I performed the initial research for this article under the auspices of the Center for Interdisciplinary Studies at the University of Bielefeld, Germany. For helpful comments on earlier drafts of the article, I am grateful to Michael Atkinson, Robert Bates, Joseph Grieco, Stephen Krasner, Tim McKeown, and the anonymous reviewers of International Organization. Of course, I am responsible for the article's remaining deficiencies.

1. See Shonfield, Andrew, Modern Capitalism (Oxford: Oxford University Press, 1965)Google Scholar.

2. For important contributions to this body of literature, see Katzenstein, Peter, ed., Between Power and Plenty (Madison: University of Wisconsin Press, 1978)Google Scholar; Katzenstein, Peter, Small States in World Markets (Ithaca, N.Y.: Cornell University Press, 1985)Google Scholar; Zysman, John, Governments, Markets, and Growth (Ithaca, N.Y.: Cornell University Press, 1983)Google Scholar; and Hall, Peter, Governing the Economy (New York: Oxford University Press, 1986)Google Scholar.

3. For recent works advocating a multivariate theory of industrial governance structures and competitive success, see Stinchcombe, Arthur, Information and Organization (Berkeley: University of California Press, 1990), especially pp. 33 and 345Google Scholar; and Porter, Michael E., The Competitive Advantage of Nations (New York: Free Press, 1990)CrossRefGoogle Scholar, especially chap. 3. Porter, however, does not treat governance structures (for example, firm strategy, structure and rivalry, patterns of interaction between related and supporting industries, and government intervention) as explananda in their own right. He instead treats them as additional inputs that explain the ultimate competitive success of industries.

4. Both dimensions were first combined in Katzenstein's, Peter “Conclusion: Domestic Structures and Strategies of Foreign Economic Policy,” in Katzenstein, , Between Power and Plenty, pp. 295–336Google Scholar.

5. For example, the following works emphasize national political institutions as at least one explanation for sectoral innovation policies: Dyson, Kenneth, “West European States and the Communications Revolution,” West European Politics 4 (Autumn 1986), pp. 11–55Google Scholar; Grant, Wyn and Streeck, Wolfgang, “Large Firms and the Representation of Business Interests in the UK and the West German Construction Industry,” in Cawson, Alan, ed., Organized Interests and the State: Studies in Meso-Corporation (Beverly Hills, Calif.: Sage, 1985), pp. 145–73Google Scholar; Katzenstein, Small States in World Markets; Katzenstein, Peter, ed., Industry and Policy in West Germany (Ithaca, N.Y.: Cornell University Press, 1989)Google Scholar; and Shepherd, Geoffrey, Duchene, Francois, and Saunders, Christopher, eds., Europe's Industries: Public and Private Strategies for Change (Ithaca, N.Y.: Cornell University Press, 1983)Google Scholar.

6. At most, sectoral distinctions are grounded in product cycle theory and subdivide industries according to their capital intensity, research and development effort, or change in product lines. See, for example, Katzenstein, Peter, “Industry in a Changing Germany,” in Katzenstein, , Industry and Policy in West Germany, pp. 21–28Google Scholar.

7. See, in particular, Atkinson, Michael M. and Coleman, William D., “Corporatism and Industrial Policy,” in Cawson, , Organized Interests and the State, pp. 22–44Google Scholar; Atkinson, Michael M. and Coleman, William D., “Strong States and Weak States: Sectoral Policy Networks in Advanced Capitalist Economies,” British Journal of Political Science 19 (01 1989), pp. 47–67CrossRefGoogle Scholar; Cawson, Alan, “Introduction: Varieties of Corporatism—The Importance of the Meso-Level of Interest-Intermediation,” in Cawson, , Organized Interests and the State, pp. 1–21Google Scholar; Wilks, Stephen and Wright, Maurice, “Conclusion: Comparing Government-Industry Relations—States, Sectors, and Networks,” in Wilks, Stephen and Wright, Maurice, eds., Comparative Government-Industry Relations (Oxford: Clarendon Press, 1987), pp. 274–313Google Scholar; and Wright, Maurice, “Policy Community, Policy Network and Comparative Industrial Policies,” Political Studies 36 (Fall 1988), pp. 593–612CrossRefGoogle Scholar.

8. Some economists have therefore advocated abandoning premature theoretical or prescriptive arguments about the linkage between governance structures and economic success in favor of qualitative case studies. See Nelson, Richard R., “Government Stimulus of Technological Progress: Lessons from American History,” in Nelson, Richard R., ed., Government and Technical Progress: A Cross-Industry Analysis (New York: Pergamon Press, 1982), pp. 481–82Google Scholar. A first step toward an analytic explanation of sectoral variance has been made by Hollingsworth, J. Roger and Lindberg, Leon N. in “The Governance of the American Economy: The Role of Markets, Clans, Hierarchies, and Associative Behavior,” in Schmitter, Philippe and Streeck, Wolfgang, eds., Private Interest Government (Beverly Hills, Calif.: Sage, 1985), pp. 221–54Google Scholar. Hollingsworth and Lindberg distinguish sectors by typical firm size and level of technological sophistication in order to predict governance structures. Unfortunately, firm size is an aspect, not a determinant, of governance structures, and the variable “technological sophistication” must be further specified, as I suggest in a subsequent section of my article.

9. See, in particular, Kurth, James R., “Industrial Change and Political Change: A European Perspective,” in Collier, David, ed., The New Authoritarianism in Latin America (Princeton, N.J.: Princeton University Press, 1979), pp. 319–62Google Scholar; Kurth, James R., “Political Consequences of the Product Cycle: Industrial History and Political Outcomes,” International Organization 33 (Winter 1979), pp. 1–36CrossRefGoogle Scholar; and Gourevitch, Peter A., Politics in Hard Times (Ithaca, N.Y.: Cornell University Press, 1986)Google Scholar. The theoretical inspiration for these studies was Gerschenkron's, AlexanderEconomic Backwardness in Historical Perspective (Cambridge, Mass.: Harvard University Press, 1962)Google Scholar.

10. Kurth, , “Industrial Change and Political Change,” p. 327Google Scholar.

11. A similar criticism applies to the most general, comprehensive, and elegant theoretical work explaining national governance structures in terms of the comparative advantages and disadvantages of economic “sectors” in world markets, Rogowski's, RonaldCommerce and Coalitions: How Trade Affects Domestic Political Alignments (Princeton, N.J.: Princeton University Press, 1989)Google Scholar. Rogowski states that the key actors are broad sectors (industry versus agriculture) and classes (capital versus labor in industry) and that their coalition potential depends on two independent variables: the relative scarcity of the factor of production controlled by each group and the expansion or contraction of the world economy. While he argues that technology and technological change play a role in constituting or dismantling collective actors and in accounting for the relative scarcity of factors of production, he never systematically develops this argument. For example, on pp. 18 and 178, he claims that improved technology has rendered the three-group model obsolete for the period since 1960, but he fails to demonstrate that technologies before 1960 were sufficiently homogeneous within the agricultural and industrial areas and sufficiently different between them to qualify each area as a sector with interests and governance structures.

12. See Piore, Michael J. and Sable, Charles F., The Second Industrial Divide: Possibilities for Prosperity (New York: Basic Books, 1984)Google Scholar. For another study with a similar thrust, see Kern, Horst and Schumann, Michael, Das Ende der Arbeitsteilung? (The end of the division of labor?) (Munich: Beck, 1984)Google Scholar. For a survey of the burgeoning literature, see Lash, Scott and Urry, John, The End of Organized Capitalism (Madison: University of Wisconsin Press, 1987)Google Scholar.

13. For the theoretical core of Piore, and Sabel's, The Second Industrial Divide, see especially pp. 14–15 and 50–65Google Scholar.

14. See ibid., p. 5.

15. Ibid., p. 43.

16. See ibid., p. 14. For the institutional arrangements conducive to craft production, see pp. 264–71.

17. Ibid., pp. 189–92.

18. For another effort to link national institutions and sectoral analysis, see Atkinson, Michael and Coleman, William D., The State, Business and Industrial Change in Canada (Toronto: University of Toronto Press, 1989)CrossRefGoogle Scholar.

19. For example, Piore and Sabel's chapter on the crisis of “Fordist” mass production (The Second Industrial Divide, chap. 7) emphasizes changing consumer demand but mentions only in a single passing phrase (p. 191) that new flexible technologies may be economically more efficient than mass production systems. By neglecting considerations of market efficiency in the design of economic institutions that are exposed to evolutionary selection and learning in the marketplace, Piore and Sabel become vulnerable to the same criticisms that institutional economics scholars have advanced against power- and culture-based theories of the firm. For these criticisms, see Williamson, Oliver, The Economic Institutions of Capitalism (New York: Free Press, 1985), pp. 231–37Google Scholar.

20. For this basic assumption, see Stinchcombe, , Information and Organizations, pp. 3–8Google Scholar.

21. A somewhat more complex division among elements of a firm's “value chain” can be found in Porter's, The Competitive Advantage of Nations, p. 41Google Scholar.

22. In The Competitive Advantage of Nations, p. 33, for example, Porter defines a sector as a “group of competitors producing products or services that compete directly with each other.”

23. My account draws primarily on the following works of Perrow, Charles: Normal Catastrophes (New York: Basic Books, 1984), chap. 3Google Scholar; and Complex Organizations: A Critical Essay, 3d ed. (New York: Random House, 1986), pp. 146–54Google Scholar.

24. See Williamson, The Economic Institutions of Capitalism.

25. A review of the older literature can be found in Perrow's, Complex Organizations, pp. 140–46Google Scholar. The most influential contribution to these “contingency theories” of organization was Thompson's, James D.Organizations in Action (New York: McGraw Hill, 1967)Google Scholar, which distinguished mediating, long-linked, and intensive technologies.

26. A logic of transaction costs was already the underlying driving concern in the older contingency theories of technology and organization, such as Thompson's, Organizations in Action, p. 57Google Scholar. Interestingly, Perrow himself does not recognize his proximity to economic theories of organization that emphasize transaction costs and in fact attacks Oliver Williamson for placing undue emphasis on transaction costs at the expense of power and domination. See Perrow, , Complex Organizations, pp. 236–47Google Scholar. In reality, however, Perrow's own theory of the interface between technology and efficient organizational forms does not rely on power and domination as a structuring principle of social institutions. Nevertheless, power and domination can be brought in as a constraint on the efficient structuring of technology-organization interfaces. The role of domestic and international regimes in shaping governance structures that diverge from efficient arrangements can be interpreted in this way, as I will argue below.

27. The governance of innovation, of course, also follows an internal trajectory. Although most technological systems start out with revolutionary invention, they often move down a learning curve toward incremental improvements. Nevertheless, even if we hold the maturity of a technology constant, critical differences between more causally complex and less causally complex systems remain.

28. For an argument about the management of research and development that is similar yet lacks the analytic elaboration and detailed comparative empirical evidence which Perrow provides, see the following works of Collingridge, David: The Social Control of Technology (New York: St. Martin's Press, 1980)Google Scholar; and Technology in the Policy Process: Controlling Nuclear Power (New York: St. Martin's Press, 1983)Google Scholar.

29. Williamson, Oliver, The Economic Institutions of Capitalism (New York: Free Press, 1985)Google Scholar.

30. For a discussion of the three contingencies, see ibid., pp. 52–61.

31. See ibid., pp. 95–98.

32. See ibid. Williamson's third condition subverting standard contracting—the frequency of interaction between suppliers and customers and the accumulation of insider's knowledge—may be more indirectly related to technological conditions. Complex interactive and tightly coupled production systems involve frequent interactions between unique actors. Moreover, interactive systems may facilitate innovation through close relations of mutual trust that allow suppliers and customers to learn from each other. See also Lundvall, Bengt-Ake, “Innovation as an Interactive Process: From User-Producer Interaction to the National System of Innovation,” in Dosi, Giovanni et al. , eds., Technical Change and Economic Theory (London: Pinter, 1988), pp. 349–69Google Scholar.

33. The importance of “learning by doing” in innovation strategies has been pointed out by Rosenberg, Nathan in Inside the Black Box (New York: Cambridge University Press, 1982), chap. 6Google Scholar.

34. See Kurth, , “Industrial Change and Political Change,” pp. 330–46Google Scholar.

35. For an analysis of the conditions conducive to normative integration, see Ouchi, William G., “Markets, Bureaucracies and Clans,” Administrative Science Quarterly 25 (Winter 1980), pp. 129–41CrossRefGoogle Scholar. Hirschman's notion of “loyalty” as the basis of coordination hints at a corresponding form of social coordination but is not sufficiently developed in his writings. See Hirschman, Albert O., Exit, Voice, and Loyalty (Cambridge, Mass.: Harvard University Press, 1970), chap. 7Google Scholar.

36. Actually, Williamson calls systems with inseparable, uncertain task structures and low asset specificity “primitive teams,” whereas the notion of “relational teams” or “clans” is reserved for configurations in which asset specificity is high (cell 2 in Figure 2). A problem here is the measurement of asset specificity. Williamson is concerned only with human asset specificity as a determinant of governance structures, whereas Perrow includes all human and nonhuman components of technological systems under the notion of loose or tight coupling. The knowledge of each individual software architect is highly asset-specific vis-à-vis the joint effort of the firm to develop a new program. The technology of writing software programs, however, is not highly asset-specific, since it can be redeployed from program to program. Investments in any individual product remain limited. When Williamson discusses physical asset specificity, he sees organizational hierarchy as a consequence. See Williamson, , The Economic Institutions of Capitalism, pp. 213–14 and 242–47Google Scholar; and Perrow, Complex Organizations.

37. See Lundvall, , “Innovation as an Interactive Process,” especially pp. 351–53Google Scholar. Because causal connections among system components are ill-understood, sociologists of science have hypothe-sized a changing logic of research from a causal analysis of mechanisms to a functional analysis of outcomes. For this “finalization” of research, see Schaefer, Wolf, ed., Finalization in Science: The Social Orientation of Scientific Progress (Dordrecht, Netherlands: Reidel, 1983)Google Scholar.

38. See Dasgupta, Partha, “The Economic Theory of Technology Policy: An Introduction,” in Dasgupta, Partha and Stoneman, Paul, eds., Economic Policy and Technological Performance (Cambridge: Cambridge University Press, 1987), pp. 15–16CrossRefGoogle Scholar.

39. In principle, uncertainty and asset specificity are constraints that increase the probability of nonstandard contracting and organizational integration. But up to what point do they do so? The development and control of complex interactive technology requires professionals, and as Williamson, acknowledges in The Economic Institutions of Capitalism, pp. 242–47Google Scholar, it is difficult to monitor professionals in strict hierarchies. Hence, integrated organizations may rely on “fiercely egalitarian,” collegial practices that are designed to promote trust among their members. At the same time, however, asset specificity in large-scale technological systems may render it impractical to confer organizational control on clans of professionals, rather than to submit them to centralized authority. It is not clear whether Williamson can identify an optimal form of organization. In Complex Organizations, chap. 7, Perrow criticizes Williamson on a number of occasions for falling victim to a functionalist logic which does not specify the mechanisms that determine the efficiency of a particular governance structure. This criticism applies with full force to configurations of large-scale technological systems with high asset specificity and uncertainty.

40. It would be mistaken, however, to assume that governments are always less risk-averse than private investors. For a critique of this assumption, see Stoneman, Paul, The Economic Analysis of Technology Policy (Oxford: Oxford University Press, 1983), pp. 118–24Google Scholar.

41. This is essentially the logic that also underlies Piore and Sabel's analysis of the capacity of advanced industrial economies to move toward flexible specialization. See The Second Industrial Divide, chap. 9.

42. It is not countries that succeed or fail in innovation policies; it is particular sectors. This point is emphasized by Porter, in The Competitive Advantage of Nations, p. 144Google Scholar.

43. Thus, the diversity of American success in industries with entirely different technological characteristics (entertainment, financial services, key areas of microelectronics, and aerospace) has to do with the regional variation of governance structures, a point emphasized by Piore, and Sabel, in The Second Industrial Divide, pp. 248–49Google Scholar, and by Porter, in The Competitive Advantage of Nations, pp. 507–35Google Scholar. Conversely, the emergence of encompassing corporatist governance structures has often been attributed to the narrow scope of relevant industrial sectors in small European democracies. See Katzenstein, , Small States in World Markets, pp. 165–70Google Scholar.

44. Here, the idea of a “punctuated equilibrium” adapted to social science explanations can serve as a useful concept. See Krasner, Stephen, “Approaches to the State: Alternative Conceptions and Historical Dynamics,” Comparative Politics 16 (01 1984), pp. 240–44CrossRefGoogle Scholar.

45. The notion of “technological trajectories” is taken from Dosi. See Dosi, Giovanni, “Technological Paradigms and Technological Trajectories: A Suggested Interpretation of the Determinants and Directions of Technical Change,” Research Policy 11 (Spring 1982), pp. 147–62CrossRefGoogle Scholar.

46. As I already argued in my critique of Piore and Sabel's The Second Industrial Divide, domestic structuralism may endogenize technology and sectoral governance structures too much. The international systemic conditions of competition and the brute opportunities of technology shape institutional responses as well. Unquestioned international hegemony may stifle any form of learning to innovate, as the example of the Chinese armament industry illustrates. See McNeil, William, The Pursuit of Power (Chicago: University of Chicago Press, 1982), chap. 2Google Scholar. Conversely, domestic vulnerability may accelerate a country's willingness to engage in institutional and technological innovation. For a general review of these linkages, see Gourevitch, Peter, “The Second Image Reversed: The International Sources of Domestic Politics,” International Organization 32 (Autumn 1978), pp. 281–313CrossRefGoogle Scholar.

47. My argument diverges from the conventional wisdom that the United States is unable to engage in effective industrial policy. At least for some sectors in which appeals to the “national interest” can serve in part as a smoke screen behind which massive sectoral development aid is disbursed, American industrial policy has been effective. Precedents include not only nuclear power and aerospace but also biotechnology (labeled cancer research), microelectronics, and a host of technologies covered by the strategic defense initiative (SDI) program of the 1980s. For a cogent review of U.S. industrial policy activism, see Vogel, David, “Government-Industry Relations in the United States: An Overview,” in Wilks, and Wright, , Comparative Government-Industry Relations (Oxford: Clarendon Press, 1987), pp. 91–116Google Scholar.

48. I draw here on the following works: Landes, David, The Unbound Prometheus (Cambridge: Cambridge University Press, 1969)Google Scholar; Freeman, Christopher, Clark, John, and Soete, Luc, Unemployment and Technical Innovation: A Study of Long Waves and Economic Development (Westport, Conn.: Greenwood Press, 1982)Google Scholar; and Freeman, Christopher, Technology Policy and Economic Performance: Lessons from Japan (London: Pinter, 1987), especially the summarizing historical typology on pp. 68–75Google Scholar. I do, however, alter Freeman's interpretations in some instances. I will refrain here from placing the surge of technologies with new qualities within the framework of Kondratieff cycles, as Freeman and others do. I find the idea of Kondratieff cycles useful, particularly in the weak version which focuses on lead countries and lead technologies and which was recently advanced by Thompson, William R. in “Long Waves, Technological Innovation, and Relative Decline,” International Organization 44 (Spring 1990), pp. 201–33CrossRefGoogle Scholar. Since Kondratieff cycles are not central for my article, I will follow the advice of an anonymous reviewer and refrain from burdening it with yet another controversial argument.

49. Here, I am following Kurth's, argument about technology and political regime structures in “Industrial Change and Political Change,” pp. 326–51Google Scholar.

50. For a comparison of British and German industrial innovation, see Landes, , The Unbound Prometheus, pp. 326 and 358Google Scholar. For a comparison of the broader industrial consequences of institutional conditions in the United States, Britain, and Germany, see Chandler, Alfred, Scale and Scope (Cambridge: Cambridge University Press, 1970), chaps. 2–4Google Scholar.

51. The modern multidivisional enterprise has been and continues to be prevalent primarily in sectors that emerged in the nineteenth and early twentieth centuries: food processing, chemicals, petroleum, primary metals, and various machine groups. See Chandler, , Scale and Scope, p. 20Google Scholar.

52. The United States, for example, acquired new capabilities following the “Sputnik shock” in 1957.

53. In France, where institutional capabilities and the lessons of World War II reinforced industrial innovation policy, industrial success has been greatest in technologies that are highly dependent on government development and procurement. See Stoffaes, Christian, “Industrial Policy in the High-Technology Industry,” in Adams, William J. and Stoffaes, Christian, eds., French Industrial Policy (Washington, D.C.: Brookings Institution, 1986), p. 45Google Scholar. Other important French export industries, such as specialty foods, clothing, and design, are of the crafts type and originated in the earliest wave of industrialization. France has always remained weak in mark II and III technologies because it had weak corporations or nonmarket networks among smaller firms.

54. See Chandler, , Scale and Scope, pp. 622–23Google Scholar.

55. According to Amendola and Gaffard, “The essence of the process of production, therefore, is no longer embodied in devices and equipment, but lies in the characteristics of the specific inputs involved which themselves contribute to defining the profile of the process and its effective articulation contextually.” See Amendola, Mario and Gaffard, Jean-Luc, The Innovative Choice: An Economic Analysis of the Dynamics of Technology (Oxford: Blackwell, 1988), p. 13Google Scholar.

56. There is reason to believe that a number of biotechnologies, especially in the field of genetic engineering, are mark IV (tight coupling and complex interaction) rather than mark V (loose coupling and complex interaction). As Perrow argues in Normal Catastrophes, pp. 293–302, these biotechnologies involve uncertain damage potentials for the environment which follow from tightly coupled processes that are irreversible once set into motion. This increases capital cost and bureaucratic management of innovation not so much in the initial stages of innovation but in comprehensive testing and safety research on new organisms. Because small venture capital firms usually cannot raise the capital and conduct long-term safety testing, many genetic engineering upstarts have been bought out by major pharmaceutical companies.

57. For the purposes of my argument, it is unimportant to examine the precise validity and comparability of these indicators. As is well known, Japanese companies have internal incentive systems that inflate the number of patent applications. See Okimoto, Daniel I. and Saxonhouse, Gary R., “Technology and the Future of the Economy,” in Yamamura, Kozo and Yasukichi, Yasuba, eds., The Political Economy of Japan, vol. 1 (Stanford, Calif.: Stanford University Press, 1987), pp. 390–91Google Scholar. Moreover, the definition of “high-technology commodities” is disputable as a measure of success. Regarding this point, see Gahlen, Bernhard, Rohmeyer, Fritz, and Stadler, Manfred, “Zur internationalen Wettbewerbsfähigkeit der deutschen Wirtschaft” (The competitiveness of the German economy), discussion paper no. 85–19, International Institute of Management and Administration, Wissenschaftszentrum Berlin, 1985Google Scholar.

58. For the free market argument, see Patrick, Hugh and Rosovsky, Henry, eds., Asia's New Giant: How the Japanese Economy Works (Washington, D.C.: Brookings Institution, 1976)Google Scholar; Schultze, Charles L., “Industrial Policy: A Dissent,” Brookings Review, Fall 1983, pp. 3–12Google Scholar; and Tretize, Philip H., “Industrial Policy in Japan,” in Devvar, Margaret E., ed., Industrial Vitalization: Toward a National Industrial Policy (New York: Pergamon Press, 1982), pp. 177–95Google Scholar.

59. See Kikkawa, Mottotada, “Shipbuilding, Motor Cars, and Semiconductors: The Diminishing Role of Industrial Policy in Japan,” in Shepherd, Geoffrey et al. , eds., Europe's Industries, pp. 236–68Google Scholar; and Eads, George C. and Yamamura, Kozo, “The Future of Industrial Policy,” in Yamamura, and Yasuba, , The Political Economy of Japan, vol. 1, pp. 423–68Google Scholar.

60. See Pavitt, Keith and Soete, Lue L. G., “International Differences in Economic Growth and the International Location of Innovation,” in Giersch, Herbert, ed., Emerging Technologies: Consequences for Economic Growth, Structural Change, and Employment (Tuebingen: Mohr, 1982), pp. 105–33Google Scholar; Fagerberg, Jan, “A Technology Gap Approach to Why Growth Rates Differ,” Research Policy 16 (Spring 1987), pp. 87–99CrossRefGoogle Scholar; and Fagerberg, Jan, “Why Growth Rates Differ,” in Dosi, et al. , Technical Change and Economic Theory, pp. 432–57Google Scholar.

61. The classic study of this topic is Johnson's, ChalmersMITI and the Japanese Miracle: The Growth of Industrial Policy, 1945–1975 (Stanford, Calif.: Stanford University Press, 1982)Google Scholar.

62. This interpretation was proposed by Piore, and Sabel, in The Second Industrial Divide, pp. 217–20 and 223–26Google Scholar, and has been fleshed out by Friedman, David in The Misunderstood Miracle: Industrial Development and Political Change in Japan (Ithaca, N.Y.: Cornell University Press, 1988)Google Scholar.

63. Regarding this argument, see Dore, Ronald, Taking Japan Seriously: A Confucian Perspective on Leading Economic Issues (Stanford, Calif.: Stanford University Press, 1987)Google Scholar; Freeman, Technology Policy and Economic Performance; and Nelson, Richard R., High-Technology Policies: A Five-Nation Comparison (Washington, D.C.: American Enterprise Institute, 1984)Google Scholar. The most detailed interpretation along these lines, providing an impressive range of evidence and analytic elaboration, is Okimoto's, Daniel I.Between MITI and the Market: Japanese Industrial Policy for High Technology (Stanford, Calif.: Stanford University Press, 1989)Google Scholar. As states, Okimito, however, he goes beyond the “synthetic literature and emphasizes the sectoral variability of Japanese success” (p. 4)Google Scholar.

64. To be fair, Okimoto's Between MITI and the Market makes some effort to identify sectoral patterns (see especially pp. 193–206). Similarly, Okimoto and Saxonhouse's “Technology and the Future of the Economy” makes an attempt to identify Japan's sectoral strengths and weaknesses, but it does not provide a theory of why Japanese success has varied across industrial sectors.

65. See Johnson, , MITI and the Japanese Miracle, pp. 307–12Google Scholar; and Okimoto, and Saxonhouse, , “Technology and the Future of the Economy,” pp. 407—9Google Scholar.

66. See Friedman, The Misunderstood Miracle; and Piore, and Sabel, , The Second Industrial Divide, pp. 216–20 and chap. 9Google Scholar. Although Piore and Sabel identify some interesting variance across countries within broadly defined industries, they make no systematic comparison of sectoral fortunes and development strategies within countries.

67. See Tyson, Laura and Zysman, John, “American Industry in International Competition,” in Tyson, Laura and Zysman, John, eds., American Industry in International Competition: Government Policies and Corporate Strategies (Ithaca, N.Y.: Cornell University Press, 1983), p. 37Google Scholar. See also Okimoto, , Between MITI and the Market, p. 25Google Scholar. For a similar assessment that Japan's industries do not perform well with highly customized products involving heavy after-sale support and small lot sizes, see Porter, , The Competitive Advantage of Nations, p. 411Google Scholar.

68. Porter, , The Competitive Advantage of Nations, p. 394Google Scholar.

69. My general characterization of Japanese institutional resources is based on the following works: Pempel, T. J., Policy and Politics in Japan (Philadelphia: Temple University Press, 1982), especially chaps. 1 and 2Google Scholar; Dore, Taking Japan Seriously; and Okimoto, , Between MITI and the Market. Within this pattern, Okimoto (pp. 193–206)Google Scholar distinguishes three different patterns of political exchange.

70. Much of Japanese innovation has been in the area of process innovation rather than product innovation. See Mansfield, Edwin, “Industrial Research and Development in Japan and the United States: A Comparative Study,” American Economic Review 78 (05 1988), pp. 223–38Google Scholar. On Japanese strategies in mass production industries, see also Porter, , The Comparative Advantage of Nations, p. 409Google Scholar; and Economist, 12 January 1991, p. 61.

71. My empirical reference for this assessment is Porter's, The Competitive Advantage of Nations, pp. 385–94Google Scholar.

72. According to Okimoto and Saxonhouse, “Progress has been slower in technologies where the theoretical parameters for problem solving are highly complex (jet aircraft design) and technological trajectories are not readily predictable (advanced software). Japanese firms are not as apt to make seminal inventions that lead to the creation of whole new industries, owing in the past to the relatively low level of government sponsorship and the (until recently) narrowly applied nature of much commercial R&D.” See Okimoto, and Saxonhouse, , “Technology and the Future of the Economy,” pp. 396–97Google Scholar.

73. The early preemption of technological alternatives in nuclear reactor development through determined state funding of light water reactors is analyzed and compared across Western countries in Bupp, Irvin C. and Derian's, Jean-ClaudeLight Water: How the Nuclear Dream Dissolved (New York: Basic Books, 1978)Google Scholar.

74. The success of nuclear power has been erroneously attributed to markets by Burn, DeLeon, and Keck. See Burn, David, Nuclear Power and Its Critics (London: Macmillan, 1978)Google Scholar; DeLeon, Peter, Development and Diffusion of the Nuclear Power Reactor (Cambridge, Mass.: Ballinger, 1979)Google Scholar; and Keck, Otto, Policy-Making in a Nuclear Program (Lexington, Mass.: Lexington Books, 1981)Google Scholar.

75. For a good account of industry-government conflicts over nuclear power, see Samuels, Richard J., The Business of the Japanese State: Energy Markets in Comparative Historical Perspective (Ithaca, N.Y.: Cornell University Press, 1987), pp. 234–45Google Scholar.

76. In 1989, the availability of Japanese commercial reactors was 72.3 percent, while the availability of American reactors was 63.3 percent. See The New York Times, 28 February 1990, pp. Cl and C17.

77. For a discussion of Japan's growing disadvantage in aerospace development, see Patel, Pari and Pavitt, Keith, “Is Western Europe Losing the Technological Race?” Research Policy 16 (Spring 1987), pp. 59–85CrossRefGoogle Scholar.

78. For a comparison of American and Japanese aircraft innovation, see Mowery, David and Rosenberg, Nathan, “Government Policy, Technical Change, and Industrial Structure: The U.S. and the Japanese Commercial Aircraft Industry,” in Langdon, Richard and Rothwell, Roy, eds., Design and Innovation: Policy and Management (New York: St. Martin's Press, 1978), pp. 71–100Google Scholar.

79. Regarding Boeing's decision to cut back on Japanese participation in the 767–X project, see The New York Times, 14 April 1990, p. 1.

80. See “Thinking Ahead: A Survey of Japanese Technology,” Economist, 1 December 1989, p. 14.

81. The following comment on Japan's technology program reflects this problem: “Japan's problems with aerospace run deeper than engineering. Its engineers are as good as any in the world; their handicap is lack of a culture that continuously challenges all design decisions, no matter how high in the hierarchy they were made. Without it, debugging aircraft design is impossible.” See Economist, 8 April 1989, p. 72.

82. For studies of Japanese semiconductor and computer development, see especially Flamm, Kenneth, Targeting the Computer: Government Support and International Competition (Washington, D.C.: Brookings Institution, 1987), pp. 125–53Google Scholar; Flamm, Kenneth, Creating the Computer: Government, Industry, and High Technology (Washington, D.C.: Brookings Institution, 1988), pp. 172–202Google Scholar; and Okimoto, Between MITI and the Market, especially chap. 2.

83. Interestingly, MITI's efforts to impose a stronger hierarchical structure on research efforts and draw activities from company laboratories into central facilities shared by several companies were strongly resisted, as was evident in the very large scale integration (VLSI) project. See Fong, Glenn R., “State Strength, Industry Structure, and Industrial Policy: American and Japanese Experiences in Microelectronics,” Comparative Politics 22 (04 1990), pp. 273–99CrossRefGoogle Scholar.

84. In a similar vein, Japanese manufacturers managed to perfect the so-called Josephson junction technology (the ultrafast semiconductor switching devices on which IBM and ATT made initial breakthroughs), but they were too impatient to fund further work in the early 1980s. This rival to conventional silicon technology was developed with $190 million of MITI funding. See Economist, 21 July 1990, p. 87.

85. Because the costs of developing new DRAMs and setting up production facilities have escalated with each generation of new devices, competition among Japanese manufacturers has become ruinous and may discourage further private developments. In the mid–1980s, it was the U.S. protectionist policy of imposing minimum prices on DRAMs sold in America that unintentionally saved Japan's industry from collapse. In the early 1990s, the new round of price cutting in 1- and 4-megabyte DRAMs appears to be leading the industry back to the brink of catastrophe. Further innovation may therefore require more state involvement, as American manufacturers have claimed for some time. See Economist, 23 February 1991, pp. 64–65; and The New York Times, 21 February 1991, p. C3. In “State Strength, Industry Structure, and Industrial Policy,” p. 294, Fong notes that there appears to be a convergence of governance structures in the American and Japanese semiconductor industries. If this is the case, the convergence is on a higher level of public involvement than in the past.

86. Toshiba has recently become willing to swap its DRAM technology for access to some of Motorola's microprocessor know-how. See Economist, 2 March 1990, p. 66. But Motorola and Intel have remained unwilling to license their most advanced microprocessors to Japanese companies.

87. Cray computers with 16 processor units will be on the market in the mid-1990s, and supercomputers with up to 256 processors are under development. See Economist, 15 April 1989, pp. 91–92. Computers with thousands of parallel processors are being developed by other firms, such as Connection Machines, which operates under contracts from the Defense Advanced Research Projects Agency (DARPA).

88. See The New York Times, 29 April 1990, p. 1.

89. See Patel, and Pavitt, , “Is Western Europe Losing the Technological Race?” p. 73Google Scholar; and Economist, “Thinking Ahead: A Survey of Japanese Technology,” pp. 12–13.

90. Regarding the HDTV competition between Japan and the United States, see International Herald Tribune, 22 03 1989, p. 1Google Scholar; and The New York Times, 28 November 1989, pp. 19 and 22.

91. Until recently, the Ministry of Health and Welfare was regulating competition and cutting prices rather than stimulating R&D and international competitiveness in the Japanese drug industry.

92. See Economist, 2 March 1991, pp. 61–62.

93. This reflects the broader problem of weak basic research in Japan and little government-industry interpenetration. See Okimoto, and Saxonhouse, , “Technology and the Future of the Economy,” pp. 403 and 412–13Google Scholar.

94. See Economist, 16 December 1989, p. 84.

95. See Sharp, Margaret, “Biotechnology: Watching and Waiting,” in Sharp, Margaret, ed., Europe and the New Technologies, pp. 179–81Google Scholar.

96. Regarding the shifting balance between basic research and development in Japanese industry, see ibid., p. 225. This shift was evident, for example, in Hitachi's 1985 decision to open an advanced research facility modeled on that of Bell Laboratories. See Economist, 11 August 1990, p. 82.

97. See Okimoto, , Between MITI and the Market, pp. 80–81Google Scholar.

98. In The Business of the Japanese State, Samuels reports similar problems in innovation programs in the energy sector, where the technological properties of the systems (tight coupling and often complex interaction) frequently place them out of the realm in which horizontal cooperative strategies are successful. Conflicts also surfaced in semiconductor research whenever MITI attempted to centralize authority over program activities. See Fong, , “State Strength, Industry Structure, and Industrial Policy,” pp. 290–93Google Scholar.

99. See Arnold, Erik and Guy, Ken, Parallel Convergence: National Strategies in Information Technology (Westport, Conn.: Quorum Books, 1986), pp. 86–87Google Scholar.

100. DARPA, for example, was responsible for the initial research on personal computers, computer graphics, computer networks, and artificial intelligence. With DARPA funding, scientists from Stanford and Berkeley launched Sun Microsystems, Inc., a major innovative computer company of the late 1980s. See The New York Times, 19 November 1989, p. E4. For a discussion of U.S. industrial policy in the information and computer area, see Arnold and Guy, Parallel Convergence, chap. 3; and Flamm, Targeting the Computer, chaps. 3 and 4. The general case for the existence of an interventionist U.S. industrial policy has been made convincingly by Vogel in “Government-Industry Relations in the United States.”

101. See, for example, Freeman, , Technology Policy and Economic Performance, pp. 78–79Google Scholar.

102. This argument can be derived from Patel and Pavitt's comprehensive overview of industrial strengths in the three blocs: “Is Western Europe Losing the Technological Race?”

103. The problems of the nuclear power industry are, however, primarily due to public safety concerns, which proved to be politically more effective in the United States than in almost any other advanced capitalist democracy. See Kitschelt, Herbert, “Political Opportunity Structures and Political Protest: Anti-Nuclear Movements in Four Countries,” British Journal of Political Science 16 (Winter 1986), pp. 57–86CrossRefGoogle Scholar.

104. For a discussion of free market conditions, see Porter, , The Competitive Advantage of Nations, pp. 384–421Google Scholar. Compare the discussion in Information and Organization, pp. 173–89, where Stinchcombe emphasizes that innovations are ultimately successful only if they affect more than the engineering component of a production system.

105. Problems of Perrow's typology include those related to the determination of technological system boundaries and to the operational measures of tight coupling and causal complexity. Perrow provides little help in addressing the first problem, yet he offers considerable elaboration of the second problem in a wealth of highly instructive case studies.

106. In a corresponding critique of historical functionalism, March and Olsen postulate that efficiency rationales govern the choice of institutions. See March, James G. and Olsen, Johan P., “The New Institutionalism: Organizational Factors in Political Life,” American Political Science Review 78 (09 1984), p. 737Google Scholar. Nevertheless, the authors may go too far in rejecting the occurrence of efficient learning. A substantive policy theory would have to identify the opportuni-ties and limitations of efficient learning.

107. On this point with respect to decision theory, see Dawes, Robyn M., Rational Choice in an Uncertain World (New York: Harcourt, Brace, Yovanovich, 1988), pp. 100–120Google Scholar; and Levitt, Barbara and March, James G., “Organizational Learning,” Annual Review of Sociology, vol. 14, 1988, pp. 319–40CrossRefGoogle Scholar.

108. I owe most of these suggestions to Timothy McKeown.

109. Elster, Jon, Making Sense of Marx (Cambridge: Cambridge University Press, 1985), p. 487Google Scholar.