«© 2013 The International Institute for Sustainable Development © 2013 The International Institute for Sustainable Development ...»
We have made a theoretical case for industrial policy, and we have sketched the major policies undertaken by countries worldwide, pointing to the fact that industrial policy in its various forms has been ubiquitous. However, when it comes to measuring the success of industrial policy, opinions differ sharply. In a recent authoritative literature survey, Harrison and Rodríguez-Clare (2009) review hundreds of empirical studies on this topic from recent decades, and are unable to give a clear answer to the (admittedly very broad) question of whether industrial policy works or not. There are several reasons why this is the case, and it is important to bear some of them in mind.
First, as is the case with all empirical policy assessments, it is impossible to accurately predict how things would have gone in the absence of the policy in question. Economists refer to this problem as the absence of a counterfactual situation. Hence, even if econometric techniques are used that allow to make a qualified guess about such a counterfactual situation, it is often impossible to precisely isolate a causal effect and to attribute a certain outcome to the policy under examination. For example, if there is a high correlation between two variables—say we observe that subsidies are highest in industries that perform worst—we may not actually uncover a causal relationship. Rather than making subsidies responsible for bad economic performance, there might be confounding factors. An obvious alternative explanation would be to turn things around and to ask whether bad economic performance causes large subsidies. This is precisely what Rodrik (2008a) criticized about studies attempting to estimate the impact of specific policies such as tariffs and subsidies on total factor productivity growth in individual sectors. Rodrik pointed out that the amount of stimulus to industries might be endogenous, increasing with sectoral backwardness, as standard industrial policy rationales would suggest (Rodrik, 2008a).
Second, even if the attribution problems were solved, one could always ask the question of whether a particular policy was an optimal one. An economist would look at the opportunity costs of a policy, meaning the entire set of alternative uses of resources spent for the policy in question. For example, would Korea have been better off stimulating domestic demand through cutting taxes, rather than using tax money to subsidize certain industries? Would Korean government officials have used their time and skills in a more valuable way, had they not been required to supervise and monitor preferential lending agreements conditioned on detailed firm-level performance criteria? Of course, data alone does not give many insights into such considerations.
Third, data collection per se is often very complicated, as countries differ widely with respect to policies on information collection and dissemination, a tendency that gets even more complicated when looking across large time spans.
IISD REPORT JUNE 2013 2013 The International Institute for Sustainable Development © Industrial Policy for a Green Economy To researchers’ chagrin, the most interesting countries (developing countries) are often those with the poorest data availability and quality.
Fourth, and probably most importantly, it is often unclear what the benchmark should be when measuring the success of a certain policy. Often there is no clear goal associated with a specific policy measure. Does the evaluating researcher look at economy-wide indicators? In this case, should she look at GDP growth per capita or a measure of structural transformation? Should she look at indices of diversification of exports, its share in total value added, or measures of technological upgrading of domestic industry? Then again, some policies may be motivated by sectoral considerations.
For example, India might be interested in promoting its information and communications technology sector. In this case, would a researcher measure whether the value added in information and communications technology has increased overall, or whether its growth has been higher than in other sectors? How to determine what level or growth rate of value added should be used as a threshold for qualifying as a success?
Economists tend to converge on two criteria that could theoretically provide for a decent cost-benefit evaluation of sectoral industrial policies, namely the Mill-Bastable test. The first part of the test was formulated by Mill (1848) and requires that protected sectors or firms ultimately become viable, even when protection is withdrawn. The second part of the test comes from Bastable (1904) and requires that the cumulative net benefits provided by the protected industry exceed the cumulative costs of protection. Empirical applications of this test are very rare, due to the fact that a strict application requires data on a level of detail that is usually beyond availability. Moreover, a thorough assessment of cumulative costs and benefits would imply a correct monetary valuation of externalities involved, as many industrial policy rationales build on the presence of these. For example, how would one accurately value the dynamic learning effects of a protected industry and its potential spillovers on other industries? Clearly, this task is very daunting, and while the test does provide for a reasonable analytical framework, its literal application remains elusive for the most part.
Owing to widely available trade data, trade policy is by far the most researched component of industrial policy.
Therefore, we will restrict our analysis to trade-related industrial policy instruments.
Trade Protectionism Assuming that there is no retaliation, trade restrictions such as tariffs and quotas can theoretically enhance welfare for the restriction-imposing country if the protected infant industry offers scope for learning and agglomeration spillovers (Melitz, 2005). Clemens and Williamson (2001) and O’Rourke (2000) find a strong positive correlation between import tariffs and economic growth across countries during the late 19th century, hypothesizing that protection allowed countries to accelerate the growth of what were then emerging sectors (industry), characterized by the type of learning effects and Marshallian externalities noted earlier. However, this type of association could not be upheld for more recent times. Reviewing nearly 200 empirical studies on the link between openness to trade and growth, Harrison and Rodríguez-Clare (2009) conclude that studies using trade volumes as a measure of openness (measured as the share of trade in GDP) generally find a positive relationship between changes in openness and growth, making a strong case for overall trade-friendly policies. However, studies that use tariffs as a measure of openness generally find inconclusive or negative effects of average tariffs on growth. Then again, looking at tariff averages might not be very revealing altogether. Lehmann and O’Rourke (2008) look at the pattern of protection during the 19th century and refine earlier findings, noting that agricultural tariffs were negatively correlated with growth, whereas the reverse was true for industrial tariffs.
IISD REPORT JUNE 2013 2013 The International Institute for Sustainable Development © Industrial Policy for a Green Economy More recently, Nunn and Trefler (2010) found strong evidence that the correlation of tariff structure across industries with those industries’ respective required skill levels had long-term growth-enhancing effects. That is, if a country’s tariff lines are designed such that higher tariffs coincide with goods that necessitate greater skill in their production, long-term GDP growth rates will be higher. They find that at least 25 per cent of the correlation between long-term growth and the skill bias in tariffs corresponds to a causal effect. Decomposing their analysis into high-skill and lowskill sectors, their analysis reveals the differential impact of tariffs on each sector: while tariffs in high-skill sectors are associated with sector growth, tariffs in low-skill sectors are associated with negative sectoral growth, somewhat echoing Lehmann and O’Rourke’s (2008) findings with 19th-century data. Overall this analysis suggests that careful sectoral targeting (skill-intensive sectors) can have significant growth-enhancing effects beyond the targeted sector, revealing the presence of inter-industry externalities in high-skill sectors.
Similarly, Estevadeordal and Taylor (2008) find that tariffs on intermediate and capital goods affect growth more negatively than tariffs on consumption goods. This finding is consistent with mainstream economic theory, stressing that firms’ access to cheaper imported inputs is vital for keeping production costs low, increasing their competitiveness on international markets. Blonigen (2013) made a similar finding more recently, analyzing industrial policies from 1975 to 2000 in the steel sectors of 21 countries. This study finds strong evidence that import protection policies and export subsidies have had significantly negative effects on the export performance of downstream sectors. Both are tools that will raise domestic prices for steel, raising production costs for domestic users in other sectors.
The fact that studies looking at average tariffs find mostly weak or even negative associations with overall growth implies that countries have used tariffs in very inefficient, even detrimental ways. Indeed, existing evidence suggests that protection is usually motivated by revenue generation, to protect special interests or for other political reasons, rather than for sound industrial policy purposes (Gawande, Krishna & Olarreaga, 2009). The highest tariff protections, as well as other targeted forms of support such as capital subsidies, have frequently been granted to declining sectors, even in countries that are often considered as industrial policy success stories, including Japan (Beason & Weinstein,
1996) and Korea (Lee, 1996).
The failure of Latin American import substitution industrialization strategy to foster emergence of competitive industries is in large part ascribed to the lack of firms’ access to cheap foreign input goods, due to inefficient protection
patterns. The policies adopted by successfully industrializing Asian countries were designed much more carefully:
“Most [rapidly industrializing Asian countries] began industrialization with a protectionist orientation and have gradually moved toward increasingly free trade. Along the way they often tapped some of the efficiency generating benefits of international competition through mixed trade regimes: they granted exporters dutyfree imports of capital and intermediate goods while continuing to protect consumer goods. Export prices were set in the international market and were often substantially less than current marginal or average costs.
Losses on export production offset profits in the protected market, while competition in the international market ensured that the firm would not suffer from loss of cost discipline.” (World Bank, 1993, p. 295) Hence, while there certainly appears to be scope for using trade protectionism as an instrument for successful IP, governments generally seem to have had great difficulty applying it optimally. Furthermore, governments today are bound by a much tighter set of trade rules as embodied in the World Trade Organization (WTO), significantly restricting the scope of their discretion in setting up trade barriers. Consequently, it is not surprising that much of the current debate on trade-related IP is nowadays focused on trade promotion policies.
IISD REPORT JUNE 2013 2013 The International Institute for Sustainable Development © Industrial Policy for a Green Economy Trade Promotion Besides the demonstrated empirical regularity that openness (measured as the sum of imports and exports over GDP) is associated with economic growth, there is increasing evidence that causality runs from trade to growth (see, for example, Brückner & Lederman, 2012). And with the decline of traditional barriers to market access throughout the world, supply-side constraints are now seen as the main obstacle that developing countries face in taking advantage of new opportunities in international markets (Cadot, Fernandes, Gourdon & Mattoo, 2011). The focus on export promotion is also aided by the advent of a new strand of international trade literature that focuses on firm behavior, whereas prior international trade theory has tended to focus on country-level aggregates. Improved data availability has recently allowed researchers to examine firm behavior more closely, providing a deeper understanding of how international trade works. It is useful to discuss a few of these new findings pertaining to IP at this point.
The number of exporting firms is typically very small compared with a country’s total number of firms. This finding dates back to Bernard, Jensen and Lawrence (1995), who examined U.S. manufacturing firms, but it has subsequently been confirmed for other countries in a number of other studies (see, for example, Eaton, Kortum & Kramarz, 2008, or Mayer & Ottaviano, 2008, for an overview). Moreover, total export value is typically concentrated in a few large, established firms. There is large turnover of firms entering into and exiting from exporting activity. The export status of newly entering firms is usually very short lived (see, for example, Eaton, Eslava, Kugler & Tybout, 2007, and Besedes & Prusa, 2006a). But if they manage to survive the first few years in the export market, new exporters quickly grow and soon make up a substantial share of total export expansion (Eaton et al., 2007, and confirmed for African exporters in Cadot et al., 2011), suggesting that improving trade duration could be an important driver of increasing export shares, making such improvements a promising target for export-promotion policies in developing countries (Cadot et al., 2011). This seems to make sense, as there appear to be structural differences in trade duration among countries and export sectors.