Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
Statistics Canada’s productivity program was initiated in the late 1940s in response to demands from the user community to provide background summary statistics that would enable analysts to track Canada’s economic progress. It was a result of recommendations from an interdepartmental committee on productivity analysis, who reviewed conceptual and measurement problems, and available data sources.
Initially, the program yielded only an estimate of labour productivity, measured as gross domestic product per worker. Statistics Canada recognized the desirability of including additional inputs and incorporating developments in the literature on productivity measurement. More comprehensive multifactor productivity measures were suggested that combined labour productivity with the productivity of other inputs (for example, capital) used in the production process. Following a feasibility study in the early 1980s, the Multifactor Productivity (MFP) Program was launched in 1987. The first MFP estimate published by Statistics Canada pertained to the 1961-to-1988 period.
Statistics Canada recognized that MFP estimates, unlike labour productivity estimates, require particular assumptions about the production process. In other words, such estimates are more of an analytical product than most other economic statistics. From the outset, the set of assumptions chosen by Statistics Canada for estimating MFP have been made explicit and documented along with the release of MFP measures.
This paper traces the evolution of the MFP Program at Statistics Canada and the improvements in MFP measurement that have been implemented over the past 25 years. Those improvements were in response to developments in the economic literature, the changing nature of the economy, better data sources, and the needs of the user community. Those changes have increased understanding of the process of economic growth in Canada and its relative productivity performance compared with its major trading partner—the United States.
This report also summarizes the research that uses alternate data and methodologies to assess the accuracy of the MFP Program and provide insights into areas that traditional international multifactor productivity programs omit. Finally, possible future directions that are being contemplated to improve the measurement of productivity at Statistics Canada are outlined.
Development of Multifactor Productivity Program at Statistics Canada
Statistics Canada uses the growth accounting framework to measure MFP growth. With this framework, economic growth can be divided into that coming from increases in capital and labour inputs, and the residual from that coming from all other sources—referred to as MFP growth. It can also be used to decompose labour productivity growth into MFP growth plus a term that depends on the growth in capital intensity (capital–labour ratios) as well as changes in the composition (quality) of labour. While the same framework continues to be used to measure MFP growth since the Program began, major improvements have been made to the measurement of capital and labour inputs.
When Statistics Canada first published MFP estimates, a simple summation of assets was used to measure capital stock, and in turn, capital input, and an aggregation of total hours worked was used to measure labour input. In 2002, the MFP measures were re-engineered. The constant quality index of capital and labour input, as proposed by Jorgenson and Griliches (1967), was introduced. The assets used to estimate capital were expanded to include land and inventories in addition to fixed reproducible assets (machinery and equipment and structures). New estimates of depreciation were introduced, which incorporated the rate of decline in asset prices after purchase; previous estimates had been based on expected lengths of lives and arbitrary assumptions about decay functions. These changes brought the Canadian practice in line with U.S. Bureau of Labor Statistics (BLS) estimates, which, in turn, were based on the work of Hulten and Wykoff (1981).
These revisions were consistent with international practice and with the recommendation of the Organisation of Economic Cooperation and Development productivity manual (OECD 2001). They also responded to users’ need for greater comparability between MFP estimates for Canada and the United States.
The comprehensive 2002 revision had two major effects. First, it shifted the components of economic growth toward input growth (both labour and capital) and away from MFP growth.
Second, it yielded more methodologically sound comparisons of MFP growth in Canada and the United States. Previous comparisons that included the effect of changes in labour and capital quality in the Canadian MFP estimates found few differences between Canadian and U.S. MFP performance between 1961 and 1997. The revised estimates, which accounted for the growth that occurred from shifting toward higher-quality inputs, showed that MFP growth in the Canadian business sector was lower than that in the United States.
Understanding productivity growth
The accuracy of the MFP Program has been assessed by examining the sensitivity of the estimates to the use of alternate parameters and by using alternate data and methodologies. First, micro-data have been used to gain a better understanding of the dynamics of productivity growth. Second, the neoclassical assumptions underlying growth accounting have been relaxed in order to examine the roles of scale economies and short-run capacity utilization fluctuations in estimates of productivity growth. Third, the asset coverage has been expanded to include assets not normally included in estimates of investment by the regular programs of statistics agencies—items such as intangibles and infrastructure—in order to improve the understanding of the process of economic growth and of the degree to which conventional measures of MFP may have shortcomings. These studies provide users with guidance on whether or how the standard measures should be used.
Future directions for the multifactor productivity program
Statistics Canada’s Multifactor Productivity Program has several research goals aimed at improving MFP measures and better understanding economic growth.
First, the Program will conduct research to evaluate how alternative measurement methods affect comparisons of productivity growth between Canada and the United States. Results so far show that the relative MFP growth performance in the two countries is robust to alternative approaches. When capital and MFP are estimated by adopting the BLS procedure to adjust the extreme internal rates of return used to calculate the user cost of capital, by adopting top-down instead of the bottom-up approach, and by using alternative methods for estimating user cost of capital, the results on relative Canada–United States MFP growth are found to be similar to those based on the official estimates in the two statistical agencies.
Over the last 50 years, labour productivity in Canada and the United States grew at about the same rate. Until the early 1980s, Canadian growth exceeded that of the United States. Since then, U.S. labour productivity growth exceeded Canadian growth, a reflection of Canada’s lower MFP growth (almost none, compared with about 1% annually in the United States).
Second, the Program will experiment with MFP estimates that include the natural resource capital in the mining sector. It has long been recommended that capital measures include natural resource capital.
Third, the Program will improve the consistency of the industry productivity KLEMS database, as part of the System of National Accounts historical revision. Greater consistency of output, inputs and productivity at the industry level will also improve MFP estimates for the aggregate business sector. The Productivity Accounts will also continue to examine how estimates can be produced for the non-business sector, particularly in health. Finally, the new National Accounts longitudinal micro-data file on firms will be used to better understand the determinants of MFP growth at the firm level as opposed to the industry level.
- Date modified: