1 1 Industrial Policies in the 21 st Century: Summarising the Conference and Looking Beyond Ha-Joon Chang Faculty of Economics University of Cambridge 1. Introduction: Backgrounds to the Conference This has been a remarkable conference, with many impressive presentations and exceptionally sophisticated discussions. It has been an occasion that befits the 60 th anniversary of one of the biggest and most successful agents for industrial policy in the world that is, the BNDES. It will be remembered as one of the most important conferences on the topic of industrial policy held in Latin America and beyond. What makes the conference even more important is that it is held at a very critical juncture in which the need for industrial policy is felt the keenest since the 1980s. First, the 2008 global financial crisis has shaken confidence in the neo-liberal model that has dominated the world in the last three decades. All over the world, people are looking for alternatives to neo-liberalism. Even industrial policy, the bête noir of neo-liberalism, is not a taboo any more, including in the US and Britain, the homes of neo-liberalism.
2 2 Second, in Brazil and the rest of Latin America, after three decades of failure of neo-liberal policies to generate growth (especially productivity growth) and structural changes, there is a revival of interests in industrial policy. Third, in the last decade or so, the rise of China has created new, large-scale challenges (tougher competitions in manufacturing export markets and in the attraction of foreign direct investment, or FDI) and opportunities (commodity price boom that provides extra resources to the exporting countries) for the rest of the developing world. Dealing with these challenges and opportunities require a good industrial policy that will direct the extra resources into the most socially-productive activities, ensure productivity growth, and provide the right collective inputs (infrastructure, skills, etc.). In this paper, I will summarise the main points of debate in this conference and critically reflect on them. 2. Definitions of Industrial Policy Definition of industrial policy has always been a major issue in the debate on it, reflecting the contentious nature of the debate. The participants in the conference agreed that there is no single best definition of industrial policy but that it may be broadly defined as a policy to promote structural change through technological upgrading and productivity increase by means of microeconomic measures that change the parameters and the incentives for actions by productive enterprises. The participants agreed that industrial policy inevitably involves some selectivity and cannot be entirely horizontal, affecting all sectors equally. First of
3 3 all, limited resources (including government attention and administrative resources) means that industrial policy cannot promote all sectors. Second, different sectors need and bring different things, so choices have to be made about what sector to promote and how to promote them. Third, it needs to be accepted that there are very few genuinely horizontal policies that affect all sectors equally. Even things like infrastructure, R&D subsidies, or education, which many people take as examples of horizontal policies involve some degree of selectivity infrastructure is mostly location specific, engineers and scientists are highly specialised, and R&D supports favour R&D-intensive industries over others. So, if selectivity is inevitable in industrial policy, it may be better if we stopped pretending that we should only do horizontal industrial policy. Pretending so will only mean that we will be discriminating between sectors without realising that we are. 3. Scope for Industrial Policy Industrial policy has traditionally been equated with promotion of manufacturing, but there is an increasing recognition that industrial policy could target some non-manufacturing sectors that have high productivity growth, greater scope for innovation, rapidly-growing demands, and high tradability. It is increasingly recognised that some producer services (e.g., engineering, design, maintenance) may have such characters. And indeed some countries (e.g., Singapore with aeroplane maintenance) have tried to promote those services through industrial policy.
4 4 Less widely recognised is the potential for high-productivity agriculture. For example, the Netherlands, despite having the 5 th highest population density in the world (excluding city states and island states), is the 3 rd largest exporter of agriculture because its agriculture had been made into a high-tech industry through a mixture of the cooperative movement and industrial policy (e.g., support for R&D, establishment of agricultural universities). While Singapore and the Netherlands show that industrial policy do not have to be targeted at a manufacturing industry, they do not suggest that countries can do away with manufacturing. The Netherlands agricultural success has been possible only thanks to its manufacturing strengths, supplying high-quality chemicals and electronics equipment (used for computer-controlled feeding). Singapore s success in producer services has been possible only because of its general strengths in manufacturing, which produce able and experiences engineers, technicians, and skilled workers. Indeed, it is increasingly recognised that producer services are likely to eventually follow their manufacturing activities, given the importance of geographical proximity, shared traditions, and continued interactions between different stakeholders in learning and innovation. Ron Bloom made this point very well in his presentation. 4. Relationship with Other Policies Industrial policy interacts with other policies, which may constrain or promote each other. An intelligent industrial policy design needs to take these interactions into account.
5 5 Industrial policy may be in direct competition with other policies that also need fiscal outlays (e.g., social policy, education policy for resources). In this case, there are trade-offs between industrial policy and other policies. Many participants, especially Nelson Barbosa Filho, pointed out that the nature of trade-offs partly depends on the time frame with which we operate. For example, devaluation has immediate effects on real wages but its impacts on productivity growth appear with a lag. Also, the success of industrial policy may be partly dependent on other policies. For example, as emphasised by Barbosa s presentation, macroeconomic policy can constrain industrial policy by restricting credits, making credits expensive, making currency overvalued, or increasing financial volatility. Alternatively, macroeconomic policy may help industrial policy by generating extra demands, either domestically (by increasing government spending and/or business investments) or internationally (by weakening the currency). However, these relationships are rather complex and context dependent (country size, currency autonomy, central bank independence, degree of initial over/under-valuation of currency, exchange rate regime, financial regulation, capital account liberalisation, wage setting regimes, etc.), so care is needed in delineating the relationships. 5. Tools of Industrial Policy The participants discussed the possible tools of industrial policy and their respective strengths and weaknesses.
6 6 Industrial policy is often equated with handing out subsidies, but it is important to acknowledge that it may be conducted through methods that do not require fiscal outlays. First, industrial policy can involve coordination between complementary activities. This is the reasoning behind the Big Push arguments. Second, it can also be achieved through competition policy in the broad sense/ This may involve not only restricting monopoly but also restrictions on excessive competition (to achieve scale economy). It may also involve allowance, or even encouragement, of other cooperative arrangements (e.g., exemption from anti-trust law for joint activities by small firms). Third, industrial policy may involve regulations on TNCs. The relevant tools include local contents requirements, conditions for technology transfer, ban on obsolete technology, requirements for export, training, or joint venture. Fourth, industrial policy may be conducted through standard setting. Quality control for exports by the government, setting of technology standards (especially in hi-tech industries), requirements for minimum scale of production are examples. Fifth, industrial policy may involve foreign exchange rationing according to industrial policy priorities. The best example is the foreign exchange budgeting exercise of South Korea between the 1960s and the 1980s, in which foreign exchange was rationed according to how much their buyers are going to contribute to industrial development (giving priority to the imports of capital goods and intermediate goods over consumer goods). All of this mean that the talk of the need for the fiscal space in the use of industrial policy may be, while important, exaggerated.
7 7 6. Factors Affecting the Success of Industrial Policy The participants discussed a large number of the factors affecting the success or otherwise of industrial policy. I put under three broad headings structural factors, implementation issues, external factors. I then divide these broad categories into several individual factors Structural Conditions Country Size Larger countries are likely to benefit more from protectionism than smaller countries, given the importance of scale economy in many industries. Brazil and Mexico in the ISI period ( ), the US in the 19 th century, Japan in the 20 th century, as well as today s China are good examples. However, if they use it selectively and if they combine with export drive, small countries can also benefit from protectionism, as seen in cases like Sweden and Finland Economic Structure Dualistic economic structure (that is, the coexistence of the modern sector and the traditional sector) may magnify the impacts of industrial policy on structural changes, as pointed out by Jaime Ros in his presentation Catch-up Industrial policy is deemed easier in catch-up economies than in the more advanced ones, as information requirements are less.
8 8 However, catch-up economies may lack the necessary administrative and institutional capabilities to conduct industrial policy well, so the net benefit of being in a catch-up position in relation to industrial policy is not necessarily clear Education It is commonly accepted that education affects the success of industrial policy by affecting the absorptive capacity of an economy. However, as Justin Lin pointed out, educational investments need to be made in accordance with the development of the productive structure, as otherwise we may end up producing a lot of highly-educated unemployed people. Few would dispute that education is important, both for its own sake and for economic development, but, as in many other things, the devil s in the details. First, in most countries, too much emphasis is put on formal education, while on-the-job training or vocational qualifications are frequently neglected or underresourced. Second, even within formal education, countries have focused either on university education or (especially when they are at lower levels of development) primary education, at the cost of secondary education, especially secondary technical education in the German mould, which may be more important for industrial development. Third, even at higher levels of economic development and even in this age of the knowledge economy, there may be simply too much university education (from the point of view of economic development, and not necessarily from the point of view of helping people realise themselves or producing good citizens). This
9 9 conjecture is most dramatically illustrated by the fact that Switzerland until recently had university enrolment ratio, at 10-15%, that is less than 1/3 the OECD average Taxes The currently popular view is that lower taxes, especially for the rich people and the corporations encourage investment and growth. However, even a casual look at the evidence suggests that this view is highly simplistic. Few investors are rushing into Albania, despite the corporation income tax rate of 10%, while few rich people migrate to Jamaica, despite it having the top income tax rate of 5%. Sweden and Finland have grown faster than the US since WWII, including the last 30 years of supposed renaissance of the US economy, despite having nearly double the tax burdens. The point is that the taxes are often spent on things that help business infrastructure, R&D subsidies, education, worker training, or even the welfare state (in so far as it promotes social peace and labour mobility through unemployment insurance and re-training schemes) Implementation Issues Targeting Targeting commonly but misleadingly known as picking winners and what Ludovico Alcorta called in the conference, more correctly, making strategic choices has been hotly disputed in all debates on industrial policy. Many people suggest that we should do away with targeting altogether and use only horizontal industrial policies that affect all sectors equally. However, this view is very limited.
10 10 First, very few things do not involve targeting at some level. As we mentioned earlier (section 2), even things like supporting education (beyond primary education), infrastructure, and R&D typical examples of horizontal industrial policy involve some degree of targeting. Targeting is a matter of degree, rather than a black-andwhite choice. Second, the private sector targets all the time. In this context, the comment by Alcorta that in designing industrial policy governments need to learn from the private sector strategic decision-making is very apt. Third, targeting fails all the time (which is in the nature of any strategic entrepreneurial choices), both in the public and the private sectors, so the crucial question is not whether targeting may fail (as it will) but how to improve the batting average of the decision-makers. Having said this, it should be accepted that no amount of effort will equalise the batting average across decision-makers. Like any entrepreneurial decision, in industrial policy decisions there will always be an element of flair that cannot be reduced to information processing and logical thinking. Fourth, it is not true that less targeted policies are necessarily more effective. Less targeted policy may require less information and are less open to corruption and lobbying (see section 6.2.6), but it is likely to have more leakages. Indeed, many opponents of industrial policy support greater targeting in social policy on the ground that the former (leakage) is a more serious problem than the latter (information requirement). Fifth, all the participants agreed that targeting needs to be realistic. For example, Lin in his presentation suggested that, based on the theory of comparative advantage, it may be a good rule of thumb to target industries that doing well in countries with 2-3 times perhaps maximum four times the per capita income of
11 11 your own country that may become internationally competitive with 3-5 years of subsidies. However, during the conference Ha-Joon Chang pointed out that, while Lin s suggestion is a sensible base line, there are many examples in which countries successfully broke into industries that were totally out of line with their comparative advantages the Japanese automobile in the 1950s and the Korean steel in the 1960s being the most dramatic examples. Also citing the Korean example, Carl Dahlman emphasised the need to take leaps. This was also echoed by the comments by Richard Kozul-Wright. Chang likened comparative advantage as the compass in your long journey for economic development the compass is vital in telling you where you stand in relation to other things, but it does not tell you where to go and how to get there. You need to decide your destination and the method of travel yourself Political Regimes A common view is that successful industrial policy requires authoritarianism based on the experiences of countries like South Korea, Taiwan, and Singapore. However, there are many countries that have successfully used industrial policy under democracy after WWII France, Japan, Norway, Finland, and Austria. Indeed, democratic consensuses are far more durable than decisions imposed by authoritarian regimes, so it cannot be presumed that a more authoritarian regime will be necessary better at implementing policies Policy Coherence and Coordination
12 12 In line with what was discussed earlier, especially in relation to macroeconomic policy, the participants agreed that industrial policy needs to be coordinated with other policies. Many industrial policy attempts have not worked well because of the silo approach taken, which often meant different parts of the government working at cross-purposes. There is a need for the so-called joined-up government The Quality of the Bureaucracy There can be no dispute that a competent bureaucracy is needed for a successful implementation of industrial policy. However, this sensible point should not be exaggerated. Even those success stories of industrial policy did not start with exceptionally capable bureaucracies. For example, until the late 1960s and the early 1970s, South Korea was sending its officials to Pakistan and the Philippines for extra training. Taiwan also had to spend a lot of time in the 1950s and the 1960s fixing its corrupt and incompetent bureaucracy, which was after all the government machinery that lost Mainland China to the Communists. Bureaucratic capabilities can be built over time, partly through deliberate investments in training and partly through learning by doing in administration. And insofar as the latter is important, countries will have to try some industrial policy, if they are to build the necessary administrative capabilities Public-Private Consultation Mechanism All the participants agreed that the quality of the public-private consultation mechanism is a crucial factor determining the success of industrial policy, as
13 13 highlighted by the experiences of industrial policy in East Asia (the famous deliberation council of Japan). Unfortunately, the conventional wisdom is that the quality of this mechanism is given by immutable factors like the past history of private-public interaction and therefore that how well a country can conduct industrial policy is more or less determined by its history. However, according to Arun Maira s excellent presentation, the recent Indian attempt shows that it is possible to deliberately construct this mechanism. Indeed, contrary to the popular belief, such mechanism was deliberately constructed and nurtured, rather than historically inherited, in the East Asian countries themselves. The participants agreed that different forms of consultation mechanism would work differently in different countries, so each country needs to construct its own mechanism. Bloom and other participants also emphasised that the mechanism needs to be flexible, as we cannot know everything in advance and therefore may need to modify the mechanism along the way Lobbying and Corruption One standard worry about industrial policy is that it may open door for lobbying and corruption, especially given its targeted nature. However, a number of participants emphasised that this is not an inevitability. For one thing, interest groups may go above their sectional interests and accept outcomes that may not be the best for them, as shown in the Mauritius case study presented by Mahamood Cheroo. For another, governments can put in mechanisms to reduce lobbying and corruption for example, by announcing clear performance indicators in advance and punishing non-performers accordingly.
14 14 Also, it was pointed out that the absence of industrial policy does not guarantee the absence of lobbying, as interest groups may lobby for less sectorspecific issues but nonetheless for their sectional interests the lobbying for inflation control by the financial industry is the classic example of this External Conditions Changes in Global Rules The participants noted that the establishment of the WTO and the proliferation of (bilateral and regional) FTAs and BITs have put serious constraints on the types of industrial policy tools that may be used. These have led to bans on quotas (and other quantitative restrictions), bans on subsidies (except for export subsidies by the LDCs and for those subsidies frequently used by the rich countries, such as those for agriculture, R&D, and regional disparity), severe restrictions on FDI, and the strengthening of intellectual property rights (IPR) protection. All of this have resulted in a significant shrinkage of policy space for the developing countries. However, many participants have pointed out that there is still policy space left and warned against these new constraints becoming excuses for policy inaction Finance-driven globalisation Many participants noted that the dominant form of globalisation of the last three decades, namely, finance-driven globalisation, has also created an international economic environment that makes industrial policy more difficult. At the micro-level, finance-driven globalisation has encouraged corporations and households to behave in more short-termist ways, which goes against long-term-
15 15 oriented industrial policy. At the macro-level, it has created numerous credit-fuelled consumption booms and busts (Chile, 1982; Mexico, Southeast Asia, Brazil, and Argentina in the 1990s; South Africa today), which make long-term-oriented policies, like industrial policy, more difficult to implement Changing Global Production Structure It was noted by many participants that the rise of global value chains, while creating some new opportunities for developing countries in some industries (now you don t have to assemble cars in order to participate in the car industry, as put by Raymond Atje), has put restrictions on what countries can do in terms of industrial policy. First, oligopolies have been strengthened at the top of the value chain, making it difficult for companies down the chain to move up the chain. However, it should be noted that in the long run even the companies at the top can lose their positions, not least because of the past success of a competitor country s industrial policy, as seen in the case of the rise of companies like Toyota, Nokia, Samsung, and Hyundai. Second, the rise of the global value chain means that benefits that are created through industrial policy are more likely to leak out of the national economy, when the beneficiaries are subsidiaries or suppliers of transnational companies (TNCs). Third, the increasing importance of global value chains makes it even more necessary to come up with policies that maximise a country s value capture without violating the new global rules. It was noted that the link in the value chain where the greatest values can be captured may not be obvious. Chang provided the example of the flat-screen TV, where the highest share of value accrues to the company manufacturing the flat glass.
16 16 Last but not least, as Kozul-Wright emphasised, FDI is a lag, rather than leading, variable and therefore recognising the importance of the global value chain does not mean that countries can develop by simply welcoming and subsidising FDI. The benefits from FDI are maximised when there are regulations to make it sure that the TNCs involved pass on knowledge and help expand local productive capabilities Changing Global Economic Relationships As noticed in the introduction, the rise of China and the relative decline of the traditional economic power centres is one major change to the global economic relationship. For other developing countries, the rise of China is both a challenge and an opportunity. It is a challenge in the sense that countries wanting to develop their manufacturing industries need to compete against China, with very high levels of skills and technologies given its level of wages. The opportunities arise from the fact that China is a huge market and also increasingly a major investor, as many resourcerich developing countries are finding out. One challenge would be for developing countries to break into the Chinese market with manufacturing exports. Another change in the global economic relationship is the increasing regional integration. A lot of globalisation in the last three decades has been driven by regional, rather than truly global, economic integration. The European integration is the most important in this, but the emergence of closer economic relationships within East Asia and the launch of the NAFTA are other examples. Some regional integration agreements among developing countries have also emerged or been revamped, such as the establishment of the ALBA in Latin America or the revamping of the East African Community in Africa.
17 17 Last but not least, there has been a move towards increasing South-South cooperation beyond regional cooperation. Especially through the WTO negotiations, G-77 and other cooperative arrangements between developing countries have been strengthened in international negotiations, with the IBSA countries (India, Brazil, South Africa) emerging as the leaders in the process. China, India, Brazil, and other key developing countries have emerged as important export markets and sources of FDI for other developing countries in recent periods. While these cooperation arrangements are still in their early stages and their impacts are yet insignificant in global terms (although they may be important for individual countries), these could be the bases for different international economic relationships in the future External shocks The participants agreed that external shocks, both positive and negative, can affect industrial policy. On the one hand, negative external shocks may derail industrial policy by creating a balance of payments problem. The best example is the 1982 Third World Debt Crisis, which followed the 1979 Oil Shock and the Volcker shock (that is, the steep increase in US interest rate under Paul Volcker s chairmanship of the Federal Reserve Board), which derailed industrial policy in Brazil and many other developing countries. On the other hand, positive external shocks for example, commodity price booms in the case of commodity-exporting countries may grant extra resources for industrial policy. In addition, it was noted by some participants that the form of a country s industrial policy matters in determining how external shocks affect its industrial policy. For example, countries with industrial policy measures geared towards promoting exports will be able to handle negative external shocks better. For another
18 18 example, a country with a good mechanism to implement industrial policy would be able to exploit positive external shocks more profitably. 7. Real-World Experiences Finally, a number of participants argued that better knowledge of real world experiences of (successful or otherwise) industrial policy is required in an intelligent design of industrial policy. In the debate on industrial policy, there has been a lot of ideologicallymotivated (or ideologically-blinded) misrepresentation of real-life cases. The most extreme example of this is the presentation of countries like Korea and Taiwan as free-market, free-trade success stories in the earlier phase of the modern debate on industrial policy, in the 1970s and the 1980s. This was quite a remarkable misrepresentation, given that anyone who had any knowledge about these economies would have known that they were no such things. Less blatant but in many ways more important is the persistent myth that most of today s rich countries, especially Britain and the US, became rich through freemarket and free-trade policies. Fortunately, it is increasingly accepted the virtually all of today s rich countries used protectionism, subsidies, regulation of FDI, deliberately lax IPR laws, and many other measures of industrial policy in their earlier stages of development. Even in terms of more contemporary success stories, there are a lot of partial and misleading presentations of evidence, not least because we are all prisoners of our theories and therefore often fail to see the parts of the reality that do not conform to our prejudices you could say that life is stranger than fiction.
19 19 For example, Singapore s free-trade policy and welcoming attitude towards FDI have blinded people from the facts that all the land in the country is publiclyowned, that the state-owned housing corporation supplies around 85% of all houses, that 22% of the country s GDP is produced by SOEs, and that its promotion of FDI is highly targeted and interventionist. (Ireland s FDI policy since the 1980s also resembled that of Singapore s, in terms of its strategic nature.) For another example, as shown in the case study presented by Cheroo in the conference, the success of Mauritius with export processing zones (EPZs) have made people forget that its government did not adopt a laissez-faire attitude but made strategic choices, including the creation of the EPZs and subsequent integration of EPZs with the rest of the economy (which involved reducing privileges for the EPZs). Most amazing of all these misrepresentations is the US. Historically, it is the home of the infant industry argument, one of the main justifications of industrial policy for developing countries Alexander Hamilton, its first Treasury Secretary invented the theory and economists like Daniel Raymond, Matthew Carey, and Henry Carey further developed the theory. It was almost always the most protected economy in the world for over a century up to the Second World War. Even after WWII, it used a lot industrial policy only that it did not call it industrial policy but R&D policy. Virtually all industries in which Americans still have technological edge were created and/or boosted by federal research funding and other industrial policy interventions (computer, semi-conductor, aircraft, internet, bio-engineering). We can also add that one of the most important US industrial policy has been to convince other countries that it does not have industrial policy and to export free-market ideologies to potential competitor countries.
20 20 8. Concluding Remarks The conference has generated a very rich and sophisticated discussions over many complex and interesting topics related to industrial policy. This summary paper could only present the most important aspects of those discussions and at that in a rather condensed and schematic fashion, but I hope it has provided a systematic overview of the discussions and thus can serve as a good starting point for more detailed discussions in the future.