White Paper: Energy Use in Food

Jan 12, 2012 | Diet

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

EARTH: THE OPERATORS' MANUAL

Martha Amram
January 2012

-----

For an introduction to energy use in food, see our Overview & Tips

-----

Of the many items we consume each day, food is one of the largest, at approximately 14 percent of annual U.S. energy consumption [1].  This short paper describes the calculations behind the personalized energy-diet calculator prepared by WattzOn that helps users understand the relative amounts of energy use to grow, process and transport their food. 

The goals of the energy-in-food calculator are:

· To provide information and create awareness about the role of food in our daily energy use.

· To provide suggestions as to how an interested consumer can reduce energy use by making alternate food choices.

· To provide information in a manner that engages consumers.

We have benefited from reviews of earlier drafts of this paper,  and from user experience with our web tools during the summer and fall of 2011. This has shown us that food choices interact with a host of other issues, including emotions about the nature of the U.S. food industry, carbon calculations and perceptions about personal diet and food choices.  For example, Americans eat a lot of sugar, and some reviewers have reacted to energy-in-food calculations with judgments about this food choice.

Based on this early experience, we have focused the calculator and this research note on the energy implications of food choices - a single dimension of a multi-dimensional issue. There are many elements of food and agriculture policy that affect climate change, diet and health, but these are beyond the scope of this work. Here, we maintain a narrow focus on personal energy use.

For most products and services, the item requiring the least amount of energy to make is also the item with the smallest greenhouse gas (GHG) emissions.  This is not true for food.  A diet heavy in beef will have higher carbon content than a vegetarian diet, but the energy required to make the two diets will not differ as much.  An important source of carbon emissions in diet is animal-based foods.  Animals themselves are producers of a significant amount of emissions as they digest food. Food processing the livestock and poultry does not require much more energy than processing in other food groups.  So while eating less meat clearly reduces the CO2e (carbon dioxide equivalent) emissions, eating less meat does not stand out as a key action to reducing energy use.

THE WEB TOOL

At the WattzOn website, users can enter data about their diet to obtain estimates of energy use and CO2e emissions from the mix of foods. Users adjust their food choices via a slider, moving the icon right or left to increase or decrease the number of daily or weekly servings of the food group.

The food groups shown are: beef, pork, poultry, fish, dairy, grains, fruits and vegetables, and sweets. We provide the increased detail on animal-based protein sources, because the relative energy impact versus CO2e impact varies sharply by protein source.

Users can enter data for a household or for an individual and display in either mode, as well. The results are shown in terms of BTUs (thermal unit) per household per year. The results can also be displayed in terms of pounds of CO2e.[2]

BASIC CALCULATIONS

The diet tool calculations are based on U.S. economy-wide data on food consumption and food production.  The basic concept is to match consumption and production by food type, thus obtaining the total energy used, to produce a total number of calories, which is then divided to obtain the energy required to produce a single calorie of that food type.  With user information, via the sliders, the annual energy use from diet choices can be estimated.

CALORIES AND SERVINGS

The USDA reports the average number of calories per capita per day by food type and the average number of servings per capita per day by food type.[3]  (The serving metric - cups, ounces, teaspoons - is also provided.)  These data are reported each year, and to match the data on energy used in food production, calorie data from 2002 is used.  There has been a slight downward trend in the number of calories and servings in recent years, but little year-to-year variation.

One of the important aspects of the USDA data is that it is "loss adjusted," meaning that food production and calories are net of waste.  For example, if 150 calories of a food type are produced, but only 100 calories are available for use by the consumer due to waste, calculations that ignore waste will be significantly off.  Food data net of waste is not always available and makes studies difficult to compare.

The Economic Research Service (ERS) of the USDA considers these loss adjustments important, as the adjusted data more closely reflect calories consumed and energy used.  However, the ERS notes that the adjustments are based on a few studies from the 1970s and 1980s, and that the food supply chain has changed significantly since then.[4]  Improved loss adjustments are an on-going project at the USDA, and, meanwhile, a recent FAO study found that losses in the North American food supply chain range from 6 percent of product in milk to 45 percent or more in vegetables (FAO, 2010).

In addition to food waste in the supply chain, there is considerable food waste at home.  The FAO study found 25 percent food waste in the home.  Parfitt et al. (2010) cite the original studies used for these conclusions, which show in-home food waste of 12-25 percent. For our calculations, we consider food calories available "at the front door".  Wasted food still required energy to produce, and, thus, no explicit data adjustment is needed.  There is, however, an opportunity to save energy by reducing in-home waste.

ENERGY AND EMISSIONS

Two key data sources are used that provide the energy consumed and the CO2e emitted in the production of food.  The first is the Environmental Input-Output and Lifecycle Cost Analysis (EIO-LCA) database, developed and maintained by the Green Design Institute at Carnegie Mellon University (CMU), which tracks energy use and CO2e emissions by industry.[5]  The second is a recent study by the USDA, Energy Use in the U.S. Food System.  Both data sets are based on the input-output tables of the U.S. economy, produced by the Bureau of Economic Analysis (BEA).[6]

Input-output tables provide a comprehensive view of the economy, tracking all inputs across all industries that are needed for the production of final goods.  The BEA maintains a 480-sector table, the most detailed analysis of its type in the world. There are 28 food sub-industries in the table.  The latest BEA input-output table is based on 2002 data.  The CMU and USDA studies are based on the BEA work, and, thus, also present data from 2002.

The EIO-LCA model is the result of more than 15 years of research at Carnegie Mellon University, and is a leading tool for lifecycle environmental impact analysis.[7]  Of particular importance for food calculations is that the EIO-LCA model tracks energy use and CO2e emissions for the entire supply chain.  For example, the estimates of the energy used to produce one pound of beef available at a local grocery store will include, among other inputs, the energy used to grow and transport the grain to feed the cattle, all energy for packaging and processing, energy for the materials to build new processing plants in the meat industry, and energy to transport the refrigerated meat to the store.  This list is not exhaustive, but merely illustrative of how the EIO-LCA model, based on economy-wide input-output tables, captures the full lifecycle energy use and CO2e emissions.  As this is the standard breadth of the EIO-LCA model, the results are comparable for each food input.

A few data details and wrinkles:  The EIO-LCA data report a single sub-industry, "animal products," instead of its sub-groups of beef, pork and other meats.  The USDA study provides this detail for energy use. The USDA study overlaps with the EIO-LCA data on other food industries, and the overlapping data are much the same.  The CO2e data for the beef and pork are from another USDA study (USDA, 2011d).  The EIO-LCA energy and CO2e data are reported in terms of dollars of industry output.  Industry output per food sub-industry is from the U.S. Census for 2002.[8]

MATCHING FOOD GROUPS TO FOOD INDUSTRIES

The USDA provides calories and servings data by food group, with detail about items within a group.  For example, under the vegetables group (48 calories per day, on average), data are available for the daily calories consumed for lettuce, tomatoes and more.

A key piece of analysis in building the diet calculator is the matching of these food groups to food industries.  This process is extremely important, as it ensures conformance between the data sets. In a few cases, a food item was moved from one food type to another, for better conformance with food industry data.  For example, ice cream is in the dairy food group for diets, but in the sweets food industry for these calculations.  Using the sub-industry detail, it was moved to the dairy industry group to create a better match between food production and food consumption.

When constructing the diet calculator, we faced a tension between ease-of-use and informative detail.  As the food group detail was adjusted to balance these factors, we also discovered that some sub-foods/sub-industries have less reliable results, due to small consumption or production.  For example, the average American eats 16 calories of fish per day, less than one serving.  Given this relatively small amount, it is no surprise that the energy used to produce a calorie is relatively high.  And, as the 16 calories are the total consumption from specific types of fish, each with an even smaller number of calories, minute changes in the number of calories consumed by fish type will have a large effect on the calculated energy use per calorie consumed.  To avoid this problem, the food sub-industry groups are aggregated, when possible.  Fish is treated as a separate source of protein, despite the potential for measurement errors, because of the strong consumer interest in this food.

The final food groups in the calculator are:

· Beef
· Pork
· Chicken
· Fish
· Grains
· Dairy
· Fruits & Vegetables
· Sweets

Not shown in the list are "added fats and oils", which is 24 percent of the calories consumed by the average American.  This is a difficult food group to deal with, as most us would not know where those nearly 600 calories go.  To best capture the full energy profile of the food industry, while not confusing users, we distributed the calories consumed from added fats and oils to other food groups by proportionately scaling up the number of servings, keeping the total calories consumed constant.  Separately, energy used and CO2e emissions from added fats and oils were distributed to other food groups in the same manner.

In addition, 9 percent of the food industry's energy use is from seasonings, sauces and miscellaneous items - again, a food group that is difficult to deal with.  The calories, energy use and CO2e were distributed to the other food and industry groups.

The adjustments for seasonings and fats preserve ease-of-use while keeping constant energy consumed, and CO2e emissions at the aggregate level.  This is a reasonable approach, as the food items (sauces, seasonings and added fats) are consumed in conjunction with the other food groups.

KEY FINDINGS

Table 1 shows the results of these calculations. Descriptions of typical serving sizes are from the Cleveland Clinic.[9]  The composition of the average American diet is shown for 2,600 total calories per day.  As mentioned earlier, these are calories before any food preparation waste in the home. Actual consumption will be less.

Note on Table 1 that the average American diet has a large number of calories from grains and sweets, and few calories from fruits and vegetables. As one shifts from the average American diet to healthier alternatives, the caloric composition shifts significantly, as do energy use and CO2e emissions.

Table 1 shows relatively high BTU per calorie produced for fish, and also high CO2e emissions for this source of protein. As mentioned earlier, this food group was broken out separately because of consumer interest, but we do not have high confidence in the data for fish.



Table 2 shows the same data in a different format, one that highlights the relative energy intensity of various food groups.  Because of concerns about the quality of the data for fish, the table was prepared both with and without the fish food group. The results for CO2e are no doubt familiar to many readers.  Beef is the largest emissions producer, by far. In contrast, the column of results for energy use in food production shows much less variation across food groups. In particular, beef is not an intense user of energy, but, in fact, is at or below average.  This leads to an interesting result: eat less beef to reduce emissions; eat less fruits and vegetables to reduce energy use.

Table 2 highlights how examining food policy through a single attribute - energy use or CO2e emissions - may miss other important goals, such as healthy eating.  Food policy has multiple objectives, and personal preference determines the balance point.  Minimizing energy use might not be the most important goal for many people.

Table 3 below provides an overall perspective and presents results that can be compared to those of other researchers.  The table shows the calories, energy use, and emissions associated with the average American diet.  Note that the average American diet is a research product of the USDA, and is not likely to be the diet of anyone you know.  It is based on food production, adjusted for losses. It is also adjusted for imports and exports. The USDA's average American diet is commonly used to show typical consumption and its validity has been extensively researched.[10]

Table 3 shows that consumption of proteins and grains are the largest energy uses in our diet. (The sum of the protein rows, from beef to fish, is 29 percent of daily BTUs on average.) The consumption of these food groups is also the largest emissions of CO2e.  Americans eat very few fruits and vegetables, only 12 percent of calories, and increasing consumption of this food group will not save energy or carbon.  Fresh vegetables, in particular, have a relatively high-energy intensity, due to the handling and refrigeration of these perishable crops.
 

COMPARISONS TO OTHER WORK

The total CO2e emissions from the food calculations are line with other carbon calculators, such as Jones and Kammen (2011), who estimate from EIO-LCA data that food emissions are 7.7 CO2e metric tons per year. (Our estimate is 9.8 CO2e metric tons per year, with the differences arising from the treatment of food away from home, miscellaneous food, the composition of the average diet, and so on.)  Similarly, a data summary from the University of Michigan shows that the average American emits 8 metric tons of CO2e per year from food choices.[11]  The same source also estimates that 48 percent of emissions arise from beef and dairy consumption, and our estimate is that 49 percent of emissions are from these sources.

The energy results from the food calculations are line with Eschel and Martin (2006) and the USDA, with about 15 percent of total energy use arising from food consumption for the average consumer.

It is important to note that congruence amongst estimates of carbon emissions and energy use for food is desirable, but may be a limited opportunity; results are highly dependent on methodology.  Researchers have approached this complicated area quite differently. Here are some of the factors to consider when evaluating studies, as often one is comparing "apples to oranges":

Defining the boundaries of the analysis. For example, the World Watch Institute includes many sources of land clearing in CO2e calculations. The CMU data uses older EPA data, and includes less of this item.

Inclusion of all food groups. For example, miscellaneous food, seasonings, and flavorings are nearly 9 percent of food industry energy, but are often omitted from food-emissions calculators.  Also, beverages are not included in diet or food industry data sources, though they consume energy and provide calories.  For example, soft drinks and alcoholic drinks are an additional 13 percent of energy use, above estimates from food alone (based on EIO-LCA data).

Fuel conversions. The EIO-LCA uses a 31 percent fuel efficiency for electricity generated from coal plants.  Other analyses do not distinguish between generation sources and keep all at 100 percent efficiency, and thus have much lower energy use and emissions.

Waste in the food supply chain. Some studies account for this and some do not.

What is the objective function? For example, monoculture has bad effects that are not related to energy or CO2e. Additionally, there are fertilizer runoff problems. These are issues that change the relative value of food groups and an informed consumer might want to consider them. Again, it is not clear that minimizing energy use is the single-decision criterion for consumption or policy-making.

Our current diet. Many studies do not start with the average American diet, and then vary food choices from this reference point. For example, eating healthy per the Choose My Plate diet recommended by the USDA will almost certainly reduce emissions - if less meat is eaten -- but, surprisingly, may sometimes increase energy use, depending on the specific fruits and vegetables and non-meat protein that replaces meat. The relative impact of diet choices must be measured against where we are, not where we wish we were.

The EIO-LCA data have been analyzed by others, and their research suggests additional recommendations. Weber and Mathews (2008) examine the transport component of energy use in the food system. They find that 83 percent of energy used comes from production, and that only 4 percent of energy is used to transport food from the producer to stores or restaurants. On average, food travels about 1,000 miles during the production process, but the authors conclude that the average consumer can save more energy or emissions by shifting away from beef or poultry one day a week than by buying local.

Typically, the food supply chain is assumed to end at the store or restaurant. The consumer then drives to those locations to purchase food. Weber and Mathews extend their analysis from simply transportation in the food supply chain, to the miles driven to purchase food. They note that reducing annual driving by 1,000 miles per year saves about the same amount of energy as does buying local. Reducing driving by 1,100 miles per year saves the same amount of energy as does shifting from animal to vegetable protein sources one day a week. Of course, the specifics of any tradeoff will depend on the particular vehicle and its mileage, but the authors point out that it might be easier to make changes via diet.

While the overall energy intensity of the U.S. economy has been declining, the food industry has been going in the opposite direction. Energy use per capital has been raising energy use in the food industry, largely from a shift to processed foods (USDA, 2010). The change reflects the decline in cooking at home; energy use from cooking, captured in utility bills, has dropped and energy use in the food supply chain has increased. Total energy may not have increased; energy consumption might just be a transfer from one sector to another. It is difficult to draw the conclusion of "buy fewer packaged foods and cook more at home to save energy."

Finally, although the data are not recent or as complete as one would like, it does appear that there is considerable opportunity to reduce energy and emissions in food consumption by wasting less.  The FAO study and the USDA estimates suggest considerable waste in the food supply chain, but importantly for these calculations, there is also considerable waste in the home.  A recent report by the USDA suggests that consumers waste a significant of calories purchased, with the exact amount varying by food group (USDA, 2011c).  So, reducing wasted food at home can make a significant savings in energy and emissions.
 

SUGGESTIONS FOR SAVING ENERGY OR REDUCING EMISSIONS

· Eat less.  One-third of adults are obese in the United States, consuming more calories than needed.[12] Eating is a very emotional issue, and we doubt that adults will eat less to save energy, but the math of food intake is clear.

· Buy less and waste less.  Estimates are that 12-25 percent of food purchased is wasted in the home. More careful food purchasing can be a large source of energy savings.

· Eat fewer grain products.  More than 35 percent of the energy in the average American diet comes from the grains food group.  Further, many grain products have added fats and oils, and/or added sugar, which are another 20 percent of energy in the diet.  In sum, 55 percent of energy and 60 percent of calories are from grains, sugars, and fats.  Because of its large share of the current American diet, this can be a place to reduce.
 

In addition, analyses by others using the same data suggest that:

· Buying local has only a small impact on energy and emissions, because most food energy is from production, not transportation.

· Driving to the store to purchase food uses more energy per year than does transporting food to the store.

· There are large waste losses in the food supply chain, and reduction of this waste throughout the industry can lower the energy content of food significantly.

· Energy use in the food industry has risen as Americans cook less. We continue to shift our purchases to prepared foods. An example is washed and bagged lettuce.

REFERENCES

BEA (2011), Benchmark Input-Output Tables for 2002. www.bea.gov/industry/io_benchmark.htm.

Eschel and Martin (2006), "Diet, Energy and Global Warming," Earth Interactions, 2006, http://pge.uchicago.edu/workshop/documents/martin1.pdf.

FAO (2010), Global Food Losses and Food Waste, downloaded from: www.fao.org/fileadmin/user_upload/ags/publications/GFL_web.pdf.

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

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.

Parfitt et al. (2010), "Food Waste within Food Supply Chains: Quantification and Potential for Change to 2050," Philosophical Transactions of the Royal Society, B, downloaded from: http://rstb.royalsocietypublishing.org/content/365/1554/3065.abstract

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.

USDA (2010), "Energy Use in the U.S. Food System," Economic Research Service, downloaded from: www.ers.usda.gov/Publications/err94/err94.pdf.

USDA (2011a). See "Average Daily per Capita Calories from the U.S. Food Availability, Adjusted for Spoilage and Other Waste," www.ers.usda.gov/data/foodconsumption/spreadsheets/foodloss/Calories.xls.

USDA (2011b), "Loss-Adjusted Availability, by Servings." www.ers.usda.gov/Data/FoodConsumption/FoodGuideSpreadsheets.htm#servings.

USDA (2011c), "Consumer-Level Food Loss Estimates and Their Use in the ERS Loss-Adjusted Food Availability Data," Economic Research Service, downloaded from: www.ers.usda.gov/Publications/TB1927/.

USDA (2011d), "U.S. Agriculture and Forestry Greenhouse Gas Inventory: 1990-2008; Chapter 2: Livestock and Grazed Land Emissions"; www.usda.gov/oce/climate_change/AFGG_Inventory/2_LivestockandGrazing.pdf.

World Watch Institute (2011), Livestock and Climate Change;
www.worldwatch.org/files/pdf/Livestock%20and%20Climate%20Change.pdf.

NOTES
[1] 14 percent (USDA, 2010)

[2] For consumer interest, other metrics may be displayed on the web tool, but as each can be translated into units of energy or CO2e, the discussion here will focus only on these two.

[3] See USDA (2011a) and (2011b).

[4] This discussion is available at www.ers.usda.gov/data/foodconsumption/FoodGuideDoc.htm.

[5] The EIO-LCA model is available in an interactive form from Carnegie Mellon University, at www.eiolca.net/cgi-bin/dft/use.pl. The producer price version of the model was used. Data on the energy use and CO2e emissions by food sub-industry were obtained, per $100 million in economic activity. The total 2002 economic activity by food sub-industry was obtained from the EIO-LCA model documentation, www.eiolca.net/docs/full-document-2002-042310.pdf. Energy use was then normalized to millions of BTUs per capita, using 2002 population estimates of 287.8 million residents (U.S. Census, http://www.census.gov/popest/data/historical/2000s/vintage_2002/index.html).

[6] The data are available at eiolca.net, USDA (2010) and BEA (2011), respectively.

[7] For an introduction to the model, see www.eiolca.net.

[8]. The industry sales data was downloaded from: http://factfinder.census.gov/servlet/IBQTable?_bm=y&-geo_id=&-ds_name=EC0231I1&-_lang=en.

[9] http://my.clevelandclinic.org/heart/prevention/weight/servingsize.aspx.

[10] There is a large literature on the two methods used by the USDA to determine what Americans eat (a food consumption survey and the production data used in this work), and much discussion about their strengths and weaknesses. Our intent here is to simply present an easy-to-understand reference point.

[11] Center for Sustainable Systems, University of Michigan. 2010. "Carbon Footprints Factsheet". Pub. No. CSS09-05.

[12] Center for Disease Control, 2010 data: www.cdc.gov/obesity/data/trends.html.