Technology and Economics (1991)

Dale W. Jorgenson

More rapid growth in productivity is essential for achieving the goals of U.S. economic policy. The slowing of U.S. economic growth in the 1970s can be attributed in large part to the decline in productivity growth. Productivity growth is an important component of the increase in our standard of living. More recently, the rise of the U.S. current account deficit in the 1980s is often ascribed to more rapid productivity growth in other countries, especially Japan and the Four Dragons of East Asia—Hong Kong, Korea, Singapore, and Taiwan. Productivity growth is necessary for enhancing our international economic competitiveness.

It is important to recognize at the outset of our discussion that the productivity problem is enormously complex, involving the performance of our whole economy and, in some ways, our whole society. The solution of the productivity problem will require rethinking our approach to economic policy. A new approach to economic policy is suggested by the idea that improving the performance of the U.S. economy requires investing in productivity. Newly available data on productivity make it possible to identify opportunities for productivity-enhancing investments. These investments will take many forms, but we can identify three that are critical to future productivity growth:

1. Tangible assets. This is the conventional meaning of investment. It includes investment in plant and equipment for the business sector, housing and consumers' durables for the household sector, and military equipment and civilian infrastructure for the government sector. It is useful to think of this as investment in hardware.

2. Intangible assets. Investments require the commitment of capital

resources and produce changes in technology that promote productivity growth. However, not all investments take the form of bricks and mortar. A rapidly growing portion of investment takes the form of research and development, advertising and marketing, and intangibles such as computer software. I will refer to investment in intangibles as software.

3. Human capital. The most important component of investment in human capital is schooling. Formal schooling extends all the way from kindergarten to the most specialized forms of higher education. This was the focus of the Charlottesville education summit organized by President Bush in the fall of 1989. However, investment in human capital also includes training on the job. Mincer (1989) has estimated that training costs in the United States amount to 35-42 percent of schooling costs. I find it useful to refer to education and training as investments in people.

My next objective is to develop a perspective on opportunities for investing in productivity. For this purpose I will summarize data on productivity that are presented in more detail in my book with Frank Gollop and Barbara Fraumeni, Productivity and U.S. Economic Growth, published by the Harvard University Press in 1987. A very important feature of these data is that we can identify specific channels for the impact of investments in hardware, software, and people on productivity growth.

Productivity is the ratio of output to input. I use gross domestic product (GDP) as a measure of output. This includes the output of all economic activities in the United States, whether conducted by Americans or foreigners. By contrast the gross national product (GNP) is output by Americans, whether at home or abroad. I will take hours worked as a measure of input, since this represents the most rudimentary measure of effort. Productivity is defined as gross domestic product per hour worked. This output is produced by combining labor in the form of hours worked with all the forms of capital I have mentioned—hardware, software, and people.

At this point I need a distinction that is crucial in interpreting productivity growth. This is between the productivity growth that can be attributed to investment and productivity growth that does not require investment. Economists refer to the first as "explained" by the commitment of greater resources and the second as the "unexplained" residual. It is important to understand the basis for this distinction, since it is central to deriving the implications of productivity growth for economic policy.

We can illustrate the distinction between explained and unexplained productivity growth by the familiar example of mechanization. If we

replace 10 people with 10 shovels by 1 person with a power-driven digging machine, we have mechanized the digging process. Economist's jargon for this transformation is to say that capital in the form of hardware is substituted for labor in the form of hours worked. This is a simple but useful illustration of productivity growth as investment in hardware. Output per hour worked increases as a result of investment in construction machinery.

A summary of U.S. productivity growth over the period 1947-1985 is presented in the following table:

The explanation of productivity growth as investment in hardware, software, and people is the key to deriving the implications for economic policy. In the example of mechanization, productivity growth results from investment in hardware. We need additional saving and investment to obtain gains in productivity. However, not all investments involve hardware and not all productivity growth involves investment. The data in the table provide the latest information on sources of growth in productivity. To interpret the results I need to relate investments in hardware, software, and people to the example of substitution of capital for labor.

First, the simplest form of substitution of capital for labor is the substitution of hardware for hours worked. In the table this is labeled "capital-labor" substitution. The quotation marks indicate that this is only one of many possible forms of substitution of capital for labor. This accounts for about one-fifth of the productivity growth (19 percent to be precise) during the postwar period. We can also substitute people for people to increase productivity. This results from hiring more highly educated and trained people in place of less well educated and trained people. Education and training both require substantial commitments of resources to produce growth in productivity. Investments in people, which are labeled growth in labor quality in the table, account for about one-fifth of productivity growth (19 percent) over the postwar period.

The third kind of substitution included in the table is substitution

among different types of hardware or tangible capital. Obviously, machines differ in effectiveness, just as people do. By increasing our investment in hardware we can obtain more effective plant and equipment, which is a very important form of substitution. This is encompassed in the measure of capital quality given in the table. Growth in capital quality accounts for more than a quarter (28 percent) of growth in productivity during the postwar period. This is almost 50 percent more important than growth in labor quality or ''capital-labor'' substitution. Growth in capital quality and "capital-labor" substitution constitute the contribution of investment in tangible assets to productivity growth. This amounts to almost half (47 percent) of productivity growth and is by far the most important source of increases in productivity.

The contribution of investments in hardware and people together make up nearly two-thirds (66 percent) of growth in productivity during the postwar period. To complete the picture, we require an estimate of the contribution of investment in software to productivity growth. For this purpose we use a "high" estimate of the contribution of research and development presented by Griliches (1988). This estimate is based on a detailed analysis of data on research and development investment by individual firms. The contribution of investment in software to productivity growth (12 percent) is considerably less than that of investment in people (19 percent). It is important to keep in mind that this can be regarded as an upper bound to the contribution of research and development expenditures.

Up to this point we have focused attention on the sources of growth in productivity that can be attributed to the commitment of additional resources through investments in hardware, software, and people. This leaves an "unexplained" residual, which accounts for about one-fifth (22 percent) of productivity growth. Abramovitz (1962) has referred to this residual as the Measure of our Ignorance. Our overall conclusion is that almost four-fifths (78 percent) of productivity growth can be explained by investments in hardware, software, and people. Growth in productivity is primarily the result of mobilizing investment resources and deploying them efficiently. The presence of the unexplained residual is useful in reminding us that additional economic research on the sources of productivity growth will be needed to provide a complete explanation.

The first line of this table shows that productivity growth has averaged 2.10 percent per year over the postwar period 1947-1985. This estimate is based on my estimates with Gollop and Fraumeni (1987), brought up to date in my recent paper (1991). Since 1979, the growth rate of productivity has been halved, running at only 1.05 percent.

This is about half the average rate of productivity growth during the postwar period, 1947-1985. An important part of the explanation of this decline is the near disappearance of the "unexplained" residual. This contribution to productivity growth has fallen from 0.46 to 0.06 percent per year and accounts for only 6 percent of productivity growth during the period 1979-1985.

For this period the relative importance of investments in hardware, software, and people in explaining productivity growth is much greater than for the entire postwar period. However, the role of these three components of investment has changed. For example, the contribution of "capital-labor" substitution, trading off additional capital against fewer hours, has fallen dramatically from 0.41 percent to 0.14 percent per year. Growth in capital quality has only declined from 0.58 to 0.31 percent per year. Taking these two contributions together, the relative importance of investment in tangible assets has declined from 47 to 43 percent of productivity growth.

Unfortunately, we do not have the data required to estimate the contribution of investment in research and development to productivity growth separately for the period 1979-1985 and the postwar period as a whole. We assume that this contribution remains unchanged, which is consistent with the findings reported by Griliches (1988). By contrast the contribution of investment in human capital has declined from 0.39 to 0.29 percent per year, but the relative importance of this investment has risen from 19 to 28 percent of productivity growth. Our assumption that the contribution of investment in research and development has remained unchanged produces a doubling of the relative importance of this form of investment. This is the consequence of the halving of productivity growth during the period 1979-1985.

We now come to the bottom line: What is the contribution that economic policy can make to productivity growth? The answers are:

1. Investment in hardware. The government can design a tax system that will promote saving and investment and ensure that capital is allocated efficiently among alternative uses. The government can also design economic regulation in such a way as to minimize loss in efficiency. The government also has special responsibilities for investment in civilian infrastructure and defense equipment. The role of these investments in productivity growth is not well understood, at least by economists, and is allocated to the unexplained residual in the estimates we have presented.

2. Investment in software. The federal government can promote investments in research and development, advertising and market-

ing, and other forms of intangible assets. The government finances an important part of research and development and conducts research in its own laboratories in such areas as health, energy, and defense. Efficient allocation of the funds available for these forms of investment is an important issue in science and technology policy.

3. Investment in people. The most important investments in people are made in the educational system. State and local governments have the main responsibility for these investments. However, the largest part of the cost of these investments is contributed by individuals through the time and effort that they devote to education.

Economists have not arrived at a complete understanding of the generation of productivity growth. However, the picture that has emerged from recent research is clear enough. Gains in productivity are due primarily to investments in hardware, software, and people. To stimulate productivity growth we need to mobilize capital through savings and deploy the resulting investments as efficiently as possible. Decisions to save and invest take place in business, household, and government sectors. Businesses accumulate tangible and intangible assets and invest in human capital through on-the-job training. Households make important investments in housing. They also invest in human capital through undertaking formal education and undergoing on-the-job training.

Ensuring continued growth in productivity involves a wide range of government policies: tax and regulatory policies for tangible assets, science and technology policies for intangibles, and education and training policies for human capital. The first part of the solution to the productivity problem is to generate new investments in hardware, software, and people. Equally important is to allocate these investments in the most efficient way. Only by using our scarce investment resources efficiently can we obtain the productivity gains that are essential to growth in our standard of living and restoration of our international economic competitiveness.

Abramovitz, M. 1962. Economic Growth in the United States. American Economic Review 52(4)(September):762-782.

Griliches, Z. 1988. Productivity puzzles and R&D: Another nonexplanation. Journal of Economic Perspectives 2(4)(Fall):9-22.

Jorgenson, D. W. 1991. Productivity and economic growth. In Measurement in Economics, E. Berndt and J. Triplett, eds. Chicago: University of Chicago Press.

Jorgenson D. W., F. M. Gollop, and B. M. Fraumeni. 1987. Productivity and U.S. Economic Growth Cambridge, Mass.: Harvard University Press.

Mincer, J. 1989. Job Training: Costs, Returns, and Wage Profiles. Department of Economics, Columbia University, September.