP. Finally, the system is notable for its far-reaching certification. Only some 10-15 percent of the age cohort leave school with no certificate or qual

DOES TECHNOLOGY POLICY MATTER? 208 ification whatsoever, compared with 20 percent in the United States and as much as 40 percent in France and the United Kingdom. Particularly in the German-speaking countries, the skill certification of large parts of the youth cohort occurs through the system of apprenticeship- based vocational education. More than 50 percent of 17-year-olds in Germany and Switzerland are enrolled in apprenticeships, compared with about 10 percent in France and the United Kingdom. These high rates of participation are encouraged both by a substantial differential between trainee wages and those of skilled craftsmen (Jones and Holenstein, 1983) and by a well-organized and extensive system for training apprentices. Thus, apprenticeships are highly structured programs of several years' duration. They include a combination of enterprise training and college education and culminate in standardized formal examinations. Moreover, completion of apprenticeships is only one stage in skill training: The classification of examination-certified vocational skins forms a continuum from the craftsman to the most highly trained engineer, and movement along this continuum is a relatively standard feature of working life.23 There is a high level of industry involvement throughout this system. In the general education sector, the main links are between industries and universities (these will be discussed below). But the core of industry involvement is in vocational education. The apprenticeship system is jointly financed and controlled by employers (acting mainly through industry associations) and local education authorities, with trade unions also providing an important input. Industry associations play a major role in defining and revising curricula and in monitoring the system's effectiveness. Combined with the emphasis on formal, written examinations, this ensures that the skins acquired are highly transferable between employers and can be adapted to improvements in the industry's technology base. Overall, this structure of investment in human capital yields two outcomes: a university system capable of keeping up with the frontiers of · science, though not necessarily pioneering their exploration, and a very high level of intermediate skills in the working population.24 The fact that these skills are certified through a standardized system of examinations erodes the advantages that internal labor markets would otherwise have had in information about individual workers' skins, and hence tends to favor smaller firms. In turn, the ongoing nature of certification encourages relatively high levels of mobility for skirted craftsmen with work experience, providing a further channel for the interfirm diffusion of technology (Glover and Lawrence, 1976; Maurice et al., 1982; Office Fédéral de I'Industrie, des Arts et Mötiers et du Travail, 1980).

DOES TECHNOLOGY POLICY MATTER? 209 Industrial Standards An emphasis on reducing transactions costs also pervades the second important feature of diffusion-oriented countries, namely, the system of industrial standardization. Of particular importance to the engineering industries, the German system of industrial standardization is unique in the range of intermediate goods and components it covers, the volume of detail it specifies (notably in relation to performance), and the legal status of its norms. This system emerged as part of a conscious effort to promote rationalization in decentralized industries.25 Though it operates as a quasi-public authority, the system is almost entirely funded and administered by industries. Although the budget of the German standards operation (DIN) is 21/2 times that of its French counterpart (AFNOR), the share of this budget provided by all levels of government is less than half that in France.26 To this must be added the considerable investment German industry makes in providing technical support for the standardization process. The immediate impact of the standardization system is to reduce transactions costs by providing clearly specified interface requirements for products. At the same time, it fulfills a quality certification function, which is especially important for industrial components. But its indirect effects may be even greater. In particular, the standardization process itself—and notably the preparation of new standards and the ongoing review of existing ones— provides an important forum for the exchange of technical information both within each industry and with its users and suppliers. Though this information is ultimately rendered public in the published specifications, the long lead times involved in drafting standards, and the relatively small share of the total information generated that is contained in the published standard, ensure that the exchange process operates as a local public good. The primary beneficiaries are the firms most actively involved in industry associations. The density of these information flows also ensures that by the time a new standard is announced, German firms are in a position to adopt it. The system of industrial standardization, in other words, functions as a means of placing ongoing pressure on firms to upgrade their products, while providing them with the technical information required to do so. Cooperative Research and Development A concern with assisting a decentralized industrial structure to adjust to changing technologies also underlies the third feature of these countries' policies, namely, the role of cooperative R&D.27 This takes two forms.

DOES TECHNOLOGY POLICY MATTER? 210 The first is close industry-university links, which have traditionally been of particular importance to the chemical industry and remain a dominant characteristic in Germany, Switzerland, and Sweden. Thus, 15 percent of university research in Switzerland is funded by industry—the highest share in the OECD and more than 3 times higher than in the United States, France, or the United Kingdom. The links go well beyond the chemical sector, as the close ties between the EFTZ in Zurich and the Swiss mechanical and electrical engineering industries attest. Similar links can be found in Sweden, notably between the technical universities and the large science-based firms. A specific feature of the German system is the role of the three large nonprofit research organizations in cooperative research. The Fraunhofer Gesellschaft, in particular, has 22 research centers, which have become increasingly involved in providing technical support to small and medium-size firms. The second major form of cooperation in R&D centers is industrywide cooperative research laboratories. These account for a considerably higher share of total R&D expenditure in Scandinavia than elsewhere. Thus, in Norway even the largest firms have only small in-house research units, and most industrial R&D is contracted out to cooperative laboratories. In Sweden an extensive network of industry or technology-specific laboratories is jointly funded by industrial firms and by the State Board for Technical Development. In addition to ongoing programs aimed at the entire population of an industry, these laboratories carry out contract research for individual firms. Similar arrangements exist in Germany and (though on a smaller scale and with considerably less government funding) for certain industries in Switzerland. The most immediate impact of the availability of these outside sources of research expertise is probably on the cost-effectiveness of R&D. They permit sharing of costly instrumentation and research facilities and allow firms occasionally to draw on specialists they could not afford to employ full-time. In this sense, their role is similar to that played by the larger U.S. technical consultancies (for example, Arthur D. Little or Battelle Laboratories) in providing support to smaller laboratories. This role may be secondary over the longer term, however, as it can be argued that the important function of cooperative research is really twofold. The first is technology transfer. Universities and cooperative research centers inevitably have a higher ratio of research to development than have the laboratories of small firms. This higher research intensity allows them to generalize, and hence transfer, the results of individual development projects from firm to firm, thus providing a degree of economies of scope to innovation programs across an industry or activity. The second function of cooperative research is technology focusing.

DOES TECHNOLOGY POLICY MATTER? 211 The process of setting research priorities for the system encourages firms to pool their perceptions of major technological threats and opportunities. This in turn feeds back into the internal R&D planning. However, the effective discharge of these functions requires that firms have a certain degree of in-house R&D capability, which they complement through recourse to external sources. Thus, the evidence for Germany suggests that the most intensive users of contract research are small and medium-size firms with an internal research unit—on average, these firms spend on external (contract) research an mount equivalent to 30 percent of their in-house R&D spending. The Role of Policy: An Example It has therefore been a major concern of policymakers, particularly in Germany, to ensure the existence of an in-house R&D capability to complement other forms of R&D. The Federal Ministry of Economics has in recent years helped finance a scheme providing a partial subsidy for the employment of research scientists and engineers in small and medium-size firms. Assessments suggest that the program has been a considerable success and that about 10 percent of the eligible firms participate. The scheme is worth examining because it provides a particularly good example of German diffusion-oriented policies and notably of what are referred to as indirect specific programs. The latter are government programs specific to the technology of a particular industry but implemented through a trade or industry association rather than by a government department. Three features of these programs stand out.28 The first is that the funds involved are small. In total, in 1985 expenditure on the R&D employment subsidy was around 420 million DM—less than 1 percent of German expenditure on R&D. Moreover, the funds were thinly spread, going to about 7,000 firms, a third of which have fewer than 50 employees. The second is the decentralized process of implementation. The major responsibility for administering the project lies not with the funding agency, but with the German Federation of Industrial Research Associations (AIF), which groups some 90 nonprofit industrial R&D associations, which in turn represent 25,000 firms in 32 industrial sectors. The AIF—70 percent of whose funds come from industry—operates some 60 research laboratories, employing 4,000 scientists and engineers. Though the AIF has operating responsibility for the project, a low level of discretionary decision making is involved. Eligibility criteria are clearly set out, and question of whether a firm is eligible is straightforward. The risks of discrimination against particular firms are therefore low. However,

DOES TECHNOLOGY POLICY MATTER? 212 being administered by the AIF provides the scheme with high visibility among industrial associations, and more than 50 percent of the firms participating in the scheme learned about it from trade associations or local Chambers of Industry and Commerce. Decentralized implementation is closely related to the third feature of the scheme, namely, the simplicity of its administrative formalities. The application forms do not call for any particular expertise—90 percent of participants completed these forms without any external assistance. This limits the fixed costs involved in participating and further reduces the risks that the program will degenerate into a privileged club. Defense Research and Development The importance accorded to the diffusion of technological skills has even affected these countries' not insignificant activities in armaments. Sweden has placed great emphasis on promoting and to some extent organizing the diffusion of defense-related technological skills into the commercial sector. By law, no Swedish company may have more than 25 percent of its business in defense. Thus, defense contractors are forced to develop civilian operations (Gansler, 1980, pp. 245-257). Specific policies have also been implemented to increase the technical capabilities of subcontractors to the larger companies involved in defense work. Financing is provided by the Swedish Industrial Development Fund. The Effectiveness of the System The diffusion-oriented countries are therefore characterized by policies that encourage widespread access to technical expertise and reduce the costs that small and medium-size firms face in adjusting to change. In essence, the policy framework serves to increase the capacity for absorbing incremental change without threatening the basic structure of industry. From this point of view, the policies have indeed been successful. It remains a striking feature of these countries that industrial production is more decentralized than it is elsewhere, notably in mechanical engineering; and that, while providing the benefits of highly focused management, decentralization does not prevent coordination of interdependent decisions and the reaping of economies of scale and scope. Though firms in these countries are smaller than their competitors overseas, higher levels of specialization minimize any relative cost disadvantage.29 The system has also functioned effectively in promoting adjustment to incremental change. New skills are transmitted relatively rapidly through labor training and re--g, as well as by interfirm labor mobility. The

DOES TECHNOLOGY POLICY MATTER? 213 standardization system itself provides an ongoing flow of technical information; and industry associations and cooperative research institutes allow for interfirm economies of scale in R&D while focusing firms' attention on emerging technologies. However, two major concerns have been expressed. First, the system as it has evolved is geared to the existing industries, which basically set the technology agenda: That is, they determine the direction of research, dominate the process of standardization, and have a large role in training and education policies. Entirely new industries and technologies may find it difficult to capture the attention they deserve. Second, even in the existing industries, the decentralized, ''bottom-up,'' approach leads to a strong emphasis on movement along technological trajectories, while reducing the visibility of, and preparedness for, major shifts in trajectories. These features--concentration on established industries and moving along set technological trajectories--are apparent in the evolution of these countries' external trade, which has been distinguished by three trends30 The first is that the diffusion-oriented countries have tended to consolidate and even sharpen their traditional patterns of specialization. They have indeed retrenched in the areas where their original performance was poor but without moving into entirely new areas of activity. Rather, their performance has remained strong in the areas where they have traditionally specialized, and within these product areas they have tended to become stronger across the board. As a result, their net exports are highly concentrated in "product clusters," mainly in products for which world demand is growing relatively slowly, so that improved performance has required a long-term gain in market share. Second, this gain in market share has occurred in products with unit values well above the average for their product category. For engineering products, around 85 percent of Swiss exports, 75 percent of German exports, and 65 percent of Swedish exports in 1970 had unit values above the average for their disaggregated product category; this compared with around 35 percent for France and the United Kingdom. Specialization in the higher-quality segments of markets has tended to increase over time. Third, and most recent, this pattern of specialization has been seriously threatened by competition from Japanese firms, which have used electronics- based technologies to challenge the European countries' traditional predominance in mechanical engineering. Lags in adjusting to shifts in technological trajectories have led to major losses of market share. These lags arise less from a lack of technological capabilities than from the conservatism inherent in industrywide decision-making processes. The Swiss watch industry and the German machine-tool industry provide striking examples in this respect.

DOES TECHNOLOGY POLICY MATTER? 214 In both cases, the research community associated with the industry was aware of the impact electronics would have—and, in fact, made important contributions to the technology. But research awareness could not be translated into industrial action—partly because of complacency among firms, but also because there were few prospects for adjusting without drastic changes in the industry structure. These changes could not be fitted into the consensus- centered decision-making process; both industries severely lost market share to their Japanese competitors. Once the loss in market share had begun to occur, however, the industries were relatively well placed to respond. The basic technological skills had been accumulated, and the mechanisms for transferring them to industry were in place. Particularly the German machine-tool industry—which benefited in the early 1980s from the effective devaluation of the DM relative to both the U.S. dollar and the yen—succeeded in reversing its loss of market share and making a quick though painful transition to the new technology. The criticism that the system slows adjustment to entirely new opportunities while reinforcing specialization in the traditional areas of activity may therefore have some foundation. As these cases bear out, however, the system's capabilities for adjustment—albeit delayed—should not be underestimated. An ongoing response to the Japanese challenge will require important changes in certain aspects of the institutional context. Thus, it has been argued that the apprenticeship system should provide a broader range of generic skills, which could be complemented through continuing vocational education. The Swedish educational reform, which has somewhat reduced the vocational component of secondary education, clearly goes in this direction (Hodenheimer, 1978). But given these changes, the diffusion-oriented countries should remain' important on the world industrial scene. JAPAN In this typology, Japan is in a class of its own. Like the countries in the first group, it has deployed coordinated efforts to advance national technological goals. At the same time, like the countries in the second group (and with a clear element of imitation from those countries), it has emphasized a broadly based capacity to diffuse innovation-related public goods. In both cases, however, the specific policies and their implementation have been modified to the requirements of the Japanese context. Two features of this context stand out. The first is that even in the recent past, Japan was at a far lower level of development than the other countries examined in this chapter (Nakamura, 1951; Shinshara, 1970). As late as

DOES TECHNOLOGY POLICY MATTER? 215 1965, Japanese GDP per capita was half the OECD average and less than one- third that in the United States. The gap in per capita GDP was closely linked to lower levels of capital and skill per unit of output throughout Japanese industry. This was accentuated by a dualistic industry structure that combined relatively high productivity in the large-firm sector with considerably lower productivity in the smaller manufacturing firms, in agriculture, and in services. More so than for the other countries in our sample, material well-being in Japan depended on whether Japanese industry could fundamentally reshape its comparative advantage in international trade, rather than simply adapt it to incremental advances in the technological base. The second factor that sets Japan apart is the relation of the state to industry (Johnson, 1982). Unlike the diffusion-oriented countries, Japan entered the 1950s with an economic bureaucracy able and willing to deploy an active strategy of industrial transformation. Compared to previous periods in its history, and even to the present day, this bureaucracy was at that time uniquely powerful relative to the other political actors on the national scene. However, particularly with the end of postwar reconstruction, the bureaucracy's power depended on its capacity to generate a consensus among the major actors, so that the "administrative guidance" it provided would be smoothly carried through into corporate decision making. This resulted in a combination of consensus-based but relatively centralized decision making with a more decentralized approach to implementation. The Development Strategy Combined, these factors have led the Japanese bureaucracy toward a development strategy that emphasizes the rapid upgrading and transformation of the nation's technological skills but does so in a manner both more decentralized and more broadly based than in the mission-oriented countries. The constraints arising from industrial dualism have led to a greater emphasis on diffusion, whereas those arising from the nature of governance have led to a greater emphasis on indirect implementation. There are three major elements to this strategy: investment in human capital, promoting activities at the "leading edge" relative to the core sector's technological capabilities, and facilitating the transfer of new technologies from the core to the periphery. Human Capital A key component of the strategy has been the progressive upgrading of Japan's base of human capital. Japan has a long tradition of engineering

DOES TECHNOLOGY POLICY MATTER? 216 training, having been among the first countries to integrate engineering into university curricula (Nakayama, 1984). But its emergence as one of the world's leading centers (at least in numerical terms) for the training of engineers is nonetheless spectacular (Table 6). This has been paralleled by a sustained increase in the average educational attainment of successive cohorts. The emphasis on the upgrading of human capital is, in many respects, reminiscent of German, Swedish, or Swiss industrialization. But, in contrast to' these countries, the expansion of the Japanese skill base has occurred on a more general and less industry-specific basis (Stevens, 1986). More particularly, the Japanese education system is one of general rather than vocational education. The growth in enrollments has consisted in increasing the share of the cohort remaining in the general stream, gradually bringing this share toward U.S. levels (Table 7). Even in postsecondary education, the level of specialization is low, and engineering training in Japan is considerably more superficial than that in Scandinavia or the German-speaking countries. As a result, the tasks of directing the labor force toward specific occupations and developing the relevant skills has largely been left to industry, and particularly to the larger, "lifetime employment" firms.31 Finns have had access to a progressively better-educated flow of labor force entrants, notably as regards general mathematical and engineering skills, but little attempt has been made in the education system to shape the capacities of students toward particular vocations. This has given the Japanese labor force a high degree of malleability, decentralizing a set of decisions that critically affect a country's technological capability. Sectoral Promotion A high degree of decentralization has also characterized the promotion of particular industries.32 Three features are important in this respect. First, the areas being promoted have generally been fairly loosely defined, cov TABLE 6 Higher Education Engineering Qualifications Country First- Per Million Below Per Million (Year) Degree Population First- Population Level Degree Level Federal 7,000 110 16,000 260 Republic of Germany (1981) United States 80,000 350 — — (1982) Japan (1982) 74,000 630 18,000 150 SOURCE: National Economic Development Office (U.K.) and Manpower Services Commission (U.K.).

DOES TECHNOLOGY POLICY MATTER? 217 ering a broad range of market segments rather than focusing on a particular product. Second, the policies adopted have primarily provided a framework within which the activity could develop, notably through import protection, restrictions on foreign direct investment, assistance in licensing overseas technology, and measures aimed at reducing the entry barriers to domestic firms. Beyond providing this framework, policies have rarely involved promotion of a particular domestic company to the apparent detriment of others. Little use has been made of "national champions," and efforts have consistently been made to diversify risk by promoting competition in the domestic market. Third, direct financial assistance has played a very limited role. Though "soft loans" have at times been important, the primary emphasis has been on nondiscretionary instruments such as tax expenditures. Whereas the volume of these tax expenditures may in certain cases have been large relative to the size of the activity being promoted, the subsidies involved have probably been small, partly because of the generally low incidence of corporate taxation. There has, of course, been a subsidy element in public procurement, but given its low defense expenditure, the Japanese government has been a relatively marginal consumer of high-technology equipment (though this is not true in some of the areas discussed below, notably telecommunications and aerospace). TABLE 7 Distribution of Students in Upper Secondary Education (Full-time and Part-time Enrollments), Around 1980-1982 Country General Education Vocational and (percent) Technical (percent) Japan 70 30 United States 76 24 Federal Republic of 21 79 Germany France 40 60 Italy 34 65 Netherlands 40 60 United Kingdom 57 43 Switzerland 25 75 Austria 17 83 Belgium 44 56 Denmark 37 63 Finland 50 50 Sweden 30 70 SOURCE: Organization for Economic Cooperation and Development. Technology Transfer The technology transfer policy itself is highly decentralized in Japan, both in implementation and funding (Ergas, 1984b, p. 22). The core of

DOES TECHNOLOGY POLICY MATTER? 218 this policy is the network of prefectural laboratories whose primary function is to provide technical assistance in developing or adapting new technologies, notably to small and medium-size firms. Central government finances haft the laboratories' capital equipment costs, and regional authorities and firms themselves provide the rest of the laboratories' income. There are now 195 regional laboratories in Japan: The 47 prefectures average four laboratories each, and each prefecture has at least one. Some laboratories axe "problem-oriented" (for example, around textiles, food, ceramics, paper, leather, metals) and are linked to the various regional industries and located accordingly. The others (about 30 percent) are more broadly based and multidisciplinary. Generally, they operate in three or four complementary fields (e.g., mechanical engineering, metals, woodworking). The 195 laboratories employ more than 5,000 research technicians and engineers. They are connected with central government laboratories, which provide high-level expertise and sophisticated equipment for R&D when needed. Moreover, the staff of the prefectural laboratories are systematically retrained by the central government to keep them abreast of the latest developments in science and technology. However, the laboratories' activities are determined mainly by their local clients. Effectiveness Given the degree of decentralization, notably in the implementation stage, the overall effectiveness of the system has mainly been due to industry's strong response to the opportunities signaled. In part, this response has reflected the high legitimacy the economic bureaucracy enjoys in Japan, so that the advice it provides is taken considerably more seriously by industry than is similar advice in other countries. This legitimacy is reinforced by the fact that a consensus of views with industry is reached well before policies are announced. However, the strength of Japanese industry's response has also been due to a set of factors that have increased the benefits and reduced the costs of exploiting new opportunities. The first, most obvious, and in some respects most pervasive of these factors is the favorable macroeconomic context. An economy where savings and investment are abundant and where consumer demand is rapidly increasing and shifting to progressively higher-quality goods provides a supportive framework for technological upgrading. This macroeconomic environment reinforces a second factor contributing to rapid adjustment, notably the low levels of social resistance to change. In addition to steady growth in employment, resistance to change has also been weakened by the lack of strong industry lobbying for declining manufacturing sectors, by the assurances of retraining provided

DOES TECHNOLOGY POLICY MATTER? 219 within the lifetime employment system of large firms, and by the "sunset" policies adopted by the Ministry of International Trade Industry (MITI) to ease the process of decline in industries—such as textiles, shipbuilding, or most recently, aluminum—that have lost their competitiveness (Launer and Ochel, 1985). These factors have also made firms less reluctant to enter new areas, since they know they will be able to withdraw if the opportunities prove ephemeral. The Role of Competition A final factor accelerating the response to new opportunities is intense rivalry between the large industrial groups. This rivalry--reflected in far- reaching price competition, in investment "races," and in competition in R&D— is accentuated by several features of the Japanese industrial environment.33 The rapid growth of demand--and the perception that growth will continue —has made oligopolistic coordination difficult, while focusing firms' attention on long-term market share rather than short-term profitability. The low cost of funds has reinforced the tendency to take a long view in investment decisions, notably by reducing the implicit discount rate for capital budgeting decisions.34 The strategy and structure of Japanese industry also tend to increase the importance of first-mover advantages, so that once a new area emerges, competition to be an early participant is intense (Kono, 1984). Although first- mover advantages in the United States are probably concentrated in the mass marketing stage, production cost factors appear to be more important in Japan. Operating with a fairly fine division of labor relative to smaller enterprises, the major Japanese firms specialize in large-scale fabrication and in mass assembly. These operations are characterized by substantial static and especially dynanic economies of scale. As a result, a large firm's unit costs are highly sensitive both to the rated capacity of its plant and to accumulated production. Given these characteristics, and especially in a rapidly growing market, the penalties of late entry are likely to exceed the costs of building ahead of demand. Market entry and capacity expansion therefore tend to occur quickly as firms seek footholds in new areas of activity.35 These pressures are accelerated by each major firm's reliance on a reasonably stable group of smaller suppliers (Imai and Itami, 1981). Unlike the situation in the United States, a late entrant in Japan cannot reduce its costs by acquiring a firm already established and experienced in the business. Moreover, its competitive disadvantage can be aggravated by the fact that its more or less fixed circle of suppliers will also lack experience

DOES TECHNOLOGY POLICY MATTER? 220 in the new area. Entering a new market early provides an insurance that the firm will not be severely handicapped should the market prove particularly promising. Lifetime Employment The lifetime employment system itself creates strong pressures for large rums to enter emerging markets. Finns committed to lifetime employment seek to diversify their portfolio of activities to cover different stages of the product life cycle, so as to stabilize employment requirements over time. The search for new areas of activities is likely to be a particularly high priority for younger professional staff, given the impact it will have on their career prospects. The lack of interfirm mobility of managerial staff (who constitute the bulk of the personnel covered by the lifetime employment system),36 and the consequent need to ensure a sufficient growth to keep internal planning resources fully employed, creates insistent pressures for diversification. However, since diversification must rely on internal expertise, it is largely confined to areas related or similar to the firm's principal activity. Japanese firms consequently tend to expand through related diversification, and the growth of conglomerates is extremely rare (Imai et al., 1984; Nonaka et al., 1983; Kono, 1984). This creates a system that breeds on itself. The drive for related diversification pushes firms' R&D efforts into adjacent areas. The fact that one firm is seen to do this propels other firms to do the same. Particularly when the technologies involved are generic, in the sense of spanning several product fields, the degree of interfirm competition for footholds in emerging product areas rapidly becomes intense (Suzuki, 1985). This increases the extent of experimentation in the Japanese market, providing a competitive advantage to the economy as a whole. Interfirm Cooperation The system involves a high degree of horizontal and vertical cooperation, mainly within each family of firms. The dual structure of Japanese industry is of obvious relevance in this respect. Three factors perpetuate this structure: the problems inherent in a system without well-developed equity markets; the high rigidities associated with internalizing activities into the larger firms, given the lifetime employment system; and an abundant supply of entrepreneurs. However, this structure could hardly survive without constant upgrading of technological capabilities in the secondary sector. This need is mainly

DOES TECHNOLOGY POLICY MATTER? 221 met by the direct technical assistance large firms supply to their smaller subcontractors (Gönenç 1984; Gönenç and Lecler, 1982), but the decentralized laboratory system discussed above also plays an important role, as do trade associations and the standardization system, both originally modeled on German lines. Another important type of cooperation among firms occurs in the context of precompetitive research, notably for the development of generic technologies. These research efforts, in particular those promoted by MITI, provide for investigatory research in the phases of R&D generating the highest content of information in the "public good." They are to some extent a substitute for university-industry links, which appear to be extremely weak in Japan. Whether they are effective in this respect is the subject of considerable controversy, but cooperative research does appear to have allowed Japanese firms to resolve some of the critical bottlenecks confronting them in the "generic technology" aspects of the industry they were entering: for example, the production of cathode ray tubes for color television receivers (Dore, 1983; Peck and Wilson, 1982). Centralized Programs These factors go a considerable way toward explaining the responsiveness of Japanese firms to the signals emerging from Japan's largely decentralized system of industrial planning. However, it would be foolish to deny that the Japanese bureaucracy has at times engaged in highly directive and centralized attempts to promote particular activities and that these attempts have involved a considerable mobilization of resources. Prominent examples are mainframe computers, central office telecommunications equipment, aerospace, and the Japanese railway plant. Policy in these areas has been similar to that in Europe, with the important difference that a greater number of competing firms have been present in each area. Despite this difference in policy design, the outcomes do not suggest a high level of policy effectiveness: Japan's output of videotape recorders far exceeds its production of computers; Japanese central office electronic switching systems have not emerged as major competitors on world markets; the Japanese bullet train is far less cost-effective than its French counterpart, the Train à Grande Vitesse; and aerospace remains a weak point in the Japanese industrial structure. Overall Impact Japan's policies have tended to be most successful when they combine three features: consensus decision making on broad goals, decentralized

DOES TECHNOLOGY POLICY MATTER? 222 implementation, and a reliance on the dynamics of competition to ensure a rapid response. Sustained by good macroeconomic management, a steady increase in human capital, and a willingness to adjust to change, this has proved to be a formidable engine of growth. In particular, it has allowed Japanese firms to modify their specialization in international trade into progressively more technologically advanced product areas (Boltho, 1975; Orléan, 1986). The striking feature of this shift is not only its breadth but its depth: Like the diffusion-oriented countries, Japan's export pattern is highly specialized, the bulk of net exports being concentrated on a small number of comodities. Unlike the diffusion-oriented countries, however, this pattern has shifted markedly over time, as the first wave of Japanese exports (mainly textiles) was replaced by a second (steel, shipbuilding), a third (automobiles), and now a fourth (electronics and machinery). Such drastic changes in international specialization have inevitably entailed major modifications in the structure of Japanese industry. The capacity of the Japanese industrial structure to carry out such shifts is what sets it apart from the other countries considered in this chapter. However, this does not imply that what has proved successful until now will remain so. Particular concern has been expressed in Japan about whether the system for promoting innovation is resilient.37 The central question in this respect is the system's continuing effectiveness once Japan arrives at the technological frontier. It is presumably easier to set broad goals in the catching- up stage of growth than in pushing beyond the state of the art. Moreover, the skills needed to implement these goals differ. Up to now, Japan has not suffered from the weakness of its scientific base, but it may prove vulnerable to a blurring of the boundaries between pure and applied research. SHIFTING AND DEEPENING: AN ATTEMPT AT SYNTHESIS Directions for Research In recent years, economists have made significant progress in analyzing technological advance as an evolutionary process—that is, a process of experimentation, selection, and diffusion.38 The work done provides a convenient analytical structure for synthesizing the arguments presented above and for examining their implications for overall economic performance. A central concern of recent analyses has been the mechanisms by which innovation shapes market structure, notably its impact on concentration and on the extent of the barriers to potential competition. The assumption

DOES TECHNOLOGY POLICY MATTER? 223 has been that this relation operates similarly from country to country; but the data presented above suggest that this is not the case. Rather the material reviewed suggests important differences between countries along three dimensions: • Who appropriates the gains from technological advantage? Is it the innovating firm alone or is it the firm and a broader group (for example, a firm's suppliers)? • To what extent are these gains cumulative and sustainable over time? Where does the process of skill accumulation occur--in the individual firm, in the industry, or in the industrial structure as a whole? • How much flexibility is there in responding to innovation? Does flexibility occur through adjustment by existing firms or through shifts in the firm population? The material reviewed also suggests that differences in each of these respects affect the evolution of each country's industrial structure. In essence, this relation operates through the balance between two (not necessarily alternative) ways of increasing the efficiency with which resources are used: shifting, or transferring resources from old to new uses; and deepening, or improving their productivity in existing uses. The greater the mobility of technical, managerial, and financial resources, the greater the contribution that shifting is likely to make to overall growth. Conversely, the greater the extent to which assets are firm-or industry-specific, the greater the importance of deepening to long-term competitiveness. This relation can be highlighted by reexamining four of the countries in our sample; these countries (the United States, France, Germany, and Japan) can be considered to be roughly representative, given the similarities between the United Kingdom and France, and between Switzerland, Sweden, and Germany. A broad characterization of the four countries is given in Table 8, which summarizes many elements of the discussion above and can be analyzed as follows. The United States can be considered paradigmatic of shifting. An extremely large applied research system, operating at the frontiers of technology, continuously generates potential new areas of commercial activity. Adjustment to these opportunities occurs through competition between firms on the open market for mobile technical and managerial skills and financial assets. The ease with which these resources can be bid out of existing uses discourages productivity-enhancing investments in skills and capabilities that are specific to a particular firm or activity, as such investment can be justified only through longer-term commitments. However, high mobility also ensures that entirely new areas of endeavor are rapidly exploited, first in the domestic market and then through world sales.

DOES TECHNOLOGY POLICY MATTER? 224 TABLE 8 Technology Systems and Industrial Structures Characteristic United France Germany Japan of System States Appropriation Firm State Finn and Industrial industry group Skill Labor Technocracy Industry and Large firm accumulation market research system Flexibility Mainly Determined Adaptation to High, but by entry through the incremental major and exit political change; low actors system intersectoral remain the flexibility same Industrial roduct Dualism Inherited ''Moving structure and cycle specialization clusters'' trade pattern In France, the transfer of resources to new activities does occur, but largely (though not solely) through major state-initiated programs aimed at both public and private markets. The technical elite, which is a more or less integral part of the state apparatus, is the essential repository of technological skills and plays the key role in designing and implementing programs. However, the concentration of power in this elite and the limited diffusion of skills and capabilities outside its area of activity has two consequences. First, the "shifting" is constrained to those pans of the economy directly affected by the large public programs. Second, the rest of the economy lacks the resources (and often the incentives) to "deepen" its competitive advantage. Germany, in contrast, is paradigmatic of deepening. Skills and resources are highly industry-specific, and their development follows paths largely charted by the industries themselves. Relations between firms, between firms and their employees, and between firms and the financial system have traditionally included long-term' commitments favoring investments in activity- specific capabilities. At the same time, high levels of education, industrial standardization, and cooperative research provide powerful mechanisms for diffusing capabilities throughout each industry, so that progress is made across a broad front. The pattern of industrial capabilities is largely inherited, yet it is constantly renewed by "doing what one has always done, but better." The distinctive feature of Japan is the extent to which it combines shifting with deepening. The key component is the large firm, closely linked to its main sources of finance and surrounded by a network of smaller suppliers. The large firm—and more generally the industrial group—seeks

DOES TECHNOLOGY POLICY MATTER? 225 to maximize the productivity with which resources are employed in existing uses. However, it also faces powerful incentives to shift its operations toward emerging areas of activity, bringing the entire industrial structure in its wake. Three factors are at work: first, the long-term nature of commitments permits productivity-enhancing investments in firm-specific skills; second, the intensity of competition between large firms encourages early entry into new markets; third, each large firm, as it moves, seeks to shift its suppliers with it. Implications for Overall Economic Performance But between what does one shift, and along what does one deepen? And what implications does the balance of shifting and deepening have for overall economic performance? The concept of a technological trajectory provides a helpful building block in exploring these questions. A technological trajectory can be defined as a path of technological development, drawing on a given set of basic scientific principles and propelled by an internal dynamic of improving performance with regard to a few key design criteria (Dosi, 1982; Nelson and Winter, 1982; Rosenberg, 1976). At the risk of considerable simplification, evolution along this path can be characterized as following an S-shaped curve (Figure 1): • The emergence phase includes experimentation among alternative design approaches, as attempts are made to identify approaches with the greatest promise for subsequent developments. • In the consolidation phase, the concentration of R&D on a few critical parameters, within the framework of a broadly set design approach, allows rapid improvement both in performance and in cost. • The maturity phase occurs as the most easily exploited opportunities have been fully used, while entirely new design approaches, possibly based on an area of applied science different from that of the original trajectory, emerge as substitutes in a growing range of uses. The development of vacuum tube technology illustrates these processes and their pattern of evolution over time (Baker, 1971; Maclaurin, 1949; Sturmes, 1958). After a phase of open experimentation, Lee de Forest's triode tube set an underlying structure for the workable amplification of small electrical signal voltages. Subsequent progress in tube technology, though yielding dramatic improvements in performance, concentrated on a few variables, such as the energy efficiency of the cathode, tube life and reliability, and automation of the manufacturing process. However, the development of solid-state semiconductor technology beginning in the

DOES TECHNOLOGY POLICY MATTER? 226 late 1940s dramatically cut across this path of improvement (Webbink, 1977). Transistor-board devices rapidly established themselves as a more reliable and space-saving alternative to the vacuum tube, with enormous potential for cost reduction through progressively larger-scale integration and automated manufacturing and testing. Figure 1 Technological trajectories. As the technology developed, so the structure of the industry changed. In the early days of the vacuum robe industry, the field was open to competition. With many differing approaches to tube design, manufacturing, and application, overall profitability in the industry was probably low, since the small number of "hits" was more than offset by high initial development costs and a large number of "misses" (de Forest himself suffering repeated bankruptcies). Profitability increased only after the basic technology had stabilized, and patents and proprietary know-how blockaded entry, weakening price competition, improving R&D focus, and allowing cost reduction as output grew. The industry's consolidation phase was dominated by a tight-knit oligopoly, including some of the largest, most technologically advanced firms of its day: General Electric, West

DOES TECHNOLOGY POLICY MATTER? 227 inghouse, RCA, and AT&T in the United States; Marconi, Siemens, and Philips in Europe. Large size and (for the time) huge R&D budgets did not allow these firms to transfer their dominance to the emerging market for solid-state devices. These drew on an applied science base quite different from that they had mastered over the years. However, the vacuum tube industry did not disappear, for four reasons: initial uncertainty about the capabilities of solid-state devices slowed substitution, the emergence of solid-state competition encouraged manufacturers to bring forward improvements in tubes, rapid growth occurred in applications where there were no practicable substitutes for tubes (notably television receivers), and new tubes were developed for applications requiring frequencies unsuitable to solid-state technologies. Substantial opportunities persisted in the industry 40 years after its technological base had been superseded, but these opportunities relied on a progressively narrower and more vulnerable base. Three broad conclusions can be drawn from this account: • The emergence of a technological trajectory is not usually associated with high rates of return on investment, given large R&D costs, the substantial risk of failure, and the intensity of competition. • It is in the consolidation phase that the greatest improvements are made in product cost and performance and the largest scope exists for supranormal profits. • As improvements in critical parameters become more difficult to achieve, the maturity phase creates new challenges for the industry, with the development of substitute products intensifying competition and increasing the importance of capturing the least vulnerable niches. Clearly these. conclusions do not have the force of laws, nor can one indiscriminately generalize from the level of individual industries to that of national industrial structures.39 Nonetheless, they suggest several hypotheses of interest: • The performance of an industrial structure specializing in the emerging phase is likely to depend first on its capacity to experiment on a broad front, thus increasing the .probability of success. ,am important factor in this respect is proximity to a pool of sophisticated customers, who can rapidly distinguish promising from less promising alternatives. Second, performance will depend on the extent to which the industrial structure can carry successes over from the emergence to the consolidation phase. However, there is no a priori reason to expect such an industrial structure to show a high rate of growth of real incomes or productivity, at least as conventionally measured (Ergas, 1979). • Conversely, an industrial structure specializing in the consolidation

DOES TECHNOLOGY POLICY MATTER? 228 phase can expect to capture substantial gains in productivity and per capita income. Whether these gains will persist, however, depends on the capacity of the industrial structure (a) to exploit the results of successive emergence phases without having fully borne their costs, and (b) to transfer resources from one technological trajectory to another as the maturity phase sets in. • To succeed, an industrial structure pursuing technological trajectories into the maturity stage will require high levels of efficiency both in R&D and in applications engineering, allowing it (a) to obtain a maximum of performance improvements out of a given path of development, thus slowing the substitution process, and (b) to retain profitability by specializing in the product segments least vulnerable to intensified competition. Nonetheless, it may be supposed that the long- term performance of such an industrial structure will be constrained by the gradual slowing of market growth and the decreasing number of technological opportunities. These hypotheses merge naturally with the country analysis presented below. Thus, the predominance of "shifting" behavior in the U.S. economy corresponds to specialization in the emergence phase of technological trajectories. The returns to this pattern of specialization are maximized by (a) the scale on which experimentation occurs, increasing the probability of success; (b) the sophistication of the U.S. market (including its public procurement component), which accelerates the process of selection among competing alternatives; (e) the rapidity with which breakthroughs in the noncommercial parts of the technological system diffuse into the sphere of commercial experimentation; and (d) the existence of a substantial pool of large U.S. firms capable of transferring the results of experimentation in the U.S. market into world sales. However, the inherent characteristics of this phase of technological evolution limit the rate of growth of per capita incomes to which it can give rise. These limits have been accentuated by the declining competitiveness of U.S. sites (though less so of U.S. firms) in the mass production operations characteristic of the consolidation phase. The "imperfect shifting" that is sometimes considered a major feature of France's technological system limits the returns obtained from concentrating on the emergence stage of technological trajectories. High levels of investment in R&D are incurred to establish a presence in this stage, although the scale of experimentation may still be too small to achieve a reasonable chance of success across the board. Even when successful outcomes are obtained, numerous factors slow their transfer from the

DOES TECHNOLOGY POLICY MATTER? 229 mission-oriented environment to that of commercial exploitation and hence the prospects for going from emergence to consolidation. The growth of French incomes has therefore depended heavily on sectors such as motor vehicles, tires, and food processing, which are outside of— and only weakly linked to—the core technological system. However, performance in these sectors has proved difficult to sustain. This is partly because the decline of traditional industries and the implicit protection accorded high-technology activities has forced other sectors to bear a disproportionate share of unfavorable macroeconomic developments. At the other extreme, the deepening processes characteristic of Germany's industrial structure are associated with far-reaching specialization in pursuing technological trajectories into their mature phases. An institutional framework that is, in many respects, uniquely suited to this pattern has allowed German industry to exploit fully the higher value-added segments of the markets in which it operates. The experience of the last decades, however, has highlighted some of the risks this pattern of specialization entails. In particular, it creates vulnerability on two fronts: • Up-market, from competitors operating in the same product markets, but exploiting new technological trajectories as they enter the consolidation phase. These competitors are well placed to provide rapid rates of increase in cost-to-performance ratios—as Japanese firms have done in numerically controlled machine tools; and • Down-market, from competitors whose technological capabilities may lag, but whose factor costs are substantially lower. The slowing of total factor productivity growth as technological opportunities along the original trajectory diminish, combined with the rationalization pressures arising from greater rivalry on world markets, could make rising living standards more difficult to achieve. This could endanger the high degree of social consensus that underpins the diffusion-oriented countries' industrial model. Between these extremes of shifting among emerging trajectories and deepening along mature trajectories, Japan has been extraordinarily successful in exploiting successive trajectories in their consolidation phase. Concentration on this phase has provided numerous advantages to Japanese firms: • By avoiding the stage of greatest technological and commercial uncertainty, the return on scarce R&D capabilities could be maximized. • Entering activities at the consolidation (rather than emergence) stage also minimized the importance of close proximity to sophisticated users —until recently a major constraint on Japanese competitiveness.

DOES TECHNOLOGY POLICY MATTER? 230 • Greatest benefit could be drawn from accumulated skills in managing large-scale fabrication and assembly processes and from the cost- reducing pressures of competition for a growing market. Rapid growth of real income has been achieved by exploiting these advantages. At the same time, Japanese firms' share of world markets has increased at the expense of firms more specialized in the emergence or maturity stages of technological development. This pattern of specialization has relied on a high degree of social consensus, which has made it possible to shift resources rapidly from one trajectory to another. It has also relied on: • an elaborate network for gathering information from abroad about emerging technologies; • the low-cost availability of the results of U.S. R&D on entirely new technologies; and • access to world markets in which to achieve economies of scale. These premises now appear vulnerable in several respects. Access to U.S. technology is not as easy as it once was, mainly because U.S. firms now consider Japanese firms as major rivals. Moreover, Japanese technological performance has passed the stage at which large improvements could be obtained simply by learning from overseas. Finally, access to world markets is threatened by the spread of protectionist measures. Nonetheless, the capacity of Japanese industry to meet these threats should not be underestimated, Japanese R&D capabilities are now more than sufficient to engage in highly advanced research, the Japanese domestic market is large and sophisticated enough to provide a good seedbed for experimentation, and Japanese firms have established the global brand image and distribution channels needed to sell a more diversified range of products internationally. This discussion suggests that there are different paths to happiness, as countries' institutional structures and social arrangements facilitate specialization in differing stages of technological evolution (see Figure 2). Each of these stages has advantages and disadvantages in providing for the growth of real income, but countries also differ in the extent to which they succeed in securing the greatest benefits from any given pattern of specialization. Over the longer term, these differences in R&D efficiency may be most important. Consider France and Germany: The French state has encouraged specialization in the emergence phase of technologies, whereas German industry has largely retained its traditional pattern of specialization. However, the disparities in performance among these countries arise less from this difference in specialization than from the efficiency with which the

DOES TECHNOLOGY POLICY MATTER? 231 potential economic gains implicit in each pattern of specialization are exploited. In other words, location on a technological trajectory may be less important than the efficiency with which the advantages of that location are pursued. This, in turn, depends on institutional features (broadly defined) that may be more or less appropriate for a given pattern of specialization. Figure 2 National strengths along technological trajectories. It is by no means obvious that the institutional features typical of one economy can be transplanted to another. But a few general factors underlie the differing outcomes countries obtain from similar patterns of specialization. It is to these factors and particularly their implications for policy that we now turn. POLICY IMPLICATIONS The dominant feature of national technological systems is diversity. This partly reflects differences in policy stance between countries, but many other factors are also at work. Examination of these factors suggests several conclusions relating to the scope and limits of technological policy.

DOES TECHNOLOGY POLICY MATTER? 232 The first and most fundamental is the dependence of technology policy outcomes on their economic and institutional environment. The policies pursued in the United Kingdom or France do not differ greatly from those of the United States, but the outcomes do. The masons for this lie partly in the details of policy design and in the manner in which policies are implemented. But deeper and more pervasive factors are of far greater significance. In pan, the U.S. advantage arises from the very size of its scientific and technological system. This ensures that mission-oriented research crowds out commercial R&D to only a limited extent and that there is a huge stock of firms and individuals capable of absorbing and commercializing the results of mission-oriented research. But this advantage of size is accentuated by other features of the U.S. system. In particular, new technological capabilities spread rapidly in the U.S. economy, both through the direct transmission of ideas—for example, between industry and university—and through the high mobility of technologically skilled personnel. Moreover, lower entry barriers into U.S. industry, combined with pressures for firms to be among the early entrants into new product markets, accelerate the transformation of technological advances into commercial innovations. In France, by contrast, several factors slow the transfer of the technological advances generated by mission-oriented research into the commercial sector. These include the paucity of contacts between universities and industries, the low mobility of scientists and engineers, the pervasive obstacles to the entry of new firms, and the protective atmosphere of government procurement in which larger firms prefer to remain.40 Those differences mean that in the United States the results of government-supported R&D diffuse quickly into the commercial sector of the economy, but in France, and even more so the United Kingdom, they remain more or less confined to their sector of origin. The Importance of Diffusion This suggests a second conclusion, which is that the key problem of technology policy (as distinguished from science policy) lies less in generating new ideas than in ensuring that they are effectively used. The "high-technology industries," however defined, are inevitably a small pan of total output; taken on its own, even predominance in these industries will have a limited impact on living standards (Nelson, 1984; Riche et al., 1983). Rather, long-term growth mainly depends on the capacity to deploy technical capabilities across a broad range of economic activities. This goal can be achieved in various ways. In the United States, the diffusion of technology is largely a market-driven process, which relies

DOES TECHNOLOGY POLICY MATTER? 233 on high levels of mobility of human and financial resources and the existence of a marketplace of ideas. In Germany and Switzerland, in contrast, organized social mechanisms for promoting technology diffusion play a more important role—these include the apprenticeship system, the system of industrial standardization, and the network of cooperative research. Seen purely in institutional terms, these experiences are not easily transferable among countries. Japan borrowed heavily from overseas in designing its institutional framework, but at an early stage of industrial development. It is questionable whether policymakers in the United Kingdom or France could quickly set up processes of industrywide technological cooperation akin to those that developed over a long period of time in the German-speaking countries. The institutional mechanisms for technology diffusion must inevitably reflect broader features of a country's economic, social, and even political environment. However, there are common elements to the countries with a record of success in technology diffusion. These elements can provide a useful indication for technology policy.. Three such elements emerge from this study. Investment in Human Capital The first element in the successful diffusion of technology is the role of investment in human capital. This investment has both a flow and a stock dimension. The flow of newly trained personnel into the active population allows the continuous upgrading of skills and capabilities. At the same time, the better educated the labor force is, the greater will be its capacity to adjust to sophisticated new techniques. Higher levels of education are also likely to make this capacity more widespread, both throughout industry and throughout the active population. Countries whose investment in human capital lacks depth or breadth may be among the pioneers in generating new technologies, given a sufficiently strong scientific elite. But as far as using these technologies is concerned, they will be disadvantaged on two counts: an inadquate rate of expansion or replacement of the skill base at the margin and difficulties in adjusting the existing stock to the demands of technological change. Moreover, their difficulties are likely to persist or even mount. The production of human capital is highly intensive in human capital, and the lags involved in correcting deficiencies in the human capital stock can be extremely long (Sandberg, 1979). Policy Decentralization A second factor in promoting diffusion relates to the design of technology policies. Whether those policies actually promote the best use of tech

DOES TECHNOLOGY POLICY MATTER? 234 nological advance appears to be closely related to the range of actors they involve--that is, to their degree of decentralization. This, it can be conjectured, occurs for three reasons. First, centralized programs frequently concentrate resources on the wrong areas. In both the United Kingdom and France, for example, excessive resources have been devoted to projects that are technologically glamorous but not economically relevant. Second, the concentration of resources on a small number of projects itself increases the risk of costly failures, particularly when each project being supported entails a high level of risk. Finally, even if successful in terms of their mediate objectives, large, centralized projects usually pose considerable problems of technology transfer once the R&D phase is completed. Program decentralization can be achieved in different ways. In the United States, the very scale of the defense R&D program is such that a fairly high level of dispersion of funds is almost inevitable, but conscious policy choices— such as the emphasis on second-sourcing and the support of R&D by new and small firms—are also significant. In Germany, Switzerland, and, to a lesser extent, Sweden, the delegation of policy-setting and implementation functions to industry associations and regional bodies averts the risks inherent in centralized, bureaucratic decision making. The Japanese emphasis on consensus probably plays a similar role. But abstracting from these differences, similarities emerge. The risks of placing too many eggs in one basket (and choosing the wrong basket at that) can be reduced by making support policy less discriminatory in the range of firms and sectors covered and by placing less emphasis on discretionary choices among alternative approaches. This implies a preference for measures with a high degree of automaticity—for example, tax expenditures. It also implies preference for the delegation of power and public support to broadly based rather than narrowly based groups—for example, to an industry or research association as a whole rather than a formal ''club'' of subsidy receivers. Traditionally, the major argument against nondiscretionary policies is that funds may be provided to firms for projects that would have been carried out in any case. Equally, the case against decentralizing decision making rests on the risk that support programs will be "captured" by organized interest groups who will abuse them to advance narrow sectional concerns. However, experience suggests that the risks of capture are greatest when decisions are highly centralized, since this usually leads to a symbiotic relationship between a small number of policymakers and a few large firms. Experience suggests further that it is in this situation that public support is most likely to become a permanent feature of the cash flow of a narrow range of privileged firms (Bauer and Cohen, 1981; Cawson et al, 1985; Cohen and Bauer, 1985; Young and Lowe, 1974).

DOES TECHNOLOGY POLICY MATTER? 235 Providing Incentives Even an improved policy framework need not lead to better performance of the incentives to make the best use of technological resources are too weak. At a most obvious level, this is a problem of ensuring that firms are exposed to competition so that ideas are quickly transferred from the research environment to that of commercial use. The problem of providing adequate incentives merits particular attention in three areas: public research laboratories and other nonprofit research institutions, publicly funded commercial R&D, and public procurement. The first of these areas should include scope—notably in the United Kingdom and France—both for reducing the share of public laboratories in government R&D expenditure and for shifting a greater part of their recurrent funding onto a matching grant basis. In the second area, opportunities should be explored for building incentives for success into the system of public support for commercial R&D—for example, by making access to continuing finance more clearly conditional on past performance. The third area, public procurement—notably of complex technological systems—too often serves to subsidize long-term inefficiency rather than to encourage the best use of resources and capabilities. Dismantling these protective devices could impose short-term costs, but these are likely to be small in relation to the longer-term benefits. In summary, it is true that the institutional framework of any one country cannot be mechanically transplanted to others. Nonetheless, comparative analysis suggests three priority areas for action: • easing constraints and rigidities that slow the diffusion of new skills and technical capabilities; • improving the human capital base while enhancing the efficiency of markets for highly trained personnel; and • increasing the extent to which technology policy relies on market signals and incentives, rather than on the administrative allocation of resources. ACKNOWLEDGMENTS The author thanks Bruce Guile of the National Academy of Engineering; Rolf Piekarz of the National Science Foundation; Professor David Encaoua, of the Direction de la Prevision, Ministére de l'Economie, des Finances, et de la Privatisation; Christian Sautter, Inspectcur Général des Finance; and P. D. Henderson, H. Fest, J. Sharer, D. Baldes, and many other colleagues at the Organization for Economic Cooperation and Development for their valuable comments on earlier drafts of this paper. Special thanks are also given to the author's colleagues Rauf Gönenç, Andreas Lindner,

DOES TECHNOLOGY POLICY MATTER? 236 Anders Reutersward, and Barrie Stevens for generously providing data and advice. However, the author wishes to stress that unless otherwise indicated, the views expressed in tiffs paper are attributable only to the author in a personal capacity and not to any institution. NOTES 1. See especially Rosenberg and Birdzell, 1986. Nove, 1983, provides an interesting comparison by the relatively sympathetic description of the functioning of a socialist economy and of its difficulty in innovating. 2. This is a key component of the classic "market failure" argument for public support for R&D. See Antonelli, 1982; Freeman, 1974; Kamien and Schwartz, 1982; Mowery, 1983a; Rothwell and Zegveld, 1981. 3. This description of the United Kingdom draws on Carter, 1981; Dickson, 1983; Hall, 1980; Henderson, 1977; Hogwood and Peters, 1985; Vernon, 1974; Young and Lowe, 1974. 4. This description of France draws on Bauer and Cohen, 1981; Cawson et al., 1985; Cohen and Bauer, 1985; Dupuy and Thoenig, 1983; Grjebine, 1983; Shonfield, 1965; Stoffaes, 1984; Vernon, 1974. 5. See especially Ponssard and Pouvoirville, 1982. The high concentration levels of overall transfers from the state to industry (including public procurement) are discussed in Centre d'Economie Industrielle, n.d., and Commissariat Général du Plan, 1979. 6. On telecommunications see Cohen and Bauer, 1985; Darmon, 1985; Ergas, 1983b; Peterson and Comes, 1985. On energy, see specifically Feigenbaum, 1985; Picard et al., 1985. 7. This discussion of the United States draws on Fox, 1974; Gansler, 1980; Nelson, 1982, 1984; Phillips, 1971; Research & planning Institute, Inc., 1980. 8. Thus, Scherer (1982) estimates that in the United States only 12 percent of 1974 defense R&D funding generated technologies that flowed directly to clearly nondefense uses. 9. Secondary effects are examined by, among others, Ettlie, 1982; Hers, 1977; Malerba, 1985; Rothwell and Zegveld, 1981; Scribberas et al., 1978; Teubal and Steinmueller, 1982. An interesting international comparison of secondary effects can be obtained by contrasting U.K. and U.S. surveys of the effects on defense funding on national semiconductor industries: Dickson, 1983; Mowery, 1983b. 10. The role of U.S. government funding in the growth of small firms is discussed in Bollinger et al., 1983; Research & Planning Institute, Inc., 1980. A survey is given in Ergas, 1984b. Defense funding of university research and its growing importance is discussed in National Science Board, 1986, chap. 2. 11. Compare Katz and Phillips, 1982, and Lavington, 1980. 12. See especially Freeman, 1971; Freeman, 1976; National Science Foundation, 1985. Compare with Wilson, 1980. 13. Compare National Science Board, 1986, p. 86 and appendix table 4-17; Le Monde, 6 February 1986. 14. See Ergas, 1984b, pp. 10-11. A fascinating case study is National Academy of Engineering, 1982. The role of scale economics in intensifying rivalry in the transition to mass production is dearly brought out by recent literature on strategic competition. See, for an excellent survey, Kreps and Spence, 1985. 15. See especially Schmalensee, 1982. Advertising-related product differentiation also

DOES TECHNOLOGY POLICY MATTER? 237 appears to be a particularly significant factor explaining persistent profitability in U.S. industry. See Geroski, 1985; Mueller, 1985. 16. Aspects of this pattern are highlighted in Prais, 1981. Robson et al. (1985) examine the diffusion of technology in the United Kingdom. See also the analysis of the United Kingdom's trade structure in Orléan, 1986. 17. See, in addition to the references in note 4 above, analyses of France's trade patterns presented in Lafay, 1985; Orléan, 1986; Vellas, 1981. 18. The results of Lipsey and Kravis, 1985, conflict with those of Dunning and Pearce, 1985, who find a sharper decline in U.S. firms' overall share of revenues and profitability. 19. The classic formulation of this process is Vernon, 1966. For empirical analysis of U.S. trade patterns, see inter alia the contrasting results set out in Hatzichronoglou, 1986; Lafay, 1985; Learner, 1984; Vernon, 1979. 20. The general characteristics of these countries are explored in Katzenstein, 1985a and 1985b. 21. On Germany and Switzerland, see Henderson, 1975; Milward and Saul, 1977. On Scandinavia, see Hecksher, 1984; Hildebrand, 1978. 22. The general characteristics of these educational systems, and international comparisons, are set out in Stevens, 1986. See also Organization for Economic Cooperation and Development, 1979; Prais and Wagner, 1983a and 1983b; Worswick, 1985. 23. A recent survey reports that in Germany 45 percent of labor force participants with vocational training at a high school level undertook continuing training during the period 1974-1979. 24. According to population census estimates, some 50 percent of the civilian labor force in Germany and Switzerland has completed an apprenticeship. See Organization for Economic Cooperation and Development, 1986. 25. The classic study is Brady, 1934. 26. Estimates are provided in Laboratorio di Politica Industriale, 1982. The literature on standardization is reviewed in Ergas, 1984b. 27. I am indebted to my colleagues in the Science, Technology and Industry Directorate of the Organization for Economic Cooperation and Development for assisting me in compiling the information presented here. 28. See especially Meyer-Krahmer et al., 1983. My colleague Andreas Lindner provided me with particularly useful information on the subjects discussed in this section. 29. See George and Ward, 1975; Prais, 1981; Pratten, 1976. Particularly useful case studies are Aylen, 1982; Daly and Jones, 1980. 30. This discussion draws on Aglietta and Boyer, 1983; Learner, 1984; Ohlsson, 1980; Orléan, 1986. A particularly useful discussion of the balance between shifting resources among competing uses, as against increasing their productivity in existing uses, is in Carlsson, 1980. 31. It has been estimated that Japanese firms' total expenditure on vocational education is 5 times greater than public expenditure on vocational education. 32. See especially Collins, 1981, 1982; Saxonhouse, 1984; Uena, 1977. 33. See Caves and Uekasa, 1976. On price competition, see Encaoua et al., 1983. 34. A theoretical model in which collusion is less stable in a growing than in a declining market is set out in Rotemberg and Saloner, 1984. Estimates of the cost of funds arc given in Ando and Auerbach, 1986. 35. Thus, in the United States, the number of large takeovers (valued at $100 million or

DOES TECHNOLOGY POLICY MATTER? 238 more) has increased steadily over the last decade, rising from 14 in 1975 to 116 in 1982; in Japan, in contrast, the number of large transfers (exceeding $50 million) has been virtually constant, with only 10 such transfers occurring in 1981. See Organization for Economic Cooperation and Development, 1984b. 36. See Tachibanki, 1984, who estimates that lifetime employment applies to no more than 10 percent of the Japanese labor force, almost entirely at higher levels of educational attainment. 37. On the blurring of frontiers between basic and applied research, see Committee on Science, Engineering, and Public Policy, 1983; on its implications for Japan, and concern about the future, see Sciences and Technology Agency (Japan), 1985. 38. Useful overviews are in Antonelli, 1982; Bollinger et al., 1983; Dosi, 1982; Kamien and Schwartz, 1982. 39. Some of the caveats in this respect are set out in Ergas, 1983a. See also Clark, 1985. 40. The fact that France and, to a lesser extent, the United Kingdom have lagged in applying competition policy to their respective national industries has also presumably been a factor reducing the pressure on firms to innovate. REFERENCES Aglietta, M., and R. Boyer. 1953. PÔles de Compétitivité, Stratégie Industrielle et Politique Macro- economique. Paris: Working Paper CEPREMAP No. 8223. Ahlström, G. 1952. Engineers and Industrial Growth. London: Croom Helm. Ando, A., and A. Auerbach. 1986. The Corporate Cost of Capital in Japan and the U.S.: A Comparison. Research Working Paper No. 1762. Cambridge: National Bureau of Economic Research. Antonelli, C. 1952. Cambiamento Tecnologico e Teoria dell'Impresa. Torino: Loescher Editore. Arocena, J. 1983. La Création d'Enteprise. La Documentation Française, 1983. Aylen, J. 1982. Plant size and efficiency in the steel industry: An international comparison. National Institute Economic Review 100 (May). Baker, W. J. 1971. A History of the Marconi Company. New York: St. Martin's Press. Bauer, M., and E. Cohen. 1951. Qui Gouveme les Groupes Industrielles? Paris: Editions du Scull. Beer, J. J. 1959. The Emergence of the German Dye Industry. Harmondsworth: Penguin. Ben-David, J. 1968. Fundamental Research and the Universities. Paris: Organization for Economic Cooperation and Development. Berger, S. D., ed. 1981. Organizing Interests in Western Europe. Cambridge: Cambridge University Press. Bollinger, L., K. Hope, and J. M. Utterback. 1983. A review of literature and hypotheses on new technology-based firms. Research Policy 12 (February). Boltho, A. 1975. Japan—An Economic Survey. Oxford: Oxford University Press. Brady, R. 1934. The Rationalization Movement in German Industry. Berkeley, Calif.: University of California Press. Brooks, H. 1983. Towards an efficient public policy: Criteria and evidence . In Emerging Technologies, H. Giersch, ed. Tubingen: J.C.B. Mohr (Paul Siebeck). Carlsson, B. 1980. Technical Change and Productivity in Swedish Industry in the PostWar Period. Stockholm: The Industrial Institute for Economic and Social Research, Research Report No. 8. Carter, C., ed. 1981. Industrial Policy and Innovation. London: Heinemann.

DOES TECHNOLOGY POLICY MATTER? 239 Caves, R., and M. Uekasa. 1976. Industrial Organization in Japan. Washington, D.C.: Brookings Institution. Cawson, A., P. Holmes, and A. Stevens. 1985. The Interaction Between Firms and the Sine in France: The Telecommunications and Consumer Electronics Sectors. Cambridge: Trinity Hall, December 10-13, 1985, mimeo. Centre d'Economie Industrielle. n.d. Quelques Réfiexions à Propos des Méanismes de Transfert Etat- Industrie Mis en Oeuvre en France et en Allemagne. Centre d'Economie Industrielle, Les Milles, mimeo. Clark, K. B. 1985. The interaction of design hierarchies and market concepts in technological evolution. Research Policy 14 (October). Cohen, M., and M. Bauer. 1985. Les Grandes Manoeuvres Industrielles. Paris: Pierre Belfond. Collins, E. 1981. Corporation Income Tax Treatment of Investment and Innovation Activities in Six Countries . Washington, D.C.: National Science Foundation. Collins, E., ed. 1982. Tax Policy and Investment in Innovation. Washington, D.C.: National Science Foundation. Commissariat Général du Plan. 1982. Aides à l'industrie. Mimeo. Committee on Science, Engineering, and Public Policy. 1983. Frontiers in Science and Technology: A Selected Outlook. New York: W. H. Freeman. Daly, A., and D. T. Jones. 1980. The machine tool industry in Britain, Germany and the United States. National Institute Economic Review 92 (May). Darmon, J. 1985. Le Grande Dérangement: La Guerre du Té1éphone, France: J.-C. Lattés. Dickson, K. 1983. The influence of Ministry of Defense funding on semiconductor research and development in the United Kingdom. Research Policy 12 (April). Dore, R. 1985. A Case Study of Technology Forecasting in Japan--The Next Generation Base Technologies Development Programme. London: The Technical Change Centre. Dost, G. 1982. Technological paradigms and technological trajectories. Research Policy 11 (June). Dunning, J. H., and R. D. Pearce. 1985. The World's Largest Industrial Enterprises, 1962-1983 . New York: St. Martins. Dupuy, F., and J.-C. Thoenig. 1983. Sociologie de l'Administration Française. Paris: Armand Colin. Earle, E. M. 1986. Adam Smith, Alexander Hamilton, Friedrich Lisk,: The economic foundations of military power. In P. Paretz, ed., Makers of Modern Strategy. Princeton: Princeton University Press. Encaoua, D., with P. Geroski and R. Miller. 1983. Price Dynamics and Industrial Structure: A Theoretical and Econometric Analysis. Paris: Organization for Economic Cooperation and Development (Economic and Statistics Department Working Paper No. 10), July, mimeo. Ergas, H. 1979. Biases in the Measurement of Real Output Under Conditions of Rapid Technological Change. Organization for Economic Cooperation and Development (Expert Group on the Economic Impact of Information Technologies, Working Party on Information, Computer and Communications Policy). Ergas, H. 1983a. The Inter-industry Flow of Technology. Paris: Organization for Economic Cooperation and Development (Workshop on Technological Indicators and the Measurement of Performance in International Trade). September. Ergas, H. 1983b. Telecommunications Policy in France. Mimeo. Ergas, H. 1984a. Corporate strategies in transition. In A. Jacquemin, ed., European Industry: Public Policy and Corporate Strategy. Oxford: Oxford University Press.

DOES TECHNOLOGY POLICY MATTER? 240 Ergas, H. 1984b. Why Do Some Countries Innovate More Than Others? Brussels: Centre for European Policy Studies Paper No. 5. Ettlie, J. E. 1982. The commercialization of federally sponsored technological innovations. Research Policy 11 (June). Feigenbaum, H. B. 1985. The Politics of Public Enterprise: Oil and the French State. Princeton: Princeton University Press. Floud, R. 1984. Technical Education 1850-1914: Speculation on Human Capital Formation. London: Centre for Economic Policy Research, April, mimeo. Forman, P. 1974. Industrial support and political alignments of the German physicists in the Weimar Republic. Minerva (January). Fox, J. R. 1974. Arming America: How the U.S. Buys Weapons. Cambridge, Mass.: Harvard University Press. Freeman, C. 1974. The Economics of Industrial Innovation. Harmondsworth: Penguin. Freeman, R. B. 1971. The Market for College Trained Manpower. Cambridge, Mass.: Harvard University Press. Freeman, R. B. 1976. The Overeducated American. New York: Academic Press . Gansler, J. S. 1980. The Defense Industry. Cambridge, Mass.: MIT Press. George, K. D., and T. S. Ward. 1975. The Structure of Industry in the EEC: An International Comparison. Cambridge: Cambridge University Press. Geroski, P. A. 1985. DO Dominant Finns Decline? University of Southampton Discussion Paper in Economics and Econometrics No. 8509, August, mimeo. Glover, I., and P. Lawrence. 1976. Engineering the miracle. New Society 30 September. Glover, R. W. 1974. Apprenticeship in America: An Assessment. Proceedings of the Industrial Relations Research Association, December. Gönenç, R. 1984. Electronisation et Réorganisation Verticales dans l'Industrie. Thése de Troisiéme Cycle, Université de Paris, Nanterre. Gönenç, R. 1986. Changing investment structure and capital markets. In H. Ergas, ed., A European Future in High Technology? Brussels: Centre for European Policy Studies. Gönenç, R., and Y. Lecler. 1982. L'électronisation Industrielle au Japon. Sciences Sociales du Japon Contemporain No. 2. Paris: Centre National de la Recherche Scientifique, Centre du Documentation Sciences Humaines, mimeo. Grjebine, A. 1983. L'état d'Urgence. Pads: Flammarion. Hall, P. 1980. Great Planning Disasters. Berkeley, Calif.: University of California Press. Hatzichronoglou, T. 1986. International trade in high technology products: Europe's competitive position. In H. Ergas, ed., A European Future in High Technology? Brussels: Centre for European Policy Studies. Hecksher, E. F. 1984. An Economic History of Sweden. Cambridge, Mass.: Harvard University Press. Henderson, P. D. 1977. Two British errors: Their probable size and some possible lessons. Oxford Economic Papers 29 (July). Henderson, W. O. 1975. The Rise of German Industrial Power 1834-1914. Berkeley, Calif.: University of California Press. Hildebrand, K. G. 1978. Labour and capital in the Scandinavian countries in the nineteenth and twentieth centuries. In P. Mathias and M. M. Postan, eds., The Cambridge Economic History of Europe: The Industrial Economies--Capital, Labour and End: Britain, France, Germany and Scandinavia. Cambridge: Cambridge University Press. Hitch, C. J., and R. N. McKean. 1960. The Economics of Defense in the Nuclear Age. Cambridge, Mass.: Rand Corporation and Harvard University Press .

DOES TECHNOLOGY POLICY MATTER? 241 Hodenheimer, A. J. 1978. Major Reforms of the Swedish Education System 1950-1975. Washington, D.C.: World Bank Staff Working Paper No. 290. Hogwood, B. W., and B. G. Peters. 1985. The Pathology of Public Policy. Oxford: Clarendon Press. Imai, K., and H. Irami. 1981. The Firm and Market in Japan—Mutual Penetration of the Market Principle and Organization Principle. Tokyo: Institute of Business Research Discussion Paper No. 104, Hitotsubashi University, June, mimeo. Imai, K., I. Nonaka, and H. Takeuchi. 1984. Managing the New Product Development Process: How Japanese Companies Learn and Unlearn. Tokyo: Institute of Business Research Discussion Paper No. 118, Hitotsubashi University, March 29, mimeo. Johnson, C. 1982. MITI and the Japanese Miracle—The Growth of Industrial Policy 1925-1975. Stanford: Stanford University Press. Jones, I., and, H. Hollenstein. 1983. Trainee Wages and Training Deficiencies: An Economic Analysis of a "British Problem." London: National Institute of Economic and Social Research Industry Series No. 12, June, mimeo. Kamien, M. I., and N. L. Schwartz. 1982. Market Structure and Innovation. Cambridge: Cambridge University Press. Katz, B. C., and A. Phillips. 1982. The computer industry. In R. Nelson, ed., Government and Technical Progress. New York: Pergamon Press. Katzenstein, P. J. 1955a. Corporatism and Change: Austria, Switzerland, and the Politics of Industry. Ithaca, N.Y.: Cornell University Press. Katzenstein, P. J. 1985b. Small States in World Markets: Industrial Policy in Europe. Ithaca, N.Y.: Cornell University Press. Kono, T. 1984. Strategy and Structure of Japanese Enterprises. London: Macmillan. Kreps, D. M., and A.M. Spence. 1985. Modelling the role of history in industrial organization and competition. In Feiwel, ed., Issues in Contemporary Microeconomics and Welfare Analysis. Cambridge: Cambridge University Press. Laboratorio di Politica Industriale. 1982. Materiali de Discussione. Bologna: Laboratorio di Politica Industriale, November, mimeo. Lafay, G. 1985. Spécialization française: Des handicaps structurels. Revue d'Economie Politique 95 (5). Launer, H., and W. Ochel. 1985. Industrielle strukturenpassung: Das Japanische modell. Ifo- Schelldienst (September 26). Lavington, S. 1980. Early British Computers: The Story of Vintage Computers and the People Who Built Them. Manchester: The Digital Press. Learner, E. E. 1984. Sources of International Comparative Advantage: Theory and Evidence. Cambridge, Mass.: MIT Press. Le Monde. 6 February 1986. M. Jean-Jacques Duby quitte le CNRS. Le Monde. 27 September 1979. Le rapport hannoun souligue la forte concentration et la faible efficacité des aides publiques à l'industrie. Liebenau, J. 1985. Innovation in pharmaceuticals: Industrial R&D in the early twentieth century. Research Policy 14 (August). Lipsey, R. E., and I. B. Kravis. 1985. The Competitive Position of U.S. Manufacturing Firms. Cambridge, Mass.: National Bureau of Economic Research Working Paper No. 1557, February. Maclaurin, W. R. 1949. Invention and Innovation in the Radio Industry. New York: Macmillan. Malerba, F. 1985. Demand structure and technological change: The case of the European semiconductor industry. Research Policy 14 (October).

DOES TECHNOLOGY POLICY MATTER? 242 Maurice, M., F. Sellier, and J.-J. Sylvestre. 1982. Politique d'Education et Organisation Industrielle en France et en Allemagne. Paris: Presses Universitaires de France. Meyer-Krahmer, F., G. Gielow, and U. Kuntze. 1983. Impacts of government incentives towards industrial innovation. Research Policy 12 (June). Milward, A. S., and S. B. Saul. 1977. The Development of the Economies of Continental Europe 1850-1914. Cambridge, Mass.: Harvard University Press. Mitchell, J. P. 1977. New Directions for Apprenticeship Policy. Worklife (January). Washington, D.C.: U.S. Department of Labor. Mowery, D.C. 1983a. Economic theory and government technology policy. Policy Sciences 12 (August). Mowery, D.C. 1983b. Innovation, market structure, and government policy in the American semiconductor electronics industry: A survey. Research Policy 12 (August). Mueller, D. 1985. Persistent Performance Among Large Corporations. Brussels: Centre for European Policy Studies, November, mimeo. Murray, C. 1984. Losing Ground: American Social Policy 1950-1980. New York: Basic Books. Nakamura, T. 1981. The Postwar Japanese Economy—its Development and Structure. Tokyo: University of Tokyo Press. Nakayama, S. (J. Dnsenbury, trans.). 1984. Academic and Scientific Traditions in China, Japan and the West. Tokyo: University of Tokyo Press. National Academy of Engineering. 1982. The Competitive Status of the U.S. Auto Industry. Committee on Technology and International Economic and Trade Issues, Automobile Panel. Washington, D.C.: National Academy Press. National Manpower Council. 1954. A Policy for Skilled Manpower. New York: Columbia University Press. National Science Board. 1986. Science Indicators--The 1985 Report. Washington, D.C.: National Science Foundation. National Science Foundation. 1985. Science and Engineering Personnel: A National Over-view. Washington, D.C.: National Science Foundation. Nelson, R. R., ed. 1982. Government and Technical Progress: Cross-industry Analysis. New York: Pergamon. Nelson, R. R. 1984. High-Technology Policies: A Five Nation Comparison. Washington, D.C.: American Enterprise Institute. Nelson R. R., and S. G. Winter. 1982. An Evolutionary Theory of Economic Growth. Cambridge, Mass.: Harvard University Press. Noble, D. F. 1977. America by Design. Oxford: Oxford University Press. Nomura Research Institute. 1983. Characteristics of Japan's Import and Export Structures. Tokyo: Nomura Research Institute. Nonaka, I., T. Kagono, and S. Sakamoto. 1983. Evolutionary Strategy and Structure—A New Perspective on Japanese Management. Tokyo: Institute of Business Research Discussion Paper No. 111, Hitotsubashi University, March, mimeo. Nove, A. 1983. The Economics of Feasible Sooialism. London: George Allen & Unwin. Office Fédéral de l'Industrie, des Arts et Métiers et du Travail. 1980. Politique Concern-ant le Marché du Travail en Suisse: Caractéristiques et Problémes, Vol. 1. Berne. Ohlsson, L. A. 1980. Engineering Trade Specialization of Sweden and Other Industrial Countries. Amsterdam: North Holland Publishing Co.

DOES TECHNOLOGY POLICY MATTER? 243 Nelson, R. R., ed. 1982. Government and Technical Progress: Cross-Industry Analysis. New York: Pergamon. Nelson, R. R. 1984. High-Technology Policies: A Five Nation Comparison. Washington, D.C.: American Enterprise Institute. Nelson R. R., and S. G. Winter. 1982. An Evolutionary Theory of Economic Growth. Cambridge, Mass.: Harvard University Press. Noble, D. F. 1977. America by Design. Oxford: Oxford University Press. Nomura Research Institute. 1983. Characteristics of Japan's Import and Export Structures. Tokyo: Nomura Research Institute. Nonaka, I., T. Kagono, and S. Sakamoto. 1983. Evolutionary Strategy and Structure—A New Perspective on Japanese Management. Tokyo: Institute of Business Research Discussion Paper No. 111, Hitotsubashi University, March, mimeo. Nove, A. 1983. The Economies of Feasible Socialism. London: George Alien & Unwin. Office Fédéral de l'Industrie, des Arts et Métiers et du Travail. 1980. Politique Concern-ant le Marché du Travail en Suisse: Caractéristiques et Problémes, Vol. 1. Berne. Ohlsson, L. A. 1980. Engineering Trade Specialization of Sweden and Other Industrial Countries. Amsterdam: North Holland Publishing Co. Organization for Economic Cooperation and Development. 1979. Policies for Apprentice-ship. Paris: Organization for Economic Cooperation and Development. Organization for Economic Cooperation and Development. 1984a. Industry and University: New Forms of Co-operation and Communication. Paris: Organization for Economic Cooperation and Development. Organization for Economic Cooperation and Development. 1984b. Mergers and Takeovers. Paris: Organization for Economic Cooperation and Development. Organization for Economic Cooperation and Development. 1986. Changes in Working Patterns and the Impact on Education and Training: Human Resources Policies and Strategies in Germany. Paris: Organization for Economic Cooperation and Development. Orléan, A. 1986. L'insertion dont les Échanges Internationaux. Economie et Statistique (January). Peck, M., and R. Wilson. 1982. Innovation, imitation, and comperative advantage. In H. Giersch, ed., Emerging Technologies: Consequences for Economic Growth, Structural Change and Employment. Tübingen: J. C. B. Mohr. Peterson, T., and F. J. Comes. 1985. An electronics dream that's shorting out. Business Week (March 4). Pham-Khac, K., and J. L. Pigelet. 1979. La Formation et l'Emploi des Docteurs és Sciences. Dossier du Centre d'Etudes et de Recherches sur les Qualifications. Paris. Phillips, A. 1971. Technology and Market Structure: A Study of the Aircraft Industry. New York: Heath Lexington Books. Pieard, J.-F., A. Beltran, and M. Bungener. 1985. Histoire(s) de I'EDF: Comment se Sont Prises les Décisions de 1946 à Nos Jours. Paris: Bordas. Ponssard, J. P., and G. de. Pouvoirville. 1982. Marché Publique at Innovation. Paris: Economica. Prais, S. J. 1981. Productivity and Industrial Structure. Cambridge: Cambridge University Press. Prais, S. J., and K. Wagner. 1983a. Schooling Standards in Britain and Germany: Some Summary Comparisons Bearing on Economic Efficiency. London: National Institute Discussion Paper No. 60. Prais, S. J., and K. Wagner. 1983b. Some practical aspects of human capital investment:

DOES TECHNOLOGY POLICY MATTER? 244 Training standards in five occupations in Britain and Germany. National Institute. Economic Review (August). Pratten, C. F. 1976. A Comparison of the Performance of Swedish and U.K. Companies. Cambridge: Cambridge University Press. Research & Planning Institute, Inc. 1980. Case Studies Examining the Role of Government R&D Contract Funding in the Early History of High Technology Companies. Cambridge, Mass.: Research & Planning Institute, Inc. Riche, R. W., D. E. Hecker, and J. U. Bergan. 1983. High technology today and tomorrow: A small slice of the employment pie. Monthly Labor Review (November). Robson, M., J. Townsend, and K. Pavitt. 1985. Sectoral Patterns of Production and Use of Innovations in the U.K.: 1945-1983 . Centre for Science, Technology and Energy Policy (Economic and Social Research Council), May 30, mimeo. Rosenberg, N., and L. E. Birdzell, Jr. 1986. How the West Grew Rich: The Economic Transformation of the Industrial World. New York: Basic Books. Rosenberg, N. 1976. Perspectives on Technology. Cambridge: Cambridge University Press. Rotemberg, J. J., and G. Saloner. 1984. A Supergame-Theoretic Model of Business Cycles and Price Wars During Booms. Cambridge, Mass.: MIT Working Paper No. 349, July, mimeo. Rothwell, R., and W. Zegveld. 1981. Industrial Innovation and Public Policy. Westport, Conn.: Greenwood Press. Ryan, P. 1984. Job training, employment practices and the large enterprise: The case of costly transferable skills. In P. Osterman, ed., Intern Labor Markets. Cambridge, Mass.: MIT Press. Saxonhouse, G. R. 1984. What is all this about ''industrial targeting'' in Japan? World Economy (September). Sandberg, L. G. 1979. The case of the impoverished sophisticate: Human capital and Swedish economic growth before World War I. The Journal of Economic History 39 (March). Scherer, F. M. 1982. Inter-industry technology flows in the United States. Research Policy 11 (August). Scherer, F. M., and D. Ravenscraft. 1984. Growth diversification: Entrepreneurial behavior in large- scale United States enterprises. Zeitschrift Fur Nationalökonomie (Suppl. 4). Schmalensee, R. 1982. Product differentiation advantages of pioneering brands. American Economic Review 72 (June). Schröder-Gudehus, B. 1972. The argument for the self-government and public support of science in Weimar Germany. Minerva. Sciences and Technology Agency (Japan). 1985. White Paper on Science and Technology 1986. Tokyo: Foreign Press Center of Japan. Scribberas, E. et. al. 1978. Competition, Technical Change and Manpower in Electronic Capital Equipment: A Study of the U.K. Mini-computer Industry. Brighton: Science Policy Research Unit. Shinshara, M. 1970. Structural Changes in Japan's Economic Development. Tokyo: Ko-nokuniya. Shonfield, A. 1965. Modern Capitalism. Oxford: Oxford University Press. Stevens, B. 1986. Labour markets, education and industrial structure. In H. Ergas, ed., A European Future in High Technology? Brussels: Centre for European Policy Studies. Stoffaes, C. 1984. Politique Industrielle. Paris: Les Cours de Droit. Sturmes, S. G. 1958. The Economic Development of Radio. London: Duckworth. Suzuki, K. 1985. An Empirical Analysis of the Interdependence of R&D Investment and

DOES TECHNOLOGY POLICY MATTER? 245 Market Structure in Japan. Tokyo: Research Institute of Capital Formation, The Japan Development Bank Staff Paper No. 4, October, mimeo. Tachibanki, T. 1984. Labour mobility and job tenure. In M. Aoki, ed., The Economic Analysis of the Japanese Firm. Amsterdam: North-Holland. Teubal, M., and E. Steinmueller 1982. Government policy, innovation and economic growth. Research Policy 11 (October). Uena, H. 1977. Conception and evaluation of Japanese industrial policy. Japanese Economic Studies (Winter 1976-1977). Vellas, F. 1981. Echanges Internationaux et Qualification du Travail. Paris: Economica. Vernon, R. 1966. International investment and international trade in the product cycle. Quarterly Journal of Economics 80 (May). Vernon, R. 1974. Big Business and the State: Changing Relations in Western Europe. London: Macmillan. Vernon, R. 1979. The product cycle hypothesis in a new international environment. Oxford Bulletin of Economics and Statistics 41 (November). Webbink. D. W. 1977. The Semiconductor Industry. Washington, D.C.: Government Printing Office. Weinberg, A.M. 1967. Reflections on Big Science. Oxford: Pergamon Press. Wilson, R. A. 1980. The rate of return to becoming a qualified scientist or engineer in Great Britain, 1966-1976. Scottish Journal of Political Economy (February):41-62. Worswick, G. D. N., ed. 1985. Education and Economic Performance. Aldershot, England: Gower Publishing Company. Young, S., and A. V. Lowe. 1974. Intervention in the Mixed Economy. London: Croom Helm.

Next: National and Corporate Technology Strategies in an Interdependent World Economy »
Technology and Global Industry: Companies and Nations in the World Economy Get This Book
×
Buy Paperback | $80.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF
  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!