Executive summary

View the most recent version.

This paper examines changes in capital intensity in Canada since 1987 and compares them with those in the United States. Capital–output ratios are used as the metric for comparison.

Capital–output ratios as summary statistics tell us about the nature of the production process. Higher capital output ratios indicate that more capital is being used in production. Changes over time in these ratios have been used to generate conclusions about capital productivity or about the extent to which technological progress is mainly labour enhancing.

When comparing capital intensity across countries, one must grapple with potential differences in methods and data sources. This paper approaches this problem by standardizing one critical set of assumptions so as to produce a more comparable set of estimates. It adopts a common set of depreciation estimates for the two countries and then estimates capital stock using the perpetual inventory technique.

Making use of a comparable set of depreciation rates changes the nature of conclusions about the capital intensity of the two countries. If comparisons of capital intensity are made with the depreciation rates that are used by Statistics Canada in its productivity program and by the Bureau of Economic Analysis in its estimates of capital stock, then the stock of capital as a share of gross domestic product (GDP), or capital intensity, is lower in Canada. However, once we impose common depreciation rates for similar asset classes, Canada's overall capital intensity is higher than that of the United States.

An examination of investment-to-GDP ratios by asset class reveals substantial differences, both in terms of level and trends. Canada's engineering investment-to-GDP ratio is much higher than that of the United States. By contrast, there has been a persistent information and communications technology (ICT) investment-to-GDP ratio gap between Canada and the United States that has increased over time. At an aggregate level, the non-ICT machinery and equipment (M&E) and building assets intensities are more alike in the two countries.

There is always the danger when working with macro data for the entire economy that important differences in the underlying structure of the economy will be missed. Therefore this paper examines the extent to which differences in capital intensity are related to differences in the underlying industry and asset structures of the two countries. It uses a decomposition analysis to examine whether there are substantial differences in the underlying components (both assets and industries) in Canada and the United States.

While Canada has a higher overall capital intensity in the business sector—because it has relatively more engineering assets (pipelines, dams, railways) and less ICT assets—the relative amounts of non-ICTM&E and building assets are more alike in the two countries. This suggests a different aggregate production function that stems from a different industrial composition or from differences in production techniques that are associated with a different economy.

To address this issue, the paper uses a shift-share decomposition analysis to examine whether the higher Canadian capital intensity is the result of differences in asset or industry composition. This analysis shows that Canada's business sector is more capital intensive primarily because of both its industry structure and its focus on engineering assets. Two sectors, the other primary sector (including mining) and utilities, are very intensive in engineering capital in Canada; together, they contributed the preponderance of the intensity effect advantage over the United States.

The industries where engineering assets are concentrated are the core infrastructure industries that provide universal services on which the rest of the economy relies—transportation, communications and energy. These industries are more important in Canada, which may reflect the fact that Canada has a comparative advantage in some natural resource sectors that are associated with these industries and that the Canadian economy is more diverse geographically and requires more of the services of these sectors per unit of GDP produced than does the United States.

The non-ICTM&E capital intensity in the Canadian business sector is 14% higher than in the United States because Canada tends to concentrate more on industries with higher non-ICTM&E capital intensity. When the difference in the industrial structures in the two countries is controlled for, the non-ICTM&E capital intensity is 12% lower in Canada.

The largest capital intensity gap exists in ICT. Canada's ICT capital intensity has been persistently lower than that of the United States since at least 1987. The gap was fairly widespread across industries in 2003, with differences in industrial structure not playing a significant role. The ICT intensity gap was particularly large in construction; transportation, warehousing and utilities; and in the finance, insurance, real estate and rental and leasing sector.

This paper also examines the relative capital intensity of the non-business sector (government, health and education) in the two countries. Capital intensity here is quite similar—at least when it comes to the use of buildings and engineering infrastructure—but it is extremely different when it comes to expenditures on M&E. The latter arises in part because of military equipment spending in the United States. Inclusion of this component in comparisons of the total economy makes it appear that Canada is deficient overall in M&E, even though this is not the case for the business sector.