Language selection


Archived — Working Paper Number 19: Inter-Industry and U.S. R&D Spillovers, Canadian Industrial Production and Productivity Growth

by Jeffrey I. Bernstein, Carleton University and National Bureau of Economic Research, February 1998


R&D investment generates new products that can be produced relatively more efficiently. Consequently, R&D activities affect living standards. A major reason for the policy focus surrounding R&D activities is that there is a public good aspect to R&D capital accumulation. The benefits of R&D effort cannot be completely appropriated by R&D performers. This means that the benefits of R&D investment spill over to other producers. In particular, there are R&D spillovers that relate to the transmission of knowledge between industries and nations. Indeed, a country's stock of knowledge depends on its own R&D investment as well as the R&D investment conducted in other nations. Thus international spillovers associated with R&D investment imply that national living standards are interdependent.

The purpose of this study is to investigate the extent to which inter-industry and intra-industry R&D spillovers exist from U.S. to Canadian industries, and to determine the production cost, factor intensity (that is, input per unit of output) and productivity growth effects associated with these spillovers. Specifically, we want to investigate how inter-industry and intra-industry spillovers from the United States affect production structures of Canadian manufacturing industries. For example, we address the question of how U.S. spillovers affect labour intensities. We also consider the effects on production efficiency from U.S. spillovers. Efficiency relates to production cost and productivity growth. Productivity growth stresses the temporal impact of R&D spillovers on efficiency, while production cost focuses on the spillover impacts at a given time.

This study is an extension of a 1994 paper by the author (the main parts of which have been published as Bernstein 1996 and 1997), which considers domestic inter-industry spillovers in conjunction with U.S. intra-industry spillovers. The present study examines the effects of spillovers on average variable cost, input-output ratios or factor intensities of labour, intermediate inputs, physical and R&D capital, and productivity growth rates for 11 Canadian industries over the period from 1966 to 1991. The industries examined are chemical products, electrical products, food and beverage, fabricated metals, non-electrical machinery, non-metallic minerals, paper and allied products, petroleum products, primary metals, rubber and plastics, and transportation equipment.

The following conclusions are reached:

  • The nature of U.S. spillovers was tested. A question arises as to whether or not U.S. inter-industry and intra-industry spillovers generate effects on the production processes of Canadian manufacturing industries. In eight of the industries examined, there are no significant international/inter-industry spillovers. In these eight industries, international spillovers are intra-industry. In three industries (namely food and beverage, fabricated metals, and rubber and plastics), international spillovers are both intra-industry and inter-industry. This conclusion is not surprising. International links would tend to be stronger within an industry rather than across industries. In addition, since domestic inter-industry spillovers are influenced by U.S. spillovers in the corresponding industry, U.S. inter-industry spillovers are indirectly related through Canadian spillovers. (Note that data in the analysis are defined at the two-digit Standard Industrial Classification. The data relate to Canadian industries and not to individual firms. Thus it is not possible to consider domestic intra-industry spillovers. By necessity, domestic intra-industry spillovers, if they exist, are assumed to be internalized within the industry data.) In two of the three remaining industries (food and beverage, and rubber and plastics), inter-industry spillovers are continental. In other words, in these two industries the source of inter-industry spillovers is defined by the combination of Canadian and U.S. R&D capital stocks. In the food and beverage industry and the rubber and plastics industry, the U.S. inter-industry spillover generates effects on factor intensities, production cost and productivity growth that are different from the effects associated with the U.S. intra-industry spillover. In the last industry, fabricated metals, international spillovers are both intra-industry and inter-industry. In this industry, the international spillover is defined by a combination of U.S. intra-industry and inter-industry R&D capital stocks.
  • Domestic R&D spillovers cause average variable cost to decrease in seven industries, namely, electrical products, food and beverage, fabricated metals, non-metallic minerals, petroleum products, rubber and plastics, and transportation equipment. The decrease ranges from 0.02 percent for fabricated metals to 0.33 percent for rubber and plastics. The largest cost decreases associated with the inter-industry spillover occur for rubber and plastics. It should be recalled that in this industry inter-industry spillovers are continental, that is both Canadian and U.S. In all cases, cost decreases are highly inelastic.
  • There is a complementary relationship between R&D capital intensity and domestic R&D capital stock in eight industries. In other words, the domestic spillover causes R&D intensity to move in the same direction as the domestic spillover. The eight industries are food and beverage, fabricated metals, non-electrical machinery, non-metallic minerals, paper and allied products, petroleum products, primary metals, and rubber and plastics. A 1.0 percent increase in the domestic spillover increases R&D intensity from a low of 0.01 percent for food and beverage to a high of 0.99 percent for rubber and plastics. With respect to the physical capital, labour and intermediate input intensities, domestic spillovers usually cause these factor intensities to decline. These results, taken together, imply that the majority of Canadian manufacturing industries are becoming more knowledge-intensive. The increase in knowledge intensity arises for two reasons. First, knowledge diffuses among Canadian industries through inter-industry spillovers. Second, in response to these spillovers, industries increase their own R&D intensities.
  • Intra-industry spillovers from the United States exert greater influence on Canadian industries than do domestic inter-industry spillovers. In the cases of the food and beverage industry and the rubber and plastics industry, inter-industry spillovers are continental; consequently, combined U.S. intra-industry and inter-industry spillovers generate greater cost reductions than Canadian inter-industry spillovers. For fabricated metals, international spillovers are both intra- and inter-industry, and these spillovers cause greater cost reductions than domestic inter-industry spillovers. International spillovers from the United States cause variable cost reductions in all industries. A 1.0 percent increase in the U.S. spillover reduces average variable cost from a low of about 0.02 percent in food and beverage to a high of about 0.78 percent in petroleum products. Therefore, at any one time the efficiency gains associated with U.S. spillovers outweigh the efficiency gains from Canadian spillovers.
  • There is generally a complementary relationship between Canadian R&D intensity and intra-industry spillovers from the United States. In two industries (food and beverage, and petroleum products), foreign intra-industry spillovers and domestic R&D intensity are substitutes for each other. In the nine cases where they are complements, the increase in R&D intensity ranges from 0.17 percent for chemical products to about 1.0 percent for transportation equipment, representing a 1.0 percent increase in the foreign spillover. In addition, U.S. spillovers increase physical capital intensities and reduce non-capital input intensities of Canadian manufacturing industries. Therefore, U.S. spillovers cause Canadian manufacturing production to become more intensive with respect to physical capital and knowledge, and less intensive with respect to labour and intermediate input.
  • For seven of the manufacturing industries examined, U.S. intra-industry R&D spillovers are the major reason for productivity gains. The percentage contributions range from around 58 percent in transportation equipment to 100 percent in petroleum products. Spillovers are also the main contributor to total factor productivity (TFP) growth for fabricated metals and for rubber and plastics. In the former case, the spillover is the combined Canadian and U.S. inter-industry spillover. In the latter case, the spillover is the combined intra- and inter-industry spillover from the United States. In two industries (chemical products, and food and beverage), output growth, through scale, dominates the elements of TFP. However, even in these industries, the U.S. spillover contributes to productivity gains. Over time, therefore, production efficiency is relatively more affected by U.S. spillovers than by spillovers from Canadian industries.
Date modified: