RESOURCE INDUSTRIES AND ECONOMIC PERFORMANCE: A BRIEF LITERATURE REVIEW

Given the substantive role played by natural resource extraction and processing activities in the Canadian economy, at least prior to the twentieth century, it is not surprising that questions about the connection between resource endowments and economic performance have motivated a particularly rich vein of research within Canadian historiography. Indeed, Canadian economic historians have been largely responsible for both the formal identification of the channels through which resource-intensive production may foster industrial expansion and diversification, and some of the more compelling expositions of the collapse of these channels in modern, industrial economic environments.

W. A. Mackintosh, Harold Innis, A. R. M. Lower, Melvin Watkins, Gordon Bertram, and Douglass North, among others, all made seminal contributions to the formulation of a model of economic growth that describes how a resource rich, but labor- and capital-poor economy may industrialize around a resource base.⁴

The “staples thesis” describes how export demand motivates the development of resource extraction and processing industries, which make both direct and indirect contributions to the intensive and extensive growth and diversification of an economy. Rather than focusing on the trigger for resource exploitation, I confine my investigation to the study of the impact that this exploitation has on an economy that has already become large, urban, and industrial. The direct contributions described by the thesis—increased employment, investment and output, productivity improvements, and resource rents—are all expected to expand the absolute size of the economy. The indirect contributions—forward, backward, and final demand linkages—are expected to promote diversification into more human- capital- and physical-capital-intensive industries by reducing domestic resource-intensive intermediate input prices, and increasing the demand for both service sector output and non-resource-intensive manufactured products.

The staples thesis continues to be widely accepted as a tool capable of describing economic development among nations that have yet to experience rapid, modern, industrial growth.⁵

American historians, for example, have long argued that the exploitation of high quality agricultural land and nonrenewable mineral resources triggered domestically unique technological change and industrial expansion between 1850–1900. H. J. Habakkuk, Robert Allen, John James and Jonathan Skinner, Louis Cain and Donald Paterson, and Gavin Wright all emphasized the great advantages that U.S. manufacturers enjoyed during the late nineteenth century due to their access to abundant, high quality domestic resources.⁶ However, almost since its inception the staples thesis has faced criticism by those who questioned the continued relevance of resource industries in economies, including the twentieth-century Canadian economy, that were diversified, urban, and wealthy. Vernon Fowke argued that as early as 1867 Canadian growth was based on agricultural production, not more narrowly defined energy, fish, forest, or mineral exploitation.⁷ W. T. Easterbrook claimed that an emphasis on resource extraction failed to capture the, “… complexities of modern industrial change.”⁸ Kenneth Buckley rejected the efficacy of the staples thesis as even a descriptive tool in the Canadian context for any era after the decline of the fur trade in the early nineteenth century.⁹ As a final example, although generally a supporter of a resource-centric approach to understanding Canadian development, Hugh Aitken discounted the importance of resource-intensive producers and export products after 1914.¹⁰

The idea that resource industries may be irrelevant in rich, industrial economies is, of course, not exclusively Canadian, nor confined to the writings of economic historians. Robert Gordon, for example, claims that the United States experienced more rapid growth than its European competitors up to 1913 due to resource advantages, but thereafter the gap between the United States and Europe continued to widen due to “technological advantages” alone.¹¹

Gordon and William Nordhaus argue that the early 1970s productivity slow down in the United States was not only highly concentrated in industries that used natural resources, particularly energy, intensively, but the productivity gains in the late 1990s appear to have been driven exclusively by human capital and technology intensive “new economy” industries.¹² Nordhaus reports that resource-intensive industries, such as energy extraction, fishing, forestry, mining, and agriculture have become serious laggards, with some of the lowest post-1990 labor productivity growth rates in the entire U.S. economy.¹³

Although their work is based on a very different methodological and theoretical approach, Jeffrey Sachs and Andrew Warner, Sachs and Francisco Rodriguez, and Richard Auty have also come to the conclusion that, at least in our contemporary economic environment, the concentration of labor and capital in resource-intensive production may be a serious constraint on economic performance.¹⁴

They have argued that the employment of inputs in resource extraction and processing activities reduces the labor and capital available for other production processes that contribute more to long-run growth and development. Sachs and Warner have famously called this the “curse of natural resources.”¹⁵

What does this very brief review of the resource-based growth literature imply about the importance of resource extraction and processing industries to the Canadian economy during the twentieth century? In one form or another much of the more recent work suggests that these industries have not made substantive direct or indirect economic contributions since the earliest years of the century, they comprise a low productivity sector that has retarded growth, and they long ago abandoned their role as a leading sector in the economy. Despite this rather pessimistic interpretation of the role played by resource exploitation, the empirical evidence presented in this article illustrates the need to exercise some caution before dismissing energy, fishing, forestry, and mining industries as insignificant in diversified, urban, and industrial nations.

DATA COMPILATION, CONSTRUCTION, AND COMPOSITION

By the end of the twentieth century information about Canadian natural resource industries was categorized and organized in a complete and transparent manner using well-documented systems such as the Standardized Industrial Classification System, or most recently the North American Industrial Classification System. However, prior to 1948 Canadian data were published under census categories that were originally established in the late nineteenth century on the basis of the characteristics of establishments' primary output products. In an effort to accommodate changes in the composition of the industry-specific series used in this article, I began with the categorization of four natural resource sectors—energy, fishing, forestry, and mining—currently employed by Natural Resources Canada and the Canadian Department of Fisheries and Oceans, and I then reconstructed these sectoral categories back through the twentieth century using the most disaggregate industry specific data available. Where this reconstruction was impossible, I dropped the industry or series from the study. Where there have been concerns about the consistency of the industrial composition, I have attempted to collect as much overlapping data as possible to facilitate the construction of an adjustment factor to account for potential changes in composition.

Given the attention that Canadian economic historians have lavished on domestic agricultural production, particularly western wheat production between 1896 and 1914, and given the intensity of land use in the production of agricultural products, it may seem odd that agriculture has not been included as a resource sector in this study. Because the original articulation of the staples thesis clearly did not consider agriculture a staple resource, because agriculture's role as a staple largely pre-dates the twentieth century, and because of the dramatic differences in the production processes involved in agriculture relative to energy, fishing, forestry, and mining—namely the lower capital intensity, the more dispersed production locations, the differences in ownership characteristics, the differences in export intensity, and the absence of industrial linkages—it does not seem unreasonable to leave the assessment of agriculture's role in twentieth-century Canadian development for a subsequent study. Other exclusions that may be noteworthy from our contemporary perspective include pipelines, electricity generation, and electricity distribution. Although there are good theoretical reasons for excluding these industries—namely their role as a substitute for transportation services in the case of pipelines, and their public rather than private ownership and operation in the case of electricity—the absence of consistent data for the first half of the century makes their inclusion undesirable for more practical reasons as well.

Virtually all natural resource industries have been, at least to some extent, vertically integrated. This implies that almost every industry has been involved in the extraction of resources from the physical environment, the primary processing of resources into intermediate inputs, and the secondary processing that prepares resources for final consumption. By again adopting the categorization currently employed by Natural Resources Canada and the Canadian Department of Fisheries and Oceans, I have further disaggregated the industry groups based on their dominant role in the transformation of in situ natural resources into intermediate inputs or goods intended for final consumption. For each industry group, therefore, I have identified three subgroups: extraction, primary processing, and secondary processing.

For the aggregate Canadian economy, the agricultural sector, the manufacturing sector, the service sector (and various less aggregated components of these sectors), and for ten resource-sector, stage-of-production aggregates (there are no secondary production industries in the energy or fishing sectors), I have constructed an estimate of the number of employees, the fixed capital stock, value added, and the share of value added paid to labor and capital for every year between 1900 and 1999.¹⁶

In general, for the years after 1921 these data were drawn from Dominion Bureau of Statistics and Statistics Canada publications. For the years prior to 1921 the data had to be reconstructed from a combination of ministry reports, Sessional Papers, the Canada Year Book, and various census reports. In most cases adequate data were available for all industries during the census years: 1901, 1911, and 1921. For many of the primary and secondary processing industries data were also available for the years in which there was a supplementary census of manufacturing: 1905 and 1915. In virtually every case interpolation across the missing years followed the methodological approach employed by Malcolm Urquhart in his reconstruction of annual aggregate GNP figures between 1870 and 1926.¹⁷ To generate some of the results discussed in this article these basic data were augmented with additional information on sector specific output prices, interest rates, GNP deflators, and capital goods prices.

DIRECT AND INDIRECT ECONOMIC CONTRIBUTIONS

When looking to identify both direct and indirect economic contributions made by resource-intensive producers, the staples thesis—as modeled by Watkins and North—provides some guidance.¹⁸

In the most basic terms the thesis suggests that if a resource industry is directly contributing to economic growth and development, then we should expect it to be responsible for a significant proportion of domestic economic activity. Economic activity can include not only the value of total income (gross national product), but also the employment of labor and capital, and the efficiency with which inputs are converted into output (total factor productivity). If a resource industry is indirectly contributing to economic performance, then we should expect it to be a leading sector, with changes in resource-intensive production chronologically preceding changes in other economic indicators. Watkins and North have provided us with an even more specific description of the channels through which these indirect contributions operate—increases in resource-intensive output should “trickle down” to the rest of the economy through reductions in resource-intensive domestic intermediate input prices, increases in non-resource-intensive manufacturing output, and increases in service sector output.

For each year between 1900 and 1999 I have calculated the proportion of Canadian GNP, aggregate employment, and the aggregate fixed capital stock that can be attributed to those industries engaged in the extraction, primary processing, and secondary processing of Canada's energy, fish, forest, and mineral resource endowments. I have compared these direct contributions to those made by non-resource-intensive manufacturing industries, agricultural producers, and service sector industries. I have also calculated the annual rate of change in TFP for each year between 1900 and 1999 for all four resource sectors at each stage of production, and I have compared the level of TFP generated by these producers to the level of TFP generated by non-resource-intensive manufacturers, agricultural producers, service sector industries, and the aggregate economy. To identify indirect contributions I have performed standard Granger causality tests to check for a statistical relationship between past changes in resource industries' output levels, and current changes in domestic resource-intensive intermediate input prices, non-resource-intensive manufacturing output, and service sector output.

Direct Contributions: Output, Employment, and Capital

In Figure 1 the share of GNP generated by resource industries, non-resource-intensive manufacturing industries, agriculture, and service industries have been plotted annually over the twentieth century. We can see that the contribution to GNP made by agriculture fell steadily over the century, with the only lull in the rate of decline spanning the period between the mid-1930s and the early 1950s. Non-resource-intensive manufacturing industries increased their contribution to GNP over the first half of the century, but their share fell all the way back to its 1900 level by the end of the century. The service sector has not only been much larger than the other sectors in the economy during the twentieth century, it steadily increased its contribution to GNP, particularly over the post–World War II period. The output contributions made by agriculture, non-resource-intensive manufacturing, and services have been included in Figure 1 for comparative purposes. In particular I wish to emphasize that relative to the other goods producing sectors, the resource industries consistently generated a large share of aggregate GNP throughout the twentieth century. This persistence in the resource industries' large output shares suggests continued, if not necessarily increasing, relevance in the post-1900 Canadian economic environment.

The greatest contribution resource industries made to domestic output over the twentieth century came in 1900, when 22.3 percent of Canada's GNP was generated by energy, fishing, forestry, and mining industries. The smallest contribution to domestic output came in 1991, when only 10.2 percent of Canada's GNP was generated by the resource producers. On average between 1900 and 1999 energy, fishing, forestry, and mining industries generated 16.7 percent of domestic GNP, but their output contribution fell by a (statistically significant) average of 0.25 percent per year between 1900 and 1999. By comparison, between 1900 and 1999 non-resource-intensive manufacturers' average contribution to GNP was just 10.8 percent, agriculture averaged only 9.6 percent, and the service sector contributed the remaining 62.9 percent.

If we compare the resource industries' annual output shares against the estimated linear time trend depicted in Figure 2, it appears that there were extended periods during which the series lies above, and subsequently below the long-run trend. This pattern in the deviations from the trend suggests cyclical movements in the resource industries' GNP contributions. To isolate the cycles that appear to be present in the data I have used a Hodrick-Prescott filter to decompose the series into its stationary and nonstationary components. The nonstationary, or smoothed, component of the resource industries' output shares (labeled “HP Filter” in Figure 2) illustrates these longer-run cyclical movements.¹⁹

We can see that prior to World War I the share of Canadian GNP generated by resource-intensive production fell quite sharply. This downturn was followed by relative stability during the interwar period, strong growth from the mid-1930s until the mid-1950s, and then a long period of decline through much of the rest of the century that was interrupted by two sharp expansionary phases during the 1970s and 1990s.

Of course, when we consider all resource industries as a group many of the more interesting chronological patterns are obscured as a result of aggregation. To hint at the rich stories that could be told about sectoral, stage of production, and (by implication) regional growth and contraction, in Table 1 I report the average annual percentage change in the proportion of GNP generated by natural resource industries disaggregated by sector and stage of production for the periods of cyclical contraction (1900–1913, 1954–1970 and 1980–1993), stability (1914–1931), and expansion (1932–1953 and 1994–1999). For comparative purposes I also include the change in the output shares for each of these periods for non-resource-intensive manufacturing, agriculture, and service industries.

The earliest period of decline in the natural resource industries' contribution to aggregate output (1900–1913) appears to have been concentrated among the extraction industries, particularly in forestry and mining.²⁰

The subsequent return to relative stability (1914–1931) can be traced to an improvement in the performance of extraction industries in forestry and mining, as well as particularly strong growth among primary processors in forestry.²¹ The expansion we observe in the aggregate series between 1932 and 1953 appears to be related not to extraction, but to primary, and to an even greater extent, secondary processing activities, although the forestry and mining industries still dominate.²² During the period from the mid-1950s until the late 1960s (1954–1970) energy industries were increasing their output shares in a sharply counter-cyclical pattern relative to fishing, forestry, and mining industries' shares, which were falling steadily over this period. Energy, particularly energy extraction, comes to dominate the “all sectors-all stages” series during its expansionary phase during the 1970s (1971–1979), and its equally dramatic contraction during the 1980s (1980–1993).²³ Interestingly, the final expansionary phase (1994–1999) was driven not by energy, but by strong growth in secondary processing among the forestry and mining industries.²⁴

Income generation is not the only direct contribution that resource industries made to the twentieth-century Canadian economy. In Figure 3 we can see that the energy, fishing, forestry, and mining industries' contribution to aggregate employment peaked in 1900 at 17.6 percent, and reached a minimum in 1995, when only 5.8 percent of Canada's workers were employed in resource industries. On average between 1900 and 1999 resource industries employed 11.3 percent of the total Canadian workforce, and like their contribution to GNP, their employment share fell by a (statistically significant) average of 0.55 percent per year over the entire period. For comparison note that between 1900 and 1999 non-resource-intensive manufacturers employed 11.3 percent of the total Canadian workforce, while agriculture employed 20.4 percent, and the service industries employed the remaining 57.6 percent, on average. Although they are broadly similar, the cyclical patterns in the resource industries' employment shares appear both more dramatic and more persistent than those observed for the output shares, with a contractionary phase from the start of the century until the mid-1930s, expansion from the mid-1930s until the early 1950s, and a long decline through the last half of the century that slowed substantially during the 1990s (See Figure 3).

During the twentieth century Canadian resource industries were considerably more capital intensive than their manufacturing or service sector counterparts, and only slightly less capital intensive than the agricultural sector. The relative capital intensity of resource production actually increased between 1900 and 1999, not only because these industries were producing a smaller share of total output, but because they were increasing their share of the aggregate capital stock. From Figure 4 we can see that in 1900 20.3 percent of Canada's fixed capital was housed in the resource sectors. Their smallest contribution to the aggregate capital stock was in 1921 at just 12.2 percent, and their largest contribution came in 1979 at 26.6 percent. On average between 1900 and 1999 the resource industries housed 18.1 percent of the total fixed capital in Canada, and this share rose by an average of 0.06 percent per year over the entire period. For fixed capital we can estimate a statistically significant positive linear time trend through the data series. For comparison note that between 1900 and 1999 non-resource-intensive manufacturers housed 4.8 percent of Canada's aggregate capital stock, agricultural producers housed 15.6 percent, and the service industries housed the remaining 61.5 percent, on average. Similar to the chronological patterns we observe among the resource industries' output and employment shares, we can again see contraction in their contribution to the aggregate fixed capital stock during the first quarter of the twentieth century, expansion during the second quarter and into the third quarter, followed by another sharp contraction over much of the last half of the century, interrupted by a brief period of growth (driven by an investment boom among energy extraction industries) from the mid-1960s until the mid-1970s.

Consideration of Table 1 and Figures 1–4 suggests that Canada's natural resource industries were responsible for substantial shares of domestic output, employment, and investment, but these shares were not chronologically nor cross-sectionally uniform. After 1930 Canada's energy, fishing, forestry, and mining industries accounted for a greater proportion of GNP, aggregate employment, and the aggregate fixed capital stock than all other goods producing sectors. In terms of the absolute size of their direct contributions, these industries appear to have been playing a substantial role in the economy throughout the entire twentieth century.

Direct Contributions: Productivity Performance

An alternate approach to the assessment of the resource industries' direct contributions to Canada's wealthy, industrialized, diversified, and urbanized twentieth-century economy is to consider these contributions in terms of productivity performance, rather than absolute size. Proponents of the resource curse and those who have emphasized the importance of new economy industries have suggested that labor and capital employed in resource-intensive industries could contribute more to the size and development of an economy if they were redistributed into other sectors that are more conducive to growth. It seems, therefore, that any questions about the importance of resource industries' direct contributions may not necessarily be about the number of workers or amount of capital they employ, but rather the amount of output produced by these workers and machines relative to what they could have produced in other sectors in the economy.

To measure the resource industries' relative efficiency I have used a Tornqvist total factor productivity index to compare TFP levels across sectors and across time. The Tornqvist index relies on the assumption that production processes may be described by a translog production function—a superlative and flexible functional form, as defined by W. Erwin Diewert.

²⁵

In effect, by using the Tornqvist index I am measuring TFP as a weighted geometric average of relative partial factor productivity ratios, in which the numerators and denominators of the partial factor productivity ratios may represent two sectors at a point in time, or a single sector in two different years. The Tornqvist index takes the form

where: A equals Hicks neutral total factor productivity parameter; VA equals value added; L equals employment; K equals fixed capital stock; sL = (wL · L) / VA; sK = 1 − sL; t equals 1900–1999; and i, j equals energy, fishing, forestry, mining, non-resource-intensive manufacturing, agriculture, services, or the aggregate economy.

Figure 5 illustrates the Tornqvist TFP index for the aggregate economy, the natural resource industries, the non-resource-intensive manufacturing industries, the agricultural sector, and the service sector for the years 1900–1999, with the aggregate economy in 1900 representing the base year and sector. This implies that the aggregate economy's index takes the value 1.00 in 1900, and all other sectors' TFP has been measured relative to this base value in each year. We can see that Canada's resource industries experienced fairly rapid TFP growth through the twentieth century, peaking at an average annual rate of 6.78 percent during the 1950s, and maintaining a long-run average growth rate of 0.68 percent per year over the full 1900–1999 period.²⁶

Among all of the sectors considered in this article, non-resource-intensive manufacturing enjoyed the highest average rate of TFP growth during the twentieth century: 1.27 percent per year. TFP within the agricultural sector grew just slightly faster than the resource industries' TFP: 0.71 percent per year between 1900–1999. The service sector experienced the slowest average TFP growth, at just 0.38 percent per year. As a whole, the Canadian economy increased the efficiency with which it converted inputs into output by an average of 0.65 percent per year over the century. Like the resource industries, the aggregate economy, non-resource-intensive manufacturing, and agriculture all experienced their most rapid TFP growth rates during the 1950s. Services' TFP growth rate peaked in the 1940s.

Because the aggregate economy in 1900 has been used as the base year and sector in Figure 5, the series depicted also reveal information about the relative productivity performance of each sector. More specifically, the position of each sector's series in Figure 5 indicates the relative level of that sector's TFP. We can see that the resource industries enjoyed TFP levels that were consistently in excess of the aggregate economy, the service sector and agriculture, whereas their ability to generate output from a given combination of inputs was approximately equal to that of non-resource-intensive manufacturing. On average over the years 1900–1999 Canadian natural resource industries were 48.9 percent more efficient than the aggregate economy, 150.0 percent more efficient than agricultural producers, 40.2 percent more efficient than service sector industries, and 3.9 percent more efficient than non-resource-intensive manufacturers.

Given these TFP figures it is difficult to be very supportive of any characterization of twentieth-century Canadian resource producers that implicitly or explicitly suggests that they comprised a low productivity sector. It appears that even in the wealthy, diversified, and industrialized twentieth-century Canadian economy resource-intensive producers were not only responsible for maintaining large employment, fixed capital, and output shares, but they were also making a significant positive contribution to the level of aggregate productivity, while maintaining TFP growth rates that were slightly higher than the aggregate economy. These direct contributions were not the only channels through which resource industries were having an impact on the economy. We may now turn to the third piece of empirical evidence that highlights the persistent and substantive economic role played by these industries.

ECONOMIC PERFORMANCE IN THE ABSENCE OF RESOURCE- INTENSIVE PRODUCTION: A COUNTERFACTUAL EXERCISE

To assess the importance of the resource industries' direct and indirect economic contributions I have performed a counterfactual exercise in which I estimate the size and rate of growth of the Canadian economy between 1900 and 1999 in the absence of these contributions.³²

This exercise has three steps. First, I derive an approximation of the importance of the resource industries' direct contributions to the Canadian economy by estimating the total value of resource rents paid to capital owners, workers, and government for every year between 1900 and 1999. Second, I derive an approximation of the importance of the indirect contributions made by the resource producers by estimating the reduction in non-resource-intensive manufacturing output and service sector output resulting from both the removal of domestic resource extraction and processing activities, and the subsequent increase in the domestic component of resource-intensive intermediate input prices. In the third and final step I simply aggregate the value of the direct and indirect contributions and report the total impact as a proportion of real Canadian GNP per capita.

To construct this estimate of the total impact that a counterfactual removal of all domestic resource-intensive production would have had on the Canadian economy I must make a series of standard (although admittedly implausible) assumptions. Wherever possible I have tried to chose assumptions that bias my estimate of the total impact of resource-intensive production on domestic GNP downwards. To determine the value of the direct contributions I follow the work of Robert Fogel, Edward Chambers and Donald Gordon, and D. McCloskey—all of whom performed very similar counterfactual exercises in which they estimated the economic importance of some sector of an economy that employed a fixed factor of production.³³

If I adopt all the standard neo-classical economic assumptions, particularly those associated with perfect labor and capital mobility between sectors of the economy and perfectly elastic labor and capital demand, then eliminating resource-intensive production simply releases labor and capital into other sectors, with no change in average wages or rental rates.³⁴ Under this scenario the only direct loss to the economy from the counterfactual removal of all energy, fishing, forestry, and mining industries would be the lost return to the fixed factor: resource rents.

Following Cherie Metcalf, resource rents may be calculated by adding government royalties and resource extraction fees to observed value added, less a counterfactual opportunity cost for labor and capital.

³⁵

This government revenue plus residual value added represents the return to resource extraction and processing that remains after paying the mobile factors of production the value of their marginal products. It is, therefore, the return to the fixed factor of production—natural resources—and it may be paid to workers through wages above the value of their marginal products, capital through rental rates above the value of its marginal product, or to governments as indirect taxes (royalties), or fees for “permits, licenses and privileges.”

Where: R equals total resource rent; VA equals value added from resource-intensive production; T equals government revenues (taxes) from natural resource extraction and processing activities; WL* equals opportunity cost of L; WK* equals opportunity cost of K; WL equals observed payment to L; WK equals observed payment to K; L equals the number of workers employed in resource-intensive production; K equals the current nominal value of fixed capital employed in resource-intensive production.

³⁶

For labor, capital owners, and government, Table 3 reports my estimate of Canadian resource rents from energy, fishing, forestry, and mining industries as a proportion of GNP, averaged over three sub-periods (1900–1939, 1940–1970, 1971–1999) and over the century as a whole. We can see that although the resource rents captured by workers employed by resource-intensive producers were substantial, accounting for 22.6 percent of total resource rents and 2.3 percent of aggregate GNP on average, they were falling over each subperiod. Capital owners captured an even larger share of total rents: an average of 68.6 percent, representing 7.0 percent of aggregate GNP; but their shares also fell between 1900 and 1999, particularly over the last 30 years of the century. Between 1900 and 1999 Canadian federal and provincial governments captured only 8.8 percent of the total resource rents on average, which amounted to 0.9 percent of aggregate GNP. However, the interesting feature of the governments' rent shares is their dramatic increase beginning in the early 1970s: in 1971 Canadian governments received 8.4 percent of total domestic resource rents; by 1982 this share had risen by a factor of more than three and a half, to 29.3 percent. The increase in resource-based government revenues during the 1970s comes almost entirely from the province of Alberta, and is therefore likely due to the increased rent capture associated with petroleum extraction.

To complete the first step in my counterfactual exercise I remove my estimate of total resource rent from Canadian GNP. The resulting preliminary counterfactual GNP figures (counterfactual 1 in Figure 6) reflect the value of the domestic resource industries' direct contributions to the Canadian economy. In this counterfactual economy, mean real GNP per capita would have been 10.3 percent lower than we actually observe, and twentieth-century real GNP per capita growth would have averaged 2.01 percent per year—less than 0.01 percent higher than we actually observe.³⁷

Therefore, it appears that the removal of the direct contributions made by Canada's natural resource industries has virtually no impact on twentieth-century growth rates, but it quite dramatically lowers Canadians' average income levels throughout the century.

Turning to the second step in my counterfactual exercise, I attempt to quantify the value of the resource industries' indirect contributions to the Canadian economy. I wish to estimate the impact that removing resource-intensive production has on the output produced by non-resource-intensive manufacturing industries and service sector industries. As I argued in the previous section, this impact operates through a reduction in demand for non-resource-intensive manufacturing and service sector output, and an increase in the domestic component of resource-intensive intermediate input prices.

The parameter estimates from equations 2–5, in conjunction with a counterfactual 100 percent reduction in resource output, facilitate the calculation of a rough approximation of the impact that the elimination of domestic resource-intensive production might have had on the value added generated by non-resource-intensive manufacturers and service sector industries, both due to reduced resource output and increased resource-intensive intermediate input prices (counterfactual 2 in Figure 6).³⁸

Again turning to Table 3 we can see that the removal of the resource industries' indirect contributions results in an additional 7.5 percent reduction in counterfactual Canadian GNP, averaged over the 1900–1999 period.

As a final step I combine and remove my counterfactual 1 and 2 estimates of the direct and indirect contributions made by Canada's natural resource industries from observed GNP. The figures in Table 3 indicate that the removal of energy, fishing, forestry, and mining industries from the Canadian economy would have reduced average real GNP per capita by 17.7 percent during the twentieth century, but the rate of change in real GNP per capita would have been virtually unchanged—increasing by just 0.06 percent per year, on average. From Figure 6 we can see that the combined counterfactual real GNP per capita follows the observed real GNP per capita quite closely, gaining only slightly during the first and third quarters of the century, and subsequently falling back between 1920–1940, 1965–1980, and 1990–1999.

To put the results from this counterfactual exercise into some comparative context: according to the Penn World Tables, in 1999 Canada enjoyed the fourth highest real GNP per capita in the world—just below Norway's and just above Denmark's. If Canada had no resource-intensive production during the twentieth century, by 1999 Canadian real GNP per capita would have fallen to eighteenth in the world, just below Germany's and just above France's. Clearly the persistence of large, efficient resource extraction and processing industries not only had no detrimental effect on Canada's twentieth-century growth performance, it also seems to have made the average Canadian considerably wealthier.