Qeh working paper series qehwps111

QEH Working Paper Series — QEHWPS111 Page 1 Working Paper Number 111 Reinventing industrial strategy: The role of government policy in building industrial competitiveness Sanjaya Lall 1 The paper reviews the nature of current globalization and the growing divergence in competitive performance in the developing world. It considers the case for industrial policy, contrasting the neoliberal with the structuralist approach. It argues that there is a valid case for selective interventions in overcoming the market and institutional failures in building the capabilities required for industrial development. It describes the strategies adopted by the Asian Tigers to build industrial competitiveness, and concludes with lessons for other developing countries. The kinds of industrial policy needed in the current setting are different from traditional industrialisation strategies, but globalization and technical change do not eliminate the need for intervention. On the contrary, given path dependence, cumulativeness and agglomeration economies, they increase it. There is a need to reconsider the rules of the game constraining the exercise of industrial policy, and for international assistance in designing and implementing appropriate policies. October 2003 This paper was prepared for the G24 Intergovernmental Group on Monetary Affairs and Development. I am grateful to Larry Westphal for discussions and detailed comments on an earlier draft, to Robert Wade for sending me pre-publication copies of papers on the issues addressed here, and to Manuel Albaladejo for help with the data. 1 QEH Working Paper Series — QEHWPS111 1. Introduction Page 2 As liberalization and globalization gather pace, concern with industrial competitiveness is growing, not just in developing countries but also in mature industrial ones. But it is the former that face the most intense competitive pressures: many find that their enterprises are unable to cope with rigours of open markets — in exporting and in competing with imports — as they open their economies. Some countries are doing very well; the problem is that many are not. Diverging industrial competitiveness in the developing world is one of the basic causes of the growing disparities in income that are now a pervasive feature of the world scene. The immense potential that globalization offers for industrial growth is being tapped by a relatively small number of countries, while liberalization is driving the wedge deeper. Much of this is widely known. The Millennium Development Goals of the United Nations were conceived to deal with just such concerns. However, there is little consensus yet on what can be done to deal with them, particularly in the industrial sphere. What can poor countries do to strengthen their industrial competitiveness in the international economic setting? Should they persist with liberalization and hope that free market forces will stimulate growth and bring about greater convergence? Or is there a need to lo ok again at national and international policy? What, in sum, is the correct role of government in stimulating industrialization and using it as an engine for growth and structural transformation? There are essentially two approaches to the issue of policy: neoliberal and structuralist. The neoliberal approach is that the best strategy for all countries and in all situations is to liberalize — and not do much else. Integration into the international economy, with resource allocation driven by free markets, will let them realise their ‘ natural’ comparative advantage. This will in turn optimize dynamic advantage and so yield the highest rate of sustainable growth attainable — no government intervention can improve upon this but will only serve to reduce welfare. In this approach, the only legitimate role for the state is to provide a stable macro-economy with clear rules of the game , open the economy fully to international product and factor flows, give a lead role to private enterprise, and furnish essential p ublic goods like basic human capital and infrastructure. This approach has the backing of the industrialized countries and the Bretton Woods institutions (which is why it is also referred to as the ‘ Washington Consensus’). It has become enshrined in the new rules of the game being formulated and implemented by the WTO. The neoliberal approach has strong theoretical premises: markets are ‘ efficient’, the institutions needed to make markets work exist and are effective, and if there are deviations from optim ality they cannot be remedied effectively by governments. The premises are a mixture of theoretical, empirical and political assumptions. Their theoretical core relies, among other things, on a restrictive view of the technological basis of competitiveness. The empirical one relies on a particular interpretation of the experience of the most successful industrializing economies, the ‘ Tigers’ of East Asia. The political element — that governments are necessarily and universally less efficient than markets — has less to do with economics than with ideology. The structuralist view puts less faith in free markets as the driver of dynamic competitiveness and more in the ability of governments to mount interventions effectively. It questions the theoretical and empirical basis for the argument that untrammelled market forces account for the industrial success of the East Asian Tigers (or, indeed, of the earlier industrialization of the presently rich countries). Accepting the mistakes of past industrialization strategies and the need for greater openness, it argues that greater reliance on markets does not pre-empt a proactive role for the government. Markets are powerful forces but they are not perfect; the institutions needed to make them work efficiently are often weak or absent. Government interventions are needed to improve on market outcomes. Structuralists also accept that some industrialization policies have not worked well in the past. To the neoliberals this is a reason for denying any role for proactive policy both in past success and in future strategy: if there are market failures, the costs are always less than those of government failures. The structuralists, on the other hand, see a vital role for policy in industrial success. For them, therefore, past QEH Working Paper Series — QEHWPS111 Page 3 policy failure is not a reason for passive reliance on deficient markets but for improving government capabilities. They note that many poor regions that have implemented neoliberal policies recently have not experienced the industrial growth or export success that characterized more interventionist economies. To them, a projection of current trends suggests that persisting with passive liberalization in the context of globalization will exacerbate rather than reverse divergence. The growing unease with the consequences of neoliberalism led the Zedillo Commission, in its ‘ Report of the High-Level Panel on Financing for Development’ to the Monterrey Conference on Financing for Development in 2002, to phrase the issue in diplomatic terms. Noting that ‘ Sadly , increasing polarization between the haves and have-nots has become a feature of our world’ it said the following on infant industry protection (a policy tool banned under the new rules): “ However misguided the old model of blanket protection intended to nurture import substitute industries, it would be a mistake to go to the other extreme and deny developing countries the opportunity of actively nurturing the development of an industrial sector” (Zedillo Commission, 2001, Executive Summary, p. 910). 2 The controversy on industrial policy, of course, is not new; it goes back decades and, in earlier guises, centuries (Reinert, 1995, Chang, 2002). Despite the frequent assertion one hears that the debate is now dead and the efficacy of free markets established beyond doubt , this is not the case. This paper shows why this is the case and suggests that the case for policy remains strong, and is in fact becoming stronger with technical change and globalization. However, the kinds of intervention needed are changing; as a structural force, globalization reduces the feasibility of some strategies while increasing that of others. Structural changes are supported by new ‘ rules of the game’ on participation in the international system. Some rules are necessary to facilitate the changes, but they must take account of the fact that the field has players of very different strengths. Imposing a level field can lead to an uneven distribution of benefits between the strong and the weak. They can constrain the ability of poorer countries to build the capabilities they need for industrialization, banning policies used with spectacular success by several countries, including the advanced ones. Before coming to the new rules and the legitimate role of policy, let us review briefly the main features of recent industrialization. 2. The new dimensions of industrial competitiveness 2. 1 Structural features Competitiveness has always mattered for industrial growth, but its nature has evolved. Rapid technical change, shrinking economic distance, new forms of industrial organization, tighter links between national value chains and widespread policy liberalization, are altering radically the nature of environment facing enterprises. Competition now arises with great intensity from practically anywhere in the world, based on a bewildering array of new technologies, advanced skills and sophisticated supply-chain and distribution techniques. To survive it, all producers must use new technologies at or near ‘ best practice’. It is organised in complex systems spanning many countries, tapping differences in costs, skills, resources and tastes to optimize the efficiency of the entire system (Radosevic, 1999). It is supported by international brands and networks with the capacity to deliver vast amounts of information at negligible cost. Manufacturing is becoming more information-intensive: larger parts of value added consist of ‘ weightless’ activities like research, design, marketing and networking. Technical change is shifting industrial and trade structures towards more complex, technology-based activities. Table 1 shows the growth of manufacturing value added (MVA) for three technological sets of activities: resource based (RB), low technology (LT) and medium and high technology (MHT). 3 For 2 3 For an interchange based on this recommendation see Wood (ed.) (2003). Rodrik (2001) raises similar issues. For a description of the categories and the rationale behind the classification see Lall (2001. a). QEH Working Paper Series — QEHWPS111 Page 4 exports the data allow us to show high technology products separately. Over the past two decades exports have grown faster than production, and complex activities have grown faster than other branches of manufacturing. Developing countries have done better in all branches than industrialized economies. Table 1: Growth of manufacturing value added and manufactured exports by technology (% per annum, 1980-2000) Activity World Industrialized countries Developing countries Manufacturing value added Total MVA RB MVA LT MVA MHT MVA Manufactured exports Total exports RB exports LT exports MHT exports manufactured manufactured manufactured manufactured 7. 6% 5. 6% 7. 4% 8. 4% 11. 5% 6. 6% 5. 2% 8. 4% 7. 3% 9. 9% 12. 0% 6. 7% 11. 4% 16. 5% 20. 2% 2. 6% 2. 3% 1. 7% 3. 1% 2. 3% 1. 8% 1. 4% 2. 6% 5. 4% 4. 5% 3. 5% 6. 8% o/w Hi-tech exports Source: Calculated from UNIDO and Comtrade data. Organizational structures and the location of production are changing in response to technical change. Industrial firms are becoming less vertically integrated and more specialized by technology. Under competitive pressure, they are scouring the world for more economical locations. Technical progress in transport and communications is shrinking economic space and allowing firms to locate p rocesses and functions in far-flung parts of the globe. Some facilities are under the control of transnationals from the industrialized countries but others are independent local firms, interwoven with the leaders in intricate webs of contractual and non-contractual relations. This ‘fragmentation’ of production is rewriting the geography of industrial activity. 4 The international fragmentation of value chains has, for economic reasons, gone furthest in activities with discrete and separable production processes and high value products. Electronics is the best example, placing production in several countries, each site specializing in a process or function according to its labour costs, skills, logistics and so on (Sturgeon, 2002). The 4 QEH Working Paper Series — QEHWPS111 Page 5 New technologies change the institutional and policy structures needed for competitiveness. For instance, countries require new skills to manage technical change, and so the institutional ability to upgrade skills (Narula, 2003). They need good technical support agencies in standards, metrology, quality, testing, R&D, productivity and SME extension. They need advanced infrastructure in information and communication technologies (ICTs). They need new rules, legal systems and agencies to encourage enterprises to build competitive capabilities and allow knowledge to flow across national boundaries. And they need to cushion the impact of new technologi s on declining activities and disadvantaged groups. It e is not easy to meet such demands, even in advanced countries — this is why most governments mount competitiveness strategies (Lall, 2001. b). Globalization also leads to greater transfer of productive factors across economies. However, though capital, technology, information and skills are more m obile they do not spread evenly over low wage locations. They go only to places where competitive production is possible, to locations that can supply the inputs and institutions needed to complement the mobile factors. It requires, in brief, the development of new industrial capabilities (Best, 2001). Cheap unskilled labour or raw natural resources are no longer sufficient to sustain industrial growth: it is strong local capabilities that determine competitive success. Even ‘simple’ entry-level industrial activities like clothing, footwear or food processing require sophisticated capabilities if they are to face global competition. However, industrial capabilitie s develop slowly, in a cumulative and path-dependent manner subject to agglomeration economies. Thus, those economies that launch on to a virtuous circle of growth, competitiveness and investment in new capabilities can carry on doing better than those that are stuck in a ‘ low level equilibrium’ and cannot muster the resources to break out. Industrial performance can diverge across countries and continue diverging over time, with no inbuilt forces to return them towards greater convergence. Reversing these trends is not easy. It calls for concerted policy action to shift economies from one growth (or rather, low growth) and technological trajectory to another. 2. 2 Rules of the game Liberalization in the developing world has been partly voluntary, partly driven by persuasion and pressures and partly enforced by changes to the rules of international economic relations. The changes have essentially been to free trade and capital flows from government interventions , strengthen private property rights and level the playing field for all economic agents. Supporting these new rules are a number of such domestic policy ‘reforms’ as liberalization of financial markets and privatization of public enterprises. Some of these changes were initiated by developing countries disillusioned with early importsubstitution industrialization strategies. Some were initiated by developed countries, the Bretton Woods agencies, and various bilateral, regional and international agreements. And some were negotiated at the international level, as in the Uruguay Round of GATT (now WTO). One effect of these changes has been to constrict policies used to promote industrial development. The most affected are: protection of infant industries, 5 performance requirements on foreign investors, export segmentation of software, business process services and other IT based activities like call centres is another manifestation of this phenomenon outside manufacturing. Fragmentation goes beyond the spread of transnational companies (TNCs). I encompasses t the closer integration of national value chains under several governance systems, with direct ownership by TNCs being at one end and loose buying relationships at the other (Gereffi et al., 2001, Humphrey and Schmitz, 2001). No new protection can be offered to products for which members have ‘ bound’ their tariffs, though if actual tariffs are lower than bound tariffs they can be raised. Export processing zones may come under the purview of the subsidies ban in the future (LDCs are exempt so far). 5 QEH Working Paper Series — QEHWPS111 Page 6 targeting and incentives and other subsidies affecting trade, 6 slack IPRs (intellectual property rights) protection to promote copying and reverse engineering and local content rules7 . The rules are too complex to be analysed here at length and their precise content is not germane to the discussion, but some general points may be noted. First, the rules on trade allow for exceptions, particularly for the least developed countries. However, the grace period allowed is coming to an end for many exceptio ns. Second, the rules carry the threat of sanctions: interventions that affect trade can lead trading partners to impose compensatory tariff or other measures. Third, more important than the specific measures undertaken till now is the underlying long-term trend towards greater liberalization. The scope and coverage of the rules are steadily increasing, and pressures for removal of policy controls are coming in many forms. It would be reasonable to project a trade regime for developing countries very simila r to that obtaining within the OECD. Policies on FDI and technology imports have undergone rapid liberalization, to a greater extent than those on trade and domestic credit. Most liberalization has occurred over the past decade or so, particularly for FDI in the industrial sector, with the pace accelerating in the 1990s. Many of the latest changes are under international commitments under the Uruguay Round; however, the trend reflects a change of attitude on the part of host countries. There are practically no policy controls left on technology transfer, in contrast to the 1970s when there were extensive interventions by governments on licensing. Some of the main issues in the multilateral agreements are as follows: 8 Services: The General Agreement on Trade in Services (GATS) covers the supply of markets by foreign firms present in those markets under WTO. Its general principles are transparency and most-favourednation (MFN) treatment (i. e. non-discrimination between firms of different origins). The GATS allows a ‘ positive list’ of permitted investments, allowing host countries freedom to exclude activities not in the list. Performance requirements on TNCs: This is treated under the Agreement on Trade -Related Investment Measures (TRIMs). TRIMs affect trade in goods and are important in that they prohibit tools traditionally widely used to extract greater benefits from FDI: local content requirements, trade balancing (extremely effective in promoting the restructuring of the Latin American automobile industry), technology transfer, local employment and R&D, and so on. Intellectual property rights (IPRs): The protection of IPRs has moved in effect from the World Intellectual Property Organisation to WTO, under the TRIPS (Trade-Related Aspects of Intellectual Property Rights) Agreement. It specifies rules on standards for protecting IPRs, domestic enforcement and international dispute settlement (UNCTAD, 1996). The most important point about the shift from WIPO to WTO is that trade sanctions can now be applied to countries deemed to be deficient protecting IPRs. 9 The implications for the developing world are worrying (Lall, 2003). While stronger IPRs may benefit the leading innovators in the developed countries, they can inhibit technological development in developing ones. They can raise the cost of formal technology transfers, by allowing technology sellers to impose stricter restrictions and by preventing copying and ‘ reverse engineering’, the source of much technological learning in newly industrialising countries. General subsidies that do not create a cost advantage for identifiable activities may not be actionable. Only subsidies given to particular activities or locations that create such an advantage are subject to potential sanctions. 7 Local content rules are actionable if there are specific subsidies or incentives linked to achieving the prescribed levels. All countries, regardless of income levels, are now subject to this restriction. 8 For a comprehensive analysis see UNCTAD (2003). 9 The WTO Agreement on Subsidies and Countervailing Duties may also affect traditional means of supporting technological activity by subsidies. Although the Agreement excludes ‘ fundamental research’ from its actionable provisions (i. e. governments may still subsidize research), the text leaves scope for interpreting what the limits of this are. In any case, R&D now comes under WTO scrutiny, and subsidies for research deemed non-fundamental could be limited in the future. 6 QEH Working Paper Series — QEHWPS111 2. 3 Trends in industrial competitiveness in the developing world Page 7 This section uses two indicators: world market shares in manufacturing value added (MVA) and in manufactured exports. Developing r egions are as follows: ‘ East Asia’ or EA includes China and all countries in the Southeast Asian region apart from Japan, while EA2 excludes China . ‘ LAC’ (Latin America and the Caribbean) includes Mexico and LAC 2 excludes it. South Asia includes the five main countries in that region. ‘ MENA’ (Middle East and North Africa) includes Turkey but not Israel (an industrialized country). ‘ SSA’ (Sub-Saharan Africa) includes S. Africa except in SSA 2. Figure 1: Developing regions’ shares of global MVA (%) 14. 0 12. 0 10. 0 8. 0 6. 0 4. 0 2. 0 0. 0 S. Asia E Asia MENA LAC SSA 1980 1990 2000 MVA: The developing world performed well in 1980-2000. Its share of global MVA rose by 10 percentage points (from 14% to 24%) and its annual rate of growth (5. 4%) was over twice the 2. 3% recorded by the industrialized world. Since this was a period of trade expansion, globalized production and liberalization, it may seem that globalisation and liberalization were conducive to development. This is not so. Success in the developing world was very concentrated (Figure 1). East Asia dominated, raising its world share from around 4% to nearly 14% — exactly the 10 point rise for the developing world as a whole. It came from behind LAC in 1980 to account for over two and a half times its share by 2000 (Figure 2). Note that EA, while strongly export-oriented, was not ‘liberal’ in the Washington consensus sense. 10 LAC, the region that liberalized the most, the earliest and the fastest, was the worst performer. 10 As is now well known, most East Asian economies used infant industry protection, export subsidies and targets, credit allocation and direction, local content rules and so on to build their base of industrial capabilities, disciplining the process by strong export orientation (Amsden, 1989, Stiglitz, 1996, Wade, 1990, Westphal, 2002, World Bank, 1993). There were different strategies within this general approach. The leading Tiger economies like Singapore, the Republic of Korea and Taiwan Province of China invested massively in human capital (particularly technical skills), fostered local R&D and built strong support institutions (Lall, 1996 and 2001. a). They tapped FDI in different ways, Singapore by plugging into global production systems and the other two by drawing on its technologies via arm’s length means like licensing, copying and original equipment manufacturing. The second wave of Tiger economies like Malaysia, Thailand, Indonesia and Philippines relied more heavily on FDI in export processing enclaves and less on building indigenous capabilities; their export success was thus largely driven by global value chains, particularly in electronics. China has a blend of different strategies, some similar to its neighbours and QEH Working Paper Series — QEHWPS111 Page 8 Figure 2: Changes in shares of global MVA (% points) 1980-90 SSA 1990-2000 MENA LAC S. Asia E Asia -2 -1 0 1 2 3 4 5 6 7 LAC and East Asia illustrate the central issues of this paper nicely. The regions had very different approaches to industrialization, initially to develop industry 11 and later to liberalize it 12 — EA has had much more strategic industrial policy than LAC. The resulting differences in outcomes are interesting, as the next two charts show. The charts separate China in EA and Mexico in LAC, both regional outliers, China because of its size, competitiveness and strong state role, Mexico because of its location and privileged access to the US market. Both have done very well in manufactured exports with a strong role for FDI, but their differences are also of interest. For instance, the link between export and MVA growth is far stronger in China than in Mexico: China is far less exposed to import competition and has used industrial others, like public enterprise restructuring, uniquely its own (Lall and Albaladejo, 2003). The region as a whole liberalized cautiously and has retained a significant role for the state. As Stiglitz says in a special contribution to the new Human Development Report, “ China and other East Asian economies have not followed the Washington consensus. They were slow to remove tariff barriers, and China still has not fully liberalised its capital account. Though the countries of East Asia ‘ globalized’, they used industrial and trade policies to promote exports and global technology transfers, against the advice of the international economic institutions” (UNDP, 2003, p. 80). Also see Rodrik (2001). 11 In the first phase, LAC, in common with most other developing regions, relied heavily on protected import -substitution, sheltering enterprises from international competition but failing to offset this with incentives or pressures to export. It did little to attract export-oriented FDI (in EPZs) and so missed the surge in global production systems in electronics. It did not deepen local technological activity (by encouraging R&D) or develop the new skills needed for emerging technologies. In concert with widespread macroeconomic (and in some cases political) turbulence, this meant that LAC failed to develop a broad base of industrial capabilities that would drive competitiveness as it liberalized. As a comparatively high wage region, LAC needed competitive advantages in complex activities to offset labour cost disadvantage vis a vis Asia. Despite its tradition of entrepreneurship and good initial base of skills, its industrial strategy failed to foster the necessary capabilities. There were exceptions, such as the automotive industry in the larger economies and resource-based activities more generally. But many such activities were not growing rapidly in world trade and, as shown below, LAC failed to increase its export market shares rapidly — the outstanding exception being Mexico, but due more to NAFTA privileges than to strategy . 12 In the second (liberalization) phase, policy reform in LAC was rapid and sweeping, with no strategy to foster competitive capabilities and target promising activities. Again, there were exceptions, including the auto industry (restructured with the help of complementation programs, banned under new WTO rules), agro-based exports in Chile or national export ‘ champions’ like Embraer in Brazil, but the general lack of strategy on industrial competitiveness meant that the region failed to catalyze export dynamism. Its main growth was in resource-based sectors where it was largely exploiting static comparative advantages Some other developing regions that also used import substitution strategies liberalized more slowly and carefully — India is a good example — and did better in terms of MVA growth (but almost as poorly in terms of export competitiveness). QEH Working Paper Series — QEHWPS111 Page 9 policy to induce greater local content in its export activity. 13 Figure 3 shows MVA market shares within the developing world for EA without China, China, LAC without Mexico, and Mexico. Figure 3: East Asia and LAC, shares of developing world MVA (%) 40% 35% 30% 25% 20% 15% 10% 5% 0% EA exc. China China LAC exc. Mexico Mexico 1980 1990 2000 Figure 4: East Asia and LAC, changes in shares of developing world MVA (%) 15% 10% 5% 0% EA 2 -5% -10% -15% 1980-90 1990-2000 China LAC 2 Mexico Figure 4 shows changes in these market shares over 1980-90 and 1990-2000. In 1980, LAC accounted for 47% of developing world MVA and East Asia for 29%; two decades later, the shares were 22% and 58% respectively. The main surge in MVA growth in EA 2 (excluding China) was in the 1980s, with a slowing down in the 1990s because of the financial crisis and the global recession. In China the trends are reversed, with the more rapid growth in the 1990s, making its share of developing world MVA higher than the rest of East Asia together. LAC2, excluding Mexico, loses MVA shares more rapidly than Mexico, with the 1980s (the ‘ lost decade’ after the debt crisis) being much worse than the 1990s. China now poses a major competitive threat to Mexico in textiles and electronics. Mexican figures suggest the loss of over 200, 000 jobs to China since 2001. See The Economist (2003) and The International Herald Tribune (2003). 13 QEH Working Paper Series — QEHWPS111 Page 10 The 1990s are illuminating for LAC industrial growth. It started the decade with considerable slack engendered by the lost decade, which favourable macro and policy conditions should have allowed it to exploit for high production and export growth. There was better macro management, widespread privatization and lowering of trade barriers. Despite these neoliberal policies, the region continued to perform poorly : LAC2 had MVA growth of only 1. 9% p. a., much lower than developing countries as a whole (6. 4%) or East Asia (9. 5%). It underperformed relative to South Asia and MENA, both highly interventionist regions. Mexico’s more robust growth of 4. 4% was largely a consequence of trade privileges over other developing regions under NAFTA — hardly a neoliberal recipe. In any case it did not match EA 2 (6. 7%) or China (13. 1%), and this despite the fact that the 1990s were a bad period for EA2, reeling from the effects of the 1997 financial crisis. Export performance: Figure 5 shows world market shares for manufactured exports for 1981-2000 and the value of such exports in 2000, separating China from East Asia 2 and Mexico from LAC 2. QEH Working Paper Series — QEHWPS111 Page 11 Figure 5: World market shares for manufactured products in 1981 & 2000, and values of manufactured exports in 2000 ($ billion) 14% East Asia excl China ($588 b.) 12% 10% WMS in 2000 China ($318 b.) 8% 6% 4% Mexico ($144 b.) MENA ($79 b.) LAC excl Mexico ($106 b.) 2% 0% 0% S. Asia ($54 b.) 1% SSA ($30 b.) 2% 3% WMS in 1981 4% 5% 6% 7% Figure 6: Changes in world market shares for manufactures (% points) 1981-90 4. 0% 3. 5% 3. 0% 2. 5% 2. 0% 1. 5% 1. 0% 0. 5% 0. 0% -0. 5% -1. 0% E Asia exc. China China S Asia MENA 1990-2000 LAC exc. Mexico Mexico SSA exc. S South Africa Africa East Asia as a whole accounted for 18. 4% of world manufactured exports in 2000, up from 6. 8% in 1981. Within it, EA2 raised its share from 5. 8% to 12. 0% and China from 1. 0% to 6. 5%. China has a much higher share of regional MVA than exports — its industry, perhaps not surprisingly in view of the size of the economy and its late entry to export markets, is less export-oriented than its neighbours’. LAC lost QEH Working Paper Series — QEHWPS111 Page 12 world market share in 1981-90 (from 3. 2% to 2. 4%) then raised it over the next decade to 5. 1%. The initial fall was due entirely to LAC 2 (from 2. 7% to 1. 9%), with Mexico holding steady at a 0. 5% share. Over 1990-2000, LAC 2 raised its share marginally while Mexico had a dramatic six-fold increase to 2. 9%. As Figure 6 shows, other regions were relatively stagnant, though each did better in the 1990s than in the 1980s. What may we conclude from these data? MVA performance is broadly correlated with manufactured export performance, though the fit is not perfect. EA 2 and Mexico fare better in exports than in MVA in the 1990s, while the opposite is true of South Asia and MENA. Neither MVA nor export growth is strongly related to liberalization in the Washington consensus sense. China, in particular, is hardly a neoliberal paradigm. Industrial success remains concentrated. Liberalization is not leading to convergence, contradicting the neoliberal premise that liberalization per se would promote industrial growth and competitiveness. 3. Why the world differs from the neoliberal ideal 3. 1 The neoclassical approach The reason why neoliberalism finds it difficult to analyse industrial development lies mainly in its treatment of technology. Developing countries are thought not to undertake significant technological activity, since they do not innovate at the frontier. The neoclassical model assumes that there are no additional costs, risks or other constraints to using technologies. Thus, it does not raise any policy issues: by assumption there can be no significant market or institutional failure. 14 Neoliberal economists accept that there is a role for the state, essentially to provide basic public goods (apart from law and order and a sound legal system and macro management). They also now accept that it has a role in providing non-selective or functional support for education, health and infrastructure. Why ‘ non-selective’? Selectivity (the support of particular activities, firms or technologies, or, crudely put, ‘picking winners’) became the arena for the industrial policy debate in the 1990s. The mid-1980s neoliberal interpretation of East Asian success, that it was due to free trade and other non-interventionist policies, was subjected to intense criticism. It was noted that most successful Asian industrialisers had been very interventionist in trade, FDI, technology transfer and domestic resource allocation. 15 The evidence was so overwhelming that the neoliberal camp was forced to admit the facts of the case. The reason why neoliberalism finds it difficult to analyse industrial development realistically lies in large part in its treatment of technology. Technology is ignored in most development analysis. Developing countries are thought not to undertake significant technological activity, since they do not innovate at the frontier and rely primarily on imported technologies. The neoclassical model assumes that there are no additional costs, risks or other constraints to using technologies. Thus, it does not raise any policy issues: by assumption there is no significant market or institutional failure. 16 14 This is as true of endogenous growth models — grounded in technical change — as it is of traditional models. Endogenous models focus on frontier innovation (the creation of new knowledge) rather than on using existing knowledge, and so simply assume that developing countries do best by opening themselves to inflows of information embodied in trade and investment. Access to new technology becomes equivalent to its effective use. The policy implications of the models that follow from externalities, increasing returns and non-appropriability in innovation apply only to advanced countries; the development implications, in so far as they are mentioned, are the same as in standard neoclassical analyses. 15 The objections to the strong neoliberal position came from such authors as Amsden (1989), Lall (1992), Pack and Westphal (1986), Wade (1990) and Westphal (1982 and 1990). 16 This is as true of endogenous growth models — grounded in technical change — as it is of traditional models. Endogenous models focus on frontier innovation (the creation of new knowledge) rather than on using existing knowledge, and so simply QEH Working Paper Series — QEHWPS111 Page 13 Neoliberal economists accept that there is a role for the state , essentially to provide basic public goods (apart from law and order and a sound legal system and macro management). They also now accept that it has a role in providing non-selective or functional support for education, health and infrastructure. Why ‘ non-selective’? Selectivity (the support of particular activities, firms or technologies, or, crudely put, ‘pickin g winners’) became the arena for the industrial policy debate in the 1990s. The mid-1980s neoliberal interpretation of East Asian success, that it was due to free trade and other non-interventionist policies, was subjected to intense criticism. It was noted that most successful Asian industrializers had been very interventionist in trade, FDI, technology transfer and domestic resource allocation. 17 The evidence was so overwhelming that the neoliberal camp was forced to admit the facts of the case. However, admitting that the most dynamic economies had ‘ picked winners’ created difficulties for neoliberals, as the normal — and in this case valid — interpretation would be that performance and policy were causally related. They responded with a ‘ moderate neoclassical’ stance (in contrast to the earlier ‘ strong neoclassical’ one that assumed all markets to be efficient) that devoted enormous effort to explaining why selectivity, while it existed, had been redundant and unnecessary (World Bank, 1993). 18 The moderate school admitted some market failures and some role for the state, but only as long as interventions were functional— it saw no valid role for policy in influencing allocation at the activity, firm or technological level The ‘ market friendly’ approach, as it was appealingly labelled, segmented market . failures not according to whether market failures existed but according to the level at which policies affected investment decisions. That neoclassical theory provides no reason for such a distinction — afte r all, if policy can correct a market failure it is justified — was countered by a political economy premise, that it was impossible for governments to mount effective selective interventions . The World Bank (1993) admitted that some selectivity may have worked in East Asia, but the circumstances had been unique. Other governments did not and could not have the kinds of capabilities needed, and so selectivity would do more harm than good. The moderate position, later termed the ‘ Washington consensus’, happily coincided with the World Bank’s own operations (in health, education and infrastructure), policy advice (greater liberalization) and structural adjustment programmes (stabilization, liberalization and privatization). The moderate position retained the simplifying assumptions of the strong neoclassical position on technology. Both used, implicitly or explicitly, the basic neoclassical model in which all markets affecting technology are ‘ efficient’. In the theoretical sense, ‘efficiency’ has stringent requirements: product markets give the correct signals for investment and factor markets respond to these signals. At the firm level there are no scale economies or externalities. Firms have perfect information and foresight and full knowledge of all available technologies. They choose the right technology if faced with free market prices. Having selected the right technology they use it instantaneously at ‘ best practice’. There are no significant learning processes, no risks, no externalities and no deficienc ies in the skills, finance, information and infrastructure available to them. assume that developing countries do best by opening themselves to inflows of information embodied in trade and investment. Access to new technology becomes equivalent to its effective use. The policy implications of the models that follow from externalities, increasing returns and non-appropriability in innovation apply only to advanced countries; the development implications, in so far as they are mentioned, are the same as in standard neoclassical analyses. 17 The objections to the strong neoliberal position came from such authors as Amsden (1989), Lall (1992), Pack and Westphal (1986), Wade (1990) and Westphal (1982 and 1990). 18 The strong neoliberal stance was that no markets failed and that there was no role for the government apart from providing basic public goods and a stable setting for market driven activity. For a critique of the World Bank (1993) publication see Lall (1996) and for a recent restatement of the moderate neoclassical position see Noland and Pack (2003). QEH Working Paper Series — QEHWPS111 Page 14 In this model, any policy intervention that affects the prices facing enterprises is by definition distorting, and moves society away from the optimum allocation yielded by free markets. 19 The critical assumption for industrial policy is the one on learning and capability building and dropping it yields very different conclusions for policy (below). But showing that there may be market failures in importing and using technology cannot establish a case for selectivity. It is also necessary to show that such failures are important in practice and not theoretical curiosities, and to establish that governments can effectively remedy them in real life, that government failures are not necessarily more costly than market failures. It is argued here that both can be shown, and the transition from an admittedly simplified neoclassical model to a universal, timeless neoliberal policy diktat is not justified in theory, history or practice. 20 To do this we turn to the structuralist approach to technology in developing countries. 3. 2 The technological capability approach How enterprises in developing countries actually use technology is analysed by a large recent literature on technological capabilities. 21 The literature is mainly empirical but has its theoretical roots in the evolutionary approach of Nelson and Winter (1982) and the modern information theory of Stiglitz. 22 It argues that industrial success in developing countries depends essentially on how enterprises manage the process of mastering, adapting and improving upon existing technologies. The process is difficult and prone to widespread and diffuse market failures, with have important implications for policy (see Box 1). Technology has strong ‘ tacit’ elements that need the user to invest in new skills, routines, and technical and organizational information. Such investment faces market and institutional failures whose remedies 19 Neoclassical economists admit the possibility of market failure arising from such textbook cases as monopoly, public goods and some externalities, although they tend to treat failures as special cases rather than the rule. The market failures that may call for selective interventions are capital market deficiencies, scale economies and externalities arising from the imperfect appropriability of investments in knowledge, technology, and skills. However, the admission that these theoretical possibilities exist does not translate into recommendations that government actually mount selective policies to overcome them (as in the World Bank, 1993). Moreover, the neglect of firm-level learning processes (below) means that the list of market failures remains incomplete – the most critical ones for developing countries are ignored. For a longer discussion see Lall and Teubal (1998). 20 Wade, in the introduction to the forthcoming new edition of his path-breaking book of industrial policy in Taiwan, Governing the Market, says: “ The remarkable thing about the core Washington Consensus package is the gulf between the confidence with which it is promulgated and the strength of supporting evidence, historical or contemporary. There is virtually no good evidence that the creation of efficient, rent-free markets coupled with efficient, corruption-free public sectors is even close to being a necessary or sufficient condition for a dynamic capitalist economy. Almost all now-developed countries went through stages of industrial assistance policy before the capabilities of their firms reached the point where a policy of (more or less) free trade was declared to be in the national interest. Britain was protectionist when it was trying to catch up with Holland. Germany was protectionist when trying to catch up with Britain. The United States was protectionist when trying to catch up with Britain and Germany, right up to the end of the World War II. 20 Japan was protectionist for most of the twentieth century up to the 1970s, Korea and Taiwan to the 1990s. Hong Kong and Singapore are the great exceptions on the trade front, in that they did have free trade and they did catch up –but they are city-states and not to be treated as economic countries. In Europe some countries abutting fast-growing centres of accumulation were also exceptions, thanks to the ‘ ink blot’ effect. But by and large, countries that have caught up with the club of wealthy industrial countries have tended to follow the prescription of Friedrich List, the German catch-up theorist writing in the 1840s: “ In order to allow freedom of trade to operate naturally, the less advanced nation [read: Germany] must first be raised by artificial measures to that stage of cultivation to which the English nation has been artificially elevated”” (Wade, 2003). For a longer historical perspective see Reinert (1995). See Lall (1992, 1996, 2001), Westphal (2002), UNIDO (2002). In his analysis of East Asian success Stiglitz (1996) argues that “… whenever information was imperfect or markets were incomplete, government could devise interventions that filled in for these interventions and that could make everyone better off. Because information was never perfect and markets never complete, these results completely undermined the standard theoretical basis for relying on the market mechanism. Similarly the standard models ignored changes in technology; for a variety of reasons markets may under-invest in research and development … Because developing economies have underdeveloped (missing) markets and imperfect information and because the development process is associated with acquiring new technology (new information), these reservations about the adequacy of market mechanisms may be particularly relevant to developing countries. ” P. 156, emphasis added. 22 21 QEH Working Paper Series — QEHWPS111 Page 15 require intervention. Many interventions have to be selective because technologies differ inherently in their tacit features and externalities. Industrial success in the developing world — and indeed in the presently developed world in its early phases of industrialization — is thus traceable to how effectively governments have overcome these market and institutional failures. Box 1: Ten features of technological learning in developing countries Technological learning is a real and significant process. It is vital to industrial development, and is primarily conscious and purposive rather than automatic and passive. Firms using a given technology for similar periods need not be equally proficient: each will be at the point given by the intensity of its capability building efforts. Firms do not have full information on technical alternatives. They function with imperfect, variable and rather hazy knowledge of technologies they are using. There is no uniform, predictable learning curve for a given technology . Each faces risk, uncertainty and cost. Differences in learning are larger between countries at differing levels of development. Firms may not know how to build up the necessary capabilities – learning itself often has to be learned. In a developing country, knowledge of traditional technologies may not be a good base on which to know how to master modern technologies. For a latecomer to a technology, the fact that others have already undergone the learning process is both a benefit and a cost. It is a benefit in that they can borrow from the others’ experience (to the extent this is accessible). It is a cost in that they are relatively inefficient during the process (and so have to bear a loss if they compete on open markets). The cost and risk depend on how new the technology is relative to the entrant’s base of knowledge, how developed factor markets are and how fast the technology is changing. Firms cope with these uncertain conditions not by maximising a well-defined function but by developing organisational and managerial routines (Nelson and Winter, 1982). These are adapted as firms collect new information, learn from experience and imitate other firms. Learning is path dependent and cumulative. The learning process is highly technology specific, since technologies differ in their learning requirements. Some technologies are more embodied in equipment while others have greater tacit elements. Process technologies (like chemicals) are more embodied than engineering technologies (machinery or automobiles), and demand different (often less) effort. Capabilities built up in one activity are not easily transferable to another. Different technologies involve different breadth of skills and knowledge, some needing a narrow range of specialization and others a wide range. Different technologies have different degrees of dependence on outside sources of knowledge or information, such as other firms, consultants, capital goods suppliers or technology institutions. Capability building occurs at all levels – shop-floor, process or product engineering, quality management, maintenance, procurement, inventory control, outbound logistics and relations with other firms and institutions. Innovation in the conventional sense of formal R&D is at one end of the spectrum of technological activity; it does not exhaust it. However, R&D does become important as more complex technologies are used; R&D is needed just for efficient absorption. Technological development can take place to different depths. The attainment of a minimum level of operational capability (know-how) is essential to all activity. This may not lead to the development of deeper capabilities, an understanding of the principles of the technology (know-why): this requires a discrete strategy to invest in deepening. The deeper the levels of technological capabilities aimed at, the higher the cost, risk and duration involved. It is possible for an enterprise to become efficient at the know-how level and stay there, but this is not optimal for its long-term capability development. It will remain dependent on other firms for all major improvements to its technologies, and constrained in what it can obtain and use. The development of know-why allows firms to select better the technologies they need, lower the costs of buying those technologies, realise more value by adding their own knowledge, and to develop autonomous innovative capabilities. Technological learning is rife with externalities and inter-linkages. It is driven by direct interactions are with suppliers of inputs or capital goods, competitors, customers, consultants, and technology suppliers. Others are with firms in unrelated industries, technology institutes, extension services, universities, industry associations and training institutions. Where information and skill flows are particularly dense in a set of related activities, clusters of industries emerge, with collective learning for the group as a whole. Technological interactions occur within a country and abroad. Imported technology provides the most important input into technological learning in developing countries. Since technologies change constantly, moreover, access to foreign sources of innovation is vital to continued technological progress. Technology import is not, however, a substitute for indigenous capability development – the efficacy with which imported technologies are used depends on local efforts. Similarly, not all modes of technology import are equally conducive to indigenous learning. Much depends on how the technology is packaged with complementary factors, whether or not it is available from other sources, how fast it is changing, how developed local capabilities are, and the policies adopted to stimulate transfer and deepening. Source: Lall (2001). QEH Working Paper Series — QEHWPS111 Page 16 The process of gaining technological mastery in a new setting is not instantaneous, costless or automatic, even if the technology is well diffused elsewhere. It is risky and unpredictable, and the process itself may have to be learnt. The cost and duration of the learning process varies by the complexity and scale of the technology; becoming an efficient garment assembler, say, is far less costly and difficult than learning to make automobiles. Moreover, the process is rife with externalities: firm do not learn on their own but in interaction with other firms (suppliers, buyers, consultants and competitors) and institutions. And it often requires inputs from factor markets: physical inputs, new skills, technical information and testing or trouble -shooting services, finance and new infrastructure. The costs of the process rise with the degree of industrial backwardness of the economy. Capability development can face market failures in building initial capacity and in subsequent deepening. Both need support, functional and selective. Support entails a mixture of policies apart from infant industry protection. 23 Take building initial capacity in new industrial activities. Free markets may not give correct signals for investment in new technologies when there are high, unpredictable learning costs and widespread externalities. This is, in modern garb, the classic case for infant industry protection : classical economists clearly recognised that in the presence of such costs, an industrial latecomer faced an inherent disadvantage compared to those that had undergone the learning process. 24 Add to this the extra costs and disadvantages faced by firms in developing countries: unpredictability, lack of information, weak capital markets, absence of suppliers, poor support institutions and so on: exposure to full import competition is likely to prevent entry i to activities with relatively difficult technologies. Yet these are the technologies n that are likely to carry the burden of industrial development and future competitiveness. Why do these interventions have to be selective? Offering uniform protection to all activities makes little sense when learning processes and externalities differ by technology, as they inevitably do. In some activities the need for protection may be minimal because the learning period is relatively brief, information easy to get and externalities limited. In complex activities or those with widespread externalities, newcomers may never enter unless measures are undertaken to promote the activity. The only complex activities where investments may take place without promotion are those based on local natural resources, if the resource advantage is sufficient to offset the learning costs. However, the processing of some resources calls for strong industrial capabilities and for a learning base; thus, both Sub-Saharan Africa and Latin America have large resource bases but advanced processing has only taken root in the latter, based on decades of capability building in import-substituting regimes. It is important to reiterate that infant industry protection is only part of industrial polic y, and by itself can be harmful and ineffective. This is so for two reasons. First, protection cannot succeed if it is not offset by competitive pressures on firms to invest in the capability building process. In fact, by cushioning the costs 23 24 See the contributions by Wade and Lall in Wood (ed.) (2003). On the case for infant industry protection John Stuart Mill, the most powerful advocate of free trade in classical economic thought, says: ” The only case in which, on mere principles of political economy, protecting duties can be defensible, is when they are imposed temporarily (especially in a young and rising nation) in the hopes of naturalising a foreign industry, in itself perfectly suitable to the circumstances of the country. The superiority of one country over another in a branch of production often arises only from having begun it sooner. There may be no inherent advantage on one part, or disadvantage in another, but only a present superiority of acquired skill and experience… But it cannot be expected that individuals should, at their own risk, or rather to their certain loss, introduce a new manufacture, and bear the burden of carrying on until the producers have been educated to the level of those with whom the processes are traditional. A protective duty, continued for a reasonable time, might sometimes be the least inconvenient mode in which the nation can tax itself for the support of such an experiment. But it is essential that the protection should be confined to cases in which there is good ground for assurance that the industry which it fosters will after a time be able to dispense with it; nor should the domestic producers ever be allowed to expect that it will be continued to them beyond the time necessary for a fair trial of what they are capable of accomplishing.” Mill (1940), p. 922, italics added. The 19th century saw intense debates, particularly in the US, on the need for infant industry protection, and most early industrializing countries used the tool extensively. QEH Working Paper Series — QEHWPS111 Page 17 of capability building, protection removes the incentive for undertaking it. One of the reasons why industrial policy failed in most developing countries is precisely that they failed to overcome this dilemma. But it is possible to do so, by strengthening domestic competition, setting performance targets and, most effectively, by forcing firms into export markets where they have to compete with best practice. Infant industry protection only works well where it is counterbalanced by such measures. Many such measures also have to be selective , since the costs of entering export markets differ by product. Thus, differentiated export targets, credits and subsidies were often used in East Asia. The second reason why industrial policy is far more than protection is the need for coordination with factor markets. Firms need many new inputs into their learning : new skills, technical and market information, risk finance, or new infrastructure. Unless factor markets can respond to these needs, protection cannot allow them to reach competitive levels of competence. And factor market interventions also have to be selective as well as functional, for three reasons. First, several factor market needs are specific to particular activities; if they lack the information or coordination to meet these needs, interventions are needed to remedy the deficiencies. For instance, the skill needs of electronics may not be fully foreseen by education markets, 25 or the financial needs emerging new technologies may not be addressed by capital markets. Second, government resources for supporting factor markets are limited, and allocating them among competing uses entails selectivity at a high level (say, between education and other uses). Third, where the government is already targeting particular sectors in product markets, factor markets have to be geared to those activities if the strategy is to succeed. The deepening of capabilities suffers similar problems. The more complex the functions to be undertaken, the higher the costs involved and the greater the factor market coordination required. Getting into production may be easy compared to design, development and innovation. Neoclassical theory accepts that free markets (implicitly in industrial economies) may fail to ensure optimal private innovative activity because of imperfect appropriability of information. However, developing countries face an additional problem. It is generally easier to import foreign technologies fully packaged than to develop an understanding of the basic principles involved — the basis of local design and development. ‘ Internalized’ technology transfer takes the form of wholly foreign-owned direct investment. This is an effective and rapid way to access new technology, but it may result in little capability acquisition in the host country apart from production skills. 26 The move from production to innovative activity involves a strategic decision that foreign investors, because of the skills and technical linkages involved, tend to be unwilling to take in developing countries. While some relocation of innovative activity is taking place (UNCTAD 2002), it is largely in advanced countries and a few newly-industrializing economies. There is, in other words, a risk of market failure in capability deepening because of the learning costs involved, similar to initial capability building. To ensure socially optimal allocation, it may be necessary to (selectively) restrict technology imports in ‘ internalized’ forms (via FDI) and promote those in 25 On the selectivity of education and training policies in East Asia, and their intimate relationship to industrial policy more narrowly defined, see Ashton et al. (1999). Also see Narula (2003). TNCs also have to undergo costly capability development in new locations but the costs are generally lower for them. They know how to go about building capabilities, have ‘ deeper pockets’, more information and better training resources. If a developing host country engages only in simple assembly operations, TNCs may be able to achieve competitive production without protection because the learning period is short and relatively predictable. However, deepening and diversification into more advanced activities or functions may need government support to improve the quality of local factors and suppliers and to induce TNCs to transfer these activities and functions. This may not involve protection if the local workforce is sufficiently skilled — the Singapore story. However, Singapore had to use a battery of selective interventions to attract and target TNCs and provide them with the factor inputs, infrastructure and incentives needed to force the pace of upgrading. FDI may reduce the need for interventions for capability building but cannot remove it altogether. Once countries move beyond simple processing, they have to provide the factors that allow TNCs to undertake complex functions efficiently. 26 QEH Working Paper Series — QEHWPS111 Page 18 ‘ externalized’ forms (licensing, equipment, imitation or OEM contracts). Over history most countries that have build strong local innovative capabilities have done it in local firms, often by restricting FDI selectively (see below). Some have done it partially by stimulating foreign investors to in vest in R&D, but this has also involved selective interventions. Thus, it is not just interventions in trade that matter but also in the way in which technologies are transferred: complete openness to internalized technology imports may not be a good thing if it truncates the process of technological deepening and internalized transfers may need to be subjected to interventions to extract greater technological benefits. Does the globalization of production change matters? The spread of integrated systems means that many technologies are now only available through FDI (Radosevic, 1999). It also means that countries that get into the low end of sophisticated activities can reap enormous export benefits. This makes the cost of restricting FDI much higher. Rapid technical change also makes it more risky to bypass global systems in building capabilities. While this is true, it does not demolish the case for policies to promote deepening. The growth of global sourcing has made it easier to become