by Jeffrey I. Bernstein, Carleton University and The National Bureau of Economic Research, June 1996
Summary
Productivity and the factors of production are the two sources of output growth in an economy. Productivity growth generally arises from technological improvements, scale economies and other sources of efficiency gains over time. Indeed, it is important for a society to have a stable and positive long-term productivity growth rate, because the associated gains lead to improvements in living standards.
It is primarily advances in the state of knowledge through technological change which determine productivity growth over long periods of time; research and development (R&D) investments contribute directly to knowledge accumulation. Investment in R&D generates new products and production processes, and thereby contributes to productivity improvements. A distinctive feature of R&D investment is that the benefits from R&D activities spill over among firms and other organizations. Productivity growth in an industry, therefore, depends on its own R&D activities, as well as on the R&D efforts of other knowledge-generating industries. This implies that productivity growth is influenced by joint cumulative R&D activity. The significance of R&D spillovers in generating productivity growth has stimulated a growing interest in the sources of R&D spillovers. It has also been observed that high-tech industries exhibit relatively high rates of productivity growth and are important sources of R&D spillovers. Firms operating in the Canadian communications and other electronicequipment industries are centres of knowledge-based activity. Hence, this industry provides an important opportunity to consider the role of R&D investment generally in improving productivity performance.
This paper addresses three major issues.
The first issue has to do with estimating the effects of R&D spillovers from the communications equipment industry on the structure of production or factor intensities (i.e., the labour-output, intermediate input-output, physical capital-output, and R&D capital-output ratios) of the Canadian manufacturing sector (measured as net of the communications equipment industry). In addition, because spillovers from the United States have significant effects on Canadian factor requirements, those emanating from the U.S. manufacturing sector are also included. In order to determine whether the production structure of the communications equipment industry differs from that of other manufacturing industries, the effects of R&D spillovers on factor intensities in the communications equipment industry are also estimated. In this case, R&D spillovers derive from both the Canadian manufacturing sector and the U.S. electrical products industry.
The second issue relates to the contribution of R&D spillovers to productivity growth. Productivity growth in Canadian manufacturing is measured and decomposed, so that the sources of growth, and especially the contribution of spillovers from the communications equipment industry and the U.S. manufacturing sector, can be determined. In addition, a similar analysis is conducted with respect to productivity growth in the communications equipment industry, where spillovers emanate from the Canadian manufacturing sector and the U.S.electrical products industry. The results of these analyses enable us to consider the extent to which R&D capital accumulation by producers in one segment of the economy influences productivity growth of producers in other industries or sectors.
The first two issues address the effects of R&D spillovers from the viewpoint of its user or receiver. Turning to the source ofspillovers, the third issue pertains to an estimation of the private and social rates of return to R&D capital. Private rates of return measure the benefits that accrue to those engaged in R&D activities; social rates of return measure the benefits that accrue to the users of the investment.
Several conclusions are reached in this paper.
First, between 1966 and 1991 the average annual rate of productivity growth in the communications equipment industry was 1.24 percent, while the growth rate for manufacturing averaged 0.50 percent. Thus, the rate of productivity growth in communications equipment was 150 percent higher than in manufacturing. Moreover, unlike manufacturing (which suffered a productivity slowdown in the post-1973 period — from 1.08 percent to 0.23 percent), the average annual productivity growth in the communications equipment industry increased by 46 percent, from 0.94 percent to 1.37 percent.
Second, as a source of spillovers, the communications equipment industry affects the production structure (i.e., factor intensities) of the manufacturing sector. Indeed, it is estimated that a 1 percent expansion of R&D capital in the communications equipment industry leads to an increase of 0.15 percent in knowledge intensity for the entire manufacturing sector. This magnitude is quite large in light of the fact that this effect emanates from a single three-digit standard industrial classification industry. In addition, spillovers from the communications equipment industry reduce factor intensities associated with physical capital, labour and intermediate inputs (such as materials).
In terms of relative importance, the R&D spillovers from the U.S. manufacturing sector generate greater effects on Canadian manufacturing factor intensities than spillovers from the communications equipment industry. It is noteworthy that the R&D capital inputs between the two North American manufacturing sectors are substitutes. A 1 percent increase in U.S. manufacturing R&D capital leads to a 0.52 percent reduction in the domestic knowledge intensity in Canadian manufacturing.
Third, factor intensities in the Canadian communications equipment industry are affected by spillovers from both Canadian manufacturing and the U.S. electrical products industry. Both sources of spillovers reduce labour and intermediate input intensities and increase the intensity of physical capital. However, the spillovers from Canadian manufacturing reduce the R&D intensity of the communications equipment industry. A 1 percent increase in the R&D capital of Canadian manufacturing leads to a 0.38 percent decline in the R&D intensity of the communications equipment industry. Combining the results of spillovers between communications equipment and manufacturing, we can see that expanding R&D capital in the communications equipment industry increases the R&D intensity of manufacturing production. This result then mitigates the need (when all other elements are held constant) for further R&D expansion in the communications equipment industry. The spillovers from the U.S. electrical products industry increase the R&D intensity in the Canadian communications equipment industry. Between these two industries, R&D capital stocks are therefore complementary. In addition, R&D capital from the U.S. electrical products industry has a greater impact on the production structure of the Canadian communications equipment industry than R&D capital from Canadian manufacturing. A 1 percent increase in R&D capital in the U.S. electrical products industry causes R&D intensity to rise by 0.65 percent in the Canadian communications equipment industry.
Fourth, the communications equipment industry is a source of important productivity gains in Canadian manufacturing. Between 1966 and 1991, about 8.5 percent of the average annual rate of productivity growth in manufacturing was accounted for by spillovers from the communications equipment industry. Moreover, this contribution increased during the post-1973 period when the productivity slowdown occurred. Thus, the spillovers were a mitigating influence on the further erosion of productivity performance in Canadian manufacturing. However, it should be recognized that the spillovers from U.S. manufacturing were the major contributor in this regard — accounting for 76 percent of the average annual rate of productivity growth in Canadian manufacturing. Spillovers from Canadian manufacturing and the U.S. electrical products industry contributed to productivity growth in the Canadian communications equipment industry. The spillovers from Canadian manufacturing accounted for only about 6 percent of productivity growth, largely dominated by the spillovers from the United States. However, the main source of productivity in the Canadian communications equipment industry was the scale economies associated with output growth, which accounted for 65 percent of productivity growth.
Fifth, the fact that the Canadian communications equipment industry and manufacturing sector are sources of productivity gains implies that there are extra-private returns to their R&D capital. The before-tax, gross of depreciation private rate of return to R&D capital averaged 17 percent between 1966 and 1991. The social rate of return pertaining to Canadian communications equipment R&D capital is estimated at 55 percent, or 225 percent higher than the private rate of return. The social rate of return associated with Canadian manufacturing R&D capital is estimated at 21 percent, or 24 percent higher than the private rate of return. These differences point to an under-investment in R&D. However, this does not mean that governments should target the Canadian communications equipment industry for special status. Although its high annual rate of productivity growth and high social rate of return distinguish this industry, there are other manufacturing industries whose social rates of return to R&D capital exceed the private rate of return.
Finally, R&D investment should be encouraged through policy instruments that focus on R&D capital formation, but these policies should not be directed towards particular industries. There are several possibilities in this regard. The government could provide information to facilitate joint ventures aimed at new product development, and joint research or "laboratory" ventures. Legislation and regulation could be amended in order to reduce the transaction costs associated with these joint ventures. Reducing the legislative and regulatory burden would also help to encourage other more indirect means of internalizing the spillovers arising from R&D. Licensing agreements are one example that come to mind.
Tax expenditures and subsidies are other policy instruments that are — and can be — directed towards R&D capital formation. It is important to recognize that any analysis of the relative costs and benefits of government tax policies aimed at R&D investment must take into account R&D spillovers. Otherwise, the benefits associated with these policies will be underestimated, not only in the way they encourage R&D investment, but also in their contribution to improving living standards through higher rates of productivity growth.