The Coming of Post-Industrial Society

by Daniel Bell

  Yet this earlier pessimism has been belied. Real wages have risen consistently in the past hundred years; the increase in the per capita income in the United States has averaged more than 2 percent a year since 1870. How did the classical economists err so badly? As Professor Lester Lave writes: “Had Ricardo been asked whether increased productivity were possible, he would probably have answered that productivity would increase if capital per worker, including land per worker, were increased.”32 But he could not give a quantitative formulation of this effect. The standard capital-output ratios (known as the Cobb-Douglas function) which have been typically used assume that output will rise 1 percent for every 3 percent of capital increase, holding labor constant. Between 1909 and 1949, the capital per man-hour employed in the private nonfarm sector of the U.S. economy rose by 31.5 percent. On this basis, the increase in goods (per capita output) should have been about 10 percent. But the startling fact is that during this period, with a capital input of 31.5 percent, output per man-hour rose not 10 percent but 104.6 percent. In short, there was an increase in productivity of 90 percent that is unexplained by the increase in capital per worker. The explanation, simple in conception but complex in detail and proof, was supplied by Robert M. Solow in a now classical (or should one say neo-classical?) article, namely, technological change.33 Technology, as we now know, is the basis of increased productivity, and productivity has been the transforming fact of economic life in a way no classical economist could imagine.

  The simple answer, however, begs a complex question. What is technological change? One can say that technological progress consists of all the better methods and organization that improve the efficiency (i.e. the utilization) of both old capital and new. But this can be many things. It can be a machine to forge car engines replacing old hand-casting methods. It can be a physical technique, such as building a ramp to move stones up a pyramid. It can be a simple sociological technique such as a rough division of labor in the construction of a shoe or a sophisticated technique of industrial engineering such as time-and-motion studies. It can be a logical analysis embodied in operations research or a mathematical formula such as linear programming which specifies new queuing tables or the production schedules in which orders are to be filled. Clearly, all of these are incommensurate. How do we combine all these different things under one rubric and seek a measurement?

  What makes the problem all the more vexing is that we are repeatedly told that we are living in a time of “constantly accelerating rate of technological change,” which is creating new and “explosive” social problems. Now, no one can deny that a good deal of technological change has taken place since World War II: atomic energy, electronic computers, jet engines are three of the more spectacular introductions of new products and processes. But the difficulty with the publicistic (and political) argument is that the word “rate” implies a measurement, that somehow the changes that are being introduced now can be measured, say, against the introduction of the steam engine, the railroad, the telephone, the dynamo, and similar technological devices of the nineteenth century. How does one distinguish the change wrought by electricity from that created by atomic energy? We cannot. Both are “revolutionary” innovations. But there is no way of matching their effects in a comparable way. More than that, many social changes are occurring simultaneously, and writers often lump these together as part of the idea of rapid technological change.

  The question of what constitutes the accelerating revolution of our time is too broad and vague. Clearly, it is in part technological. But it is also political in that, for the first time, broadly speaking, we are seeing the inclusion of the vast masses of people into society, a process that involves the redefinition of social, civil, and political rights. It is sociological in that it portends a vast shift in sensibility and in mores: in sexual attitudes, definitions of achievement, social ties and responsibilities, and the like. It is cultural, as we have already noted. Clearly there is no simple conceptual way to group all these together and find a common mensuration. It is simply impossible—if we restrict ourselves to the idea of “change”—to measure “the pace of change.” There is no composite index; we have to be more delimited.

  If we are to restrict ourselves to measuring the technological dimension, and ask about the rate of change, we have, first, to return to the realm where its values (beginning first in the monetary sense) are expressed—to economics.

  For the economist, a technological change is a change in the “production function.”34 Simply stated, the production function is a relationship between inputs and outputs which shows, at any point in time, the maximum output rate which can be obtained from the given amounts of the factors of production. In the simplest cases, the factors of production are assumed to be capital and labor, and the production function would show the most effective combination (the optimal proportions) of factors at given costs.35 Increases in real income per man-hour are a function of both relative increases in capital and the more efficient utilization of resources. Classical economic theory emphasized that the higher levels of real income are generated by increasing the capital stock. But real wages might also be increased by an upward shift in the production function as a result of research that leads to better combinations of resources, new techniques, and so on. In fact, we assume today that technological change, rather than capital stock, is the more effective determinant of higher real wages. Robert Solow, for example, in his aforementioned 1957 paper created an “aggregate production function” (which has been criticized for its assumptions of homogeneity and high elasticity of substitution of capital for labor and capital for capital), which sought to separate the increase in productivity caused by growth in capital from that due to technological change. He found that in the period from 1909 to 1949 the capital increase accounted for approximately 12.5 percent of increased productivity, while technological change accounted for 88 percent.36

  In the broader context, we have to move from production functions to the measures of productivity computed in regular time-series. Conventionally, the gross measurement of technological change is the year-by-year change in the output per man-hour of labor, or what economists call a partial productivity index. It is arrived at by dividing the market value of goods and services produced during a given year (in the economy as a whole, or in a particular industry) by the number of man-hours it has taken to produce them. Productivity so defined in no way identifies whether the increased efficiency has been brought about by new machinery or by a more skilled labor force, or even by a speed-up of work done on the job. Still, if we are to consider the question whether technological change has been vastly accelerating in recent years and at what rate, this is the only consistent measure we have.

  The most comprehensive study of labor productivity in recent years, by John W. Kendrick,37 draws the following conclusions:

  First, during 1889–1957, the nation’s real output per man-hour of work rose at an average rare of between 2 and 2.5 percent. These gains have been widely diffused, resulting in a rapid increase in real hourly earnings and a decline in working hours of between 20 and 30 percent since the turn of the century. According to Kendrick, there was a break after World War I. During 1889–1919, output per man-hour rose at an average rate of 1.6 percent; during 1920–1957, it grew at an average rate of 2.3 percent per year. The reasons for this increase are by no means clear. Kendrick suggests that it may have been due to the spread of scientific management, the expansion of college and graduate work in business administration, the spread of organized research and development, and the change in immigration policy. A similar picture is obtained if one uses the “total” productivity index for computations. The total productivity index, developed by Evsey Domar, relates changes in both labor and capital inputs rather than in labor inputs alone. Kendrick, using this index, estimated that during 1889–1957, total productivity for the entire domestic economy increased by 1.7 percent and that in the period following World War I, this rose to 2.1 percen
t.

  In these measures, the usual assumption is that technological change is essentially “organizational,” i.e. that technological progress consists of better methods and organization that improve the efficiency of both old capital and new. If one tries to measure that aspect of change that is due directly to machinery, rather than just to methods (i.e. time-and-motion studies, linear programming, etc.), these. changes must be capital-embodied if they are to be utilized. For example, the introductions of the continuous wide strip mill in the steel industry and the diesel locomotive in railroads necessitated large capital investments, and we can thus “factor out” those proportions of productivity due to machinery. In a study based on capital-embodied change, published in 1959, Solow estimated that the rate of technological change in the private economy during 1919–1953 was 2.5 percent per year.38

  Most of the analytical and specialized estimates stop at a period of ten years ago. Has there been a recent increase? In the early 1960s, the seemingly persistent high unemployment rates (averaging about 6 percent) gave rise to fears that a rapid increase in automation (which necessarily would be reflected in an accelerating rate of productivity) was responsible for the unemployment. A number of economic writers prophesied so dazzling a rate of productivity increase that the economy would be unable to absorb the new production without making a sharp separation between income and work. In 1965, President Johnson appointed a National Commission on Technology, Automation, and Economic Progress to report on the question, and after a year of study, the Commission concluded that the arguments had been greatly exaggerated. The report stated: “In the 35 years before the end of the Second World War, output per man-hour in the private economy rose at a trend rate of 2 percent a year. But this period includes the depression decade of the 1930s. Between 1947 and 1965 productivity in the private economy rose at a trend rare of about 3.2 percent a year, or an increase of more than 50 percent. If agriculture is excluded, the contrast is less sharp, with the rate of increase 2 percent a year before the war and 2.5 percent after.” 39

  If one moves to the more refined indexes, Kendrick and Sato found that the average annual rate of increase of total productivity in the private domestic economy during the 1948–1960 period was 2.14 percent as compared with a rate of 2.08 percent for the larger 1919–1960 period. Richard Nelson, assuming that technological change was organizational, estimated the average rate of technological change as 1.9 percent in 1929–1947; as 2.9 percent in 1947—1954 and 2.1 percent in 1954—1960. Thus, while there is some evidence that the rate of technological change has been higher since World War II, the difference is much smaller than that indicated by the behavior of output per man-hour.40

  As the President’s Automation Commission concluded:

  Our study of the evidence has impressed us with the inadequacy of the basis for any sweeping pronouncements about the speed of scientific and technological progress. ... Our broad conclusion is that the pace of technological change has increased in recent decades and may increase in the future, but a sharp break in the continuity of technical progress has not occurred, nor [since most major technological discoveries which will have a significant economic impact within the next decade are already in a readily identifiable stage of commercial development] is it likely to occur in the next decade.41

  THE FORECASTING OF TECHNOLOGY

  Even though it is difficult to demonstrate that the “rate” of technological change has leaped ahead substantially in the past decades, it is undeniable that something substantially new about technology has been introduced into economic and social history. It is the changed relationship between science and technology, and the incorporation of science through the institutionalization of research into the ongoing structure of the economy, and, in the United States, as a normal part of business organization. Two things, therefore, are new: the systematic development of research and the creation of new science-based industries.

  Classical economists, even as late as John Stuart Mill, held that population and land were the limiting variables on economic growth and that eventually a prudent economy would end in a “stationary state.”42 Marx, to the contrary, saw that the dynamic of capitalist society was, necessarily, accumulation but that monopoly would inevitably slow down the rate of growth and that the system itself might even break down because of its “contradictions.” Several generations of post-Marxian economists have, in turn, expected a new state of “economic maturity” based either on the exhaustion of markets and investment opportunities in new lands (the theme of imperialism), the slowdown of population growth (a favorite theme of economic pessimists of the 1930s, vide Alvin Hansen 43), or the ending of new technological advances as the “long waves” of business activity because of the waning impetus of the railroad, electricity, and the automobile.

  The bogey of “economic maturity” has by now been largely dispelled. And the principal reason is the openness of technology. In his Capitalism, Socialism and Democracy, published in 1942, Joseph Schumpeter wrote: “We are now in the downgrade of a wave of enterprise that created the electrical power plant, the electrical industry, the electrified farm and home and the motorcar. We find all that very marvelous, and we cannot for our lives see where opportunities of comparable importance are to come from.”

  Though Schumpeter was pessimistic about the future of capitalism (because of the bureaucratization of enterprise and the hostility of the intellectual), he did have a clear view of the promise of technology. Thus he added: “As a matter of fact, however, the promise held out by the chemical industry alone is much greater than what was possible to anticipate in, say, 1880. ... Technological possibilities are an uncharted sea ... there is no reason to expect slackening of the rate of output through exhaustion of technological possibilities.” 44

  In the quarter of a century since Schumpeter made those prophetic remarks, two changes have occurred. One has been the systematic joining of science to invention, principally through the organization of research and development efforts. The second, more recent change has been the effort to “chart the sea” of technology by creating new techniques of technological forecasting which will lay out the future areas of development and which will allow industry, or society, to plan ahead systematically in terms of capital possibilities, needs, and products. This new fusion of science with innovation, and the possibility of systematic and organized technological growth, is one of the underpinnings of the post-industrial society.

  Earlier inventions and innovations were not tied to scientific research. As Nelson, Peck, and Kalachek have observed:

  Compare Watt’s utilization of the theory of latent heat in his invention of the separate condensing chamber for steam engines, or Marconi’s exploitation of developments in electromagnetism with Carothers’ work which led to nylon, Shockley’s work which led to the transistor, or recent technological advances in drugs and military aircraft. In the earlier cases the scientific research that created the breakthrough was completely autonomous to the inventive effort. In the later cases, much of the underlying scientific knowledge was won in the course of efforts specifically aimed at providing the basic understanding and data needed to achieve further technological advances. Carothers’ basic research at Du Pont which led to nylon was financed by management in the hope that improvements in the understanding of long polymers would lead to important or new improved chemical products. Shockley’s Bell Telephone Laboratories project was undertaken in the belief that improved knowledge of semiconductors would lead to better electrical devices.45

  But, they continue, the new industries of the 1970s—the polymers and plastics, electronics and optics, chemicals and synthetics, aerospace and communications—are all integrally science-based.

  The science-based technologies and industries have a great advantage in achieving major advances in products and processes. Research aimed at opening up new possibilities has substituted both for chance development in the relevant sciences, and for the classical major inventive effort aimed a
t cracking open a problem through direct attack. The post World War II explosion of major advances in electronics, aircraft, missiles, chemicals, and medicines, reflects the maturing of the science base in these industries, as well as the large volume of resources they employ to advance technology. Many of the products of the science-based industries are the materials used by other industries, and their improvement has led to rapid productivity growth in many sectors of the economy. The more important new consumer goods have come either directly from these industries or through incorporation into new products by other industries of the materials and components created by the science-based sector.46

  The role of “research and development” as a component of scientific and economic activity will be discussed in a later section. In this discussion on the measurement of technology, we can turn to the new kinds of knowledge represented by technological forecasting.

  Can we predict today better than before—at least in relatively chartable fields such as technology? Three factors distinguish the prediction of today from that of the past: (1) the awareness of the complex differentiation of society and the need, therefore, to define the different kinds of systems and their interrelations; (2) the development of new techniques, primarily statistical and often mathematical, which facilitate the ordering and analysis of data so as to uncover the different rates of change that obtain in different sectors of society; and (3) the sheer quantity of empirical data which allow one to see the detailed components of sectors and to plot their trends in consistent time-series.