The Canadian Productivity Review

Volume 2007
Number 7

Abstract

Introduction

Provincial productivity growth,
1997 to 2005

Provincial labour productivity
1997 and 2005

Conclusion

References

Figures

More information

PDF version

Other issues
in this series

About this
series

Provincial labour productivity, 1997 and 2005

Differences in levels of provincial labour productivity stem from many factors, such as differences in industrial structure, urbanization, prices and technology use.¹ Figure 2 reports provincial labour productivity in 1997 and 2005, presented in the familiar west-to-east configuration. These data are reported in basic units of productivity — the amount of constant dollar GDP (gross domestic product) produced per hour worked. The dashed lines denote the national productivity averages in 1997 and 2005 respectively.

Nationally, real GDP per hour worked increased from $35.6 in 1997 to $40.5 in 2005 (both figures are measured in 1997 prices). Labour productivity was highest in Alberta and Ontario, two provinces that exceeded the Canadian average in both years. Productivity in British Columbia mirrored the national average in 1997 and, to a lesser extent, in 2005. Real GDP per hour worked in Quebec stood just below the Canadian average in 1997, and, by 2005, had declined slightly relative to the national level. In 1997, Saskatchewan's labour productivity was lower than Quebec's, but stronger productivity growth in Saskatchewan brought its real GDP per hour worked into line with Quebec's by 2005. Labour productivity in Manitoba and in the three Maritime provinces remained well back of labour productivity in the other provinces in 2005.

Not surprisingly in light of its high growth rate, the most dramatic shift in relative labour productivity during this nine-year period occurred in Newfoundland and Labrador. In 1997, its GDP per hour worked stood below Saskatchewan's, and well below Quebec's and British Columbia's. By 2005, Newfoundland and Labrador produced more real GDP per hour worked than any of these provinces.

In Figure 3, we examine shifts in relative labour productivity more directly by indexing provincial levels to the national average in each year. Provinces are sorted leftward to rightward by their relative labour productivity in 1997, ranging from Prince Edward Island (72% of the national average) to Alberta (108% of the national average). The left and right bars in each pairing denote relative labour productivity in 1997 and 2005 respectively. Figure 3 also plots the average annual rate of growth in labour productivity for each province from 1997 to 2005 (which we connect from province to province to highlight the size of these growth differentials). The left axis is relative labour productivity (Canada = 1), while the right axis is the average annual growth rate.

Alberta, Ontario, British Columbia and Quebec — the four provinces with the highest levels of real GDP per hour worked in 1997 — saw no improvement in their relative productivity from 1997 to 2005. In Alberta, real GDP per hour worked declined from 108% of the national average in 1997 to 104% in 2005 (when measured in 1997 prices). Relative labour productivity in Quebec and British Columbia also fell slightly over this period, to 95% and 98% of the 2005 Canadian average respectively.² In Ontario, relative labour productivity was virtually identical in 1997 and 2005. At 4% above the national average, Ontario's real GDP per hour worked was, in 2005, the same as Alberta's.

Stronger average productivity growth in Saskatchewan led to an improvement in its relative position. Saskatchewan's real GDP per hour worked increased from 92% of the Canadian average in 1997 to 95% in 2005.

Relative labour productivity in New Brunswick and Nova Scotia remained stable from 1997 to 2005, as growth rates in these provinces mirrored the national rate. In 2005, real GDP per hour worked in New Brunswick and Nova Scotia stood at 83% and 81% of the national average, respectively. Slower average productivity growth in Prince Edward Island saw relative productivity decline slightly, to 70% of the national average.

These relative estimates help bring the large productivity gains in Newfoundland and Labrador into clearer view. At 89% of the Canadian average in 1997, labour productivity in Newfoundland and Labrador was already substantially higher than in the rest of Atlantic Canada. By 2005, this gap had widened substantially, as labour productivity in Newfoundland and Labrador equalled the national average.

The case of Newfoundland and Labrador represents the only substantial shift in the implicit productivity rankings reported in Figure 3. In 1997, Newfoundland and Labrador's real GDP per hour worked was slightly lower than Saskatchewan's, and substantially lower than Quebec's. By 2005, Newfoundland and Labrador stood well ahead of Saskatchewan and Quebec in terms of real GDP per hour worked, behind only Ontario and Alberta.

To this point, we have focused strictly on the changes in labour productivity that are associated with underlying changes in real, or constant dollar, GDP. We have done so in order to remove the impact of relative price changes from our provincial comparisons. Below we examine how level estimates for 2005 differ when they are based on the actual price conditions that existed in that year.

Figure 4 converts the labour productivity differentials reported in Figure 3 into percentage deviations from the national average in 1997 and 2005. Three estimates are reported for each province. The first two bars depict deviations in GDP per hour worked from the national average in 1997 and 2005, using 1997 dollars in each case. The third bar re-estimates the deviation in 2005 based on 2005 prices (i.e., substituting nominal dollar GDP for constant dollar GDP). Differences between the second and third bars occur because relative nominal prices in 2005 are not the same as relative real prices in 2005 (which are measured in 1997 constant dollars). As in Figure 3, provinces are ranked leftward to rightward (lowest to highest), based on their labour productivity in 1997.

For two resource-rich provinces — Alberta and Newfoundland and Labrador — the impact of relative price changes is very substantial. As noted earlier, Alberta's relative labour productivity actually declined from 1997 to 2005 when its output is measured in constant (1997) dollars — from 8% to 4% above the national average. But the price of its natural resources (primarily oil and gas) has increased relative to the commodities produced in other provinces, leading to a dramatic increase in relative performance when viewed in nominal terms. In 2005, Alberta's nominal dollar GDP per hour worked was 35% higher than the national average.

Improvements in the relative position of Newfoundland and Labrador, already well apparent in real terms, are far more dramatic when GDP per hour worked is evaluated in nominal terms. When changing relative prices are taken into account, Newfoundland and Labrador's GDP per hour worked increased to 19% above the national average.

The results for Saskatchewan, Ontario and Quebec also warrant emphasis, as changing price conditions in these provinces affect perceptions of relative performance. Saskatchewan's relative productivity increased in real terms, from 8% below the national average in 1997 to 5% below in 2005 (when measured in 1997 dollars). But when the latter estimate is recalculated using 2005 prices, Saskatchewan's GDP per hour worked actually exceeds the national average by a slight margin (1.2%).

These nominal dollar adjustments have the opposite effect in Ontario and Quebec. In real terms, Ontario's relative labour productivity remained virtually unchanged in 1997 and 2005, at around 5% above the national average. But when nominal prices are used, Ontario's GDP per hour worked in 2005 is 2% below the Canadian average. In Quebec, the decline in relative labour productivity that is apparent in real terms is accentuated by the shift to nominal prices. Quebec's GDP per hour worked in 2005 was 5% below the national average when measured in 1997 prices — this falls to 8% when 2005 prices are used.

1. Several Statistics Canada research papers have examined the factors that underlie provincial differences in labour productivity and GDP per capita. We summarize several of these below.
   
   Baldwin et al. (2001) used decomposition methods to disentangle the extent to which productivity differences across provinces are due to (a) industry-mix effects or (b) real productivity effects (the differences that persist after controlling for the impact of industry structure). The authors found that, on balance, real productivity effects account for more of the variation in productivity levels across provinces than do industry-mix effects.
   
   Other papers have examined the role of labour productivity in explaining provincial differences in GDP per capita. Baldwin, Brown and Maynard (2005) evaluated the extent to which provincial differences in GDP per capita are related to differences in labour productivity or work intensity. The results show that these two factors reinforce one another-provinces with lower (higher) GDP per capita tend to have both lower (higher) labour productivity and lower (higher) work intensity. In an earlier study, Baldwin et al. (2004) found that differences in GDP per capita were predominantly associated with differences in labour productivity, although "labour market conditions…also prove[d] to be an influential factor" (p. 21). This analysis also differentiated (as does this paper) between real effects and relative price effects. In a recent study, Beckstead and Brown (2005) show that provincial differences in GDP per capita also depend, in part, on basic differences in urban–rural composition.



2. It is worth noting that, for most provinces, the absolute magnitude of these shifts in relative labour productivity is qualitatively small. Small shifts, such as the slight declines in relative productivity apparent for British Columbia, Quebec and Prince Edward Island, should be interpreted with some measure of caution, as all point estimates of productivity are subject to statistical error. For background on the precision of productivity measures, see Baldwin and Harchaoui (2001). For a more general discussion of measurement issues, developed within the context of international comparisons, see Baldwin et al. (2005).