The traditional method for analyzing productivity in economics has been to use a mathematical construct, called a production function, to interrelate the quantity of output produced with quantities of inputs utilized - primarily capital and labor.2 Labor is measured in terms of full-time-equivalent employees per year, and capital in terms of how much the equipment and facilities would cost if they were rented for a year. Productivity is said to increase when more output can be produced with the same amounts of inputs. When observed changes in productivity can be associated with R&D activities, the value of those productivity changes is considered a return to R&D investment.
With regard to scientific research, productivity changes may best be associated with the link between (1) new equipment, facilities, and organizational structures, and (2) cost reductions in the supply of existing goods and services. Studies in this area range from examinations of a specific innovation and the particular effects that it had to examinations of aggregate estimates of productivity growth for an entire economy as a result of all R&D performed. These studies focus on particular types of research, such as academic, government supported, and private, and on types of economic effect, such as cost reduction and quality improvement to the organization performing the research, as well as spillover effects that benefit those who did not pay for the research.
Terleckyj (1980, p. 376) divided the effects of private R&D into two groups: "(1) direct increases in productivity of industries conducting the privately financed R&D, and (2) indirect increases in productivity of industries purchasing capital and intermediate inputs from the industries conducting the privately financed R&D." He found the latter effect to be greater, on average, across all industries. In another study, Levy and Terleckyj (1982) examined government financed R&D and observed that it had the effect of stimulating additional private R&D expenditure. Similar complementarity between government and private R&D was observed again by Leyden and Link (1991).3
Economic analysis of R&D investments has progressed over the past 30 years, and although there are differences in estimates of the exact levels of returns, the leading researchers in the field agree that R&D offers high private and social returns in terms of high productivity. (See Rates of Return.) It should be noted, however, that the precise magnitude of these returns cannot be measured without the use of simplifying assumptions in the analysis. A recent survey article by Nadiri (1993) examined 63 studies in this area published by prominent economists, mostly in reference to the United States, but also in reference to Japan, Canada, France, and Germany. Looking at the results of these studies, he concluded that R&D activity renders, on average, a 20-to 30-percent annual return on private (industrial) investments. R&D renders a much greater return to society overall. Estimates of these "social rates of return" range from 20 to 100 percent, with an average of approximately 50 percent (Nadiri, 1993). These figures for the aggregate effects of R&D are consistent with the high returns often found in the analysis of individual innovations. For example, Mansfield (1994) examined three studies of sets of specific innovations, which reported social rates of return of 56, 70, and 99 percent. Recent work by Brynjolfsson and Hitt (1993) found the returns from information technology to be quite high, exceeding 80 percent per year from 1987 to 1991 (Magnet, 1994). Trajtenberg (1990), whose work is described in greater detail below, found the social rate of return from innovations in a particular type of medical equipment (computed tomography scanners) to be as high as 270 percent per year. This is not to say that every research project has a high, or even a positive, rate of return, but portfolios of scientific research projects selected for analysis have the rates of return cited above.
Several recent studies on U.S. industries estimate average rates of return from R&D investments. (See text table 8-1.) The rates of return are for individual firms ("firm-level" studies) or for industries as a whole ("industry-level" studies, often by standard industrial classification [SIC]). These rates differ, depending on the time period being studied, the type of data, and the analytical method employed. Nevertheless, taken together, they reveal that R&D does tend to render high rates of return.
Griliches (1994) examined data on different industries from 1958 to 1989 and provided strong evidence that industrial productivity increases with increased R&D expenditure as a proportion of sales volume. He noted that the computer industry is a major outlier in the data. It has both the highest productivity growth and the highest ratio of R&D to sales. With computers left out of the data, the relationship between R&D and productivity is not as strong, though it is still substantial. Although R&D is a statistically significant, contributing factor to productivity growth, productivity differences across sectors of the economy cannot be explained by R&D alone.
Because of difficulties in assessing the economic value of new and improved goods and services, it has always been difficult for economists to measure productivity precisely. Higher quality goods and services can have the same prices or be measured in the same quantities as lower quality goods and services. As a result, the real economic value of new goods and services is sometimes underestimated, along with the productivity of the industries that produce them (Gordon, 1990).
Many other types of measurement problems can arise. Suppose a newly developed computer costs the same amount to produce and to operate as its predecessor, but it performs twice as many calculations per second. Intuitively, the new computer is twice as productive as the former one.4 Information about product characteristics is often unavailable, however, and even when it is available, its relationship to productivity can be extremely complicated. The productivity of the improved computer may be less than twice that of the former computer or more than twice that of the former computer, depending on how much of the capacity is tapped. Without such information, it is difficult for any researcher to determine the extent to which productivity increases are attributable to the new computer.
Traditionally, economists have evaluated a product's value using price. However, for instance, a new computer's price is not a good indicator of its productive value. Price depends on several factors not necessarily related to the computer's productivity, such as the costs of producing the unit, the competitiveness of the market, and speculation among computer buyers about the future availability of another even better computer (which would lower the price of the existing computer by reducing demand for it). Finally, the improved productivity of the firm that purchases the new computer may be strongly interrelated with the application of new software associated with the new computer, as well as its employees' programming skills.
These concerns regarding productivity measurement, especially as they relate to the quantification of new forms of capital equipment, have long been recognized by economists. In one of the earliest works on the productivity effects of R&D, Solow (1957, p. 312) notes that his analysis is limited by "the profound difficulties that stand in the way of giving any precise meaning to the quantity of capital." Given the complexity of production processes, the difficulties of calculating productivity changes attributable to R&D involve not only informational difficulties, but often analytical and philosophical ones as well (Blaug, 1992; Boskin and Lau, 1994; Chase, 1979; Eichner, 1983; Griliches, 1992). On the other hand, such limitations have tended to produce underestimates, more than overestimates, of the importance of R&D (Nordhaus, 1994). Consequently, the high rates of return to R&D observed by economists serve as evidence of the importance of scientific research to the Nation's economy.
When investment in physical capital or in inventories of raw materials yields a high return, it is often due, indirectly, to scientific or engineering advances, even though the investment itself does not fall under the heading of R&D. In effect, R&D conducted in one industrial sector has beneficial spillover effects in another, in the form of returns to investors and consumer satisfaction. These ripple effects have continual and strong influence on investment opportunities, many of which are not directly associated with scientific research.
These results should be viewed in a larger context. Although average rates of return to R&D investment have been estimated to be quite high, economists have also observed equally high, and sometimes even greater, rates of return for other types of investment by firms. Non-R&D investments could include, for example, enhancement of productive capacity through the acquisition of new machinery, advertising, marketing research, building up of inventories, hiring consultants to improve managerial practices, and speculative purchases of assets in anticipation of price increases. Because results of research cannot always be predicted and often require a long time to develop, individual R&D investment decisions carry an element of risk and, in many cases, R&D may not be the most profitable investment a firm could make; the best type of investment would depend on the circumstances particular to a firm.
3 The literature is quite extensive in this area, and therefore, an adequate literature review would not be possible within the confines of this chapter.
4 In a more complicated analysis, one might consider the economic concept of diminishing returns, in which the additional calculations are not as valuable as the first set of calculations.
| | | | | |
| | | |
Page | Previous | Next | Chap | Refs | Seind96 | Abbrev | Help | SRS |