WPS3839 The case for industrial policy: a critical survey* Howard Pack Kamal Saggi The Wharton School Department of Economics University of Pennsylvania Southern Methodist University 1400 Steinberg Hall-Dietrich Hall PO Box 750496 3620 Locust Walk Dallas, TX 75275-0496 Philadelphia, PA 19104-6372 E-mail: ksaggi@smu.edu E-mail: packh@wharton.upenn.edu Abstract: What are the underlying rationales for industrial policy? Does empirical evidence support the use of industrial policy for correcting market failures that plague the process of industrialization? To address these questions, we provide a critical survey of the analytical literature on industrial policy. We also review some recent industry successes and argue that only a limited role was played by public interventions. Moreover, the recent ascendance of international industrial networks, which dominate the sectors in which least developed countries have in the past had considerable success, implies a further limitation on the potential role of industrial policies as traditionally understood. Overall, there appears to be little empirical support for an activist government policy even though market failures exist that can, in principle, justify the use of industrial policy. World Bank Policy Research Working Paper 3839, February 2006 The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Policy Research Working Papers are available online at http://econ.worldbank.org. *This paper has been partly funded by the Development Research Group of the World Bank. 1. Introduction Many nations in recent years have encountered great disappointment with the results of pursuing conventional economic policies that Williamson crystallized and named the Washington Consensus. Although few countries ever followed the pristine form of this consensus, some countries in East Asia adhered to many (but hardly all) of its components and experienced extraordinarily rapid growth for a period of three decades or more. Though there was a brief and sharp recession in some of these countries during the crisis of 1997 to 1999, most have rebounded with the exception of Indonesia. Yet other nations that have gotten their macroeconomic and trade regimes much closer to the idealized consensus than the Asian countries did have failed to experience comparable growth. In many Latin American nations and in some African ones as well, there is an understandable search for the magic bullet and many policy makers have expressed interest in some form or other of industrial policy. Few phrases elicit such strong reactions from economists and policy-makers as industrial policy. As Evenett (2003) notes, the term `industrial policy' means different things to different people. According to us, industrial policy is basically any type of selective intervention or government policy that attempts to alter the sectoral structure of production toward sectors that are expected to offer better prospects for economic growth than would occur in the absence of such intervention, i.e., in the market equilibrium. Policies designed to improve the productivity of individual sectors and firms are a subsidiary but often pursued objective. Given this definition, it is not surprising that those who believe strongly in the efficient working of markets view any argument in favor of industrial policy as fiction or, worse, an invitation for all types of rent seeking activities. On the other side, people who believe market failures are pervasive think that any path to economic development requires a liberal dose of industrial policy. In this paper, we address arguments for and against industrial policy and then ask whether empirical evidence helps settle the debate in favor of one group or another. As is often the case, evidence does not come out strongly in favor of one group. While there certainly exist cases where government intervention co-exists with successes, in many instances industrial policy has failed to yield any gains. Above all, the real issue is that 2 the relevant counterfactuals are not available. Consider the argument that Japan's industrial policy was crucial for its success. Since we do not know how Japan would have fared under laissez-faire, it is difficult to attribute its success to its industrial policy. Maybe it would have done still better in the absence of industrial policy or maybe it would have done much worse. Given this basic difficulty, we can only hope to obtain indirect clues regarding the efficacy of industrial policy. Direct evidence that can `hold constant' all of the required variables (as would be done in a well specified econometric exercise) simply does not exist and it is unlikely that it will ever exist ­ perhaps that is why the debate over industrial policy has remained unsettled and may remain so in the future. The paper is organized as follows. In section 2 we critically analyze the major conceptual arguments in favor of industrial policy. Since the infant industry argument for trade protection anticipates most of the rationales for industrial policy, we begin with a discussion of this argument. Then, in section 3, we examine how the case of India's successful software industry fits into the arguments for industrial policy. In section 4, we ask how the expansion of international production networks has altered the case for industrial policy. In section 5, we provide some concluding remarks where we also comment on the issue of `policy space.' 2. Why industrial policy? In this section, we present a detailed discussion of the various theoretical arguments in favor of industrial policy. At a general level, there is room for government intervention either when markets are characterized by some distortions (such as externalities or presence of market power) or because they are incomplete (for example futures markets for many goods simply do not exist). As is known from one of the basic theorems of welfare economics, under such market failures, a competitive market system does not yield the socially efficient outcome. In the end, any argument for industrial policy is a special case of this general argument. Three arguments in favor of industrial policy have received the most attention. The first is derived from the presence of knowledge spillovers and dynamic scale economies; the second stems from the presence of coordination failures while the third concerns informational externalities. Before discussing these arguments in detail, it is 3 useful to begin with a rather well known argument for trade protection that is in many ways the precursor for modern arguments for industrial policy. In fact, as will become clear shortly, at least two of the aforementioned arguments for industrial policy play an important role in the infant industry argument for trade protection. A. The infant industry argument: a precursor of modern industrial policy The infant industry argument is one of the oldest arguments for trade protection and is perhaps the only such argument that is not dismissed out of hand by trade economists. The most popular (and the simplest) version of the argument runs as follows. Production costs for newly established domestic industries in a country may be initially higher than those of well-established foreign competitors due to their greater experience. However, over time, new domestic producers can experience cost reductions due to learning by doing (i.e. they enjoy dynamic scale economies) and can end up attaining the production efficiency of their foreign rivals. However, due to the initial absence of experience, if domestic industry is not protected from foreign competition, it will never take off and if dynamic scale economies are strong enough, temporary protection of the domestic industry can be in the national interest. A stronger version of the argument states that the domestic industry might even be capable of attaining production costs below its foreign rivals if it is given sufficient protection. In this version of the argument, true comparative advantage lies with the domestic industry so that temporary protection may actually be in the global interest ­ consumers in the rest of the world can also benefit from the eventual lower production cost of the domestic industry. In an influential paper, Baldwin (1969) provided an incisive criticism of the infant industry argument. He argued that "if after the learning period, unit costs in an industry are sufficiently lower than those during its early production stages to yield a discounted surplus of revenues over costs (and therefore indicate a comparative advantage for the country in the particular line), it would be possible for firms in the industry to raise sufficient funds in the capital market to cover their initial excess of outlays over receipts." In other words, as Baldwin points out, the period of learning during which domestic firms are unprofitable can be treated as an initial fixed cost that can be recovered once the industry is globally competitive. If future returns indeed outweigh 4 initial losses, capital markets would finance the necessary investment needed by the domestic industry. It is obvious, but worth stressing, that if future returns fall short of initial losses, the industry should not be established in the first place. A frequently cited counter to Baldwin's criticism (and one that he acknowledged) is that capital markets might be imperfect and therefore the industry may fail to gain the required financing. For example, a proponent might appeal to the presence of informational asymmetries: unlike producers, investors may not know that the industry is profitable in the long run and therefore fail to provide the capital needed to cover the initial costs. However, such an argument defies credibility since it requires one to believe that firms that have not even begun to produce the good in question know more about their prospects than those whose main objective is to find profitable uses for their excess capital and have analyzed and financed similar projects. Even if one grants the presence of asymmetric information, why cannot potential producers convey such information to likely investors? While the infant industry argument assumes that it is known with certainty that the industry in question will eventually be profitable, it seems more likely that the prospects for most new industries are uncertain and no one really knows whether or not a particular infant industry will in fact be profitable in the future. Under such circumstances, capital markets would require compensation for the risks involved and the interest rates required might make the investment unprofitable. But efficiency requires that those bearing risks should be compensated and there is no market failure if the underlying problem is that investors do not provide the necessary capital because they perceive the rewards to be not commensurate with the risks they are asked to bear. Nevertheless, the assumption of omniscient financial intermediaries should be viewed with some degree of skepticism. From early bubbles such as the Tulip mania to the Asian Crisis of the late 1990s it is clear that financial actors are often deficient. In the case of countries that suppressed the financial sector and directed industry and firm specific loans as a part of industrial policy, the banking sector was itself in need of significant improvement in operating procedures much as industrial firms were. Thus, the argument that if there were opportunities they would be exploited by investors might be a weak link in Baldwin's argument. On the other hand, it also implies that any selective economic policies would have to simultaneously address the weakness of this sector 5 along with that of manufacturing or other services. Indeed there might be an argument for initially strengthening the banking sector, perhaps by allowing foreign financial intermediaries into the country, before pursuing targeted sectoral policies. In any case, as Baldwin notes, if there indeed is a problem with capital markets, policy ought to target that specific problem as opposed to providing trade protection for an industry that is claiming to be profitable but cannot convince investors that such is the case. In today's world of global capital markets, the simple version of the infant industry argument runs into a real difficulty ­ since there are foreign producers that have achieved low production costs due to their head start, investors can determine the prospects for the domestic infant industry from the experience of existing foreign producers. If domestic investors lack such information, surely foreign investors ought to have it. Why cannot the borrowing be international rather than local? A potential answer to this question is that investors may believe that just because an industry has succeeded abroad does not necessarily imply that it will also succeed at home. But this explanation can be consistent with the very hypotheses underlying the infant industry argument only if investors are not fully rational. This point is related to the idea that a country may not really know where its comparative advantage lies and we discuss this in greater detail in section 2D. What light has formal analysis shed on the simple infant industry argument? A seminal paper by Bardhan (1971) noted that the infant industry argument is dynamic in nature and that "any elaboration of this idea involves explicitly dynamic analysis, and it has hardly been integrated into the main corpus of trade theory which is mostly comparative-static in nature." Bardhan (1971) provided the first dynamic model of learning by doing in an open economy and derived the optimum extent and time path of protection to the learning industry. In his model, there are two goods c and m and two factors of production capital and labor with constant returns to scale in production of both goods. The learning effect is assumed to depend upon the cumulated volume of industry output in good m and it shifts out the production function for the good in a Hicks neutral fashion.1 Bardhan models learning by doing as a classic positive externality that depends 1Bardhan's model is in the spirit of the original learning-by-doing model of Arrow which posited learning that occurred in the machine producing sector. Some of the endogenous growth literature also posits such effects. However, much of the literature on technological innovation summarized in Evenson and Westphal 6 upon the size of the learning industry: the higher the cumulative output of the industry, the more productive the technology of each individual firm. When learning is unbounded, Bardhan shows that it is socially optimal to subsidize the infant industry and that the time profile of the optimal subsidy depends upon initial conditions ­ when the initial stock of experience is relatively small, the optimal subsidy decreases over time to a stationary rate otherwise it increases over time. However, the infant industry argument is predicated on the idea that the learning effect is temporary. To allow for this, Bardhan also considers optimal policy under the scenario where the learning effect is present only until cumulated industry output falls below a certain threshold. Under such bounded learning, the optimal subsidy is positive only till the threshold level of cumulative output is reached after which zero subsidy is optimal. Finally, Bardhan also considers the case where the foreign industry is also subject to learning and the domestic industry enjoys spillovers from foreign learning and shows that it might still be optimal to subsidize the domestic infant industry. However, his framework does not capture the idea that international spillovers may partially substitute for domestic learning since the learning effect function contains the stock of domestic and foreign outputs as separate arguments and the relationship between the two is not really considered.2 Succar (1987) extends Bardhan's (1971) analysis to allow the learning in one sector to generate spillovers for both sectors thereby providing an inter-industry spillover rationale for the infant industry argument. As might be expected, the optimal path of output subsidy in the presence of both intra and inter-industry learning by doing (LBD) economies lies above that when only the former are present. However, the presence of such economies is not sufficient to justify intervention. As Succar astutely notes, the discounted stream of productivity gains generated by LBD in the infant industry should outweigh the discounted stream of subsidies or else intervention would be socially undesirable.3 Furthermore, like Bardhan, Succar assumes that industry learning and firm learning are one and the same. Such need not be the case and firms may learn more from (1995) and Ruttan (2002) shows that learning can occur in all sectors, a fact that would enormously complicate the results of much of the literature. 2Pack and Saggi (2001) explore the implications of the provision of free technology by the purchasers of a firm's exports, a further complication. 7 their own experience relative to that of other firms. Of course, is this is true then models of perfect competition would have to be abandoned ­ once a firm realizes that its own output lowers its future costs, price taking behavior cannot be assumed since the firm would never be in equilibrium. But it is worth noting that the presence of imperfect competition does not necessarily weaken the argument for infant industry intervention ­ firms with market power typically produce too little output and an even larger subsidy may be called for to correct the underproduction problem. According to Succar (1987), the presence of inter-industry LBD economies considerably strengthens the case for intervention in favor of infant industries and there can be an irreversible locational advantage in producing goods characterized by such LBD economies domestically. The intuitive idea underlying Succar's model is that the production of capital goods can enhance growth by acting as an "informal learning center where technical skills are required" thereby contributing to a country's technical infrastructure.4 Such improvements in the skill base of workers complement investments in human capital and can help further the industrialization process in developing countries. The distinction between firm and industry level LBD becomes quite important when one considers the fact that firms are heterogeneous in nature. Suppose some firms are more efficient at learning than others. Under such a scenario, optimal subsidies would necessarily have to be non-uniform and the government is unlikely to possess the information needed to implement an optimal subsidy program. Given the information problem, it might make sense for the government to adopt a uniform policy even though it may not be first best. While in theory one can design mechanisms that result in firms revealing their learning capabilities but the practical relevance of such mechanisms is far from clear. As one might expect, there is more to the infant industry argument than the `simple version' formalized by Bardhan (1971) and Succar (1987). As Baldwin notes, there are four versions of the infant industry argument that are a bit more nuanced: (a) 3It is likely that this criterion has been satisfied by the European Airbus effort, widely considered a major example of a successful industrial policy. However, one also needs to account for the cost of distortions that are generated by the taxes needed to finance the subsidies paid. 8 acquisition of knowledge involves costs but yet knowledge may not be appropriable by an individual firm (i.e. knowledge spillovers may discourage investment in acquisition of knowledge) - this is the standard argument for subsidizing R&D; (b) firms may provide costly on the job training but may not be able to prevent the diffusion of such knowledge via movement of workers (i.e. there might be a free-rider problem in worker training) - while firm specific training involves no potential externality, general training can lead to externalities that would justify subsidies; (c) static positive externalities in the production of a good may justify trade protection and (d) determining profitability of a new industry might require a costly investment the results of which may become freely available to potential competitors ­ in other words, investment into new industries might result in informational externalities that make it difficult for investors to earn a rate of return that is high enough for the initial investment to be justified. This is precisely the argument that has been formalized by Hausmann and Rodrik (2003) although they call it the process of `self-discovery' ­ i.e. the process of determining what you can produce profitably at world prices. The simple version of the infant industry argument does not really specify how learning occurs ­ i.e. it just assumes that dynamic scale economies somehow will be realized by the infant industry. Of course, learning is rarely exogenous and it usually requires considerable effort and investment on the part of firms (Pack and Westphal, 1986). If such investments are to be made, firms need to be able to appropriate the benefits of the knowledge gained. As is well known, knowledge is a non-rival good and, once created, any number of agents can use it simultaneously. If firms cannot prevent the leakage of knowledge that is costly to create, then they will have little incentive to create such knowledge in the first place. In other words, property rights over knowledge may not be enforceable and this can create a rationale for government intervention (this is one reason why we have intellectual property rights protection). As Baldwin notes (1969), many types of knowledge acquisition are not subject to the externality described above since entrepreneurs can often prevent the leakage of their knowledge to potential competitors. Similarly, if there are only a few firms in the 4Succar's emphasis on the capital goods sector is similar in spirit to Arrow's learning by doing model and endogenous growth models such as Romer's (1986) model that employed it as a building block. 9 industry, inter-firm negotiations should help offset the externality problem (see Coase, 1960). But what if many rival firms benefit from the investment undertaken by a knowledge acquiring firm and nothing can be done to prevent such diffusion? Is government intervention justified then? As is clear, trade protection is certainly not called for ­ a tariff does nothing to solve the basic externality problem. In fact, trade protection may very well worsen the problem. As Baldwin (1969) notes: "...the speed at which firms respond to market opportunities is a function of the level of profit prospects. A duty will make it worthwhile for firms to incur the costs of acquiring the knowledge discovered by other firms (if it is not completely free) faster (italics added) and also to move into production more rapidly..." A production subsidy to the entire sector will also fail to remedy the externality. What is needed are subsidies to initial entrants into the industry that help create new knowledge and discover better production technologies. As in the case of R&D subsides, governments should target the marginal rather than inframarginal research. In the case of new firms, it takes time to discover whether a new idea or technology is socially valuable and the adoption of a novel technology by others is in fact the strongest proof of its social value. Thus, such a policy of rewarding early entrants requires an accurate forecast of the social value of their inventions and discoveries ­ a process that can be fraught with failure. Not only that, given the uncertainty associated with new technologies, a delayed pattern of adoption might even be socially optimal. B. Knowledge spillovers, dynamic scale economies, and industrial targeting One of the most powerful ideas in all of economics is the notion of comparative advantage. Ever since David Ricardo, it has been well known that under free trade a country can increase its national income (and welfare) by moving resources into sectors in which its opportunity cost of production is lower than its trading partners. But is this prescription sufficient to generate economic growth? Perhaps not. Allocating resources according to comparative advantage can only ensure static efficiency and in no way guarantees dynamic efficiency. Succar (1987) argues "...the comparative advantage theory is a static construct that ignores forward linkages exist between present choices and future production possibilities. Therefore it cannot guide the pattern of international specialization when there are asymmetric learning opportunities associated with the 10 production of different goods and/or use of certain techniques. Promotion of industries which generate substantial learning by doing economies should be an integral part of a strategy of human capital formation in LDCs." In other words, Succar argues for some sort of industrial targeting although her model does not explicitly deal with this issue. Even if one accepts the premise that certain industries are more likely to generate spillovers (based on knowledge diffusion or other factors), can policy be designed to encourage the `right' industries? The ideal but rarely attained goal of industrial policy is the development of a general-purpose technology. DARPA, a small unit within the U.S. Department of Defense that generated and financed a portfolio of projects, is widely credited with having been the key contributor to the development of the internet, the demand for this innovation being derived from the need to maintain communications during an assault on the U.S. This breakthrough was clearly fundamental and has social benefits many-fold the cost of the DARPA effort. This instance of success addressed a market failure, namely, the social benefits of research were much larger than the anticipated private benefits. Moreover, DARPA foresaw a potential need that may have escaped the purview of private firms. While the internet was a major technological breakthrough and suggests the potential gains from such an activity, it is useful to remember that, by their very nature, the discovery of such "general purpose technologies" is a rare event. The informational constraints facing policy-makers pursuing industrial policy are severe and any realistic model of industrial targeting needs to account for them. In a recent paper, Klimenko (2004) models industrial targeting as an optimal experimentation strategy of a government that lacks information about the set of industries in which the economy has comparative advantage with respect to rest of the world. He examines the set of industries in which a country will specialize as a result of such policy. In his model, for any set of targeted industries, it is possible to know whether or not a country will specialize in this set with positive or zero probability. He shows that an optimally designed industrial policy can actually lead a country to specialize in sectors in which it does not have comparative advantage. The key issue is the beliefs of the policymaker. Depending on the nature of such beliefs, a country can end up abandoning the industries in which it has "true" comparative advantage. 11 Furthermore, Klimenko argues that the policy-maker may stop looking for better targets when the favored industries perform well enough. He interprets this outcome as a failure of industrial targeting policy even though it may not appear as such. In fact, he goes on to show that, despite the existence of market failures, the outcome of the learning process through private experimentation (without any assistance from the government) can even yield outcomes that are closer to the full information social optimum. Klimenko's rigorous analysis of this issue underscores our intuitive argument that the relevant counterfactuals are unavailable and what may appear to be a successful industrial policy may not be the first best outcome from a country's perspective ­ merely doing something well need not imply one cannot be better at something else. While Klimenko's model considers the case where the government does not have information about the industries in which the country has comparative advantage, Dinopoulos et. al. (1995) examine industrial targeting when the government lacks information about the local industry's learning curve and provide a framework that synthesizes industrial targeting with learning by doing of the type analyzed by Bardhan (1971) and Succar (1987). More specifically, in their model a domestic monopolist faces a competitive foreign fringe that has exhausted all learning economies. While the current production cost of the domestic firm is assumed to exceed that of the foreign fringe, the future relative cost of the domestic firm can be lower if its learning curve is steep enough. They show that under such circumstances, the optimal local policy is a subsidy to the domestic firm that increases with the amount by which the domestic firm's price, post- learning, falls below the cost of the foreign industry. An intriguing finding of the model is that the optimal policy provides a subsidy to the domestic firm precisely when it would learn a sufficient amount to actually drive the foreign fringe from the domestic market ­ under such a scenario, despite the fact that a domestic monopoly results, the price in the domestic market is below the competitive price set by the foreign industry. The idea is that a monopolist is preferable to a competitive industry if its price falls below that of the competitive industry's cost (price). But there is a difficulty with this argument -- given the option of exporting, why is the price lower in the domestic market? C. Coordination failures as a rationale for industrial policy The basic idea behind the coordination failure argument for industrial policy is 12 that many projects require simultaneous investments in order to be viable and if these investments are made by independent agents there is little guarantee that, acting in their own self interest, each agent would choose to invest. As Scitovsky (1954) noted, reciprocal pecuniary externalities in the presence of increasing returns can lead to market failure because the coordination of investment decisions requires a signaling device to transmit information about present plans and future conditions and the pricing system is not capable of playing this role. Pack and Westphal (1986) argue that such pecuniary externalities related to investments in technology are pervasive in industrialization. They provide an example of two infant industries (say A and B) where industry A produces an intermediate that is required in industry B and neither industry is profitable if it is established alone. However, if both industries are established together, then both are profitable implying that it is socially optimal to indeed establish both. Of course, the problem is that without explicit coordination between investment decisions this outcome would not be obtained. Okuno-Fujiwara (1988) presents a formal model of such interdependence between industries and the coordination failure that can result from such interdependence. He considers an economy with three goods: x, y, and z where good z serves as a numeraire and is produced under perfect competition with constant returns to scale. Labor is the only factor of production and the price of good z equals the wage rate. Good x is produced by a competitive industry and it requires good y as an intermediate. The technology for good y exhibits large economies of scale and the industry is assumed to be oligopolistic where the number of firms is endogenously determined to ensure zero profits in equilibrium. A coordination problem arises in the industry because the derived demand for the intermediate good y depends upon its price, which in turn determines incentives for entry into the intermediate sector. If y-producers anticipate low demand for their good, given the fixed costs of entry, few of them would want to enter implying a higher price for the intermediate which may then make industry x unsustainable ­ the key assumption here is that the intermediate good y must be locally supplied. On the other hand, if y producers are assured of a high demand for their product, more of them would enter and such entry would lower its price which would then allow the high demand for 13 the intermediate to be sustained.5 Okuno-Fujiwara (1988) shows that there is no unique equilibrium in a small open economy with the above production structure. In the bad equilibrium, the economy ends up specializing in good z where in the good equilibrium it produces both goods x and y and exports good x to the rest of the world (where the latter equilibrium Pareto dominates the former). Turning to policy analysis, Okuno-Fujiwara (1988) suggests that three types of traditional government intervention can help ensure that the good equilibrium is realized: (a) the government can provide a production subsidy to either x or y industry (or both) thereby causing the two sectors to expand or (b) provide an export subsidy to the x sector; or (c) shut off international trade. However, he notes that trade protection can be effective only if the autarkic equilibrium production of good x is sufficiently large ­ something that is less likely to be true of small developing countries. In addition to traditional industrial policies (a) and (b), Okuno-Fujiwara (1988) also suggests that the government can play a coordinating role between x and y producers by facilitating information exchange between them: "...a government may collect information about planned production of the x-good and planned demand for the y-good depending upon various price levels of the intermediate good. It then collects information about planned production of the y-good. Like planning procedures of centralized economies, a government may continue such information exchange until it can obtain a consistent plan of the entire economy."6 He is not arguing for a one-time information exchange but rather repeated exchanges that help resolve the coordination failure. It is difficult to believe that policy- makers can effectively execute such information exchange between disjoint industries about whose day-to-day business they may know very little. Furthermore, since coordination failures are likely to be pervasive, the above policy prescription suggests a massive role for government intervention in the process of industrialization. Okuno- Fujiwara himself is skeptical whether the mechanisms captured by his model and the policy prescriptions that emerge from his analysis had any practical analog in the 5As will be discussed below, good x could be produced by multinationals that establish local production, thus obviating the coordination problem. 6Much of the effort of MITI and the Ministry of Finance in Japan can be described as the interchange of information among firms and the interaction with the government to reduce any obstacles to the realization of consistent plans. The same is true of French indicative planning of the 1950s and 1960s. As noted 14 Japanese experience. In a paper along the lines of Okuno-Fujiwara (1988), Rodrik (1996) argues that for coordination failures between upstream and downstream industries to exist, it is necessary that there be some type of scale economies in production and that imperfect tradability holds across national borders of some of the goods, services, or technologies associated with manufacturing. Of course, as noted above, both of these conditions are met in Okuno-Fujiwara (1988)'s model. But Rodrik's analysis adds value in two respects. First, in his model the intermediate good sector is characterized by monopolistic competition rather than oligopoly. Second, he suggests an interesting interpretation of the nontradable intermediate goods sector: "...my preferred interpretation is to think of these inputs as representing different categories of specialized skill labor." The idea is that a worker's decision to acquire any skill depends upon demand for that skill and it is indeed quite costly or simply infeasible to import labor services should certain skills be in short supply locally. In Rodrik's model, there are two final goods sectors: the high-tech sector requires a range of intermediate inputs that are produced under increasing returns to scale and are non-tradable (as in Ethier, 1982) whereas the low-tech sector uses only labor and capital. For the high-tech sector to be viable in an economy, it needs to produce a sufficiently large number of intermediate inputs. As a result, if the entry decisions of intermediate goods producers are not coordinated, the economy can easily get stuck in an equilibrium where only the low-tech good is produced even though the high-tech good is profitable.7 The problem is that no intermediate goods producer has an incentive to begin production unless a relatively large number of intermediates are already being produced. If one takes Rodrik's interpretation of specialized intermediates as being some types of skilled labor, the coordination problem may be less intractable than he suggests. For example, in some countries there has been a cycle of emigration and repatriation of those who are highly educated as happened in India's software sector, discussed below, as well as Korea and Taiwan (China).8 In all three cases many students acquired more advanced earlier, it is difficult to assess whether such sector specific targeting was successful. For an extensive review of the empirical evidence on Japan, see Noland and Pack, 2003. 7A similar argument has also been made by Rodriguez-Clare (1996b). 8A detailed study of the Silicon Valley-Taiwan (China) nexus is provided by Saxenian (2001). 15 education than could be used at the time of their graduation, emigrated and later played a critical role in the expansion of their respective countries. The synchronization of education and the demand for it is almost certainly impossible to plan but recognizing the role of emigration and potential return mitigates the problem though clearly the first best solution of harmonization would be more desirable especially in light of potential permanent residence abroad. It is worth noting, that like Okuno-Fujiwara (1988), Rodrik (1996) is quite hesitant to offer strong policy recommendations based on his analysis and he concludes that government intervention designed to resolve such coordination failures "must be judged a risky strategy". Thus the World Bank's (1993) well-known report on the East Asian miracle argues that East Asian efforts to coordinate investment decisions led to a number of inefficient industries. While the theoretical rationale for redressing coordination failure appears to be sound, the argument rests on certain key assumptions, particularly that the organization of production activity is exogenously given. Why would industries whose profitability is so intimately intertwined not find ways to help coordinate decisions as is the case of the many international supply networks (Gereffi and Memedovic, 2003, Sturgeon and Lester, 2002, 2003)? For example, vertical integration between intermediate and final goods producers can help resolve some coordination problems although there are clearly limits to the extent to which organizations can adjust their scale and scope in order to solve coordination problems. At some point, all firms have to interact with others via the market. But long-term contracts between firms have been used to solve problems of relation specific investments in many industries. It is not clear why contracts cannot play the same role in the context of coordination failures. Perhaps the biggest problem with the coordination failure argument is that it relies heavily on the assumption of non-tradable intermediate inputs, partly reflecting the fact that much of the early literature was based on the example of steel and autos circa 1960 rather than the products in which transportation costs for the intermediate are likely to be low. Virtually all of the models make this assumption despite the fact that the majority of international trade is in intermediate goods. Thus, the coordination failure argument runs up against the central fact around which much of the `new' trade theory has been built. 16 This is no small contradiction and if the coordination failure story has to be rescued it needs to appeal to nontradable services as in Rodriguez-Clare (1996a). But the problem then is that the case for industrial policy on the basis of coordination failures is quite thin if inward foreign direct investment (FDI) is feasible and/or permitted. If local firms do not produce sufficient number of intermediates due to coordination failures, why can't they be produced by foreign multinationals that are surely not dependent upon the production structure of any one economy? In small developing countries, a large-scale investment by a multinational can create sufficient demand for intermediates and easily resolve the coordination problem. In fact, this is partly what the literature on backward linkage effects of FDI argues (see Markusen and Venables, 1999 and Rodriguez-Clare, 1996a). It is quite unlikely that multinational firms would be hostage to the type of coordination problems that confront small producers in developing countries. Indeed, the huge growth in the importance of international supply chains established by MNCs has become one of the most visible features of industrial growth in the last decade (Sturgeon and Lester, 2002). In section 2E we further discuss the role multinational firms can play in determining the overall case for industrial policy. D. Informational externalities In a recent paper, Rodrik (2004) has argued that the traditional view of industrial policy (based on technological and pecuniary externalities) is out-dated and does not capture the complexities that characterize the process of industrialization. According to Rodrik (2004), industrial policy is more about eliciting information from the private sector than it is about addressing distortions by first-best instruments. He envisions industrial policy as a strategic collaboration between the private and public sectors the primary goal of which is to determine areas in which a country has comparative advantage. The fundamental departure of this viewpoint from classical trade theory is that entrepreneurs may lack information about where the comparative advantage of a country lies. Or more to the point, at the micro level, entrepreneurs may simply not know what is profitable and what is not. In the presence of informational externalities, a free rider problem arises between initial investors and subsequent ones. Suppose no one knows whether activity x is profitable or not and the uncertainty can only be resolved by making a sunk investment 17 that cannot be recovered in case the outcome turns out to be unfavorable. If there is free entry ex post, no entrepreneur may be willing to make the investment required to discover the profitability of activity x: if someone does make the investment and the activity turns out to be profitable, other entrepreneurs will be attracted to the same activity thereby eliminating all rents. It is worth noting that Baldwin's (1969) classic paper anticipates Rodrik's argument almost exactly. He wrote "...suppose, for example, that a potential entrant into a new industry, if he could provide potential investors with a detailed market analysis of the industry, could borrow funds from investors at a rate that would make the project socially profitable. However, should this information become freely available to other investors and potential competitors, the initial firm might not be able to recoup the cost of making the market study....under these circumstances the firm will not finance the cost of the study, and a socially beneficial industry will not be established." Similarly, in the context of adoption of high yielding varieties of crops by farmers in developing countries, Besley and Case (1993) note that late adopters may learn from early adopters when "a technology is of uncertain profitability, some potential adopters may wait until they observe whether others have fared well by using it" and that such "externalities are potentially important in agricultural technology adoption." Given the importance of this argument for the debate on industrial policy, it is useful to consider the framework presented in Hausmann and Rodrik (2003) in some detail. They consider a small open economy comprised of two sectors: traditional and modern. Labor can move freely across the two sectors. The production technology in the traditional sector is constant returns to scale and the presence of a fixed factor generates diminishing returns to labor. In the modern sector (that consists of many goods), there are constant returns to scale in production but the cost of production of a good depends upon an unobserved productivity parameter (i) that becomes known only when the production of a good is attempted (something that requires a time period in which resources must be utilized but no production takes place ­ this is what Baldwin called a `market study'). The basic idea here is that entrepreneurs lack information about the profitability of production of various goods in the modern sector and this information can be obtained only by undertaking a sunk investment. After uncertainty regarding i is resolved, entrepreneurs compare their production 18 costs with world prices and produce those goods for which they make monopoly profits (which accrue for length of time T -- call this the monopolization period). Of course, once information becomes public (which it does in period three when the monopolization period has elapsed) there is further entry (into goods that yield positive profits) until all profits are competed away to zero. Hausmann and Rodrik (2003) analyze the laissez-faire equilibrium of the above model and compare it to the social planner's problem in order to derive the market failures that result from the presence of informational externalities. They argue that the market equilibrium is deficient in two respects. First, the level of investment and entrepreneurship delivered by the market does not coincide with the social optimum because the entrepreneurs care only about profits and not about economy-wide benefits of their investment. It is notable that if the monopolization period is very long, the market economy can actually deliver too much investment in the modern sector as opposed to too little. This suggests that in economies where firms face substantial entry barriers, the underinvestment problem noted by Hausmann and Rodrik (2003) is not likely to be serious. For example, the industrial licensing regime pursued by India during the first forty or so years after independence made it quite difficult for firms to enter new markets. And the recent literature on the business climate that emphasizes other factors that discourage investment such as the time it takes to obtain business permits, telephone lines, and other utility hookups further discourages excessive investment in the modern sector (World Bank, 2004).9 Such barriers should have helped protect rents for those that did manage to enter profitable markets. The second market failure identified by Hausmann and Rodrik (2003) is that the market equilibrium yields too little specialization ­ all activities that turn out to be profitable are sustained whereas optimality requires that only the one with the highest return be pursued. In other words, in their model, while it is optimal for the small open economy to only produce the good for which the profit margin is the highest, the market solution allows all those that make positive profits to stay in business during the period of 9If there is concern about excessive investment, some aspects of the adverse business environment may unwittingly be a second best policy. 19 monopolization. This result reflects the general equilibrium nature of their model and the fact that they consider a small open economy. To see this, first note that the modern sector draws resources out of the traditional sector and optimality requires that these resources be utilized where they generate the largest profits (which happens in the modern good for which the productivity parameter (i is the highest). Second, since the country's output of a good does not affect world price, one can never have a situation where the mark-ups across different goods are equalized. Clearly, if world prices changed with a country's exports/output, complete concentration in the modern sector need not obtain. A more likely scenario would be that a country should produce higher quantities of modern goods for which it has a more favorable productivity draw and lower quantities of other goods. Hoff (1997) argues that if initial producers benefit subsequent producers, the case for subsidizing initial producers hinges very much on the assumption that the externalities operate in a deterministic fashion (i.e. do not involve any uncertainty). She constructs a model where initial entrants provide information that is socially valuable by reducing uncertainty for potential followers regarding production conditions. In her model, factors that increase the informational barrier to entry can actually imply a lower optimal subsidy for the infant industry. By contrast, in most existing models, the externalities are assumed to remove all uncertainty as opposed to reducing it. Since Hoff's model is clearly more realistic, it is notable that her results weaken the case for subsidizing an infant industry. E. The international dimension: role of exports and FDI For small developing countries, the case for industrial policy is rarely a purely domestic one. International considerations are fundamental in many respects but the role of exports (on the part of domestic firms) and inward foreign direct investment (FDI) has received considerable attention. A potential rationale for industrial policy in the context of exports arises when product quality is unknown to foreign consumers. The informational asymmetry can lead to market failure that can then potentially justify some form of intervention. To make the argument concrete, we consider below some of the more influential models of industrial policy based on asymmetric information in the context of exports. Mayer (1984) argues that when foreign consumers lack information about quality, an industrial policy of 20 subsidizing exporters can indeed be a first-best policy. In his two-good model, foreign consumers can discern the quality of good one only after purchasing and consuming it whereas the quality of the second good can be determined through inspection alone (if the first good has been consumed). The intuitive idea behind this assumption is the firms that produce good one generate positive externalities for those that produce good two by `cultivating' the foreign market. Mayer's view could underlay the important role assigned by many analysts to publicly supported trading companies in Japan and the Republic of Korea (Lall and Keesing, 1992). However, this conclusion has been criticized by Grossman and Horn (1988) since in Mayer's model firms do not choose the quality of their products (i.e. it is exogenous) and consumers have pessimistic expectations that are adjusted upwards with experience. By contrast, Grossman and Horn (1988) assume that firms choose the quality of their products and that consumers form expectations regarding quality using available information and being cognizant of the decision problem facing firms ­ i.e. in their model, consumers have rational expectations. Furthermore, the model of Grossman and Horn (1988) permits free entry and they focus on the domestic market where quality of foreign goods is known whereas the domestic industry is an infant. However, their model can easily be interpreted as one of export promotion of a domestic infant industry. The modeling differences would not be terribly important were it not for the fact that the more complete model of Grossman and Horn (1988) seems to completely overturn the case for a subsidy when an industry faces informational barriers to entry. They note that adding an explicit process of reputation acquisition leads to the result that infant industry protection actually reduces welfare despite the presence of information asymmetry. The logic is as follows. When consumers have rational expectations, they pay a price that leaves them indifferent between importing and buying the average domestic product. Since the marginal entrant produces goods of lower quality than the average firm, new entry provides negative social surplus. Furthermore, temporary protection does not alter the incentives for reputation acquisition since it benefits all firms. A major reason Grossman and Horn's (1988) results differ from those of Mayer (1984) is that they focus on reputation acquisition at the firm level whereas Mayer 21 focuses at the country level ­ i.e. in the Grossman/Horn view, Toyota can affect only its own reputation in foreign markets whereas in Mayer's model, the experience with Toyota also determines how foreign consumers view other Japanese companies such as Honda. The difference matters because returns to reputation acquisition are appropriable in the Grossman and Horn model whereas they are not in the Mayer model. It is worth noting that the conclusions of Grossman and Horn (1988) are not the ones reached by Bagwell and Staiger (1989) who argue that if asymmetric information blocks the entry of high quality firms, export subsidies can improve welfare by breaking the entry barrier facing high quality firms. Thus, whether or not an export subsidy is desirable hinges very much on the nature of the distortion that is caused by the presence of asymmetric information.10 Now we discuss how an argument for industrial policy might arise in the context of FDI. Policy intervention with respect to FDI has a long history and the rationale for such intervention has frequently been the effects of FDI on productivity of local firms via technology transfer as well as linkage effects. The literature on FDI, technology transfer, and linkages has been surveyed extensively in Saggi (2002) and here we confine ourselves to those aspects of FDI that relate intimately to local industrial development and its linkage effects since these correspond quite well to the coordination failure rationale for industrial policy. There exists a voluminous informal as well as empirical literature on backward linkages. For example, the 1996 issue of the World Investment Report was devoted entirely to the effects of FDI on backward linkages in host countries. However, analytical models that explore the relationship between multinationals and backward linkages in the host country are hard to come by. Two prominent examples of such models are Markusen and Venables (1999) and Rodriguez-Clare (1996a). Both these studies provide important insights regarding the two-way relationship between multinationals and linkages. In the models of Rodriguez-Clare (1996a) and Markusen and Venables (1999) the intermediate goods sector is monopolistically competitive so that the effects of foreign investment occur via altering the incentives for entry into such markets. In both models, Ethier's 10See also Raff and Kim (1999) for further investigation of the export subsidy result in the presence of oligopoly into which there is no free entry. 22 (1982) formulation of the so-called love-of-variety production function for final goods is at the heart of the interaction between multinationals and local suppliers. These models emphasize the demand-side effects of multinationals' entry on the host economy: multinationals generate derived demand for intermediate goods, thereby promoting industrial development of the intermediate goods sector in the host country. This demand creation effect exists in both Rodriguez-Clare (1996a) and Markusen and Venables (1999). In addition, the Markusen and Venables model also allows for a competition effect wherein the entry of a multinational hurts its local rivals. In Rodriguez-Clare (1996a), the host country is assumed to be in a `bad' equilibrium where the final good is produced only by multinationals. As a result, the competition effect is absent in his model (which has substantial richness along other dimensions). The key trade-off underlying Rodriguez-Clare's model is as follows. On the one hand, since multinationals produce complex goods, they tend to have a higher demand for specialized intermediate inputs per unit of labor relative to domestic firms. On the other hand, by assumption, only multinationals have access to intermediates produced abroad and so they demand only a fraction of required intermediates from the host country. As a result, multinationals enhance backward linkages only when it is relatively expensive to source from abroad.11 As noted earlier, common problem with analytical models in this area (as well those dealing with coordination failures) is the assumption that intermediates are nontradable. These models shut off trade in intermediates and then use FDI as the channel that either provides some of those intermediates or increases demand for local intermediate goods producers (since multinationals are not allowed to import intermediates by assumption). As a result, they are likely to overstate the impact of multinationals on industrial development. Suppose one accepts the optimistic view regarding the effects of FDI (indeed the evidence discussed below suggests that there are reasonable grounds for doing so). Does this have implications for industrial policy? Our answer is a qualified yes. Basic 11The focus on domestic demand for intermediates is appropriate only if the MNC exports and therefore it can be assumed that the intermediates are supplied efficiently. If production is destined for the domestic market and the final goods sector is protected, local suppliers, especially if there are local content 23 economic theory tells us that it is optimal to subsidize an activity if it generates positive externalities -- i.e. the activity benefits agents other than those directly involved in the activity itself. The potential for positive externalities from FDI surely exists and available evidence exists that often this potential is realized. Incentives to attract FDI may be justified on the grounds of such externalities from inward FDI. However, such policies are not typically what proponents of industrial policy have in mind ­ indeed the thrust of such arguments is typically in favor of encouraging the development of indigenous firms. It is worth keeping in mind that investment incentives and tax breaks to multinational investors work against their local competitors. Thus, if there exist local firms that could potentially compete with multinationals, the adverse effect of tax incentives to multinationals on such firms needs to be taken into account. The efficacy of investment incentives is also dubious ­ such policies can end up transferring rents to foreign investors without affecting their investment decisions. One final insight from the literature on linkage effects of multinational firms is worth mentioning. As Rodriguez-Clare (1996a) notes, we should "...expect multinationals to generate more linkages when they come from regions that are farther away." The implication for industrial policy with respect to multinationals is clear ­ if a country does want to attract FDI, it ought to focus on firms that come from far off regions and are interested in serving the local market. In a recent paper, Barrios et. al. (2004) construct a model wherein the competition effect generated by a multinational is eventually dominated by the positive externalities it generates. Using plant level panel data from the manufacturing sector in Ireland (a country whose economic development has been influenced greatly by multinational firms) they show that such a model indeed describes the Irish experience with FDI. Alfaro and Rodriguez-Clare (2003) use plant level data from several Latin American countries and they make the important point that many empirical studies testing existing theoretical models often use inappropriate measures to evaluate the linkage effects of multinationals. More specifically, empirical studies often use the share of inputs purchased locally by a multinational to measure its impact on linkages. They argue that requirements, may have costs above world prices, raising the possibility that greater linkages imply lower GDP per capita. 24 the proper measure (as implied by theory) is the ratio of value of inputs bought domestically to the total workers hired by a multinational. The distinction between the two types of measures is important because multinationals typically source a lower percentage of their inputs locally relative to their local competitors. However, it does not imply that their linkage effects are necessarily negative since their production techniques might require more inputs in relation to the workers they hire. In their empirical work, Alfaro and Rodriguez-Clare find that the linkage coefficient of multinationals is actually higher than that of local firms in Brazil, Chile, and Venezuela whereas it is no different (statistically) in Mexico. Recent case-study evidence regarding the linkage effects of FDI is provided by Mexico's experience with the maquiladora sector and its automobile industry. Mexico started the maquiladora sector as part of its Border Industrialization Programme designed to attract foreign manufacturing facilities along the US-Mexico border. Most maquiladoras began as subsidiaries of US firms that shifted labor-intensive assembly operations to Mexico because of its low wage relative to the US. However, the industry evolved over time and the maquiladoras now employ sophisticated production techniques, many of which have been imported from the US. In certain key respects, Mexican experience in the automobile industry is quite representative of its overall industrial development. In keeping with the overall strategy of import substitution, during the 1960s and 1970s, Mexico imposed domestic content requirements on multinationals in the automobile industry. These requirements were less stringent than those imposed in Brazil: whereas Brazil required car manufacturers to have a domestic content of over 90%, Mexican policy imposed a domestic content requirement of "only" 60%. Incidentally, another important difference between Mexican and Brazilian policies was that Mexico did not allow foreign firms to vertically integrate with their local suppliers while Brazil did. The goal of this nationalistic stance was to ensure that domestic firms captured the benefits generated by the backward linkages of FDI. An interesting aspect of Mexico's export performance requirement in the automobile industry (as noted by Moran, 1998) was that, unlike countries such as Malaysia, Mexico did not require that foreign firms export a particular product (such as a finished car) but only that the value of exports be a specified ratio relative to imports. 25 Clearly, such a policy let the car companies decide what to export and what not, leaving them free to make their own calculations based on comparative advantage considerations.12 The export performance of the industry has improved remarkably in recent years: between 1990 and 1998, the share of exports in output has increased from 3.7% to a remarkable 68.6%. Mexico's experience in the automobile industry is also illustrative of how FDI can contribute to industrial development in the host country though Mexico's favorable experience was facilitated by the NAFTA agreement (Laderman, Maloney, and Serven, 2003). Initial investments by US car manufactures into Mexico were followed by investments not only by Japanese and European car manufacturers but also by firms who made automobile parts and components. As a result, competition in the automobile industry increases at multiple stages of production thereby improving efficiency. Such a pattern of FDI behavior (i.e. investment by one firm was followed by investment by others) probably reflects strategic considerations involved in FDI decisions. Most multinational firms compete in highly concentrated markets and are highly responsive to each other's decisions. An important implication of this interdependence between competing multinationals is that a host country may be able to unleash a sequence of investments by successfully inducing FDI from one or two major firms. However, the concentration of inward FDI into a handful of LDCs suggests that only a few countries can benefit from this process ­ Tanzania and Egypt are not China. Extensive backward linkages resulted from FDI in the Mexican automobile industry: within five years of investments by major auto manufacturers, there were 300 domestic producers of parts and accessories, of which 110 had annual sales of more than a million dollars. As per Peres Nunez (1990), multinationals in the Mexican automotive sector conducted production audits, held weekly coordination meetings, and provided technical training to their suppliers. Foreign producers also transferred technology to such domestic suppliers: industry best practices, zero defect procedures, production audits etc. were introduced to domestic suppliers thereby improving their productivity and the quality of their products. As a result of increased competition and efficiency, Mexican exports in the automobile industry boomed. 12Such a policy largely, but not entirely, precludes the adverse effects cited in the previous footnote. 26 A recent comprehensive case study of the effects of Intel's investment in Costa Rica by Larrain et. al. (2000) finds evidence that local suppliers benefited substantially from Intel's investment. Similar evidence exists for other sectors and countries and such evidence is discussed in great detail in Moran (1998 and 2001). For example, in the electronics sector, Moran (2001) notes that in Malaysia, foreign investors helped their local subcontractors keep pace with modern technologies by assigning technicians to the suppliers' plants to help set up and supervise large-volume automated production and testing procedures. In a broader study, Batra and Tan (2002) use data from Malaysia's manufacturing sector to study the effect of multinationals on inter-firm linkages and productivity growth during 1985-1995. Their results show that not only are foreign firms more involved in inter-firm linkages than domestic firms but also that such linkages are associated with technology transfer to local suppliers. Such technology transfers were found to have occurred through worker training and the transmission of knowledge that helped local suppliers improve the quality and timeliness of supply. 13 Javorcik (2004) examines backward linkages and technology spillovers using data from the Lithuanian manufacturing sector during the period 1996-2000. She finds that firm productivity is positively affected by a sector's intensity of contacts with multinational customers but not by the presence of multinationals in the same industry. Thus, her results support vertical spillovers from FDI but not horizontal ones. Furthermore, she finds that vertical spillovers are realized only when the technological gap between domestic and foreign firms is moderate. Blalock (2001) uses a panel dataset from Indonesian manufacturing establishments to check for the same effects. He finds strong evidence of a positive impact of FDI on productivity growth of local suppliers showing that technology transfer from multinationals indeed takes place. He also plausibly suggests that since multinationals tend to source inputs that require simple technologies relative to the final products they produce, local firms that manufacture such intermediates may be in a better position to learn from multinationals than those that compete with them. 13Pack and Saggi (2001) analyze the willingness of purchasers to provide such technology transfer given the potential sales to competitors of the original purchaser. 27 This review of arguments for industrial policy indicates the enormous difficulties of implementation of industrial policies. The range and depth of knowledge that policy makers would have to master to implement a successful policy is extraordinary. They would have to understand the relevance of, and be accurately informed about, a huge range of complex questions and have the ability to accurately evaluate very subtle differences. A subset of the issues on which policy makers would have to be knowledgeable derived from the preceding discussion includes: 1. whether consumers learn the quality of a good only after consuming rather than inspecting it; 2. whether firms that are trying to reduce production costs also begin a simultaneous effort to improve their product's quality to obtain a better reputation. 3. which firms and industries generate knowledge spillovers 4. which firms and industries benefit from dynamic scale economies ­ what is the precise path of such learning and the magnitude of the cost disadvantage at each stage of the learning process 5. which sectors have a long-term comparative advantage 6. knowledge of the size of scale economies of different firms and sectors in order to facilitate investment coordination 7. the potential effects of FDI or international trade in solving some of the coordination problems, including a detailed knowledge of which of tens of thousands of intermediates are tradable 8. a better sense than individual firms possess about their potential competitiveness 9. the nature and extent of capital market failures 10. the magnitude and direction of inter-industry spillovers 11. the relative amount of learning by individual firms from others and from their own experience 12. the extent to which early entrants generate benefits for future entrants 13. the extent of heterogeneity of firms' learning abilities 28 14. a forecast of which firms can create new knowledge and discover better production methods. 15. the spillover effects of FDI as well as the likely intensity of their purchase of domestic intermediates It is possible that government officials might be this omniscient but the performance of the portfolio managers in developed country stock markets suggests that few of the very well trained (and remunerated) equity analysts can evaluate even much more certain and grosser characteristics of existing firms and industries with long track records. Nor do industrial firms themselves have the ability to successfully forecast such developments. Acknowledging that a first best policy would argue for the government to address such market failures or externalities, the task is daunting. Quite apart from the dangers of optimal policy being subverted by industries and firms that would benefit, the sheer knowledge and skill requirements would exceed that possessed by almost any institution including the best consulting firms. On a far more circumscribed set of tasks, measuring and explaining the sources of lower total factor productivity for a small number of sectors in South Korea and Brazil relative to the United States, McKinsey & Co., a preeminent consulting firm spent several years and employed dozens of people whose qualifications exceed those possessed by officials in almost any developing country (McKinsey Global Institute, 1998a, 1998b). No study has attempted to assess whether governments have been successful in mastering any of these 15 issues (or others that can be derived from our discussion) that have to be addressed. The evaluation of industrial policy has to evaluate its success by the realized results of either firms or industries that have been encouraged. The underlying market failures or externalities that contributed to the decision to foster a firm or sector cannot be identified from the policy such as subsidized directed credit. Only the effects of the policy can be assessed. We now turn to this task. 3. Does industrial policy work? As noted earlier, it is impossible to offer a single agreed-upon counterfactual to evaluate the past success of industrial policy targeted to individual industries. Thus there 29 have been a number of research strategies pursued to provide an empirical evaluation of industrial policy. These have been reviewed in Noland and Pack (2003). Not all of the methods have been explored for all countries. Researchers have examined, inter alia, the impact of: (1) trade protection; (2) subsidies to R&D; (3) general subsidies; and (4) preferential lending rates on the evolution of productivity and capital accumulation. Few of the empirical analyses find that sectoral targeting has been particularly effective. Consider some of the evidence. In Japan, more than 80 percent of on-line budget subsidies were devoted to agriculture, forestry, and fisheries in the 1955-80 period, the peak of Japan's industrial policy efforts.14 Implicit tax subsides for investment were highest in the mining sector, and quite low in the high technology sectors. Government subsidies to R&D were also small. Unless elasticities of investment and R&D with respect to subsides were implausibly high, their effect was limited. Industries that were encouraged did not experience significantly faster rates of TFP growth than others, and R&D subsidies were largely ineffective. Beason and Weinstein (1996) examined the connection between industrial policy and sectoral TFP growth in Japan. Working with a 13 sector sample for the period 1955- 1990, they fail to uncover evidence that preferential policies (measured by the effective rates of protection, taxes, or subsidies) targeted sectors with increasing returns to scale or that they contributed to the rate of capital accumulation in sectors or their TFP growth. They did find some evidence that prior to the first oil shock, industrial policy targeted sectors with high labor usage. Lawrence and Weinstein (2001) extended this research employing a slightly different data set and found that differential corporate tax rates had an impact on sectoral TFP growth, while direct subsidies and subsidized loans did not. Moreover, they find the paradoxical result the effective rate of protection was negatively associated with sectoral TFP growth and that imports, not exports, were positively associated with TFP growth. The latter result can be explained by noting that there are at least two channels through which imports could contribute to increasing productivity. The first allows domestic producers to use new, improved, or highly specialized intermediate inputs to which they would not otherwise have access. The second is imports compete with 14The following paragraph are based on Noland and Pack, ibid. Chapter 2. 30 domestic products and their availability acts as a constant spur to domestic producers to cut costs and improve quality. Lawrence and Weinstein divide imports into "competitive" and "noncompetitive" imports and in the case of Japan, find evidence to support the second hypothesis. From this they conclude that Japan's growth would have been even faster if it had cut tariffs and exposed a greater share of its domestic producers to foreign competition.15 Lee (1996) following a method broadly similar to Beason and Weinstein finds a similar lack of impact of Korean industrial policies on sectoral capital accumulation or TFP. Pack (2000) follows a different path, assuming that TFP in favored manufacturing sectors was in fact increased in both Japan and Korea and estimates how much of an impact even an assumed successful policy could have had on the growth of gross domestic product. The most favorable estimate is a roughly .5 percentage point increase out of a total GDP growth rate of roughly 10 percent over the relevant periods. While this is significant, it is hardly the magical key to accelerated growth. It is possible that the impact of industrial policy is manifest largely in sectors that purchased inputs from the promoted sectors, even if the latter did not themselves benefit. However, Pack (2000) finds that sectors that were encouraged had few linkages with non- favored sectors via input-output relations and there is little evidence of labor flowing from favored to neglected sectors, a likely mechanism for the transmission of knowledge. Nevertheless, as noted at the beginning of this paper, the difficulty of constructing a single agreed upon counterfactual precludes a robust conclusion. Moreover, all of the empirical analysis examines the contemporaneous impact of policies, for example, did Korean industries that were encouraged experience greater TFP growth in the period during which the stimulation occurred, 1973-85. Someone doubting these results could point to the performance of some Korean firms such as Samsung and LG in the following two decades in such diverse product lines as plasma TVs, RAM chips, and cellular phones, and attribute these later successes to the earlier stimulation the firms received for other product lines. (In the typology above, the Korean Economic Planning 15Japan's Ministry of Finance apparently agrees. In a June 2002 report issued by its Policy Research Institute, it opines that "the Japanese model was not the source of Japanese competitiveness but the cause of our failure" and specifically argues that sectors sheltered by MITI became bloated and inefficient, while 31 Board had mastered skills 2, 4, 8, 12, and 14). These more recent efforts by the firms that allowed them to succeed could be attributed, in this interpretation, to their earlier growth in other product categories. In this view, learning to perform R&D on microwaves, had future carryover effects on plasma TV. Fully resolving divergent views is impossible. Even detailed firm histories by Kim (1997) or Hobday (1995) do not suggest only one line of causality. Nevertheless, even if it should be shown that the success of a few firms could be attributed to earlier encouragement by the government, the aggregate effects just cited suggest there was not a major impact at the national level during the main period of growth acceleration. And any such effects would have to be weighed against the negative long run impacts in the financial sector cited by those skeptical of industrial policy. For example, the Asian financial crises of the late 1990s and Japan's stagnation since 1990 can be interpreted as partly the result of the earlier government direction of lending that minimized the need of banks to learn modern techniques of evaluating individual projects and managing the riskiness of their overall portfolio. 4. New industrial policy Recent discussion of "new" industrial policy including the desirability of fostering learning and obtaining benefits from agglomeration economies offered by industrial clusters has received little systematic empirical evaluation. In principle, the development of clusters could facilitate growing productivity through the provision of overhead services by the organizers plus the interaction of the firms choosing to enter the cluster. Thus clusters could offer an alternative to dependence on either buyer or manufacturer led networks. The benefit of clusters may arise from face-to-face interactions that are productivity enhancing (interactions between software writers and chip manufacturers), a pool of workers with the relevant skills, and reduced transportation costs. Individual market agents may not be aware of the externality they generate for others and this provides an additional market failure that could in principle be addressed by public intervention. In the U.S., where there is a favorable environment for the policy-induced those exposed to international competition tended to be more market-aware, efficient, and profitable (Issei Morita, Financial Times, 27 June 2002). 32 generation of agglomeration, many states and metropolitan regions have attempted to attract firms in similar industrial niches in order to achieve a "critical" mass. Notable successes have occurred in the research triangle in North Carolina centered on Duke University, the University of North Carolina, and North Carolina State University. Others such as the Route 128 corridor in Massachusetts and Silicon Valley were generated spontaneously largely as a result of the presence of major research universities, an innovative and entrepreneurial faculty and high quality PhD students. Moreover, both agglomerations were contiguous to major cities, Boston and San Francisco, attractive consumption venues to the highly educated employees. In contrast, the presence of Cornell University, the University of Rochester, and Rensselaer Polytechnic Institute, along with such major research organizations as IBM and Corning have not engendered a comparable agglomeration in upstate New York, despite efforts by the state government to staunch the significant decline of the region in employment. More generally, state governments and region wide efforts to attract firms to science parks have not been generally successful. The difficulty of replicating Silicon Valley in the U.S. is paralleled by the absence of major success stories in developing countries.16 The rapid development of the software sector in Bangalore and other cities in India, discussed below, appears to be the outcome of the existence of a large group of well educated English speaking students and the entrepreneurial abilities of a small group of residents combined with the awareness of their existence on the part of the large Indian expatriate community, particularly in Silicon Valley. Government participation was non-existent ­ a critical communications satellite was financed by Hewlett-Packard. Positive government efforts followed the "takeoff" of the sector. Of course, publicly financed education institutions generated the fundamental resource, educated workers. This might be considered a generic policy not specifically targeted to the software sector but there was no explicit effort to galvanize the agglomeration economies that have since developed. There are interesting descriptions of a number of clusters in OECD nations but few normative evaluations of their success employing social cost benefit analyses or even 16The Hsinchu Science Park in Taiwan may be an exception though there has been no rigorous evaluation of its benefits. 33 grosser measures such as growth of exports relative to firms outside of the cluster but in the same sector.17 However, some insights can be obtained about whether some recent success stories in Asia conform to the contours of the new industrial policy. We consider in detail the evolution of the Indian software sector centered in Bangalore. The development of the software sector was attributable primarily to activities of private actors. Its success reflected a complex set of interactions between domestic and foreign responses to perceived opportunities. Many of the same patterns, with different details, can be documented for other success stories such as the Hsinchu Science Park in Taiwan (Saxenian 1999, 2001), the Special Economic Zones in China (Rosen, 1999, Huang, 2002), and Bangladesh's rise as a clothing exporter (Rhee, 2000). In the Indian software sector and the Bangladeshi garment sector, the initiating force was private, the government playing almost no role except for the fundamental one in India of providing good education, a policy that does not fall into the domain of selective industrial policy. In Taiwan (China) the establishment of a science park and legislation in China to allow special economic zones to attract FDI were due to an initial government stimulus. A critical input for the success was foreign participation that dealt with some of the roles cited above as requisites of industrial policy (source of new technology, facilitation of learning, source of new product ideas, centralized marketing allowing economies of scope, and coordination of entry of complementary firms). In China, the SEZs mimicked the effect that would have arisen from a free trade policy, i.e., it negated previous adverse public policies. It did not discriminate among sectors. The decision to foster a science park by Taiwan (China) comes closer to a proactive industrial policy. Unfortunately, the experience at Hsinchu has not been systematically evaluated. Many nations have attempted to use export-processing zones of one form or another to catalyze foreign direct investment and perhaps generate agglomeration economies. Evaluation of these suggests that while potentially a useful instrument, they have had indifferent results.18 Success stories can be pointed to in Korea and Taiwan (China) in the 1950s and early 1960s, and of course in the special economic zones of China. But there have been more than a thousand such efforts. There are few clues in the 17Humphrey and Schmitz 2002 provides an extensive survey of the empirical literature on clusters as well as a useful discussion of whether they offer a locally controlled alternative to participation in networks. 34 existing literature about why some EPZs have been successful, while most have failed. A. The Indian software sector In India, the preconditions for the development of the software sector were high quality education in junior colleges and universities financed by the government. A critical role was played by university graduates who went abroad for further training, remained as expatriates in the high tech sector, and later returned home or interacted intensively with newer Indian firms. The lamented brain drain became, with some lag, a paradoxical source of strength and a critical catalytic input (this was also true in the case of Hsinchu). In the 1980s there were a growing number of programming graduates and many were underemployed. There were a large number of graduates at levels ranging from post secondary technical schools to those trained at the Indian Institutes of Technology. Almost all of the students trained in programming had been educated in English. The government's continuing investments in education had resulted in over 1,800 educational institutions and polytechnics producing 70,000 to 85,000 computer science graduates every year.19 Many Indian graduates also had a second university degree or post-graduate degree from the United States or the United Kingdom, often in computer technology20. Other Indian software programmers received training in private software institutes to keep abreast of the latest developments in the software industry and acquire a breadth of software skills. Hence, many were familiar with major computer hardware systems (IBM, UNISYS, DEC, HP and DG21), computer-aided software engineering tools, object- oriented programming, graphical user interface and client networking22. An effort to insure greater local competence in computer production inadvertently contributed to the growth of software. In 1977 increasing pressures to increase domestically controlled computer production led the Indian government to demand that IBM allow Indians to hold 60% of its equity if it wished to continue its operations in India. IBM responded by shutting down its operations in India within 6 months. IBM's departure turned out to be a boon to the Indian computer software industry if not to the 18An extensive set of references is provided in World Bank, 2004. 19James (2000) and India Express (2000). 20Deshmukh (1993). 21Lakha (1990). 35 hardware sector. Without being constrained to proprietary operating systems used in IBM computers, software professionals turned to the cheaper open platforms for programming, especially UNIX. When U.S. firms decided to migrate from mainframe and minicomputer-based systems to UNIX-based client server systems in the 1980s, Indian programmers' adeptness with UNIX relative to their Western counterparts, who were stuck with the "conceptual baggage of legacy systems"23 made them highly sought after. Indian programmers were thus better prepared for many applications than programmers in other developing countries that may have had similar wage structures. In the 1970s the Indian government standardized on UNIX and since it purchased one-half of all computers, its demand contributed to the growth of these skills among programmers. Government demand rather than FDI provided the basis for the expansion of output in the sector, an unintended industrial policy similar to that pursued by the Japanese government in the supercomputer sector (which did not succeed). Nevertheless, the major impetus to demand came from abroad in the form of a set of "accidents." In the 1990s the ratio of world prices for programming services relative to those in India increased due to a global shortage of programmers and the demands for solutions to the anticipated Y2K problem. Enterprising businesses in India capitalized on this opportunity by setting up firms that were essentially employment agencies. Indian software programmers were hired by local firms on behalf of clients in the United States on short-term contracts (either for a fixed period of time or on a project basis) to provide onsite services. `Bodyshopping', as this practice was called, became the predominant mode of Indian software exports because the development work was performed on the client's premises, saving software firms the high costs of acquiring computer hardware. NASSCOM, the software trade association reported that the software sector earned $2.5 billion from Y2K billing from 1996 to 1999, a critical period in the growth of the industry. As late as 1988 software exports had been less than $200 million but had risen to $3.6 billion by 1998, accounting for over 10 percent of total Indian exports. Indian software firms also benefited from another serendipitous event, the European Union's move to the Euro. Many Indian software professionals have been 22Lekshman Pai (1998) 23Udell (1993). 36 actively involved in adapting existing computer systems and databases to accommodate the Euro. Between 2000 and 2002, it is estimated that India earned approximately $3 billion in revenues from these Euro-related IT projects24. Clearly a contributory factor was the low relative level of programming costs in India that conferred a Ricardian comparative advantage in some sub-sectors of software. As late as 1995, after substantial wage increases because of a rising demand for Indian software, the annual wages of Indian software professionals were only 14% to 59% that of their counterparts in Switzerland, USA, Canada and the UK. Given the skills of Indian programmers, these cost savings led firms in some of the industrial countries to outsource their software development requirements to India. While the precise sequence of events leading to a focus on say India rather than the Philippines which exhibited even lower wages is not easily reconstructed given the familiarity with English of educated Filipinos but the Indian expatriate community in the U.S. was certainly a part of the process. In addition, Indian programming firms had engaged in capital stretching efforts due to tariffs on computers that permitted the inefficient domestic hardware industry to charge high prices. Even after restrictions on hardware imports were relaxed in the late 1980s and early 1990s, the tariffs imposed on the imports made it too expensive to acquire state-of-the art computer systems or to upgrade frequently. With less than frontier computers, Indian software developers "worked hard and successfully to get every available bit and byte out of what computing power they had."25 This equipped Indian software programmers with the ability to write the optimized programs that set them apart from their potential low cost competitors, an unintended indirect benefit of India's ISI policy. The circumstances just described suggest the relevance of Brian Arthur's view that idiosyncratic events, tariffs on imported hardware, the Y2K problem, and the shift to the Euro, exert positive feedback and generated a succession of mutually reinforcing benefits. In terms of industrial policy, of whatever form, it seems unlikely that any government could have foreseen and acted upon the demand generated by Y2K and the Euro or anticipated the impact on programming skills of expensive computer hardware. 24IDSA (2000). 25McDowell (1995) 37 The government's main contribution had been to provide high quality university education.26 B. The foreign role One of the major contributors to the Indian software sector was the large number of expatriate Indian IT professionals located in Silicon Valley. The prominence of Indian expatriates in Silicon Valley has been remarkable. In 1998, 774 (or 9%) of the high technology firms were led by Indian CEOs.27 Many of them helped to convince large firms such as Oracle, Novell, and Bay Networks to establish operations in India.28 The impact of this community can be seen in the ownership structure of Indian firms in 2000 shown below. Forty eight percent of Indian software firms were foreign owned, joint ventures, or owned by Indian nationals with intensive participation by foreigners. Although foreign wholly owned firms only make up a small fraction of the software firms in India, they account for a disproportionately large share of the investment made by the software industry. Ownership structure of Indian software firms in 200029 100% Indian 52.1 100% foreign 12.1 Joint Venture 8.9 Indian with foreign connections 24.2 These and other expatriates also invested in India, started firms, helped raise U.S. venture capital, provided expertise, and convinced venture capital firms to operate in India or pay greater attention to opportunities in India.30 In recent years, non-resident Indians, NRIs, have gone a step further in assisting the Indian software industry. Aware 26Some observers felt this was an incorrect allocation of education funds and the returns would have been greater to more extensive and higher quality primary and secondary education. The success of the software sector does not disprove this earlier view. For example, the favorable effect on the adoption of the green revolution package on the income of Indian farmers of even lower levels of education are well established. 27James, 2000. 28Saxenian, 1999. 29Balasubramanyam (1997). 30This suggests that the brain drain must be viewed dynamically. 38 of the obstacles some Indians face in raising capital for their software startups they actively raised venture capital from U.S. investment firms and organized conferences in the U.S. to heighten the awareness of the potential of India's software industry31. A group of NRIs in the U.S. also founded TIE (The IndUS Entrepreneurs), a mentoring group for technology entrepreneurs in India to network with their counterparts in the U.S. and learn from their experiences32. Finally, NRIs were actively involved in lobbying efforts urging the Government of India to revamp its telecommunication policies and other regulations that have impeded the growth of the Indian software industry33. Foreign direct investment accounted for a large percentage of early investment in the sector. For example, in 1996, foreign companies accounted for 70% of the investment in software development in Bangalore.34 And this contribution understates the true impact. Texas Instruments (TI), the first foreign firm to establish an offshore software facility in Bangalore in 1984 after IBM's departure, augmented Bangalore's inadequate land-based telecommunication infrastructure by investing in its own satellite communications network, in conjunction with Videsh Sanchar Nigam (VSNL), the government's overseas communication agency35. Some of TI's lines were later leased to other software firms, enabling them to expand their India-based operations instead of relying solely on onsite services abroad. Until the government built software technology parks in the 1990s linked to earth stations and other telecommunications infrastructure, TI's satellite network remained an important driving force behind the offshore development of software exports. Once U.S. based firms had become interested in India, Bangalore's reputation for technical excellence and its abundant supply of IT graduates from its 3 universities, 14 engineering colleges and 47 polytechnic schools36 made it a natural choice for foreign companies to locate their software firms. The foreign role has been of major importance as it provided much of the infrastructure and international knowledge that allowed Indian firms to exploit 31Kripalani (2000). 32Biers and Dhume (2000). 33Kirpalani (2000). 34The Economist, 1996. 35The telecommunications industry in India is state-controlled, hence the need for TI to procure the services of VSNL instead of a private firm. 39 international opening. In addition, Indian software firms also benefited from foreign joint ventures and partnerships because they create markets for Indian software exports. At the same time these firm provide distribution networks for Indian firms attempting to move upstream. A foreign firm typically outsources its software requirements to its Indian partner, thus providing it a direct and steady source of income. Moreover, partnerships with foreign firms add to the credibility to an Indian firm and act as endorsements of its quality and reliability without government encouragement. The advantage seems to have been firm specific, ala Grossman-Horn rather than Mayer. Thus, other foreign firms looking to outsource their software development would invariably choose a software firm with a proven track record with another foreign company. Lastly, for small Indian firms attempting to move out of the low-end of the software business by venturing into software packages, having foreign partners is an asset because of their established distribution networks, knowledge of the recent trends in the software market (due to proximity to the demand in the U.S.) and significantly lower marketing costs. Since as much as 70% to 80% of the final price of a software package arises from marketing37, small Indian firms without a known brand, an extensive sales network or sufficient revenue find it more profitable to sell its packages via a foreign collaborator. How does this experience of a very successful sector square with the many strands of new industrial policy? All of it was privately initiated, governments at various levels becoming involved only after the success of the sector was evident. The industry expanded on the basis of comparative advantage and never needed any protection. Indeed, one advantage of the software sector was that its inputs, largely downloads from satellites and its output, uploaded to satellites, could not easily be taxed by the Indian authorities. A symbiosis of foreign and domestic firms was critical. Though there was clearly an agglomeration of firms in Bangalore, this was the effect of the presence of the graduates of the local education system. But this was achieved spontaneously without government direction. Foreign contracts rather than government subsidies provided the basis for international exploration of markets. There is no evidence of government initiation or preference. 36Stremlau (1996a). 37Lakha (1994). 40 5. Is industrial policy still relevant? From Hamilton and List to contemporary discussions of industrial policy, the implicit framework has been that of a firm producing tradable goods at an initial cost disadvantage given the limited industrial history of the country, learning to become more efficient, and then competing with imports in the local market or successfully exporting. The marketing of the efficiently manufactured product was implicitly assumed to be routine. Reduction of production costs whether through internal learning-by-doing or through spillovers within industrial clusters was viewed as paramount. In discussions of post-war Asian experience some attention was given to the catalytic role of Japanese, Korean, and Taiwanese trading companies in assembling large quantities of goods and achieving scale economies in marketing but this activity was not given center stage (Lall and Keesing, 1992). Even if countries could now pursue the export oriented policies of four decades ago in Korea and Taiwan (China), it is not clear that they would be efficacious given the changed nature of both retailing and production networks. In the last two decades there has been a shift in the institutional mechanism of international trade. Two types of organization have evolved: (a) international production networks, IPN, in which a producing firm organizes large numbers of suppliers in a number of locations; (b) buyer-led networks in which large retail chains provide specifications for the desired final product and encourage suppliers in developing countries to organize their own production system that most often includes large numbers of local subcontractors.38 These networks have become increasingly important, and are dominant in clothing and electronics and growing in importance in other products such as automotive components.39 In East Asia in recent years components "constitute at least a fifth of manufacturing exports and ... have typically grown 4-5 percent faster than overall trade in East Asia" (Yusuf et. al., 2003, p. 272). One effect of the growing importance of IPNs is their efficiency at organizing production and continuously reducing costs so that the global price that non-member 38A good description of these alternatives and evidence on their quantitative importance is given in Gereffi, 1999. See as well Yusuf et. al. (2003) Chapter 7. Sturgeon and Lester (2002), and UNCTAD, 2001 provide evidence on the empirical importance of the IPNs. 41 firms must compete with shifts down rapidly. Infant firms undergoing learning face other hurdles: rapidly improving quality; changing characteristics of existing products (Ernst, 2002); and an array of new goods that compete with existing ones. For firms attempting to enter export markets, it cannot be assumed that simply achieving low cost is sufficient to realize foreign sales. There is no guarantee that lead firms will be able to identify one or two firms in a small African nation. The existence of supply networks imposes a significant challenge to LDC firms that are not embedded in such a network as the lead firms usually succeed in generating higher performance in design, engineering, the effective use of information and communication technology, and the ability to coordinate production in several locations. (Yusuf et. al., p. 278).40 Further militating against the classical view of infant industries is the change in the nature of retailing. Consider a mundane product such as socks that can be produced efficiently with relatively labor intensive technology. Huge retailers such as Walmart and Target buy these in quantities of millions that typically exceed the production capacity of small (by international standards) industries. The special economic zones in China have become a series of clusters that produce enormous quantities of socks, ties, and other clothing. Retailers and wholesalers place very large orders that are well beyond the production capacities of smaller firms even if these have learned sufficiently rapidly to become cost competitive in relatively small quantities. "These days buyers from New York to Tokyo want to be able to buy 500,000 pairs of socks all at once, or 300,000 neckties, 100,000 children's jackets..." (Barboza, 2004). European firms buy smaller, more varied products but expect local suppliers to provide "in-house design and sample making capabilities that would allow them to translate and adapt the design from Europe." (Sturgeon and Lester, 2002, p. 49). In textiles, clothing, electronics, auto parts and other sectors being a part of an IPN is critical to exporting and upgrading of quality. Firms that are not part of such networks may not succeed even if they are as efficient as members in production costs. Local participants in the network must "label track, respond to product orders in real time 39A good introduction is given by Sturgeon and Lester (2003), and UNCTAD, 2001 provide evidence on the empirical importance of the IPNs. 40 Pack and Saggi (2001) provide a theoretical framework for understanding the willingness of foreign firms to offer such assistance to local firms who could, in principle, switch to other networks. 42 on the basis of style, color, fabric, and size; exchange information on an ... electronic basis, provide goods to a retailer's distribution center that can be efficiently moved to stores ... including containers with bar codes concerning contents" (Yusuf, p. 283). These requirements, now fairly standard in many product areas, suggest that successful penetration of OECD markets will become increasingly difficult for nations that have not yet industrialized. In electronics, an important efficient, labor-intensive growth sector in the past for many of the Asian countries, much of the production is now carried out by contract manufacturers whose size has grown enormously in the last decade. Firms such as Solectron and Flextronics now undertake activity that was formerly under the aegis of major developed country firms who have outsourced the activity. Examining the location of several activities: headquarters, manufacturing, materials purchasing and management, new production introduction centers, and after sales repair centers, Sturgeon and Lester (2002) show that most of these activities of Solectron, the largest of the contract manufacturers, take place in developed countries or the more advanced semi- industrialized nations contiguous to them such as Mexico, Puerto Rico, Romania, and Turkey. Ernst, 2002 (p. 24) confirms these results and points out that specialized clusters in countries such as the Nordic nations, the U.S., France, and Germany are major sources as are Singapore, Hungary, Israel, Korea, and Taiwan (China). Poorer countries even if they have a potential cost advantage after a long learning period will have trouble breaking into these existing circuits. Moreover, China and India present formidable competitors as demonstrated by the concern over the termination of the multi-fiber agreement and the potential losses incurred by nations that formerly had guaranteed access to OECD markets. While it might be argued that the two giant nations will encounter rising wages and thus enter more capital and technology intensive sectors, providing room for new countries, both still have hundreds of millions of workers, largely in the rural sector, who remain poor and will keep a lid on the real wage faced by industrialists over the next decades, implying a continuing supply of low cost products in many sectors. While in principle poorer nations can find a niche in which they have a comparative advantage, finding them is likely to require a vector of skills that are best nurtured by membership in a 43 production network or direct interaction with large retailers. Designing a proactive system to foster industrial growth was difficult at a simpler time when countries such as Japan, Korea, and Taiwan could rely on comparative advantage to enable them to successfully enter the world market. Four decades later in a world of international production networks, very fast innovation with dramatic declines in product prices, rapidly changing product characteristics, new products that quickly lead to the obsolescence of older ones (flat panel computer monitors vs. CRTs), and the premium on the ability to rapidly communicate electronically, it may be beyond the competence of any government to help their domestic firms foresee and successfully deal with more than a small fraction of the unknowable changes that will affect their future trajectory. Viewed from the perspective of potential government policies, the growing importance of production networks suggests an array of potential interventions. Korea and Taiwan (China) had numerous trading companies that aggregated orders of local manufacturers, following the Japanese model of the shosha soga. Most of these arose spontaneously from private efforts. Governments could attempt to encourage the development of trading companies as there may be a market failure given the characteristic that setup costs for such a firm may be significant but marginal costs of adding firms to the network may be small. Such trading firms would operate across clusters of manufacturing firms. Again, this assumes that there are capital market failures that preclude a nascent trading firm from obtaining finance. Other policy issues arise. Will government-sponsored clusters be as effective in generating sustained improvements in product development, quality upgrading, and growing efficiency in order to continue to compete on the world market or will firms within clusters improve faster by becoming part of networks? There is some anecdotal evidence that international networks attempt to limit the extent of upgrading, especially in higher value added segments of design. If this is so, one is back to a situation of deciding whether to promote specific activities within the entire production nexus but this is surely beyond the capacity of all but the most competent of governments.41 Taiwan's experience in the Hsinchu Science Park may be provide a counter-example, so far 41A good discussion of this issue is contained in Humphrey and Schmitz, 2002. 44 unconfirmed by systematic evidence. 5. Concluding remarks Does the current policy landscape of the multilateral trading system even permit developing countries to pursue industrial policy? Should it? It is clear that developing countries today have to contend with several multilateral agreements that the rich countries did not have to consider when they themselves were developing. Have the constraints and disciplines imposed by WTO agreements such as TRIPS and the TRIMS become too restrictive to allow developing countries to chart their preferred course to economic development? This is a difficult question but it cannot be dismissed out of hand. Certainly the international policy environment today imposes constraints on the use of national policies that were absent even 15 years ago and the constraints are backed up by the potent dispute settlement procedure of the WTO.42 The experience in a number of countries in the last two decades suggests that private firms often have been successful at the kind of learning strategies that Succar and earlier analysts were advocating as the government's role. The growth of the Indian software sector, Bangladesh's clothing industry, and China's special economic zones was driven primarily by private sector agents (often from abroad). In the first two the main role of the government was benign neglect while in the latter the Chinese imitated the earlier success of Singapore by enabling the location of foreign investment in enclaves that were well provided with infrastructure. Much of the earlier investment came from overseas Chinese. There was not a search in any of these cases that identified firms or industries with high learning potential and likely spillovers. In Bangladesh and China the firms brought standard technology but importantly extensive marketing networks. Standard comparative advantage can explain the pattern of sector choice. Compared with the exceptionally complex process of either picking sectors (or firms) or the process of allowing firms to identify their own competitive advantage, it seems much more efficient in the current state of intensifying world competition and the growing importance of extensive and complex supply networks to allow foreign firms to facilitate the reduction 45 of costs in the host economy. This would suggest a change in focus from even the new industrial policy to one that focused on negotiation with MNCs on issues ranging from environmental regulation and taxes to efforts to ensuring local learning. The difficulty with this approach is the limited amount of FDI going to LDCs ­ many countries in Africa, the Middle East, and Latin America continue to receive very little. This may be due to their overall economic prospects given their policies. But in these economies hewing to some of the major tenets of the Washington Consensus while recognizing some of its weaknesses might prove a better investment of limited government competence and legitimacy than the extraordinarily complex strategies required by either the new or old industrial policy. In addition, there exist several working groups that are charged with examining the case for multilateral rules on the `Singapore' issues. What will be the eventual impact of these working groups is difficult to say but it is clear that the mere existence of such groups is a reflection of the strong undercurrents that are at play in the world trading system. 42For further discussion see Noland and Pack, 2003, Chapter 5. 46 References Alfaro, Laura and Andres Rodríguez-Clare, 2004. "Multinationals, Linkages, and Economic Development." Economia, Spring, 113-169. Bagwell, Kyle and Robert W. Staiger, 1989. "The Role of Export Subsidies When Product Quality is Unknown." Journal of International Economics 27: 69-89 Balasubramanyam, V and Ahalya Balasubramanyam (1997). "International Trade in Services: The Case of india's Computer Software." The World Economy, v.20 no.6 (September 1997), pp.829-843. Barboza, David, 2004, "In Roaring China, Sweaters are Wests of Socks City," New York Times, December 24. Bardhan, Pranab, 1971. "On Optimum Subsidy to a Learning Industry: An Aspect of the Theory of Infant-Industry Protection." International Economic Review 12: 54-70. Barrios, Salvador, Holger Gorg, and Eric Strobl, 2004. "FDI, Competition, and Industrial Development in the Host Country." CORE discussion paper 2004/11. Barry, Frank and John Bradley, 1997. "FDI and Trade: The Irish Host Country Experience." Economic Journal 107: 1798-1811. Batra, Geeta and Hong W. Tan, 2002. "Inter-Firm Linkages and Productivity Growth in Malaysian Manufacturing." International Finance Corporation, World Bank, Washington, D.C. Beason, Richard and David E. Weinstein. 1996. Growth, Economies of Scale, and Targeting in Japan (1955-1990), Review of Economics and Statistics, 78:2 286-95. Blalock, Garrick, 2001. "Technology from Foreign Direct Investment: Strategic Transfer Through Supply Chains." Mimeo, Cornell University. Besley, Timothy and Anne Case, 1993. "Modeling Technology Adoption in Developing countries." American Economic Review 83: 396-402. Biers, Dan and Sadanand Dhume (2000). "In India, a Bit of California." Far Eastern Economic Review v. 163 no.44 (November 2, 2000), pp.38-40. Coase, Ronald, 1960. "The Problem of Social Cost." Journal of Law and Economics III: 1-44. Deshmukh, Vikas, 1993. "Bangalore: India's Hi-Tech Birthplace." Economic Reform Today, no.3, 1993. Dinopoulos, Elias, Lewis, Trace R., and David E. M. Sappington, 1995. "Optimal Industrial Targeting with Unknown Learning-by-Doing." Journal of International Economics 38: 275-295. Economist, 1996. "Software in India: Bangalore Bytes." The Economist, March 23, 1996. Ernst, Dieter, 2002, "Global Production Networks in East Asia's Electronics Industry, East-West Center, Honolulu, Hawaii. 47 Evenett, Simon. "Study on Issues Related to a Possible Multilateral Framework on Competition Policy," WTO paper WT/WGTCP/W228. Ethier, Wilfred J., 1982. "National and International Returns to Scale in the Modern Theory of International Trade." American Economic Review 72: 389­405. Foster, Andrew D. and Mark R. Rosenzweig, 1995, "Technical Change and Human Capital Returns and Investments: Evidence from the Green Revolution," American Economic Review 86: 931-53. Gereffi, Gary, 1999, "International trade and industrial upgrading in the apparel commodity chain." Journal of International Economics, 48: 37-70. Gereffi, Gary and Olga Memedovic (2003), "The Global Apparel Value Chain: What prospects for upgrading by developing countries." UNIDO Sectoral Studies Series Grossman, Gene and Henrik Horn, 1988. "Infant Industry Protection Reconsidered: The Case of Informational Barriers to Entry." Quarterly Journal of Economics 103: 767-787. Hausmann, Ricardo and Dani Rodrik, 2003. "Economic Development as Self- Discovery." Journal of Development Economics 72: 603-633. Hoff, Karla, 1997. "Bayesian Learning in an Infant Industry Model." Journal of International Economics 43: 409-436. Huang, Yasheng, 2003, Selling China, Cambridge University Press, 2003. Humphrey, John and Hubert Schmitz, 2002, "Governance and Upgrading: Linking Industrial Cluster and Global Value Change Research," Institute for Development Studies, University of Sussex, Working Paper 120. James, David, 2000. "India Starts Up." Upside (Foster City), v.12 no.4, pp265-269. Javorcik, Beata S., 2004. "Does Foreign Direct Investment Increase the Productivity of Domestic Firms? In Search of Spillovers through Backward Linkages." American Economic Review 94, 605-627. Klimenko, Mikhail, 2004. "Industrial Targeting, Experimentation and Long-run Specialization." Journal of Development Economics 73: 75-105. Kripalani, Manjeet, 2000. "A Typhoon of Venture Capital?" Business Week January 31, 2000, pp.28. Laderman, Daniel, William F. Maloney, and Luis Serven, 2003. Lessons from NAFTA for Latin American And Caribbean Countries, Washington, The World Bank. Lakha, Salim, 1990. "Growth of the Computer Software Industry." Economic and Political Weekly, January 6, 1990, pp.49-56. Lall Sanjaya, and Donald Keesing, 1992. "Marketing Manufactured Exports from Developing Countries: Learning Sequences and Public Support," in G. Helleiner, ed. Trade Policy, Industrialization, and Development: New Perspectives, Oxford, Clarendon Press. Larrain B., Felipe, Luis F. Lopez-Calva, and Andres Rodriguez-Clare, 2000. "Intel: A 48 Case Study of FDI in Central America," CID Working Paper No. 58, Harvard University. Lakha, Salim, 1994. "The New International Division of Labour and the Indian Computer Software Industry." Modern Asian Studies, v. 28 no. 2 (1994), pp.381-408. Lawrence, Robert Z. and David E. Weinstein. 2001. "Trade and Growth: Import Led or Export Led? Evidence from Japan and Korea," in Joseph E. Stiglitz and Shahid Yusuf eds. Rethinking the East Asian Miracle. Oxford: Oxford University Press. Lekshman Pai and Uday Lal, 1998. "A Major Exporter Turns Inward." Computerworld (Framingham), v. 32 no. 14 (April 6). Lee, Jong-wha. 1997. "Government Interventions and Productivity Growth in Korean Manufacturing Industries," Journal of Economic Growth. 1:3 391-414. Markusen, James R., and Anthony Venables, 1999. "Foreign Direct Investment as a Catalyst for Industrial Development." European Economic Review 43: 335-56. Mayer, Wolfgang, 1984. "The Infant-Export Industry Argument." Canadian Journal of Economics 17: 249-269. McDowell, Stephen D, 1995. "The Decline of the License Raj: Indian Software Export Policies." Journal of Communication v. 45 no. 4 (Autumn), pp.25. McKinsey Global Institute, 1998a, Productivity Led Growth in Korea, McKinsey & Co, Washington D.C. and Seoul. McKinsey Global Institute, 1998b, Productivity- the Key to an Accelerated Development Path for Brazil, McKinsey & Co, Washington D.C. and Sao Paulo. Moran, Theodore, 1998. Foreign Direct Investment and Development, Washington DC: Institute for International Economics. Moran, Theodore, 2001. Parental Supervision: The New Paradigm for Foreign Direct Investment and Development. Washington DC, Institute for International Economics. Moran, Theodore, 2004. "How Does Foreign Direct Investment Affect Host Country Development: Do We Already Know the Answer? Using Industry Case Studies to Make Reliable Generalizations," in Magnus Blomstrom, Edward Graham, and Theodore Moran, eds., The Impact of Foreign Direct Investment on Development: New Measures, New Outcomes, New Policy Approaches, Institute for International Economics, Washington DC. Noland, Marcus and Howard Pack, 2003. Industrial Policy in an Era of Globalization: Lessons from Asia, Washington, D.C. Institute for International Economics, 2003. Pack, Howard and Larry Westphal, 1986. "Industrial Strategy and Technological Change: Theory versus Reality." Journal of Development Economics 22: 87-128. ........, 2000, "Industrial Policy: Growth Elixir or Poison?" World Bank Research Observer, 15:47-68. Pack Howard and Kamal Saggi, 2001. "Vertical Technology Transfer via International Outsourcing." Journal of Development Economics 65: 389-415. 49 Peres Nunez, Wilson, 1990. Foreign Direct Investment and Industrial Development in Mexico. Paris: OECD. Raff, Horst and Young-Han Kim, 1999. "Optimal Export Policy in the Presence of Informational Barriers to Entry and Imperfect Competition." Journal of International Economics 49: 99-123. Rhee, Yung W., 1990, "The Catalyst Model of Development: Lessons from Bangladesh's Success with Garment Exports," World Development, vol. 18, no. 2, February 1990, pp. 333-46. Rodriguez-Clare, Andres, 2004a. "Microeconomic Interventions After the Washington Consensus." Inter-American Development Bank, Washington DC. Rodriguez-Clare, Andres, 2004b. "Clusters and Comparative Advantage: Implications for Industrial Policy." Inter-American Development Bank, Washington DC. Rodriguez-Clare, Andres, 1996a. "Multinationals, Linkages, and Economic Development." American Economic Review 86: 852-74. Rodríguez-Clare, Andrés, 1996b. "The Division of Labor and Economic Development." Journal of Development Economics 49: 3-32 Rodrik, Dani, 2004. "Industrial Policy for the Twenty-First Century." John H. Kennedy School of Government, Harvard University, mimeo. Rodrik, 1996. "Corodination Failures and Government Policy: A Model with Applications to East Asia and Eastern Europe," Journal of International Economics 40: 1-22. Rosen, Daniel H., 1999. Behind the Open Door: foreign enterprises in the Chinese Marketplace. Washington D.C., Institute for International Economics. Saggi, Kamal, 2002. "Trade, Foreign Direct Investment, and International Technology Transfer: A Survey." World Bank Research Observer 17: 191­235. Saxenian, AnnaLee, 1999. Silicon Valley's New Immigrant Entrepreneurs. San Francisco: Public Policy Institute of California, June 1999. Saxenian, AnnaLee, 2001, "Taiwan's Hsinchu Region: Imitator and Partner for Silicon Valley," Stanford Institute for Economic Policy Research, Stanford, CA. Scitovsky, Tibor, 1954. "Two Concepts of External Economies." Journal of Political Economy 62: 143-151. Stremlau, John, 1996. "Dateline Bangalore: Third World Technopolis." Foreign Policy 102: 152 - 159. Sturgeon, Timothy and Richard Lester, 2002, "Upgrading East Asian Industries: New Challenges for Local Suppliers," Cambridge, Mass. Industrial Performance Center, MIT. Sturgeon, Timothy and Richard Lester, 2003. "The New Global Supply-Base: New Challenges for Local Suppliers in East Asia", MIT Industrial Performance Center, Working Paper 03-001. 50 Succar, Patricia, 1987. "The Need for Industrial Policy in LDCs ­ A Restatement of the Infant-Industry Argument." International Economic Review 28: 521-534. Udell, Jon (1993). "India's Software Edge." Byte, v. 18, no. 10 (September 1993). United Nations Conference on Trade and Development (UNCTAD), 1996. World Investment Report: Promoting Linkages. New York: United Nations. World Bank, 2004, World Development Report, The World Bank, Washington D.C. Yusuf, Shahid et. al., 2003, Innovative East Asia, Washington, D.C., The World Bank and Oxford University Press. 51