Economic growth, by which I mean increases in per capita income, can be achieved by bringing more resources into productive use or by deriving more output from a given amount of resources. During the nineteenth century, U.S. residents' incomes rose through tapping vast agricultural and mineral resources as well as by borrowing technology from Europe and inventing technology at home. But in this century, and for the foreseeable future, economic growth must come primarily from increased productivity, or extracting more output from a given set of inputs.

Productivity growth, however, has slowed down markedly over the past quarter-century. Output per labor hour, the standard productivity measure, grew at an average rate of 3.1 percent per year between 1947 and 1973, but at only 1.3 percent per year between 1973 and 1994.

Not surprisingly, this disappointing record, with its implications for economic well-being, has caused concern.

The retardation of productivity growth has numerous causes. With the highest absolute productivity levels among the world's industrialized nations both now and in 1947, the United States must push harder on the frontiers of technology to achieve rapid productivity growth than less advanced nations. Productivity and its growth depend not just on technology; workers' education and skills, the sectoral composition of national output (among agriculture, manufacturing, and services), the quality of business management, the competence of government regulation, and the business cycle all play a role. Most critically for business and for government policies that directly affect business, productivity growth depends on the level of investment that expands the stock of capital with which workers do their jobs.

The Role of Technology

Determining the role of the various factors raising productivity is further complicated by powerful interaction effects. A well-educated workforce is more adept at creating various technological advances that make productivity growth possible; it is also better able to implement the new technologies when they become available. Many technological innovations are "embodied" in new machines and processes that require complementary capital investment before they can work their magic. This embodiment makes it particularly difficult to sort out the effects of technology from the effects of capital investment.

Our imperfect understanding of "embodied" technical change underlies many of the disagreements among economists concerning the role of technology in economic growth. This confusion is critical because technological progress in the United States typically originates in one industry and is sold to another in the form of improved machines, software, material inputs, and the like (see flow chart). Because the interaction effects are hard to identify, quantitative studies often fail to give technological progress sufficient credit for raising U.S. productivity.

Most studies of the links between new technology and productivity growth use expenditures on research and development as a surrogate measure of technological change. R&D performed in an industry is assumed to enhance productivity in that industry only. But since most industrial R&D is embodied, the buying and using industries, and also individual consumers, capture the increased productivity or quality-enhancing benefits. A new turbojet engine thus elevates productivity in the air transport industry; a new microprocessor raises the productivity of computer users; and a new antidepressant improves consumers' quality of life along with their productivity as workers. Most studies overlook these links and thus underestimate technology's contribution to productivity growth in both the user industries and the economy at large.

Studies that acknowledge these links also are hampered by statistical measurement problems. Take as an example a better machine that augments the productivity of user industries. The value of that contribution is partly captured by the makers of the machine in the form of a higher price. But some of the added value must be passed on to the users, for otherwise they would purchase the innovation at best reluctantly.

Government statistical agencies have trouble measuring actual quality improvements in their price indices and typically underestimate the value of technologically new products. So economists who use official price statistics to measure real output and discover the sources of economic growth attribute too much to investment per se and too little to technological enhancements.

Policy Considerations

When we turn our attention to policies that could raise the nation's rate of economic progress, arguments about the respective roles of technology and investment might not be so important. This is so because technology embodied in capital goods dominates interindustry technology flows. In 1974, an estimated 45 percent of all industry R&D went into developing capital goods that were sold to other industries.

Policies that stimulate capital investment enhance the rate at which new technology is brought into use and thereby raise the rate of productivity growth. Initiatives that lower the real cost of capital make capital goods investments turn the corner to profitability earlier than they otherwise would. Such policies include reducing taxes on capital or a firmer assurance of inflation-free growth.

The effect of lower capital costs tends to be stronger for decisions to replace long-lived assets that have not yet reached their physical life limits. Examples include a middle-aged blast furnace or a programmable machine tool. The effect is less significant for relatively short-lived assets like trucks or high-tech products such as computers, for they are likely to be replaced by improved models whether or not interest rates vary by a percentage point or two.

Policy-induced capital cost reductions thus mainly affect longer-term investment, which in recent years accounted for less than half of all U.S. business plant and equipment investment. There is very little solid evidence, however, to tell us exactly how much productivity might rise in response to a policy-induced reduction in capital costs.

R&D itself is also investment. Its volume depends in part upon the cost of investable funds, which can be influenced by government policies. The most direct way to stimulate productivity-enhancing R&D would be to put the R&D tax credit, applicable at diverse rates since 1981, on a permanent legislative footing. Taxing capital gains at a preferential rate, or reducing the inflation tax on capital gains through indexation, would also encourage investment in high-technology startup companies, which are responsible for a disproportionate share of technological innovations in U.S. industry.

How much such policies will stimulate R&D and raise productivity remains unclear. Most high-technology ventures depend in complex ways upon the parallel advance of scientific knowledge, which is largely indifferent to federal tax policy (although not spending policy). And high-technology success rates are modest; the most successful 5 percent of venture capital start-ups yield 40 to 50 percent of the total capital gains from venture investments. During the past decade, funds for high-technology venture investment have tended toward glut supply conditions owing in large part to the relaxation of "prudent investor" constraints on institutional investors. These investors, moreover, are generally untaxed and thus less directly influenced by R&D and capital gains tax preferences.

Technological advances are without doubt the most important contributor to economic growth today. They can be influenced by governmental policies that directly encourage industrial R&D and by policies that increase investment in plant and equipment. But while the possibilities for raising productivity growth rates through such policies are clear conceptually, their effects are very hard to predict quantitatively.

Technical progress and productivity growth are far more difficult to accelerate than a car or space vehicle. While speeding the output of important innovations is critically important to our economic prospects, actually doing so poses difficult challenges.

F.M. Scherer is Larsen professor of public policy and management at Harvard University.