«James A. Robinson Harvard University Department of Government and IQSS May 2009 Paper prepared for the 2009 World Bank ABCDE conference in Seoul June ...»
Industrial Policy and Development: A Political
James A. Robinson
Department of Government and IQSS
Paper prepared for the 2009 World Bank ABCDE conference in Seoul June 22-24.
In this paper I discuss the role of industrial policy in development. I make five main
arguments. First, from a theoretical point of view there are good grounds for believing that
industrial policy can play an important role in promoting development. Second, there certainly are examples where industrial policy has played this role. Third, for every such example there are others where industrial policy has been a failure and may even have impeded development (though the counter-factual is complicated). Fourth, the difference between these second and third cases rests in the politics of policy. Industrial policy has been successful when those with political power who have implemented the policy have either themselves directly wished for industrialization to succeed, or been forced to act in this way by the incentives generated by political institutions.
My fifth point, which is derivative from the first four, is that economists and international institutions have to change the way they think about “industrial policy”. To really promote industrialization in a society we need a positive theory of the political equilibrium of that society which leads to particular policy choices. To give policy advice that would foster industry, one has to understand this political equilibrium and either attempt to change it or work within the environment it generates. This is a very different way of thinking about what industrial policy means.
Unfortunately, as things stand, while we have a good normative theory of industrial policy we have a woefully inadequate positive theory which can help explain why industrial policy was adopted and apparently so successful in Taiwan, for example, and such a disaster in Ghana. It is towards building such a theory that research should focus and without it I argue that advocating industrial policy in the traditional sense as a solution to poor countries’ problems involves a quite large leap of faith.
Before delving into the consequences of industrial policy and my arguments in more detail it is important to have some sort of definition of what industrial policy is. I take it to mean that the government deliberately attempts to promote industry. Nevertheless, there are many ways in which this can be done and many things that can count as ‘industrial policy’.
This ranges from tariff and trade policy (protection), through tax relief, subsidies of various forms, export processing zones, to state ownership of industry. The way that Joseph Stalin promoted industrialization in the Soviet Union in the 1930s was completely different to the way that General Park Chung Hee did so in South Korea in the 1960s. Moreover, any of these policies may have promoted industrialization inadvertently, rather than deliberately.
This is not an issue in the Soviet or South Korean case since the governments overtly committed themselves to industrialization. In other cases, however, whether or not industrialization occurred intentionally is not obvious. One example would be the Calico Acts passed by the British Parliament in 1701 and 1721 with subsequent amendments. These acts raised prohibitive tariffs on cotton goods imported into Britain from India and even banned the wearing of garments made out of Indian fabrics (calicos). In fact it was not until 1774 that the wearing of all cotton cloth was legal. By this time, of course, a rather vibrant British cotton industry had emerged. The traditional interpretation of these reforms rested on the notion that 18th century British governments were laboring under the doctrine of ‘mercantilism’ – an incorrect theory of how the economy worked. According to this view it took a revolution in ideas, induced by Adam Smith and David Ricardo amongst others which led to the final repeal of these acts in 1774. A more positive explanation for the introduction of these reforms was that they were advocated by the English wool and linen industries which were suffering from Indian competition (Mokyr, 1999, pp. 50-51). Neither view suggests that there is any connection between the Calico Acts and industrialization.
However, the Calico Acts, in conjuncture with the Statue of Monopolies of 1623 which made it more or less impossible to establish domestic monopolies in Britain, allowed the initially uncompetitive British cotton industry to develop without facing international competition. Though we do not know what would have happened in the absence of protection, it is obvious that the Calico Acts raised the return to investing in the cotton industry and may have played an important role in stimulating investment in the industry which sparked the British industrial revolution.
Were the Calico Acts an “industrial policy”? If either of the first two views were correct I would argue no, the fact that industrialization was stimulated was an unintended byproduct. Nevertheless, neither of these views provides a compelling understanding of economic policies after the Glorious Revolution in 1688. In fact, as Pincus (2009) convincingly shows, the Glorious Revolution was led by a Whig coalition which quite definitely and explicitly had the object of stimulating ‘manufactures’ in other words industrializing. To this end they started the Bank of England, facilitated the development of the transportation sector via canals and turnpike roads, reorganized the tax system and changed commercial policy. In fact, the Calico Acts was part of a vector of policies which probably constitute one of the world’s most successful and most consequential industrial policies (the ‘mother of all industrial policies’?).
There are quite a few existing approaches to the role of industrial policy in the development literature. Early work during the 1940s and 1950s, by Rosenstein-Rodan, Myrdahl, Nurkse, Hirschman and others associated development with industrialization but argued that due to various types of market failures one could not expect this to happen automatically in poor countries. Hence there was an important role for the government to simulate industrialization with an industrial policy. A particularly important sub-set of these ideas were those due to Singer and Prebisch emphasizing dynamic comparative advantage and the need to close the economy for some period to develop an internationally competitive industrial sector.
These ideas were part of mainstream development economics until they came under sustained attack from scholars with a public choice bent in the late 1970s and 1980s.
Scholars such as Anne Kruger (1993) and Deepak Lal (983) (anticipated to a large extent by Peter Bauer) argued that industrial policy had not worked and indeed could not work because government failures were always worse than market failure. One should forget about industrial policy or for that matter any other policy intervention to solve problems of development and instead focus on creating free markets and a nightwatchman state. This literature was certainly correct in pointing to some very unsuccessful instances of industry policy in developing countries. However, it was rather selective in its focus. Moreover, the theoretical argument that government failures are always worse than market failures seems more ideological than based on either theory or evidence.
Inevitably, therefore, this view did not remain convincing for long, even if it had a large impact on development agencies in the 1980s. The most damning evidence against it came from a series of important interpretations of the ‘East Asian Miracle’ economies by Johnson (1982), Amsden (1989), Wade (1990) and World Bank (1993). These works all put successful industrial policy at the heart of the post-war economics successes of Japan, South Korea and Taiwan, respectively. This research, and much other like it, stood the public choice view on its head more or less arguing that market failures were always worse than government failures and that industry policy was a powerful tool to promote economic growth. This interpretation of the East Asian experience heavily influenced the famous World Bank report in 1992.
Though the evidence in these studies is compelling, the evidence on unsuccessful industrial policy is equally compelling. This suggests that neither extreme view is correct.
Industrial policy can sometimes work, but sometimes not. What distinguishes these cases?
An obvious difference is the type of industrial policy adopted in different cases was very different. In Latin America, for example, it came in the form of Import Substituting Industrialization (ISI) with domestic markets closed to international competition. In South Korea and Taiwan, the model was instead export based with incentives created to induce the development of export industries (though it is also true that the domestic market was protected).
Why were such different strategies chosen to promote industry? As with the explanation for the Calico Acts, the impulse of many scholars is to attribute this variation in policies to variations in ideas. Krueger (1993), for example, argues that Latin American countries were led by erroneous economic theories into adopting the model of import substitutions industrialization. Possibly, East Asian politicians had better (or maybe different and luckier) economic advisors than those who worked in Latin America. Ultimately, variation in the adoption and success of different industrial policies is explained by differences in the ideas and ideologies of different policymakers or their economists. Thus Stalin had an industrial policy of a particular form because of his socialist ideology, while Mauritius had a successful export processing zone because Nobel Laureate Sir James Meade (1961) persuaded the governing Mauritius Labour Party that it was a good economic policy for the country..
An alternative approach to explaining this variation is set out in Rodrik (2007).
Rodrik’s basic argument is that industrial policy is potentially very powerful but one size does not fit all. To successfully promote development industrial policy has to be tailored to the specific context or institutions of a country, or to use the terminology introduced by Hausmann, Rodrik and Velasco (2007) has to be sensitive to the “binding constraints”.
According to this view different countries could adopt identical policies with very different results since they had different sets of market failures. Why would some countries adopt policies suited to their market failures and not others? The main reason is uncertainly about what the binding constraints are. Either South Korea was very lucky in being able to understand this or it had (again) better economists who managed to understand this.
Ghanaian industrial policy failed because some academic scribbler (actually Sir Arthur Lewis, see Lewis, 1953), persuaded the government of Kwame Nkrumah to adopt an industrial policy which was not the right one given Ghana’s circumstances.
In this essay I lay out a completely different way of thinking about the evidence on industrial policy. I agree that there are many market failures in the world, that there can be important externalities from having a thriving industrial sector and that potentially industrial policy can be a powerful tool to promote rapid economic growth and development. I also believe that this was the case in South Korea, Taiwan and many of the other cases studied by the revisionists in the late 1980s. I think the balance of evidence suggests that these scholars were right to attribute a powerful causal role to industrial policy (though admittedly we have no definitive econometric evidence on this). However, I also believe that industrial policy can totally fail, as it did in Ghana in the 1960s and all over Latin America from the 1940s onwards. But the difference between these cases is not that the Japanese or South Koreans got lucky, were clever or had better economists advising them, it was because the political equilibrium of these societies differed.
I shall argue that to begin to think properly about industrial policy we need to start with a positive theory of such policy. It is remarkable the extent to which the economics literature on industrial policy has avoided proposing a positive theory of industrial policy as an equilibrium outcome. Take the re-assessment of Soviet industrialization developed by Allen (2003). Allen’s whole point is to argue that Stalin’s industrialization policy was actually optimal in a poor country will bad initial institutions. There is no argument, however, which would lead us to expect the Stalinist regime would adopt an efficient policy (indeed, a vast mass of evidence suggests that efficiency was not high on the criteria which Stalin’s regime maximized – see Gregory and Harrison, 2005). The normative approach to explaining policy assumes that politicians choose policy in a socially optimal way, something hard to believe about the Soviet Union in the 1920s and 1930s. Another striking example of this approach in action comes from the experience of policy reform in transition economies in the 1990s.
While Russian reform was done badly, Chinese dual track reform is characterized as clever (see Lau, Qian and Roland, 2000, for the argument that Chinese reforms were cleverer than Russia reforms). Of course the Great Leap Forward during the 1950s is not viewed as clever, but a big mistake. Though Shleifer and Treisman (2000) did try to analyze Russian policy reforms as an equilibrium outcome, they still argued that they were (constrained) efficient. In fact the evidence suggests that the reason that the Chinese did things differently from the Soviets was not because they were cleverer (see Shirk, 1993, Naughton, 2007). Chinese policy reform was an equilibrium outcome, as was Soviet reform. The fact that the transition policy differed was because though both regimes may have been avowedly communist and (less avowedly) authoritarian, in fact the political equilibria in these societies were quite different. For example, the strength of the central state and the party was very different. The problem with industrial policy in Ghana in the 1960s was not that it was not clever, or appropriate, or needed, but rather that it was adopted in political circumstances where it had no chance of succeeding in actually industrializing the country.