Sources of Technical Change:
Induced Innovation, Evolutionary
Theory, and Path Dependence
This is an appropriate time to take stock, as economists, of our understanding of the determinants of the rate and direction of technical change. The 1960s through the 1980s produced considerable new theory and empirical insight into the process of technical change. In the 1960s and 1970s, major attention was focused on the implications of changes in demand and in relative factor prices. In the late 1970s and early 1980s, attention shifted to evolutionary models inspired by a revival of interest in Schumpeter's insight into the sources of economic development. Since the early 1980s, these models have been complemented by the development of historically grounded “path-dependent” models of technical change.
Each of these models has contributed substantial insight into the generation and choice of new technology. It appears to me, however, that each research agenda is approaching a dead end. In this chapter I argue that the three models— induced, evolutionary, and path-dependent—represent elements of a more general theory. The purpose of this chapter is to review the development of the three models to identify their complementarity and to suggest how they might be incorporated into a more general theory.
Three major traditions of research have attempted to confront the impact of change in the economic environment on the rate and direction of technical change. The “demand-pull” tradition has emphasized the relative importance of market demand on the supply of knowledge in inducing advances in technology. A microeconomic approach was built directly on an early observation by Sir John Hicks (1932:124–125) that a change in the relative prices of factors of production is a spur to invention. A macroeconomic growth theoretic tradition stems from attempts by economic theorists to understand the apparent stability in factor shares in the American economy during the twentieth century in spite of the very large substitution of capital for labor. In the language of the “new growth economics,” each of these approaches could be described as an attempt to interpret the process of technical change as at least partially endogenous.
Schumpeter, whose writings have been exceptionally important in influencing the way economists think about technical change, made a sharp distinction between invention (and the inventor) and innovation (and the innovator): “Innovation is possible without anything we should identify as invention, and invention does not necessarily induce innovation but produces itself . . . no economically relevant effect at all” (Schumpeter 1934:84). The Chicago sociologist S.C. Gilfillan viewed invention as proceeding under the stress of necessity, with the individual innovator being an instrument of luck and process (Gilfillan 1935).
In his now classic study of the invention and diffusion of hybrid maize, Griliches demonstrated the role of demand in determining the timing and location of invention (Griliches 1957). Schmookler, in a massive study of patent statistics for inventions in four industries (railroads, agricultural machinery, paper, and petroleum), concluded that demand was more important in stimulating inventive activity than advances in the state of knowledge (Schmookler 1962, 1966). The Griliches–Schmookler demand-induced model received further support from papers by Lucas (1967) and Benzion and Ruttan (1975, 1978) that showed technical change to be responsive to aggregate demand. In the mid-1960s, Vernon (1966, 1979) introduced a demand-pull model to interpret the initial invention and diffusion of consumer durable technologies (e.g., automobiles, television, refrigerators, and washing machines) in the United States versus other developed countries. His interpretation came just as the United States was about to lose its dominance in several of these technologies to Japan.
Arguments about the relative importance of the role of demand-side forces and supply-side forces, such as advances in knowledge, in inducing advances in technology were intensified in the late 1960s. A study conducted by the Office of the Director of Defense Research and Engineering (1969) purported to show that the significant “research events” contributing to the development of major weapons systems were predominantly motivated by military need rather than disinterested scientific inquiry. This view was challenged in studies commissioned by the National Science Foundation that, not unexpectedly, found that science events were of much greater importance as a source of technical change (Thirtle and Ruttan 1987:6–11).
In a review of the “demand-pull/supply-push” controversy, Mowery and Rosenberg argue that much of the research purporting to show that technical change has been demand induced is seriously flawed (Mowery and Rosenberg 1979). They insist that the concept of demand employed in many of the studies has been so broad or imprecise as to embrace virtually all possible determinants. Rosenberg also insists that the demand-pull perspective has ignored “the whole thrust of modern science and the manner in which the growth of specialized knowledge has shaped and enlarged man's technological capacities” (Rosenberg 1974). Research conducted from a demand-pull perspective appears to have atrophied since the late 1970s, partly as a result of the Rosenberg criticism.
Careful industry studies, such as the study of innovation in the chemical industry by Walsh, suggest that both “supply and demand factors play an important role in innovation and in the life cycles of industries, but the relationship between the two varies with time and the maturity of the industrial sector concerned” (Walsh 1984:233). A rigorous econometrics study by Scherer (1982) that simultaneously tests both the demand-induced and supply-push hypotheses across a broad range of industries confirms the earlier Schmookler finding of a strong association between capital goods investment and invention. However, Scherer found a weaker association between demand pull and industrial materials inventions. He also found that the introduction of an index of technological opportunity based on the richness of an industry's knowledge base added significantly to the power of his model to explain differences in the level of inventive activity among industries.
It should no longer be necessary to insist that basic research is the cornucopia from which all inventive activity must flow to conclude that investment in the generation of scientific and technical knowledge can open up new possibilities for technical change. Nor should it be necessary to demonstrate that advances in knowledge, inventive activity, and technical change flow automatically from changes in demand to conclude that changes in demand represent a powerful inducement for the allocation of research resources.
Modern interest in the effect of factor endowments on the direction of technical change dates to the early 1960s. Hicks had previously suggested:
The real reason for the predominance of labor saving inventions is surely that which was hinted at in our discussion of substitution. A change in the relative prices of the factors of production is itself a spur to invention and to inventions of a particular kind—directed at economizing the use of a factor which has become relatively expensive. (Hicks 1932:124–125)
Hicks’ suggestion received implicit assent but little attention until the early 1960s. In his work on the theory of wages, Rothschild repeated the Hicks argument (Rothschild 1956:118, 176). In a book on economic growth, Fellner argued that firms with some degree of monopsony power had an incentive to make “improvements” that economized on the progressively more expensive factors of production, and that expectations of future changes in relative factor prices would be sufficient to induce even firms operating in a purely competitive environment to seek improvements that would save the more expensive factors (Fellner 1956:220–222; see also Fellner 1961, 1962).
An intense dialogue concerning the issue of induced innovation by economic theorists in the 1960s and early 1970s was triggered by Salter's explicit criticism of Hicks’ induced technical change hypothesis. Salter insisted that “at competitive equilibrium each factor is being paid its marginal value product; therefore all factors are equally expensive to firms” (Salter 1960:16). He went on to argue that “the entrepreneur is interested in reducing costs in total, not particular costs or capital costs. When labor costs rise any advance that reduces total cost is welcome, and whether this is achieved by saving labor or saving capital is irrelevant” (Salter 1960:43–44; see also Blaug 1963). It is difficult to understand why Salter's criticism attracted so much attention except that students of economic growth were increasingly puzzled about why, in the presence of substantial capital deepening in the US economy, factor shares to labor and capital appeared to remain relatively stable. The differential growth rates of labor and capital in the US economy were regarded as too large to be explained by simple substitution along a neoclassical production function. Below, I discuss a series of papers published in the 1960s that successfully refuted Salter's argument and established a solid theoretical and empirical foundation for the induced technical change hypothesis.
The debates about induced technical change centered on two alternative models— one a growth theoretic approach and the other a microeconomic version. The most formally developed version was the growth theoretic approach introduced by Kennedy (1964). The Kennedy article initiated an extended debate on the theoretical foundations and the implications of incorporating the process of induced technical change into the theory of economic growth (Samuelson 1965, 1966; Kennedy 1966; Drandakis and Phelps 1966; Wan 1971).
In the Kennedy model the initial conditions include (1) given factor prices, (2) an exogenously given budget for research and development (R&D), and (3) a fundamental trade-off (a transformation function) between the rate of reduction in labor requirements and the reduction of capital requirements. The model assumes a production function with factor-augmenting technical change. Kennedy cast his analysis in terms of the effect of changes in relative factor shares rather than changes in relative factor prices on bias in invention because of the growth theory implications.
The following example from Binswanger represents an intuitive interpretation of the Kennedy model:
Suppose it is equally expensive to develop either a new technology that will reduce labor requirements by 10 percent or one that will reduce capital requirements by 10 percent. If the capital share is equal to the labor share, entrepreneurs will be indifferent between the two courses of action. . . . The outcomes of both choices will be neutral technical change. If, however, the labor share is 60 percent, all entrepreneurs will choose the labor reducing version. If the elasticity of substitution is less than one, this will go on until the labor and capital shares again become equal, provided the induced bias in technical change does not alter the (fundamental) trade-off relationship between technical changes that reduce labor requirements on the one hand, or capital requirements on the other. (Binswanger 1973, 1978a:32)
The Kennedy variant of induced innovation was subsequently incorporated into neoclassical growth theory (Wan 1971). Nordhaus notes
Until recently, only Harrod-neutral (or purely labor-augmenting) technological change could be introduced into neoclassical growth without leading to bizarre results. Neoclassical growth models were “saved” from such restrictiveness by the introduction of the theory of induced innovation. Under the usual neoclassical assumptions and, in addition, when the innovation possibility curve takes the form assumed by Kennedy and Samuelson the system settles down into a balanced growth path exactly like that of the labor-augmenting case. (1973:209)
By the early 1970s the growth theoretic approach to induced technical change was under severe attack (Wan 1971; Nordhaus 1969:93–115, 1973; David 1975:44–57). Nordhaus notes that in the Kennedy model no resources are allocated to inventive activity. A valid theory “of induced innovation requires at least two productive activities: production and invention. If there is no invention then the theory of induced innovation is just a disguised case of growth theory with exogenous technological change” (Nordhaus 1973:210). He further notes that the Kennedy innovation-possibility frontier (IPF) implies that the rate of capital-augmenting technological change is independent of the level of labor augmentation. Thus, as technological change accumulates there is no effect on the trade-off between labor- and capital-augmenting technological change (Nordhaus 1973:215). He insisted that the model is “too defective to be used in serious economic analysis” (Nordhaus 1973:208). The growth theoretic version of induced innovation has never recovered from the criticism of its inadequate microeconomic foundation.1
A microeconomic approach to induced innovation, built directly on Hicksian foundations, was developed by Ahmad (1966). His criticism of the growth theoretic approach initiated a vigorous exchange (Fellner 1967; Ahmad 1967a, 1967b; Kennedy 1967). In his 1973 critique, Nordhaus mentioned that Ahmad was the only person to attempt to formulate the theory of induced technical change along microeconomic lines, but he did not comment explicitly on the Ahmad paper or on the subsequent exchange.2
In his model, Ahmad employed the concept of a historic innovation-possibility curve (IPC). At a given time there exists a set of potential production processes, determined by the basic state of knowledge, available to be developed. Each process in the set is characterized by an isoquant with rather narrow possibilities for substitution. Each process in the set requires that resources be devoted to R&D before the process can be actively employed in production. The IPC is the envelope of all unit isoquants of the subset of those potential processes which the entrepreneur might develop with a given amount of R&D expenditure.
Adapted from Ahmad (1966), Figure 1 amended.
Assume that It is the unit isoquant describing a technological process available in time t and that IPCt is the corresponding IPC (Figure 2.1). Given the relative factor prices described by line PtPt, It is the cost-minimizing technology. Once It is developed, the remainder of the IPC becomes irrelevant because, for period t + 1, the IPC shifts inward to some IPCt+1. This occurs because it would take the same R&D resources to go from It to any other technique on IPCt as to go from It to any technique on IPCt+1. If factor prices remain unchanged and technical change is neutral, the new unit isoquant will be It+1 on IPCt+1. If, however, factor prices change to Pt+1Pt+1, then it is no longer optimal to develop It+1. Instead, a technological process corresponding to some I′t+1 becomes optimal. In the graph, Pt+1Pt+1 corresponds to a rise in the relative price of labor. If the IPC has shifted neutrally, I′t+1 will be relatively labor saving in comparison to It.
Ahmad's graphical exposition is a useful illustration of the induced innovation process of a one-period microeconomic model in which a firm or a research institute has a fixed exogenous budget constraint. When research budgets are no longer fixed, a mathematical exposition is more convenient (Binswanger 1978a:26–27).3 In a multiple-period model the shift from It to I′t+1 would occur in a series of steps in response to incremental shifts from Pt to Pt+1. One way of describing such an incremental process would be to appeal to “learning-by-doing” and “learning-by-using” concepts (Arrow 1962; Rosenberg 1982).
The initial dialogues about the logic of the Kennedy–Samuelson–Weizscker growth theoretic and the Hicks–Ahmad microeconomic approaches to induced technical change were conducted within the confines of the standard two-factor (labor and capital) neoclassical model. Among economic historians there has been an ongoing debate about the role of land abundance on the direction of technical change in the industrial sector. Among agricultural economists there has emerged a large literature on the bias of technical change along mechanical (labor-saving) and biological (land-saving) directions.
Habakkuk (1962) argued that the ratio of land to labor, which was higher in the United States than in Britain, raised real wages in American agriculture and thereby increased the cost of labor to manufacturers. He argued, in effect, that in the nineteenth century, the higher US wage rates resulted not only in the substitution of capital for labor (more capital) but in induced technical changes (better capital) biased in a labor-saving direction as well. The issue became controversial among economic historians even before they became fully sensitive to the emerging theoretical debates of the 1960s concerning the issue of induced technical change or the earlier empirical work by Hayami and Ruttan (1970, 1971).
The criticisms of the Rothbard–Habakkuk labor scarcity theses by Temin (1966) and the debates that his criticism engendered (Fogel 1967; Ames and Rosenberg 1968; David 1973, 1975:24–30) focused primarily on the issue of the impact of land abundance on the substitution of capital for labor— the “more capital” rather than the “better capital” part of the thesis. David argued that economic historians “steered away from serious re-evaluation of the proposition about the rate and bias of innovation, precisely because standard economic analysis was thought to offer less reliable guidance there than on questions of the choice of alternative known techniques of production” (David 1975:31).
David insisted that the argument could not be resolved without a more intensive mining of the historical evidence. But recourse to measurement could not be expected to get very far without a theoretically grounded definition of an operational concept that distinguishes between choice of technology and technical change and between bias in the direction of technical change and the rate of technical change. David argued that this can be done by embracing “the concept of a concave, downward sloping ‘innovation-possibility frontier’ . . . along the lines of the neoclassical theory of induced technical progress due to Kennedy, Weizsacker and Samuelson” (David 1975:32). He then went on to argue along the same lines as Wan (1971) and Nordhaus (1973) that the particular pattern of changes in macro-production relationships observed in the United States could not be rationalized within the framework of a stable IPF: “While shifts of the innovation-possibility frontier are entirely conceivable, the necessity of accepting their occurrence in this context signifies a practical failure of the underlying theoretical construct. For the latter treats the position of the frontier as established autonomously for each economy, and has no explanation to offer for it” (David 1975:33).
David also insisted that bias in the direction of technical change could only be understood by building a theory of induced innovation on microeconomic foundations consistent with engineering and agronomic practice. To David this also meant abandoning both neoclassical growth theory and the neoclassical theory of the firm. Furthermore, it would be necessary to incorporate the intimate evolutionary connection “between factor prices, the choice of technique and the rate and direction of global technical change” (David 1975:61).
In attempting to develop a non-neoclassical “evolutionary and historical” approach to induced innovation, David introduced the concepts of (1) linear fixed-coefficient processes or techniques from activity analyses (which he credits to Chenery) and (2) a latent set of potential processes that could be designed with the currently existing state of knowledge (which he attributed to Salter). He then added (3) localized learning, which directs technical change toward the origin along a specific process ray (which he attributes to Stiglitz), and (4) a probablistic learning process that is bounded by transition probabilities that depend on the firm's initial technical state (David 1975:57–86).
The model of the search process appears to have been inspired by the Nelson– Winter evolutionary model (see Section 2.6). David insisted, however, that his transition probabilities, in which past states—the firm's initial myopic selection of a technical process—influence the future course of development, are “clearly non-Markovian” (David 1975:81). He also distinguished the mechanism that accounts for the evolutionary nature of technical change from the form employed by Nelson and Winter.
David differentiated his approach from neoclassical production theory by suggesting that substitution may involve an element of innovation. This is similar to the mechanism that Ahmad (1966) and Hayami and Ruttan (1970) had earlier employed to account for the shift in the IPC (or in David's terms, the frontier production function). It should be viewed as an extension rather than an alternative to the neoclassical model.
When he turned to the technical relationships among natural resources, labor, and capital, drawing on the work of Ames and Rosenberg (1968) and his own earlier work (David 1966), David argued that in the mid-nineteenth century mechanical technology and land were complements: “The relevant fundamental production functions for the various branches of industry and in agriculture did not possess the property of being separable in the raw materials and natural resource inputs; instead the relatively capital-intensive techniques . . . were also relatively resource using” (David 1975:88). Greater availability of natural resources facilitates the substitution of capital for labor. “Thus, even if the same labor/capital price ratios had faced producers in Britain and America, the comparatively greater availability of natural resources would have suggested to some American producers the design and to others the selection of more capital-intensive methods. . . . In America the on-going capital formation spurred by the greater possibilities of jointly substituting natural resources and capital for labor may well have been responsible for driving up the price of labor from the demand side” (David 1975:89–90). The formal introduction of the role of relative resource abundance (or scarcity) clearly represents an important extension as compared to the traditional two-factor (labor and capital) neoclassical models. But the primary significance is that David opened the door to, and identified most of the elements of, what has since become known as the path-dependent model of technical change (David 1975:65, 66).
There are substantial differences in the extent to which the several induced technical change models have been tested against empirical data. The demand-induced model was developed in close association with empirical studies and was not subjected to formal modeling or theoretical critique until fairly late (Lucas 1967; Mowery and Rosenberg 1979).4 The growth theoretic version of factor-induced technical change has produced very little empirical research. The only test against empirical data seems to have been that of Fellner. Fellner interpreted his results as indicating that, except during periods of very rapid increase in rising capital intensity, and hence rapidly rising demand for labor, the induced labor-saving bias was sufficient to prevent the labor share from rising (Fellner 1961).
In contrast, the microeconomic version of factor-induced technical change has stimulated a wide body of applied research. The first formal test based directly on microeconomic foundations was the Hayami–Ruttan test against the historical experience of agricultural development in the United States and Japan (Hayami and Ruttan 1970). It seemed apparent that neither the enormous differences in land/labor ratios between the two countries nor the changes in each country over time could be explained by simple factor substitution. Hayami and Ruttan employed a four-factor model in which (1) land and mechanical power were regarded as complements and land and labor as substitutes, and (2) fertilizer and land infrastructure were regarded as complements and fertilizer and land as substitutes.
The processes of advance in mechanical technology in the Hayami–Ruttan model are illustrated in Figure 2.2a. i*0 represents the IPC at time zero; it is the envelope of less elastic unit isoquants that correspond, for example, to different types of harvesting machinery. The relationship between land and power is complementary. Land-cum-power is substituted for labor in response to a change in the wage rate relative to an index of land and power prices. The change in the price ratio from BB to CC induces the invention of labor-saving machinery—say, a combine for a reaper.
The process of advance in biological technology is illustrated in Figure 2.2b. Here, i*0 represents an IPC that is an envelope of relatively inelastic land-fertilizer isoquants such as L0. When the fertilizer/land price ratio declines from bb to cc, a new technology—a more fertilizer-responsive crop variety—represented by cc is developed along i*0. Since the substitution of fertilizer for land is facilitated by investment in land and water development, the relationship between new fertilizer-responsive varieties and land infrastructure is complementary.
In Figure 2.2 the impact of advances in mechanical and biological technology on factor ratios is treated as if the advances were completely separable. This is clearly an oversimplification. It is not essential to the Hayami–Ruttan induced technical change model that changes in the land/labor ratio be a direct response to the price of land relative to the wage rate (Thirtle and Ruttan 1987:30, 31).
Source: Hayami and Ruttan (1985:91).
The econometric tests conducted by Hayami and Ruttan suggested that the enormous changes in factor proportions that occurred during the process of agricultural development in the two countries represent “a process of dynamic factor substitutions accompanying changes in the production function induced by changes in relative factor prices” (Hayami and Ruttan 1970:1135).
The initial Hayami–Ruttan article and the further exposition in their book Agricultural Development (1971) became the inspiration for a large number of empirical tests of the microeconomic version of the induced technical change hypothesis in the agricultural and natural resource sectors. Binswanger advanced the methodology for measuring technical change bias with many factors of production (1974a, 1974b). In a 1987 literature review, Thirtle and Ruttan (1987) listed 29 empirical studies of induced technical change in agriculture. Most of the studies drew their inspiration from the initial study by Hayami and Ruttan (1970). Thirtle and Ruttan also listed 38 empirical studies in the industrial sector. The initial studies of biased technical progress change in industry typically did not involve direct tests of the induced technical change hypotheses. By the late 1970s and early 1980s, however, a substantial number of studies, some stimulated by the rise in energy prices in the 1970s, involved direct tests of the induced technical change hypotheses. Within the industrial sector the evidence is strongest in those industries using natural resources and raw materials (Jorgenson and Fraumeni 1981; Wright 1990; Jorgenson and Wilcoxen 1993). As of the mid-1980s, the evidence of tests of the induced technical change hypotheses in agriculture, both in the United States and abroad, was sufficient to support the view that changes (and sometimes differences) in relative factor endowments and prices exert a substantial impact on the direction of technical change.5
The modern revival of interest by economists in an evolutionary theory of technical change derives largely from a series of articles by Nelson and Winter in the mid-1970s (Nelson and Winter 1973, 1974, 1975, 1977; Nelson et al. 1976).6 These articles in turn served as the basis for the highly acclaimed book An Evolutionary Theory of Economic Change (Nelson and Winter 1982). The theory advanced by Nelson and Winter has been identified by the authors as “Schumpeterian” in its interpretation of the process of economic change. In much of the literature that has drawn its inspiration from Nelson and Winter, “evolutionary” and “Schumpeterian” have been used interchangeably.7 The second cornerstone of the Nelson–Winter model is the behavioral theory of the firm, in which profit-maximizing behavior is replaced by decision rules that are applied routinely over an extended period of time (Simon 1955, 1959; Cyert and March 1963).
The Nelson–Winter evolutionary model, particularly Chapters 9–11, jettisons much of what they consider to be the excess baggage of the neoclassical microeconomic theory: “the global objective function, the well-defined choice set, and the maximizing choice rationalization of firm's actions. And we see ‘decision rules’ as very close conceptual relatives of production ‘techniques’ whereas orthodoxy sees these things as very different” (Nelson and Winter 1982:14). The production function and all other regular and predictable behavior patterns of the firm are replaced by the concept of “routine”: “. . . a term that includes characteristics that range from well-specified technical routines for producing things, procedures for hiring and firing, ordering new inventory, or stepping up production of items in high demand to policies regarding investment, research and development (R&D), or advertising, and business strategies about product diversification and overseas investment” (Nelson and Winter 1982:14). The distinction between factor substitution and shifts in the production function is also abandoned. The two fundamental mechanisms in the Nelson–Winter model are the search for better techniques and the selection of firms by the market (Elster 1983:14). In their models the microeconomics of innovation is represented as “a stochastic process dependent on the search routines of individual firms” (Dosi et al. 1992:10). The activities leading to technical changes are characterized by (1) local search for technical innovations, (2) imitation of the practices of other firms, and (3) satisfying economic behavior.
In their initial models, a firm's search for new technology—whether generated internally by R&D or transferred from suppliers or competitors—is set in motion when profits fall below a certain threshold. The models assume that in this search the firms draw samples from a distribution of input–output coefficients (Figure 2.3). If A is the present input combination, then potential input coefficients are distributed around it such that there is a much greater probability of finding a point close to A than finding one far from it. The search is local. Once the firm finds a point B it makes a profitability check. If costs are lower at B than at A, the firm adopts the technology represented by point B and stops searching. Otherwise, the search continues. Thus, the technology described by the point B, input–output and factor ratios will be accepted if labor is relatively inexpensive, that is, if relative prices are described by line CD. However, if labor is relatively expensive, as described by line C′D′, the firm will reject the technology at point B and continue to search for another technology until it finds another point, say B′. The technology at point B′ will be labor saving relative to that at B.
Adapted from Nelson and Winter (1975:472).
The stochastic technology search process is built into a model with many competing firms. All profits above a “normal” dividend—investors are satisficers rather than optimizers—are reinvested so that successful firms grow faster than unsuccessful ones. The capital stock of the economy is determined by the total investment by all firms. Labor supply is elastic to the firm.
Simulation runs rather than formal analysis or tests against historical experience are employed to demonstrate the plausibility of the models. The simulations start from an initial point where all firms are equal. The model endogenously determines the output of the economy, the wage-rental rate, and the capital accumulation rates. Nelson and Winter have used a series of variations of their basic model to explain how changes in market structure influence the rate of technical change, the direction of technical change, and the importance of imitation and innovation.
When firms check the profitability of the alternative techniques that their search processes uncover, a higher wage rate will cause certain techniques to fail profitability tests they would have passed at a lower wage rate and enable others to pass tests they would have failed at a lower wage rate. The latter will be capital intensive relative to the former. Thus a higher wage rate nudges firms in a direction that is more capital intensive than that in which they would have gone. Also, the effect of a higher wage rate is to make all technologies less profitable (assuming, as in their model, a constant cost of capital), but the cost increase is proportionately greatest for those that involve a low capital/labor ratio. Since firms with high capital/labor ratios are less adversely affected by high wage rates then those with low capital/labor ratios, capital-intensive firms will tend to expand relative to labor-intensive ones. For both these reasons, a higher wage rate will tend to increase capital intensity relative to what otherwise would have been obtained (Nelson and Winter 1974:900). The responsiveness of the capital/labor ratio to changes in relative factor prices is rather striking because, except for the profitability check, search (or research) outcomes are random (Nelson and Winter 1982:175–184), and the inducement mechanism comes about through competition, survival, and growth rather than through efforts to maximize profits.
The early Nelson–Winter models were criticized for the “dumb manager” assumption in which the search (or research) process is triggered only when profits fall below a threshold level. For example, “here we assume that firms with positive capacity do not search if they are making positive or zero profits; they satisfice on their prevailing routines” (Nelson and Winter 1982:149). An implication is that an increase in demand for an industry's product can lead to a reduction in research effort. This was hardly consistent with either historical evidence (Schmookler 1966) or with the Schumpeterian perspective. The restriction was relaxed in the second round of Nelson–Winter models by the explicit introduction of directed research. As the wage/rental ratio rises, research effort is allocated, as in the microeconomic induced technical change model, to sampling the spectrum of potential capital-intensive techniques (Nelson and Winter 1975, 1977).
Winter has devoted considerable attention to extensions of the initial Nelson– Winter models. In a 1984 article, for example, he abandons the assumption of the level playing field in which the initial conditions are the same for all firms. The basic model is augmented by a model that includes entirely new firms. Winter uses this expanded model to explore the growth path of two industrial regimes. One is an “entrepreneurial regime,” which he identifies with the early Schumpeter of The Theory of Economic Development (1934; originally published in German, 1911). The second is a “routinized regime,” which he identifies with the Schumpeter of Capitalism, Socialism and Democracy (1950). The entrepreneurial regime model is designed so that innovations are primarily associated with the entry of new firms. In the routinized regime, innovations are primarily the result of internal R&D by established firms. Several suggestions for further extension of the Nelson–Winter models to include the creation of new industries, interaction among industries, and product innovation and imitation, for example, have been summarized and extended by Andersen (1994:118–131).
It is important to clarify the role of historical process in the Nelson–Winter evolutionary models. The condition of the industry in each time period shapes its condition in the following period:
Some economic processes are conceived as working very fast, driving some of the model variables to (temporary) equilibrium values within a single period (or in a continuous time model, instantaneously). In both the entrepreneurial and routinized Schumpeterian models, for example, a short-run equilibrium price of output is established in every time period. Slower working processes of investment and of technological and organizational change operate to modify the data of the short-run equilibrium system from period to period (or from instant to instant). The directions taken by these slower processes of change are directly influenced by the values taken by the subset of variables that are equilibrated in the individual period or instant. (Winter 1984:290)
I find it difficult to resolve the question of why there have been so few efforts by other scholars to advance the Nelson–Winter methodology,8 or to test the correspondence between the plausible results of the Nelson–Winter simulations and the historical experience of particular firms or industries.9 Simulation is capable of generating a wide range of plausible behavior. But the hypotheses generated by the simulations have seldom been subjected to rigorous empirical tests. The closest Nelson and Winter come to empirical testing is the demonstration that it is possible to generate plausible economywide growth paths or changes in market share.
The argument that technical change is “path dependent” was vigorously advanced by Arthur and several colleagues in the late 1970s and early 1980s (Arthur 1989, 1990, 1994; Arthur et al. 1987).10 In the mid- and late 1980s, David presented the results of a series of historical studies—of the typewriter keyboard, the electric light and power supply industries, and others—that served to buttress the plausibility of the path-dependence perspective (David 1985, 1986, 1993; David and Bunn 1988). As noted previously, the emphasis on path dependence in David's more recent work represents an extension of his earlier research on the relationship between labor scarcity and modernization in nineteenth-century America (David 1975). This earlier work was strongly influenced by Arrow's article on learning by doing (Arrow 1962) and by Habakkuk's historical research on British and American technology in the nineteenth century (Habakkuk 1962).
The work by Arthur and his colleagues has emphasized the importance of increasing returns to scale as a source of technological “lock-in.” In some nonlinear dynamic systems, positive feedbacks (Polya processes) may cause certain patterns or structures that emerge to be self-reinforcing. Such systems tend to be sensitive to early dynamical fluctuations (Figure 2.4). Often there is a multiplicity of patterns that are candidates for long-term self-reinforcement; the accumulation of small events early on “pushes” the dynamics of technical choice into the orbit of one of these and thus “selects” the structure that the system eventually “locks into” (Arthur et al. 1987:294).11
Source: Arthur (1989:120).
The authors provide an intuitive example. Think of an urn of an infinite capacity:
Starting with one red and one white ball in the urn, add a ball each time, indefinitely, according to the following rule. Choose a ball in the urn at random and replace it; if it is red, add a red; if it is white, add a white. Obviously this process has increments that are path dependent—at any time the probabilities that the next ball added is red exactly equals the proportion red. . . . Polya proved in 1931 that in a scheme like this the proportion of red balls does tend to a limit X1 and with probability one. But X is a random variable uniformly distributed between 0 and 1. (Arthur et al. 1987:259)
Thus in an industry characterized by increasing returns, small historical or chance events that give one of several technologies an initial advantage can (but need not) “drive the adoption process into developing a technology that has inferior long-run potential” (Arthur 1989:117).
The historical small events that result in path dependence are “outside the ex ante knowledge of the observer—beyond the resolving power of his model or abstraction of the situation” (Arthur 1989:118). Arthur employs a series of progressively complex models to simulate situations in which several technologies compete for adoption by a large number of economic agents. Agents have full knowledge of the technology and returns functions but not of the events that determine entry and choice of technology by other agents. His analysis is carried out for three technological regimes (constant, increasing, and diminishing returns) with respect to four properties of the paths of technical change (predictable, flexible, ergodic, and path efficient).12 The only unknown is the set of historical events that determine the sequence in which the agents make their choices. The question he attempts to answer is whether the fluctuations in the order of choice will make a difference in final adoption shares. Arthur's simulations emphasize the importance of increasing returns as a necessary condition for technological lock-in:
Under constant and diminishing returns the evolution of the market reflects only a priori endowments, preferences, and transformation possibilities; small events cannot sway the outcome. . . . Under increasing returns, by contrast, many outcomes are possible. Insignificant circumstances become magnified by positive feedbacks to “tip” the system into the actual outcome “selected.” The small events in history become important. . . . (Arthur 1989:127)
The network externalities are important not only because of their impact on the direction or path of technology development, but also because they represent a source of market failure—welfare losses that cannot be resolved by normal market processes—and hence call for public intervention (Arthur 1994:9–10). Other factors that may contribute to lock-in include adjustment costs, switching costs, and the costs of maintaining parallel technologies (David 2001).
In Technical Choice, Innovation and Economic Growth, David (1975) characterizes his work as an evolutionary alternative to neoclassical theory. As noted previously, he explicitly rejected the Fellner and Kennedy versions of the induced technical change approach to the analysis of factor bias. He also rejected the early work of Nelson and Winter as being “fundamentally neoclassical-inspired” (David 1975:76).13 But he shares the Nelson and Winter view that the neoclassical model is excessively restrictive, as factor substitution typically involves not simply a movement along a given production function but an element of innovation leading to a shift in the function itself. He does assume that the firm has knowledge of available (or potentially available) alternative technologies and chooses rationally from among them.
David's early analysis of factor bias was remarkably similar to the Hicks– Ahmad–Hayami–Ruttan interpretation of the process of induced technical change. And, in spite of his assertion in Technical Choice (1975) that the future development of the system depends not only on the present state but also on the way the present state evolved, I find his research on path dependence in the 1980s to be a distinct departure from his research on factor bias in the 1970s.
In his research from the mid- and late 1980s, David employs historical analysis of a series of technical changes—the typewriter keyboard, the electric light, and power supply industries—to buttress the plausibility of the path-dependence perspective. His now classic paper on the economics of QWERTY (the first six letters on the left side, topmost row of the typewriter and now the computer keyboard) explored why an inefficient (from today's perspective) typewriter keyboard was introduced and why it has persisted.14 David's answer is that an innovation in typing method, touch typing, gave rise to three features “which were crucially important in causing QWERTY to become ‘locked in’ as the dominant keyboard arrangement. These features were technical inter-relatedness, economies of scale, and quasi-irreversibility of investment” (David 1985:334). Technical interrelatedness refers to the need for system compatibility—in this case, the linkage between the design of the typewriter keyboard and typists’ memory of a particular keyboard arrangement. Scale economics refers to the decline in user cost of the QWERTY system (or any other system) as it gained acceptance relative to other systems. The quasi-irreversibility of investments is the result of the acquisition of specific touch-typing skills (the “software”). These characteristics are sometimes bundled under the rubric of positive “network externalities.”
As David has drawn increasingly on Arthur's path-dependence model, his research has moved even further in the direction of interpreting the QWERTY-like phenomenon in dynamic systems characterized by network externalities and path-dependent technical change as a dominant paradigm for the history of technology (David 1993:208–231).15 This paradigm would seem particularly apt at a time when the impact of scale economies on productivity growth has been rediscovered and embodied in a “new growth economics” literature (Romer 1986; Lucas 1988; Barro and Sala-i-Martin 1995).16 But Arthur's results suggest some caution is necessary: “Increasing returns, if they are bounded, are in general not sufficient to guarantee eventual monopoly by a single technology” (Arthur 1989:126). And there is substantial empirical evidence that scale economies, which often depend on prior technical change, are typically bounded by the state of technology (Levin 1977:208–221).
Both induced innovation and evolutionary theory suggest that as scale economies are exhausted (and profits decline) the pressure of growth in demand will focus scientific and technical effort on breaking the new technological barriers. Superior technologies that lost out as a result of chance events in the first round of technical competition have frequently turned out to be successful as the industry developed.17 And induced technical change theory suggests that research effort will be directed to removing the constraints on growth resulting from technological constraints or inelastic (or scarce) factor supplies.18
The transition from coal to petroleum-based feedstocks in the heavy organic chemical industry is a particularly dramatic example. From the 1870s through the 1930s, German leadership in the organic chemical industry was based on coal-based technology. Beginning in the 1920s with the rapid growth in demand for gasoline for automobiles and trucks in the United States, a large and inexpensive supply of olefins became available as a by-product of petroleum refining. By the end of World War II, the US chemical industry had shifted rapidly to petroleum-based feedstocks. In Germany this transition—impeded by skills, education, and attitudes that had been developed under a coal-based industrial regime—was delayed by more than a decade. By the 1960s, however, Germany was making a rapid transition to the petroleum-based feedstock path of technical change in heavy organic chemicals (Grant et al. 1988; Stokes 1994).
In this section I first summarize my assessment of the strengths and limitations of each of the three models of technical innovation. I then outline the elements of a more general theory. I would like to make clear to the reader my particular historical and epistemological bias: Departures from neoclassical microeconomic theory, when successful, are eventually seen as extensions and become incorporated into neoclassical theory. Thus, for example, the microeconomic version of induced technical change can now be viewed as an extension of, rather than a departure from, the neoclassical theory of the firm.19
One common theme pervading the three approaches to understanding sources of technical change is the disagreement with the assumption in neoclassical growth models that a common production function is available to all countries regardless of human capital, resource, or institutional endowments. It should now be obvious that differences in productivity levels and rates of growth cannot be overcome by the simple transfer of capital and technology. The asymmetries between firms and between countries in resource endowments and in scientific and technological capabilities are not easily overcome. The technologies that are capable of becoming the most productive sources of growth are often location specific (Kenney and von Burg 2001). A second common theme is an emphasis on microfoundations. This emphasis on microfoundations is common to the approaches that have abandoned neoclassical microeconomics as well as to those that have attempted to extend neoclassical theory (Dosi 1995). This stands in sharp contrast to the limited attention to microfoundations in neoclassical growth theory (Ruttan 1998, 2001).
The primary limitation of the growth theoretic version of the induced innovation model is the implausibility of the innovation-possibility function. The shape of the function is independent of the bias in the path of technical change. As technical change progresses, there is no effect on the “fundamental” trade-off between labor- and capital-augmenting technical change. Thus, as Nordhaus notes, the growth theoretic approach to induced innovation fails to rescue growth theory from treating technical change as exogenous. It has not produced empirical research and is no longer viewed as an important contribution to growth theory.
The primary limitation of the microeconomic version is that its internal mechanism—the learning, search, and formal R&D processes—remains inside a black box. The model is driven by exogenous changes in the economic environment in which the firm (or public research agency) finds itself. But the process of technical change is, itself, not entirely endogenous. Exogenous advances in scientific and technological knowledge can, for example, open up new opportunities for technical change. The microeconomic model has, nevertheless, produced a substantial body of empirical research and has helped to clarify the historical process of technical change, particularly at the industry and sector levels both within and across countries.
The strength of the evolutionary model lies precisely in the area where the microeconomic induced innovation model is weakest. It builds on the behavioral theory of the firm in an attempt to provide a more realistic description of the internal workings of the black box. The Nelson–Winter evolutionary approach has not, however, become a productive source of empirical research. The results of the various simulations are defended as plausible in terms of the stylized facts of industrial organization and of firm, sector, and macroeconomic growth. It is possible that the reason for the lack of empirical testing is that the simulation methodology lends itself to the easy proliferation of plausible results. At present, the evolutionary approach must be regarded as a “point of view” rather than a theory (Arrow 1995).
The strength of the path-dependent model lies in the insistence of its practitioners on the importance of the sequence of specific micro-level historical events.20 In this view, current choices of techniques become the link through which prevailing economic conditions may influence the future dimensions of technology and knowledge (David 1975:39, 57). However, the concept of technological lock-in, at least in the hands of its more rigorous practitioners, applies only to network technologies characterized by increasing returns to scale. In industries with constant or decreasing returns to scale, historical lock-in does not apply.
There can be no question that technical change is path dependent in the sense that it evolves from earlier technological development. In spite of somewhat similar motivation, the path-dependent literature has not consciously drawn on the Nelson–Winter work for inspiration (Arthur 1996). It is necessary to go beyond the present path-dependent models, however, to examine the forces responsible for changes in the rate and direction of technical change. But there is little discussion of how firms or industries escape from lock-in. What happens when the scale economies resulting from an earlier change in technology have been exhausted and the industry enters a constant- or decreasing-returns stage? At this point in time it seems apparent that changes in relative factor prices would, with some lag, have the effect of bending or biasing the path of technical change along the lines suggested by the theory of induced technical change. Similarly, a new radical innovation may, at this stage, both increase the rate and modify the direction of technical change.
The study of technical change in the semiconductor industry by Dosi (1984) represents a useful illustration of the potential value of a more general model. The Dosi study is particularly rich in its depth of technical insight. At a rhetorical level, Dosi identifies his methodology with the Nelson–Winter evolutionary approach. In practice, however, he utilizes an eclectic combination of induced innovation, evolutionary, and path-dependence interpretations of the process of semi-conductor technology development. A more rigorous approach to the development of a general theory of the sources of technical change will be required to bridge the three “island empires.”
A first step toward developing a more general theory of technical change is to integrate the “factor-induced” and the “demand-induced” models (Ruttan and Hayami 1994:180). Binswanger, drawing on Nordhaus (1969:105–109) and Kamien and Schwartz (1969), has sketched outlines of how a more general model can make both the rate and direction of technical change endogenous (Binswanger 1978b:104–110).
If one assumes decreasing marginal productivity of research resources in applied research and technology development and, in addition, incorporates the effects of changes in product demand, then growth (decline) in product demand would increase (decrease) the optimum level of search and research expenditure. The larger research budget, induced by growth in product demand, increases the rate at which the meta-production function shifts inward toward the origin. Even when the initial path of technological development is generated by “technology push,” factor market forces often act to modify the path of technical change. Differential elasticities of factor supply result in changes in relative factor prices and direct research effort to save increasingly scarce factor supplies. The result is a non-neutral shift in both the neoclassical and the meta-production functions.21
More recently, Christian has elaborated the Binswanger model and analyzed more formally the innovators’ decision to conduct R&D directed toward process innovation (Christian 1993). As yet, however, there has been no attempt to implement empirically an integrated factor- and demand-induced innovation model.
A second step would be the integration of the induced technical change and the path-dependent models. As noted earlier, David has pointed to the persistent failure to replace the inefficient QWERTY layout of the typewriter and computer keyboards with the more efficient DSK keyboard. Wright (1990) has suggested that the historical resource intensity of American industry, based on domestic resource abundance, has been an important factor in weakening the capacity of American industry to adapt to a world in which lower transportation costs and more open trading systems have reduced the traditional advantage of US-based firms. If this perspective is correct, Japan's industrial success may be attributed to its historical resource scarcity.
The difference in perspective seems to hinge on how the elasticity of substitution changes over time in response to changes in resource endowments or relative factor prices. David has shown how localized induced technical change can lead to path-dependent technical change (1975:65–68). The effect of localization is to lower the elasticity of substitution and lock in the trajectory of technical change. As relative factor prices continue to shift, however, it is difficult to believe technological competition would not result in a “bending” of the path of technical change in the direction implied by changing factor endowments. The path-dependent and the induced innovation models are appropriately viewed as complementary rather than as alternative interpretations of the forces that influence the direction of technical change.
The path-dependent model will remain incomplete, however, until it is more fully integrated with the microeconomic version of the induced technical change model and with the Nelson–Winter evolutionary model. Development of an industry seldom proceeds indefinitely along an initially selected process ray (Landes 1994). As technical progress slows down or scale economies erode, a shift in relative factor prices can be expected to induce an intensified search for technologies along a ray that is more consistent with contemporary factor prices.22 At the theoretical level, research should be directed toward the development of a succession of more fully integrated models of the sources of technical change. At the empirical level, research should be directed toward more comprehensive testing of the induced, evolutionary, and path-dependent models. There is a substantial empirical literature on induced technical change in agricultural economics and economic history, but only limited efforts have been made to test the evolutionary and path-dependent models against historical experience. Metaphor is not enough!
2.10 Induced Technical Change and Endogenous Economic Growth
Since the late 1980s, students of economic growth have been engaged in a reevaluation of neoclassical growth models. The re-examination was stimulated by concern that the neoclassical growth theory was inconsistent with the evidence of a lack of convergence of growth rates among rich and poor countries (Baumol 1986; Dollar and Wolff 1993; Ruttan 1998). One result of this re-examination has been the emergence of a new generation of endogenous growth models.
The primary focus of the new “macroeconomic endogenous” growth models is to attribute differences in growth performance among countries to factors such as investment in human capital, learning by doing, scale economies, and technical change (Romer 1986, 1994; Lucas 1988; Ruttan 1998). In the initial Romer– Lucas framework, the accumulation of human capital adds to the productivity of the person in whom it is embodied.23 But the general level of productivity rises by more than can be accounted for or captured by the person or firm that makes each particular investment. Gains in scale economies are enhanced by the integration into multinational trading systems of economies that are human-capital intensive (Grossman and Helpman 1992).
There has also been a renewed interest in the theory of induced technical change. This renewed interest is reflected in the subsequent chapters of this book. It has been stimulated, at least in part, by the focus on endogenous technical change in the new growth literature. It has also been stimulated by a growing concern with a series of natural resource issues such as the effects of global climate change discussed in this book. The new endogenous growth literature has yet, however, to incorporate the richness and depth of understanding of the sources of technical change that the three traditions reviewed in this chapter have achieved (Bardhan 1995; Ruttan 1998). Like the older neoclassical growth literature, its focus is on the proximate sources of growth rather than the sources of technical change. A major challenge for the future is to integrate the insights about endogenous growth gained from the theoretical and empirical research conducted within the induced technical change, the evolutionary, and the path-dependence theories with new insights into the relationship between human capital, scale, and trade opened up by the macro-endogenous growth models.
The author is indebted to Esben Sloth-Anderson, W. Brian Arthur, Erhard Bruderer, Jason E. Christian, Paul A. David, Jerry Donato, Giovanni Dosi, Laura McCann, Richard Nelson, Nathan Rosenberg, Tugrul Temel, Michael A. Trueblood, Andrew Van de Ven, and Sidney Winter for comments on an earlier draft of this paper. Earlier versions of this paper have been presented in seminars at the International Institute for Applied Systems Analysis (IIASA), at the University of Minnesota Economic Development Center, and at the Hong Kong University of Science and Technology. The research on which the paper is based was supported, in part, by a grant from the Alfred P. Sloan Foundation. Earlier versions of this paper have appeared in Ruttan (1996, 1997, 2001). For critical reviews, see Dosi (1997) and Wright (1997).
1. Zvi Griliches pointed out to me (in conversation) that another reason for the decline in interest among economic theorists was the difficulty, noted by Diamond et al. (1978:125–147), of simultaneously measuring the bias of technical change and the elasticity of substitution among factors. This problem had, however, already been solved (Binswanger 1974c; Binswanger and Ruttan 1978:73–80, 215–242). For a more recent discussion see Haltmaier (1986).
2. It is interesting to speculate on what the course of induced innovation theory might have been if the Ahmad article had, as it might have, appeared first. The initial drafts of the articles were written while Kennedy was teaching at the University of the West Indies (Kingston) and Ahmad was teaching at the University of Khartoum (Sudan). Ahmad submitted his article to the Economic Journal in 1963. Kennedy served as a reviewer of the Ahmad article. His article, which was published in 1964, was originally written as a comment on the Ahmad article. Ahmad's article was rewritten, resubmitted, and published in 1966 (Ruttan and Hayami 1994:24).
3. Binswanger authored a series of articles in the mid-1970s that were fundamental in the formalization of the theoretical foundations of the microeconomic model (1974a, 1974b). The material in these two articles is incorporated in Binswanger and Ruttan (1978). See also Kamien and Schwartz (1969, 1971).
4. At the time the article was written, Hayami and Ruttan were familiar with the growth theoretic literature by Fellner, Kennedy, and Samuelson, but not with the Ahmad article and his subsequent exchange with Fellner and Kennedy. The inspiration for the 1970 Hayami–Ruttan paper was the historical observations about the development of British and American technology by Habakkuk (1962). See Ruttan and Hayami (1994).
5. Olmstead and Rhode (1993) have criticized the Hayami and Ruttan work on both conceptual and empirical grounds. At the conceptual level they find confusion between the relative factor “change variant” used in explaining productivity growth over time within a given country and the “level variant” of the model used in analysis of international productivity differences. They also argue, on the basis of regional tests in the United States, that the induced technical change model holds only for the central grain-growing regions. In a later paper (Olmstead and Rhode 1995) using state-level data, they found somewhat stronger support for the induced technical change hypothesis. For further criticism and a defense, see Koppel (1995).
6. Nelson and Winter identify Alchian (1950) and Penrose (1952) as representing direct intellectual antecedents of their work. For the theoretical foundations of the Nelson– Winter collaboration, see Winter (1971). For the historical and philosophical foundations, see Elster (1983:131–158) and Langolis and Everett (1994:11–47). Witt has assembled many of the most important articles in the field of evolutionary economics in a collection of readings, Evolutionary Economics (1993). For a review of recent evolutionary thought about economic change, see Nelson (1995:48–90).
7. The Nelson–Winter model departs in its treatment of the linkage between invention and innovation. For Schumpeter there was no necessary link between invention and innovation (Ruttan 1959). Nelson and Winter employ the term “evolutionary” metaphorically: “We emphatically disavow any intention to pursue biological analogies for their own sake” (1982:11). Nelson and Winter regard their approach as being closer to Lamankianism than Mendelianism. Yet their description of the evolutionary process of firm behavior and technical change as a Markov process and their use of the Markov mechanism in their simulation are analogous to the Mendelian model.
8. For a useful interpretation and extension, see Andersen (1994). Andersen's work is particularly helpful in clarifying the “poorly documented” computational steps of the Nelson–Winter models. Andersen supplements the mathematical notation employed by Nelson and Winter with an algorithmically oriented programming notation. An appendix, “Algorithmic Nelson and Winter Models” (pp. 198–219), is particularly useful.
9. A large body of empirical research on technical change that can be categorized as broadly Schumpeterian or evolutionary in inspiration has emerged since the mid-1970s [see the review by Freeman (1994)]. The point I am making, however, is quite different. There has been very little effort to use the simulation models to generate hypotheses about the process of technology development and then to either identify historical counterparts or test the outcomes against historical experience in a rigorous manner. The one exception with which I am familiar is the Evenson–Kislev (1975:140–155) stochastic model of technological discovery. The model was used to interpret the stages in sugarcane varietal development. The Evenson–Kislev model did not, however, draw directly on the Nelson–Winter stochastic model.
10. Arthur encountered unusual delay before his work was accepted in a leading economics journal. His 1986 Economic Journal paper was initially submitted to the American Economic Review in 1983. It was rejected by the American Economic Review twice and by the Quarterly Journal of Economics twice and accepted by the Economic Journal only after an appeal. By the time the paper was finally accepted in the Economic Journal, referees were noting that the path-dependence idea was already recognized in the literature (Gans and Sheperd 1994:173).
11. There has been considerable confusion regarding the interpretations of path dependence and the meaning of “lock-in” in the literature. David has recently provided explicit definitions: “A path dependent stochastic process is one whose asymptotic distribution evolves as a consequence (function of) the process's own history” (David 2001:5). “The term lock-in is a vivid way to describe the entry of a system into a . . . region . . . that surrounds a locally (or globally) stable equilibrium. When a dynamic system enters such a region it cannot escape except through the intervention of external force, or shock” (David 2001:10).
12. “A process is predictable if the small degree of uncertainty built in ‘averages away’ so that the observer has enough knowledge to predetermine market shares accurately in the long run; flexible if a subsidy or tax adjustment to one of the technologies’ returns can always influence future market choice; ergodic (not path dependent) if different sequences of historical events lead to the same market outcome with probability one; . . . and path efficient if at all times equal development (equal adoption) of the technology that is behind in adoption would not have paid off better” (Arthur 1989:118, 199).
13. For a further comparison of the David and Nelson–Winter evolutionary approaches, see Elster (1983:150–157). Elster notes that David regards the Nelson–Winter model as evolutionary, but ahistorical. In his view, it differs from the neoclassical model only in its conception of microeconomic behavior. It is ahistorical since, in David's view, the Markovian-like transition probabilities depend only on the current state and not on earlier states of the system. Elster rejects David's criticism of Nelson and Winter on the basis that, for the past to have a causal influence on the present, it must be “mediated by a chain of locally causal links.” Thus, since all the history that is relevant to the prediction of the future is contained in the state description, if the present state is known, prediction cannot be improved by considering the past history of the system (Elster 1983:157).
14. Liebowitz and Margolis (1990, 1994) argue that David's version of the history of the market's rejection of the supposedly more efficient Dvorak keyboard represents bad history. Given the available knowledge and experience at the time QWERTY became dominant, it represented a rational choice of technology. For a vigorous response to the Liebowitz and Margolis criticism, see David (2001).
15. Both Arthur and David emphasize the role of network externalities in locking in inferior technological trajectories. In their review of the creation and growth of Silicon Valley, Kenney and von Burg note that while the evolution of Silicon Valley venture capital has a path-dependent history, “it is difficult to imagine a more efficient system for formulating high technology startups” (2001:144).
16. Scale economics have become the new “black box” of contemporary growth theory. It is difficult to believe that much of the productivity growth that is presumably accounted for by scale economies is not the disequilibrium effect of prior technical change (Landau and Rosenberg 1992:93; Liebowitz and Margolis 1994:139).
17. See, for example, the exceedingly careful study of technological substitution in the case of Cochlear implants by Van de Ven and Garud (1993) and Garud and Rappa (1994). The Cochlear implant is a biomedical invention that enables hearing by profoundly deaf people. The industry is characterized by the conditions that David and Arthur identify with technological lock-in. Yet in spite of initial commercial dominance, the “single-channel” technology was completely replaced by the “multiple-channel” technology. For other cases, see Foray and Grübler (1990), Cheng and Van de Ven (1994), and Liebowitz and Margolis (1992, 1995).
18. The development of semiconductor technology as a replacement for vacuum tubes for amplifying, rectifying, and modulating electrical signals is an example of a shift in technological trajectories induced by technological constraints (Dosi 1984:26–45). The development of fertilizer-responsive crop varieties represents an example of a shift in technological trajectories induced by changes in resource endowments (Hayami and Ruttan 1985:163–198). The gas turbine's emergence from a niche technology to become an important source of electric power generation since the early 1980s was induced, in part, by the exhaustion of scale in steam turbine generation (Islas 1997:49–66).
19. Nelson and Winter attempt to confront this problem by arguing that there are two alternative views of neoclassical theory. One is the more rigorous “literal” view. The other is termed the “tendency” view. Applied economists with a primary interest in interpreting economic history or behavior tend to employ the “tendency” view, while theorists who are more concerned with the formal properties employ a more literal interpretation. They identify evolutionary theory with the “tendency” view (Nelson and Winter 1975:467).
20. This emphasis creates an important opportunity to incorporate the contributions of noneconomists (particularly historians and other social scientists) along with those of economists into a more comprehensive understanding of the sources of technical change (Kenney and von Burg 2001).
21. In Binswanger, both the production function and the meta-production function are neoclassical. David (1997) noted that if the advances in technology are highly “localized,” in the sense suggested by Atkinson and Stiglitz (1969), the neoclassical assumption is inappropriate. See also David (1975:65–68) and Antonelli (1995:1–18).
22. See, for example, the patterns of factor substitution in the transition in primary energy sources and transportation infrastructure (Grübler and Nakicenovic 1988:13–44; Nakicenovic 1991).
23. The initial models are frequently referred to as AK models after the assumed production function (Y = AK). In expanded versions of the model, K can be thought of as “a proxy for a composite of capital goods that includes physical and human components” (Barro and Sala-i-Martin 1995:146). In a retrospective assessment, Romer notes, “My interpretation . . . was that investments in physical capital tended to be accompanied by investments in new ideas. Looking back . . . it has pushed the discussion away from knowledge and ideas and toward a more narrow focus on the marginal productivity of capital” (Romer 1994:558).
Ahmad, S., 1966, On the theory of induced innovation, Economic Journal, 76:344–357.
Ahmad, S., 1967a, Reply to Professor Fellner, Economic Journal, 77:662–664.
Ahmad, S., 1967b, A rejoinder to Professor Kennedy, Economic Journal, 77:960–963.
Alchian, A.A., 1950, Uncertainty, evolution and economic theory, Journal of Political Economy, 58:211–222.
Ames, E., and Rosenberg, N., 1968, The Enfield arsenal in theory and history, Economic Journal, 78:730–733.
Andersen, E.S., 1994, Evolutionary Economics: Post-Schumpeterian Contributions, Pinter Publishers, London, UK.
Antonelli, C., 1995, The Economics of Localized Technological Changes and Industrial Dynamics, Kluwer Academic Publishers, Dordrecht, Netherlands.
Arrow, K., 1962, The economic implications of learning by doing, Review of Economic Studies, 29:155–173.
Arrow, K., 1995, Viewpoint, Science, 267(March 17):1617.
Arthur, W.B., 1989, Competing technologies, increasing returns, and lock-in by historical events, The Economic Journal, 99(March):116–131.
Arthur, W.B., 1990, Positive feedbacks in the economy, Scientific American, 262(February):92–99.
Arthur, W.B., 1994, Increasing Returns and Path Dependence in the Economy, The University of Michigan Press, Ann Arbor, MI, USA.
Arthur, W.B., 1996, Letter, October 17.
Arthur, W. B., Ermoliev, Y.M., and Kaniovski, Y.M., 1987, Path dependence processes and the emergence of macro-structure, European Journal of Operational Research, 30(June):294–303.
Atkinson, A.B., and Stiglitz, J.E., 1969, A new view of technological change, Economic Journal, 79(September):573–578.
Bardhan, P., 1995, The contribution of endogenous growth theory to the analysis of development problems: An assessment, in J. Behrman and T.N. Srinivasan, eds, Handbook of Development Economics, Vol. III, Elsevier, Amsterdam, Netherlands.
Barro, R.J., and Sala-i-Martin, X., 1995, Economic Growth, McGraw Hill, New York, NY, USA.
Baumol, W., 1986, Productivity growth, convergence and welfare: What the long run data show, American Economic Review, 76:1072–1085.
Benzion, U., and Ruttan, V.W., 1975, Money in the production function: An interpretation of empirical results, Review of Economics and Statistics, 57:246–247.
Benzion, U., and Ruttan, V.W., 1978, Aggregate demand and the rate of technical change, in H.P. Binswanger, and V.W. Ruttan, eds, Induced Innovation: Technology Institutions and Development, The Johns Hopkins University Press, Baltimore, MD, USA.
Binswanger, H.P., 1973, The Measurement of Biased Efficiency Gains in US and Japanese Agriculture to Test the Induced Innovation Hypothesis, Ph.D. dissertation, North Carolina State University, Raleigh, NC, USA.
Binswanger, H.P., 1974a, A microeconomic approach to induced innovation, Economic Journal, 84:940–958.
Binswanger, H.P., 1974b, A cost function approach to the measurement of elasticities of factor demand and elasticities of substitution, American Journal of Agricultural Economics, 56:377–386.
Binswanger, H.P., 1974c, The measurement of technical change biases with many factors of production, American Economic Review, 64:964–976.
Binswanger, H.P., 1978a, Induced technical change: Evolution of thought, in H.P. Binswanger and V.W. Ruttan, eds, Induced Innovation: Technology, Institutions and Development, The Johns Hopkins University Press, Baltimore, MD, USA.
Binswanger, H.P., 1978b, The microeconomics of induced technical change, in H.P. Binswanger and V.W. Ruttan, eds, Induced Innovation: Technology, Institutions and Development, The Johns Hopkins University Press, Baltimore, MD, USA.
Binswanger, H.P., and Ruttan, V.W., 1978, Induced Innovation: Technology, Institutions and Development, The Johns Hopkins University Press, Baltimore, MD, USA.
Blaug, M., 1963, A survey of the theory of process-innovation, Economica, 63:13–32.
Cheng, Y.-T., and Van de Ven, A.H., 1994, Learning the Innovation Journey: Order out of Chaos?, Discussion Paper 208, October, University of Minnesota Strategic Management Center, Minneapolis, MN, USA.
Christian, J.E., 1993, The Simple Microeconomics of Induced Innovation, Department of Agricultural Economics, University of California, December 23, Davis, CA, USA (mimeo).
Cyert, R.M., and March, J.G., 1963, A Behavioral Theory of the Firm, Prentice Hall, Englewood Cliffs, NJ, USA.
David, P.A., 1966, Mechanization of reaping in the antebellum Midwest, in H. Rosovsky, ed., Industrialization in Two Systems: Essays in Honor of Alexander Gerschenkron, John Wiley, New York, NY, USA.
David, P.A., 1973, Labor Scarcity and the Problem of Technological Practice and Progress in the Nineteenth Century, Research Paper 297, Harvard Institute of Economic Research, May, Cambridge, MA, USA.
David, P.A., 1975, Technical Choice, Innovation and Economic Growth, Cambridge University Press, Cambridge, UK.
David, P.A., 1985, Clio and the economics of QWERTY, American Economic Review, 76:332–337.
David, P.A., 1993, Path dependence predictability in dynamic systems with local network externalities: A paradigm for historical economics, in D. Foray and C. Freeman, eds, Technology and the Wealth of Nations: The Dynamics of Contracted Advantage, Pinter Publishers, London, UK.
David, P.A., 1997, Letter, October 24.
David, P.A., 2001, Path dependence: Its critics and the quest for historical economics, in P. Garrouste and S. Ioannides, eds, Evolution and Path Dependence in Economic Ideas: Past and Present, Edward Elgar Publishing, Cheltenham, UK.
David, P.A., and Bunn, J.A., 1988, The economics of gateway technologies and network evolution: Lessons from electricity supply history, Information Economics and Policy, 3:165–202.
Diamond, P., McFadden, D., and Rodriguez, M., 1978, Measurement of factor substitution and bias of technological change, in M. Fuss and D. McFadden, eds, Production Economics: A Dual Approach to Theory and Applications, Vol. 2, North Holland, Amsterdam, Netherlands.
Dollar, D., and Wolff, E.N., 1993, Competitiveness, Convergence and International Specialization, The MIT Press, Cambridge, MA, USA.
Dosi, G., 1984, Technical Change and Industrial Transformation, St. Martins Press, New York, NY, USA, pp. 26–45.
Dosi, G., 1988, Sources, procedures, and microeconomic effects of innovation, Journal of Economic Literature, 26:1120–1171.
Dosi, G., 1995, Letter, February 15.
Dosi, G., 1997, Opportunities, incentives and the collective patterns of technological change, Economic Journal, 107:1530–1547.
Dosi, G., Giannetti, R., and Toninelli, P.A., eds, 1992, Technology and Enterprise in a Historical Perspective, Oxford University Press, Oxford, UK.
Drandakis, E.M., and Phelps, E.S., 1966, A model of induced invention, growth and distribution, Economic Journal, 76:823–840.
Elster, J., 1983, Explaining Technical Change, Cambridge University Press, New York, NY, USA.
Evenson, R.E., and Kislev, Y., 1975, Agricultural Research and Productivity, Yale University Press, New Haven, CT, USA.
Fellner, W., 1956, Trends and Cycles in Economic Activity, Henry Holt, New York, NY, USA.
Fellner, W., 1961, Two propositions in the theory of induced innovations, Economic Journal, 71:305–308.
Fellner, W., 1962, Does the market direct the relative factor saving effects of technological progress?, in R.R. Nelson, ed., The Rate and Direction of Inventive Activity: Economic and Social Factors, Princeton University Press for the National Bureau of Economic Research, Princeton, NJ, USA.
Fellner, W., 1967, Comment on the induced bias, Economic Journal, 77:664–665.
Fogel, R.W., 1967, The specification problem in economic history, Journal of Economic History, 27:283–308.
Foray, D., and Grübler, A., 1990, Morphological analysis, diffusion and lock-out of technologies: Ferrous casting in France and the FRG, Research Policy, 19:535–550.
Gans, J.S., and Shepherd, G.B., 1994, How are the mighty fallen: Rejected classic articles by leading economics, Journal of Economic Perspectives, 8:165–179.
Garud, R., and Rappa, M., 1994, A socio-cognitive model of technology evolution: The case of Cochlear implants, Organization Science, 5:344–362.
Gilfillan, S.C., 1935, The Sociology of Invention, Folliett, Chicago, IL, USA.
Grant, W., Patterson, W., and Whitston, C., 1988, Government and the Chemical Industry: A Comparative Study of Britain and West Germany, Oxford University Press, Oxford, UK.
Griliches, Z., 1957, Hybrid corn: An exploration in the economics of technological change, Econometrica, 25:501–522.
Grossman, G., and Helpman, E., 1992, Innovation and Growth in the Global Economy, The MIT Press, Cambridge, MA, USA.
Grübler, A., and Nakicenovic, N., 1988, The dynamic evolution of methane technologies, in T.H. Lee, H.R. Linden, D.A. Dryfus, and T. Vasco, eds, The Methane Age, Kluwer Academic Publishers, Dordrecht, Netherlands.
Habakkuk, H.J., 1962, American and British Technology in the Nineteenth Century, Cambridge University Press, Cambridge, UK.
Haltmaier, J., 1986, Induced Innovation and Productivity Growth: An Empirical Analysis, Federal Reserve Board Special Studies Paper 220, February, Washington, DC, USA.
Hayami, Y., and Ruttan, V.W., 1970, Factor prices and technical change in agricultural development: The United States and Japan, 1880–1960, Journal of Political Economy, 78 (September–October):1115–1141.
Hayami, Y., and Ruttan, V.W., 1971 and 1985, Agricultural Development: An International Perspective, The Johns Hopkins University Press, Baltimore, MD, USA.
Hicks, J., 1932, The Theory of Wages, 1st ed., Macmillan, London, UK (2nd ed., 1963).
Islas, J., 1997, Getting round the lock-in in electricity generating systems: The examples of the gas turbine, Research Policy, 26:49–66.
Jorgenson, D., and Fraumeni, B.M., 1981, Relative prices and technical change, in E.R. Bendt and B. Fields, eds, Modeling and Measuring National Resource Substitution, MIT Press, Cambridge, MA, USA.
Jorgenson, D., andWilcoxen, P.J., 1993, Energy, the environment, and economic growth, in A.V. Kneese and J.L. Sweeney, eds, Handbook of Resource and Energy Economics, Vol. III, Elsevier Science Publishers, Amsterdam, Netherlands.
Kamien, M., and Schwartz, N., 1969, Optimal induced technical change, Econometrica, 36(1):1–17.
Kamien, M., and Schwartz, N., 1971, The theory of the firm with induced technical change, Metroeconomica, 23:233–256.
Kennedy, C., 1964, Induced bias in innovation and the theory of distribution, Economic Journal, 74:541–547.
Kennedy, C., 1966, Samuelson on induced innovation, Review of Economics and Statistics, 48:442–444.
Kennedy, C., 1967, On the theory of induced innovation—A reply, Economic Journal, 77:958–960.
Kenney, M., and von Burg, U., 2001, Paths and regions: The creation and growth of Silicon Valley, in R. Garud and P. Karnoe, eds, Path Dependence and Path Creation, Lawrence Earlbaum and Associates, New York, NY, USA.
Landau, R., and Rosenberg, N., 1992, Successful commercialization in the chemical process industries, in N. Rosenberg, R. Landau, and D.C. Mowery, eds, Technology and the Wealth of Nations, Stanford University Press, Stanford, CA, USA.
Landes, D.S., 1994, What room for accident in history? Explaining big changes by small events, Economic History Review, 47:637–656.
Langolis, R.N., and Everett. M.J., 1994, What is evolutionary economics?, in L. Magnusson, ed., Evolutionary and Neo-Schumpeterian Approaches to Economics, Kluwer, Dordrecht, Netherlands.
Levin, R.C., 1977, Technical change and optimal scale: Some evidence and implications, Southern Economic Journal, 44(October):208–221. Liebowitz, S.J., and Margolis, S.E., 1990, The fable of the keys, Journal of Law and Economics, 33(April):1–325.
Liebowitz, S.J., and Margolis, S.E., 1992, Market Processes and the Selection of Standards, March, Working Paper, University of Texas at Dallas School of Management, Dallas, TX, USA.
Liebowitz, S.J., and Margolis, S.E., 1994, Network externality: An uncommon tragedy, Journal of Economic Perspectives, 8(2):133–150.
Liebowitz, S.J., and Margolis, S.E., 1995, Path dependence, lock-in, and history, Journal of Law, Economics, and Organization, 11 (April):205–226.
Lucas, R.E. Jr., 1967, Tests of a capital-theoretic model of technological change, Review of Economic Studies, 34:175–180.
Lucas, R.E. Jr., 1988, On the mechanics of economic development, Journal of Monetary Economics, 22:3–42.
Mowery, D.C., and Rosenberg, N., 1979, The influence of market demand upon innovation: A critical review of some recent empirical studies, Research Policy, 8:103–153.
Nakicenovic, N., 1991, Diffusion of pervasive systems: A case of transport infrastructures, Technological Forecasting and Social Change, 39(March/April):181–219.
Nelson, R.R., 1995, Recent evolutionary theorizing about economic change, Journal of Economic Literature, 33(March): 48–90.
Nelson, R.R., and Winter, S.G., 1973, Toward an evolutionary theory of economic capabilities, American Economic Review, 63:440–449.
Nelson, R.R., and Winter, S.G., 1974, Neoclassical vs. evolutionary theories of economic growth: Critique and prospects, Economic Journal, 84:886–905.
Nelson, R.R., and Winter, S.G., 1975, Factor price changes and factor substitution in an evolutionary model, Bell Journal of Economics, 6:466–486.
Nelson, R.R., and Winter, S.G., 1977, Simulation of Schumpeterian competition, American Economic Review, 67:271–276.
Nelson, R.R., and Winter, S.G., 1982, An Evolutionary Theory of Economic Change, Harvard University Press, Cambridge, MA, USA.
Nelson, R.R., Winter, S.G., and Schuette, H.L., 1976, Technical change in an evolutionary model, Quarterly Journal of Economics, 40:90–118.
Nordhaus, W.D., 1969, Invention, Growth and Welfare: A Theoretical Treatment of Technical Change, The MIT Press, Cambridge, MA, USA.
Nordhaus, W.D., 1973, Some skeptical thoughts on the theory of induced innovation, Quarterly Journal of Economics, 87:208–219.
Office of the Director of Defense Research and Engineering, 1969, Project Hindsight: Final Report (HINDSIGHT), Washington, DC, USA.
Olmstead, A.L., and Rhode, P., 1995, Induced Innovation in American Agriculture: An Econometric Analysis, Institute of Government Affairs, University of California, Davis, CA, USA (mimeo).
Penrose, E.T., 1952, Biological analogies to the theory of the firm, American Economic Review, 42:804–819.
Romer, P., 1986, Increasing returns and long-run growth, Journal of Political Economy, 94:1002–1037.
Romer, P., 1994, Idea gaps and object gaps in economic development, Journal of Monetary Economics, 8:3–22.
Rosenberg, N., 1974, Science, invention and economic growth, Economic Journal, 84(March):90–108.
Rosenberg, N., 1982, Learning by using, in N. Rosenberg, ed., Inside the Black Box: Technology and Economics, Cambridge University Press, Cambridge, UK.
Rothschild, K., 1956, The Theory of Wages, Blackwell Publishers, Oxford, UK.
Ruttan, V.W., 1959, Usher and Schumpeter on invention, innovation, and technological change, Quarterly Journal of Economics, 73:596–606.
Ruttan, V.W., 1996, Induced innovation and path dependence: A reassessment with respect to agricultural development and the environment, Technological Forecasting and Social Change, 53(1):41–59.
Ruttan, V.W., 1997, Induced innovation, evolutionary theory and path dependence:Sources of technical change, Economic Journal, 107:1520–1529.
Ruttan, V.W., 1998, The new growth theory and development economics: A survey, Journal of Development Studies, 35:1–26.
Ruttan, V.W., 2001, Technology, Growth and Development: An Induced Innovation Perspective, Oxford University Press, New York, NY, USA.
Ruttan, V.W., and Hayami, Y., 1994, Induced innovation theory and agricultural development: A personal account, in B. Koppel, ed., Induced Innovation Theory and International Agricultural Development: A Reassessment, The Johns Hopkins University Press, Baltimore, MD, USA.
Salter, W.E.G., 1960, Productivity and Technical Change, 2nd ed., Cambridge University Press, New York, NY, USA.
Samuelson, P.A., 1965, A theory of induced innovation along Kennedy–Weizsacker lines, Review of Economics and Statistics, 47:343–356.
Samuelson, P.A., 1966, Rejoinder: Agreements, disagreements, doubts and the case of induced Harrod-neutral technical change, Review of Economics and Statistics, 48:444–448.
Scherer, F.M., 1982, Demand pull and technological inventions: Schmookler revisited, Journal of Industrial Economics, 30:225–237.
Schmookler, J., 1962, Determinants of industrial invention, in R.R. Nelson, ed., The Rate of Direction of Inventive Activity: Economic and Social Factors, Princeton University Press, Princeton, NJ, USA.
Schmookler, J., 1966, Invention and Economic Growth, Harvard University Press, Cambridge, UK.
Schumpeter, J.A., 1934, The Theory of Economic Development, Harvard University Press, Cambridge, MA, USA.
Schumpeter, J.A., 1950, Capitalism, Socialism and Democracy, Harper, New York, NY, USA.
Simon, H.A., 1959, Theories of decision making in economics, American Economic Review, 49:253–283.
Stokes, R.G., 1994, Opting for Oil: The Political Economy of Technical Change in the West German Chemical Industry, 1945–1961, Cambridge University Press, Cambridge, UK.
Temin, P., 1966, Labor scarcity and the problems of American industrial efficiency, Journal of Economic History, 26:277–298.
Thirtle, C.G., and Ruttan, V.W., 1987, The Role of Demand and Supply in the Generation and Diffusion of Technical Change, Harwood Academic Publishers, London, UK.
Van de Ven, A., and Garud, R., 1993, Innovation and industry development: The case of Cochlear implants, in R. Burgelman and R. Rosenbloom, eds, Research on Technological Innovation, Management and Policy, Vol. 5, JAI Press, Greenwich, CT, USA.
Vernon, R., 1966, International investment and international trade in the product cycle, Quarterly Journal of Economics, 80:190–207.
Vernon, R., 1979, The product cycle hypothesis in a new international environment, Oxford Bulletin of Economics and Statistics, 40:255–267.
Walsh, V., 1984, Invention and innovation in the chemical industry: Demand-pull or discovery-push?, Research Policy, 13:211–234.
Wan, H.Y., Jr., 1971, Economic Growth, Harcourt Brace Javonovich, New York, NY, USA, pp. 215–226.
Winter, S.G., 1971, Satisficing, selection and the innovating remnant, Quarterly Journal of Economics, 85(May):237–261.
Winter, S.G., 1984, Schumpeterian competition in alternative technological regimes, Journal of Economic Behavior and Organization, 5:287–320.
Witt, U., 1993, Evolutionary Economics, Edward Elgar Publishing, Aldershot, UK.
Wright, G., 1990, The origins of American industrial success, 1889–1940, The American Economic Review, 80(September):651–667.
Wright, G., 1997, Toward a historical approach to technological change, Economic Journal, 107:1560–1566.
Throughout this volume the terms “technical change” and “technological change” are generally used interchangeably. In Chapters 2 and 3, the use of the term “technical change” reflects the long-standing terminological tradition in the field of the economics of technical change. We have not attempted to standardize the usage across chapters, however, and instead chose to respect the terminological choices of the authors. Eds.