From YalePressWiki

Contents 1 Chapter 1: Plurality, Choice, and the Politics of Industrial Innovation 1.1 A Brief History 1.2 The State of the Literature 1.2.1 Late-Development and Developmental-State Theories 1.2.2 Globalization as Fragmentation: Global Production Network Theory 1.2.3 Systems-of-Innovation Theories 1.3 Building the Theory 1.3.1 A New Model of Rapid Innovation–Based Industrialization Politics 1.3.2 Different Choices: The Varied Paths of Industrial Growth 2 Notes

[edit] Chapter 1: Plurality, Choice, and the Politics of Industrial Innovation

Action, the only activity that goes on directly between men without the intermediary of things or matter, corresponds to the human condition of plurality, to the fact that men, not Man, live on the earth and inhabit the World. While all aspects of the human condition are somehow related to politics, this plurality is specifically the condition—not only the conditio sine qua non, but the conditio per quam—of all political life.
—Hannah Arendt, The Human Condition

This book is about choice. Its main argument is that, contrary to what we are led to believe, the current processes of intensified globalization give emerging economies a larger number of economic-development alternatives than they have had since World War II. This is especially true, I argue, in the case of rapid innovation–based (RIB) industries. A general truism today is that both the onslaught of international economic forces and the fragmentation of production limit the power of states to set unique courses of successful economic growth. And yet in this book I argue that these same conditions have given states more choices of action than ever before, for the increasing complexity and openness of the world allow developing states that wish to engage with the international economic system a far larger number of entry points than in any other period.

This does not mean that emerging economies have an easier task of development, or that success, even if and when it is achieved, is without cost. The options facing states today are real, and each has its own negative as well as positive consequences. It is true that old methods of industrialization no longer work. However, I contend that this means only that the role of the state in economic development has changed, not that its power has diminished. The world has evolved, and with it we need to transform our understanding of industrial development.

One of the most unexpected developments of the 1990s is that firms in a number of emerging states not known for their high-technology industries in the past moved to the forefront in new information technologies (IT). Even more surprising from the point of view of comparative political economy theories, the IT industries of these countries seem to embody different business models and to carve out different positions in the global production network of the U.S.-dominated IT industry. Of these emerging economies, the Taiwanese, Israeli, and Irish indigenous IT industries, riding what seems to be one of the greatest waves of industrial innovation ever, have become hotbeds of new technology–based firms (NTBF). Some of these firms have become global players over the past ten years.

That very different IT industries working in the same sectors and markets have grown in emerging economies challenges two central ideas in the comparative political economy literature. First, it suggests that there is a way for less-developed countries to become successful players in rapid innovation– based industries, something that was not considered plausible in the past. Second, it suggests that there are multiple choices for industrial development. The cases of Ireland, Israel, and Taiwan demonstrate that states and societies, even under conditions of intense globalization in one of the most globalized industries, can pursue different strategies of economic growth.

Contemporary political economy literature cannot satisfactorily explain many of the changes in industrial systems in the past few decades. This is especially true in the case of rapid innovation–based industries in emerging economies. Contemporary political economists take a dismal view of the prospects for RIB industrial leapfrogging by backward societies—that is, transformation of such societies to front-rank competitors in technological innovation in the span of one generation. While each of the standard theories has a particular approach, each unites in pessimism about the possibility of using government policies to leapfrog in rapid innovation–based industries.^[1] Furthermore, none of the theories can accommodate a situation in which a few emerging countries, each employing very different industrial policies, planned and implemented by very different bureaucracies, with very different labor market regulations, financial institutions, and educational and training systems, achieve success at the same time in the same research and development (R&D)–intensive sectors.

Against these bleak predictions, the stories of Israel, Ireland, and Taiwan offer cases of successful rapid innovation–based industrial growth in the time span of one generation in countries with very different political and institutional systems. These stories offer us multiple puzzles, the answers to which are intimately linked to important debates in comparative political economy. First, comparing these three stories provides us with insights into a highly contested issue of economic growth—the role of the state in the development of high-technology industries in less-developed economies. Second, comparing the three very different paths of science-and-technology industrial policy regimes and development bears upon the question of whether states and societies have choices in their economic development strategies. Finally, comparing the three different positions and relationships of the local IT industry with the global IT production networks offers us an understanding of how globalization, and especially the radical transformation of production, influences development opportunities of peripheral economies.

Politics is about crafting and picking alternatives. Politics matters, therefore, only if choices can be made and if making them has real consequences. In the past three decades students of politics have done their best to convince us that choosing between alternatives, even if choices exist, no longer truly matters. Nowhere is this more evident than in comparative political economy, particularly in the study of industrial development, where the plurality of positive actions has been diminished to one. Both the neoclassical economists, in their belief that the only choice is to let the market rule, and the developmental statists, who argue that only specific structures of states bring forth industrial success, have marginalized the importance of picking between true alternatives. Furthermore, even the varieties of capitalism and the development-path literatures, by focusing on how different established institutions grant different comparative and competitive advantages, offer path-dependent explanations that maintain diversity only by limiting choice.

Yet I believe that choices still exist, that states and societies, through the political process of crafting and picking alternative modes of action, can follow diverse paths and still achieve industrial success. Furthermore, this book demonstrates that the outcome of each path is a different socioindustrial system with different strengths, different weaknesses, and a different distribution of the fruits of success. With this book I set out to confirm that emerging economies have a plurality of options with regard to their industrial development. Moreover, I advance a framework of thinking about how different choices lead to long-term consequences and the development of successful and radically different industrial systems. I strive in short to give politics—the art and profession of creating alternatives and the social struggles of choosing between, and acting upon, them—the importance that it seems to have lost in the social sciences.

To this end, one goal of this book is to expand our understanding of the role of the state in the development of the rapid innovation–based industries— not only to clarify what options less-developed states have to spur and sustain IT industrial growth but also to analyze and map the limits of state actions and the dynamics of state-industry interactions. To do so, I focus my analysis on the ways in which the three states have influenced the skills and capabilities acquired and developed by the private IT industry. Another goal is to explain how the different approaches of the three states shaped the IT industries and influenced the particular development path of, and the particular capabilities developed by, each industry. In doing so, I seek to move beyond the question of whether the state has a role in industrial development. I propose a theory concerning the influence of particular policies on the development of rapid innovation–based industries in emerging economies. While much of the literature in comparative political economy sees the existence of different industrial and capitalistic systems as givens, and explains their influence on the behavior of firms and economies under the same global conditions, in this book I seek to understand how different industrial systems are created in the first place.^[2] I argue that state actions give rise to different hightechnology industries, and that these industries not only utilize different business models, position themselves differently in the global production network, and acquire different skills and capabilities, but also have very different strengths and weaknesses with regard to their sustainability and profitability. In addition, I demonstrate that these industries allocate the economic benefits of success very differently within their host societies, and thus choices are made, actions are taken, and these choices and actions have consequences.

Thus the story is both simple and complex. I argue that in the case of rapid innovation–based industries two major facts not accounted for in the literature— that new technologies in themselves have become the final product, and that production is increasingly fragmented internationally—have altered the way in which industrial production is taking place. Accordingly, these changes transform the ways in which emerging economies can take part in the international economy.

With the fragmentation of production and the rise of global production networks, no longer do whole industries locate in one specific geographical location. Instead, a particular set of activities takes place in one country, while others take place in different ones (Arndt and Kierzkowski 2001a, Gereffi 1994, Gourevitch 2000, Henderson et al. 2002, Sturgeon 2000, 2002, 2003). Consequently, states now excel in a definite set of activities rather than in specific industries. For example, Taiwan, China, Israel, and the United States all have large and successful semiconductor industries; however, each of these countries specializes in a particular set of activities and has continued to develop a significantly different set of capabilities.

If in the past, late-developing countries were expected to utilize a strategy of imitation, technology transfer, and competition on the basis of scope and scale, aiming to build whole industries that can compete in global markets, this strategy has become suspect because production in all those industries had become fragmented (Amsden 1989, 2001, Gerschenkron 1962, Johnson 1982, Wade 1990). Moreover, if in the past one could claim that states need to develop capabilities that encompass whole industries, a much more plausible strategy today would be to develop specialized capabilities in particular stages of production.

In the case of RIB industries, new technologies are themselves the products. Therefore a development strategy that builds on technology transfer and imitation, assuming that both markets and products are known—and hence strategic planning is feasible—does not work. In our transformed world, states need to develop the capacity to innovate. Yet the point is that with the fragmentation of production, emerging economies now have many different entry points around which to innovate and develop their own unique industrial strengths, inspiring innovation with many different faces. Building on insights from the development state, globalization, and systems of innovation theories, I develop in this book a framework to explain both how different choices result in specific industrial systems and why the same state can both succeed and fail in different subsectors of the IT industry.

Following the neo-development-state theories, I contend that state-industry embeddedness is critical (Ansell 2000, Evans 1995, O’Riain 2000, 2004). However, I also contend that there are different modes of embeddedness, and that the process of becoming embedded is at least as important as the final structural result. Hence each mode and process of embeddedness leads to a different industrial system with diverse strengths, weaknesses, and relationships with the global markets. There is no one structural form to which all successful countries must adhere.

Furthermore, what influences the specific processes of state-industry interlinking, as well as the specific mode of interaction between the local and global IT industry and financial markets, depends on:

1. the ways in which the state acts to acquire the necessary knowledge and skills;
2. the ways in which the state acts in order to solve the inherent market failures in industrial R&D, specifically, by lowering the risk in order to spur private actors to enter the industry; and
3. the ways in which the state acts to interlink the local economy with the leading multinational corporations (MNCs) and the global financial markets, both within and outside its national borders.

These state actions, in turn, depend on the political processes behind state action. Hence while this book refutes the specific argument of Alexander Gerschenkron about the advantages of economic backwardness, it does urge a return to the fundamental Gerschenkronian insight. Since national economic backwardness equals powerlessness in the international arena, states do not, and should not, wait idly hoping that the miraculous power of the market will throw some economic growth their way. Hence economic development should be seen as part of the national building project, a rite of passage that many states must pass in order to preserve their independence (Gerschenkron 1962). Consequently, how and why states act reflect the ideals that their societies have of themselves and the role of the state within this vision; each state’s interpretation of its fundamental security threats, both internal and external; and their position within the international system.

Before I fully elaborate on the theoretical framework and arguments of this book, I have chosen to present the three cases and follow with an open debate with the literature, which reveals how I reached the above arguments.

[edit] A Brief History

Ireland, Israel, and Taiwan were poor, peripheral, and technologically backward societies in 1950. They all had mixed track records of success in more traditional industries. At the end of the 1960s, state-led initiatives to develop indigenous high-tech industries appeared in all three societies. At that point, all three were very similar along most of the important economic dimensions, including size and population. All three had low skill intensity in their labor force (defined as the number and percentage of scientists and engineers in the population), a relatively low percentage of high school graduates, poor communication, and physical infrastructures and a high dependence upon agriculture.

All three states followed similar infrastructure policies. Each improved its educational system, leading to high rates of skill formation between the 1960s and the 1980s. All three improved their physical infrastructures and, after failed attempts at direct bureaucratic control, created public telecommunication companies that vastly improved their line-subscription penetration rates. Deregulation and privatization of the telecommunication market followed, leading to high rates of wireless and Internet penetration. Finally, unlike Japan and South Korea, all three countries based their growth on small and medium-sized enterprises rather than on big corporations.^[3]

Despite similarities among their macro-level and infrastructure policies, however, their micro-level policies—those at the level of industry and firm— were distinctly different. Since the late 1960s Ireland has focused mainly on foreign direct investment (FDI)–based industrial development policies. Israel has focused on inducing industrial R&D activities through government grants, with project ideas originating solely in private industry. In Taiwan, the ruling party, the Kuomintang (KMT), mistrusted big private industry and feared the rise of competing powers. Hence it prevented the creation of overarching conglomerates, controlled the inflows and outflows of investment, and relied on such public research agencies as the Industrial Technology Research Institute (ITRI) and the Institute for Information Industry (III) to lead R&D efforts and diffuse the results throughout private industry.

Moreover, the structure of the state in all three countries had varied significantly. Ireland is a model of an English-style Weberian bureaucracy with its separation between engineering and management—that is, between industrialdomain knowledge and high-level bureaucrats and politicians (Rose 1981, 1986). Taiwan is a model example of a technocratic Weberian bureaucracy in the Japanese style (Kim et al. 1995). All high- and midlevel bureaucrats, as well as their ministers, have graduate educations in engineering and thus are quite versed in their industrial-domain knowledge. Israel has a much more chaotically structured bureaucracy. The key principle behind its establishment was to balance bureaucratic professionalism and the ability of politicians to implement their will. The result is similar to the American bureaucracy: elected politicians bring with them many of the top executives, including all the directors general of state ministries, from outside the state structure (Bowsher 1991, Deri 1993, Heclo 1977, Pfiffner 1992). These differences significantly influence the politics of industrial development in each country. Moreover, their existence severely weakens arguments that focus strictly on specific state structure.

It is important to underscore the critical role of the state in all three cases. For many years, the private market did not possess the necessary skills and capabilities to successfully create and manage R&D-based IT companies. Private actors—in particular, investors—shied away from even attempting to do so. Moreover, even after more than a few successes, such as the spin-off of UMC from ITRI in Taiwan or the several initial public offerings (IPOs) of Israeli firms on New York stock exchanges by 1989, private financial institutions were unwilling to invest in the IT industry. Thus the state was the only actor able and willing to start the process of skills and capabilities development in an attempt to spur the growth of the industry. Moreover, being small also posed significant challenges for all three states when they opted to develop their high-technology industries, challenges that larger countries like South Korea, India, and China did not face.^[4] Lastly, while Israel, Ireland, and Taiwan are, on the aggregate, successful, they also had their fair share of failures. Consequently, looking at both hardware and software subsectors of the IT industry in the three cases gives us an added advantage of an in-case comparison and a fuller understanding of both failures and successes.

Today the IT industries of all three countries are undoubtedly successful.^[5] Yet the three IT industries are markedly different in the way they work, the business models they use, and the way they relate to the global IT production network.

Israeli firms have succeeded in both the software and the hardware subsectors. The industry’s role in the global IT production networks is one of a supplier of high-end new technologies (or of new products based on new technologies). Thus Israel’s success is almost solely based on intensive R&D. The quintessential Israeli company is Checkpoint, the inventor and market leader of what has become the ubiquitous network security product: the firewall. Checkpoint was founded by three friends who, in the late 1980s (many years before the ascendancy of the Internet), realized that the move to large-scale networking computing would necessitate new security products. Understanding that the market was not yet ready for their invention, the three bided their time working for other companies until they joined forces to found NSK, later to be renamed Checkpoint, in 1993. Since its inception, Checkpoint has not only invented and created its own market niche but also organized international standard-setting bodies around its own technologies.

In Taiwan the software industry development has stagnated, and the IT hardware industry is embedded within the global IT production network in a very different way from that of Israel. Taiwan’s industry bases its success on innovation in product design and manufacturing, and focuses on secondgeneration innovation R&D. The archetypal Taiwanese company is Taiwan’s Semiconductor Manufacturing Corporation (TSMC). Similar to most of the leading Taiwanese semiconductor companies, TSMC began life as a research effort of Taiwan’s public research organization, ITRI. ITRI’s launching of TSMC has completely transformed the working of the global semiconductors industry, not by inventing new technology or products but by inventing a new way to organize integrated circuit (IC) chip production. The company, led by Morris Chang, the head of ITRI before resigning to become TSMC’s chairman, was spun off to become the world’s first pureplay foundry—that is, a company whose sole business is fabricating chips out of other companies’ IC designs. Pureplay foundries receive codified designs from the design houses and fabricate their chips for them, thus accelerating specialization in both design and fabrication. By allowing small teams of designers to produce their own chips, the pureplay foundries have unleashed immense innovative potential, transforming the global industry as well as Taiwan’s. Today TSMC, together with another ITRI spin-off, UMC, are by far the world’s largest pureplay foundries.

In Ireland, even after more than forty years of industrial policies that brought MNCs to open manufacturing facilities there, the indigenous IT hardware sector growth has not been significant, and it is the local software industry that has risen to prominence. The prototypical Irish company is Iona. Created by four computer science lecturers at Trinity College, Dublin; boosted by the founders’ ability to raise EU grants, Iona was established in 1991 to implement into products a new middleware standard, CORBA, which the group has been involved with. Iona’s successful IPO on NASDAQ in 1997, at the time the fifth-largest software IPO on the exchange, has transformed the perception of the Irish indigenous software industry within and outside of Ireland. However, while Iona is at its core a product company, most of its revenues throughout its history have been generated by its services arm.

In all of these stories, politics plays an important role. In a similar way to Japan in the late nineteenth century and Germany before it, these three young societies faced a harsh reality in which a failure to build successful national industries would have meant the destruction of their societies as independent social units. Israel, Ireland, and Taiwan were all societies that had only recently gained the right to independent existence after long, bitter, and bloody struggles. The leaders who directed the three societies to their current success— politicians, civil servants, and businessmen and entrepreneurs—were all of a generation keenly aware of the price of failure.

Thus the ways in which Ireland, Israel, and Taiwan have been building their national IT industries, and shaping the relations of these with the global markets, can be understood only as part of national efforts. These have been endeavors in which Ireland, Israel, and Taiwan strived not only to stay independent but also to continuously win legitimacy as a coherent national unit, both internally and externally. Economic and industrial growth has been an important part of the national and international process with which these societies have been defining and redefining themselves. Analyzing the roles of the state in the growth of the IT industry, therefore, gives us another mirror in which to see how national sociopolitical identity is shaped in a globalizing world—a continuation of the constant sociopolitical struggle to define what it is to be Irish, Israeli, and Taiwanese.

In the bulk of the book I aim to show that politics—creating, choosing between, and acting on, different alternatives—has consequences. Accordingly, I have chosen to analyze how different policies in turn are structured and shape different national industrial systems. Equally important is how specific critical decisions were shaped, in part, as an element in this nation-building political process. Thus I will briefly explore how Ireland’s economic policy was shaped within the context of its long dependency on Britain and its struggle to create itself anew; how science and technology policies were formed by both the Israeli-Arab conflict and the ongoing quest for a national Jewish identity; and how the Taiwanese IT industry has been shaped by its formal international political isolation and the need to craft and present a new and better model of “a China.”

In this exploration, I have used the comparative case-study method in an inductive-iterative theory-building effort (Eisenhardt 1989, King et al. 1994, Lin 1987, Ragin 1987, 1994, Van Evera 1997, Yin 1994). Hence the theoretical and research traditions from which this book arises are based on the assumption that only an intimate connection with empirical reality allows the researcher to develop a relevant and valid theory (Glaser and Strauss 1967). To this end, the data analyzed in this book come from a database of 482 interviews conducted between December 1999 and December 2003, as well as from official and industrial statistical sources.^[6] The interviews were conducted with founders and executives of IT companies, top civil servants in all the developmental and science–and-technology industrial agencies, academics, and venture capitalists (VCs). All developmental agencies were visited at least once and their top officials interviewed. In addition, interviews were conducted with the specific labs and departments within each agency. For both the hardware and software sectors, a list of companies was compiled from public and official records. All the major Irish, Israeli, and Taiwanese companies in the two sectors were interviewed, together with a large sample of smaller companies chosen from companies that were not doing only bespoke development or consultancy. Finally, a snowball technique was used to tease out companies that were not included in the original lists but were deemed innovative by their peers. Thus our sample was intentionally biased toward the more successful and R&D-based companies, giving us, if anything, an overoptimistic view. In all company visits I tried to interview the top management, preferably the CEO-president and/or the firm’s founders.^[7]

[edit] The State of the Literature

No single strand of political economy literature offers a comprehensive answer to the puzzles that motivate this book. The first puzzle is the role of the state in the development of high-technology industries in less-developed economies, under the conditions of intensified globalization and fragmented production. The second puzzle is how such different development paths of the hightechnology industry appeared, and whether they suggest that multiple developmental choices are available to emerging economies.

It was therefore necessary for me to develop a theoretical approach based on the arguments of several theories. For each question it was necessary to glean insights from a different body of literature. From the developmental-state and late-development theories I sought to understand the role of the state in economic growth in emerging economies; from globalization theories, specifically global production network and industrial fragmentation, I looked for a way to understand development in the radical new conditions caused by the rise of global production networks; and from the systems-of-innovation theories I sought to develop a fresh understanding of how different institutional systems affect the innovative capacity of industrial systems, and how different decisions about resources, industrial R&D, and industrial relations influence the dynamic creation of firms’ capabilities. Systems-of-innovation theories also allow me to present a theory of how the co-evolution of state-industry interactions influences different development paths.

In this section I explore these theories and the answers they provide to the puzzles of R&D-based industrial development of emerging economies. Tying these answers together with research-based insights, I then present the full account of my approach in the concluding part of this chapter.

[edit] Late-Development and Developmental-State Theories

One prominent school of thought that attempts to explain the role of the state in emerging countries’ economic growth is late development. This paradigm originates in the work of Alexander Gerschenkron on the industrialization of the “late developing” European countries—for instance, France, Germany, Russia, and Italy (Gerschenkron 1962). In Economic Backwardness in Historical Perspective, Gerschenkron presents an institution-based view of how a backward state’s leadership can use the advantages of relative backwardness to attain rapid economic growth. He argues that the more economically backward a nation, the more state intervention is needed if it is to grow. An economic historian of Europe, Gerschenkron acknowledges that the task of economic growth should be positioned within the process of modern nation-state building. Late-developing states needed a more rapid industrialization than the pioneering nations, as economic backwardness equals political and military weakness.

One of Gerschenkron’s main arguments is that states have different paths of industrial development built around different institutions, depending on the timing of industrialization. Gerschenkron points to the appearance of a new model of industrial banks in France and Germany and the transfer of this model to Italy and Russia as they reached their rapid industrializing phase. Advancing a linear theory of economic development that treats it as a process with specific stages, Gerschenkron contends that among the advantages of backward countries is that product markets are already developed and defined by the pioneering countries, which have already invested in the R&D to develop the manufacturing technologies.^[8] A backward country, unlike a pioneering country, has the double advantages of knowing the market, and accordingly being able to predict needs fairly accurately, while at the same time having a clean slate with regard to manufacturing infrastructure, so that its industry can invest in the latest technologies. This allows backward countries to reach a scale and scope that the pioneering countries cannot match, as they have already invested in an array of smaller and older manufacturing facilities.

To use these advantages, backward economies need two institutional capabilities: planning capability (therefore a professional and capable bureaucracy) and access to large amounts of patient investment capital—that is, financiers willing to invest large sums for long periods of time. Under the adverse condition of a backward economy, this is more likely to occur with active state involvement. A relatively backward economy, if it manages to put these two together, can invest in large-scale factories with the latest manufacturing technology. It can achieve economies of scale and outdo the leader in manufacturing capacity and global market share. Gerschenkron stipulates the development of specific and new institutional systems, different from the institutional system of early developing countries.

The idea that backward economies have a specific path of development that depends on the timing of their industrialization is still prominent in the literature. For example, Alice Amsden argues:

If industrialization first occurred in England on the basis of invention, and if it occurred in Germany and the United States on the basis of innovation, then it occurs now among “backward” countries on the basis of learning. The paradigm of late industrialization through learning generalizes to a diverse assortment of countries . . . but in all cases industrialization has come about as a process of learning rather than of generation of inventions or innovations. Learning, moreover, has been based on a similar set of institutions. (Amsden 1989, p. 4, italics added)

Even the most recent proponents of the developmental state theories bow to the notion that common timing necessarily means a common model of development.^[9]

Gerschenkron’s model provides the basis for a more recent version of the late-development theories, the developmental-state theory. Most writers in this tradition focus on the rapid growth of Japan after World War II and on the rise of the Asian newly industrialized countries (NIC), also known as the Tigers or the Dragons. The latter group of states has been described as latelate- developers. One of the pioneers of this approach was Chalmers Johnson (Johnson 1982). Johnson situated the industrial and economic growth of Japan as a part of its national-political struggle and development process. In MITI and the Japanese Miracle, Johnson uses Gerschenkronian logic to argue that what enabled Japan to attain such rapid growth after its devastation in the Second World War was a unique developmental agency called the Ministry of International Trade and Industry (MITI). Johnson argues that MITI acted as a pilot agency, not only strategically planning the development of Japan, relying on the fact that both the technologies and the major markets for its final products were already well developed, but also using control over foreign exchange to encourage the entrance and growth of a few large conglomerates to compete in each of the strategic industries chosen.

Hence, Johnson argues, the fact that the Japanese state had a capable bureaucracy, and a specific structure with a pilot developmental agency, allowed Japan to become a plan-rational economy, enabling it to generate an exportoriented catch-up industrialization that became the basis for its “economic miracle” after World War II.^[10] Some writers have applied Johnson’s arguments to South Korea. Others claim that Taiwan also used a similar strategy even if in a less-state-controlled way (Amsden 1989, 2001, Amsden and Chu 2003, Cheng 1990, 1993, Fields 1995, Kohli 1994, Park 2000, Wade 1990, Woo- Cumings 1991).^[11]

In sum, the developmental-state theories have built on Gerschenkron’s theory of relative backwardness to present a model of development that emphasizes the role of the state in a national effort of creating an export-based industrial system and facilitating industries that excel in technology transfer– based catching-up. Thus they advance an argument about the need for specific state structure that enables emerging economies to utilize a particular strategy of development. This is a strategy of state-led development based on longterm industrial planning and the nurturing of a few large industrial conglomerates operating across a broad array of industries with some managed competition. This creates a system that tends toward large investment in the latest manufacturing technologies to reach scope and scale. Its underlying assumption is that industrial development is achieved by the growth of a few big vertically integrated firms that manufacture complete products and are not only competing among themselves but also able to directly compete with foreign companies in the world markets.^[12]

The industrial systems built as a result of these policies are systems that foster industrial research that focuses on improvement in process and manufacturing technologies, and on incremental changes to products and technologies originally innovated and designed elsewhere. Moreover, this model of development is based on the assumption that advanced economies are willing to grant the emerging economies access to their market and to their technologies, even when their trade balances with the same emerging economies are deteriorating. In addition, a key element in all these theories is a particular model of professional elite bureaucracy, as was idealized by Weber (Weber 1952, 1958, 1999).

Yet none of the assumptions of this model holds in the case of R&Dintensive and rapid innovation–based industries. First, looking at the industrial landscape, the market is not already well developed. Therefore strategic planning by the state, based on the fact that both the products and their markets are defined, is not as useful. Second, the rate of technological innovation is so fast that industrial systems based on incremental innovations on technologies and products developed elsewhere find it difficult to compete and develop comparable products quickly enough. Third, products are no longer being manufactured by vertically integrated firms. Many of the leading MNCs in all industries, American as well as European, are shedding most of their manufacturing capacities and moving to manage a global production network where products are manufactured by stages in geographically distant locations. Finally, when the industry itself becomes the creation and rapid application of new technologies, a strategy that is based on catching up and massive long-term investment in large-scale manufacturing facilities does not grant such a big advantage.

Last, but not least, Israel, Ireland, and Taiwan, while developing at the same time and in the same industrial sectors, have profoundly different industrial systems and capabilities, employing different business models. Furthermore, their state structures and state-industry relationships could not be more diverse. Hence we need to develop a new understanding, a framework which would take the Gerschenkronian insights about the role of the state in industrial development but reshape them it into the new reality, where a linear model of development is not necessarily true, and where multiple options and state structures exist.

Indeed, some of these same concerns—especially the relative economic stagnation of many of the Asian NICs and Japan toward the end of the 1990s, their difficulties to succeed in more innovative industries, and the realization that under the new international conditions and the rise of China it would be difficult for other nations to emulate their growth strategies—brought a new version of the development state theories to the fore. Proponents of the neo-developmental state theories, building on earlier research by critics of the strong developmental state theory, such as Richard Samuels, or on the revised theories offered by some of the developmental statists themselves, such as Peter Evans, proposed a restructured theory of state-industry interaction in industrial development (Amsden and Chu 2003, Ansell 2000, Calder 1993, Chibber 2002, 2003, Evans 1995, O’Riain 2004, Samuels 1988, 1994). The adjectives describing this newly found category of developmental state proliferated. However, be it “the flexible developmental state,” “the neo-developmental state,” “the networked polity,” “the developmental networked state,” or “the embedded autonomy industrial bureaucracy,” the same broad model is advanced. This model suggests that for a state to initiate successful industrial development, especially in a technologically intensive industry, it must cultivate interaction and a dynamic division of labor with the local industry. To accomplish this, the state needs to have and retain an ability to make and implement decisions in the national interest; be informed about the needs, abilities, and difficulties of the industry so that it can tailor its policies accordingly and refrain from policy initiatives that limit the ability of the industry to develop capabilities on which it can base long-term growth; and change its policies in tandem with the changing needs of the industry. Rather than long-term planning, the state needs a flexible structure that enables it to quickly change and implement different policies as quickly as possible when industrial conditions change.^[13] In addition, the neodevelopment argument implicitly calls for the state to manage a process in which it lets the industry gain more and more power to decide its own future as it finds its feet and grows. Hence the state needs to be able to change its role from that of initiator and leader to that of a supporting actor.^[14]

However, the neodevelopmental arguments are still structural and hence fail to take the new plurality of the world into account. The neodevelopmental writers all contend that if a state manages to have the specific necessary state structure, economic development will follow. The main difference between the various neodevelopmental statists lies in the specification of the necessary structure, not in the theoretical argument or causal mechanisms.

The neodevelopmental statists argue that in order to succeed in the developmental efforts in high-technology industries, the bureaucracy needs, first, to have multiple ties with, and embeddedness within, industry and finance, both locally and globally.^[15] Indeed, recent writers in this line argue that the main attribute of a successful developmental state in the 1990s was its network structure. They argue that the ability of the state to successfully motivate industrial development stems from a structure in which different agencies are deeply networked into different social-industrial-capital networks (Ansell 2000, O’Riain 2004).

Nonetheless, they still view embeddedness as one specific option. Thus, within their framework, a state can either be a networked polity or not. The reality, however, is much more complex. There are many ways by which state and industry can interlink, and each one of them necessitates a different division of labor and gives rise to different industrial capabilities. Furthermore, the political process of becoming embedded itself has direct influence on both the success and the shape of the industry to be, in particular, when we take into account the duality of this embeddedness both within and outside the national borders. This is a critical point, which the neodevelopmental statists at best disregard.

Second, as soon as we discuss the need for a flexible bureaucracy that understands and quickly and correctly reacts to the needs of an ever more technologically sophisticated industry, the issue of skills comes into play. To be effective in the intricate dance of embeddedness and rapid innovation–based industrial growth, the bureaucracy needs skills that enable it to understand the complex and changing needs of a highly technological industry. This requirement, coupled with the need to deeply attach itself within the industry, means that we can no longer view the developmental bureaucracy as an ideal-type Weberian bureaucracy with long stable career patterns and skills and capabilities taught solely within the organization. In addition, if the arguments are that (a) the developmental agencies need to be able to act independently from other state agencies and (b) the developmental agencies need to be flexible and dynamic, constantly changing their policies in step with the needs of the specific and different industries each agency aims to nurture, it follows that we must also change our view of the state.

Thus, in contrast to the developmental- and neo-developmental-state arguments, we no longer can view the state as a unitary actor. The neodevelopmental theories themselves lead us to focus on specific parts within the organization of the state and analyze each one in line with the specific policy domains and industrial sectors we study. If we are to follow the neodevelopment arguments to their ultimate conclusion, the state itself is to be viewed as multiple groups of development bureaucracies, each with unique capabilities, structures, and powers, bundled and embedded within society. Indeed, even a cursory look at Ireland, Israel, and Taiwan reveals three very different bureaucratic structures and cultures, with very different politics of industrial development.^[16]

Nonetheless, in much the same way as the old development statists, the new developmental-state theorists still portray the state as a unitary and underdefined variable: the state is portrayed either as a success, and therefore a flexible/ networked/embedded developmental state, or as a failure, and hence not a neodevelopmental state. An example is Sean O’Riain’s recent study of the Irish software industry, where he defines a flexible development state: “The flexible developmental state (FDS) is defined by its ability to nurture post- Fordist networks of production and innovation, to attract international investment, and to link these local and global technology and business networks together in ways that promote development” (O’Riain 2000, italics added).

The tendency toward a success-or-failure definition of states is the result of two common traits of recent studies. The first is the use either of a macro level of analysis of industrial development or of partial sectoral analysis looking at one specific industry or region and generalizing to the national level. The second trait is the tendency to focus on either successes or failures without attempting to analyze these at the micro level. This tendency can be attributed to the underdefined nature of the neo-developmental-state concept, which necessitates the classification of the cases studied using only a success-failure dichotomy. This is true even in studies that urge us to look at the different bureaucratic constellations of agencies and power.

Thus, for example, Vivek Chibber, comparing the development efforts of India and South Korea, argues that the Korean state was successful and that the Indian state was a failure. He does so even when arguing that Korean development efforts were successful because the Korean state had a “nodal” development agency with power not only over the industry but also over other agencies. Thus even with an argument locating the reason for success at the level of industrial agencies and constellation of power within the state, Chibber still treats the state as a unitary actor and industrial success as a binary success-or-failure variable (Chibber 2002).

I find this especially surprising because the neodevelopmental statists’ arguments, in a very similar line of logic with our proposed framework, strongly suggest that the same state, having a structure where different agencies are embedded into different social-industrial-capital networks during their attempts to nurture diverse industries, can both succeed and fail in different sectors. It is quite possible that it is no longer the “state” that is neodevelopmental or not, but rather different parts of it. If indeed there are many ways to organize ties between the state and industry, many ways in which to organize R&D efforts, and no way in which to fully and strategically plan the industrialization process, then industrial systems that grow out of these diverse efforts will be different. The neodevelopmental statists, however, still propose one “best” model of development, focusing their explanations around specific state structures. Therefore their argument marginalizes the importance of politics.^[17]

Another way in which the neodevelopmental statists diminish the role of politics in their theories is by not accounting for the dynamic process of coevolution within which the state builds its “networked structure” and embeddedness. Even if the argument is that there is only “one” best way to be embedded, then it is important to explain how this embeddedness takes place.

It becomes crucial to explain this once the assumption is made that there is more than one way to be “networked” and that the different ways in which the state relates to industry have long-term consequences. Ignoring the politics of state-industry co-evolution is a significant oversight in the case of emerging economies. Since private industry lacks the skills, the capital, and sometimes the will to enter new high-technology industries, many times the primary role of the state is to directly intervene and create the industry. Hence it is imperative for sustained industrial growth that the state manages the dynamic political process of moving from a position of hierarchical power into a position where it acts more as the supporting actor. These processes are critical in molding the final shape of each particular industrial system, and as such they need to be accounted for. To be useful any theoretical framework needs to account not just for the idea of embeddedness but for the processes that lead, or fail to lead, to different kinds of embeddedness within and outside the state’s border.

Since the old and neo-developmental-state theories are not focused on explaining the various outcomes of diverse industrial policy regimes, they do not supply tools to understand and predict these differences and to gauge their influence and importance. Instead, their main efforts are devoted to showing the importance of the state in industrial development and to distinguishing their model of state-industry interactions from that of the old development state theories. Indeed, “The State” might have been brought back “in,” but only at the price of throwing politics “out” (Evans et al. 1985).

In order to understand how different policy choices lead to different industrial systems, we need to merge our insights about the plurality of embeddedness models with an understanding of rapid innovation–based industrial production. Although the neo-developmental-state writers claim to explain growth in post-Fordist industries, their argument proceeds without attention to the radical ways in which industrial production has changed since World War II. Recall that the old developmental school was built around the coherent Gerschenkronian institutional theory of industrial development in relatively backward societies, and it was built on an understanding of industrial technology and production management. Thus the old developmental-state school not only explained but also predicted a whole set of complementary institutions seen in many successful NICs, such as high saving rates and the strategy of industrial conglomerates–based growth. In contrast, the neodevelopmental statists do not propose or integrate a theory of economic and industrial development. This explains why such writers cannot account for a particular set of institutions arising in a specific society or industry, nor for how certain institutions influence economic growth trajectories and behavior in global markets.

This is a major lapse in the neo-developmental-state arguments, as it also denies them one of the mainstays of the developmental-state theories: the intricate explanation of the specific ways in which developmental states’ control over finance allowed them to influence industrial development (Zysman 1983). Many developmental-state writers have gone as far as to call control over finance the “nerves” of the developmental state; others have argued that answers to questions about financial resources and policies provide the best indicators for states’ ability to realize industrial goals (Evans et al. 1985, Woo-Cumings 1991, 1999). It is a glaring omission that the neodevelopmental school does not have a coherent theory of how states’ different financial initiatives shape the growth of different high-technology industries.

To summarize, by analyzing and debating the late-development school, we have gained an understanding of some of the components of the needed theoretical framework. Now we know much more about the need for embeddedness as well as what level of analysis we should seek. Moreover, we now understand that there is a need to gain insights into how different processes of embeddedness, influenced by the varying structures of bureaucratic skills and information and the different roles played by the state to interlink both locally and globally, lead to particular industrial systems. Lastly, we understand that to do so we have to gain a new understanding of industrial production, particularly in the case of rapid innovation–based industries.

[edit] Globalization as Fragmentation: Global Production Network Theory

The intensified movement of capital, trade, services, and, to some degree, people across national borders, starting in the 1970s, gave rise to a multitude of studies. Of particular interest among these studies is a specific line of research that tries to understand the international reconfiguration of industrial activities. Pioneered by economic geographers, these studies highlight two interlinked, novel features of the latest globalization: the growth of worldwide production networks, and the increasing fragmentation or “deverticalization” of the production process. Interestingly, this process is accompanied by a new pattern of geographical industrial clustering, that of similar activities of the production process (Arndt and Kierzkowski 2001a, Feenstra 1998, Gereffi 1994, 1996, Sturgeon 2000, 2002).

Today we see rising spatial specialization in particular stages of the product manufacturing chain and a rising international interfirm (not just intrafirm) trade in components (not only in final products). Within the electronics industry, for example, there has been a major transformation in the way final products are manufactured and sold. In the United States, new leading companies such as Dell, Sun, Cisco, or Microsoft’s Xbox division do not even have their own manufacturing facilities but concentrate on R&D, high-level product design, sales, marketing, and, at most, final assembly. Old companies such as Hewlett-Packard (and Compaq before their merger), 3Com, and IBM have been shedding a large percentage of manufacturing capabilities and opting for manufacturing outsourcing. Even such components as hard-drives are now manufactured in discrete stages in different locations globally (Gourevitch 2000, Kenney and Florida 2004, McKendrick et al. 2000, Sturgeon 2000, 2003). In fact, a whole new class of high-technology companies—the so-called fabless IC design houses—took their name from their lack of facilities to fabricate their own chips.

Many writers use the term product chain to describe this growing “deverticalization.” However, product chain may imply linear manufacturing processes of specific products from basic inputs to final assembly. The reality, in contrast, is of manufacturing processes built from multiple relationships between suppliers, each with a different power structure between actors involved in the production of numerous products. These products may be the final products and/or components for other products. For these reasons I prefer to use the term production networks, which implies many suppliers with different relationships producing multiple products. Especially in the case of the IT industry, the term global production networks describes reality much better than product(s) chains, with its linear implications.^[18]

There are many causes of the growing fragmentation of production: lower transportation costs, lower telecommunication and information costs that in turn decrease the cost of off-site coordination and control, technological advances that enable the codification of component manufacturing and allow off-site manufacturing, and the de- and reregulation of trade. This fragmentation process has also been the cause of, and in turn is further expanded by, the delinking of the innovation, design, and marketing and sales processes from the production process. The delinking of production stages is now evident in almost every industry in the world, from textile to auto manufacturing, and from toolmaking to electronics (Fuller et al. 2003, Gereffi 1999, Gereffi and Korzeniewicz 1994, Herrigel and Wittke 2004, Sturgeon 2000, 2003, Sturgeon and Lester 2004, Yeats 2001).

The growth of the global production network raises questions about its effects on developmental strategies of nations and regions (Arndt and Kierzkowski 2001b, Bathlet et al. 2002, Berger 2006, Gereffi 1999, Gereffi et al. 2005, Henderson et al. 2002, Ruane and Gorg 2001, Sturgeon and Lester 2004). It is becoming more and more apparent that what are moving from country to country are specific phases of production rather than the production of complete products. Thus product-cycle theories, which predict that as products and industries become old and commoditized they will move from the more-advanced economies to the less-developed, are becoming less relevant (Akamatsu 1962, Cumings 1984, Vernon 1966).

This fragmentation occurs not only in production but also in R&D (Dunning 1994, Gassmann and von Zedtwitz 1999, Gerybadze and Reger 1999, Pearce 1999, Reddy 2000). A study by the MIT Industrial Performance Center analyzing the results of two surveys conducted in 1991 and 1999 on all the companies in western Europe, Japan, and the United States with R&D expenditures topping $100 million in 1991 current dollars (totaling 244) found a remarkable growth in the tendency of big companies to rely on external sources of R&D. While in 1991 fewer than 10 percent of American firms, 20 percent of European firms, and 40 percent of Japanese firms reported a high reliance on external sources for R&D, by 1999 the figure was above 80 percent for all surveyed firms. Moreover, the geographical range and distance of firms’ sources of R&D had grown tremendously. The average in 1991 of the percentage of total R&D activity outside the home region was 26.8 percent for European companies, 24.3 percent for American firms, and 4.6 percent for Japanese ones; in 1999 European and American companies reported estimates of more than 33 percent of total R&D done outside the home region, and Japanese companies reported more than 10 percent (Roberts 1999).

There has been no study that systematically examines how product fragmentation shapes the development of emerging industrial countries. Such an account would need to explain why some activities occur in specific locations, as well as clarify whether particular development strategies resonate better with specific stages in the global product networks. What the literature offers are only suggestions for research frameworks on the governance and power relations in different types of product chains, or case studies about particular industries in specific locations.

For our purpose a promising point of departure is to look at the two dynamic processes that reinforce fragmentation: production-stage specialization and capability building, and production-stage economies of scope and scale. These are the main reasons for the rapid diffusion of fragmentation and the growth of global production networks. If we can analyze these processes to gain an understanding of how different emerging countries’ industrial policies give these countries advantages in specific stages of production, we will have the beginning of an answer.

Production-stage economies of scale and scope is the term I use to describe the process by which, once a specific production network fragments into discrete stages, suppliers in each stage, by pooling the demand of many customers, create economies of scope and scale that in-house manufacturing divisions cannot. These economies of scope and scale enable suppliers to become more efficient and allow them to profitably operate on margins that are significantly lower than those achieved by in-house manufacturing divisions. This in turn allows them to further lower their prices while offering the same or even higher quality, further speeding the trend toward outsourcing of this stage’s manufacturing activities (Sturgeon 2000, 2002).

Production-stage specialization is the term I use to describe the process by which product fragmentation leads companies to develop superior capabilities in particular stages or components of the product network. A recent example of this in the IT industry is provided by the pureplay foundries. Pureplay foundries, such as the Taiwanese TSMC and UMC, are semiconductor companies that deal solely in chip fabrication, and whose revenues come from fabricating chips according to the designs of their customers. Such specialization enables companies to become better and more efficient in this narrow set of activities.^[19] It also helps them to acquire specialized capabilities and knowledge that more–vertically integrated firms could not. These capabilities, once acquired, enable these firms to excel in innovation around the particular production stages and set of components they focus on, using their superior skills and unique knowledge. Over time these two related advantages, in skills and innovation capabilities, grant these companies even more advantages vis-à-vis in-house division of vertically integrated companies. We observe this process both in high-technology areas—for example, in the manufacturing of specialized cards or chips for the PC, such as graphic cards or memory chips—and in more traditional industries, such as the bicycle, where the product network fragmentation allowed one company, Shimano, to become the innovator and market leader in drive-train components (Galvin and Morkel 2001).

The fragmentation of production and the growth of global product chains suggest that there are multiple entry points and ways to succeed, even in the same industry and even during the same time period. Global production networks allow emerging economies as many different innovation-based entry points to the global IT industry as there are stages in the global IT industry production network.

I argue that the decisive factors that influence the entry point of each national industry are:

1. the relationships of the industry with global markets, specifically with the MNCs that build, manage, and to a large degree control these product chains, and
2. the development of capabilities that enable the industry to excel in specific activities which are more or less suitable to specific stages of production.

Thus, focusing on the development and growth of global production networks, theories of product fragmentation enable us to formulate arguments as to how different states’ science and technology policies build national industries that do better in different production stages.

Nonetheless, in order to complete our framework, we still need to account for how state-business interaction and the implementation of specific policy regimes translate into institutional systems that, through time, facilitate the development of particular capabilities and skills by private firms and motivate them to employ specific business models. We need to understand how different industrial innovational capabilities are created, maintained, and employed.

[edit] Systems-of-Innovation Theories

From the point of view of comparative political economy literature, the systems-of-innovation theories should be seen as part of the varieties-of-capitalism theories. Those theories assume many different models of national economies, each with distinct institutional systems. These systems confer different economic and industrial advantages and disadvantages on companies operating within them. As a result, these companies, in turn, behave differently even when operating in the same product markets and under the same international conditions.

Systems-of-innovation theories aim to explain how systematic differences in industrial as well as science-and-technology institutional systems give rise to industries with specific industrial technological-innovation capabilities. Systems-of-innovation theories were first developed in the 1980s by evolutionary economists (Carlsson et al. 2002, Edquist 1997, Freeman 1987, Lundvall 1992, Nelson 1993). The writers in this tradition build on Schumpeterian economics and on institutional economic theories. They argue not only that innovation is the main source of long-term economic growth but also that there is a major role for policy in R&D and science. Government has such a major role, they argue, for two reasons: first, the innovation process itself is iterative and cooperative in nature; hence there is a significant role for public actors (Braczyk et al. 1998, Cooke and Morgan 1998, Lester and Piore 2004, Morgan 1997, Piore et al. 1994, Piore and Sabel 1984, Teubal 1996, 2002). Second, building on Arrow’s seminal paper, they argue that scientific research, especially industrial R&D, represents a clear case of market failure. This market failure occurs because the indivisibility, inappropriability, and high uncertainties of R&D lead private investors to allocate suboptimal amounts of finance to research under conditions of perfect free competitive markets (Arrow 1962, Schumpeter 1961 [1934]).^[20]

Researchers in the systems-of-innovation tradition have argued that the different relationships between government, firms, research institutions, suppliers, and customers influence the ways in which new technologies are created and/or diffused throughout the industrial system. The relationships also affect the rates at which new generations of products using the latest technology are produced (Braczyk et al. 1998, Breschi and Malerba 1997, Lundvall et al. 2002).

These constellations influence firm behaviors via three main mechanisms. The first comprises the shaping of firm capabilities, especially their dynamic economic capabilities and innovation-production/absorption capacities—that is, their capabilities to produce new products and technologies or to infuse new technologies into their existing products and manufacturing systems (Carlsson and Eliason 1994, Carlsson et al. 2002, Cohen and Levinthal 1989, 1990). Second, institutional structures influence firm behavior by offering different opportunity structures and different resources which influence the profitability of certain activities (Carlsson 1995, Carlsson and Eliason 1994, Giuliani 2002, Morgan 1997, Teubal 2002, Teubal et al. 1991). Third, institutional structures influence firms’ R&D behavior by infusing the system with specific research paradigms and business models that move certain R&D activities and procedures to the forefront and others to the background (Dosi 1982, Samuels 1994).

There are two main strands of systems-of-innovation theories. The American strand focuses more on the influences emanating from the different ways in which R&D is organized and financed (Nelson 1993). The European school, originated at Ålborg, has been more focused on explanations based on backward links between producers, and on the flows of information from users (or user sectors) to producers, and has a stronger emphasis on the role of the quality of demand on the development of capabilities; that is, greater attention is given to the correlation between the sophistication level of customers and users and the overall innovation capability of their suppliers (Lundvall 1992, Lundvall et al. 2002).^[21]

Systems-of-innovation theories implicitly argue that three main variables explain the amount and intensity of industrial R&D of each system, all intimately connected to the fact that industrial R&D is a semipublic good. The first variable is the location of industrial R&D activities within the industrial system. The levels of industrial R&D sophistication and capabilities differ according to whether most of the industrial R&D is conducted by the firms themselves or by public and semipublic organizations like research institutions and universities. Systems-of-innovation theories contend that there are major differences between systems depending on the location (public or private) and the identity of the agents that typically conduct most of the R&D. While most writers do not deal with these issues in detail, the logical conclusion of their argument is that the more R&D that is conducted by private firms, the more private industry develops sophisticated R&D capabilities.

The second variable is financing. Different financing modes affect not only the location of R&D activities but also the focus (time horizons and product versus process innovation) and amount of industrial R&D. Again, systemsof- innovation theories do not fully elaborate their arguments on the subject but rather suggest that differences between public and private financing, the financial gains that investors are looking for, and the origin of the financiers directly influence the R&D capabilities and business models employed in each industrial system. The more resources that are available for innovative product R&D, the argument goes, the more R&D is conducted and the more sophisticated capabilities are developed. Similarly, the more there are sectoral limits on R&D activities, which inhibit the development of broader or multidisciplinary capabilities and knowledge, the more the scope of R&D capabilities will be limited. Following the same logic, an implicit conclusion is that the more that financing is linked with global capital and exchange markets, the more investors will seek financial exits that allow them to channel the profits back home.

Accordingly, I argue that foreign or foreign-financed VCs will push firms into business models that aim at either going public on the leading stock exchanges or being bought by an MNC. As these tend to utilize a business model emphasizing new-product innovation, I contend that the more that the local industry is dependent on global venture capital financing, the more the industry will focus on product innovation and follow a business strategy aimed at foreign IPOs or merger-and-acquisition deals.^[22]

The third explanatory variable highlighted by systems-of-innovation theories is the industrial-opportunity structure: the composition of the local industry, including the links among producers, between producers and customers, and between the local industry and the global production networks, in particular, with the MNCs that control them. Different opportunity structures affect the capabilities and business models employed by firms by giving different industrial systems access to different information and skills, as well as by giving each local industry different sets of customer focus, business prospects, and market-niche openings. A critical factor in this configuration concerns the local leading companies. In addition, as the main market for IT has always been the United States, and as American MNCs (with some exceptions) also control the global markets, a key to understanding the different opportunity structures is to look at the different relationships that the local IT industry has with these MNCs, both within and outside national borders. This has become especially true with the growing fragmentation of production.

MNCs (and leading local firms) are critical for the overall capabilities composition of each national IT industry in part because they are the main customers of new local IT companies. MNCs are also key actors in the transfer and development of two critical spheres of skills and capabilities. First, the R&D labs of MNCs develop products and components on the technological cutting edge. Consequently, they are a major source of cutting-edge technological information and knowledge and, even more important, of world-level R&D-conducting capabilities for the industrial system in which they operate. Second, MNCs are the best organizations in the world in designing and developing new products. Therefore their R&D divisions diffuse into the industrial systems within which they are located the capabilities needed to design new products as well as better knowledge of the customers in the final market.

For these reasons, major differences in industrial-development patterns are to be found in IT industries with different leading companies and different relationships and links with the leading MNCs. If the MNCs’ subsidiaries and leading companies within a specific country are conducting manufacturing activities with a focus on process and manufacturing innovation, then their outputs into and inputs from the industrial systems are significantly different than if most of the MNCs’ subsidiaries and leading local companies conduct critical R&D activities. Thus the outputs that MNCs and leading local companies seek from local firms and the inputs they put into the local industrial systems, as well as the ability of local firms to sell their own products and services to these MNCs globally, vary depending on the activities MNCs perform locally.

For example, in the case of MNCs’ manufacturing subsidiaries, local firms might find it easier to sell manufacturing services or become component suppliers to MNCs than to become suppliers of new technologies and innovations. The inputs of the leading domestic companies and MNCs into the local industrial system would be focused around manufacturing and process information and skills, including critical innovational capabilities around these activities. For the same reasons, local firms will find it harder to sell the same MNCs new technologies, and the MNCs will diffuse less R&D and fewer product-design capabilities into the local industrial system.

In short, the global production network and the systems-of-innovation theories give us a better understanding of the institutional underpinning of industrial development and innovation in a world of geographically fragmented industrial production. Using these insights, together with the lessons we have learned from late-development and developmental-state theories, we can now go back to our framework and develop a more comprehensive argument about the role of the state in the development of rapid innovation–based industries in emerging economies.

[edit] Building the Theory

In this book I develop an approach to understanding the role of the state in industrial development in a world of globalized, fragmented production. I do so by presenting arguments about the roles and structure of the state in the development of high-technology industries in emerging economies. Hence we must identify first how states can organize and interact with private industry to promote the growth of innovative high-technology industry, and then the political process of this interaction. Second, we must understand how specific policies and state-business interactions produce different national industrial systems within the same industrial sector.

[edit] A New Model of Rapid Innovation–Based Industrialization Politics

From analyzing and debating late-development and developmental-state theories, we have learned that, in the development of high-technology industries, the state should act as the flexible facilitating agent, not as an overall commander. The state should no longer methodically plan the development of strategic industries by choosing specific products and product niches, forcing private companies to enter them and supplying some of the necessary finance and technologies. Instead, the state should concentrate on creating more broadly defined technological capabilities, and should focus on motivating private agents to work in these areas and to collaborate with one another and with the state. The state’s role is no longer to make the decisions and compel private companies to follow them, but rather to motivate private companies to make long-term commitments to operate in rapid innovation–based industries and activities.

Furthermore, in the case of rapid innovation–based industrialization, the state needs not only to spur the creation of the industry but also to quickly grant the industry control if not full jurisdiction as it develops. At the same time, the state still needs to retain enough knowledge and skills to devise and implement policies that help the industry to grow and maintain its competitiveness in the world of constant technological change.

We should picture the state, therefore, as the national management consultant and central arbitrator and promoter of the industry, with the distinction between public and private roles and power becoming more ambiguous. Stateindustry relations, I argue, should be seen as a division of labor, of the tasks each side takes to reach the same national goal of economic growth. In this way the state is seen as embedded—that is, as an important agent but not the overall leader—in a network of private and public agents that together makes the industrial system.

In contrast to the neodevelopmental writers, I contend that in order to gain full understanding, we should fully account for the importance of sectoral politics. I argue that it is not enough to see embeddedness as the one final result. First, we need to explain how embeddedness develops in various industrial sector and the political process behind it. In addition, we need to confront the realization that there are multiple forms of state-industry networked relationships, each leading to a different industrial system. For example, we need to be able to explain why the Taiwanese hardware industry flourished while the software industry languished, although the two sectors had the exact same formal state-industry structures and rules of conduct.

Hence our framework needs to take into account the many differences between the developmental bureaucracies of different states, their policies, and the ways these policies are implemented. Furthermore, it needs to explain how these different modes of state-industry embeddedness and unique bureaucratic structures and cultures bring about different trajectories of industrial development. Failing to offer an explanation for these is a critical fault of the current developmental-state theories. In this book I explain these phenomena.

Lastly, we are missing a coherent framework that can help us to understand how and why, through the course of rapid innovation–based industrial development, a particular set of institutions emerges in a specific society or industry, as well as how certain institutions shape economic growth trajectories and behavior in global markets. The framework provided in this book rectifies this gap by utilizing insights from the systems of innovation and global production network theories.

There is a need for a new model that delineates the role of the state in the politics of industrialization. While this theoretical framework should be built on the neo-developmental-state arguments, it should take into account the multiple ways in which successful development can simultaneously take place within the same industries; accordingly, there are alternatives, with different consequences with which to create and from which to choose.

The idea of network development serves as the basis for my argument. My framework proposes that in successful cases of rapid innovation–based industrial development, state agencies will first aim at creating a set of firms and industrial actors, then seek to develop a thoroughly multiplexed network among these actors, and between these actors and the international financial and production markets and networks (Ansell 2000, Locke 1995, O’Riain 2004, Teubal et al. 1991).

In effect, successful cases of rapid innovation–based industrial development are the result of different political processes, in which the state’s first role is as a key actor in the creation of a network. At first, the network is hierarchical, but in the course of co-evolution I expect to find the network becoming denser and more heterarchial or policentric and international, with the state moving from a position of controlling the network to a position of centrality— that is, becoming an important node within a bigger and much more multiplexed, larger, and diffused network. Consequently, the state’s main role becomes more that of a facilitator-organizer than of an overall commander. Further, I argue that the different ways local and sectoral politics shape the development of these networks vastly influence the industrial structure and capabilities.

Rapid innovation–based industrial development failure takes two forms. The first occurs because of the inability or unwillingness of the state to facilitate and participate in the growth of this industrial sector. An example of that kind of a failure would be the state’s ignoring the industry, largely refusing to participate in its growth or to channel the resources necessary to that effect.

A second kind of failure occurs when the state actively participates in the creation of the network but then resists relinquishing its power, trying instead to retain its position of power and control, aiming to maintain hierarchical relationships with private industry.

Overcoming these dangers is not easy and requires much more than state will. Indeed, to successfully manage the dual-staged process above, the state needs to acquire two assets: (a) sufficient technological and scientific skills, knowledge, and information to make informed decisions in the case of ever more complex industrial technologies development; (b) multiple relationships with the industry that enable the development agencies to be constantly informed about the shape and needs of the industry, as well as to implement its decision without resorting to coercion. Moreover, it needs to have and preserve its ability to make decisions independently. As I have noted, acquiring these two assets, managing these transitions, and developing the needed industry-state co-evolution processes suggest a model of the state that is incompatible with the Weberian model of the ideal-type bureaucracy. In addition, as there are many ways to acquire the needed skills and create embeddedness within the industry, as well as many ways to manage the relationship with the industry, even within the same state, it follows that there are many different models of the new development state. The interventionist who looks for the recipe for success may be disappointed that there are multiple options available to the state. However, for the state that is striving for economic growth, the temporal and structural indeterminacy of solutions is a hopeful message.

We can now propose a new model of successful state efforts for rapid innovation– based industrialization. The new model suggests a bureaucracy that is less Weberian and isolated but much more fragmented. However, it is also more technologically and scientifically knowledgeable and skillful, as well as closer to industry. The state’s structure must permit the developmental agencies to react quickly to the needs of the industry, allowing the state to nurture industrial development without itself seizing control over that development. While the old model conceptualized the state as a unitary actor with personnel recruited, trained, and indoctrinated within the bureaucracy, the new views the state less as a monolith and shifts attention to its different agencies. It follows that differences in the ways the bureaucracy is constructed, the social origin and educational background of its personnel, and its ability to shape the decision making process must be analyzed.

There are many ways in which the developmental bureaucracy can acquire the needed skills and embeddedness within industry (fig. 1.1). One approach is for the state to be involved as deeply as possible in the industrial R&D process, bringing the technology-creating agents into its own structure. Of the three cases, Taiwan’s utilization of two state-controlled public research institutions— the Industrial Technology Research Institution (ITRI), and the Institute for Information Industry (III)—as the main impetus for IT industrial growth, is an example of such a strategy. A second approach is to create a development bureaucracy with porous borders and to employ a “revolving door” recruitment-and-training strategy, enabling scientists and industry leaders to move back and forth from state to private industry. Of our cases, the way in which Israel’s main developmental agency—the Office of the Chief Scientist in the Ministry of Trade and Industry (OCS)—has been structured is an example of such a strategy.^[23] The third approach is a variant of the second, in which the state attempts to preserve the cohesiveness and Weberian principles guiding the developmental agency by creating specific subunits, usually with a more limited or advisory role, in which regular recruitment and training procedures do not apply and into which industry insiders are recruited on a temporary basis. Of our cases, Ireland’s utilization of such subunits, of which the National Software Directorate has been the most influential, within larger development agencies—the Industrial Development Agency (IDA) and Enterprise Ireland (EI)—that embody Weberian ideals, is an example of that strategy.^[24]

The new framework differs from the old in the ways in which the developmental agencies create, conduct, and maintain their relationships with industry. In the old model the developmental state was portrayed as having strong relations with the leading big firms. While dialogue was maintained, the state had the authority and responsibility of guiding and controlling the industry’s development. In the new model, the state aim is to become bundled within, and create tight relational networks with, the industry; industry-state dialogues are to be between equals; and the state’s role is more of a facilitator than a planner and overall leader. However, unlike in the old theories, which described a fairly straightforward relationship and power structure, there are many ways in which the state can accomplish its roles and create the relationships that the new framework emphasizes. The shaping of these choices is done through a constant political process.

The choices available to the state can be conceptualized along two orthogonal axes. On one end of the first axis, measuring control, are attempts by the state to get maximum control over development and direct industrial R&D efforts, from basic research all the way to specific products. On the other end of this axis we have a state formulating high-technology industry–specific policies only after a few companies have already proven the growth potential of specific subsectors, with the state’s direct assistance limited to more mature firms.

State decisions as to the degree to which it should target sectors and technologies can be conceptualized on a second axis. On one end are states that formulate policies down to the level of defining specific generic products and technologies. On the other are states that see their role mainly as assisting in the realization of decisions made by private firms, employing horizontal technological policies that are neutral with regard to technologies and sectors (Teubal 1983, 1997). In between there are states that target specific sectors— for example, software—but do not attempt to target specific technologies or define future products.

In our cases we have an example of the three main positions on these two axes, both extremes and the middle (fig. 1.2).

On the one hand we have Taiwan, where the state has been targeting specific technologies and products, authorizing and financing its public research institutions to develop them to the stage of working product prototypes and then either deliver the result to the industry or spin off the research teams as companies to commercialize the results. On the other hand, we have Israel, with a development agency that defined its role as fixing the market failures associated with industrial R&D and maximizing product R&D activities. The Israeli state followed almost ideal-type neutral horizontal technological policies. It implemented a policy of giving R&D grants for product ideas developed by private companies and entrepreneurs in every industrial branch, helping companies in all stages of development. In the middle is Ireland, which did not have a specific industrial policy for rapid innovation–based industries until after the indigenous software industry achieved some success globally. Ireland then restructured its development agencies and policies to focus on specific industries and tailored its financial instruments to companies that, while still young, have already moved beyond the seed stage.^[25]

The last dimension in which the new model of the state differs significantly from the old is in its role in relation to the global industry and financial markets. In the old model the state had three main tasks. The first was to organize and facilitate technology transfer deals from the leading MNCs to local industry. The second was to use its control over foreign currency to shape industrial development, granting it to chosen firms who were exporting successfully and denying it to others. The third task of the state was to prevent MNCs from gaining control of too large a percentage of the domestic market and crowding out local infant industries.

In our framework, the state’s roles are much more complex. In a world of fragmented industrial production, vertical integration is no longer the sole option, nor necessarily the best one. Instead, there are many ways in which the local industry can become a part of the global production networks. In each of these, the local industry needs to develop specific relationships with the MNCs and with the global financial markets, and for each kind of relationship different technological and managerial skills are necessary. In addition, as the local industry is based more and more on indigenous R&D activities, the role of the state in strategically managing technology transfer, while it might still prove important in the early stages, is diminishing. These decisions relate to the heart of national politics: questions about national ownership, identity, and foreign relations, as well as financial security and independence.

Again, it is useful to think about state choices for enhancing the local industry’s specific links with, and embeddedness within, the global industry and financial markets as two orthogonal axes. The first looks at the state’s behavior toward foreign technology-based firms, in particular the MNCs that develop and control the global production networks. On one extreme, we have states that look at foreign firms as the main mode for industrial upgrading and focus on persuading foreign firms and MNCs to locate manufacturing facilities within their borders. They hope that these firms’ subsidiaries, once established, will become embedded within the domestic industry. On the other extreme we have states that focus most of their efforts on persuading MNCs to locate their R&D labs within their borders, and that pay close attention to the development of specific relationships of local companies with their foreign partners.

On one extreme we have Ireland, which until the early 1990s focused its industrial policy almost solely on bringing MNCs to open manufacturing facilities in Ireland, without paying much attention to creating strong relationships between them and the local industry, within or outside Ireland. On the other extreme lies Israel. By generating collaborative projects between Israeli IT firms and American companies, the Israeli state has established relationships in which domestic firms focus on product R&D and their American partners focus on sales and marketing. Israel’s developmental agencies have also been attempting to lure foreign firms to open R&D centers in Israel. Taiwan has chosen a middle course, first urging MNCs to open manufacturing facilities in Taiwan, but then goading them into buying a growing percentage of their components from, and outsourcing their manufacturing activities to, Taiwanese companies.^[26] At the same time, Taiwanese developmental agencies have tried to help local companies produce more components and improve their manufacturing-and-design-services capabilities.

Relationships with global financial markets are better conceptualized along a second axis. On one end we have states that try to link their local industry with global financial markets. They court foreign investors not only to back local firms but also to be the financiers of the local VC industry, and encourage local companies to get publicly listed on foreign stock exchanges. On the other end we have states that try to limit, as much as possible, the financing of their high-technology industry to local investors, and try to limit, if not outright prevent, the public listing of local companies on foreign stock exchanges. In between, we have states that either are less active or aim to strike a balance between the two (fig. 1.3).

Of our three cases, Israel has done the most to lure foreign VC investors and urge public listing on foreign exchanges. State efforts have been a main cause behind the current situation: the vast majority of Israeli start-up financing, either directly or through local VC funds, is foreign. In addition, Israeli companies all aim to be listed on foreign exchanges, making Israel, with only six million people, the state with the largest number of companies listed on NASDAQ other than the United States and Canada. On the other extreme is Taiwan, focusing all its technology financing initiatives on the local investment community, and strongly advancing the Taiwanese Stock Exchange. Ireland has chosen the middle road: channeling European Union (EU) money to the local VC industry but creating the new funds mostly through local financial institutions.

[edit] Different Choices: The Varied Paths of Industrial Growth

Conceptualizing the new model of state-industry relationships helps us to elucidate the different ways in which emerging states can structure themselves and their relationship with the local industry in order to nurture and support the growth of the high-technology industry. We can now consider how the different directions taken by states influence their high-technology industrial development path. Using these insights we can explain how in the cases of Taiwan, Ireland, and Israel specific state policies and behavior have helped to spur the development of different IT industries and to generate the relative success and failure of each subsector within each of the three cases.

The spatial clustering of similar production activities and the underlying linked dynamics of production-stage specialization and production-stage economies of scale and scope point us to several domains of interest. First, specific industrial R&D, dynamic economic capabilities, and business models are more suitable for certain stages of the production process. Second, the analysis of the dynamics of production-stage specialization and economies of scale and scope indicates that industrial systems that specialized in a specific stage of production tend, if successful and with time and effort, to become ever more competitive and innovative in these specific stages. Consequently, if we want to understand the different evolutionary processes of high-technology industrial growth we need to understand, for each of our cases:

• how industrial R&D and dynamic economic capabilities are created, improved, and maintained;
• the growing utilization of certain business models; and
• the development of specific relationships that the local industry has with the global industrial and financial markets.

In effect, we need to analyze the development of industrial R&D capabilities, capacities, and opportunity structures in each of our cases.

In order to understand the growth of industrial R&D capabilities, we compare the different location of the R&D conducting agents, the R&D resources available and the way in which firms are financed, the scope and limits of R&D, and the industries’ customers and relationships.

With regard to opportunity structures, infrastructural variables being more or less equal, two main variables—the industry’s customers and markets and the ways in which the industry is financed—strongly affect the business models and strategies firms and entrepreneurs employ. This leads industrial agents to focus on the building of specific innovative and manufacturing capabilities. In a world in which most of the trade in IT is done through global production networks, there are two key factors in determining their opportunity structure: the relationships that local IT firms have with local leading firms and with foreign companies, especially leading MNCs; and the sources and forms of finance. One question is what forms of financing are open to technological entrepreneurs? Another question is whether the source of the capital is local or global and how deeply it is tied to foreign-exchange markets.

We can now focus on four critical decision areas to analyze how different behavior and structure of the state influences the development of the IT industry’s R&D capabilities in each of our cases:

1. Decisions by the state on how to acquire the necessary R&D skills are influencing the identity of the agents conducting industrial R&D. The more the state bureaucracy encompasses R&D conducting agents within its structure, the fewer industrial R&D activities private firms are encouraged to take.
2. Decisions by the state over the level of control over the technological development path of the industry, including the decision of how, and whom, to finance, bear on both the R&D resources available to the industry and the scope of R&D activities taken.
3. State efforts toward developing the first local leading companies have long-term consequence for the industry’s opportunity structure.
4. Decisions by the state concerning the involvement of foreign firms and investors within its national borders, as well as decisions about whether to enhance specific relationships between local and foreign companies outside states’ national borders, affect the resources, the particular feedback and information the industry gets from its main customers, and the diffusion and development of specific innovative capabilities.

It is my contention that private firms in the high-technology industry of emerging economies have more limited R&D capabilities when the state has tried to control the industrial R&D development path, has encompassed the R&D conducting agents within its structure, has supplied only limited resources, has refrained from urging foreign companies to conduct R&D activities within its borders, and has limited the involvement of foreigners as financiers. Yet neither Taiwan nor Ireland nor Israel followed these idealized paths. In each case a particular mix of policies and initiatives across these different dimensions has been played out. Therefore the full answer requires a more refined level of comparison, fully elaborated in Chapters 2–4.

I have three goals for this book. The first is to expand our understanding of the role, and the limits, of the state in the development of rapid innovation– based industries in less-developed economies. The second is to explain how the different approaches taken by Israel, Ireland, and Taiwan shaped the IT industries, influenced the particular development path of and the particular capabilities developed by each industry, and moved each industry into a different position in the IT industry’s global production network. Lastly, by showing that emerging countries have real alternatives with regard to RIB industrial growth, I argue that politics and the political struggles of crafting and choosing between these alternatives, as well as the politics of state-industry co-evolution, are important not only in explaining the difference between success and failure but in identifying the reasons behind the diverse paths to success. By shifting our explanations from timing and structure reasoning, I aim to show that both “state” and “politics” need to be explained if we want to understand both policy formulation and its outcomes.

[edit] Notes

1. ↑ For the purpose of this book, success in rapid innovation–based industrial growth is defined as the development of an industry based around NTBFs, which contribute a significant amount of growth in gross domestic product (GDP), exports, and employment, making the industry a critical sector of the national economy. I use the common definition of NTBF from the literature: a company whose products are new technologies or are based on their own and others’ R&D effort to commercialize applications of new technology, or a company whose main revenue stream is based on R&D efforts to develop new technologies. I further elaborate on the definition by treating an IT NTBF as a company whose main product or products are either a new technology or a new application of new technology whose main function is the gathering, transferring, processing, or presenting of information, and where the company is conducting the R&D that leads to the product itself, either with or without cooperation with other organizations.
2. ↑ For two recent examples of such research, one from the varieties of capitalism perspective and one from the development-paths perspective, see Guillen 2001, Hall and Soskice 2001.
3. ↑ On Japan’s and South Korea’s conglomerate-based industrial development, see, for example, Amsden 1989, Anchordoguy 1989, Calder 1993, Evans 1995, Fields 1995, Johnson 1982, Okimoto 1987.
4. ↑ For a collection of theoretical and empirical articles elaborating more on the argument that small states will have significant and unique problems in developing innovative high-technology industries, see Freeman and Lundvall 1988.
5. ↑ In Israel annual overall sales in the high-tech electronic industry, excluding software, more than septupled from 1986 to 2004, from $2.003 billion to more than $15 billion. Exports rose even faster, from $1.135 billion to $13 billion. Software sales had the highest growth rates, rising from overall sales of $380 million in 1989 to more than $4 billion in 2004. Software exports grew from $5 million in 1984 to more than $3 billion in 2004. Ireland, a country that lost 1.1 percent of its total population in 1989 due to economic emigration, has seen its IT industry achieve remarkable growth. Software was the main sector where indigenous companies, and not only Irish subsidiaries of MNCs, achieved worldwide success. The number of indigenous software companies in Ireland rose from 291 in 1991 to 630 in 1998, while the total number of firms rose from 365 to 760. Revenues in 2004 reached a total of 16.9 billion euros, out of which exports were around 16 billion euros. The total sales of Irish-owned companies in 2004 were 1.369 billion euros, a marked improvement from 1991, when total revenues were 2.198 billion euros, out of which only 191 million euros came from the Irish-owned sector. In Taiwan, a country in which the first IC chip was fabricated in a demonstration fabrication facility of a public research institution only in 1977, its IT companies manufactured, in 2001, 70 percent of global production of motherboards, 55 percent of laptops, 56 percent of LCD monitors, 51 percent of color-display-tube monitors, and 36 percent of digital still cameras. In addition, with 225 companies and NT$147.8 billion in sales, Taiwan’s IC design subsector is the second largest in the world after the United States’. In 2002 the total sales of the Taiwanese electrical and electronics industry reached almost $88 billion USD, and on top of that the semiconductor industry added another $21.4 billion. Sources: Israel, CBS 2001, 2004a, IAEI 2006, IASH 2003, MOTI 1998; Ireland, IDA 2000, NID 2006; Taiwan, ITRI 2003, MoEA various years. As the SEC filing of the leading Israeli software firms were higher than the reported sale figures of the Israeli Association of Software Houses, there is reason to believe that the figures above are underestimated.
6. ↑ Of the 482 interviews (not counting repeats), 137 were conducted in Ireland or with Irish companies and officials in the United States, 110 were conducted in Israel or with Israeli companies and officials in the United States, and 235 were conducted in Taiwan or with Taiwanese companies and officials in the United States. Unless otherwise stated, data on company profiles and information were gathered through interviews, SEC filings, the Hoover online database, and companies’ Web sites. Because many interviewees from private companies have requested anonymity not only for themselves but for their firms, I decided that when possible I would not specify which of the companies described in this book were interviewed.
7. ↑ While the interviews were open ended, an interview theme was adhered to consisting of forty questions organized around eight subthemes, with all questions touched upon during the one- to two-hour-long interview. The interviews’ transcripts were then uploaded onto a server running custom-made software that allows access to the extended database through a Web portal with enhanced search engine capabilities.
8. ↑ Together with Rostow, Gerschenkron gave a paradigmatic all-powerful stature in economic development studies to the notion that economic development should be seen almost as an organic process that must pass through certain specific stages before achieving full blossom, with the ability to innovate on the scientific edge being the last capability to develop (Rostow 1960). A more recent proponent of the linear model in regard to innovation and the IT industry in Asia was Linsu Kim, with his influential model of innovational catch-up progression (Kim 1997). For an argument that changes in the 1990s made this linear model of sequential innovational progression less compelling in Asia, see Ernst 2005b.
9. ↑ See, for example, O’Riain 2004, pp. 6–8, or Amsden, proposing the same theoretical argument in her recent works (2001, p. 2).
10. ↑ Johnson’s pilot agency argument has a long history of being reinvented and elaborated under different names in the literature; for two recent examples, see Chibber 2002, Weiss 1998.
11. ↑ Among Japanese specialists a disagreement with Johnson’s assessment of the state’s extreme power and importance evolved. For examples of studies that strongly disagree with Johnson, see Calder 1993, Freeman 1987, Okimoto 1987, Samuels 1988, 1994; for examples of studies that only slightly disagree with Johnson, or even agree with him, see Anchordoguy 1989, Chinworth 1992. Nonetheless, all these studies agree with the basic Gerschenkronian model of industrial growth, and all agree that the state was highly important in the development of the industry in Japan.
12. ↑ Parts of this version of this economic-growth theory have been sanctified by the World Bank (WB 1993).
13. ↑ There are some differences among the neo-developmental-state theorists, in particular between those who argue that the state should continue to advance national champions, and those who argue for a strategy based around SMEs. It is unclear whether these differences arise because of the different context of location, timing, and industries that the different writers researched. However, in the critical part of theorizing about the optimal role and behavior of the state in its attempt to spur the growth of high-technology industries and in their treatment of the state for all practical purposes as a unitary actor, the authors are similar enough to be treated as advancing one model.
14. ↑ The reason why I argue that this point is implicit and not explicit in the neo-developmental-state theories is that while some writers—for example, Peter Evans—present a typology of the different roles of the state, none of the neodevelopment statists pays enough attention to the political process of the state initiating the creation of the industry and then managing its own retreat from a commanding and planning position to a more supporting and less controlling role. Nonetheless, these exact political processes determine both whether sustained success is achieved, and how different models of industrial systems are shaped.
15. ↑ As such the neo-developmental-state theorists are different from the earlier critics of the strong developmental-state argument, such as Samuels’s model of the politics of reciprocal consent. These earlier critics’ main aim was to reinquire into the exact roles of the state and the nature of state-business relations in the development of industries that are not based on rapid innovation, at a period before the advent of globalized fragmented production, not to propose a new model of the developmental state suitable for rapid innovation–based industries (Samuels 1988).
16. ↑ Just using the simple Evans-Ruach measurement of “Weberianism” in the structure of the bureaucracy in regard to meritocratic recruitment and career patterns reveals some surprises. While Taiwan and Ireland score very high (12 and 12.5 respectively on a 14-point scale), Israel’s score is a relatively low 7 (Evans and Rauch 1999, O’Riain 2004, pp. 145–146). Israel’s score equals that of Morocco and Turkey and is below that of Mexico, Brazil, Sri Lanka, and Egypt. (Ireland’s score was evaluated by a different study using the Evans-Ruach measurement, an evaluation which is strengthened by my own analysis; see Chapter 4.) This result suggests that the strong coherent Weberian bureaucracy hypothesis does not hold a lot of generalizable explanatory power over the three cases. This is before we even note that bureaucracies are very different in their traditions and social roles, as well as in the social origins, educational backgrounds, and influence on the political decision making process of their members.
17. ↑ For example, Sean O’Riain, a neodevelopmental statist who specifically looks at the growth of SMEs in the software industry of Ireland, still argues for one specific model of development and state behavior that he terms either as Flexible Developmental State or as Developmental Networked State, which is different from the old developmental-state model that he terms Bureaucratic Developmental State (O’Riain 2000, 2004).
18. ↑ For papers outlining the different terminology used in the literature, see Henderson et al. 2002, Sturgeon 2001.
19. ↑ In effect this process takes Adam Smith’s argument about the division of labor, specialization, efficiency, and the size of the market one step further (Smith 1937, in particular book 1, chapters 1 and 3).
20. ↑ Market failure is defined as a case where due to lack, or asymmetries, of information and/or the specific characteristics of the industry or activity, we should expect a free market to allocate resources in suboptimal fashion.
21. ↑ For an excellent example of a theoretical model for the importance of the creation of a users-innovators community to the successful implementation and continuous improvement of new technology, see Teubal et al. 1991.
22. ↑ Systems-of-innovation theories usually do not pertain to the important national difference between the regulation of venture capital and stock exchanges. Thus while they argue for a more general observation, claiming that VC financing leads to the rapid diffusion and use of American-like start-up businesses models, I prefer to refine their argument. Thus I claim that only VC funds whose source of finance is the global capital markets and that are linked to the leading global stock exchanges would affect the business development in such a way.
23. ↑ Since 2003 the ministry is called the Ministry of Trade Industry and Employment.
24. ↑ In 2003 the National Software Directorate was renamed the National Informatics Directorate.
25. ↑ That is not to say that the state was not important in the early development of the industry in Ireland; indeed, the Industrial Development Agency (IDA)—then focused mainly on foreign direct investment (FDI)—was critical in the industry’s development in the 1980s through its Enterprise Development Plan. However, as the name suggests, the aim of that policy initiative was to develop business in all sectors of the economy, without regard to R&D, and its main goal remained job creation.
26. ↑ In 2001 the Taiwanese Department of Industrial Technology (DOIT) embarked on an initiative specifically to lure MNCs to open R&D centers in Taiwan. This initiative has been quite successful, with thirteen MNCs opening facilities in Taiwan within two years and ten more MNCs on the waiting list. Indeed, if this new trend is sustained, and these MNCs consistently conduct high-level R&D activities in Taiwan, a very interesting future research project would be to gauge their influence on the Taiwanese IT industry after a decade or so of operations.

This site made possible through the generous support of the Alfred P. Sloan Foundation