Economic InputOutput Life-Cycle Assessment of ... - ACS Publications

23 Jan 2007 - Policy Analysis. Economic Input-Output Life-Cycle. Assessment of Trade Between. Canada and the United States. JONATHAN NORMAN,...
0 downloads 0 Views 279KB Size
Policy Analysis Economic Input-Output Life-Cycle Assessment of Trade Between Canada and the United States JONATHAN NORMAN, ALEX D. CHARPENTIER, AND HEATHER L. MACLEAN* Department of Civil Engineering, University of Toronto, 35 St. George Street, Toronto, Canada, M5S 1A4

With increasing trade liberalization, attempts at accounting for environmental impacts and energy use across the manufacturing supply chain are complicated by the predominance of internationally supplied resources and products. This is particularly true for Canada and the United States, the world’s largest trading partners. We use an economic input-output life-cycle assessment (EIO-LCA) technique to estimate the economy-wide energy intensity and greenhouse gas (GHG) emissions intensity for 45 manufacturing and resource sectors in Canada and the United States. Overall, we find that U.S. manufacturing and resource industries are about 1.15 times as energyintensive and 1.3 times as GHG-intensive as Canadian industries, with significant sector-specific discrepancies in energy and GHG intensity. This trend is mainly due to a greater direct reliance on fossil fuels for many U.S. industries, in addition to a highly fossil-fuel based electricity mix in the U.S. To account for these differences, we develop a 76 sector binational EIO-LCA model that implicitly considers trade in goods between Canada and the U.S. Our findings show that accounting for trade can significantly alter the results of life-cycle assessment studies, particularly for many Canadian manufacturing sectors, and the production/consumption of goods in one country often exerts significant energy- and GHG-influences on the other.

Introduction Growing concerns about accelerated climate change, conservation of non-renewable resources, and security of energy supplies among industrialized countries has placed a renewed emphasis on understanding the total energy use and greenhouse gas (GHG) emissions resulting from industrial production. In recent years, this understanding has improved with the development of novel Life-Cycle Assessment (LCA) techniques that incorporate economic input-output accounts to reliably estimate the total energy use and GHG emissions across an industry’s entire supply chain (1, 2). However, in an era of ever-increasing trade liberalization, attempts at accounting for total supply chain effects are complicated by the predominance of internationally supplied resources and products. * Corresponding author phone: 416-946-5056; e-mail: hmaclean@ ecf.utoronto.ca. 10.1021/es060082c CCC: $37.00 Published on Web 01/23/2007

 2007 American Chemical Society

Consideration of trade is important for highly interdependent economies such as those of Canada and the United States, the world’s largest bilateral trading partners. Imports of goods to the U.S. totaled $1,670 billion in 2005, of which 17% originated in Canada; while imports of goods to Canada totaled $314 billion in 2005, of which 56% originated in the U.S. (3). (All dollar values are reported in U.S. dollars). This trading volume reflects a high degree of integration between Canadian and U.S. industry, which is particularly true for the manufacturing sectors, and extends down the entire manufacturing supply chain. The automotive industry is a prime example: every vehicle assembled in North America contains nearly $1,250 of Canadian parts (4), without considering the indirect Canadian content of the parts themselves (e.g., the metals, castings, and energy used to create the parts). This type of cross-border economic reliance between Canada and the U.S. is not unique to the automotive sector; it is common among many North American industries. Moreover, this reliance has been generally increasing in the decade following the North American Free Trade Agreement (3). Given this high degree of economic integration, it is important to consider that Canadian and U.S. industries will often exhibit significant differences in production structure and overall manufacturing energy intensity (5). In this regard, it is logical to expect that international differences in resource use, energy efficiency, environmental legislation, and production methods may result in dissimilar life-cycle environmental impacts of industrial production between those nations. Indeed, ongoing efforts studying aggregate sectors of the global economy have hinted at potentially significant differences in sector-specific energy intensity and carbon intensity between industrialized nations (6, 7). Accounting for these differences in the context of Canada and U.S. trade for both environmental policy analysis and LCA is particularly compelling. This is because, despite the differences in industrial production, the two countries share a high degree of policy and geographic integration: in other words, Canada and the U.S. share not only economies, but also airsheds and energy distribution systems, and they cooperate frequently on environmental and energy policy as a result of this closeness. Clearly, the foregoing has significant implications for North American LCA practitioners, manufacturers, and policy-makers. There are two important reasons for this: First, conducting an LCA for a product or industry that ignores the effects of trade may result in significant over- or underestimates of the true environmental impacts. Second, production or consumption of products in one country can result in significant environmental impacts imputed on the other via the cross-border procurement of goods and services. Quantifying these cross-border effects is thus important to shed light on these interdependencies and enable more accurate environmental analyses/LCAs of products. Despite this importance, the overall understanding of the effect of international trade systems on the life-cycle environmental impacts of production is presently limited (8), particularly for North America. While recent and important efforts are underway to examine the environmental implications of and trade in manufactured products and associated with household consumption, notably in European nations (9-13), as well as in South America (14), the energy use and GHG emissions associated with trade relationships between Canada and the U.S. remain largely VOL. 41, NO. 5, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1523

unexplored, particularly in the context of LCA. Our work builds upon these studies and earlier studies that have examined pollution linkages associated with NAFTA (15, 16), as well as more recent studies of environmental load displacements between nations (17). While these studies have made important contributions to the energy policy field, there has been little quantification of the energy and GHG linkages between Canada and the U.S., or sector-specific differences in energy and GHG intensity between the two countries, which has important implications for North American LCA. While the possibility of these international differences between Canada and the U.S. has been alluded to in the energy policy literature (5, 18), this issue has not been studied in detail, particularly not in the context of LCA. Based on the foregoing, the objectives of this paper are the following: (1) to identify differences in industrial energyuse intensity and GHG-emissions intensity between Canada and the U.S. and interpret the major causes for differences; (2) to examine the implications of these differences for LCA based on the level of trade between Canada and the U.S. for different sectors of the economy through the development of a bi-national U.S./Canada economic input-output (I/ O)-based LCA model; and (3) to illustrate how this model can be used to quantify the cross-border energy use and GHG emissions that result from one nation’s demand for a product, using a case-study of motor vehicles. Overall, we will shed light on actual differences in energy use and GHG emissions in Canadian and U.S. industry, and demonstrate a method of quantifying the effects of trade in LCA studies.

Methodology We employ an Economic Input-Output Life-Cycle Assessment (EIO-LCA) modeling technique to estimate the total energy use and GHG emissions resulting from industrial production in Canada and the U.S. EIO-LCA is based on an environmental I/O modeling approach first proposed by Leontief (19) and is described in detail in the Supporting Information. The implementation of a comprehensive EIOLCA modeling approach for the US economy is described in refs 20 and 21. The basic EIO-LCA method couples national I/O tables, which quantify the economic interdependencies between industrial sectors in matrix form, with emissions and energy use data for each sector that are normalized to economic output. EIO-LCA thus enables an analyst to consider the environmental effects throughout the economy (i.e., the direct and indirect effects across the supply chain) that result from a change in output for a particular industry (2). Several other works have expanded on the EIO-LCA framework such as that conducted by Suh on hybrid models (22, 23) and by several research groups on multi-regional models (13, 24, 25). National EIO-LCA Models. National EIO-LCA models of the above form are currently maintained and described in detail for the U.S. economy by the Green Design Institute at Carnegie Mellon University (26), and for the Canadian economy by researchers at the University of Toronto (27). For this study, the existing national models are revised to be directly comparable between the Canadian and U.S. economies for 1997 (the most recent year of available high-quality and comparable data). The model revisions are described in detail in the Supporting Information. We base these revisions on updated economic (technical) coefficient matrices for 1997, obtained in summary form from the U.S. Bureau of Economic Analysis (28) and Statistics Canada (29), which we then aggregate to a total of 76 comparable economic sectors in accordance with the 1997 North American Industrial Classification System (NAICS). Each economic matrix is 1524

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 5, 2007

multiplied by normalized vectors of energy use and GHG emissions coefficients (specified as the amount of energy use and GHG emissions per dollar of output for each industrial sector). This multiplication yields the total direct and indirect energy used and GHGs emitted for a given level of demand for an industry’s products. All sectors considered in our analysis are shown in Table 1, along with their respective NAICS codes. It is noted that normalization of the energy use and GHG emissions coefficients was facilitated by the fact that that environmental and economic datasets in both countries are commonly reported by NAICS conventions, enabling broad data comparability across sectors between the countries. For both national models, energy use is defined as the total fuel use by sector (including coal, natural gas, motor gasoline, diesel fuel, aviation fuel, fuel oil, liquid petroleum gas, coke, and non-fossil generated electricity) converted to terajoules (TJ). Greenhouse gases correspond to the total global warming potential (GWP) of combined carbon dioxide, methane, and nitrous oxide emissions by sector, reported in terms of CO2 equivalents (CO2e) calculated in accordance with the conventions established by the Intergovernmental Panel on Climate Change (26, 27). Energy use and GWP data for the United States by NAICS sector for 1997 were provided by the Green Design Institute (30, 31), based on publicly available data from American statistical agencies, while those for Canada were obtained from Statistics Canada’s Environmental Accounts Division (32). These data were normalized to economic output for each sector considered using data obtained from the U.S. Bureau of Economic Analysis (33) and Statistics Canada (29). Overall, the quality of the environmental data is expected to be quite high, since both countries place a good deal of emphasis on the maintenance of energy and GHG data, particularly in the current era of international attention to these metrics. Furthermore, the environmental datasets used are broadly consistent with data relied upon for national and international policies such as the Kyoto protocol. As such they represent a reasonable basis upon which to develop a policy analysis and LCA tool such as EIO-LCA. The national EIO-LCA models are used to quantify the energy use and GHG emissions associated with the production of products from 45 agricultural, resource, and manufacturing sectors. The total direct and indirect effects associated with manufacturing these products are calculated for each country and adjusted by annual purchasing power parity exchange rates (PPPs) for 1997. PPPs equalize the purchasing power of different currencies in their home countries for a given basket of goods (34). The adjustment therefore allows for a more realistic comparison of the industrial emissions/energy use resulting from dollars spent in Canada and the U.S. To our knowledge, this is the first comparison of industrial energy intensity and GHG intensity between Canada and the U.S. that considers indirect (i.e., total supply chain) effects and that maintains a high level of sectoral detail in the analysis. Bi-National Canada-U.S. EIO-LCA Model. We create a bi-national Canada-U.S. EIO-LCA model by linking the national models described above through trade flows for each industrial sector. The full description of the methodology for linking the national models is found in the Supporting Information. We summarize key aspects in the current section. The methodology for linking the models is based on a Chenery-Moses multiregional I/O framework (35), modified to account for competitive imports (36), which are included by convention in the published economic I/O matrices. Using this approach, we define an import coefficient for each industry, Ti, based on the assumption that the bilateral import volume for each sector in a country is

TABLE 1. Industries Considered in the EIO-LCA Models by NAICS Code NAICS

Description

NAICS

Description

11A0 1130 1140 1150 2111 2121 2122 2123 2131 2211 221A 2301 2302 2303 3110 3121

Crop & Animal Production Forestry & Logging Fishing, Hunting & Trapping Support Activities for Agriculture Oil & Gas Extraction Coal Mining Metal Ore Mining Non-Metallic Mineral Mining Support Activities for Mining Electric Power Generation Natural Gas Distribution & Utilities Residential Construction Non-Residential Construction Repair Construction Food Manufacturing Beverage Manufacturing

335A 3361 336A 3364 336B 3370 3390 4100 4A00 4810 4820 4830 4840 4850 4860 48B0

3122 31A0 3150 3160

Tobacco Manufacturing Textile & Textile Products Clothing Manufacturing Leather & Allied Products

4910 4920 4930 51A0

3210 3221 3222

Wood Product Manufacturing Pulp, Paper & Paperboard Mfg. Converted Paper Product Mfg.

5120 5131 513A

3231 3241 3251 3252

5A01 5A02 541A 5418 5610

Administrative and Support Services

3254 325A 3260 3270 3310 3320 3330 3341 334A

Printing & Related Support Activities Petroleum & Coal Products Basic Chemical Manufacturing Resin, Synthetic Rubber, Artificial/Synthetic Fibers and Filaments Mfg. Pesticides, Fertilizer & Other Agricultural Chemical Mfg. Pharmaceutical & Medicine Mfg. Miscellaneous Chemical Product Mfg. Plastic & Rubber Product Mfg. Non-Metallic Mineral Product Mfg. Primary Metal Manufacturing Fabricated Metal Product Mfg. Machinery Manufacturing Computer & Peripheral Equipment Mfg. Electronic Product Manufacturing

Electrical Equipment & Component Mfg. Motor Vehicle Manufacturing Motor Vehicle Body, Trailer, & Parts Mfg. Aerospace Product & Parts Mfg. Other Transportation Equipment Mfg. Furniture & Related Product Mfg. Miscellaneous Manufacturing Wholesale Trade Retail Trade Air Transportation Rail Transportation Water Transportation Truck Transportation Transit and Ground Passenger Transportation Pipeline Transportation Scenic and Sightseeing Transportation and Support Activities for Transportation Postal Service Couriers and Messengers Warehousing and Storage Publishing Industries, Information Services and Data Processing Services Motion Picture & Sound Recording Industries Radio and Television Broadcasting Pay TV, Specialty TV and Program Distribution and Telecommunications Finance, Real Estate & Leasing Insurance Professional, Scientific & Technical Services Advertising and Related Services

5620 6100 62A0 6220 7100 7200 8110 81A0 8130

3352

Household Appliance Manufacturing

GS

Waste Management & Remediation Services Educational Services Health Care Services (except Hospitals) and Social Assistance Hospitals Arts, Entertainment and Recreation Accommodation and Food Services Repair and Maintenance Personal Laundry Services & Private Households Religious, Grant-Making, Civic, and Professional and NFP Organizations Government Services

3253

proportional to that country’s demand for the sector (import proportionality assumption):

Ti )

mi n

∑z

ij

(1)

+ yi

j)1

where mi refers to the dollar value of bilateral imports for industry i; Σzij refers to the total intermediate (interindustry) demand for industry i, in dollars; yi refers to the final (consumer) demand for industry i, in dollars, and n is the number of sectors (industries). Precedence for the use of import proportionality assumptions in interregional studies is found throughout the literature that examines the crossborder impacts of interregional trade (37, 38). Furthermore, it is notable that the assumption is often relied upon by statistical agencies when compiling national accounts (39). Trade (import) data for 1997 were obtained for the Canadian industrial sectors from Industry Canada (40) and for the U.S. industrial sectors from the U.S. Census Bureau (41). It is noted that trade in the service sectors generally accounts for less than 10% of total trade between Canada and the U.S. (3). Because of its lesser significance, and due to limitations in service trade data, we do not consider trade in service sectors and maintain our focus on industrial/ manufacturing sectors throughout this analysis.

The national economic (technical) coefficients are premultiplied by the import coefficients in vector form and linked in a multiregional I/O framework as follows:

[ ] [ XC

XU

)

]

-1

I - (I - Tˆ U,C)AC - cU,C Tˆ C,U AU I - (I - Tˆ C,U)AU - cC,U Tˆ U,C AC

[

(I - Tˆ

U,C

(I - Tˆ

C

U,C

Tˆ C,U YU

U

C,U

Tˆ U,C YC

)Y + c

C,U

)Y + c

]

(2)

where the superscripts U and C represent the U.S. and Canada, the superscripts U,C and C,U represent trade flows from the U.S. to Canada and vice-versa, in dollars; X represents the vector of national economic output; T represents the diagonal matrix of import coefficients; A represents the national technical coefficients matrix; Y represents a specified vector of final demand, in dollars; c represents the currency exchange rate; and I represents the identity matrix. The system is solved to yield the total economic output resulting in both countries for a given final demand scenario in each country (adjusted for the international exchange rate between Canada and the U.S. for 1997). One country’s demand for a particular industry’s products results in domestic production, as well as cross-border demand and production in the other country. Equation 2 captures these relationships by considering infinite rounds of spending to VOL. 41, NO. 5, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1525

FIGURE 1. Comparison of industrial energy intensity between Canada and the United States. determine the total resulting production required by all sectors in each country (i.e., total Canadian output and U.S. output). It should be noted that the model treats imports from the rest of the world as if they were produced in the domestic country, so that the life-cycle emissions and energy use contribution from foreign imports are not lost in the overall LCA. The resulting economic output vector for each country is multiplied by the national vectors of energy use and GHG emissions coefficients (as described in the previous section) to yield the total energy use and GHG emissions by sector. For example, the total energy use is calculated as follows:

ET ) Eˇ CXC+ Eˇ UXU

(3)

where ET refers to the vector of total energy use (i.e., energy used across the U.S. and Canada) by sector, in TJ; Eˇ C and Eˇ U are vectors of energy use coefficients for Canada and the U.S., whose elements consist of the energy use in TJ per dollar of economic output for each sector; and, XC and XU refer to the vector of total economic output for each of Canada and the U.S. as determined from eq 2. It should be noted that the overall bi-national EIO-LCA model relies on proportionality assumptionssmost notably that trade in an industry is proportional to the domestic demand for its products, and that energy use/GHG emissions vary in direct proportion with economic output. While these are simplifications of reality, they enable the model to reliably estimate life-cycle environmental burdens across multiple industries in an economy, and across international borders, which would be difficult otherwise. The bi-national model’s effectiveness is evidenced by the fact that similar multiregional models have recently been used successfully to estimate the environmental burdens of trade in other areas of the world (11, 13, 25). We use the bi-national EIO-LCA model to examine the importance of considering trade between Canada and the U.S. when evaluating the life-cycle impacts of products purchased in either country. We compare the binational model results for each sector with the results of the national models alone to reveal the difference in GHG emissions and energy use. Finally, we employ the bi-national model to quantify the cross-border energy influences (“leakages”) and 1526

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 5, 2007

GHG emissions “leakages” that result from the production of automobiles in each country [building on the concept of carbon leakages as has been described by the Organization for Economic Cooperation and Development (42)].

Results and Discussion Comparing Industrial Siblings: Energy Use and GHG Intensity in Canada and the United States. Figures 1 and 2 show the total (direct + indirect) energy use and GHG emissions, respectively, associated with one million dollars (1997 $U.S.) of production for different industrial sectors of the economy. We have calculated the national “energy intensities” and “GHG intensities” for 45 different industrial sectors using our single nation EIO-LCA models and have directly compared the results for Canada and the United States. The industrial sectors cover agriculture, mining, utilities, construction, and manufacturing. We have also examined the relative breakdowns of fuel sources used to supply energy for the major aggregate industrial sectors, which are shown in Figure 3. The results show important similarities as well as differences between the two countries. Generally speaking, the calculated sectoral energy and GHG intensities for each country are roughly comparable across broad classes of production. For instance, both Canada and the U.S. exhibit similarly low energy and GHG intensities for light manufacturing sectors (NAICS 3110-3160), such as clothing and textile production, and secondary manufacturing sectors (NAICS 3320-3390), such as motor vehicle or appliance production. These are in sharp contrast with the comparably high intensities for resource-intensive sectors like mining (NAICS 2111-2131), utilities (NAICS 2211-221A), or production of chemicals (NAICS 3241-3254). While the broad comparability of the findings may not be unexpected, it does provide some assurance about the validity of each model’s data sources, which is important since the datasets used are unique to each country and taken from unrelated statistical agencies. However, even with relative congruity across broad classes of production, specific and important differences in energy and GHG intensity between Canada and the U.S. are clearly evident. Given the level of trade between the two nations,

FIGURE 2. Comparison of industrial greenhouse gas intensity between Canada and the United States.

FIGURE 3. Final fuel mix differences between Canada and the United States by major sector. these differences can be highly relevant to industry and government decision-makers who might otherwise assume homogeneity in manufacturing energy intensities across the U.S. and Canada. While it is beyond the scope of this paper to dissect the international differences in detail, it is informative to examine the overall trends and the nature of some of the more significant discrepancies. The international variability in energy intensity and GHG intensity observed in Figures 1 and 2 appears to be caused mainly by “realworld” differences in production methods/resource use between Canada and the U.S. and, to a lesser extent, by “apparent” differences caused by data limitations/comparability issues. We have broken down the main sources of the differences into three broad categories: fuel use differences, industrial structure differences, and economic differences, each of which is discussed below. Fuel Use Differences. Differences in fuel sources for energy supply appear to be the most significant factors influencing the international variability in energy and GHG intensity between Canada and the U.S. With few exceptions, the U.S. manufacturing sectors exhibit higher energy and GHG intensity than their corresponding Canadian sectors, a discrepancy that appears to stem mainly from significant differences in fuel use. The most obvious specific case of fuel

use differences is electric power generation (NAICS 2211), in which the U.S. generates more than three-quarters of its electricity from fossil fuels (primarily coal and natural gas) (43), while Canada obtains almost two-thirds of its electricity supply from hydroelectric sources (44). As a result, electricity generation in the U.S. is found to be almost 1.5 times as energy-intensive and more than twice as GHG-intensive as power generation in Canada. This difference in electricity generation fuel mix has significant implications for the overall GHG intensity of the U.S. economy. Indeed, using our data to calculate the total energy use and GHG emissions for all sectors of the economy (including the service sectors), normalized to gross domestic product (GDP), we find that total sectoral activity in the United States is almost 1.3 times as GHG-intensive as the same level of activity in Canada, yet only 90% as energy-intensive. In general, this implies that Canada’s industrial and service sectors, while energyintensive, typically rely on less carbon-based, GHG-intensive fuels than do those of the U.S. This economy-wide discrepancy between Canada and the United States appears to be due to a combination of factors, including high levels of electricity use by service sectors (offices, retail, etc.) in both countries, as well as significant fuel use differences in manufacturing between the countries. VOL. 41, NO. 5, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1527

Other factors, such as cheaper transportation fuel prices in the U.S., may also play a part in the higher overall GHG intensity of the U.S., although these factors are more difficult to verify. Nonetheless, the international fuel use differences are seen in aggregate form in Figure 3, which shows that manufacturing sectors in Canada are far more likely to rely on the nation’s cleaner (and often cheaper) electricity mix for their energy requirements, while those in the U.S. are more likely to rely directly on fossil fuels (notably natural gas, gasoline/fuel oil, and coal) to power their operations. As a result, the agricultural, mining, construction and manufacturing sectors combined are found to be 1.15 times more energy intensive and 1.3 times more GHG-intensive per equivalent dollar of production in the U.S. than in Canada. Overall, the fuel use differences shown in Figure 3 appear to be the most significant factor contributing to the “energy gap” and “GHG gap” between Canada and the U.S. This conclusion is borne out by an analysis of the carbon intensity (7) of the different fuels used by U.S. and Canadian industrial sectors: overall, the U.S. fuel mix is found to be almost 1.3 times as carbon intensive as Canada’s fuel mix. Furthermore, data published by the International Energy Agency similarly suggest that U.S. GHG emissions per amount of energy used is 1.25 times greater than Canada’s. These data, while not a perfect basis for decomposition, do suggest that the vast majority of differences we identify in apparent energy and GHG intensity (i.e., normalized to economic output) between the two countries are attributable to fuel use differences. Industrial Structure Differences. Differences in energy and GHG intensity between the U.S. and Canada are also influenced by differences in industrial structure, which can be viewed in two ways: international variability in processes and production methods (which result in “real-world” differences in intensities) and product mix differences between the countries (which result in “apparent” differences in intensities). A likely example of the former is Canada’s energy-intensive oil sands production, which accounts for more than 1/4 of Canadian crude oil production but essentially none in the U.S. (45); the high levels of energy required for oil sand extraction likely explains much of the cross-border discrepancy in energy intensity and GHG intensity for oil and gas extraction (NAICS 2111). On the other hand, product mix differences between the countries result in more of a data-induced error caused by different sets of commodities being produced under the same industry classification. While product mix differences do not truly reflect “real world” differences in energy and GHG intensity between Canadian and U.S. industries, they are only likely to be an issue in highly aggregated sectors or certain manufacturing sectors with many hundreds of products involving different production techniques. Because of the reasonably high degree of sectoral detail employed, the errors induced by product mix differences in other sectors are likely to have been minimized. Nonetheless, sectors that may be affected by product mix issues should be interpreted with care. Based on levels of aggregation and high levels of product variability that would impact energy use and/or GHG emissions per dollar of product produced, we expect that sectors to treat carefully would include the agricultural sectors (NAICS 11A0 to 1150), pulp and paper manufacturing (NAICS 3221), chemical manufacturing sectors (NAICS 3251 to 325A), plastic and rubber product manufacturing (NAICS 3260), motor vehicle parts manufacturing (NAICS 336A), and miscellaneous manufacturing sectors (NAICS 336B and 3390). Economic Differences. Since the energy and GHG intensities we are considering are normalized to economic output (production) data, some of the discrepancies apparent in Figures 1 and 2 are likely influenced by economic differences between the nations, including international variability in commodity prices (such as natural gas or electricity prices) 1528

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 5, 2007

and differences in sectoral productivity. While we have adjusted the national energy and GHG intensities using PPP exchange rates to account for some of these effects (34), other factors, such as a well-known industrial productivity gap between Canada and the U.S. (46), may still influence the interpretation of the results. For instance, it is important to consider that as a result of the productivity gap, fewer physical goods may be produced in Canada per dollar of economic output, which may be deflating the apparent energy and GHG intensities of Canadian production. Similarly, international price variability for the goods produced could also impact the interpretation of intensities. While it is beyond the scope of this paper to examine these effects in any detail, it is emphasized that these are nonetheless “real world” differences that deserve to be reflected in the bi-national EIO-LCA model. It is notable that some of the variability observed in Figures 1 and 2 may also be attributable to anomalies in international classification. Overall, however, the likelihood of these anomalies is quite low since we have relied on standardized data sources classified by homogeneous NAICS codes. In fact, the only significant classification anomaly that has affected our results is in the Natural Gas Distribution sector (NAICS 221A), which appears on the figures to be significantly more energy and GHG intensive in the U.S. This is largely due to incompatible data classification for energy use and GHG emissions in the natural gas distribution and transmission industries between Canada and the U.S. (32, 47). Overall, the classification discrepancy will have minimal impact on the final results of this study, since the variance is accounted for through proportionally lower intensities in the U.S. “pipeline transportation” sector (NAICS 4860) (47). Nonetheless, greater cooperation in the classification of environmental data collected between the countries would assist in resolving this type of classification discrepancy in the future. It is also worth noting that some of the discrepancy in energy and GHG intensity for the Natural Gas Distribution industry is attributable to more tangible “real world” factors, such as an (on average) 150% greater distribution pipeline travel per volume of natural gas consumed in the U.S. (48, 49), which results in increased fuel consumption and line losses. Interestingly, it is often reported that Canada is more energy-intensive than the United States and roughly equal in terms of GHG intensity (e.g., 6, 7). While these reports may differ somewhat from our findings, it is important to consider that our method is focused specifically on the agriculture, mining, manufacturing, and service sectors of the economy, and intentionally does not include the household sector (personal transportation, heating, cooling, etc) and various fugitive emissions such as those from landfill sites. Furthermore, we rely on an amalgamation of highly disaggregated sector-specific datasets, which contrasts with the more top-down approach that is typically used for national GHG reporting. This approach is advantageous both in terms of accuracy and sectoral detail for trading sectors of the economy, particularly to ensure that the analysis remains focused on national datasets that are comparable, since each nation tends to report differently on non-sectoral emissions (e.g., personal transportation and fugitive sources of emissions) (7, 32). The overall implication of our findings is that various sectors of the economy may be disproportionately GHG intensive and energy intensive, and may be significantly different between nations. In the case of the U.S. and Canada, we have found that U.S. industry as a whole tends to be more GHG intensive than Canadian industry, although certain factors like productivity differences between Canada and the U.S. may influence our findings. Implications of Canada-U.S. Trade for Life-Cycle Assessment. It is clear from the foregoing comparison that significant sector-specific differences do exist in industrial

FIGURE 4. Difference in results for key sectors using the Bi-national EIO-LCA Model versus the National EIO-LCA Model for purchases made in Canada (left) and in the U.S. (right). Only differences greater than 5% are shown energy intensity and GHG intensity between Canada and the U.S. The majority of these differences appear to be caused by differences in production methods, fuel use variability in manufacturing sectors, and differences in production costs/ pricing structure. These differences are lost, however, when using standard LCA methods that do not implicitly account for trade between Canada and the U.S. In particular, given the high level of Canada-U.S. trade that exists for many of these industrial sectors, it may be important for LCA practitioners to consider the effect of this trade on their calculations. To explore this issue, we have employed the bi-national Canada-U.S. EIO-LCA model to examine the importance of bilateral trade on the life-cycle energy use and GHG emissions associated with demand for goods in each country. Figure 4 considers the life-cycle energy use and GHG emissions resulting from a $1 M purchase from each sector within Canada (left side of the graphic) and the U.S. (right side of the graphic)sit shows the percentage difference between the bi-national EIO-LCA model results as compared to the Canadian and U.S. single-nation EIO-LCA models (only sectors exhibiting greater than a 5% difference are shown). This comparison is informative, since the single nation EIOLCA models assume that any imported commodities have been produced domestically. In other words, they ignore the effects of imports and exports (19). By comparing the binational model with the single nation Canadian EIO-LCA model, Figure 4 thus shows the importance of accounting for trade with the U.S. when looking at the life-cycle impacts of purchases from a sector in Canada, and vice versa. For example, the life-cycle energy use and GHG emissions associated with a purchase from the Canadian Computer Equipment Manufacturing sector (NAICS 3341) are 77% higher and 113% higher, respectively, when trade with the U.S. is considered. In general it can be seen from Figure 4 that the most important trade influences on life-cycle energy use and GHG emissions are evident in the primary and secondary manufacturing sectors, which is reflective of the significant fuel

use differences and relatively high levels of trade involved with these sectors. In the context of the previous section, these LCA differences in manufacturing should not be surprising. To continue with the example of the computer industry, for instance, Canada is significantly reliant on the U.S. (and abroad) for the majority of its components for computer systems (50), which are significantly more energyintensive to produce (e.g., the production of a computer chip requires far more energy than the assembly of an end-product computer). Thus, while systems integration is common in Canada, many of the more significant environmental realities associated with component manufacturing would be lost in a Canadian-only EIO-LCA analysis of computers. It is also apparent that the consideration of trade is much more important to Canadian LCA practitioners than U.S. practitioners. Goods purchased in the Canadian economy are more likely to have been produced in the U.S., which, due to the heavier reliance in the U.S. on fossil fuels (as described above), typically results in higher energy use and GHG emissions. Thus demand for products associated with manufacturing sectors in Canada exhibits up to a 50-100% higher energy use and GHG emissions when trade with the U.S. is considered. While the U.S. economy is more selfreliant, certain sectors, most notably transportation equipment manufacturing (NAICS 336), fishing/trapping (NAICS 1140), and electric power generation (NAICS 2211) have important life-cycle influences from north of the border, which show between 10% and 30% differences in overall life-cycle impacts when trade with Canada is considered. It appears therefore, that the consideration of trade can often have an important influence on sector-specific LCA results in U.S./Canada. Thus for sector-specific analysis, the binational EIO-LCA model has the potential to be an important addition to the LCA practitioner’s toolkit, since it can implicitly account for cross-border differences in resource use and production methods. It also has the significant advantage of being able to quantify the cross-border impacts of product purchases and production decisions that one country has on the other, which is highlighted in the following VOL. 41, NO. 5, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1529

FIGURE 5. Life-cycle energy use and GHG emissions resulting from $25,000 in motor vehicle purchases in Canada for the top ten contributing sectors. case study of motor vehicle manufacturing. Case Study: Motor Vehicle Manufacturing. Motor vehicle production forms a significant component of both the U.S. and Canadian economies. However, the sector’s energy use characteristics are quite different in each nation, with U.S. auto manufacturers considerably more reliant on fossil fuels for energy than Canadian manufacturers, who utilize a higher proportion of electricity generated from relatively clean sources (30, 32). As a result, motor vehicle production in Canada is less GHG intensive (which is reflected by the lower Canadian intensity for NAICS 3361 shown previously in Figure 2). Cross-border trade in the sector is significant: 77% of cars purchased in Canada are assembled in the U.S. (totaling about 1.1 million vehicles), while 10% of cars purchased in the U.S. are assembled in Canada (totaling about 1.5 million vehicles), based on 1997 data (51). Furthermore, trade is also significant between the countries further down the supply chain, including trade in motor vehicle parts (components for the motor vehicle parts, etc.) and other durable goods required for the industry, which can result in significant crossborder indirect effects. These direct and indirect implications of trade are particularly important from the perspective of responsible consumer purchasing (i.e., purchases made by both industry supply chain managers and end-use consumers). As we will show below, a significant proportion of the life-cycle energy use and GHG emissions resulting from an average automobile purchased in Canada ultimately reside in the U.S. Given the Canadian auto sector’s high dependence on U.S. manufacturing, we focus our example on the life-cycle impacts of automobile demand from Canada. Figure 5 shows the major life-cycle energy use and GHG emissions (broken down by the top ten contributing sectors of the economy) associated with a $25,000 (1997 $ U.S.) automobile purchase in Canada (adjusted for producer prices and by PPP for international comparability). Importantly, the figure also shows the country where the GHG are emitted. For example, the purchase of an “average” automobile in Canada results in almost 1.35 T CO2e from the electric power generation sector, 83% of which is ultimately emitted in the U.S. In other words, Figure 5a shows that the purchase of an automobile in Canada results in a significant “GHG emissions leakage” onto different sectors of the U.S. economy. Overall, purchasing an automobile in Canada results in a total of 5.3 T CO2e in the U.S. (which corresponds to 74% of the total resulting GHG emissions in both countries). Figure 5b shows a similar, but less pronounced, trend for energy use, with a significant 1530

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 5, 2007

“energy leakage” likewise exerted on certain U.S. industries. It is interesting to note from Figure 5b that, as might be expected, the energy use from local service sectors, such as advertising (NAICS 5418), is seen almost entirely within Canada (i.e., the country where the demand originated). This is not surprising when one considers that direct energy use within the sector accounts for 74% of the total energy intensity for the sector. This concept of international “GHG emissions leakages” and “energy leakages” is highly relevant from the perspective of extended producer and consumer responsibility, which is highlighted in recent studies of the environmental impacts of trade (13, 17). With growing concern about GHG emissions, new emphasis is being placed on responsible purchasing and supply chain management. With global outsourcing however, the supplier manager of a national firm, individual, or government often does not consider the GHGs emitted outside of their own jurisdiction, which (as was shown for the case of automobiles) can be far more significant than those emitted domestically. In reality, countries and industries that are not domestically reliant for certain inputs can result in significant GHG emissions and energy “leakages” on their trading neighbors. The bi-national EIO-LCA model has the potential to be useful for policy-makers and industrial managers who need to quantify the total environmental burdens on the other country to develop appropriate policy responses that address transboundary impacts.

Policy Implications We have shown that for many industries, significant “realworld” differences in energy-use intensity and GHG-emissions intensity exist between Canada and the U.S. These differences can have a significant impact on LCA results. This is particularly true from the perspective of Canadian industries, which are often highly reliant on U.S. producers for their inputs. Overall, because of Canada’s relatively open economy and economic reliance on the U.S., it is clear that environmental policy, LCA studies, and industrial production decisions must be formulated with consideration given to trade, and in particular, trade with the U.S. Similarly, we have also shown that consumption and production of goods in one of the two nations can have an indirect, but often significant, GHG and energy “leakage” due to the strong Canada/U.S. economic interdependencies. The quantification of these leakages may prove particularly relevant to policy-makers in terms of analyzing joint (international)

implementation programs to reduce GHG emissions at the industry level, or by informing the development of North American emissions trading programs. As a final note, it is important to recognize a growing dichotomy between environmental policy formulation and economic reality: while Canada and the U.S. are becoming increasingly integrated economically, environmental policies are often formulated independently at the national (or subnational) level. Indeed, in this post-Kyoto ratification era, environmental policy divergence is becoming more apparent between Canada and the U.S. The results of this paper and the binational EIO-LCA model itself can assist policy-makers to move a step closer in bridging the analytical gap between local environmental policy and the economic realities of the North American trading block. Armed with better knowledge of the international environmental/energy efficiency discrepancies at the industry level, policy-makers will be more able to identify leading and lagging sectors, and can thus prioritize their responses appropriately. Furthermore, with the ability to quantify the energy and GHG leakages that are associated with industrial demand between Canada and the U.S., North American decision-makers can better evaluate the joint national and international environmental impacts of various production policies or demand-management scenarios.

Acknowledgments We thank the Natural Sciences and Engineering Research Council (Canada) and the University of Toronto for support, Thomas Ferguson for research assistance, and Scott Matthews of Carnegie Mellon University for his valuable advice and contribution of data to this study.

Supporting Information Available Methodological description for implementation of bi-national EIO-LCA in Canada and United States. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Matthews, H. S.; Small, M. J. Extending the Boundaries of LifeCycle Assessment through Environmental Economic InputOutput Models. J. Ind. Ecol. 2000, 4 (3), 7-10. (2) Hendrickson, C.; Horvath, A.; Joshi, S.; Lave, L. Economic InputOutput Models for Environmental Life-Cycle Assessment. Environ. Sci. Technol. 1998, 32 (7), 184A-191A. (3) Department of Foreign Affairs and International Trade, Trade and Economic Analysis Division. NAFTA @ 10 - A Preliminary Report; Minister of Public Works and Government Services Canada, 2003; Cat. No. E2-487./2003. (4) Scotia Economics. Canadian Auto Report; January 29, 2003. (5) Theriault, L.; Sahi, R. Energy Intensity in the Manufacturing Sector: Canadian and International Perspectives. Energy Policy 1997, 25 (7-9), 773-779. (6) International Energy Agency. Energy Balances of OECD Countries 2001/2002; IEA: Paris, 2004. (7) International Energy Agency. CO2 Emissions From Fuel Combustion 1971-1999; IEA: Paris; 2001. (8) United Nations University, Institute for Advanced Studies. Industrial Trade Systems and Life-Cycle Assessment. Adv. Perspect. 1999, 3, 6-7. (9) Statistics Sweden. Environmental Impact of Swedish Trade; Report 2002:2; Stockholm, 2002. (10) Hertwich, E. G.; Erlandsen, K.; Sørensen, K.; Aasness, J.; Hubacek, K. Pollution embodied in Norway’s import and export and its relevance for the environmental profile of households; Report IR-02-073; International Institute of Applied Systems Analysis: Laxenburg, Austria, 2002; pp 63-72. (11) Peters, G.; Briceno, T.; Hertwich, E. Pollution Embodied in Norwegian Consumption; NTNU Industrial Ecology Programme working paper 6; Trondheim, Norway, 2004. (12) de Haan, M. Disclosing international trade dependencies in environmental pressure indicators: the domestic consumption perspective; Proceedings of the 14th International Conference on Input-Output Techniques, Montre´al, October 10-15, 2002.

(13) Lenzen, M.; Pade, L.-L.; Munksgaard, J. CO2 Multipliers in MultiRegion Input-Output Models. Econ. Syst. Res. 2004 (2), 16 (4), pp 391-412. (14) Machado, G. V. Energy Use, CO2 Emissions and Foreign Trade: An IO approach applied to the Brazilian case; Proceedings of the 13th International Conference on Input-Output Techniques, Macerata, Italy, August 21-25, 2000. (15) Antweiler, W. The Pollution Terms of Trade. Econ. Syst. Res. 1996, 8 (4), 361-65. (16) Reinert, K. A.; Roland-Holst, D. W. Industrial Pollution Linkages in North America: A Linear Analysis. Econ. Syst. Res. 2000, 13 (2), 197-208. (17) Muradian, R.; O’Connor, M.; Martinez-Alier, J. Embodied Pollution in Trade: Estimating the ‘Environmental Load Displacement’ of Industrialized Countries. Ecol. Econ. 2002, 41, 51-67. (18) Price, L.; Michaelis, L.; Worrell, E.; Khrushch, M. Sectoral trends and driving forces of global energy use and greenhouse gas emissions. Mitigation Adaptation Strat. Global Climate Change 1998, 3, 263-319. (19) Leontief, W. Environmental Repercussions and the Economic Structure: An Input-Output Approach. Rev. Econ. Stat. 1970, 52, 262-277. (20) Lave, L.; Cobas, E.; Hendrickson C.; McMichael, F. C. Using Input-Output Analysis to Estimate Economy-Wide Discharges. Environ. Sci. Technol. 1995, 29 (9), 420A-426A. (21) Hendrickson, C.; Lave, L.; Matthews, H. S. Environmental Life Cycle Assessment of Goods and Services: An Input-Output Approach; Resources for the Future: Washington DC, 2006. (22) Suh, S.; Lenzen, M.; Treloar, G. J.; Hondo, H.; Horvath, A.; Huppes, G.; Jolliet, O.; Klann, U.; Krewitt, W.; Moriguchi, Y.; Munksgaard, J.; Norris, G. System Boundary Selection in LifeCycle Inventories Using Hybrid Approaches. Environ. Sci. Technol. 2004, 38 (3), 657-664. (23) Suh, S. Functions, Commodities and Environmental Impacts in an Ecological-Economic Model. Ecol. Econ. 2004, 48 (4), 451467. (24) Peters, G.; Hertwich, E. Production Factors and Pollution Embodied in Trade: Theoretical Development; NTNU Industrial Ecology Programme working paper 5; Trondheim, Norway, 2004. (25) Munksgaard, J.; Wier, M.; Lenzen, M.; Dey, C. Using InputOutput Analysis to Measure the Environmental Pressure of Consumption at Different Spatial Levels. J. Ind. Ecol. 2005, 9 (1-2), 169-186. (26) Carnegie Mellon University, Green Design Institute. On-line EIO-LCA modeling tool; via http://www.eiolca.net; Accessed November 15-30, 2004. (27) Bjorn, A.; Declercq-Lopez, L.; Spatari, S.; MacLean, H. L. Decision Support for Sustainable Development Using a Canadian Economic Input-Output Life-Cycle Assessment Model. Can. J. Civil Eng. 2005, 32, 16-29. (28) U.S. Bureau of Economic Analysis, Industry Economic Accounts. 1997 Benchmark Summary Tables; via http://www.bea.doc.gov/ bea/dn2/i-o.htm#benchmark; Accessed October 10, 2004. (29) Statistics Canada, Input-Output Division. Inputs and outputs, by industry and commodity, L-level aggregation and North American Industry Classification System (NAICS) 1997; via http:// cansim2.statcan.ca/; Accessed September 1-15, 2004. (30) Final Energy Vector 2004. Compiled by Carnegie Mellon University Green Design Institute, Personal correspondence with Dr. Scott Matthews, August 2004. (31) Adjusted GWP Vector 2004. Compiled by Carnegie Mellon University Green Design Institute, Personal correspondence with Dr. Scott Matthews, August 2004. (32) Statistics Canada. Canadian System of Environmental and Resource Accounts, Material and Energy Flow Accounts,; Cat No. 16-505.-GPE; 2004. (33) U.S. Bureau of Economic Analysis, Industry Economic Accounts. Gross Output by Industry; via: http://www.bea.doc.gov/bea/ industry/gpotables/; Accessed October 15, 2004. (34) Organization for Economic Development and Co-operation. Purchasing Power Parities, Main Economic Indicators; via: http://www.oecd.org/std/ppp; Accessed June 12, 2004. (35) Chenery, H. B.; Clark, P. G. Interindustry Economics; Wiley: New York, 1959. (36) Hayami, H.; Nakamura, M. CO2 Emissions of an Alternative Technology and Bilateral Trade Between Japan and Canada: Relocating Production and an Implication for Joint Implementation; Proceedings of the 14th International Conference on Input-Output Techniques, Montreal, October 10-15, 2002. VOL. 41, NO. 5, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1531

(37) Wonnacott, R. J. Canadian-American Dependence; An InterIndustry Analysis of Production and Prices; North Holland: Amsterdam, 1961. (38) Carter, H. O.; Ireri, D. Linkage of California-Arizona Input-Output models to analyze water transfer patterns. In Applications of Input-Output Analysis; Carter, A. P., Brody, A., Eds.; NorthHolland Publishing Company: Amsterdam, 1970. (39) Ghanem, Z. Statistics Canada’s Input-Output Models: The Mathematical Framework; Industry Account Division, Statistics Canada, 2005. (40) Industry Canada. Trade Data On-Line; via: http://strategis.ic.gc.ca/sc_mrkti/tdst/engdoc/tr_homep.html; Accessed October 16, 2004 (41) U.S. Census Bureau, Foreign Trade Division. 1997 Import and Export Statistics by NAICS for Canada and World; obtained by special order; Washington, DC, November 4, 2004. (42) Organisation for Economic Co-operation and Development. Carbon Emission Leakages: a General Equilibrium View; Economic Department Working Papers no. 242; May 2000. (43) Energy Information Administration. United States Country Analysis Brief, January 2005; U.S. Department of Energy: Washington, DC; via: http://www.eia.doe.gov/emeu/cabs/ usa.html; Accessed February 10, 2005. (44) Energy Information Administration. Canada Country Analysis Brief, January 2005; U.S. Department of Energy: Washington,

1532

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 5, 2007

(45) (46)

(47)

(48) (49)

(50) (51)

DC; via: http://www.eia.doe.gov/emeu/cabs/canada.html; Accessed February 10, 2005. Natural Resources Canada. Energy in Canada 2000; Government of Canada, 2000. Rao, S.; Tang, J.; Wang, W. Measuring the Canada-U.S. Productivity Gap: Industry Dimensions. Int. Productivity Monit. 2004, 9, 3-14. Carnegie Mellon University, Green Design Institute. Revising the EIO-LCA Energy, Conventional Pollutants, and Global Warming Potential Vectors; via: http://www.eiolca.net/remakingenergy.pdf; Accessed November 4, 2004. Alberta Energy. Natural Gas Commodity Information, 2004; via: www.energy.gov.ab.ca; Accessed February 10, 2005. Pipeline 101. Overview of Natural Gas Pipelines, 2004; via www.pipeline101.com/Overview/natgas-pl.htm; Accessed February 10, 2005. Stanisic, John. IDC Market Research Consultants; Personal communication, October 12, 2004. Desrosiers Automotive Consultants. Global Sales and Production. Desrosiers Automotive Reports June 2002, 16 (12), 7-10.

Received for review January 15, 2006. Revised manuscript received September 28, 2006. Accepted November 30, 2006. ES060082C