57756 WO R K IN G PAP E R N O .57 From Growth Theory to Policy Design Philippe Aghion Steven Durlauf WORKING PAPER NO. 57 From Growth Theory to Policy Design Philippe Aghion Steven Durlauf © 2009 The International Bank for Reconstruction and Development / The World Bank On behalf of the Commission on Growth and Development 1818 H Street NW Washington, DC 20433 Telephone: 2024731000 Internet: www.worldbank.org www.growthcommission.org Email: info@worldbank.org contactinfo@growthcommission.org All rights reserved 1 2 3 4 5 11 10 09 08 This working paper is a product of the Commission on Growth and Development, which is sponsored by the following organizations: Australian Agency for International Development (AusAID) Dutch Ministry of Foreign Affairs Swedish International Development Cooperation Agency (SIDA) U.K. Department of International Development (DFID) The William and Flora Hewlett Foundation The World Bank Group The findings, interpretations, and conclusions expressed herein do not necessarily reflect the views of the sponsoring organizations or the governments they represent. The sponsoring organizations do not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the sponsoring organizations concerning the legal status of any territory or the endorsement or acceptance of such boundaries. All queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 2025222422; email: pubrights@worldbank.org. Cover design: Naylor Design About the Series The Commission on Growth and Development led by Nobel Laureate Mike Spence was established in April 2006 as a response to two insights. First, poverty cannot be reduced in isolation from economic growth--an observation that has been overlooked in the thinking and strategies of many practitioners. Second, there is growing awareness that knowledge about economic growth is much less definitive than commonly thought. Consequently, the Commission's mandate is to "take stock of the state of theoretical and empirical knowledge on economic growth with a view to drawing implications for policy for the current and next generation of policy makers." To help explore the state of knowledge, the Commission invited leading academics and policy makers from developing and industrialized countries to explore and discuss economic issues it thought relevant for growth and development, including controversial ideas. Thematic papers assessed knowledge and highlighted ongoing debates in areas such as monetary and fiscal policies, climate change, and equity and growth. Additionally, 25 country case studies were commissioned to explore the dynamics of growth and change in the context of specific countries. Working papers in this series were presented and reviewed at Commission workshops, which were held in 2007­08 in Washington, D.C., New York City, and New Haven, Connecticut. Each paper benefited from comments by workshop participants, including academics, policy makers, development practitioners, representatives of bilateral and multilateral institutions, and Commission members. The working papers, and all thematic papers and case studies written as contributions to the work of the Commission, were made possible by support from the Australian Agency for International Development (AusAID), the Dutch Ministry of Foreign Affairs, the Swedish International Development Cooperation Agency (SIDA), the U.K. Department of International Development (DFID), the William and Flora Hewlett Foundation, and the World Bank Group. The working paper series was produced under the general guidance of Mike Spence and Danny Leipziger, Chair and Vice Chair of the Commission, and the Commission's Secretariat, which is based in the Poverty Reduction and Economic Management Network of the World Bank. Papers in this series represent the independent view of the authors. From Growth Theory to Policy Design iii Abstract This paper focuses on how growth theory can guide growth policy design. It first argues that policy matters for growth, in particular when policy variables are interacted with countryspecific variables (financial development, institutional environment, technological development, and so forth). Second, it argues that the Schumpeterian paradigm does a better job at delivering policy prescriptions that vary with country characteristics. Third, it discusses the advantages and drawbacks of growth regression analysis. Finally, it briefly describes and then questions the recently proposed "growth diagnostic" approach. iv Philippe Aghion and Steven Durlauf Contents About the Series ............................................................................................................. iii Abstract ............................................................................................................................iv 1. Introduction ..................................................................................................................7 2. What Can We Learn from New Growth Theory? ...................................................8 3. From Growth Theory to Growth Policy..................................................................18 4. Conclusion ..................................................................................................................32 References ....................................................................................................................... 33 From Growth Theory to Policy Design v From Growth Theory to Policy Design Philippe Aghion Steven Durlauf 1 1. Introduction One indicator of a good theory is the extent to which the predictions delivered by the theory can be validated or falsified by the available evidence. New growth theories can potentially qualify on that ground, as they deliver predictions on the relationship between growth and a series of micro and macroeconomic variables that can be confronted by crosscountry, crossindustry, or crossfirm (panel) data series. In this paper, we put growth theory to another test, namely that of delivering relevant growth policies across various types of countries. For example, imagine a policy adviser, who faces a disappointing growth performance (in particular growth has been persistently slow in the EU, with less than 2 percent a year against 3 percent in the United States between 1995 and 2006). Or imagine the policy adviser having to explain why Latin American countries, which had comparable or higher per capita GDP levels than their Asian counterparts in the early 1960s, have grown at a much slower rate ever since. Thus, while Latin American countries have grown at less than 3 percent on average between 1960 and 2000, Singapore has grown at 7 percent, the Republic of Korea at 6.5 percent, Hong Kong, China at 6 percent, Thailand at 5 percent, and China now grows at annual rates beyond 10 percent. Disappointingly, classical growth models, in which growth is primarily driven by (physical and/or human) capital accumulation, do not appear to have much to tell us on this question. The neoclassical growth model would try to explain the lower growth rate in Europe by comparing its savings rate or the capitallabor ratio to those in the United States. But savings rates tend to be higher in Europe and the same is true for capitallabor ratios, and this has been the case for more than one decade and even during times where per capita GDP in Europe was growing either faster than or at the same rate as in the United States. Similarly, capital output ratios are about the same in the big Latin American countries as in the OECD countries, namely at 0.31. Philippe Aghion is Robert C. Waggoner Professor of Economics in the Economics Department at 1 Harvard University. Steven Durlauf is Kenneth J. Arrow Professor of Economics at University of WisconsinMadison. From Growth Theory to Policy Design 7 Also, at first sight the standard textbook innovationbased model(s) cannot account for the recent growth gap between Europe and the United States, given that the property rights and innovation subsidies stressed by these models are reasonably well established in Europe, and that Europe invests almost as large a fraction of GDP on research and development (R&D) as the United States (2 percent versus 2.5 percent). Moreover, these models do not seem to explain why European labor productivity growth was much higher than U.S. growth during the 1960s and 1970s (3.5 percent versus 1.4 percent on average during the 1970s), given that R&D investments were higher in the United States than Europe throughout this period. Nor can R&D investments fully explain the divergence between Asia and Latin America: the former grew at a very high rate long before they invested in R&D. In this paper we argue that modern growth theory can in fact provide relevant tools to analyze the determinants of growth in different types of countries, and that it generates a systematic approach to growth policy design. In Section 2, we offer a summarized presentation of new growth theory. In Section 3 we discuss how growth theory can be used to design policy. In particular we contrast the growth diagnostic approach by Hausmann, Rodrik, and Velasco (2005) with an alternative approach based on growth regressions suggested by the theory. Section 4 builds on this approach to draw some guidelines of comprehensive growth policy packages for Europe or Latin America. Section 5 goes in more details into the methodological issues raised by growth regressions. Section 6 concludes. 2. What Can We Learn from New Growth Theory? In this section we present a very brief overview of new growth theory paradigms, and of the policy lessons each of those can deliver. A natural starting point for growth analysis is to specify an aggregate production function which describes how domestic output flow is generated from a given stock of production factors. Thus we can write: Y AF K , L , where K is the capital stock, L is the stock of labor, and A is a productivity factor which reflects the existing stock of knowledge and the resulting efficiency of capital and labor in producing final output. Growth in output thus results: (i) from the accumulation of production factors K or L (one could also include human capital H on top of K and L); and/or (ii) from increases in the productivity factor A, that is, from productivity growth. If the production function F is homogeneous of degree one, one can rewrite the above equation as: 8 Philippe Aghion and Steven Durlauf y AF k ,1 Af k , where Y y L is the per capita output flow and K k L is the stock of capital per capita, and f displays decreasing returns to capital accumulation; that is, the more capital has already been accumulated, the smaller the increase in per capita output generated by accumulating one more unit of capital. The seminal contribution by Solow (1956) was the first to show that absent technical progress--that is, in the absence of growth in A--per capita output y might grow in the short run but not in the long run: eventually, accumulating more capital will entail more capital depreciation than what it can bring in terms of added output. The history of the past 50 years offers a perfect illustration of that view: for example, the Soviet Union grew fast during the postwar period where growth was primarily driven by factor accumulation, itself financed by (forced) domestic savings. But as of the late 1960s, growth tapered off. Similarly, factor accumulation facilitated by the Marshall Plan explains part of the growth miracle of the "Trente Glorieuses" in Western Europe. However, Western Europe ceased to grow fast as of the mid1970s. The contribution of new growth theory is to explain productivity growth, that is, the growth in A resulting from innovations. The first version of endogenous growth theory was the socalled AK theory. In AK models, sustained productivity growth results from contemporaneous or intertemporal externalities between individuals who accumulate physical or human capital. Thus in effect these models lump together the physical and human capital whose accumulation is studied by neoclassical theory with the intellectual capital that is accumulated when technological progress is made. Indeed Lucas's (1988) influential contribution followed Uzawa (1965) in explicitly assuming that human capital and technological knowledge were one and the same. When this aggregate of different kinds of capital is accumulated there is no reason to think that diminishing returns will drag its marginal product down to zero, because part of that accumulation is the very technological progress needed to counteract diminishing returns. According to the AK paradigm, the way to sustain high growth rates is to save a large fraction of GDP, some of which will find its way into financing a higher rate of technological progress and will thus result in faster growth. Thrift and the capital accumulation are the keys, not novelty and innovation. AK From Growth Theory to Policy Design 9 theory thus formalizes the ideas behind the World Bank consensus policies after WWII, according to which the problem of economic development was underaccumulation of capital and the cure was to subsidize and give aid to large investment projects. The second wave of endogenous growth theory consists of socalled "innovationbased" growth models, which themselves belong to two parallel branches. One branch is the model of Romer (1990), according to which aggregate productivity is a function of the degree of product variety. Innovation causes productivity growth in the productvariety paradigm by creating new, but not necessarily improved, varieties of products. This paradigm grew out of the new theory of international trade, and emphasized the technology spillovers according to which the productivity of resources devoted to developing new product varieties was greater the greater the variety of products that have already been developed. The other branch of innovationbased theory, developed in Aghion and Howitt (1992) and subsequently elaborated in Aghion and Howitt (1998), grew out of modern industrial organization theory, and is commonly referred to as "Schumpeterian" growth theory, because it focuses on qualityimproving innovations that render old products obsolete, and hence involves the force that Schumpeter called "creative destruction." This second wave of endogenous growth models puts the investment side at the heart of the growth process. In particular, economic policies and institutions influence growth by influencing entrepreneurs' incentives and/or ability to make innovative investments. 2.1. The AK Paradigm The AK paradigm is neoclassical growth theory without diminishing returns. The theory starts with an aggregate production function, which, in the case where knowledge externalities among firms exactly compensate for the diminishing returns to capital accumulation by individual firms, ends up being linear and homogeneous in the stock of capital: Yt AKt (1) with A as a constant. Output is in turn divided between consumption and investment: Yt Ct It (2) and net investment is investment minus depreciation at the fixed rate : Kt I t Kt . (3) Thus the growth rate of output is the same as the growth rate of capital, which in turn depends on thrift. 10 Philippe Aghion and Steven Durlauf In early versions of the theory,2 thrift is represented by a fixed saving ratio: s 1 Ct Yt , in which case the economy's growth rate is: g sA . Alternatively, Romer's (1986) version represented thrift as intertemporal utility maximization à la Ramsey, in which a representative household maximizes: Ct1 e t dt 0 1 subject to the production function (1), the law of motion (3), and an individual version of the resourcebalance constraint (2): Yt C t 1 I t Tt , where is an investment subsidy (or tax, if negative) financed by the lumpsum tax T. The Euler equation implied by this problem determines immediately the economy's growth rate as: A g 1 C g. C An immediate implication of this model is that a higher saving rate s, or higher investmentsubsidy rate , encourages capital accumulation and therefore growth. There are several problems with AK theory. First, it constitutes a "one size fits all" approach to the growth process. It applies equally to countries that are on the leading edge of the world technology frontier and to countries that are far behind. Like the neoclassical theory of Solow and Swan, it postulates a growth process that is independent of developments in the rest of the world, except insofar as international trade changes the conditions for capital accumulation.3 In particular, the theory is not helpful in understanding how the institutions and policies that were so successful in promoting growth immediately following WWII, when Europe was far below the frontier, turned out to produce relatively poor performance since the 1990s. Second, in AK models growth is primarily driven by the savings side of the economy: policies affect growth by affecting aggregate savings, which are equal to aggregate investment. This view becomes problematic, for example when discussing the relationship between macroeconomic volatility and growth. Third, AK models cannot explain convergence. Indeed, a country's or region's growth rate is independent from its 2 See Frankel (1962). 3 See for example Acemoglu and Ventura (2002). From Growth Theory to Policy Design 11 current capital stock, which in that model measures its stage of development. Thus, countries with low levels of capital stock do not grow faster than more advanced countries. Compared to the Solow model, AK models can "explain" longrun growth but at the expense of losing convergence. 2.2. The Schumpeterian Paradigm Schumpeterian theory begins with a production function specified at the industry level: Yit Ait K it , 0 1 1 (4) where Ait is a productivity parameter attached to the most recent technology used in industry i at time t. In this equation, Kit represents the flow of a unique intermediate product used in this sector, each unit of which is produced onefor one by capital. Aggregate output is just the sum of the industryspecific outputs Yit . Each intermediate product is produced and sold exclusively by the most recent innovator. A successful innovator in sector i improves the technology parameter Ait and is thus able to displace the previous innovator as the incumbent intermediate monopolist in that sector, until displaced by the next innovator. Thus the first key implication that distinguishes the Schumpeterian paradigm from the AK and productvariety models is that faster growth generally implies a higher rate of firm turnover, because this process of creative destruction generates entry of new innovators and exit of former innovators. Although the theory focuses on individual industries and explicitly analyzes the microeconomics of industrial competition, the assumption that all industries are ex ante identical gives it a simple aggregate structure. In particular, it is easily shown that aggregate output depends on the aggregate capital stock Kt according to the CobbDouglas aggregate perworker production function: Yt At1 Kt (5) where the laboraugmenting productivity factor At is just the unweighted sum of each sectorspecific Ait . As in neoclassical theory, the economy's longrun growth rate is given by the growth rate of At , which here depends endogenously on the economywide rate of innovation. There are two main inputs to innovation; namely the private expenditures made by the prospective innovator, and the stock of innovations that have already been made by past innovators. The latter input constitutes the publicly available stock of knowledge to which current innovators are hoping to add. The theory is quite flexible in modeling the contribution of past innovations. It encompasses the case of an innovation that leapfrogs the best technology available before the innovation, resulting in a new technology parameter Ait in 12 Philippe Aghion and Steven Durlauf the innovating sector i, which is some multiple of its preexisting value. And it also encompasses the case of an innovation that catches up to a global technology frontier At , which we typically take to represent the stock of global technological knowledge available to innovators in all sectors of all countries. In the former case the country is making a leadingedge innovation that builds on and improves the leadingedge technology in its industry. In the latter case the innovation is just implementing technologies that have been developed elsewhere.4 For example, consider a country in which in any sector leadingedge innovations take place at the frequency n and implementation innovations (or imitations) take place at the frequency m . Then the change in the economy's aggregate productivity parameter At will be: At 1 At n 1 At m At At and hence the growth rate will be: At 1 At gt At n 1 m at1 1 (6) where: at At At is an inverse measure of "distance to the frontier." The second term reflects the existence of knowledge spillovers across countries: the further below the frontier a country is, that is the smaller at , the faster the country can grow us by imitating or adapting leadingedge technologies. Thus the theory incorporates Gerschenkron's "advantage of backwardness"5 in the sense that the further the country is behind the global technology frontier (that is, the smaller is at ) the faster it will grow, given the frequency of implementation innovations. As in Gerschenkron's analysis, the advantage arises from the fact that imitations allow the country to make larger quality improvements the further it has fallen behind the frontier. Thus the Schumpeterian paradigm can explain convergence, but one primarily based on knowledge spillovers rather than on diminishing returns to capital accumulation as in the neoclassical model. The above growth equation also illustrates the fact Schumpeterian theory provides a framework in which the growth effects of various policies are highly 4 This flexibility of the Schumpeterian framework does not lead to a theory in which anything can happen. For example, in Aghion et al. (2005a), the effect of competition on growth in the Schumpeterian paradigm is either monotonic or invertedU shaped, but cannot be of any other form. Similarly, the effect of entry at the frontier on productivity growth is always more (and not less) positive in sectors initially closer to the frontier. 5 See Gerschenkron (1962). From Growth Theory to Policy Design 13 contextdependent. In particular, the Schumpeterian apparatus is well suited to analyze how a country's growth performance will vary with its proximity to the technological frontier at , and what kinds of policy changes are needed to sustain convergence as the country approaches the frontier. We could take the critical innovation frequencies m and n that determine a country's growth path as given, just as neoclassical theory often takes the critical saving rate s as given. However, Schumpeterian theory goes deeper by deriving these innovation frequencies endogenously from the profit maximization problem facing a prospective innovator, just as the Ramsey model endogenizes s by deriving it from household utility maximization. This maximization problem and its solution will typically depend upon institutional or policy characteristics of the economy such as property right protection, product market competition, education policy and the composition of education spending, financial development and the financial system, market size, the organization of labor markets, trade policy, and so forth, in a way that may or may not vary with the country's distance to the technological frontier a. Thus, quite independently from a country's level of development the theory clearly predicts that innovation and therefore productivity growth and convergence should be fostered by: (i) better protection of (intellectual) property rights, as this will improve the extent to which successful innovators can appropriate the rents from their innovation; (ii) better financial development, as tight credit constraints will limit individuals' ability to finance a new innovative idea;6 (iii) a higher stock of educated labor, as this will improve individuals' ability to imitate more advanced technologies or to innovate at the frontier building on giants' shoulders;7 and (iv) macroeconomic stability: by ensuring low (riskadjusted) equilibrium interest rates, it will encourage individuals to engage in longterm growthenhancing investments. However, the effects of product market competition, trade policy, the composition of education spending, or the organization of labor markets or that of financial systems turn out to depend heavily upon a country's or region's or sector's level of development as measured by at . Typically, the closer a country or sector is to the technological frontier, the more growthenhancing are the effects of: (i) product market competition8 and entry:9 here the idea is that competition and entry encourages innovation by firms that are already sufficiently advanced that they can escape competition by innovating; (ii) higher education, which is more important for frontier innovation than for imitation, and therefore becomes increasingly important for growth as a country 6 See DurlaufJohnson (1995) and AghionHowittMayer (2005) for analyses of club convergence. In the latter, club convergence results from the interplay between credit constraints and knowledge spillovers. 7 See NelsonPhelps (1966), BenhabibSpiegel (1994), and KruegerLindahl (2001). 8 See AghionBloomBlundellGriffithHowitt (2005). 9 See AghionBlundellGriffithHowittPrantl (2005). 14 Philippe Aghion and Steven Durlauf approaches the world technology frontier;10 (iii) stock market finance, which is particularly adapted for innovative projects that are highly uncertain and do not necessarily carry collateral; (iv) democracy, which favors entry that in turn is growthenhancing in advanced sectors as discussed above;11 and (v) more decentralized firms: delegating initiative to managers downstream is important when a firm introduces new frontier technologies on which information upstream is limited.12 Take as given that the institutions or policies that favor implementation innovation (that is, that lead to firms emphasizing m at the expense of n ) are not the same as those that favor leadingedge innovation (that is, that encourage firms to focus on n ). Then, to sustain a high growth rate as it develops (that is, as at increases), a country will have to shift from implementationenhancing institutions to innovationenhancing institutions. As formally shown in AcemogluAghionZilibotti (2002), failure to operate such a shift can prevent a country from catching up with the frontier level of per capita GDP. AAZ refer to this as a nonconvergence trap phenomenon. In Sapir et al. (2003) we argued that this failure largely explains why Europe stopped catching up with U.S. per capita GDP since the mid 1970s. How about growth rates? Suppose that the global frontier grows at the exogenous rate g. 13 Then equation (6) implies that in the long run a country that engages in implementation investments (with m >0) will ultimately converge to the same growth rate as the world technology frontier. That is, the relative gap at that separates this economy from the technology frontier will converge asymptotically to the steadystate value: m a ^ (7) g m n 1 which is an increasing function of the domestic innovation rates and a decreasing function of the global productivity growth rate. The economic force underlying this convergence in growth rates is again Gerschenkron's advantage of backwardness, according to which a country that is growing slower than the frontier rate g , and which is therefore falling further behind the frontier, will therefore experience an increase in its growth rate. Now, can we explain why, since the mid 1990s, the EU is growing at a lower rate than the United States? A plausible story, which comes out naturally from the above discussion, is that the European economy caught up technologically to 10 See VandenbusscheAghionMeghir (2004) and AghionBoustanHoxbyVandenbussche (2005). 11 See AghionAlesinaTrebbi (2007). 12 See AcemogluAghionLelargeVan ReenenZilibotti (2007). 13 Howitt (2000) shows how the global growth rate can be endogenized as a function of innovation rates in sectors and all countries. From Growth Theory to Policy Design 15 the United States following WWII but then its growth began to slow down before the gap had been closed, because its policies and institutions were not designed to optimize growth when close to the frontier. That by itself would have resulted in a growth rate that fell down to that of the United States but no further. But then what happened was that the IT revolution resulted in a revival of g in the late 1980s and early 1990s. Since Europe was as not well placed as the United States to benefit from this technological revolution, the result was a reversal of Europe's approach to the frontier, which accords with the Schumpeterian steady state condition (7). Furthermore, the fact that Europe is not adjusting its institutions in order to produce the growthmaximizing innovation policy acts as a delaying force on growth convergence towards the United States.14 Neoclassical theory might explain the convergence phase of Europe towards the United States, but not the reversal that has occurred since. 2.3. The Product-Variety Paradigm The other branch of innovationbased growth theory is the productvariety model of Romer15 (1990), which starts from an EthierDixitStiglitz production function16 of the form: Nt Yt K di 0 it in which there are Nt different varieties of intermediate product. By symmetry, the aggregate capital stock Kt will be divided up evenly among the Nt existing varieties equally, which means we can reexpress this production function as: Yt N t1 Kt . (8) 14 Endogenizing m can also generate divergence in growth rates. For example, human capital constraints as in Howitt and MayerFoulkes (2005), or credit constraints as in Aghion, Howitt, and MayerFoulkes (2005), make the equilibrium value of m increasing in a, which turns the growth equation (6) into a nonlinear equation. That m be increasing in a, in turn follows from the assumption that the cost of innovating is proportional to the frontier technology level that is put in place by the innovation (Ha and Howitt (2005) provide empirical support for this proportionality assumption) whereas the firm's investment is constrained to be proportional to current local productivity. Then, countries very far from the frontier and/or with very low degrees of financial development or of human capital will tend to grow in the long run at a rate which is strictly lower than the frontier growth rate g . However, our empirical analysis in this paper shows that this source of divergence does not apply to EU countries. 15 The semiendogenous model of Jones (1995), in which longrun economic growth depends uniquely on the rate of population growth, might be thought of as a fourth paradigm, but it has nothing useful to say about growth policy since it predicts that longrun growth is independent of any policy that does not affect population growth. It does imply that innovation affects growth during the transition to the long run, but in that context it behaves just like the productvariety model that we discuss in this section. 16 See Dixit and Stiglitz (1977). 16 Philippe Aghion and Steven Durlauf According to (8), the degree of product variety Nt is the economy's labor augmenting productivity parameter, and its growth rate is the economy's long run growth rate of per capita output. Product variety raises the economy's production potential in this theory because it allows a given capital stock to be spread over a larger number of uses, each of which exhibits diminishing returns. The driving force of longrun growth in the productvariety paradigm is innovation, as in the Schumpeterian paradigm. In this case however innovations do not generate better intermediate products, just more of them. Also as in the Schumpeterian model, the equilibrium R&D investment and innovation rate result from a research arbitrage equation that equates the expected marginal payoff from engaging in R&D to the marginal opportunity cost of R&D. But the fact that there is just one kind of innovation, which always results in the same kind of new product, means that the productvariety model is limited in its ability to generate contextdependent growth, and is therefore of limited use for policy makers in Europe. In particular, the theory makes it very difficult to talk about technology frontier, distance to the frontier, and convergence. Thus, it has little to say about how the kinds of policies appropriate for promoting growth in countries near the world's technology frontier may differ from those appropriate in technological laggards, and thus to explain why Asia is growing fast with policies that depart from the Washington consensus, or why Europe has grown faster than the United States during the first three decades after WWII but not thereafter.17 In addition, nothing in this model implies an important role for exit and turnover of firms and workers; indeed increased exit in this model can do nothing but reduce the economy's GDP, by reducing the variety variable Nt that uniquely determines aggregate productivity according to the production function (8). As we argue in Aghion and Howitt (2006), these latter implications of the productvariety model are inconsistent with an increasing number of recent studies demonstrating that labor and product market mobility are key elements of a growthenhancing policy, particularly in sectors or countries near the technological frontier. 17 For example, Helpman (1993) uses the productvariety approach to construct a twocountry model in which innovation takes place only in the North and imitation only in the South. But although policies would then have different growth effects depending on whether implemented in the North or South, there is nothing in this analysis that links a given country's position as imitator or innovator to any productivity gap; instead it is just assumed that some countries cannot imitate and some cannot innovate. Thus there is nothing in the approach that would imply a change in appropriate institutions or policies as the country closed the gap, let alone allow for one country to leapfrog another. From Growth Theory to Policy Design 17 3. From Growth Theory to Growth Policy In a recent and influential paper entitled "Growth Diagnostics," Hausmann, Rodrik, and Velasco (2005), henceforth HRV, have proposed an attractively simple methodology to design growthenhancing policy. In this section we will first summarize the methodology, then point at some of its potential limitations, and then propose an alternative approach based that uses growth regressions themselves suggested by the theory, particularly the Schumpeterian paradigm outlined above. The HRV Approach to Growth Diagnostics The approach HRV start from the relevant observation that growthenhancing policies should vary from one country or region to another. For example, while growth in the United States and other industrialized countries over the past 10 years appears to have benefited from market deregulations and privatizations, in Asian countries including China high growth rates have been promoted under limited competition or limited privatizations. The next question then is: can one use existing new growth theory to provide a flexible guide to growth policy making, one that fully takes crosscountry data into account? HRV provide a positive and attractively simple answer to this question. They start from the Euler equation in graduate textbooks, which expresses the growth rate g of an economy as: g r 1 , where is the intertemporal elasticity of substitution, r is the rate of return on capital, is the tax rate, and is the individual rate of time preference taken to be equal to the world rate of interest. The rate of return on capital r is itself endogenous and depends in particular upon the country's total factor productivity (TFP) level a, the degree of externalities in production , the tax rate on capital , and the availability of complementary factors x: r r a , , x . These two equations narrow down the set of potential suspects to be considered when designing growth policy. The next question is: how can one identify the most binding constraint to growth, that is, the actual suspects? Here HRV propose a simple solution that avoids any use of econometrics, namely to use price comparisons to infer the importance of each potential constraint to growth. To illustrate their methodology, HRV consider a few Latin American examples, including Brazil and El Salvador. 18 Philippe Aghion and Steven Durlauf In Brazil, returns to capital are high (with a net interest margin equal to 11.5 percent in 2001). This leads HRV to point at the low level of local savings (with very negative public savings) and the high tax rates as the main constraints on growth (the importance of the former is further supported by the positive and significant correlation between the interest rate and the current account deficit over time). The rate of return on education is also high in Brazil, which suggests that the rate of return on capital, and thereby growth, could be further increased by investing more on education. However, the argument goes, the already high rate of return on capital suggests that investing in education may not be a priority in Brazil. In El Salvador, interest rates are low (a net interest margin equal to 3.7 percent in 2001), but so is the tax rate on capital. Is the lack of education responsible for the rate of return on capital? The HRV answer is no, given that the rate of return on education in El Salvador is low. Nor is there a lack of contractual enforcement that would reduce profitability. Lack of savings cannot be the binding constraint either; otherwise the interest margin would be high. Having failed to identify true obstacles to growth in El Salvador, HRV mention the "absence of profitable investment opportunities" as yet another potential suspect to consider. Now suppose we used the same growth diagnostic approach to deal with the slow EU growth problem. The return to education is lower in the EU than in the United States, which HRV would not interpret as an indication that education is the most binding constraint to growth. Rather, they would presumably point at the high European tax rates as the main suspect, and thereby follow Prescott (2004) in advocating lower tax rates as the primary cure to the growth problem in the EU. Discussion The simple and ingenious approach proposed by HRV raises a number of questions, first on individual country cases and then at a more fundamental level on the methodology itself. Looking first at examples, do we really believe that education is not a binding constraint for growth in the EU and that the main problem there is taxation? A recent study (see Aghion, Cette, Cohen, and Pisani (2007)) based on crosscountry panel regressions of growth on education, regulation indexes, and their interactions with countries' levels of technological development, suggests that the following: (i) France could gain more than half of a percentage point of growth by investing as much as Sweden in higher education; (ii) France could increase growth by an additional half of a percentage point by reducing regulations on goods and labor markets down to Scandinavian levels. Taxing corporate profits may of course discourage entrepreneurs from investing in the country, yet we observe relatively high growth rates in high incometax Scandinavian countries such as Denmark, Sweden, or Finland. From Growth Theory to Policy Design 19 At a more fundamental level, we see at least two problems with the HRV methodology. First, equilibrium prices do not necessarily reflect a constraint on growth. Consider first interest rates. A low interest rate does not mean that the local credit market is not constrained. In fact, low interest rates may reflect a high degree of credit rationing, as shown by Aghion and Bolton (1997). Indeed, the more restricted the access to credit (that is, the more individuals are barred from undertaking their own projects), the more supply of loanable funds there will be in the economy, as all creditrationed individuals will end up lending to a few entrepreneurs. But this in turn should result in a lower domestic equilibrium interest rate. Next, consider the rates of return on labor, which are measured by the socalled Mincerian wages--that is, by the foregone wage income of one more year in education at different levels of education. Mincerian wages of course provide some useful indication on the marginal value of private investments in education in different fields and at different levels of education. However, a big shortcoming of the Mincerian approach is that the Mincerian wage does not account for externalities, and in particular for the intertemporal knowledge externalities that lie behind the positive relationship between education and growth. That intertemporal externalities matter is evidenced by the large effects of education on growth (see Aghion, Boustan, Hoxby and Vandenbussche 2005), in comparison to the recent study by Acemoglu and Angrist (2004), which pointed at the absence of significant contemporaneous externalities in education. More generally, current prices reflect a current state of the economy. They do not inform directly about the growth dynamics that would result for various types of policies. A second problem with the HRV approach is that it cannot lead to growth prescriptions that would affect the demand side and the supply side of markets simultaneously. Thus, for example HRV would never recommend that a country invest in education (thereby increasing the supply of research labor) and at the same time invest in structural reforms that increase the profitability of innovations (thereby fostering the demand for R&D labor by firms).18 3.2. Use New Growth Theory and Growth Regressions Instead A natural alternative to the above methodology is to use the theory outlined in Section 2 to construct growth regressions, which are meant to inform us directly about the impact of different institutions or policies on growth. Some main issues in using growth regressions In this section, we discuss some issues that arise when using growth regressions to evaluate policies. We describe the baseline crosscountry regression and 18 Incidentally, HRV would never recommend more active competition policies whose effect in the simple growth paradigm they consider is simply to reduce the rate of return on capital. 20 Philippe Aghion and Steven Durlauf identify some interpretation issues that naturally arise when the model is considered against the background of economic theory. The basic cross country growth regression, as developed by Barro (1996), may be written as: gi Xi Zi i , (9) where gi is real per capita growth across some fixed time interval, Xi is a set of regressors suggested by the Solow growth model (population growth, technological change, physical and human capital savings rates transformed in ways implied by the model), Zi is a set of additional control variables suggested by new growth theories, and i is an error. The distinction between Xi and Zi is important in econometric practice because while variables are essentially constant across empirical studies, there is no consensus on which Zi variables should be included. The use of crosscountry growth regressions has been criticized at several levels. Here, we note four criticisms that are germane to the empirical work we wish to discuss and which flow naturally from the use of these regressions to make substantive claims. First, the regressors in (9) may be endogenous, so that the interpretation of OLS regressions using this structure begs the question of whether the regression coefficients reflect causality or correlation. From the perspective of economic theory, this worry is an obvious one. Variables such as the savings rate or population growth rate are (aggregations of individual) choices and so a theory that treats them as exogenous is not only incomplete but susceptible to classical simultaneous equations bias. The natural solution to this problem is the use of instrumental variables. The determination of appropriate instrumental variable choices is difficult because of the fact that many candidate instruments are themselves possible growth determinants. For example, geographic variables are used as instruments for international trade in some studies while the variables are treated as growth determinants in others. Another example is ethnic diversity. This issue should not be interpreted as a general rejection of instrumental variables methods in growth regressions. Rather it suggests that there are difficult judgments involved in their use. Brock and Durlauf (2001a,b) and Durlauf, Johnson, and Temple (2005) provide an extensive discussion. A second problem concerns interactions. Linear specifications treat each growth determinant is acting separately from the others. This is inconsistent with the essentially nonlinear structure of new growth theories, in which the different determinants interact. This problem has often been addressed using products of variables as additional regressors. This use of products of variables is, we believe, a sensible way to begin to understand interactions. That said, we believe an important outstanding question is how to develop richer conceptions of interactions. The addition of From Growth Theory to Policy Design 21 products of variables may be understood as a step towards a secondorder Taylor series approximation of a general nonlinear growth model. It is unclear as to whether, for the sorts of interactions that theory suggests exist, this approximation is accurate. While there has been some interest in using semiparametric methods to allow for richer structures (early examples include Durlauf and Johnson, 1995; Liu and Stengos, 1999), the focus of this literature has been on the modeling of heterogeneity in the coefficients in (9). The one systematic evaluation of interactions between theories using semiparametric methods that we know of is Tan (2006). We note that semiparametric and nonparametric methods require larger data sets than are conventionally available for growth contexts. A third problem concerns the appropriate level of aggregation. New growth theories such as the innovation models focus on interactions that are defined with respect to firms and industries, not aggregate economies. For this reason, identification of these effects is problematic when the exclusive focus in on aggregate data. The reason for this is that the endogeneity of those aggregates that produce spillovers (such as human and physical capital) is difficult to disentangle from the effects of other aggregate determinants, such as the saving rate. In our view, the use of disaggregated data is a key to making progress on many of the empirical questions in growth. It is surprising that so little interaction exists between the growth literature as it relates to spillovers and the literature on social interactions, in which similar phenomena are studied and a host of identification issues have been analyzed since Manski's seminal work (1993). See Brock and Durlauf (2001a,b) for a synthesis of much of the literature and Brock and Durlauf (2007) for recent partial identification results. One message of the social interactions literature is that disaggregated data is not a panacea, as issues of selfselection and unobserved aggregate variables make the development of persuasive claims about spillovers difficult. Durlauf's (2002) criticisms of the use of aggregate data to identify spillovers were based on the relative strength of the necessary assumptions relative to those that are needed in disaggregated studies; substantive assumptions are still needed. A final problem is residual heterogeneity. Economic theory does not impose any natural homogeneity on the residuals in the model (9). The reason for this, using a term of Brock and Durlauf (2001a,b), is that new growth theories are openended in the sense that different theories are logically compatible. From the perspective of the substantive information that a researcher brings to a problem, the key issue concerns the requirement that each i in the growth regression is exchangeable, which in essence means that given the specification chosen for (9), a researcher is willing to assume that he has no basis for distinguishing the residuals of one country versus another. This is problematic given the knowledge that is available to a researcher about each country. Is it appropriate to disallow the possibility that over the last halfcentury there is a component of South 22 Philippe Aghion and Steven Durlauf Africa's growth that is countryspecific in the sense that it is not captured by any of the controls one introduces? One solution to this problem that has been widely adopted is to generalize the analysis to a panel context, so that countryspecific fixed effects may be allowed for. The limitations imposed by the incorporation of fixed effects in a panel analysis are well known. Differencing of panel data to eliminate the fixed effects makes inference difficult if the controls in the model only move slowly over time. And there are cases where these methods may not be available. As argued by Durlauf, Johnson, and Temple (2005): The fixedeffects identification strategy cannot be applied in all contexts. Sometimes a variable of interest is measured at only one point in time. Even where variables are measured at more frequent intervals, some are highly persistent, in which case the withincountry variation is unlikely to be informative. At one extreme, some explanatory variables of interest are essentially fixed factors, like geographic characteristics or ethnolinguistic diversity. Here the only available variation is "between country", and empirical work will have to be based on crosssections or pooled crosssection timeseries. In the remaining part of this subsection, we consider particular growth regressions and ways to address some of the issues raised above. Crosscountry regressions and beyond Barro (1991), Barro and SalaiMartin (1995), and Barro (1996) regress the average growth rate of (per capita) GDP across countries over potential growth determinants suggested by endogenous growth theory such as education spending, government expenditure, democracy measures, and so forth, measured on average over the period under consideration. As discussed above, this type of regression raises several issues. In particular, are we capturing the causality from policy variables to growth, or instead the reverse causality from growth to the level of these variables? For example, if more education is growth enhancing, at the same time a fastergrowing country can afford to invest more on education as it becomes wealthier. To address this endogeneity problem in the relationship between growth and education, AghionBoustanHoxbyVandenbussche (2005) use political economy variables (membership in appropriation committees for research education, existence of twoyear or fouryear colleges in the district of state's education head, and so forth) as potential instrumental variables for education spending or attainment. Similarly, to deal with the endogeneity problem in the relationship between growth and competition, AghionBoustanHoxby Vandenbussche (2005) use policy variables such as the implementation of the singlemarket program in the EU, or the extent of privatizations. A second problem with crosscountry regressions is that they do not control for country fixed effects. To deal with this latter criticism, one needs: (i) to move From Growth Theory to Policy Design 23 from pure crosssectional to panel regressions, that is, regressions where growth over a succession of limited subperiods (typically between one and five years) is regressed over a set of policy variables averaged over the same subperiods, or over the values of these variables at the beginning of the subperiod; (ii) to control for country fixed effects. The problem is often that the crosscountry panel data are too sparse for regression results to hold with the same degree of significance once one controls for country fixed effects. For example, VandenbusscheAghion Meghir (2005) show that the positive correlation between education and growth in the crosscountry OECD panel regression becomes insignificant once we fully controlled for country fixed effects. Similarly, the positive correlation between democracy and growth in Barro (1996) is not robust to the inclusion of country fixed effects, as pointed out by Acemoglu et al. (2007). Another example is Easterly (2005), who shows that once we control for institutions (or for the macroeconomic policy in the country being reasonable), the correlations between growth and policy variables such as inflation, openness, black market premium, and so forth may also become insignificant. A natural way out of this second problem is to move from crosscountry to more disaggregated crossindustry or even firmlevel data. However, there are policies such as education or market regulations that are often uniform across sectors within a given country. This in turn implies that to assess the effects of such policies, one cannot completely disregard crosscountry regressions. As discussed above, one way out is to interact the policy variables with other state variables, for example the country's distance to the technological frontier, or the country's level of financial development, or other institutional characteristics of the economy. For example, consider the relationship between volatility and growth. While no significant relationship would survive the restriction to "reasonable" countries (as in Easterly (2005)), introducing the interaction between volatility and financial development restores the significance of the righthand side coefficients (see AghionAngeletosBanerjeeManova (2005)): namely, in a country with low financial development, macroeconomic volatility is clearly detrimental to growth, whereas this is no longer true in countries with high degrees of financial development. Using crosssectoral variations We have just seen that introducing interaction terms on the righthand side of the growth regression can help restore the significance of correlations coefficients between some policy variables and growth. But this may not be enough. In particular, if a policy variable (such as financial development or the degree of democracy) does not vary much over time in the various countries in the sample, then it is hard to obtain correlations between this variable and growth that are robust to controlling for country fixed effects. 24 Philippe Aghion and Steven Durlauf An ingenious way out, suggested by Rajan and Zingales (1998), is to interact that variable with a sectoral characteristic (in RZ it is the degree of dependence upon external finance in the corresponding sector in the United States). This reflects the extent to which the policy variable (for example, financial development) affects investment and growth in that particular sector. Thus, using the RZ method, AghionAlesinaTrebbi (2007) show that democracy is more growthenhancing in sectors that are both more dependent on external finance and closer to the technological frontier. Also using the RZ method, AghionFallyScarpetta (2007) regress the postentry growth of firms (measured here by firms number of employees after six years of existence since entry, divided by the number of employees upon entry) over financial development and labor market regulations, respectively interacted with the sector's dependence upon external finance and the sector's capitallabor ratio (the idea being that a section with lower capitallabor ratio is more vulnerable to labor market regulations). Their main conclusion is that financial development, both bank credit and stock market finance, beats labor market regulations as a determinant of postentry growth of firms. Using crossfirm variations One can go even further and look at firmlevel data when data are too sparse at the country or even sectoral level. Thus, when analyzing the relationship between credit constraints, volatility, and the growthenhancing nature of investment using crosscountry panel data, AghionAngeletosBanerjeeManova (2005) would only get significant results by using structural investment as a proxy for growthenhancing investment. One problem there is that R&D panel data are not available for more than a handful of countries. However, once we move to firmlevel data, one can get more precise results. Thus, using a firmlevel French panel data set, Berman et al. (2007) could look at the relationship between macroeconomic volatility, R&D investments, and growth. Their main result is that R&D investments become more procyclical the more credit constrained the firm is. Moving from countrylevel to firmlevel data here has two additional virtues besides the fact that we can take advantage of detailed firmlevel R&D panel data. First, we have more observations, which in turn allows us to control for more variables without losing the significance of main coefficients. Second, instead of using private credit over GDP at country level as the measure of financial development, one can construct a more precise and firmspecific measure of credit constraint: namely, the amount of unpaid trade credits by the firm, information which in France is available to banks and affects directly the firm's ability to obtain new bank loans. From Growth Theory to Policy Design 25 3.3. Europe versus the United States and Latin America versus Asia How can we foster growth in Europe? Between 1995 and 2006, we observe persistently slow growth in the EU15 (less than 2 percent a year against 3 percent in the United States between 1995 and 2006). This is in sharp contrast with the 1945­75 period where labor productivity growth was much higher in Europe than in the United States (for example, the EU was still growing at an average of 3.5 percent per year on average during the 1970s versus 1.4 percent in the United States). How can we explain this reversal of fortune? The model in Section 2 suggests an explanation, namely that Europe has exhausted the growth potential generated by factor accumulation and imitation and needs now to adapt its institutions so as to foster innovation. Innovation has increasingly become a priority. In 1945, Europe's stock of physical capital had been largely destroyed and its technological knowledge as reflected by its average level of per capita GDP was far behind per capita GDP in the United States. At that time, what Europe needed for growth was to accumulate capital and to imitate or adapt technological innovations made elsewhere. And this is what Europe did quite successfully during the "trente glorieuses," with the support of economic institutions and policies that were adapted to those goals, in particular: limited competition in the product market; large firms financed by banks and by government subsidies; educational systems emphasizing primary, secondary, and specialized undergraduate education; and rigid labor markets that would privilege human capital investment within firms over mobility. However, by the late 1980s Europe had largely caught up with the world technology frontier in terms of its capitallabor ratio and also in terms to its per capita GDP level. This in turn implied that Europe had largely exhausted capital accumulation and technological imitation as its main sources of growth, and had to turn to an alternative source, namely innovation, that is, the ability for firms and workers to move rapidly into new activities or to improve production processes. In the meantime the IT revolution resulted in a revival of U.S. growth in the late 1980s and early 1990s. Since Europe did not have the institutions and policies to benefit from this technological revolution, the result was a reversal of Europe's approach to the frontier, and the fact that Europe is not adjusting its institutions in order to produce growthmaximizing innovation policy acts as a delaying force on growth convergence towards the United States. A first way to foster innovation is to invest more in R&D. As we all know, EU countries invest less than 2 percent of their GDP in R&D whereas the United States invests between 2 and 3 percent. Investment in R&D becomes more essential as industries move closer to their technological frontier. This is evidenced by looking at the correlation between distance to frontier and R&D 26 Philippe Aghion and Steven Durlauf intensity at the industry level using data from the OECD sectoral database previously analyzed by (among others) Griffith et al. (2004). A natural (inverse) proxy for distance to frontier is "proximity to frontier" for an industry, PTGFict , defined as TFP in industry i in country c at time t divided by the highest TFP in industry i at time t in the sample. We obtain estimates of proximity to frontier as well as data on R&D intensity, RDict (R&D divided by sales) for the years 1974­90. AcemogluAghionZilibotti (2002) reports the correlation between these two measures with or without controlling for country and industry effects. The first three columns use a measure of PTFict without correcting for differences in skills and hours, while the last three columns use a measure that corrects for these differences. All columns show the same pattern of a statistically significant positive correlation between proximity to frontier and R&D intensity: industries closer to their respective frontier are more R&D intensive. Moreover, as an industry approaches the world technology frontier more rapidly than others it becomes relatively more R&D intensive. These results are consistent with the view that R&D is more important in industries or countries closer to the world technology frontier. Therefore, the lesson is: now that they have moved closer to the world technological frontier, EU countries should invest more in R&D. A naive view, however, is that R&D subsidies would be sufficient to foster innovation and productivity growth: that is, just invest more in R&D here and there and the economy will grow faster. In the remaining part of this note we will discuss five indirect sources of innovation and growth: competition/entry, education, financial development, labor market regulations, and the conduct of macroeconomic (particularly fiscal) policy over the business cycle. Competition and entry. As stressed by the Sapir report (see Sapir (2003)), competition policy in Europe has emphasized competition among incumbent firms but paid insufficient attention to entry. Entry, as well as exit and turnover of firms, are more important in the United States than Europe. For example, 50 percent of new pharmaceutical products are introduced by firms that are less than 10 years old in the United States, versus only 10 percent in Europe. Similarly, 12 percent of the largest U.S. firms by market capitalization at the end of the 1990s had been founded less than 20 years before, against only 4 percent in Europe, and the difference between U.S. and European turnover rates is much bigger if one considers the top 500 firms. The higher entry costs and lower degree of turnover in Europe compared to the United States are an important part of the explanation for the relatively disappointing European growth performance over the past decade, as has been shown in empirical work by Nicoletti and Scarpetta (2003). However, as shown by AghionBloomBlundellGriffithHowitt (2005) and AghionBlundellGriffithHowittPrantl (2005), entry and competition are more growthenhancing for industries that are closer to their technological frontier. More specifically, these papers look at patenting rates within a panel of UK From Growth Theory to Policy Design 27 manufacturing firms over the period 1973­92 as a function of the degree of competition in the industry. They show that that if we restrict the set of industries to those that are closer to their world frontier than the median, the upward sloping part of the invertedU relationship between competition and innovation is steeper than when we consider the whole sample of industries. What is true for competition is also true for entry. Namely, entry has a more positive effect on growth for industries that are close to the technological frontier than for those that are not. During the immediate postWWII period European firms were predominantly technological laggards whose innovation rates would have been diminished by very intense competition. Thus for some time the relatively noncompetitive nature of Europe was favorable to innovation and productivity growth by European firms. However, as Europe has moved closer to the global technological frontier, competition and entry have become increasingly important catalysts for innovation and productivity growth. Education. Is the European education system growth maximizing? A first comparison of the United States to the EU in 1999­2000 shows that 37.3 percent of the U.S. population aged 25­64 has completed a higher education degree, against only 23.8 percent of the EU population. This educational attainment comparison is mirrored by that on tertiary education expenditure, with the United States devoting 3 percent of its GDP to tertiary education versus only 1.4 percent in the EU. Is this European deficit in tertiary education investment a big deal for growth? Recent analyses by VandenbusscheAghionMeghir (2004), and more recently the cross­U.S. state analysis by AghionBoustanHoxbyVandenbussche (2005), ABHV, suggest that the deficit is important. Namely, as Europe has moved closer to the world technological frontier, it should invest more in tertiary education in order to increase its innovative potential. In particular, ABHV show that not investing in higher education is more damaging to growth, the closer a U.S. state is to the world technology frontier. What is true between countries is also true among regions within a country. For example, AghionBoustanHoxbyVandenbussche (2005) show that an additional $1,000 per person in research education spending raises a U.S. state's productivity growth rate by 0.27 percent if the state is at the frontier, whereas it raises it by only 0.09 percent if the state is far from the frontier. Europe's emphasis on primary and secondary education was fine as long as it was technologically far from the United States and therefore relying more on imitation as a main source of growth. Now that the growth potential of imitation is wearing out, it becomes more urgent to invest more in higher education in order to foster innovation. Other evidence shows that the IT and globalization waves of the 1980s further increased the growth potential of higher education investment in all OECD countries. 28 Philippe Aghion and Steven Durlauf Credit constraints versus labor market regulations. Labor market rigidities are often presented as the main impediment to firms' entry, mobility, and post entry growth. As it turns out, financial constraints are at least if not more important. The study by AghionFallyScarpetta (2007) mentioned above shows that credit constraints act as a main barrier to the entry and postentry growth of small firms, whereas labor market regulations inhibit the entry of larger firms. Macropolicy. There is currently a debate on the conduct of macroeconomic policy in the euro area. In a nutshell, it has been noticed that structural budget deficits and shortterm interest rates fluctuate much less over the cycle in the European Monetary Union than in the United States and United Kingdom, and some policy makers have raised the concern that this in turn may inhibit growth in the euro area. Are those concerns justified? Mainstream macroeconomists would say no. Indeed, a common prejudice in macroeconomics is that there is a perfect dichotomy between: (i) macroeconomic policy (budget deficit, taxation, money supply), taken to affect primarily the short run and the primary aim of which is to stabilize the economy; and (ii) long run economic growth, which is either taken to be exogenous or to depend only upon structural characteristics of the economy (property right enforcement, market structure, market mobility, and so forth). The only link between macro policy and longrun growth that most policy makers believe in is that growth requires macroeconomic stability, everything else remaining equal. However, one might depart from that view. According to Schumpeter, recessions would provide a cleansing mechanism for correcting organizational inefficiencies and for encouraging firms to reorganize, innovate, or reallocate to new markets. The cleansing effect of recessions is also to eliminate those firms that are unable to reorganize or innovate. Now, if firms could always borrow enough funds to either reorganize their activities or move to new activities and markets, and the same was true for workers trying to relocate from one job to another, the best policy recommendation would be that governments do not intervene over the business cycle, and instead let markets operate. However, things become quite different when credit market imperfections prevent firms from innovating and reorganizing in recessions. For example, suppose that the borrowing capacity of firms is proportional to their current earnings. In a recession, current earnings are reduced, and therefore so is the ability of firms to borrow in order to innovate. But then, a countercyclical budgetary policy may foster innovation and growth by reducing the negative consequences of a recession (or a bad aggregate shock) on firms' innovative investments. For example, the government may decide to increase the volume of its public investments, thereby fostering the demand for private firms' products. Or the government may choose to directly increase its subsidies to private enterprises, thereby increasing their liquidity holdings and thus making it easier for them to face idiosyncratic liquidity shocks without having to sacrifice R&D or other types of longerterm growthenhancing investments. From Growth Theory to Policy Design 29 That countercyclical government support to the enterprise sector may be growthenhancing is suggested by the work of Berman et al. (2007) mentioned above, which uses French firmlevel data set to show that R&D investments are more hurt in downturns in firms that are more credit constrained. Recent work by Aghion and Marinescu (2007) provides consistent evidence to the effect that the correlation between countercyclical budgetary policy and growth is more positive the lower a country's level of financial development. Interestingly, budgetary policies are currently far less countercyclical in the EU than in the United States, even though the latter is more financially developed. Indeed, the ratio of private credit to GDP in the EU is 0.76 against 1.32 in the United States. This difference reflects differences in stock market and venture capital market development, both of which are far more developed in the United States. As it turns out, both the structural deficit and the real interest rates vary much less over time in the euro area than in the United States, and the above discussion suggests that the absence of an active (or reactive) macro policy in the euro area is a potential source of growth deficit in this region. Latin America versus Asia What can Latin American countries do to get closer to the high growth rates experienced by Asian countries? As we mentioned in the introduction, capital output ratios are the same in OECD countries. Yet, the framework in Section 2 suggests several complementary levers to foster growth in that region. Savings. In a recent paper, AghionCominHowittTecu (2009) argue that domestic savings can serve as collateral to attract foreign direct investment, and that foreign direct investment in turn is more growthenhancing for countries that are not too close to the technological frontier as it brings knowledge from the frontier to those countries (in terms of the model in Section 2, domestic savings and FDI enhance the implementation rate m ). This prediction is tested using a crosscountry panel regression where growth is regressed over the domestic savings rate and its interaction with the country's proximity to the frontier. The interaction coefficient between savings and proximity is found to be positive and significant, which in turn means that growth in middleincome countries such as those in Latin America would benefit from a higher savings rate. To show how much growth potential can be gained on that front, it suffices to note that average private savings in Latin America over the period 2000­02 were only 7.8 percent of GDP, versus 28.1 percent in the Asian tigers including China. Education. Education is another lever of growth that has been largely underexploited or misexploited in Latin America. That education spending and attainment has substantial growth effects is shown in several studies, including VandenbusscheAghionMeghir (2004) and AghionBoustanHoxbyVandenbussche (2005) mentioned above. However, if we look at the BarroLee schooling data in 1985 for the population aged between 25 and 64 years old, we find a schooling index of only 3.49 in Brazil and 6.45 in Chile, versus 11.79 in the United States. 30 Philippe Aghion and Steven Durlauf Not only do Latin American countries underinvest in education, they also misinvest in education. Indeed, the analysis in VandenbusscheAghionMeghir (2004) suggests that given Latin American countries' distance to the world productivity frontier, they should invest more, not less, in primary and secondary education than OECD countries, and they should invest less, not more, in tertiary education. However, if we look respectively at spending in primary, secondary, and tertiary education in percentage of per capita GDP, we find that Brazil invests 10.8 percent in primary, 11.2 percent in secondary, and 48.87 in tertiary. By comparison, the United States invests 21.6 percent in primary, 24.4 percent in secondary, and 26 percent in tertiary. Thus, Brazil overinvests in tertiary education, which we take as resulting from a long history of excessive income and wealth inequality and of capture of the political system by the well off, who would enjoy almost exclusive access to higher education. We do not recommend that Latin American countries give up on higher education altogether, but that they place more emphasis on primary, secondary, and undergraduate education. Competition. Existing growth analyses suggest that competition and entry tend to enhance innovation and growth, especially in sectors and countries that are not too far from the technology frontier. One may argue that Latin America is not so close to the frontier, and that the same argument that led us to advocate more emphasis on lower education would also push for not emphasizing competition policy. However, if we look at the recent liberalization experience in India (see AghionBurgessReddingZilibotti (2005a,b)) we find that while liberalizing entry has enhanced productivity growth more in more advanced sectors than in sectors further below the frontier, overall this has enhanced growth in the Indian manufacturing sector. The same should presumably be true in Latin American countries which are, if anything, closer to the world productivity frontier than India was in 1985 or in 1991 when delicensing and trade liberalization reforms took place. The socalled Frazer index (which ranks countries according to their levels of product market deregulations), suggests that much growth potential can be gained in Latin America by fostering competition and entry: Brazil is ranked 88th, Argentina is ranked 74th, Mexico is ranked 60th, and only Chile does relatively fine, being ranked 20th. Financial development. There exists a large literature on the positive effects of finance on growth and convergence, also based on new growth theory. Here the idea is simply that financial constraints limit firms' ability to undertake innovative or implementation activities; see Levine (2005) for a firstrate survey on the question. Here again, Latin American countries are dramatically lagging behind. If we look at the ratio of private credit to GDP as the measure of financial development, we find a ratio of only 0.345 in Brazil, 0.244 in Argentina, and 0.184 in Mexico; but we find 0.683 in Chile, which is close to 0.76 in the euro area but still far below 1.42 in the United States. From Growth Theory to Policy Design 31 4. Conclusion In this paper we have argued that new growth theory combined with suitable growth regressions has the potential to deliver growth policy recommendations. In particular we have argued that growth in the EU should benefit from a more active competition and entry policy, a bigger emphasis on higher education, and a more proactive macroeconomic policy over the cycle. Similarly, although more econometric work needs to be done in this case, we have argued that growth in Latin America should benefit from a combination of higher domestic savings, higher investments in education starting with primary and secondary education, from opening up competition and entry (with complementary policies to help workers relocate from lagging sectors to more advanced sectors), and from higher financial development. However, so far we have taken a Pigovian approach and ignored the political costs of implementing various reforms. A necessary next step is to look at more specific policy instruments and for each instrument calculate a ratio of the additional growth potential associated with this instrument with respect to the cost of implementing the corresponding policy. This in turn would enable us to "rank" the reforms, that is, to get a more precise view as to what should be undertaken first, or as to which reforms should be implemented jointly because of complementarities in their growth impacts. Also, we have only briefly touched upon the difficult issues raised by the use of growth regressions. These are discussed in detail in Durlauf, Johnson, and Temple (2005). 32 Philippe Aghion and Steven Durlauf References Acemoglu, Daron, Aghion, Philippe, and Fabrizio Zilibotti (2002). "Distance to Frontier, Selection, and Economic Growth." NBER Working Paper No 9066. National Bureau of Economic Research, Cambridge, MA. Acemoglu, Daron, Philippe Aghion, Claire Lelarge, John Van Reenen, and Fabrizio Zilibotti (2007). "Technology, Information and the Decentralization of the Firm." Quarterly Journal of Economics 122(4): 1759­ 1799. Acemoglu, Daron, and Josh Angrist (2000). "How Large Are Human Capital Externalities? Evidence from Compulsory Schooling Laws." NBER Macroannual, 9­59. Acemoglu, Daron, and Jaume Ventura (2002). "The World Income Distribution." Quarterly Journal of Economics 117(2): 659­694. Aghion, Philippe, Alberto Alesina, and Francesco Trebbi (2007). "Democracy, Technology, and Growth." Mimeo, Harvard. Aghion, Philippe, and Patrick Bolton (1997). "A Theory of TrickleDown Growth and Development." Review of Economic Studies 64(2): 151­172. Aghion, Philippe, Marios Angeletos, Abhijit Banerjee, and Kalina Manova (2005). "Volatility and Growth: Credit Constraints and ProductivityEnhancing Investment." Mimeo, HarvardMIT. Aghion, Philippe, Nick Bloom, Richard Blundell, Rachel Griffith, and Peter Howitt (2005). "Competition and Innovation: An InvertedU Relationship." Quarterly Journal of Economics 120 (May 2005): 701­28. Aghion, Philippe, Richard Blundell, Rachel Griffith, Peter Howitt, and Susanne Prantl (2004). "Entry and Productivity Growth: Evidence from MicroLevel Panel Data." Journal of the European Economic Association, Papers and Proceedings 2: 265­276. ------ (2005), "Entry, Innovation, and Growth: Theory and Evidence from UK FirmLevel Data." Mimeo, HarvardUCL. Aghion, Philippe, Leah Boustan, Caroline Hoxby, and Jerome Vandenbussche (2005). "Exploiting States' Mistakes to Evaluate the Impact of Higher Education on Growth." Mimeo, Harvard. Aghion, Philippe, Robin Burgess, Stephen Redding, and Fabrizio Zilibotti (2005a). "Entry Liberalization and Inequality in Industrial Performance." Journal of the European Economic Association, Papers and Proceedings. ------ (2005b). "On the Unequal Effects of Liberalization: Theory and Evidence from Indian Delicensing Data." Mimeo, LSEHarvardIIES. From Growth Theory to Policy Design 33 Aghion, Philippe, Gilbert Cette, Elie Cohen, and Jean Pisani (2007). Les Leviers de la Croissance Francaise. La Documentation Francaise. Aghion, Philippe, Diego Comin, Peter Howitt, and Isabel Tecu (2009). "When Domestic Savings Matters for Economic Growth." Mimeo. Harvard. Aghion, Philippe, Thibault Fally, and Stefano Scarpetta (2007). "Credit Constraints as a Barrier to the Entry and PostEntry Growth of Firms." Mimeo, Harvard. Aghion, Philippe, and Peter Howitt (1992). "A Model of Growth through Creative Destruction." Econometrica 60: 323­351. ------ (1998). Endogenous Growth Theory. Cambridge, MA: MIT Press. ------ (2006). "Appropriate Growth Policy." Schumpeter Lecture, Journal of the European Economic Association, Papers and Proceedings. Aghion, Philippe, Peter Howitt, and David MayerFoulkes (2005). "The Effect of Financial Development on Convergence: Theory and Evidence." Quarterly Journal of Economics 120: 173­222. Aghion, Philippe, and Ioana Marinescu (2007). "Cyclical Budgetary Policy and Productivity Growth: What Do We Learn from OECD Panel Data?" Mimeo, Harvard. Barro, Robert (1991). "Economic Growth in a Cross Section of Countries." Quarterly Journal of Economics 106(2): 407­43. ------ (1996). "Democracy and Growth." Journal of Economic Growth 1: 1­27. Barro, Robert, and Xavier SalaiMartin (2005). Economic Growth. Cambridge, MA: MIT Press. Barro, Robert, and JongWha Lee (2000). "International Data on Educational Attainment: Updates and Implications." Working Paper No 42, Center for International Development at Harvard University. Benhabib, Jess, and Mark Spiegel (1994). "The Role of Human Capital in Economic Development: Evidence from Aggregate CrossCountry Data." Journal of Monetary Economics 34: 143­173. Berman, Nicolas, Laurent Eymard, Philippe Aghion, Philippe Aszenazy, and Gilbert Cette (2007). "Credit Constraints and the Cyclicality of R&D: Lessons from French FirmLevel Panel Data." Mimeo, Banque de France. Bils, Mark, and Peter Klenow (2000). "Does Schooling Cause Growth?" American Economic Review 90: 1160­1183. Blundell, Richard, Rachel Griffith, and John Van Reenen (1995). "Dynamic Count Data Models of Technological Innovation." Economic Journal 105: 333­344. 34 Philippe Aghion and Steven Durlauf Brock, W., and S. Durlauf (2001a). "Growth Empirics and Reality." World Bank Economic Review 15(2): 229­272. ------ (2001b). "InteractionsBased Models." In Handbook of Econometrics, J. Heckman and E. Leamer, eds., Amsterdam: NorthHolland. ------ (2007). "Identification of Binary Choice Models with Social Interactions." Journal of Econometrics 140(1): 57­75 Burnside, C., and D. Dollar, (2000). "Aid, Policies, and Growth." American Economic Review 90(4): 847­868. Dixit, Avinash K., and Joseph E Stiglitz (1977). "Monopolistic Competition and Optimal Product Diversity." American Economic Review 67(3): 297­308. Durlauf, S. (2002). "On the Empirics of Social Capital." Economic Journal 112: 459­ 479. Durlauf, S., and P. Johnson (1995). "Multiple Regimes and CrossCountry Growth Behavior." Journal of Applied Econometrics 10(4): 363­384. Durlauf, S., P. Johnson, and J. Temple (2005). "Growth Econometrics." In Handbook of Economic Growth, P. Aghion and S. Durlauf, eds. Amsterdam: North Holland. Easterly, William (2005). "National Policies and Economic Growth: A Reappraisal." Forthcoming in the Handbook of Economic Growth, edited by P. Aghion and S. Durlauf. Frankel, Marvin (1962). "The Production Function in Allocation and Growth: A Synthesis." American Economic Review 52: 995­1022. Gerschenkron, Alexander (1962). Economic Backwardness in Historical Perspective. Harvard University Press. Griffith, Rachel, Steve Redding, and John Van Reenen (2004). "Mapping the Two Faces of R&D: Productivity Growthin a Panel of OECD Industries." Review of Economic and Statistics 86: 883­895. Ha, Joonkyung, and Peter Howitt (2005). "Accounting for Trends in Productivity and R&D: A Schumpeterian Critique of SemiEndogenous Growth Theory." Mimeo, Brown University. Hausmann, R., D. Rodrik, and A. Velasco (2005). "Growth Diagnostics." Manuscript, John F. Kennedy School of Government, Harvard University. Helpman, Elhanan. 1993. "Innovation, Imitation, and Intellectual Property Rights." Econometrica 61(6): 1247­1280. Howitt, Peter (2000). "Endogenous Growth and CrossCountry Income Differences." American Economic Review 90: 829846. From Growth Theory to Policy Design 35 Howitt, Peter, and David MayerFoulkes (2005). "R&D, Implementation and Stagnation: A Schumpeterian Theory of Convergence Clubs." Journal of Money, Credit and Banking 37: 147­177. Jones, Charles (1995). "R&DBased Models of Economic Growth." Journal of Political Economy 103: 759­784. King, Robert, and Sergio Rebelo (1993). "Transitional Dynamics and Economic Growth in the Neoclassical Model." American Economic Review 83: 908­931. Krueger, Alan, and Mikael Lindahl (2001). "Education for Growth: Why and For Whom?" Journal of Economic Literature 39: 1101­1136. Levine, Ross (2005). "Finance and Growth: Theory and Evidence." In P. Aghion and S. Durlauf, eds, Handbook of Economic Growth. Elsevier. Liu, Z., and T. Stengos (1999). "NonLinearities in Crosscountry Growth Regressions: A Semiparametric Approach." Journal of Applied Econometrics 14(5): 527­38. Lucas, Robert (1988). "On the Mechanics of Economic Development." Journal of Monetary Economics 22: 3­42. Manski, C. (1993). "Identification of Endogenous Social Effects: The Reflection Problem." Review of Economic Studies 60(3): 531­542. Mankiw, Gregory, David Romer, and David Weil (1992). "A Contribution to the Empirics of Economic Growth." Quarterly Journal of Economics 25: 275­310. Nelson, Richard, and Edmund Phelps (1966). "Investment in Humans, Technological Diffusion, and Economic Growth." American Economic Review, Papers and Proceedings 61: 69­75. Nickell, Steven (1996). "Competition and Corporate Performance." Journal of Political Economy 104: 724­46. Nicoletti, Giuseppe, and Stefano Scarpetta (2003). "Regulation, Productivity and Growth." Economic Policy 36: 11­72. Rajan, Raghuram G., and Luigi Zingales (1998). "Power in a Theory of the Firm." Quarterly Journal of Economics 113(2): 387­432. Ramey, Garey, and Valerie Ramey (1995). "CrossCountry Evidence on the Link Between Volatility and Growth." American Economic Review 85: 1138­1151. Rodrik, Dani (2005). "Growth Strategies." Forthcoming in the Handbook of Economic Growth, edited by P. Aghion and S. Durlauf. Romer, Paul (1986). "Increasing Returns and LongRun Growth." Journal of Political Economy 94: 1002­1037. ------ (1990). "Endogenous Technical Change." Journal of Political Economy 98: 71­102. 36 Philippe Aghion and Steven Durlauf Sapir, Andre, et al. (2003). An Agenda for A Growing Europe. Oxford University Press. Solow, Robert (1956). "A Contribution to the Theory of Economic Growth." Quarterly Journal of Economics 70: 65­94. Tan, C. M. (2006). "No One True Path to Development: Uncovering the Interplay Between Geography, Institutions, and Ethnic Fractionalization in Economic Development." Mimeo, Tufts University. Uzawa, Hirofumi (1965). "Optimum Technical Change in an Aggregative Model of Economic Growth." International Economic Review 6: 1831. Vandenbussche, Jerome, Philippe Aghion, and Costas Meghir (2004). "Growth, Distance to Frontier and the Composition of Human Capital." Mimeo, HarvardUCL. From Growth Theory to Policy Design 37 Eco-Audit Environmental Benefits Statement The Commission on Growth and Development is committed to preserving endangered forests and natural resources. The World Bank's Office of the Publisher has chosen to print these Working Papers on 100 percent postconsumer recycled paper, processed chlorine free, in accordance with the recommended standards for paper usage set by Green Press Initiative--a nonprofit program supporting publishers in using fiber that is not sourced from endangered forests. For more information, visit www.greenpressinitiative.org. The printing of all the Working Papers in this Series on recycled paper saved the following: Trees* Solid Waste Water Net Greenhouse Gases Total Energy 48 2,247 17,500 4,216 33 mil. *40 inches in height and 6­8 Pounds Gallons Pounds CO2 Equivalent BTUs inches in diameter The Commission on Growth and Development Working Paper Series 48. Macroeconomic Policy: Does it Matter for Growth? The Role of Volatility, by Antonio Fatás and Ilian Mihov, February 2009 49. Current Debates on Infrastructure Policy, by Antonio Estache and Marianne Fay, April 2009 50. Explaining China's Development and Reforms, by Bert Hofman and Jingliang Wu, March 2009 51. Growth Challenges for Latin America: What Has Happened, Why, and How To Reform The Reforms, by Ricardo Ffrench-Davis, March 2009 52. Chile's Growth and Development: Leadership, Policy-Making Process, Policies, and Results, by Klaus Schmidt-Hebbel, March 2009 53. Investment Efficiency and the Distribution of Wealth, by Abhijit V. Banerjee, April 2009 54. Africa's Growth Turnaround: From Fewer Mistakes to Sustained Growth, by John Page, April 2009 55. Remarks for Yale Workshop on Global Trends and Challenges: Understanding Global Imbalances, by Richard N. Cooper, April 2009 56. Growth and Education, by Philippe Aghion, April 2009 57. From Growth Theory to Policy Design, by Philippe Aghion and Steven Durlauf, July 2009 Forthcoming Papers in the Series: Eight Reasons We Are Given Not to Worry about the U.S. Deficits, by Jeffrey Frankel (June 2009) A New Bretton Woods? by Raghuram G. Rajan (June 2009) Electronic copies of the working papers in this series are available online at www.growthcommission.org. They can also be requested by sending an e-mail to contactinfo@growthcommission.org. T his paper focuses on how growth theory can guide growth policy design. It first argues that policy matters for growth, in particular when policy variables are interacted with country-specific variables (financial development, institutional Commission on Growth and Development environment, technological development, and so forth). Second, it argues that Montek Ahluwalia Edmar Bacha the Schumpeterian paradigm does a better job at delivering policy prescriptions Dr. Boediono that vary with country characteristics. Third, it discusses the advantages and Lord John Browne drawbacks of growth regression analysis. Finally, it briefly describes and then Kemal Dervis ¸ questions the recently proposed "growth diagnostic" approach. Alejandro Foxley Goh Chok Tong Philippe Aghion, Professor of Economics, Harvard University Han Duck-soo Steven N. Durlauf, Professor of Economics, University of Wisconsin-Madison Danuta Hübner Carin Jämtin Pedro-Pablo Kuczynski Danny Leipziger, Vice Chair Trevor Manuel Mahmoud Mohieldin Ngozi N. Okonjo-Iweala Robert Rubin Robert Solow Michael Spence, Chair Sir K. Dwight Venner Hiroshi Watanabe Ernesto Zedillo Zhou Xiaochuan The mandate of the Commission on Growth and Development is to gather the best understanding there is about the policies and strategies that underlie rapid economic growth and poverty reduction. The Commission's audience is the leaders of developing countries. The Commission is supported by the governments of Australia, Sweden, the Netherlands, and United Kingdom, The William and Flora Hewlett Foundation, and The World Bank Group. www.growthcommission.org contactinfo@growthcommission.org