ARTICLE IN PRESS

Energy Policy 33 (2005) 197208

Consumer lifestyle approach to US energy use and the related CO2 emissions
Shui Bina,*, Hadi Dowlatabadib
a

Center for Integrated Study of the Human Dimensions of Global Change, Department of Engineering and Public Policy, Carnegie Mellon University, Pittsburgh, PA 15213, USA b Sustainable Development Research Institute, University of British Columbia, 1924 West Mall, Vancouver, Canada BC V6T 1Z2

Abstract Historically, a sectoral approach (based on the industrial, transportation, commercial, and residential sectors) has shaped the way we frame and analyze issues of energy conservation and CO2 mitigation. This sectoral categorization, however, is limited in its capacity to reveal the total impacts of consumer activities on energy use and its related environmental impacts. In this paper, we propose an alternative paradigm, called the Consumer Lifestyle Approach (CLA), to explore the relationship between consumer activities and environmental impacts in the US. Estimates based on our methodology reveal that more than 80% of the energy used and the CO2 emitted in the US are a consequence of consumer demands and the economic activities to support these demands. Direct inuences due to consumer activities (home energy use and personal travel) are 4% of the US GDP, but account for 28% and 41% of US energy use and CO2 emissions, respectively. Indirect inuences (such as housing operations, transportation operations, food, and apparel) involve more than twice the direct energy use and CO2 emissions. Characterization of both direct and indirect energy use and emissions is critical to the design of more effective energy and CO2 emission policies. It may also help erode the false dichotomy of them versus us (industrial polluters versus consumers) references to the locus of responsibility for control of energy use and CO2 emissions. r 2003 Elsevier Ltd. All rights reserved.
Keywords: Household; Consumption; Embodied energy

1. Introduction Historically, a sector-based approach (i.e., industrial, transportation, commercial and residential) has shaped the way we frame and analyze issues of energy use and CO2 emissions. For example, in the US for 1997, the industrial sector was the most energy-intensive (38% of US total) and CO2-intensive (33%), while the residential sector ranked only fourth (11%) and third (19%), in energy use and CO2 emissions respectively (EIA, 2000b, c). In the sector-based approach, the residential sector is the only sector to directly reect consumer activities which merely refer to end-uses of home energy, such as space heating, water heating, and appliance use. Therefore, the sectoral approach is limited in its ability
*Corresponding author. National Center for Atmospheric Research, Advanced Study Program, P.O. Box 3000, Boulder, CO 80307, USA. Tel.: +1-303-497-2889; fax: +1-303-497-2889. E-mail addresses: shuibin@cmu.edu (S. Bin), hadi@sdri.ubc.ca (H. Dowlatabadi). 0301-4215/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0301-4215(03)00210-6

to mirror the total impacts of consumer activities on energy use, and the related environmental impacts. Furthermore, it may help propagate a false dichotomy among consumers of them versus us (industrial polluters versus consumers) references of responsibility about energy use and the related CO2 emissions. In the past two decades, the role of consumers and their consumption patterns have attracted increasing attention of and discussions among researchers. In the late 1980s, some researchers brought the concept of lifestyle into the study of personal energy consumption. As a consequence of this innovation, Lee Schipper et al. concluded that: about 4555% of total energy use is inuenced by consumers activities for personal transportation, personal services, and homes. They argued that signicant changes in energy demand will be driven by: the mix of personal activities and their locations ybesides energy prices and incomes (Schipper and Bartlett et al., 1989). In order to understand household energy consumption patterns, Vringer and Blok estimated energy requirement of households in the

ARTICLE IN PRESS
198 S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208

Netherlands (Vringer and Blok, 1995), Lenzen assessed the energy use and greenhouse gases attributable to consumer activities in Australia (Lenzen, 1998), Weber and Perrels quantied the impact of lifestyle factors on the 1990s and 2010s energy demand and related emissions in West Germany, France and the Netherlands (Weber and Perrels, 2000), Pachauri and Spreng studied energy requirement of households in India (Pachauri and Spreng, 2002), and Reinders et al. (2003) conducted a study to estimate direct and indirect energy requirement of households in 11 European countries. Building on these past studies on energy use and consumer activities, we propose a consumer-oriented integrated assessment framework for analysis of energy use and CO2 emissions, viz. the Consumer Lifestyle Approach (CLA). We also offer a transparent methodology linking various consumer activities to economic activity, energy use and CO2 emissions. Our efforts intend to shed light on the following questions: (1) How do we study consumer-related energy use and the related environmental impacts? (2) What is the role of consumer lifestyle choices in the US economy, energy use and CO2 emissions? (3) How much of the total US energy consumption and CO2 emissions is a product of consumer activities? (4) What are the most energy/ CO2-intensive consumer activities? In Section 2, we introduce the design and various components of the Consumer Lifestyle Approach, as well as the concept of direct and indirect inuences of consumer activities. Then, a CLA study, re-estimating US energy use and CO2 emissions, is presented. Its

estimation methodologies and the corresponding results are documented in Sections 3 and 4, respectively. The derived policy implications are discussed in Section 5. Conclusions are provided in the nal section.

2. The concept 2.1. Consumer lifestyle approach The term Consumer in CLA refers to the entity that purchases and uses products and services for the purpose of individual or household consumption. Lifestyle is a way of living that inuences and is reected by ones consumption behavior. The basic premise underlying consumer lifestyle research is that by understanding consumers we can design better public policies. Understanding consumers is made complicated by many interacting factors, many of which evolve over time in response to circumstances. The proposed CLA attempts to offer an interdisciplinary framework which explicitly acknowledges the multitude of interacting factors and groups these under ve major headings (see Fig. 1): (1) External environmental variables, such as cultural inuences and technology development, which form external context to a consumers decision process. (2) Individual determinants, such as attitudes and beliefs, which are psychological variables inuencing an individual consumers decision making. (3) Household characteristics, such as household size, housing type and size, income and location, which

Fig. 1. A framework for the proposed consumer lifestyle approach. Note that: (1) The 1. External Environment is partly from Fig. 1-1. A simplied decision process framework for studying Consumer Behavior (Loudon and Bitta, 1993, p. 22). The 2. Individual Determinants is partly from Fig. 1 Rationality for Economists (McFadden, 1999). The dashed line indicates feedback from Consequences to other CLA components.

ARTICLE IN PRESS
S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208 199

Fig. 2. Direct and indirect inuences. Notice that: (1) The subjects of Purchase component may be a product or a service; the subjects of Use component may involve direct energy inputs, services or products that do not require energy inputs; the subjects of Disposal component may be recycled, buried or incinerated. (2) About 28% of US energy is directly consumed by household activities (direct inuences), 57% of the total for household-related economy activities (indirect inuences), and about 15% is for non-household related consumption activities, such as national defense. The details of methodology and results can be referred to Sections 3 and 4. (3) Government and non-prot related activities are labeled as non-household here. Detailed information on their energy use and emissions is not part of this study, while its contribution is marked using double lines.

form household context for a individual consumers decision making. (4) Consumer choices, such as purchases and use of services and equipment. (5) Consequences, such as resource use and related environmental impacts, which are the results of consumer activities. The consumer choices studied here refer to consumer purchase and use behaviors, and the consequences studied here refer to the energy use and the related carbon dioxide emissions caused by consumer activities. In this paper, we employ various data sources and models to quantify the relationship between consumer choices and consequences at a macro-level.

2.2. Direct and indirect inuences Within the component of consumer choices, household and individual activities (purchase and use) have direct or indirect impacts on energy use and related CO2 emissions. If a consumers activity leads to energy consumption and CO2 emissions while the product or service is in use, these are called direct (on-site) inuences, where energy consumption and CO2 emissions occur in the preparation (production and delivery) of a product or service and before its use are called indirect (embodied) inuences.

Fig. 2 demonstrates the scopes of direct and indirect inuences as dened and estimated in this study.1 Notice that energy and emissions related to end-of-life processes are included if associated with economic activity (e.g., as input costs to recycled material). However, informal economy and government related direct and indirect impacts, i.e., composting, landll emissions, and government services are not yet characterized. In CLA, the terms of direct and indirect are dened from a consumers perspective. To distinguish a consumer activity causing to direct or indirect inuences is to see if the energy is being used and the related CO2 is emitted at the same time of using or before (or after) a consumer uses a product or service. For example, driving a car leads to direct inuences as gasoline is used and CO2 is emitted while a consumer operates the car. However, for this to happen, there are indirect impacts related to the manufacturing of cars, their maintenance and insurance, provision of a road infrastructure, and discovery, production, transport and sale of gasoline to consumers taking a ride in their car. In CLA, consumer activities are categorized into three levels. At the highest level, all consumer activities are categorized into two groups: direct inuences and indirect inuences. Direct inuences include home energy use and personal travel, while indirect inuences
1 For detailed methodologies and results, please refer to Sections 3 and 4.

ARTICLE IN PRESS
200 S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208 Table 1 Consumer consumption activities categorization Consumer activities categorization Direct inuences Home energy Space heating Air conditioning Water heating Refrigeration Other appliances and lighting Long distance by automobiles and trucks Long distance by air Long distance by others Short distance by automobiles and trucks Short distance by others Shelter, utilities, etc. Vehicle purchase (net), gasoline and motor oil, other vehicle expenses, etc. Food at home, food away from home Men and boys, women and girls, etc. Health insurance, drug, etc. Fees and admissions, magazines, etc. Personal insurance and pensions Education, tobacco, etc. Sources Residential Energy Consumption Survey (Department of Energy, 1999), or RECS

Personal travel

American Travel Survey (Department of Transportation, 1995), or ATS; Transportation Energy Data Book (Oak Ridge National Laboratory, 1999), or TEDB

Indirect inuences

Housing operations Transportation operation Food and beverage Apparel and services Health care Entertainment Personal Insurance Others

Consumer Expenditure Survey (Department of Commerce, 1999), or CES

include housing operation, transportation operation, food consumption, etc. Within these broad categories two nested levels of activity are dened with increasingly detailed information for each activity (data permitting). A complete list of consumer activities characterized using available data for the US is presented in Table 1.

consumer expenditures on housing operation, transportation operation, food, clothing, and recreation. 3.1. EnergyCO2 approach for direct inuences 3.1.1. Home energy Fuel consumption information at home is well documented in the Energy Information Administration (EIA)s RECS database. For example, RECS provides residential energy consumption data on ve end-uses (space heating, water heating, air conditioning, refrigeration, other appliances and lighting) by four fuel types (electricity, natural gas, fuel oil, and liquid petroleum gas) (EIA, 1999). The CO2 coefcients for each fuel type are also estimated and published by the EIA (EIA, 2000a). The formula used to estimate the total CO2 emissions from ve residential end-uses is presented below: T HOME CO2 XX Fuelm CO2 Coefficientm n HH ;
n m

3. Methodology The existing US data sources (listed in Table 1) do not provide ready information regarding CO2 emissions from consumer activities: (1) the Residential Energy Consumption Survey (RECS) provides information about home energy consumption; (2a) the American Travel Survey (ATS) reports on household travel patterns, while (2b) the Transportation Energy Data Book (TEDB) provides energy consumption in the transportation sector; and (3) the Consumer Expenditure Survey (CES) collects expenditure data characterizing buying habits. Two approaches are used here to estimate the total energy use and CO2 emissions from consumer activities. One uses actual and estimated fuel use and carbon content of these, the other uses economic data on consumer expenditures and an economic inputoutput matrix of energy use (and CO2 emissions) associated with the economic activity supporting that expenditure. The energyCO2 accounting framework is used for direct inuences while the expenditure based IO analysis is used to ll in the indirect inuences of

where T HOME CO2 (in tons of CO2 equivalent) is the annual CO2 emissions of home energy in US; n refers to the type of end-use at home, such as space heating and water heating; m refers to the fuel type used in each end use, such as electricity and natural gas; Fuelm (GJ) is the annual energy consumption by fuel per household; CO2 Coefficientm (metric ton/GJ) is the CO2 coefcient by fuel type; and HH is the number of households. HH 101:5 million households in 1997.

ARTICLE IN PRESS
S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208 201

3.1.2. Personal travel Although many people believe that energy consumption due to personal travel is a signicant fraction of the total US picture, few studies have estimated its magnitude. In this paper, the estimation of energy use from personal travel is reported. TEDB provides each vehicles (automobiles, motorcycles, trucks, buses, air, rail-passenger, and water-recreational) energy consumption by fuel type (gasoline, diesel fuel, LPG, jet fuel, and electricity) (Oak Ridge National Laboratory, 1999), while ATS offers information on the percentage of personal travel use with each vehicle type (Bureau of Transportation Statistics, 1997a, b). The integration of the two information sources produces estimations of energy consumption from personal travel. The formula for estimating energy use of personal travel by a transport mode for long and short distance travel is T PT ENERGYk XX Fuelij DRatioi VRatioij ;
i j

where T PT ENERGYk (EJ) is the annual energy use by all transport modes for long or short distance travel in US; k refers to the fraction of long or short distance travel; j refers to the fuel type, such as gasoline and diesel; i refers to the transport mode, such as automobile and bus; Fuel j (EJ) is the energy use of a vehicle mode i for personal travel using fuel j ; VRatioij is the ratio of personal use for a vehicle mode i using fuel j ; DRatioi is the fraction of long and short distance traveled by transport mode. For example, about 87% of energy use by automobile and truck is for short distance travel (less than 100 miles), and close to 100% of energy use by air is for long distance travel (large than 100 miles). CO2 emissions from personal travel are calculated as the product of the fuel consumed and its corresponding CO2 coefcient. 3.2. Using the EIO-LCA to estimate indirect inuences Inputoutput analysis has been widely recognized as a popular tool to estimate energy use, greenhouse gas emissions, pollutants embodied in consumer goods and services on a macro-scale. Similar analyses have been conducted in assessing the indirect energy requirements of households in Netherlands (Vringer and Blok, 1995), Australia (Lenzen, 1998), West Germany and France (Weber and Perrels, 2000), and India (Pachauri et al., 2002). The Environmental InputOutput Life Cycle Analysis (EIO-LCA) model developed at Carnegie Mellon University is a suitable and convenient tool to estimate energy use and related greenhouse gas emissions embodied in US domestic goods and services. It has adjunct tables that represent various environmental

efuents from each industry and economic activity associated with dollar value of output (Green Design Initiative, 2000; Henderickson and Horvath, et al., 1998). In this study, an average annual households consumer expenditure is used as inputs to the EIO-LCA model. The Bureau of Labor Statistics provides detailed information on the consumer expenditure survey recounting expenditures under 70 different categories (Bureau of Labor Statistics, 1999). Since these categories are not a one-to-one match to the 485 commodity categories in the EIO-LCA, we examined each of the consumer expenditure categories and developed a mapping from these to appropriate commodity categories dened by the Department of Commerce. This mapping function translated the 70 consumption categories into 62 commodity outputs, allowing us to use the EIO-LCA to estimate indirect energy use from consumer expenditures.2 The full list of indirect inuences of consumer activities is presented in Appendix A. The formula for estimation of energy consumption and CO2 emissions for indirect inuences is X CPI 1992j Xi CES 1997j Tji ; CPI 1997j j X EIi Xi HH ; T IND Energy X T IND CO2 CIi Xi HH ;
i i

where i refers to an industrial product or a service category in EIO-LCA model; j refers to an expenditure category in the Consumer Expenditure Survey; Xi ($) is a households expenditure on products or services i in the EIO-LCA; CES 1997j ($) is the consumer expenditure on product j in 1997; CPI 1992j and CPI 1997j are the consumer price indices in 1992 and 1997 for a product or service j ; Tji is the transformation matrix of consumer expenditure category j to an industrial product or a service category i; T IND Energy (EJ) is annual energy consumption caused by indirect inuences; EIi (thousand J/$) is the energy intensity of industrial output i; T IND CO2 (million metric ton of CO2 equivalent) is annual CO2 emissions caused by indirect inuences; and CIi (metric ton/$) is CO2 intensity of industrial output i: An example of the estimation of annual indirect inuences from food consumption per household is illustrated in Table 2. We dissect each recorded element of food expenditures in the CES and nd an equivalent class of economic activities in the EIO-LCA. Then by correcting for the base-year of expenditure measures we can estimate indirect energy use and CO2 emissions
2 The base year for the EIO-LCA when our study was conducted is 1992.

202

Table 2 Estimation of indirect inuences of food consumption for a US household in 1997

Consumer expenditures in 1997 Items 1997 price Consumer price indexes 1992 CPI 1997 CPI CPI items 1992 price (chained) EIO-LCA model EIOLCA sector name Energy coefcient (GJ/$) CO2 coefcient (kg/$) Estimation results Energy consumption (GJ) 60 41 5 2 3 12 4 2 2 3 1 CO2 emissions (kg) 3953 2675 303 129 174 773 231 137 108 178 81

Food Food at home Cereals and bakery products Cereals and cereal products Bakery products Meats, poultry, sh, and eggs Beef Pork Other meats Poultry Fish and seafood

4801 2880 453 161 292 743 224 157 96 145 89

S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208

153.3 150.4

169.5 181.1

146 243

Cereals breakfast foods Bread, cake, and related products Meat animal Meat animal Miscellaneous livestock Poultry and eggs Prepared fresh or frozen sh and seafoods Poultry and eggs Dairy farm products Fruits Vegetables Dehydrated fruits, vegetables, and soups Dehydrated fruits, vegetables, and soups Bottled and canned soft drinks Eating and drinking places Wines, brandy and brandy, spirits

0.0140 0.0110

0.884 0.718

ARTICLE IN PRESS

132.3 127.8 130.7 131.4 151.7

136.8 155.9 144.4 156.6 177.1

Meats

217 129 87 122 76

0.0170 0.0170 0.0190 0.0229 0.0170

1.065 1.065 1.244 1.466 1.069

Eggs Dairy products Fruits and vegetables Fresh fruits Fresh vegetables Processed fruits

33 314 476 150 143 102

108.3 128.5 184.2 157.9 137.7

140 145.5 236.3 194.6 148.8

26 277 117 116 94

0.0229 0.0219 0.0249 0.0100 0.0160

1.466 1.412 1.563 0.634 1.05

1 6 7 3 1 2

37 392 429 183 74 99

Processed vegetables

80

128.8

147.2

70

0.0160

1.05

74

Other food at home Food away from home Alcoholic beverages Total

895 1921 309 5110

136.8 138.9 147.3

158.1 154.7 162.8

Food at home Dinner

774 1725 280

0.0150 0.0110 0.0110

1.005 0.741 0.741

12 19 3 63

778 1278 207 4160

ARTICLE IN PRESS
S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208 203

associated with each element of food expenditures. Since the use of inputoutput model, the energy use and CO2 emissions resulting from manufacture, transportation, and marketing have been allocated into the estimated indirect inuences of food consumption.

Table 3 US Energy consumption and CO2 emissions by consumer activities, 1997 Consumer activities Direct inuences Home energy Space heating Other appliances and lighting Water heating Refrigeration Air conditioning Personal travel Short distance by automobiles and trucks Long distance by air planes Long distance by automobiles and trucks Long distance by others Short distance by others Indirect inuences Housing operation Transportation operation Food and beverages Apparel and services Others Personal insurance and pensions Entertainment Health care Total direct and indirect Energy use (EJ) 28.3 10.9 5.5 2.4 2.0 0.5 0.4 17.4 12.7 2.2 1.9 0.3 0.3 56.4 25.6 17.4 6.4 2.6 1.6 1.1 1.1 0.5 84.6 CO2 emissions (Mt CO2) 2230 1046 338 381 157 89 81 1184 855 149 130 25 25 3289 1411 954 426 167 117 81 76 56 5519

4. Results and discussions 4.1. The national picture A summary of results from our analyses of US energy use and CO2 emissions is presented in Table 3. Direct energy use (28.3 EJ) is only half as large as indirect inuences of consumer expenditures (56.4 EJ), and the related CO2 emissions due to direct inuences (2230 Mt CO2-e) are only 70% of indirect inuences (3289 Mt CO2-e). The rank of energy- and CO2-intensive consumer activities is depicted in Figs. 3 and 4, respectively. Housing operation (indirect inuences) is the most energy- and CO2-intensive among all consumer activities (including direct and indirect inuences), which is mainly contributed by the consumption of utilities (about 60% of housing operation). For direct inuences, personal travel is the most energy- and CO2-intensive (17.4 EJ and 1184 Mt CO2-e), much of which is caused by short distance travel by automobiles and trucks (about 72%). Comparing Figs. 3 and 4, a reader may nd that the direct energy use of personal travel is about 50% higher than home energy, while the related direct CO2 emissions of personal travel are only 13% higher than home energy. The percentage differences results from the large portion of electricity consumed in homes. About two-third of fossil energy used to produce electricity is lost in production, transmission and distribution. This leads to electricity being about three times as carbon intensive as the primary energy source used in its production. Since US has 101.5 million households and there are 2.64 individuals in average household in 1997, a reader can get calculation results about annual energy use and CO2 emissions per capita and a household from the information provided in Table 3. 4.2. Consumer activities, the economy, trade, energy use, and CO2 emissions What are the roles of consumer activities in US economy, energy use and CO2 emissions? This section is devoted to mapping the ndings from the CLA reestimation study onto nationally aggregated measures of economic activity, energy use, and CO2 emissions. Table 4 documents share of consumer expenditures in

GDP, and ratio of energy use and CO2 emissions by consumers to the US total. The rst row of Table 4 shows the US GDP, and shares of that GDP directly represented by expenditures on home energy and personal travel in the rst three columns. In the fourth column the indirect inuence of other consumer expenditures on GDP is represented. We observe that while the share of direct energy expenditures by households is only 4% of the GDP, the shares due to all other consumer expenditures (related to indirect inuences) is 10 times higher at 42%. When we look at energy use we see that approximately 11% is accounted for by consumption for home energy and 18% for personal travel. The indirect impact of consumer choices accounts for a further 57% share of the energy used in the US. Thus, consumer lifestyle decisions account for 85% of all energy use in the US when direct and indirect usages for home energy, personal travel and other expenditures are combined. We should add at this point that while the InputOutput models are based on balanced national accounts, the augmented matrices used to estimate energy use and emissions from such activities are not corrected for embodied energy or pollution in exported and imported goods and services. Therefore, the values reproduced in

ARTICLE IN PRESS
204 S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208

Fig. 3. Rank of energy use by US consumer activities, 1997.

Fig. 4. Rank of CO2 emissions by US consumer activities, 1997.

Table 4 reect the fraction of energy used assuming imported goods and services similarly classied in the Consumer Expenditure Survey have the same energy and CO2 intensity as their domestically produced alternatives. In the nal section of Table 4, we present estimates of the share of consumer activities in total US CO2 emissions. These gures show the relatively high carbon intensity of electricityused in homesleading to a relatively higher share for home energy in CO2 emissions than in energy use. In the nal column we also nd the paradox of total emissions from consumer expenditures accounting for 102% of US CO2 emissions. This is explained by the high emissions intensity of the products imported into the US (using US emissions characteristics). The calculated embodied carbon in these imports (as if they had been made in the US)

is not part of the US domestic emissions accounts and hence the nding of household consumption accounting for 45% of GDP, 85% of energy use, and 102% of CO2 emissions. These gures not only highlight the true magnitude of energy use and emissions by US consumers, but also reveal a large leakage of CO2 through imports of goods serving the needs of US consumers.

5. Policy implications 5.1. Direct and indirect inuences The ndings from our re-estimation study demonstrate that the indirect inuences of consumer choices represent a larger share of US energy use and CO2

ARTICLE IN PRESS
S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208 Table 4 The role of household consumption activities in the US picture
Conventional Indicators for U.S. Economy, Energy Consumption, and CO2 Emissions U.S. GDP - 8111 billion U.S. current dollar (Total - 100%) personal consumption (68% of GDP) gross private domestic investment (15%) net exports of goods and services (-1%) government consumption expenditures and gross investment (18%) U.S. Energy Consumption - 99 EJ (Total 100%) residential (11% of Total) transportation (26%) commerical (25%) industrial (38%) U.S. CO2 Emissions - 5426 MMTG (Total - 100%) residential (19%) transportation (32%) commerical (25%) industrial (33%) N.A. 19% 100% 22% N.A. 68% N.A. N.A. 41% N.A. N.A. N.A. N.A. Consumer Consumption Activities (CLA Approach) Direct Influences Indirect Total (direct Home Personal Subtotal Influences + indirect) Energy Travel 2% 3% 2% 3% 4% 5% 42% 62% <100% <100% < 74% 11% 100% 18% N.A. 66% N.A. N.A. 28% 57% N.A. <34% <100% <100% 60% N.A. <32% <100% <100% N.A 102% N.A 85% N.A. 45% 67%

205

Note that: (1) Personal consumption is conceptually different to household consumption mentioned in this study. Personal consumption expenditure is collected by the Bureau of Economic Analysis at the US Department of Commerce from US manufacturers. It covers not only household consumption, but also non-household consumption such as consumption from non-governmental organizations. (2) The darkest shaded areas with N.A. indicate an absence of information from which to make calculations, the lightly shaded areas specify a low degree of uncertainty in the presented estimates, and the italic gures show the upper bound of our best estimates. (3) The dark area with white italic gures (in the rows of US energy consumption and US CO2 emissions) indicate the listed ratio should be interpreted as a reference point because actual comparisons are inappropriate for conceptual reasons. These conceptual challenges arise from the fact that US total energy use and emissions data do not account for imports of manufactured goods and services. The consumer expenditure data however includes imported products and energy used to produce these is calibrated to manufacturing patterns in the US, rather than the location of origin.

emissions than the direct inuences that are so often the focus of public policy discussions. This is a far cry from the traditional sector-based presentations that attribute so much of the energy use and emissions under headings that appear to be outside the control of households. Direct inuences, accounting for 28% and 41% of the total US energy consumption and CO2 emissions respectively, are caused by consumers use and purchase behaviors. The estimation of direct inuences sets the upper bound for possible changes of use behavior that may be partly achieved by regulation, persuasion, or market interventions. By combining these gures with an assessment of the possibility of bringing about such changes, we can estimate payoff from investment in such interventions (as compared to technology-based initiatives). Indirect inuences, twice as large as direct inuences, are related to consumers purchase behavior. The reduction of energy use and CO2 emissions for indirect inuences would be mainly achieved by green design (such as employing efcient technologies and processes) in the supply chain. So, the related policies can be aimed at making production processes more

efcient, and making the durable consumer goods more efcient. The potential for the latter to reduce energy use and CO2 emissions is, on average, roughly half the former. Therefore, policymakers need to avoid two obvious pitfalls: (1) Mandating early capital stock turnover, in pursuit of use phase efciency, may lead to needless energy use and CO2 emissions in the manufacturing phase of a products lifecycle. (2) The ip side of the coin is mandating more robust technologies that last long after they could be replaced by better technologies offering combined energy and emissions savings in both use and manufacturing phases. One of the main purposes of this paper is to demonstrate that polices of energy conservation and carbon mitigation should aim at the aggregate of direct and indirect energy use and CO2 emissions. We use the following two examples to illustrate that missing the integration of both inuences in policy design may lead to slow or even opposite direction of progress. Example 1, in US, the production of an average vehicle is associated with GHG emissions of 12.25 t of CO2 equivalent. Based on a service life of 14 yr, this

ARTICLE IN PRESS
206 S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208

indirect emission is equivalent to 0.875 t/yr. When considering a policy to reduce emissions (direct inuences) from private transportation by scrapping inefcient cars, a policy maker should not ignore the indirect emissions associated with manufacturing a car. Early retirement of a car in order to reduce in-use CO2 emissions will therefore be only advisable if the savings from in-service emissions exceeds 0.875 t CO2/yr. Therefore, it is easy to justify the early retirement of a car achieving only 12 L/100 km with one that is 25% ! Standard more efcient and achieves the US CAFE (9 L/100 km). However, the next 25% change in efciency from 9 to 6.75 L/100 km is only just above the pro-rated GHG emissions embodied in the vehicle.3 However, one thing we need to address here is that building cars of greater longevity is a double-edged sword that can reduce pro-rated indirect emissions but may lead to a protracted lock-in to a less efcient capital. Example 2, the end-use efciency of electric appliances is often higher than gas-red appliances which perform the same functions (i.e. while electric water heaters are over 90% efcient, gas-red units are rarely better than 80%). If a policy maker focuses only on the direct inuences, he may be in favor of promoting electric units while their indirect energy use and CO2 emissions are more than three times higher than gas-red units. It is suggested that if policies are aimed at direct impacts alone, there is a high likelihood of signicant unintended negative consequencesthe indirect effects are twice as large and often go unassessed. We also suggest that it is important for policy makers to recognize that the portfolio of capital stock represented by the indirect inuences of consumer expenditures has many different lifetimes and that policies aimed at changing these will have to be pursued for corresponding periods if the policy is to bear fruit. For example, we could aim for more efcient housing structures by enforcing more stringent building standards. But it is important to realize that the payoff to such measures will take many decades to be realized fully. Moreover, policy makers should be aware of the dynamics of responses by consumers in their choices, and by the economy in how these are met, needs to be kept in focus when designing policy. These dynamic factors will determine the pace at which policy objectives can be met and at what cost. From a consumer perspective, the concepts of direct and indirect inuences would further stipulate learning a consumers his/her own use and purchase patterns, both of them incurring direct and indirect inuences. It may
These calculations are based on US averages where the cost of a new car in 2001 was $20,000, it was expected to be driven 12,000 miles per year and in service for 14 yr.
3

help them be aware that their increasing travel mileage (both direct and indirect inuences) and quick switching to a new model of car (indirect inuences) because of fashion change will put them on a hot seat labeled high carbon emitter. 5.2. Target activities Identifying target activities (the most energy-intensive or/and the most carbon-intensive consumer activities) is one of important outputs from application of CLA methodology. This identication helps policy makers and researchers employ scarce resources in the most efcient way. Our ndings (see Table 3) suggest that nancial investment, technology development, capability building and research focus should be addressed to services and products related to home (home energy use and housing operation) and personal transportation (personal travel and transportation operation) for energy conservation and carbon mitigation. The reduction potential of energy use and carbon emissions from food and clothing consumption is too limited to be pursued. This identication also helps lay people become aware of the level of impacts associated with each of their consumption activities. Most lay people, even when motivated to become more environmentally responsible, are not equipped with such knowledge. Some think that turning off lights, recycling cans, and becoming vegetarians are the most signicant behavioral choices they can make. Although these behaviors often reduce energy use and CO2 emissions, their overall effect is minimal. Meanwhile, unwitting persistence in other activities, such as frequent air-travel and using electricity as a source for heat can lead to signicant primary energy use and CO2 emissions. We suggest that well-designed and credible information dissemination programs, based on the outputs of consumer-impact studies, especially containing the information relating to their daily life and the corresponding environmental impacts, will more effectively engage the public in the issues of energy conservation and carbon mitigation.

6. Conclusions In this paper, we propose an interdisciplinary framework to study consumer-related issues of energy use and carbon dioxide emissions. We apply this approach in an estimation of the relationship between US consumers direct and indirect energy use and CO2 emissions. Based on the ndings presented in this paper, we conclude that studying the specics of consumer choices and activities is of critical importance to devising

ARTICLE IN PRESS
S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208 207

successful policies for energy conservation and CO2 mitigation. We elaborated the importance of addressing the aggregate of direct and indirect energy use and carbon emissions in the policy design, and further note that the relevant policies should aim to reduce both direct and indirect energy uses and emissions. We also highlighted the issue of leakage due to trade. The US is perhaps an extreme example of this phenomenon, but our calculations show how consumer expenditures while only 45% of GDP and 85% of the energy use account for more than 100% of the CO2 emissions reported in the US emissions inventory. Clearly, there is a need to develop CO2 inventories using a concept similar to GNP and GDP to differentiate between within border production and within border consumption. Moreover, we suggested the output of consumer impact studies could be a good resource for information dissemination programs. We hope that CLA may be a useful addition to traditional methods employed in the study of energyenvironment problems and policy design. Whether this tool, through being designed around specic and familiar household activities, is more useful to consumers remains to be studied. Our hope is that CLA approaches will not only help policy makers design more effective policies but enhance public awareness and engagement in these issues. Acknowledgements We are very grateful to Baruch Fischhoff for his valuable comments and inputs. In addition, we thank the following people whose questions, comments and information helped shape our research: Mike Decay, Lester Lave, Granger Morgan (at Carnegie Mellon University), Bill Keil (Bureau of Labor Statistics), Robert Lotta (Energy Information Administration). This research was made possible through support from the Center for Integrated Study of the Human Dimensions of Global Change. This Center has been created through a cooperative agreement between Carnegie Mellon University and the National Science Foundation (SBR-9521914). All views expressed herein are those of the authors and all remaining errors are a reection of their fallibility. Appendix A Consumer activities categorization of indirect inuences Food and beverage Food at home (cereals and bakery products/meats, poultry, sh, and eggs/dairy products/fruits and vegetables/non-alcoholic beverages)

Food away from home Alcoholic beverages Housing operation Shelter Utilities, fuels, and public services (electricity/natural gas/fuel oil and other fuels/telephone, and water and other public services) Household operations Housekeeping supplies Household furnishings and equipment (household textiles/furniture/oor coverings/major appliances/small appliances, misc. houseware,/miscellaneous household equipment) Apparel and services (men and boys/women and girls/ children under 2 years old/footwear/other apparel products and services) Transportation operation Vehicle purchases (net outlay) (cars and trucks, new/cars and trucks, used) Gasoline and motor oil Other vehicle expense (vehicle nance charges/maintenance and repair/vehicle insurance/rent, lease, licenses, other/public transportation) Health care (health insurance/medical services/drugs and medical supplies) Entertainment (fees and admissions/television, radios, and sound equipment/other equipment and services (pets, toys, playground equipment, sports, exercise, etc.)/reading) Personal insurance and pensions Others (miscellaneous/tobacco/education/personal care products and services/cash contribution). References
Bureau of Labor Statistics, 2001. Consumer Expenditure in 1999. http://www.bls.gov/cex/csxann99.pdf. Bureau of Transportation Statistics, 1997a. 1995 American Travel Survey-Prole. Bureau of Transportation Statistics, US Department of Transportation, 15pp. Bureau of Transportation Statistics, 1997b. Long-distance leisure travel in the United Statestravel prole for American Travel Survey. Bureau of Transportation Statistics, US Department of Transportation, 19pp. Chris H., Horvath, A., et al., 1998. Economic inputoutput models for environmental life-cycle assessment. Environmental Science and Technology 32(7) 184A191A. EIA, 1999. A look at residential energy consumption in 1997. Energy Information Agency, Department of Energy. EIA, 2000a. Full combustion carbon coefcients. Energy Information Administration, Department of Energy (Table B1). http:// www.eia.doe.gov/oiaf/1605/ggrpt/tblb1.html.

ARTICLE IN PRESS
208 S. Bin, H. Dowlatabadi / Energy Policy 33 (2005) 197208 Oak Ridge National Laboratory, 1999. Transportation Energy Data Book: Edition 19, Center for Transportation Analysis Energy Division, Oak Ridge National Laboratory, 300pp. Pachauri, S., Spreng, D., 2002. Direct and indirect energy requirements of household in India. Energy Policy 30, 511523. Reinders, A.H.M.E., Vringer, K., et al., 2003. The direct and indirect energy requirement of households in the European Union. Energy Policy 31, 139153. Schipper, L., Bartlett, S., et al., 1989. Linking life-styles and energy use: a matter of time? Annual Review of Energy 14, 271320. Vringer, K., Blok, K., 1995. The direct and indirect energy requirements of households in the Netherlands. Energy Policy 23 (10), 893905. Weber, C., Perrels, A., 2000. Modeling lifestyle effects on energy demand and related emissions. Energy Policy 28, 549566. EIA, 2000b. Energy consumption by sector, 19492000. Annual Energy Review, http://www.eia.doe.gov/emeu/aer/pdf/pages/sec2 4.pdf. EIA, 2000c. Carbon dioxide emissions from energy consumption by sector, 19801999. Annual Energy Review, http://www.eia.doe.gov/emeu/aer/pdf/pages/sec12 5.pdf. Green Design Initiative, 2000. Environmental I/O lifecycle analysis, green design initiative. Carnegie Mellon University. http://www. eiolca.net. Lenzen, M., 1998. Primary energy and greenhouse gases embodied in Australian nal consumption: an inputoutput analysis. Energy Policy 26 (6), 495506. Loudon, D.L., Bitta, A.J.D., 1993. Consumer Behavior. McGraw-Hill, Inc., New York. McFadden, D., 1999. Rationality for economists? Journal of Risk and Uncertainty 19, 73105.