White Paper: Energy Use in Products and Services

Jan 12, 2012 |

This white paper was developed for GHSP Inc., to support the PBS series:

EARTH: THE OPERATORS' MANUAL

by Martha Amram

------

The energy consumed in goods and services is virtually invisible to consumers but constitutes more than half of the U.S.'s annual energy use.  Our homes and closets are filled with things;  we purchase items at the drugstore or mall, but seldom do we consider the energy used to make them. In addition, if someone asked us to purchase the items that had less energy used in their manufacture, we would not know which these are.

Despite its invisibility, energy use in goods and services is the largest category of personal energy consumption.  Goods and services include all manufactured items, imports, and services such as insurance, banking and education.  While food is considered a manufactured product in national income accounts, and thus part of goods, it is analyzed separately for the personalized energy calculator on the WattzOn website.  Similarly, the personal travel elements of the transportation sector are omitted from goods and services, but the other sector elements, such as commercial driving, are included in the goods and services category.

This white paper describes how the energy content of goods and services is calculated. It also discusses the challenges of creating product-level estimates of energy used for goods and services using a series of examples. The examples highlight the role for simple but effective energy-saving actions, such as recycling and buying less.

Because product-level comparisons cannot be made, energy use from goods and services feels "indirect" to the consumer; it is as if this energy consumption is beyond personal control. In contrast, one can choose to drive less, a "direct" decision to reduce energy use.[1]  For this reason, personal energy use from goods and services is not prominently featured on the WattzOn website, whose focus is on actionable recommendations.

This white paper proceeds as follows.  First, there is a short overview of the calculations.  It is followed by a short summary of the difference in energy intensity for goods and services, demonstrating that energy savings opportunities are in the product arena, despite the fact that we spend far more on services.  A third section presents several examples of product-level estimates that illustrate the unreliable nature of this type of calculation, due to lack of data and an immature methodology.  The final section provides supplementary details on the calculations used in the personalized energy calculator.

A SHORT OVERVIEW OF CALCULATIONS

This whitepaper is part of a series that describes the calculations behind a personalized energy calculator to measures one's energy footprint. The emphasis is on the word personalized, as individual choices create huge variation in energy use by individuals or households.  A second reason for a personalized calculator is the desire to demonstrate to consumers that there are product choices that use less energy so that a consumer can make purchasing decisions accordingly.

Unfortunately, the data do not support the personalization of consumption for goods and services or the product-level estimates needed to compare items within a product category.  Instead, an estimate is made of the average per capita energy use - a single number that is applied to every individual.

The calculation method is simple: Begin with an economy-wide estimate of energy use.  Remove energy use that has been accounted for elsewhere in the personalized energy calculator (flying, driving, household utility bills, and food). Add imports, assuming that energy use per dollar of import is the same as energy use per dollar of U.S. production.  Divide the adjusted total by population.  The result is energy use per capita, per year for goods and services. Our estimate is that energy use in goods and services is 224 million BTUs per capita per year.

ENERGY USE IN SERVICES

The label "goods and services" may be unfamiliar to many readers, and it is worthwhile to lay out some of the basic facts about these sectors.  The goods category consists of manufactured products, and as Table 1 shows, these are relatively high in energy use. Services include hotels, insurance, banking and so on.

Table 1 suggests that there may be more energy saving opportunities in goods/manufacturing than exist in services.  Table 2 shows this same notion a bit differently, presenting the energy intensity index for seven industries, three in the service sector and four in manufacturing.  The energy intensity index measures energy used per dollar of output. Energy intensity indices are calculated by the U.S. Department of Energy to track energy efficiency by industry, as a change in the index reflects relatively more or less energy use in production.[2] 

The industries shown in Table 2 illustrate relatively high energy-using industries, focusing on areas in which consumption choices may have a large impact on energy use.

The data show little variation in the energy intensity of the industries in the service sector but considerable variation in manufacturing. Services constitute a much larger share of consumer spending than do manufactured products, yet Table 2 suggests that there is much less energy saving opportunity in buying from a "green" insurance company than there would be in buying a "green" paper or paint product.

EXAMPLES OF ENERGY USE IN GOODS AND SERVICES

The lack of personalization in our estimate of energy used in goods and services is bothersome. The energy used by these sectors is large, and consumption can be expected to vary across individuals. Furthermore, the purpose of the personalized energy calculations is to provide recommendations to save energy, and it is difficult to link the indirect energy use from goods and services to anything other than general advice.

Because these results are rather unsatisfying, this section of the paper explores product-level calculations of energy use. Recent examples are presented and illustrate the challenges of creating a product-level estimate.

FAT TIRE AMBER ALE

The first example is beer, as this particular good illustrates a key challenge in calculating energy use or CO2e emissions by product: What are the boundaries of the analysis? In 2008, New Belgium Beer commissioned a study of the CO2e emissions created in the manufacturing of one of its products, Fat Tire Amber Ale. [3]

A key conclusion of the study is that Fat Tire's direct manufacturing emissions make up only 5% of the total.  Forty-eight percent of emissions are incurred indirectly, from inputs, and 47% of emissions are downstream in transportation, wholesale centers, and retail outlets. In particular, retail was defined to include refrigeration in retail stores, at 28% of emissions. (Other energy use in the store, such as lighting, was not included.)  This scope of the analysis is a bit odd, as typically, energy use in a store is included in the retail sector data, not in manufacturing. The Fat Tire example highlights how product-level results will always vary depending on the boundaries of analysis.

When the retail outlet component is removed, the Fat Tire Amber Ale CO2e results can be compared to those obtained for the brewery-manufacturing sector from the EIO-LCA model.[4]   This comparison is shown in Table 3.  What is striking is that the proportions of energy use and CO2e emissions identified are not similar even across the two data sources.  Again, this can be attributed to the boundaries of analysis problem. The Fat Tire analysis includes CO2e from waste and excludes much of the infrastructure investments (such as the manufacturing plant).  The lack of similarity in results, therefore, is not surprising;  Hendrickson et al. (2006) note how difficult it is to define the boundaries of analysis.

LIFECYCLE ANALYSIS OF FRIDGES AND AUTOS

The typical consumer does not buy a new car every year.  As with many products, a car is purchased only occasionally.  Additionally, while it is relatively easy to calculate the energy use or emissions per gallon of gasoline, what are the emissions created from the vehicle itself?  Are some cars better in this regard than others?

Hendrikson et al. (2006) use the EIO-LCA model to create a lifecycle model of energy use for a mid-sized passenger car.  They find that a whopping 85% of energy used is in driving.  Vehicle manufacture is only 7% of energy used. The remaining energy use is in the form of auto repairs over the car's life and the energy use of the consumer auto insurance industry.  The results are similar for CO2e emissions, as the majority of the energy use is from gasoline.  This breakdown between energy use/emissions in operations and manufacturing is similar to that found in many other studies.

This finding suggests that rather than worry about energy used within the vehicle manufacturing process, a consumer's best, most effective choice to reduce energy is to buy a car with high miles per gallon (mpg). Going from 20 mpg to 30 mpg cuts lifetime gasoline use by a third. No choice about vehicle model can deliver a similar-sized energy saving result.

Refrigerators are another purchase for which an energy-reducing consumer might want to balance annual energy consumption against the energy used to make and dispose of the product. In this case, the energy from using a refrigerator over its life is smaller than that used to make the item.  Horie (2004) models the consumer's purchase strategy and finds that a replacement cycle of every eight years or so minimizes energy use, even after accounting for energy used in production. He also finds, however, that this strategy does not minimize costs, and is, in fact, 20 - 30% more expensive.

The reason? Refrigerator technology improved so dramatically during the period of Horie's study, that one could save energy by buying a new refrigerator before the previous one wore out. The energy savings are present, but not financially attractive.

The comparison of purchase decisions for autos and refrigerators shows that energy savings do not always translate into monetary savings, making purchasing decisions a bit more complex. It is only after undertaking a detailed category analysis, that rather simple recommendations can be made: To save energy and money, buy a high-mpg car. To save energy and money, wait until your refrigerator wears out, and then buy an energy-efficient model.

CARBON EMISSIONS FROM COCA-COLA PRODUCTS

The British arm of Coca-Cola commissioned a study on the CO2e emissions created from the production of its most popular drinks.[5]  Table 4 reports the results.

The table shows that packaging is the largest contributor to emissions, as there is little variation across the drinks within a specific packaging type, but enormous variation across the packaging types.  The data suggest that an emissions-reducing consumer should select products packaged in aluminum cans. The company also notes that it has reduced packaging in its products by 20% since 2007.

It may seem surprising to discover that aluminum is the "greenest" packaging, as the aluminum industry is known for high-energy use. For example, in the U.S., nearly 1% of the energy used in the U.S. economy is in aluminum manufacturing.[6] However, energy use for recycled aluminum inputs consists of only 5% of that of newly manufactured aluminum.  Currently 55% of aluminum cans are recycled in the U.S., and aluminum cans now contain more than 50%-recycled input.[7]

As Coca Cola notes in its comments about these results, responsible recycling is one of the most effective actions a consumer can take to reduce emissions (and also energy), and it would appear that the company's results demonstrate that with high levels of recycling, even an energy-intensive product can provide the best choice.

CLOTHING AND SHOES

Two ambitious companies, Patagonia and Timberland, have developed product-level energy and emissions data.  Patagonia attempts to capture the energy used to manufacture and transport products from around the world to a central location.[8] Timberland does the same.[9] In the details of their methodologies, a difference arises: Patagonia does not include the energy content of materials used, such as the petroleum feedstock in polyester or the animal emissions for wool or goose down, while Timberland does in fact include this indirect energy use.

The results? Patagonia reports that 12 kWh are needed to make a t-shirt, but only 9 kWh to make a goose down sweater. Thirty-five (35) kWh are needed to make a dress. CO2e emissions are largely proportional to energy use, as the energy content of animal products is excluded. Timberland, however, finds that the CO2e emissions of shoes range from 22 to 220 pounds. Leather is the key driver of higher emissions. (The Timberland carbon accounting model attributes 7% of a cow's emissions to leather.) And Timberland found that only 5% of the emissions came from transport.

In both cases, the companies went to considerable effort and expense to provide useful information to consumers and to their staffs. However, the methodology differs on such a key, critical point-whether animal emissions are included-that the product-level results are company-specific and cannot be compared to one other. Consider the goose down jacket offered by Patagonia. Under the Timberland methodology, this would be the most carbon-intense product, but under the Patagonia methodology, this jacket uses less energy and emits less CO2e than even a t-shirt!

SYSTEMATIC PRODUCT-LEVEL ANALYSIS

The examples above illustrate some of the challenges in preparing a product-level analysis of energy and CO2e emissions. These examples also illustrate the confusion around methods, results, and sensible consumer actions.

Researchers at UC Berkeley have attempted to undertake a more systematic estimate of product-level CO2e emissions. Jones and Kammen (2008) mix product/process detail and results from the EIO-LCA model. They deconstruct a list of products into elements of manufacture, transportation, wholesale and retail, and then apply industry-specific emissions data by supply-chain stage, creating CO2e emissions estimates for 1000 products. This method produces emissions estimates for goods and services that are lower than the EIO-LCA estimates for manufactured goods by 30% and produces estimates for services that are higher than the EIO-LCA emissions estimate by 15 percent.

Jones and Kammen (2011) extend this work and find that 44% of emissions per person are from goods and services. However, their calculations lead to an estimate of total emissions, 48 metric tons per year, which is significantly higher than other reference data and which put the figure at about 20 metric tons per year. This makes it difficult to interpret their result for goods and services.

Jones and Kammen (2011) also estimate CO2e emissions from goods and services by income level. They use the Consumer Expenditure Survey and assume that emissions scale linearly with expenditures. In Jones and Kammen (2008), the authors note that the Consumer Expenditure Survey underestimates consumer expenditures by 30%. As such, in sum, it seems that their calculations could result in a significant underestimate of the percent of emissions from goods and services. Their work is pioneering, and the confusion of results simply illustrates the topic's challenges.

SUMMARY

In sum, the field of product-level energy and emissions calculation is unsettled. The issues are difficult, and an initial attempt at a systematic approach has produced a total emissions estimate far higher than other existing estimates in the literature. Clearly, there is not yet a reliable way to include this type of calculation in the personalized energy calculator.

Of equal importance are the myriad of surprising implications from the examples above:

· Aluminum cans are relatively energy efficient … If you recycle
· Want to save on the energy used in your shoes? ... Don't buy leather.
· Want to save energy while driving? ... Buy a high-mpg vehicle.
· Want to save money? … Keep your refrigerator until it wears out.
· Want to save energy? ... Replace your fridge every eight years or so.

The point is that the implications of these detailed models and calculations are sensible and somewhat predictable. That does not take away, however, from their impact.

ENERGY ACCOUNTING

Given the lack of opportunity for a product-level energy analysis, we now turn to an analysis of energy use that is based on aggregate data. As Weber (2009) notes, there are two main schools of energy use accounting in the literature. One branch uses data from the Energy Information Administration (EIA), which presents data by five sectors: residential; commercial; industrial; transportation, and electric power.  The second branch of analysis uses a detailed mapping of the U.S. economy, such as the EIA-LCA model, which is based on BEA input-output tables, which in turn captures economic activity for the 480 largest industries in the NAICS accounts. (NAICS is the North American Industry Classification System, used to classify businesses for data reporting on U.S. economic activity.)

Weber notes that these are two distinct literatures, and it is common not to reference studies of the other body of work. Weber, Mathews et al. (2010) show that the energy use from the two accounting systems reconciles to within one percentage point.

For the estimate of goods and services for the energy gauge, it is then reasonable to use the EIA data to create an estimate of national energy use in goods and services.

CALCULATION DETAIL

Table 5 shows the estimate of per-capita energy use in goods and services from the aggregate, economy-wide approach.

The inputs for food, flying, and driving are estimated from the other elements of the personalized energy calculator, for an average American. Home energy consumption is captured separately in the personal energy calculator, and so the EIA data on the residential sector is omitted from the total.

Household food includes the elements of the food industry captured in the diet section of the personalized energy calculator. This is a smaller industry definition than that used by the USDA. Its definition includes energy use in restaurants and home food preparation, but in the personalized energy calculator, these are included in goods and services and residential utility bills, respectively.

Despite this data adjustment, the results presented here are very similar in proportion to those of Jones and Kammen (2011). The ratio of the energy use in the food industry to energy use in goods and services before imports, 59%, is the same as their reported results, although their estimates are based on the EIO-LCA model, and the estimates here are based on EIA data. This congruence of results indicates that the two studies are directionally correct and that this aspect dominates any particulars about data adjustments and sources.

Imports in 2010 were 15.9% of GDP, and it is assumed that the energy used in imports is the same as in domestically produced goods and services, as measured by energy expenditures per dollar of revenue. This is a common assumption in the literature due to lack of alternate data.

The aggregate calculation method used here treats energy use for goods and services as essentially a very large residual. Several direct energy-using choices have been personalized and calculated; energy use for goods and services constitutes the remainder. A key advantage of this aggregate approach is a full accounting of energy use in the economy.

REFERENCES

Gardner and Stern (2009), "The Short List: The Most Effective Actions U.S. Households Can Take to Curb Climate Change," Environment, published in 2008; updated in 2009. Downloaded from http://www.environmentmagazine.org/Archives/Back%20Issues/September-October%202008/gardner-stern-full.html

Hendrickson et al..(2006), Environmental Life Cycle Assessment of Goods and Services, An Input-Output Approach. Resources for the Future, 262 pages.

Horie (2004), "Life Cycle Optimization of Refrigerator-Freezer Replacement", Center for Sustainable Systems, University of Michigan, downloaded from http://css.snre.umich.edu/css_doc/CSS04-13.pdf

Jones and Kammen (2008), "Consumer Oriented Lifecycle Consumption of Food, Goods and Services," The Berkeley Institute of the Environment, UC Berkeley, downloaded from http://repositories.cdlib.org/bie/energyclimate/jones_kammen_mcgrath_030308

Jones and Kammen (2011), "Quantifying Carbon Footprint Reduction Opportunities for U.S. Households and Communities," Environmental Science & Technology, downloaded from http://pubs.acs.org/doi/abs/10.1021/es102221h

Weber and Mathews (2008), "Food-Miles and Relative Impact of Food Choices in the U.S.," Environmental Science and Technology. Downloaded from psufoodscience.typepad.com/psu_food_science/files/es702969f.pdf

Weber, Mathews et al. (2010), "The 2002 US Benchmark Version of the Economic Input-Output Life Cycle Assessment (EIO-LCA) Model," Green Design Institute, Carnegie Mellon Institute, downloaded from http://www.eiolca.net/docs/full-document-2002-042310.pdf

NOTES

[1] Gardner and Stern (2009) also use the term "indirect" energy use in regard to energy consumption from the goods and services sector. Unfortunately, there is some room for confusion, as much of the literature uses the term "direct energy use" to mean energy used to make the good or service and "indirect energy use" to mean the energy used to make the inputs for producing the good or service. We hope to avoid confusion via context.
[2] See "Energy Intensity Indicators in the U.S.," http://www1.eere.energy.gov/ba/pba/intensityindicators/
[3] The Carbon Footprint of Fat Tire Amber Ale," http://www.newbelgium.com/Files/the-carbon-footprint-of-fat-tire-amber-ale-2008-public-dist-rfs.pdf
[4] This model has also been used for energy and emissions calculations for the food component of the expanded energy gauge.
[5] The results and discussion can be found at http://www.cokecorporateresponsibility.co.uk/big-themes/energy-and-climate-change/product-carbon-footprint.aspx
[6] Manufacturing Energy Consumption Survey, EIA. The data are found here: http://205.254.135.24/emeu/mecs/iab98/aluminum/energy_use.html (1998 data)
[7] See "Aluminum Can Life-Cycle Analysis," Aluminum Association, download here: http://www.patagonia.com/us/footprint/index.jsp
[8] The products are shown at http://www.patagonia.com/us/footprint/index.jsp and the methodology is found at http://www.patagonia.com/pdf/en_US/method_for_cost5.pdf
[9] Timberland's scorecard is found here: http://community.timberland.com/Earthkeeping/Green-Index. Timberland's experience with its calculations is profiled in "Six Products, Six Carbon Footprints," The Wall Street Journal Online, March 1, 2009: http://online.wsj.com/article/SB122304950601802565.html