Industrial Structural Change and the Post-2000 Output and Productivity Growth Slowdown: A Canada-U.S. Comparison

Almon, Michael-John, Tang, Jianmin, International Productivity Monitor

THE INDUSTRIAL STRUCTURE of the Canadian and U.S. economies have each been in a pro longed period of evolution that has become even more pronounced in the wake of the 2008 financial crisis and subsequent economic downturn that a number of journalists have taken to calling "The Great Recession". (2) While the two firmly interconnected economies face a host of similar internal trends (such as the declining share of the labour force concentrated in primary and secondary industries) and external pressures (the increasing competitive pressures from emerging markets among them), the reactions and adjustments of various industries have not necessarily been of uniform direction and magnitude on both sides of the border. (3)

Canada's business sector has generally underperformed its U.S. counterpart in two key (and related) economic indicators over the past two decades: real gross domestic product (GDP) growth and aggregate labour productivity growth (see Summary Table). (4) In order to understand the driving forces behind what appears to be divergent performances of two highly integrated economies, it is not enough to merely know that the Canadian business sector has underperformed, but becomes necessary to

identify the sources of growth in both countries. To that end, we analyse the industry contributions to real GDP growth and to aggregate labour productivity growth in Canada and the United States from 1987 to 2008. We also compare and contrast the results in order to identify which industries are more important as sources of growth and to better understand the shift in the composition of the two economies.

The de-industrialization of advanced economies has been a well-publicized trend in the postwar period, with the share of the economy accounted for by manufacturing and primary industries diminishing over the decades as more and more economic activity occurs in serviceproducing industries. (5) Caves (1980) has shown that the growth in the demand for goods is outpaced by the growth in the demand for services as economic gains lead to rising real incomes, but the recent acceleration of this trend likely has more to do with the unbalanced foreign and domestic supply and demand conditions that have emerged in recent years. This lack of balance results in uneven changes in the relative price of real output and that, in turn, leads to an adjustment in how production resources are allocated across industries. A simplified model of the economy developed by Baumol (1967) suggested, at its core, that resources will be absorbed predominantly by "stagnant" industries and have a downward impact on the overall labour productivity growth rate of the economy. It should be noted that this particular result, often called "Baumol's cost disease," is still subject to rather vigorous dispute and discussion (Nordhaus, 2006).

In order to best capture this facet of the conversation, the decompositions of industry-level contributions for both real GDP and labour productivity growth are conducted using a decomposition method developed by Tang and Wang (2004, 2010). (6) We feel that this particular framework is the most appropriate way to decompose industry-level contributions due to its ability to effectively utilize the implicit information inherent in the chain-Fisher index method of computing real economic activity, while simultaneously preserving the additivity feature of traditional decompositions that use fixed-weight real GDP.

In the case of real GDP growth, this decomposition technique allows us to identify how much of the contribution to growth stems from the quantity effect and how much from the price effect. The use of the chain Fisher index to con struct real GDP, as is done by both the Canadian and U.S. statistical agencies, (7) results in the inclusion of the value of production as well as the real quantity of products or services produced in the economy in real GDP, but it also results in a loss of additivity to real GDP, increasing as one moves away from the base year. (8) As a result, most traditional decompositions eschew the chain Fisher index method of calculating real GDP because of the additivity issue and instead use the Laspeyres index fixedweight method. Industry-level outputs over an entire observation period are therefore evaluated at their output prices in the base year and ignore any relative price effect. A consequence of this choice, however, is that the importance of industries that have experienced declining output prices will be overemphasized and, conversely, industries with increasing prices will be underemphasized.

Therefore, traditional methods of decomposition are likely to misestimate the contribution of industries to real GDP growth and labour productivity growth (see Jorgenson and Stiroh, 2000; Stiroh, 2002; Faruqui etal., 2003; Jorgenson, 2004; Ho, Rao, and Tang, 2004; and Sharpe and Thomson, 2010). Moreover, the trend in aggregate real output is particularly sensitive to the base year, with similar periods of decomposition able to provide conflicting results if different base years are selected. This problem may become particularly acute if there are significant technological advances over the period that result in sharp declines in output prices for an industry relative to the total economy, as has been observed of information and communication technologies, particularly over the 1990s and 2000s. (9)

The framework developed by Tang and Wang allows us to take advantage of the relative weighting of goods and services in real GDP that arises from the natural valuation of output made by producers and consumers. Consequently, it captures some of the forces that may be causing a shift in the industrial structure of the business sector and avoids the pitfalls outlined above. The incorporation of the price effect in measuring industry-level contribution to real GDP (and, by extension, labour productivity) is consistent with Diewert (2008) and with the economic approach outlined by Diew ert (2002). (10)

In the case of real GDP growth, this means that an industry contributes through two separate streams: real output and output price. For example, if demand for the products of an industry increase, it will likely result in a positive contribution through both streams. Relative output prices will increase to reflect higher demand, resulting in a positive price effect, and will likely spur the industry to increase output, thus resulting in a positive output effect. This is explored in section three, which details the decomposition technique for real GDP growth by industry in the business sector.

Like real GDP growth, industry-level contributions to aggregate labour productivity growth in Canada and the United States can be identified as belonging to two distinct sources: a pure productivity effect and a shift effect. The contribution from the pure productivity effect is due to productivity growth within the industry, while the shift effect stems from the reallocation of production resources and the change in output valuation across industries with differing productivity levels. Thus, the shift effect here captures the change in the "economic significance" of industries in terms of resource use and output valuation. It differs from the reallocation effect by which the traditional terminology only means the effect from the reallocation of production resources. The methodology and results of the decomposition of aggregate business sector labour productivity are explained in more detail in section four.

Using this more detailed technique, we decompose real GDP growth and aggregate labour productivity growth for Canada and for the United States over the 1987-2000 and 20002008 periods into their component industrylevel contributions and further separate that contribution into the aforementioned effects. (11) This level of detail allows us to not only identify the individual role of each industry in the aggregate performance of the Canadian and U.S. business sector, but also to better understand the nature of the forces driving that performance.

The results of the decomposition indicate that Canada's economic slowdown between the two periods is largely due to the manufacturing sector, through both possible channels, while much of the slowdown in real GDP growth in the U.S. business sector can be traced to a decreased contribution from service-producing industries. In terms of aggregate labour productivity growth, Canada's extraction industries cannot be cited as the reason behind the post-2000 malaise, as has often been the case in traditional decompositions, despite a negative contribution through the pure productivity effect. It is the poor productivity performance of industries in the Canadian manufacturing sector that can again be highlighted as the main culprit of Canada's slowdown in aggregate labour productivity growth from the 1987-2000 period to the 2000-2008 period. However, in explaining the persistence of the gap in Canada-U.S. productivity growth, the manufacturing sector cannot be isolated as the singular cause. While it remains true that the Canadian manufacturing sector can shoulder a fair share of the blame for the poor relative performance of the business sector as a whole, Canada's service sector also continued to contribute far less to labour productivity growth than its U.S. counterpart.

An Overview of Industry Structure Change in Canada and the U.S.

Before decomposing the industry-level contributions to real GDP and aggregate labour productivity growth, it is essential to obtain a sense of how the industry structure of Canada and the United States has transformed over the 1987-2008 observation period. In order to achieve this, we examine the relative size of each industry in the business sector of Canada and the United States in terms of total hours worked, identifying the importance of the industry to the business sector as an employer, and in terms of nominal value added, demonstrating the direct impact of the industry on the business sector in nominal output. These two measures of industry size are explored for both Canada and the United States within this section, in addition to an examination of the contribution of each industry to employment growth in the business sector. (12) A third size measure employed in our analysis, combining both a labour component and an output price component to determine the relative importance of the industry for aggregate business sector labour productivity, is discussed in section four.

The observation period is divided into two main periods throughout this paper, 1987 to 2000 and 2000 to 2008. The data sources are discussed in Appendix A. All industries include private as well as non-private activities (if applicable). (13) The "business sector" is total economy minus public administration and owner-occupied dwelling. Thus, our aggregate "business sector" differs from the traditional business sector that only includes private activities. Despite this departure, for simplicity, we continue to refer the aggregate as the business sector.

Canada

There have been some important shifts in the composition of the Canadian business sector over the 1987 to 2008 period. In terms of hours worked, manufacturing and primary industries (such as agriculture, fishing, and forestry) have all declined in relative importance (Table 1). (14) Agriculture, forestry, fishing and hunting went from accounting for 6.0 per cent of all hours worked in 1987 to only 2.8 per cent in 2008. While 18.6 per cent of all hours worked in the business sector in 1987 were in the manufacturing sector, this was reduced to only 12.3 per cent by 2008. The services sector share of total hours worked increased 7.9 percentage points, from 65.4 per cent in 1987 to 73.3 per cent in 2008. This is consistent with the trend among most advanced economies towards a more service-oriented business sector.

Between 1987 and 2008, the total hours worked in the Canadian business sector grew at an average annual rate of 1.60 per cent, with the rate of expansion slightly faster from 1987 to 2000 (1.66 per cent) than it was from 2000 to 2008 (1.49 per cent) (Table 2). The vast majority of this growth originated in the service sector, which was the source of nearly nine-tenths of all the growth in hours worked over the entire period. Half of the contribution from the service sector was attributable to three industries: education, health care, and social assistance; professional and business services; and administrative and waste management. The manufacturing sector was a net drag on the growth rate for total hours, despite positively contributing to the growth rate between 1987 and 2000. In the 1987-2000 period, nine of sixteen manufacturing industries negatively contributed to growth. In the 2000-2008 period, thirteen of sixteen industries reported contracting total hours worked, with the most significant downward period-to-period adjustment coming in transportation equipment manufacturing. The contribution to the growth of hours worked from the mining and construction sectors grew from 1987-2000 to 2000-2008, increasing the importance of these sectors as manufacturing became a net drag on growth and the contribution from service-producing industries diminished.

In terms of the share of nominal value added (Table 3), the manufacturing sector experienced a drop from 20.6 per cent in 1987 to 13.6 per cent in 2008 similar to the overall decrease observed in the share of hours worked. (15) But this obscures the fact that much of the manufacturing sector's decline in relative importance came in the 2000s . In fact, the manufacturing sector held a 22.0 per cent share of nominal value added produced in the Canadian business sector in 2000, up 1.4 percentage points from 1987. Within the manufacturing sector itself, the transportation equipment manufacturing industry, which is primarily composed of the auto and aerospace sectors in Canada, closely tracks the pattern of rise and then decline of the sector as a whole and is perhaps driving much of this trend. In 1987, transportation equipment represented 2.4 per cent of all business sector value added and 12 per cent of all value added produced within the manufacturing sector. By 2000, coinciding with dramatic increases in the annual sales of motor vehicles over the late 1990s, transportation equipment manufacturing industries comprised 3.9 per cent of value added in the business sector and 18 per cent of the nominal value added produced in manufacturing. In 2008, on the heels of a period of declining sales in North America (the worst of which were to take place in 2009), the share of value added produced by transportation equipment manufacturing had contracted to only 1.6 per cent and once again represented only 12 per cent of the output in a Canadian manufacturing sector that was much smaller than it had been eight years prior.

Moreover, service sector industries did not experience a dramatic increase in the relative size of their value added corresponding to an expansion in total hours worked. Between 1987 and 2008, the share of the service sector in terms of nominal value added rose from 59.2 per cent to only 62.3 per cent.

Mining (including oil and gas extraction) in Canada saw the most notable increase in relative size, due primarily to the oil and gas extraction industries. Mining represented 5.6 per cent of the nominal value added produced by the Canadian business sector in 1987 and by 2008 it had more than doubled its share to 11.8 per cent.

The United States

The composition of total hours worked in the U.S. business sector has largely followed the same pattern as in Canada (or, more likely, the Canadian economy has realigned in response to changes in the U.S. economy) (Table 1). Agriculture, forestry, fishing, and hunting declined from 4.1 per cent of total hours in 1987 to 2.7 per cent in 2008 and the share represented by the manufacturing sector fell from 20.7 per cent to 12.7 per cent over the same period. Services rose from 66.4 per cent to 75.7 per cent.

The average annual growth for hours worked in the U.S. business sector was 1.11 per cent between 1987 and 2008, lower than the growth rate observed in Canada over the same period (Table 2). While growth was strong in the United States from 1987 to 2000 (1.78 per cent), it slowed down significantly over the 2000-2008 period (0.02 per cent). As in Canada, the service sector was the primary source of employment growth, responsible for nine-tenths of the growth rate in the 1987-2000 period and experienced a sizeable decline in its total contribution from one period to the next (from 1.70 percentage points per year in the 1987-2000 period to 0.46 percentage points per year in the 2000-2008 period). The magnitude of this decline far exceeded that in Canada (0.29 percentage points). The U.S. manufacturing sector had an even more negative contribution to growth in total business sector hours worked than its Canadian counterpart and was negative in both the 1987-2000 period and the 2000-2008 period, but to a far greater degree in the 20002008 period. The U.S. mining sector was less of a source of growth than it was in Canada and the construction sector declined in importance between the 1987-2000 period and the 20002008 period.

While the change in the industry-level composition of nominal value added in Canada did not track the shares of total hours worked represented by each industry, most notably in the mining sector, the United States demonstrated a more stable connection between the two (Table 3). Manufacturing value added, which experienced a smoother drop in relative size in the United States than was experienced in Canada, fell from 22.1 per cent of value added in the U.S. business sector in 1987 to 15.3 per cent in 2008. Only computer and electronic equipment manufacturing experienced an increase between 1987 and 2000 (from 2.2 per cent to 2.3 per cent) that was then followed by a fall in relative size (to 1.8 per cent in 2008). However, even this decline in the relative importance of the industry over the 2000-2008 period compares favourably to the contraction experienced by its Canadian counter-part after 2000.

Services increased in relative importance from 64.5 per cent of business sector nominal value added in 1987 to 72.8 per cent in 2008, led by the largest service-producing industry in the United States: finance, insurance, real estate (FIRE) and management of companies. FIRE and management services industries represented 15.8 per cent of business sector value added in 1987 and increased to 18.2 per cent in 2000 and to 19.0 per cent in 2008. In Canada, the share of output represented by FIRE and management also increased, but at a much slower pace.

The U.S. mining sector, at 2.0 per cent of nominal output in 1987, saw its importance diminish to 1.3 per cent in 2000, but then it rebounded to 2.5 per cent by 2008 on the back of a global commodities boom. Both the oil and gas extraction industry and other forms of mining followed the overall sector trend.

Industry Contribution to Real GDP Growth

As noted in the introduction, industry-level contributions to real GDP growth are calculated based on a decomposition technique developed by Tang and Wang (2010). Essentially following a top-down approach, the technique decomposes aggregate real GDP growth into industry components, taking into account the effects from changes in both supply and demand conditions.

The Decomposition Technique

Define V, [v.sup.r] and P as nominal GDP, real GDP and the GDP deflator, respectively. (16) In addition, let v-[v.sup.r.sub.i], and [p.sup.i] be the nominal value added, real value added, and value added deflator for industry i. The sum of industry nominal value added is equal to nominal GDP, that is, V = [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]. (1)

For a given year, real GDP can be decomposed into its industry components:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

where [??] is the real price of value added for industry i, defined as

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.].

Equation (1) shows that real GDP can be expressed as the weighted sum of the value added quantities of its constituent industries. These weights are the real prices of value added. Thus, the formulation values industry real output more when its relative price rises and less when its relative price falls.

Real aggregate GDP growth from year s to year t, where t > s, can also be decomposed into industry growth components:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (2)

Where = [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] is the nominal value added share in total GDP,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] is real value added growth of industry i over the period from s to t, and

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

is the per cent change in real value added price of industry i over the period from s to t.

Equation (2) shows that each industry contributes to real GDP growth through an increase in real output and/or a rise in real output price. The contribution is weighted by its share of nominal or real GDP at the beginning period.

The two terms on the right-hand side, called the quantity effect and the price effect, respectively, measure the contributions of growth in quantity produced in the industry and a rise in real output price. The sum of the price effect over the industries is positive if increases in real output prices occur in industries with relatively high value added levels and/or relatively high value added growth.

As previously noted, the above decomposition technique has two desirable properties. First, it is consistent with real GDP in the chain Fisher index. In addition to contributing to real GDP growth through a change in output quantity, an industry also contributes positively (or negatively) to real GDP growth when the real output price rises (or falls). Thus, it allows one to identify the sources of each industry's contribution to real GDP growth: quantity effect or price effect. Second, it is additive for any long period as it is not necessary for year s and t to be adjacent.

Changes in the prices and quantities produced of a good or a service in a competitive market are determined by a change in its demand and supply conditions. Given demand conditions, technological progress or a reduction in input cost will cause a positive shift (downward) in the supply curve of an industry. This leads to an increase in output and a decrease in price. And the opposite is true when the supply curve shifts upward, as it might as a result of a decline in production efficiency or an increase in input cost. Similarly, when there is a change in consumers' tastes or income or external forces (e.g. international trade), the demand for the product will change. Given supply conditions, an increase in demand (a positive upward shift in the demand curve) for an industry's output will lead to an increase in both its quantity and price, and the opposite is true for a decrease in demand.

If an industry in a competitive product market experiences a positive demand shift and a positive supply shift, there will be an increase in quantity; but the net effect on price will depend on the relative strength of the two shifts. If the demand shift is stronger, one will observe an increase in price and if the supply shift is stronger, one will observe a decline in price. Similarly, if the industry experiences a negative shift in both demand and supply, there will be a decline in quantity but the net effect on price will again depend on the strength of one shift against the other. In the remaining two possibilities, where the shifts are in opposite directions, the effect on price can be determined, but the effect on quantity depends on the relative strength of the shifts in demand and supply. Because it is difficult to untangle effects from demand and supply shifts in a given period, this paper only addresses the net shift experienced by each industry through the quantity effect and the price effect.

Industry Contribution to Output Growth in Canada and the United States

The decomposition technique, equation (2), is applied to the business sector in Canada and the United States, using a comparable data set on value added and labour input for the two countries, as is discussed in Appendix A. For both Canada and the United States, industry real value added growth is presented in Table 4. Of course, the rate of growth of real value added in an industry does not necessarily provide much information on its contribution to real GDP growth rate, as this contribution is also dependent on the change in the industry real value added price (Table 5). An industry with positive real value added growth and a decline in real value added price may be a net drag on real GDP growth. The decomposition results of industry contribution to real GDP growth are presented in Table 6 for Canada and Table 7 for the United States.

Canada

Examining the industry contribution to real GDP growth in the Canadian business sector we observe that the biggest contributor to growth across both time periods was the services sector (Table 6). Of the 2.87 per cent annual growth rate reported in the business sector between 1987 and 2000, 1.77 percentage points (almost two-thirds of all growth) was from the services sector. Within the service sector, the largest sources of growth were finance, insurance, real estate (FIRE) and management services industries, …

Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: Industrial Structural Change and the Post-2000 Output and Productivity Growth Slowdown: A Canada-U.S. Comparison. Contributors: Almon, Michael-John - Author, Tang, Jianmin - Author. Magazine title: International Productivity Monitor. Issue: 22 Publication date: Fall 2011. Page number: 44+. © Not available. COPYRIGHT 2011 Gale Group.

This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.