Technology and growth
The Sources of Economic Growth, by Richard R. Nelson. Cambridge, Mass.: Harvard University Press, 1996, 328 pp.
Jane Sneddon Little
Although many economists agree that technology, broadly defined to include organizational ability and culture, is a major contributor to economic growth and that investments in physical and human capital and R&D are likely to entail sizable mutually reinforcing spillovers, our understanding of the interactions among the ingredients of growth and of the relative importance of the various components of technology remains remarkably limited. Columbia University's Richard Nelson has been a trailblazer who can help guide exploration of these promising areas. The 10 previously published essays in The Sources of Economic Growth cover a range of topics, including the author's perspectives on the shortcomings of orthodox growth theory, Joseph Schumpeter's understanding of growth as a disequilibrium process, patent policy, the links between science and invention and between U.S. universities and industrial R&D, the rise and fall of U.S. technological leadership, and differences in national innovation systems. Although the topics are diverse, the essays all serve to support Nelson's major theses-that technological advance is the primary driver of economic expansion and that social institutions mold and are molded by innovation and growth.
Because Nelson sees innovation as the key to growth and stresses the likelihood of strong interactions among the variables determining growth, his views are, he points out, more like those of the new growth theorists (such as Paul Romer of Stanford Unversity and Gene Grossman of Princeton University) than of the neoclassical economists (including Robert Solow at MIT and Dale Jorgenson at Harvard). Whereas the former seek to model the effort required to advance technology and view growth as an endogenous process, the latter see technology as a residual that explains the part of the increase in output that cannot be credited to the accumulation of physical or human capital. Nelson, however, distinguishes his views from those of the new growth theorists who, he claims, view the economy as shifting in predictable ways from one equilibrium to another. He sees the economy as being in a constant state of disequilibrium, engendered by technological change that involves a path-dependent but fundamentally uncertain evolutionary process. Thus, history matters, but so, Nelson argues, do institutions, because innovation and growth proceed through the interactions of a complex set of organizations-public and private, rival and cooperative. As the foregoing may suggest, although the book does not contain a single mathematical equation, it is highly theoretical and the approach is analytical rather than anecdotal.
One model's limitations
The first chapter, "Research on Productivity Growth and Productivity Differences: Dead Ends and New Departures," is one of the most thought-provoking and helpful pieces in the collection. Originally published in the Journal of Economic Literature in 1981, the chapter serves as a reminder to current researchers that many scholars writing in the 1950s and earlier made useful observations about the growth process that are worth revisiting. Nelson uses this literature review to argue that the orthodox model underlying most research on productivity growth over time or across countries is superficial or even misleading in several respects. He suggests that research based on the neoclassical model has reached a stage of sharply diminishing returns, not just because it ignores important variables such as management skills or institutional characteristics, but also because the model obscures some of the central features of productivity growth. For instance, treating technology as a freely available public good may actually impede analysis of how technology is created and spread.
Starting with a look at firm-level productivity, Nelson underscores the importance of managerial skills, labor relations, and the norms of the shop floor, as well as the technology found in blueprints and work designs. As he points out, all of these diverse factors find their way, undistinguished, into the residual in orthodox growth models. Turning next to the process of technological advance, Nelson stresses the fundamental uncertainties facing inventors and investors, the competitive nature of much R&D activity, and the importance of learning by doing as a complement to R&D. In the case of technology diffusion, Nelson reminds us that in the original neoclassical model, technological innovations are immediately available across the entire capital stock, whereas in the "vintage capital" version, which accounts for the age of the capital stock, the pace of capital investment limits the pace of diffusion. But neither variant acknowledges the role of uncertainty, the proprietary nature of some innovations, and the feedback between R&D and experience with using the innovation.
Finally, in reexamining the sources of growth, Nelson emphasizes that if factors such as human and physical capital and technology are complements, then the growth of one input augments the marginal contribution of the others, and dividing up the credit for growth (as is done in a growth-accounting exercise) makes little sense. In his view, capital equipment "carries" new technology, and educated workers facilitate learning by doing and technology diffusion. Nelson ends the essay by suggesting an evolutionary growth model that recognizes uncertainty, choice, competition, and imitation. However, as he never describes this model in any detail, his contention that the biggest impediment to using such a model for empirical work is the lack of an appropriate microlevel database seems an overstatement. Moreover, this impediment no longer looks quite as formidable as it did in the early 1980s, now that researchers are beginning to have access to the wonderfully rich data in the Census Bureau's Longitudinal Research Database, which in turn is linked to the Worker-Employer Characteristics Database and the Survey of Manufacturing Technologies, among other data series. Using these sources, researchers can at least start to explore the links between worker education, say, and manufacturers' capital spending patterns and technology adoption.
The nature of technical advance
Several of Nelson's essays on the innovative process and the links between science and technology should particularly interest policymakers as well as academics. Whereas macroeconomists using orthodox growth models have had little to say about the nature of innovation, economic historians have offered a wealth of fascinating anecdotes that underscore the fundamental role of uncertainty in technological progress. This emphasis on the extremely unpredictable nature of technlogical advance tends to discourage efforts to analyze the process and suggests a strictly limited role for public policy. Although Nelson also stresses the importance of uncertainty (in the sense that innovators cannot know all the options available), his analytic approach curbs the role of chance, allowing him to make some useful observations about how innovation tends to proceed.
In "The Role of Knowledge in R&D Efficiency," for instance, Nelson demonstrates that better understanding of effective R&D tests and strategies, the results of recent R&D efforts, and the available scientific and technological building blocks improve R&D efficiency by narrowing the set of candidate projects and broadening the set of possible components for future innovations. The relevant knowledge derives from academic research in the basic and applied sciences in universities and corporate laboratories and from the R&D process itself. Nelson is particularly intrigued by the observation that basic science occurs in corporate as well as university settings and that researchers from rival firms find ways to share the science resulting from their work, to the social good, while keeping its applications proprietary. He also suggests that scientists are often able to predict quite closely the nature of the practical advances permitted by their basic research. Nevertheless, although Nelson advocates that the close links between university and industry found in defense, health, and agriculture be broadened, he warns against asking scientists to make production or commercial decisions they are ill-equipped to handle. Universities excel at research and training, he maintains; industry at product and process development and improvement. Fruitful cooperation requires recognizing the relative strengths of both types of institution.
Policymakers concerned with granting or interpreting patents might be particularly interested in "On Limiting or Encouraging Rivalry in Technical Progress: The Effect of Patent-Scope Decisions." This essay argues that a broad patent discourages outsiders from participating in subsequent rounds of innovation and thus slows progress. Moreover, if a technology is cumulative or involves a system of components developed independently, then broad patents may create an innovative impasse.
Nelson finds broad patents in science-based technologies such as biotechnology and optics especially problematic. Although evolutionary growth theories suggest that technological races in which many inventors see the same goals and use approximately the same means are not very common, they are not unusual in the science-based technologies, which, Nelson believes, are becoming increasingly important. When an invention simply represents a first "practice" of leads provided by publicly funded scientific breakthroughs, then it is especially important that a patent's scope be limited. Nelson notes that seeking patents "on the prospects defined by the identification and purification of particular DNA fragments reveal(s) the problem at its worst."
The key role of firms
The final pair of essays deals with international differences and convergence in innovation systems and economic performance. If I could arrange it, all officials handling technology and trade policy would read "National Innovation Systems," a summary of a study of innovation in 15 countries. From this study, Nelson concludes that the bulk of the work required by innovation must be done by firms themselves; that exposure to international competition benefits innovation, even in the United States; and that the record of national industrial and protection policies promoting high-tech industries is at best uneven. The most promising role for government seems to be in promoting the development of strong skills and encouraging firms to compete in world markets.
The other essay, "The Rise and Fall of American Technological Leadership," coauthored with Gavin Wright, seems dated, however. The authors argue, plausibly enough, that this country's postwar technological dominance reflected its large domestic market and its relatively large investments in scientific and technical education and R&D. Their claim that the globalization of commerce, investment, and technology has eroded this country's unique position is also persuasive. Still, although the authors' comments about the decline of U.S. technological leadership and its relatively slow growth over the past two decades may have fit the mood of the early 1990s, they do not match current perceptions that Japan has become a mature economy encumbered with many highly inefficient industries, such as distribution and financial services; that the Europeans are struggling with extremely high unit labor costs and thus high unemployment rates; and that the United States has undisputed technological leadership in newly important areas such as information technology, communications, and biotechnology. This misreading reflects the authors' failure to follow their own advice and give proper weight to the macroeconomic environment. For example, they offer this country's declining share of global high-tech exports between 1970 and the late 1980s as evidence of its eroding technological leadership without noting the impact of the dollar's huge appreciation in the first half of the 1980s. Focusing on manufactured exports from our increasingly service-based economy, they also overlook the fact that shifts of resources from declining to growing sectors must affect our trade data. Finally, they ignore the impact of Japan's regulatory system and of Europe's inflexible labor markets even though Nelson flags these institutions as being generally important in determining a country's innovative prowess.
This abrupt change in perceptions concerning the United States' relative strength illustrates how hard it is to analyze the sources of economic and technological growth. As Nelson writes, the essays provide "an analytic framework, not wide enough to encompass all of the variables and relationships that are likely to be important, not sharp enough to tightly guide empirical work, but broad enough and pointed enough to provide a common structure in which one can have some confidence." They start us down the road most students of economic growth want to travel, but the road is long and full of twists and branches.
Jane Sneddon Little is an assistant vice president and economist at the Federal Reserve Bank of Boston.