Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 187
V. National Technology Policies and International Friction: Theory, Evidence, and Policy Options This section of the Report addresses the theory and empirical evidence of national technology policy in globalized markets, and it deals with approaches to multilateral conflict resolution in selected policy areas giving rise to international friction: subsidies and public procurement, market access and structural impediments, and dumping, antidumping, and competition poli- cies. Prepared as an analytical basis for deliberations of the Steering, Com- mittee, it is intended to give a theoretical and empirical underpinning to the Committee's Recommendations and Findings. Economic theory and empirical evidence suggest that governments could usefully intervene in high-technology in two ways. First, they could act as a neutral agent that creates the necessary credibility, commitment, arid mutual trust among private companies to facilitate their cooperation in hiah-risk, high-volume R&D. Second, if in view of the externalities involved an element of subsidization is to be added, this should be done in a nondis- criminatory fashion. A favorable tax treatment of R&D may be the most appropriate tool to achieve this task. In practice, whether justified on eco- nomic grounds or not, governments do engage in targeted industrial and technology policies. This gives rise to recurrent trade friction. To resolve that friction this section of the Report suggests, among other measures, an open subsidies club where access to the public funding of *Section V, copyright 1996 by the Hamburg Institute for Economic Research and the Kiel Institute for World Economics. All rights reserved. 187
OCR for page 188
88 CONFLICT AND COOPERATION high-technolo~y development should be open to domestic as well as foreign firms on a reciprocal basis. To overcome structural impediments to high- technology trade and investment, the multilateral route offered by the so- called nonviolation clause of Article 23 GATT should be given preference over unilateral approaches. In order to reduce the potential for protectionist misuse of the antidumping instrument, proceedings should make explicit reference to the state of competition in the relevant exporting, and importing country markets, thus introducing elements of competition policy into the proceedings. To meet specific antitrust concerns in high-technology, relat- ing inter alla to network externalities, systems leverage, standardization and innovation cartels, the paper supports the Draft International Antitrust Code (DIACJ proposed by an international group of legal experts. The section was jointly produced by staff members of the Kiel Institute of World Economics (IfW) and the HWWA Institute for Economic Re- search. The first chapter, The Economics of Technology Policy in Globalized Markets, and the first part of the second chapter, Approaches to Conflict Resolution: Agenda for Action, were written by Karl-lIeinz Paque, Jurgen Stehn, and Ernst-Jurgen Horn of IfW. The latter two sections of chapter 2 were written by Hans-Eckart Scharrer and Georg Koopmann of HWWA. The second section draws on the papers and discussions at the three confer- ences held in Hamburg, Washington, D.C., and Kiel, and it greatly ben- efited from the discussions of the Steering Committee. The authors were also primarily responsible for the scientific programs of the two German conferences and the editing of the proceedings, which are published as separate volumes.* Their collaborative effort, in partnership with their counterparts at the NRC, has been critical to the success of this joint ven- ture in transatlantic research cooperation. Erhard Kantzenbach Project Co-Chazrman *See George Koopman and Hans-Eckart Scharrer (eds.), The Economics of High-Technology Competition and Cooperation in Global Markets. Baden-Baden, 1996, and Horst Siebert (ed.), Towards a New Global Framework for High-Technology Competition. Tubingen, forthcoming.
OCR for page 189
V. JOINT HWWA-IFW ANALYSIS: NATIONAL TECHNOLOGY POLICIES AND INTERNATIONAL FRICTION: THEORY, EVIDENCE, AND POLICY OPTIONS THE ECONOMICS OF TECHNOLOGY POLICY IN GLOBALIZED MARKETS 190 Analytical Benchmarks, 190 Endogenous Growth Theory, 190 Strategic Trade Theory, 193 Empirical Evidence, 194 External Learning and Growth, 195 The Case of the Semiconductor Industry, 195 Knowledge Diffusion: International and Interregional, 198 Strategic Trade Policy: The Aircraft Industry, 200 Technology Policy: Some Cautious Conclusions, 202 APPROACHES TO CONFLICT RESOLUTION: AGENDA FOR ACTION 206 Subsidies and Public Procurement, 206 Toward an Open Subsidy Club, 206 Public Procurement: Conflict Resolution by National Courts, 210 Market Access and Structural Impediments, 215 Tariffs and Nontariff Barriers, 215 Technical Norms and Standards, 221 Dumping, Antidumping, and Competition Policies, 223 Private versus Public Conduct in Trade and Competition, 223 Dumping and Antidumping in Hi=,h-Technology, 226 The Uruguay Round Approach to Antidump~n~,, 228 Reforming Antidumping Policy, 231 Toward Multilateral Competition Rules, 233 REFERENCES 237 189
OCR for page 190
190 CONFLICT AND COOPERATION The Economics of Technology Policies in GIobalized Markets ANALYTICAL BENCHMARKS To a large extent, international frictions in markets for high-technology products are the consequence of some form of unilateral state intervention. Using a large variety of instruments, many governments grant support to specific branches of economic activity whose growth is deemed to be ben- eficial or even crucial for the long-term performance of the national economy. The policy toolbox ranges from overt protectionism and subsidization to more covert barriers to market entry such as discrimination in public pro- curement, product standards, and distributional networks. Whatever their specific shape may be, the instruments are usually part of a more or less coherent growth strategy that is well summed up under the terms "technol- ogy policy" or "industrial policy.") Many of the more popular arguments for technology policies in high- technology markets can easily be refuted with standard economic reasoning and are therefore no serious candidates for further scrutiny.2 The more powerful case for technology policy and the one that is most likely to dominate future debate on trade frictions in high-technolo=,y markets comes from two modern strands of economic research that are both analytically important and politically relevant: endogenous growth theory and strategic trade theory. Endogenous Growth Theory Some strands of endogenous growth theory-notably the pioneering work by Grossman and Helpman (1991) and by Rivera-Batiz and Romer (1991a, l991b)3-have focused on the link between international trade integration ~ We use these terms interchangeably in the following paragraphs, with both terms de- scribino a government policy that grants support to a (specific) high-technology industry. 2 A standard undergraduate textbook on international trade P. Krugman, and M. Obstfeldt, International Economics: Theory and Policy, 3rd ed. New York, 1994, pp. 278-281-does an excellent job in refuting some common errors and misapprehensions. 3 G.M. Grossman and E. Helpman, Innovation and Growth in the Global Economy, Cam- bridge, Mass., 1991; F.L. Rivera-Batiz, P.M. Romer, '`Economic Integration and Endogenous Growth,' Quarterly Journal of Economics, vol. 106 (1991) pp. 531-556, and "International Trade with Endogenous Technological Change," European Economic Review, vol. 35 (1991), pp. 971-1004. For a summary evaluation of the policy relevance of the new growth theories, see also F.L. Rivera-Batiz, The Economics of Technological Progress and Endogenous Growth in Open Economies. Paper presented at the conference The Economics of High-Technology
OCR for page 191
THE ECONOMICS OF TECHNOLOGY POLICIES IN GfOBAf/ZED MARKETS 191 and the concomitant chances of sectoral specialization patterns on one side and the long-run growth prospects of a country on the other. This work contains a vast array of rather complex analytical insights that may also be of interest for selected questions of economic policy. With some coura- geous simplifications, the policy gist of the work as far as it is relevant for technology policy in advanced economies-may be summarized as follows.4 International trade patterns in (free) high-technology markets are deter- mined by comparative advantages. More than in other markets, however, these comparative advantages may be affected by the resources that differ- ent economies devote to industrial research, i.e., to investment in the cre- ation of new knowledge. But not each and every piece of new industrial knowledge affects trade patterns; whether it does depends, most impor- tantly, on the subsequent diffusion of this knowledge, i.e., in economists' jargon, on the degree of localization of technological spillovers. In this respect, two polar cases deserve attention as conceptual benchmarks: knowledge as a global good and knowledge as a national good. If the spillovers are essentially global, i.e., if competitors In all relevant countries have access to any addition to the (common) knowledge pool, wherever it comes from, there will be no lasting effects of national research efforts on trade patterns, and one is back at standard explanations of trade in terms of national endowments, with ordinary production factors, notably labor, human capital, and physical capital. In these circumstances, national technology policy makes little economic sense because what it helps to create in terms of new knowledge will easily diffuse outside the country, without giving domestic producers a viable and persistent advantage over foreign competitors. If, on the other hand, knowledge spillovers remain geographically con- centrated and thus essentially national in scope, a cumulative process of what may be called "national learning" may then set in and drive a widen- ing competitive wedge between the respective national industry and the rest of the world. In the extreme, a country starting with a tiny and accidental technological lead may eventually dominate the relevant world market be- cause it profits alone and persistently from its own knowledge creation, Competition and Corporation in Global Markets held at the HWWA-Institute in Hamburg, 9-3 February 1995; and R. Baldwin, The Economics of Technological Progress and Endogenous Growth in Open Economies. Comment presented at the conference '`The Economics of High- Technology Competition and Cooperation in Global Markets," HWWA-Institute in Hamburg. 2-3 February 1995; L. Soete, Technology Policy and the International Trading System: Where Do We Stand ? Paper presented at the conference `'Towards a New Global Framework for High- Technology Competition," in Kiel, 30-31 August 1995. 4 A detailed survey of results is given by M. Stolpe, Technology and the Dynamics of Specialization in Open Economies, Kiel Studies 271, Tubingen, 1995, Section B. IV. pp. 41~7.
OCR for page 192
92 CONFLICT AND COOPERATION which allows it to lower costs, to raise quality levels, and to introduce new products and production processes. In these circumstances, a national tech- nology policy may well have a powerful rationale: if private producers do not make the socially optimal decisions in economists' jargon, if there is a market failure government may step in to initiate the virtuous high-tech- nology growth circle. Is the market likely to fail? Among the cases of potential market failure that have been identified in the literature, two stand out in importance, one focusing on positive externalities of private R&D spending,, the other on negative ones. The argument on positive externalities recognizes a ten- dency toward private underinvestment in R&D due to any positive spillover contributions to the (nationals stock of knowledge that cannot be appropri- ated in a private calculus of profit maximization. The government is then called upon to raise the level of research effort in the industry at hand. The one on negative externalities identifies a tendency toward ovennvestment due to inefficient parallel research: with a limited common pool of potential discoveries and innovations at any point in time, a successful innovation is likely to reduce the prospective commercial value of research efforts by other firms, a form of negative (pecuniary) externality that is not properly taken into account by private a=,ents.5 The government may then be called upon to reduce, or better, to bundle and focus research efforts so as to ensure a maximum expected social rate of return and a minimum dead- weiaht loss. Which instruments of intervention should the government use? In the case of positive externalities, the modern theories of endogenous growth recommend public support directly of R&D activities rather than general protectionist measures for the respective industrial branch, say, in the form of production subsidies or tariffs. This is in line with the more traditional welfare analysis of externalities in trade theory,6 though the underlying rationale of the result is somewhat different: while traditional theory wants the government to avoid static allocative losses, the modern theory wants it to avoid a (=,rowth-hindering) resource competition for skilled labor that may be used both in manufacturing and in research. E.g., production subsi- dies or tariff protection for the production of a hi~h-technolo~y good may induce highly qualified engineering personnel to move from research into production, thus increasing the cost of R&D and reducing the country's potential for growth. 5 This (pecuniary) externality is a typical result of models of (Schumpeterian) creative destruction as pioneered by P. Aphid and P. Howitt, "A Model of Growth through Creative Destruction," Econometrica, vol. 60 (1992), pp. 323-351. 6 See, e.g., the classical statement by M. Cordon, Trade Policy and Economic Welfare, Oxford. 1974, pp. 257-264.
OCR for page 193
THE ECONOMICS OF TECHNOLOGY POLICIES IN GLOBALIZED MARKETS 193 A similar argument for direct intervention applies in the case of negative externalities, although it has not yet been spelled out in detail in the litera- ture: if a "bundling" of R&D efforts is required to reduce inefficient paral- lel research, it should be done by allowing firms to cooperate so as to rationalize and coordinate some of their research investment, possibly under government auspices and encouragement. Strategic Trade Theory Some modern strands of trade theory beginning with the pioneering, work by Brander aIld Spencer (1983, 1985) and Dixit and Kyle (198517 have focused on the international rivalry for monopoly rents in world mar- kets that operate under conditions of highly imperfect competition, usually involving only very few producers from different countries. This rivalry can have the characteristic that possibly accidental initial advantages of one firm lead to high monopoly profits because potential competitors are de- terred from market entry by high start-up costs and/or the narrowness of the prospective market. Once again, the details of the relevant theoretical models are complex, but the case for government intervention in the foe of so-called strategic trade policy is simple and straightforward: by granting a temporary subsidy to a newcomer, the government may turn potential competitors into actual ones, thus breaking up a (quasi-)monopolistic market position of the domi- nant foreign producerks) and shifting at least some of the monopoly rents from one country to the other. If, in the end, national subsidy costs are lower than the gain In rents, then the active policy stance pays off from a national point of view. Note that, in the technology policy debate, the case for a strategic trade policy hardly comes neatly separated from arguments based on endo:,enous growth theory. E.t,., the case for the European Union's long-standin=, effort to break up the American quasi monopoly in the world market for commer- cial aircraft has been consistently justified on two grounds: shifting, rents across the Atlantic and laying the ground for the expansion of a hi:,h- technolo~y industry that was regarded as greatly beneficial in teas of internal and external learning processes for future economic growth in Eu- rope. In fact, the American lead in aircraft and related technologies was widely identified as being the consequence of a costly learnin;, process that any prospective European competitor would still have to go through. 7 J.A. Brander, B. J. Spencer, '`International R&D Rivalry and Industrial Strategy," Review of Economic Studies, vol. 50 (1983), pp. 707-722, and "Export Subsidies and International Market Share Rivalry," Journal of international Economics. vol. 16 (1985), pp. 83-100; A.K. Dixit and A.S. Kyle, "The Use of Protection and Subsidies for Entry Promotion and Deter- rence," American Economic Review, voh 75 (1985), pp. 139-152.
OCR for page 194
94 CONFLICT AND COOPERATION EMPIRICAL EVIDENCE What do we know about the prospects for a successful technology policy in actual reality? Or more specifically: what do we know about the extent to which there are cumulative processes of external learning that are geo- :,raphically concentrated? And what do we know about the extent to which there are monopoly rents to be shifted around internationally? It may not come as a surprise that any serious economist can hardly avoid answering all of these questions with "very little." Our current ignorance simply reflects the enormous problems of empirically isolating and measuring, the relevant phenomena, not to speak of identifying causal relationships. To be sure, there is a large literature of individual case studies, in which selected hi;,h-technology branches or markets are picked to demonstrate the success or failure of particular government measures. In the context of the bilateral trade disputes between the United States and Japan, some of these studies have received considerable public attention, notably the book by Tyson (19921.8 These studies suffer from major methodological deficien- cies: despite their wealth of information and interpretation, they offer nei- ther a theory-based empirical account of the diffusion of knowledge or the shiftin;, of rents, nor a reasonably specified counterfactual scenario that would allow pinning, down with some conceptual precision to what extent government intervention has in fact altered the path of economic history. The example of Japanese industrial policies is notorious in this respect.9 To qualify as a valuable piece of economic analysis, it is not sufficient to demonstrate that there was some intervention in some industrial branch and that producers in this branch were increasingly successful in world markets; rather, it has to be shown that the relevant path of events was sufficiently different from what could have been expected on the basis of factor endow- ments and learnin;, processes that went on ~nvw~v in~lenendentiv of ~ov ernment action. ~- ~I ~ i or-- -A -- ~- It is an a priori open question whether the remarkable world market penetration that Japanese industry achieved in the mass production of cam- eras, automobiles, and semiconductors was the "natural crop" of the spe- cific engineering skills that the Japanese education and training system tends to provide (like, say, the German or the Swiss one) or whether it was the ~ L.D. Tyson, Who's Bashing Whom ? Trade Conflict in High-TechrzoZogy Industries. Washington, D.C.. 1992. See also, most recently, S. Cohen and P.H. Chin, Tipping the Balance: Trade Conflicts and the Necessity of Managed Competition in Strategic Industries. Paper presented at the conference "Towards a New Global Framework for High-Technology Competetion, Kiel, 3~31Au;,ust 1995. 9 On Japanese industrial policies in the relevant periods, see K. Yamamura, "Caveat Emptor: The Industrial Policy of Japan,'' in P. Krugman (ed.), Strategic Trade Policy and the New International Economics, Cambridge, Mass., 1986.
OCR for page 195
THE ECONOMICS OF TECHNOLOGY POLICIES IN GLOBALIZED MARKETS 195 "artificial consequence" of smart technology policy initiatives- or both. To come closer to an answer, one would have to study carefully the dynam- ics of comparative advantages over time and across countries, testing for elements of path dependence and hysteresis in specialization patterns of production and trade,~° and linking this to government interventions. Clearly, the case studies available so far fall well short of this standard. Apart from the more descriptive case studies, however, there is a small but growing literature that makes a serious econometric attempt at identify- ing phenomena that are relevant for technology policymaking. In the fol- lowing paragraphs, we will provide a brief review of important recent pieces of empirical analysis- first of those concerning external learning effects and growth, and second of those concerning rent-shifting through strategic trade policy. ~ ~ External Learning and Growth There has been a large number of studies over the last three decades which, in one way or another, try to measure and quantify spillover effects of R&D. While many of them are flawed and subject to a variety of reservations, they do on the whole support the view that R&D spillovers are positive and quantitatively important. We shall focus strictly on those (relatively few and recent) studies that investigate the regional and sectoral incidence of learning effects and externalities, which is the core issue for the design of technology policies. The Case of the Semiconductor Industry In recent years, the semiconductor industry has been at center stage of the technology policy debate, not least because it is widely regarded as a strategic industry in the sense that the production of semiconductors in- volves strong learning effects (and thus cost reductions) over time, and that semiconductors are used as high-technology inputs in virtually all branches 10 An econometric step in this direction has been made by M. Stolpe, Technology and the Dynamics of Specialization in Open Economies, Kiel Studies 271, Tubingen, 1995, chaps. 3 and 4. 11 For a more extensive survey, see K.H. Paque, "Technologie, Wissen und Wirtschaftspolitik- Zur Rolle des Staates in Theorien des endogenen Wachstums," Die Wel~wirtschaft 3 (1995), pp. 237-253. 12 This is also the conclusion of the most comprehensive survey of this literature to date: Zvi Griliches, "The Search for R&D Spillovers," Scandinavian Journal of Economics, vol. 94 (1992), Supplement, pp. 2407.
OCR for page 196
96 CONFLICT AND COOPERATION of any modern economy.~3 Irwin and Klenow (1994a)~4 deliver the first serious econometric attempt to estimate the actual magnitude of internal and external learning effects as well as the geographic concentration of externalities for the semiconductor industry. On the basis of a large set of quarterly data for the period 1974-1992 and for roughly thirty semiconductor producers in the United States, Europe, and East Asia (Japan and South Korea), the study provides estimates of how strongly the product price depends on past production of semiconductors (1) in the firm itself, (2) in the country where the firm is located, and (3) in the world as a whole. In the absence of cost data, the product price is taken as a proxy for "dynamic" marginal production cost,~5 i.e., the marginal cost that takes into account the discounted value of all future cost reductions due to production experience. Cumulative past production is taken as a proxy for the cumulative production experience and thus for the level of technical knowledge, i.e., the stage of the learning process. Also, an explicit distinc- tion is made between cumulative experience within a production line-i.e., within the current generation of memory chips and the experience with earlier production lines, i.e., with older memory chip generations. The results of Irwin and Klenow (1994a) are remarkable and important. They find strong learning effects within each chip generation on average a 20 percent cost reduction, with a doubling of output over time. However, they find no significant learning effects between chip generations: any new quality stage of technological development in the form of a new generation of memory chips begins with a level playing field. They also find rather powerful externalities: additional output (and thus additional experience) of other firms leads to a cost reduction of roughly one-third of a corresponding output increase of the firm itself. What they do not find, however, is a significant difference between international and intranational externalities: knowledge spillovers as far as they exist appear to be undisturbed by national borders. Also, the authors of the study do not find any significant , ~ 1' See T. F. Bresnahan and M. Trajtenberg, General Purpose Technologies: "Engines of Growth"? NBER Working Paper No. 4148, National Bureau of Economic Research, Cam- bridge. Mass., August 1992; and E. Helpman and M. Trajtenber ,, A Time to Sow and a Time to Reap: Growth Based on General Purpose Technologies. NBER Working Paper No. 4854, National Bureau of Economic Research. Cambridge, Mass., September 1994, who provide a rigorous analysis of the link between economic growth and the expansion of what they call '`;,eneral purpose technologies," i.e., technologies that are "extremely pervasive" in the sense of being used as inputs by a wide range of sectors of an economy. 14 D.A. Irwin and P.J. Klenow, "Leaming-by-Doing: Spillovers in the Semiconductor Industry," Journal of Political Economy, vol. 102 (1994a), pp. 1200-1227. 15 Strictly speaking, it is an "adjusted" product price that is used: Irwin and Klenow, op. cite pp. 1212-1213, assume Cournot-competition and adjust the price accordingly, thus takin;, account of changing, monopolistic mark-ups over the product cycle.
OCR for page 197
THE ECONOMICS OF TECHNOLOGY POLICIES IN GLOBALIZED MARKETS 197 difference in the structural parameters of learning and of external effects between Japanese and other firms. All in all, these results cast serious doubt on whether the popular idea that there are powerful first-mover advantages for early market entrants is really compatible with the empirical record of the semiconductor industry. Prima facie, the simple observation that Japanese semiconductor produc- ers after a prolonged period of spectacular world market penetration lost world market shares back to American competitors by the late 1980s, seems to point in the same direction. However, the interpretation of this fact is complicated by two major policy shifts that may have influenced the course of events, namely the conclusion of two U.S.-Japanese semiconductor trade agreements, which brought some protectionist relief to American producers, and the establishing of SEMATECH, the joint industry-=overnment research consortium in the American semiconductor industry. With respect to the trade agreements, there is by now a broad consensus that they came too late" the first in 1986-to prevent a fundamental re- structuring of American industry away from the mass production of DRAMsi6 to more profitable market segments (e.g., design-intensive chips). With the benefit of hindsight, one can say that this turned out to be a blessing be- cause of the rising tide of competitive pressures from South Korean produc- ers that forcefully entered the DRAM market and undermined the dominat- iIlg Japanese position.~7 With respect to the economic impact of SEMATECH, there has been a first econometric attempt at estimation, again by Irwin and Klenow (1994b).~8 Usin:, data on American semiconductor firms, i.e., on those which were members of SEMATECH in the relevant period and those which were not (and which, statistically, form a control group), the authors estimate the effects of the consortium on the members' R&D spending, profitability, investment, and productivity. Their results turn out to be conclusive only for the effect on R&D spending: they indicate that SEMATECH has re- dur~d R&D spending by roughly 300 million U.S. dollars per year. In terms of the two alternative theoretical interpretations of technology policy presented above,~9 this supports the view that SEMATECH has served as an i6 Dynamic Random Access Memory chips. 17 On the genesis and the details of the American-Japanese semiconductor agreement, see D.A. Irwin, Trade Politics and the Semiconductor Industry. NBER Working Paper No. 4745. National Bureau of Economic Research, Cambridge, Mass., May 1994, pp. 69-71; for a sum- mary of opinions, see I. Maitland, "Who Won the Industrial Policy Debate?" Business & the Contemporary World 1 (1995) pp. 83-95, notably pp. 88-89. 18 D.A. Irwin, and P.J. Klenow, High Tech R&D Subsidies: Estimating the Effects of SEMA TECH. NBER Working Paper No. 4974. National Bureau of Economic Research, Cambridge, Mass., December 1994(b). 19 See the section "Endogenous Growth Theory" above.
OCR for page 198
198 CONFLICT AND COOPERATION instrument to focus and bundle research, i.e., to reduce inefficient parallel efforts, rather than to expand the research scope into ranges that would otherwise not have been profitable for private member firms to explore.20 While SEMATECH may thus have helped firms to coordinate their re- search efforts efficiently, there is by now broad agreement that this piece of positive technology policy can hardly be made responsible for the bulk of the recovery of the American semiconductor industry.2i Instead, the timely restructuring of the industry appears to have done by itself the major part of the work. In view of the econometric evidence about the powerful interna- tional diffusion of technological knowledge in this high-technology indus try, one should not be surprised to see one country's producers recover quite successfully even from rather deep adjustment crises.22 Knowledge Diffusion: International and Interregional Coe and Helpman (1993123 attempt to quantify international knowledge spillovers between industrial economies.24 On the basis of annual data for the period 1971-1990 and twenty-two mostly-OECD countries, they esti- mate how strongly total factor productivity in any country depends on the "research capital stock,' (1) in this country itself and (2) in all other coun- tries. The domestic "research capital stock" is defined as total national R&D expenditure, accumulated over time using a certain rate of "knowl- e~lge depreciation;" the total foreign research capital is calculated as a weighted average of that of all other countries, with the weights being bilateral im port quotas, i.e. roughly speaking the degree of trade integration. 20 Irwin and Klenow (1994b) speak of support for the "sharing hypothesis" as against the "commitment hypothesis." 21 See, inter alla, the assessment in The Economist, 2 April 1994, "Uncle Sam's Helping Hand," pp. 91-93 22 It is most unfortunate that no methodologically comparable studies are available for evaluating specific technology policies of other countries, notably those of Japan and the European Union. This is so because, to our knowledge, no data are available that would allow the setting up of control groups of firms that are not covered by the relevant policy and the quantifying of the effect of government intervention against this background. Thus the many available accounts of European technology policy e.g., H. Klodt, Wettlauf um die Zukunit, Kieler Studien No. 206, Tubingen' 1988; H. Klodt et al., ForschungspolitiZc unter EG-Kontrolle. Kieler Studien No. 220, Tubingen, 1988; and, more recently, H. Klodt and J. Stehn et al., Strr~krurpolztik der KG, Kieler Studien No. 249, Tubingen, 1992, pp. 98-114 and 152-16 have to remain largely descriptive and interpretative. 23 D.T. Coe, and E. Helpman, International R&D Spillovers, NBER Working Paper No. 4444, National Bureau of Economic Research, Cambridge, Mass., August 1993. 24 A methodologically comparable study that focuses, however, on spillovers from rich to poor countries is D.T. Coe, E. Helpman, and A.W. Hoffmaister, North-South R&D Spzllovers, CEPR Discussion Paper No. 1133, Centre for Economic Policy Research, London, February 1995.
OCR for page 199
THE ECONOMICS OF TECHNOLOGY POLICIES IN GLOBALIZED MARKETS 199 The authors find that total factor productivity in a country depends strongly on the domestic research capital: according to their estimates, a one percent increase of the research capital stock leads to a 0.25 percent productivity increase. They also find that foreign research capital matters as well, but that it matters comparatively more for smaller countries than for large ones. In fact, for small highly integrated economies, research efforts in the main trading-partner countries appear to matter at least as much as corresponding efforts at home. Apparently, trade between advanced economies- and with it international capital movements appear to be powerful channels for trans- mitting technological knowledge across borders, a result which tends to support the estimates and message of Irwin and Klenow (1994a) in the narrower context of the semiconductor industry. On interregional knowledge spillovers, there are two major studies that are directly relevant to the policy questions at hand. Jaffe, Trajtenberg, and Henderson (1993~25 exploit the geographic information contained in U.S. patent statistics to draw econometric inferences on the extent to which the diffusion of knowledge remains localized after an innovation has been made in some part of the country. The central piece of information on which they build their econometrics is the geographic pattern of patent citations to be found in new patent applications, which can serve as a kind of roadmap to track knowledge spillovers. While the methodology is rich and complex in detail,26 the results are rather clearcut. First of all, they point toward quite strong localization effects, with the degree of localization being higher for patents of private firms than for those of universities, which is plausible because purely aca- demic research results are likely to be circulated more openly than the knowledge created in private research laboratories. Second, for any patent, localization tends to decline over time, but very slowly: even a decade after a patent is granted, the geographic diffusion pattern remains little changed for all practical purposes. And third, the high degree of geographic local- ization is not matched by a corresponding degree of sectoral localization: whatever the grouping of patents into technological or industrial segments, there is always a remarkably high share of citations that refer to patents in very different fields. This casts some doubt on the common assumption that cumulative learning processes within a well-defined high-technology industry are a valid rationale for technology policy. 25 A.B. Jaffe, M. Traj~enberg, and R. Henderson, "Geographic Localization of Knowledge Spillovers as Evidenced by Patent Citations," Quarterly Journal of Economics, vol. 108 (1993), pp. 577-598. 26 See gaffe, Trajtenberg, and Henderson, op. cit., pp. 58~585. The main methodological problem is to separate genuine spillovers from correlations that are due to a preexisting pattern of geographic concentration of technology-relevant activities.
OCR for page 200
200 CONFLICT AIVD COOPERATION This last result has received independent support from another strand of research on interregional knowledge spillovers, which applies ideas of en- dogenous growth theory to the study of urban agglomerations. In a major contribution on the economic determinants of city growth, Glaeser et al. (1992~27 test two competing hypotheses, one identifying knowledge spillovers within an industry, the other one spillovers between industries as the driving force of output and employment growth. Using a large data set on the structure of industry for 170 metropolitan areas in the United States for the period 1956-1987, the authors estimate to what excerpt the long-term growth of a city industry was positively or negatively correlated with a number of characteristics of local industry structure, most importantly the decree of specialization (~) of the respective industry and (2) of the rest of the urban economy. The results are again rather clearcut: other things being equal, an indus- try tends to grow faster in cities where it is still underrepresented and where the rest of the urban economy has a low de Free of industrial specialization. Hence, as in the research on patent knowledge diffusion, an empirical case can be made for inter-industrial rather than intra-industnal spillovers domi- nating the picture.28 Strategic Trade Policy: The Aircraft Industry Rent-shiftin=, in high-technology markets through deliberate government intervention has been an explicit aim of policy in one particular branch of economic activity: the aircraft industry. Naturally, it has been this industry and in particular the economics of the European launching of Airbus as a competitor of Boeing that was subjected to empirical analysis, notably by 97 E.L. Glaeser, H.D. Kallal, A. Scheinkman, and A. Shleifer? `'Growth in Cities," Journal of Political Economy, vol. 100 (1992), pp. 1 126-1 152. 28 Note, however, that by not specifying any proxy variable for the state of "knowl- edoe" the methodology used by Glaser et al. (1992) may be subject to the criticism that it does not allow to discriminate between technological externalities and mere pecuniary exter- nalities, which work through the price mechanism and do less easily qualify as a rationale for technology policy. The same problem plagues another strand of research which tries to iden- tify intersectoral external effects by examining, the input-output relationships between indus- tries and estimating to what extent the productivity in one industry is affected by output changes in other industries. By this token, R.J Caballero, and R.K. Lyons, "The Case for External Effects," in A. Cukierman, (ed.), Political Economy, Growth, and Business Cycles, Cambridge, Mass., 1992, pp. 117-139, find sizable externalities, but it remains unclear of what kind. On these methodological problems in detail, see M. Stolpe, Technology and the Dynam- ics of Specialization in Open Economies, Kiel Studies 271, Tubin;,en, 1995, Section B. V, pp. 51-54.
OCR for page 201
THE ECONOMICS OF TECHNOLOGY POLICIES IN GLOBALIZED MARKETS 201 Baldwin and Krugman (1988) and Klepper (1990, 19941.29 For a number of reasons ranging, from the duopolistic market structure to simple lack of data, all these empirical studies consist of model calibration and simulation rather than econometric estimation of theory-based parameters.30 This is why all results must be interpreted with utmost caution: because they de- pend crucially on the models' assumptions and imputed parameters. The policy-relevant core of these studies consists in answering the fol- lowina double question: to what extent and for what economic reason did the subsidized market entry of Airbus-and thus the transformation of the aircraft market from a (prospective) monopoly of Boeing, into a transatlantic duopoly lead to a redistribution of producer and consumer surplus in the United States, in Europe, and in the rest of the world? Despite considerable methodological differences, the answers to these questions do not differ greatly between the relevant studies. Briefly summarized, the answers read as follows. The market entry and continued market presence of Airbus has led or will lead to a loss of pro- ducer surplus in the United States due to the reduction of (monopoly) prof- its and a gain of consumer surplus all over the world due to lower aircraft prices. However, European "consumers/taxpayers" are likely to end up worse off because the additional tax burden may well overcompensate for the gain in consumer surplus. Also, world welfare as a whole is likely to be reduced because the loss of producer surplus in the United States may well overcompensate for the worldwide gains In consumer surplus (net of the tax burden), a result which reflects the enormous importance of economies of scale in the aircraft industry: given very sharp cost reductions through learning effects, the socially optimal outcome for the world as a whole may simply be a monopoly. The central conclusion from these results is that, in economic terms, the European Airbus venture is better interpreted as a worldwide anti-monopoly policy than as a transatlantic rent-shiftin=,: European consumers/taxpayers foot the bill of breaking an American monopoly to the unambiguous advan 29 R. Baldwin, P.~. Krugman, "Industrial Policy and International Competition in Wide- Bodied Jet Aircraft," in R. Baldwin (ed.), Trade Policy Issues and Empirical Analysis, Chi- cago, 1988; G. Klepper, "Entry into the Market for Large Transport Aircraft," European Economic Review, vol. 34 (1990), pp. 775-803; "Industrial Policy in the Transport Aircraft Industry," in P. Kru~,man, and A. Smith (eds.), Empirical Studies of Strategic Trade Policy, Chicago, London, 1994, pp. 101-126. A methodologically comparable analysis of the market for small-size aircraft is provided by R. Baldwin, and H. Flam, "Strategic Trade Policies in the Market for 30~0 Seat Commuter Aircraft," Weltwirtschaftliches Arch~v/Review of World Eco- nomics, vol. 125 (1989), pp. 484-500. 30 On the specific methodological problems in this field of empirical research, see P. Kru;,man, "Introduction," P. Krugman and A. Smith, op. cit., pp. 4-5.
OCR for page 202
202 CONFLICT AND COOPERATION tage of consumers in the rest of the world.3i Whether one likes this policy or not, it has very little to do with the idea of a genuine strategic trade policy that postulates a national interest in rent-shifting as the normative basis for government intervention, and not an altruistic policy stance vis-a- vis the rest of the world. Needless to say, one may put forward other arguments for fostering a European aircraft industry as an important high-technology branch, but these usually lead into the realm of technological spillovers (see "External Learn- ing and Growth" above). To our knowledge, there have been no systematic empirical assessments of the aircraft industry with respect to technological externalities. Casual observations-e.g., the apparent failure of Germany's Daimler-Benz to profit from so-called synergy effects between aircraft and motor car production32 suggest that these effects have been grossly over- estimated in the past and have lured some firms into a path of diversifica tion that turned out to be unprofitable in the longer run. There is a more general conclusion to be drawn from the empirical stud- ies on rent-shifting that goes well beyond the aircraft market. Obviously, the aircraft industry is almost a textbook case for an industry with strong economies of scale and, consequently, very few commercial players: and fat monopoly margins: if even in this industry it is very hard to channel rents into the pockets of producers in the intervening country, how can one ever arrive at a powerful case for government intervention on these grounds in other industries that are much further away from the conditions of a natural monopoly? One may suspect that the chances to do so are small at best. TECHNOLOGY POLICY: SOME CAUTIOUS CONCLUSIONS Despite the vast uncharted territories that still await future research- notably in the field of econometrically scrutinizing the various dimensions of knowledge diffusion-some preliminary conclusions may be drawn from the available evidence surveyed above. 3i Note, however, that this result holds only if Boeing would in fact have become a monopoly without the subsidized market entry of Airbus. Yet this is an open question because Boeing's intra-American competitor McDonnell Douglas might have remained in the market in this case. A model calibration by Klepper (1990, 1994), which explicitly puts the actual Airbus entry against a duopoly with equal market shares of Boeing and McDonnell Douglas, yields the counterintuitive result that a duopoly Boeing/Airbus leaves Boeing with higher profits than the duopoly Boeing/McDonnell Douglas because the latter involves more equal market shares and thus forces Boeing further up the average cost curve. Whatever one may think of this peculiar result it depends crucially on the imputed market share of McDonnell Douglas, which is purely speculative the Airbus venture loses its antimonopoly rationale once it is assumed that there is no monopoly in the counterfactual scenario anyway. 32 See, e.g., The Economist, 20 May 1995, "Schrempp Cocktail" (p. 70), and 18 November 1995, "A Tale of Two Conglomerates" (p. 20) and "Dismantling Daimler" (p. 79).
OCR for page 203
THE ECONOMICS OF TECHNOLOGY POLICIES IN GLOBALIZED MARKETS 203 (1) Knowledge spillovers appear to be a pervasive feature of modern economies that is likely to be of great importance for economic growth. There is still much less clarity about how the knowledge diffusion actually works, although there are some patches of relevant evidence: studies on the interreg~orZal diffusion process point to strong localization effects in terms of geography, but not in terms of sectors of economic activity; in turn, studies on international diffusion point to powerful spillovers across na- tional borders, be it on a macroeconomic level or for a selected important high-technolo=,y industry such as semiconductors. The apparent contradic- tion between the empirical results from interregional and international stud- ies may have its cause in a genuine "globalization" of knowledge flows: it is not implausible to suspect that the leading high-technology centers in different countries may be better linked in terms of communication than the high-techr~ology centers and the periphery in one single (lar:,e) country. (2) The aim of shifting monopoly rents between countries does not by itself make a case for government intervention. The one instance where it was obviously relevant the aircraft industry is history, since Airbus is by now an established competitor of the former (quasi-) monopolist Boeing. Whether the launching of Airbus was a policy success depends on the crite- ria used. Be that as it may, no comparably structured high-technology branch, where fat monopoly profits accrue in just one country, is In sight in the near future. (3) The actual record of government intervention in high-technology industries is very hard to evaluate because, usually, no sensible counterfactual scenario can be made available. However, there is one important case where such a scenario has been at least tentatively constructed: SEMATECH. The relevant analysis of SEMATECH indicates that the consortium helped to reduce arguably inefficient parallel research, thus pointing to a posi- tive role for the government as an agent that bundles and focuses rather than expands research efforts. Back to the basic question: is there a theoretically sound and empirically supported rationale for government intervention in high-technology mar- kets? Given the extent of our ignorance of the precise determinants and structure of knowledge flows, it is hard to avoid the conclusion that a government in an advanced economy has exactly the same information problem as the empirical economist in tracing the relevant knowledge flows. Hence a fine-structured industrial policy targeted at selected high-technology in- dustries can hardly be recommended.33 On the other hand, the observed regional localization of knowledge spillovers may suggest that a "techr~ol 33 This conclusion is almost a commonplace among, economists. See, e.,,., the policy conclusions by one of the main advocates of endogenous growth theory: G.M. Grossman, "Promoting, New Industrial Activities: A Survey of Recent Arguments and Evidence," OECD Economic Studies, vol. 14 ( 1990), pp. 1 17-1 19.
OCR for page 204
204 CONFLICT AND COOPERATION ogy policy" aimed at generally supporting the (otherwise suboptimal) growth of high-technolo=,y agglomerations, can make economic sense. The prob- lem is, of course, how this should be done with a minimum risk of incurring, deadweight losses for taxpayers and consumers. Some reasonable :,uide- lines may read as follows. (1) In selecting instruments of government intervention, one should aim at supporting R&D itself, and not output or trade, because output subsi- dies and trade policies invariably have undesirable allocative side effects, not to speak of the potential for political frictions that may come to the fore once countries engage in protectionist warfare in high-technolo=,y markets. (2) When choosing R&D for government support, one should consider whether the apparent market failure cannot be corrected through what may be called "government coordination" rather than through subsidies. Finding an economically efficient level and structure of industry research by jointly launching particularly large and risky projects or by weeding out costly parallel research is eventually in the interest of all firms concerned; and at least in high-technolo~y markets that are characterized by (not too wide) oligopolies, firms may be ready to cooperate in research. FIence the gov- ernment may simply serve as a positively neutral agent that creates the necessary credibility, commitment, and mutual trust among the private par- ties to make the joint venture possible at all. With the benefit of hindsight, SEMATECH may in fact be interpreted in this way, given its apparently successful record and its 1994 decision to continue its operation into the second half of the l990s, but to renounce government money, which had previously made up about half of its funds. Of course, even such a modest government role creates problems. First, it requires a precise definition in antitrust law, under which in precompetitive circumstances an R&D joint venture does not fall under the ban on collu- sion. In practice, any exemption clause will be abused to some extent, so the likely damage of abuse will have to be put onto the debit account of any (government-sponsored) cooperation. Second, it requires a decision as to what extent foreign firms are permitted to participate. Again, there is a conflict: on one side, the very rationale of technology policy is to foster national hi:,h-technology agglomerations, and not the ones in other coun- tnes; on the other hand, the participation of foreign firms may be the only feasible way to tap a foreign stock of knowledge that may be crucial to obtain or preserve the competitive edge of domestic industry. Even if foreign firms are excluded from participating in government-sponsored co- operation, however, they may acquire the relevant knowledge by taking over a fib that does participate.34 In practice, this can hardly be avoided 34 The history of SEMATECH is telling in this respect. See The Economist, 2 April 1994, "Uncle Sam's Helping Hand" (p. 93).
OCR for page 205
THE ECONOMICS OF TECHNOLOGY POLICIES IN G~OBALIZED MARKETS 205 unless a government is prepared to make very serious inroads into the free- dom of capital movements. (3) If an element of subsidization is to be added to the package of government R&D support, it should be done in a nondiscriminatory fashion. This means roughly speaking-that any dollars or DM spent on R&D should be subsidized at the same rate, no matter which branch of economic activity it is invested in. This of course means that high-technology branches with high ratios of R&D spending to value added receive a higher subsidy per unit of value added; but this "discrimination" is perfectly compatible with the externality-based logic of the subsidy, because they are also the branches whose output is likely to be the furthest below the social optimum. In practice, a favorable tax treatment of R&D may be the most appropriate tool to achieve this task. Of course, one may wonder whether most industrialized countries in- cluding the United States and Germany do not already have an implicit R&D subsidy implemented in their tax codes, although maybe for reasons that have nothing to do with a conscious and deliberate effort to support R&D. An implicit R&D subsidy can be recognized in two distinct elements of the tax code.35 First, labor costs incurred in R&D can be deducted as current expenditure, like labor costs incurred in production, although they are economically more like an investment in future knowledge creation and thus, in a neutral tax code, would have to be treated like an investment in physical capital, which can be deducted only over time, according to some schedule of depreciation. Second, the value of the knowledge output that is not sold in the market for patents and licenses, but kept for exclusive use in the company, is not counted as an asset for tax purposes. This amounts to a positive discrimination of the knowledge stock vis-a-vis that part of the physical capital stock that is produced by the firm itself. While these tax rules may not be optimal from an economic perspective, they do constitute a considerable element of R&D support and maybe a starting-point for a "non-mercantilistic technology policy." 35 For details, see M. Stolpe, "Industriepolitik aus Sicht der neuen Wachstumstheorie," Die Wel~wirtschaft, 3 (1993) pp. 369-370.
Representative terms from entire chapter: