2 Canada-U.S. Productivity Comparisons

View the most recent version.

The traditional notion that Canadian industry has long been inefficient can be found both in the work that emphasizes the link between the tariff and the growth of manufacturing and in various Canada-U.S. productivity comparisons. A number of studies have compared relative productivity for the post World War II period and concluded that a substantial gap existed. Most of these studies have relied on partial labour productivity measures to infer inefficiency. Fullerton and Hampson (1957) measured relative output per worker in secondary manufacturing between Canada and the United States for 1953. Standardizing labour input by hours worked, and adjusting for price differences, they found that Canadian productivity was only 61% of the American average. For primary industries, the gap in favour of the United States was only 10%.

In a study for the Economic Council of Canada, E.C. West (1971) working with a sample of 30 industries for 1963 found that, after correcting for input and output price differences, Canada's value-added per worker was between 72% and 77% that of the U.S., depending on whether Canadian or American outputs were used to construct the price deflator. Working with a sample of 33 industries, and adjusting for output and material prices, Frank (1977) estimated the gap in value-added per worker to be 65% in 1967 and 80% of the American levels in 1974. Caves et al. (1980) using a broader sample of 84 matched Canadian/American industries, and assuming that Canadian outputs and inputs were priced up to the tariff level when making price corrections, estimated Canadian value-added per worker to be 63% of the American level in 1967.

In contrast to the post 1945 period, there are relatively few studies of productivity in the Canadian manufacturing sector in earlier periods (Maddison, 1952; Maddison, 1953; Sutton, 1953; Dales, 1966). Existing studies suggest that productivity in Canada was substantially below that of the United States in this period. The Economic Council of Canada (1965, p. 54) calculated that real GNP per capita in Canada was between 67% and 77% of the American GNP per capita between 1920 to 1960. Sutton (1953, p. 197) estimated Canadian output per employed person as only 83% that of the United States in the non-agricultural sector and 74% overall for the period 1929-1933. While these studies do not focus directly on manufacturing per se, the disadvantages for Canada as a whole have no doubt influenced the view that the Canadian economy and the manufacturing sector have long been relatively inefficient compared to their American counterparts.⁶

In what is one of the better known Canada/U.S. studies for the pre-1945 period, Dales (1966) argues that the Canadian tariff led to the development of a large portion of the manufacturing sector—secondary manufacturing—that suffered a comparative disadvantage relative to U.S. standards.⁷ Our interest is with this index of relative real value-added per employee. Dale's estimates indicate that average productivity for secondary manufacturing in Canada varied between 75% and 85% of that for total manufacturing in the United States between 1926 and 1939. His figures were not corrected for relative prices. These results have been widely interpreted as indicating that Canadian secondary manufacturing was less efficient than the United States both, in the interwar, as well as in the post 1945 period.

Most of these historical studies focus on one or other variant of a partial labour productivity measure.⁸ Partial productivity measures do not take into consideration other inputs (capital, fuel, materials), nor the type of substitution that is possible among inputs. Total factor productivity measures offer potentially superior estimates of relative efficiency, but at a cost—more factors must be measured, amongst which relative capital estimates provide some of the greatest measurement problems. For purposes of comparing productivity over time, partial measures may be adequate—if factor intensities, the quality of factors and plant size relative to minimum efficient scale all remain relatively unchanged. But partial factor productivity measures may be less adequate when analyzing cross-countries differences. When factor intensities or plant sizes differ as they do across countries, they will be quite misleading when it comes to understanding differences in technical efficiency.

The measurement problems associated with moving from partial to total factor productivity statistics have meant that even the more recent studies of Canada/U.S. productivity (Caves et al., 1980; Saunders, 1980; Bernhardt, 1981) have tended to concentrate on partial measures. Even where rough estimates of capital are provided, the Canada/U.S. total factor productivity estimates invariably receive less emphasis, usually as a secondary reference (West, 1971; Frank, 1977). More recently, better estimates of relative capital intensity have permitted more accurate estimates of relative Canada/U.S. differences in TFP levels (Baldwin, Gu and Yan, 2008).

This study is aimed at overcoming the deficiency of partial productivity measures previously used to characterize pre World War II efficiency in manufacturing by developing a total factor productivity measure for a large number (137) of matched Canadian/U.S. industries in 1929. By choosing to calculate total factor productivity measures for individual industries, we avoid the potential aggregation bias that is associated with using total economy aggregates that apply to the manufacturing sector as a whole.⁹

 

^6. See West (1971, p. 1) for a statement that Walters (1968) concluded lower Canadian income levels were the result of generally lower efficiency in Canada.

^7. Dales focused on allocative rather than technical efficiency in that he used ratios of Canada/U.S. labour productivity relative to relative wage rates to infer inefficiency.

^8. An exception is the study by Maddison (1953); but it concentrates on Canada/U.K. relative productivity in 1948 and does not therefore focus on the key differences between Canada and the United States which provide the focus of this study.

^9. For a brief description of the bias in performing TFP analyses at the level of the aggregate economy, see Hulten (2001).