Shifting Diets for a Sustainable Food Future 0

Transcript

1 EMBARGOED FOR 4.20.16 AT 12:01 AM EDT. WORKING PAPER Installment 11 of “Creating a Sustainable Food Future” SHIFTING DIETS FOR A SUSTAINABLE FOOD FUTURE JANET RANGANATHAN, DANIEL VENNARD, RICHARD WAITE, PATRICE DUMAS, BRIAN LIPINSKI, TIM SEARCHINGER, AND GLOBAGRI-WRR MODEL AUTHORS SUMMARY CONTENTS How can shifting diets—the type, combination, and quantity of foods people consume—contribute to a Summary...1 sustainable food future? Building on the United Nations Diet Matters on the Menu...14 Food and Agriculture Organization’s (FAO) food demand Converging Diets... 21 projections, we estimate that the world needs to close a 70 Diet Shift 1: Reduce Overconsumption of Calories... 25 percent “food gap” between the crop calories available in 2006 and expected calorie demand in 2050. Diet Shift 2: Reduce Overconsumption of Protein by Reducing Consumption of Animal-Based Foods...31 The food gap stems primarily from population growth Diet Shift 3: Shift from Beef Specifically...42 and changing diets. The global population is projected to Effects of the Diet Shifts in 2050... ...49 grow to nearly 10 billion people by 2050, with two-thirds of those people projected to live in cities. In addition, Shifting Strategies for Shifting Diets...50 at least 3 billion people are expected to join the global Shift Wheel: A Framework for Shifting Consumption... 52 middle class by 2030. As nations urbanize and citizens Recommendations...63 become wealthier, people generally increase their calorie Call to Action...65 intake and the share of resource-intensive foods—such Appendices...66 as meats and dairy—in their diets. At the same time, technological advances, business and economic changes, References...73 and government policies are transforming entire food Endnotes...82 chains, from farm to fork. Multinational businesses are increasingly influencing what is grown and what people Working Papers contain preliminary research, analysis, eat. Together, these trends are driving a convergence findings, and recommendations. They are circulated to toward Western-style diets, which are high in calories, stimulate timely discussion and critical feedback, and to protein, and animal-based foods. Although some of this influence ongoing debate on emerging issues. Working shift reflects health and welfare gains for many people, papers may eventually be published in another form and the scale of this convergence in diets will make it harder their content may be revised. for the world to achieve several of the United Nations Sustainable Development Goals, including those on Ranganathan, J. et al. 2016. Suggested Citation: hunger, healthy lives, water management, climate change, “Shifting Diets for a Sustainable Food Future.” Working Paper, and terrestrial ecosystems. Creating a Sustainable Food Future Installment 11 of . Washington, DC: World Resources Institute. Accessible at http://www.worldresourcesreport.org. Note: All dollars are US dollars. All tons are metric tons (1,000 kg). “GHG” = greenhouse gas. “CO e” = carbon dioxide equivalent. “Kcal” = kilocalorie, also referred to as simply “calorie.” 2 | 1 | April 2016 WORKING PAPER

2 Efforts to close the food gap have typically focused on What are the trends in calorie consumption and increasing agricultural production. However, relying solely why do they matter? on increased production to close the gap would exert pres - There is a global trend toward overconsumption of sure to clear additional natural ecosystems. For example, calories, even though many people around the world to increase food production by 70 percent while avoiding remain hungry. In 2009, per capita calorie consumption further expansion of harvested area, crop yields would exceeded average daily energy requirements in regions need to grow one-third more quickly than they did during containing half of the world’s population. Globally, there the Green Revolution. In short, yield increases alone will are now two-and-a-half times more overweight than likely be insufficient to close the gap. undernourished people. More than one in three adults are overweight. While per person calorie availability may To help provide a more holistic approach, the World be peaking in developed countries, it is rising across the Resources Report, Creating a Sustainable Food Future , developing world, particularly in emerging economies like and a series of accompanying working papers propose a China and Brazil. Once considered a high-income-country and consumption-based solutions. In menu of production- problem, the numbers of obese or overweight people are this paper, the last in the series, we assess the role of one - now rising in low- and middle-income countries, espe - consumption-based solution: shifting the diets of popula cially in urban areas. tions who consume high amounts of calories, protein, and animal-based foods. Specifically, we consider three Overconsumption of calories widens the food gap and interconnected diet shifts: - drives unnecessary use of agriculture inputs and unnec essary environmental impacts. It also contributes to Reduce overconsumption of calories. 1. people becoming overweight and obese, harming human Reduce overconsumption of protein by reducing 2. health and contributing to rising healthcare costs and lost consumption of animal-based foods. productivity. The related economic and healthcare costs are enormous. For example, the global economic cost of Reduce consumption of beef specifically. 3. obesity was estimated to be around $2 trillion in 2012, roughly equivalent to the global cost of armed conflict or For each shift, we describe the issue it addresses, why it smoking. matters, and the relevant trends. We use the GlobAgri model to quantify the land use and greenhouse gas What are the trends in protein consumption consequences of different foods, and then analyze the and why do they matter? per person and global effects of the three diet shifts on agricultural land needs and greenhouse gas emissions. Overconsumption of protein occurs in all of the world’s We find that these diet shifts—if implemented at a - regions, and it is rising in developing and emerging econo large scale—can close the food gap by up to 30 percent, mies. In 2009, the average person in more than 90 percent while substantially reducing agriculture’s resource use of the world’s countries and territories consumed more and environmental impacts. With the food industry in protein than estimated requirements. Global average pro - mind—particularly the retail and food service sectors—we tein consumption was approximately 68 grams per person introduce the Shift Wheel, a framework that harnesses per day—or more than one-third higher than the average marketing and behavioral change strategies to tackle daily adult requirement. In the world’s wealthiest regions, the crucial question of how to shift people’s diets. We protein consumption was higher still (Figure ES-1). conclude with four recommendations to help shift diets and apply the Shift Wheel. In addition, the share of animal-based protein is grow - ing in people’s diets relative to that of plant-based pro - tein. Between 1961 and 2009, global average per person availability of animal-based protein grew by 59 percent, while that of plant-based protein grew by only 14 percent. Looking forward, total consumption of animal-based food is expected to rise by nearly 80 percent between 2006 | 2

3 Shifting Diets for a Sustainable Food Future Figure ES-1 | Protein Consumption Exceeds Average Estimated Daily Requirements in All the World’s Regions, and is Highest in Developed Countries g/capita/day, 2009 Animal-based protein Sub-Saharan OECD (other) Former US & Canada Brazil Plant-based protein Africa Soviet Union 100 India European China Latin America Middle East & Asia Union North Africa (ex. Brazil) (ex. China & India) 90 80 70 60 Average daily 50 protein requirement 40 30 20 10 0 0 1 3 2 4 5 6 7 Population (billions) Source: GlobAgri model with source data from FAO (2015) and FAO (2011a). Width of bars is proportional to each region’s population. Average daily protein requirement of 50 g/day is based on an average adult body weight of 62 kg (Walpole et al. 2012) and recommended protein intake of 0.8 g/kg body weight/day (Paul 1989). Individuals’ energy requirements vary depending on age, gender, height, weight, pregnancy/lactation, and level of physical activity. and 2050. Although per person animal-based food con - What are the trends in beef consumption and sumption may be peaking in developed countries where why do they matter? consumption is already high, it is projected to rise in Beef consumption is rising in emerging economies and is developing countries, especially in emerging economies showing signs of peaking in some developed countries. In and in urban areas. Brazil, per person beef availability (and probably con - sumption) has increased steadily over the past decades, Like overconsumption of calories, overconsumption and is now more than three times the world average, of protein widens the food gap. Furthermore, animal- having surpassed the United States in 2008. In China, per based foods are typically more resource-intensive and - person beef availability is still only half of the world aver environmentally impactful to produce than plant-based age, but is growing. In India, growing demand for dairy foods (Figure ES-2). Production of animal-based foods products is spurring an expansion in the cattle population, accounted for more than three-quarters of global agri - although beef consumption remains low. In the United cultural land use and around two-thirds of agriculture’s States, per person annual beef consumption has declined production-related greenhouse gas emissions in 2009, 27 percent since the 1970s. Global demand for beef is pro - while only contributing 37 percent of total protein con - jected to increase by 95 percent between 2006 and 2050, sumed by people in that year. Because many animal-based with much of this growth in countries where current per foods rely on crops for feed, increased demand for animal- person consumption is low, such as China and India. based foods widens the food gap relative to increased demand for plant-based foods. | 3 | April 2016 WORKING PAPER

4 Production of Animal-Based Foods is Generally Figure ES-2 | PR OTEIN CONSUMED CAL ORIE CONSUMED More Impactful on the Planet than Plant-Based Foods PER T ON PR ORIES CONSUMED OCAL PER MILLION KIL OTEIN CONSUMED ANIMAL-B ASED PLANT ASED ANIMAL-B PLANT -BASED -BASED 3 3 1, 000 m t CO e t CO e 1, 000 m ha ha 2 2 120 3,000 250 180 15 10 LAND USE (ha) LAND USE (ha) Pasture Pasture Cropland Cropland 3 3 TER CONSUMPTION (1,000 m FRESHWA ) FRESHWA ) TER CONSUMPTION (1,000 m ater Rainw ater Rainw 100 2,500 150 ation Irrig ation Irrig 12 8 200 GHG EMISSIONS (t CO e) e) GHG EMISSIONS (t CO 2 2 hang e Land-use c e hang Land-use c Agricultural production Agricultural production 2,000 80 120 9 6 150 60 1,500 90 6 4 100 40 1,000 60 3 50 2 500 20 30 0 0 Pulses Rice Maize Roots & Wheat Beef y Wheat Dair Poultr Pork Eggs y Fish Rice Rapeseed y Beef Pulses Ro ots & Sug Fish ar Maize Eggs Pork Poultry Dair So ybean Oil Fruits & Sunflow er med) (far Tubers & Mustard Tubers med) (far Seed Oil Vegetabl es Seed Oil Sources: GlobAgri model (land use and greenhouse gas emissions), authors’ calculations from Mekonnen and Hoekstra (2011, 2012) (freshwater consumption), and Waite et al. (2014) (farmed fish freshwater consumption). Data presented are global means. Entries are ordered left to right by amount of total land use. Indicators for animal-based foods include resource use to produce feed, including pasture. Notes: Tons of har vested products were converted to quantities of calories and protein using the global average edible calorie and protein contents of food types as reported in FAO (2015). “Fish” includes all aquatic animal products. Freshwater use for farmed fish products is shown as rainwater and irrigation combined. Land use and greenhouse gas emissions estimates are based on a marginal analysis (i.e., additional agricultural land use and emissions per additional million calories or ton of protein consumed). Based on the approach taken by the European Union for estimating emissions from land-use change for biofuels, land-use change impacts are amortized over a period of 20 years and then shown as annual impacts. Land use and greenhouse gas emissions estimates for beef production are based on dedicated beef production, not beef that is a coproduct of dairy. Dairy figures are lower in GlobAgri than some other models because GlobAgri assumes that beef produced by dairy systems displaces beef produced by dedicated beef-production systems. | 4

5 Shifting Diets for a Sustainable Food Future Production of Animal-Based Foods is Generally Figure ES-2 | CAL OTEIN CONSUMED ORIE CONSUMED PR More Impactful on the Planet than Plant-Based Foods (continued) ON PR PER T ORIES CONSUMED OCAL PER MILLION KIL OTEIN CONSUMED PLANT ASED ANIMAL-B -BASED PLANT -BASED ASED ANIMAL-B 3 3 1, 000 m t CO e t CO e 1, 000 m ha ha 2 2 3,000 120 250 180 15 10 LAND USE (ha) LAND USE (ha) Pasture Pasture Cropland Cropland 3 3 ) TER CONSUMPTION (1,000 m FRESHWA TER CONSUMPTION (1,000 m FRESHWA ) ater Rainw ater Rainw 100 2,500 150 ation Irrig Irrig ation 12 8 200 e) GHG EMISSIONS (t CO GHG EMISSIONS (t CO e) 2 2 e hang Land-use c e Land-use c hang Agricultural production Agricultural production 2,000 80 120 9 150 6 60 1,500 90 6 4 100 40 1,000 60 3 2 50 500 20 30 0 0 Pulses Fish Beef Rice Maize Roots & Wheat Wheat y Pork Poultr y Dair Eggs Rice Rapeseed Dair Fish Poultry ots & y Ro Pulses Maize ar Sug Pork Eggs Beef So Fruits & Sunflow er ybean Oil (far med) Tubers & Mustard Tubers (far med) Seed Oil Vegetabl es Seed Oil Sources: GlobAgri model (land use and greenhouse gas emissions), authors’ calculations from Mekonnen and Hoekstra (2011, 2012) (freshwater consumption), and Waite et al. (2014) (farmed fish freshwater consumption). Notes: Data presented are global means. Entries are ordered left to right by amount of total land use. Indicators for animal-based foods include resource use to produce feed, including pasture. Tons of har vested products were converted to quantities of calories and protein using the global average edible calorie and protein contents of food types as reported in FAO (2015). “Fish” includes all aquatic animal products. Freshwater use for farmed fish products is shown as rainwater and irrigation combined. Land use and greenhouse gas emissions estimates are based on a marginal analysis (i.e., additional agricultural land use and emissions per additional million calories or ton of protein consumed). Based on the approach taken by the European Union for estimating emissions from land-use change for biofuels, land-use change impacts are amortized over a period of 20 years and then shown as annual impacts. Land use and greenhouse gas emissions estimates for beef production are based on dedicated beef production, not beef that is a coproduct of dairy. Dairy figures are lower in GlobAgri than some other models because GlobAgri assumes that beef produced by dairy systems displaces beef produced by dedicated beef-production systems. | 5 | April 2016 WORKING PAPER

6 At the global level, beef production is a major driver of Beef is one of the least efficient foods to produce when agricultural resource use. One-quarter of the Earth’s considered from a “feed input to food output” perspective. landmass, excluding Antarctica, is used as pasture, and When accounting for all feeds, including both crops and forages, by one estimate only 1 percent of gross cattle feed beef accounts for one-third of the global water footprint - calories and 4 percent of ingested protein are converted of farm animal production. Although some beef produc to human-edible calories and protein, respectively. In tion uses native pasture, increases in beef production now rely on clearing forests and woody savannas. Ruminants, comparison, by this estimate, poultry convert 11 percent of of which beef is the most commonly produced and con feed calories and 20 percent of feed protein into human- - edible calories and protein. Because of this low conversion sumed, are responsible for nearly half of greenhouse gas emissions from agricultural production. Given the envi efficiency, beef uses more land and freshwater and generates - more greenhouse gas emissions per unit of protein than any ronmental implications of rising demand for beef, reduc - other commonly consumed food (Figure ES-2). ing its consumption will likely be an important element to limiting the rise of global temperatures to 1.5 or 2 degrees Celsius, in line with international goals. Diet Shifts and Scenarios Modeled in this Paper Table ES-1 | AFFECTED POPULATION SCENARIO NAME SCENARIO DESCRIPTION (MILLIONS), 2009 DIET SHIFT 1: Reduce overconsumption of calories 1,385 Eliminate Obesity and Recognizing that reducing overconsumption of calories can contribute to reducing overweight and obesity, this scenario eliminates obesity and halves the number of overweight people by Halve Overweight reducing calorie consumption across all foods. 1,046 Halve Obesity and Similar to the above scenario, this scenario halves the number of obese and Halve Overweight overweight people. DIET SHIFT 2: Reduce overconsumption of protein by reducing consumption of animal-based foods 1,907 Ambitious Animal In regions that consumed more than 60 grams of protein (from animal and plant sources Protein Reduction combined) and more than 2,500 calories per person per day, protein consumption was reduced to 60 grams per person per day by reducing animal-based protein consumption (across all animal- based foods). Overall, global animal-based protein consumption was reduced by 17 percent. 437 Traditional In regions that consumed more than 40 grams of animal-based protein and more than 2,500 Mediterranean Diet calories per person per day, half of the population was shifted to the actual average diet of Spain and Greece in 1980. Overall calorie consumption was held constant. 437 Vegetarian Diet In regions that consumed more than 40 grams of animal-based protein and more than 2,500 calories per person per day, half of the population was shifted to the actual vegetarian diet as obser ved in the United Kingdom in the 1990s. Overall calorie consumption was held constant. DIET SHIFT 3: Reduce beef consumption specifically 1,463 Ambitious Beef In regions where daily per person beef consumption was above the world average and daily per person calorie consumption was above 2,500 per day, beef consumption was reduced to the Reduction world average level. Overall, global beef consumption was reduced by 30 percent. 1,952 Shift from Beef to In regions where daily per person beef consumption was above the world average, beef consumption was reduced by one-third and replaced by pork and poultry. Overall calorie Pork and Poultry consumption was held constant. 1,952 Shift from Beef to In regions where daily per person beef consumption was above the world average, beef Legumes consumption was reduced by one-third and replaced with pulses and soy. Overall calorie consumption was held constant. | 6

7 Shifting Diets for a Sustainable Food Future Highlights of the results are summarized below. What would be the effects of applying the three diet shifts to high-consuming populations? The agricultural land use and greenhouse gas emissions Shifting the diets of high-consuming populations could associated with the average American diet were nearly significantly reduce agricultural resource use and double those associated with the average world diet, with environmental impacts. We used the GlobAgri model to 80 to 90 percent of the impacts from consumption of analyze the effects of the three diet shifts on agricultural animal-based foods. land use and greenhouse gas emissions in 2009. For each - We found that producing the food for the average Ameri of the three shifts, we developed alternative diet scenarios, can diet in 2009 required nearly one hectare of agricul - ranging from “realistic” to “ambitious” (Table ES-1). In - tural land, and emitted 1.4 tons of carbon dioxide equiva each scenario, we assumed that crop and livestock yields e), before accounting for emissions from land-use lent (CO and trade patterns remained constant at actual 2009 2 change. These amounts of land use and greenhouse gas levels. We altered food consumption levels among the emissions were nearly double those associated with the world’s high-consuming populations, but did not alter the average world diet that year (Figure ES-3). Animal-based diets of the world’s less wealthy. None of the scenarios foods (shown in red, orange, and yellow in Figure ES-3) sought to turn everyone into a vegetarian. accounted for nearly 85 percent of the production-related - greenhouse gas emissions and nearly 90 percent of agri We conducted two types of analysis using 2009 food cultural land use. Beef consumption alone (shown in red) consumption data: accounted for nearly half of the US diet-related agricul - tural land use and greenhouse gas emissions. First, we quantified the per person effects of applying ▪ the diet scenarios in Table ES-1 to the consumption Furthermore, factoring land-use implications into pattern of a high-consuming country—the United agricultural greenhouse gas emissions estimates shows States (Figure ES-3). This analysis shows how, among a fuller picture of the consequences of people’s dietary high-consuming populations, the three diet shifts choices. For example, if an additional person eating the could significantly reduce per person agricultural land average American diet were added to the world population use and greenhouse gas emissions. in 2009, the one-time emissions resulting from converting a hectare of land to agriculture to feed that person would Second, we quantified the global effects of applying the ▪ e. This amount is equal to 17 be about 300 tons of CO - diet scenarios to people currently overconsuming calo 2 times the average US per person energy-related carbon ries or protein, or who are high consumers of beef, to dioxide emissions in 2009. In other words, the emissions show the aggregate effects of the diet shifts across large from clearing additional land to feed an additional person populations. The scenarios affected the diets of between eating the US diet are equal to 17 years’ worth of an 440 million and 2 billion people (Figure ES-4). average American’s energy-related CO emissions. 2 Shifting the diets of high consumers of animal-based foods could significantly reduce per person agricultural land use and greenhouse gas emissions. When applied to the average American diet in 2009, the Vegetarian Diet Ambitious Animal Protein Reduction and scenarios reduced per person land use and agricultural greenhouse gas emissions by around one-half—or down to around world average. The three scenarios that reduced consumption of beef—just one food type—reduced per person land use and greenhouse gas emissions by 15 to 35 percent. Figure ES-3 shows the effects of the three diet shifts on per person agricultural land use and greenhouse gas emissions when applied to the average American diet. | 7 | April 2016 WORKING PAPER

8 Figure ES-3 | Shifting the Diets of High Consumers of Animal-Based Foods Could Significantly Reduce Per Person Agricultural Land Use and GHG Emissions per capita values, 2009 Plant-Based Foods Other Animal-Based Foods Dairy Beef REDUCE OVERCONSUMPTION OF PROTEIN BY REDUCE OVERCONSUMPTION OF CALORIES REDUCING CONSUMPTION OF ANIMAL-BASED FOODS US (TRADITIONAL US (ELIMINATE OBESITY US (AMBITIOUS ANIMAL US (HALVE OBESITY US (REFERENCE) & HALVE OVERWEIGHT) PROTEIN REDUCTION) & OVERWEIGHT) MEDITERRANEAN) DAILY FOOD 2,796 CONSUMPTION 2,904 2,520 2,904 2,726 (KCAL) 0.93 0.90 0.85 0.53 0.96 AGRICULTURAL LAND USE (HECTARES) 1.5 1.4 1.3 1.3 1.2 1.2 0.9 GHG EMISSIONS 0.8 FROM AGRICULTURAL 0.6 PRODUCTION (TONS CO E) 2 0.3 0 20 15.2 14.3 14.7 15 13.5 GHG EMISSIONS 10 FROM LAND-USE 8.4 CHANGE E) (TONS CO 2 5 0 Source: GlobAgri model. All "US" data are for United States and Canada. Land-use change emissions are amortized over a period of 20 years and then shown as annual impacts. Calculations assume global average Note: e emitted) for all food types. “Other animal-based foods” includes pork, poultry, eggs, fish (aquatic animals), sheep, and goat. efficiencies (calories produced per hectare or per ton of CO 2 | 8

9 Shifting Diets for a Sustainable Food Future Shifting the Diets of High Consumers of Animal-Based Foods Could Significantly Figure ES-3 | Reduce Per Person Agricultural Land Use and GHG Emissions (continued) per capita values, 2009 Plant-Based Foods Dairy Other Animal-Based Foods Beef REDUCE OVERCONSUMPTION OF PROTEIN BY REDUCE CONSUMPTION OF BEEF SPECIFICALLY REDUCING CONSUMPTION OF ANIMAL-BASED FOODS US (AMBITIOUS US (SHIFT FROM BEEF US (SHIFT FROM US (VEGETARIAN) WORLD (REFERENCE) TO PORK AND POULTRY) BEEF REDUCTION) BEEF TO LEGUMES) DAILY FOOD CONSUMPTION 2,904 2,433 2,904 2,904 2,834 (KCAL) 0.50 0.49 0.82 0.64 0.83 AGRICULTURAL LAND USE (HECTARES) 1.5 1.1 1.2 1.2 0.8 0.9 0.9 GHG EMISSIONS FROM AGRICULTURAL 0.6 0.6 PRODUCTION (TONS CO E) 2 0.3 0 20 15 13.2 13.0 GHG EMISSIONS 10.2 10 FROM LAND-USE 7.9 7.6 CHANGE E) (TONS CO 2 5 0 GlobAgri model. Source: All "US" data are for United States and Canada. Land-use change emissions are amortized over a period of 20 years and then shown as annual impacts. Calculations assume global average Note: efficiencies (calories produced per hectare or per ton of CO e emitted) for all food types. “Other animal-based foods” includes pork, poultry, eggs, fish (aquatic animals), sheep, and goat. The 2 vegetarian diet scenario, which uses data from Scarborough et al. (2014), includes small amounts of meat, as “vegetarians” were self-reported. | 9 | April 2016 WORKING PAPER

10 2. REDUCE OVERCONSUMPTION OF PROTEIN BY REDUCING Reducing animal-based food consumption results in This diet CONSUMPTION OF ANIMAL-BASED FOODS. significant savings in global agricultural land use. shift resulted in the largest benefits, as it applied When applied globally to populations overconsuming to a relatively large population and across all calories or protein, or who are high consumers of beef, animal-based foods. the diet scenarios could spare between 90 million and 640 million hectares of agricultural land. The Ambi - 3. REDUCE BEEF CONSUMPTION SPECIFICALLY. This diet tious Animal Protein Reduction scenario—which shifted shift resulted in significant benefits, and would be the diets of nearly 2 billion people in 2009—spared 640 relatively easy to implement, since it only affects one million hectares of agricultural land, including more than type of food. Additionally, some high-consuming 500 million hectares of pasture and 130 million hectares countries have already reduced per person beef of cropland. This area of land is roughly twice the size of consumption from historical highs, suggesting that India, and is also larger than the entire area of agricul - further change is possible. tural expansion that occurred globally over the past five scenario Ambitious Beef Reduction decades. Notably, the The diet shifts can also help close the gap between crop spared roughly 300 million hectares of pasture—an calories available in 2006 and those demanded in 2050. amount similar to the entire area of pasture converted With a projected 25 percent of all crops (measured by from other lands since 1961. calories) dedicated to animal feed in 2050, we calculate scenario Ambitious Animal Protein Reduction that the These results suggest that reducing consumption could reduce the food gap by 30 percent—significantly of animal-based foods among the world’s wealthier reducing the challenge of sustain ably feeding nearly 10 populations could enable the world to adequately feed billion people by mid-century. 10 billion people by 2050 without further agricultural expansion. Curbing agricultural expansion would also Will the diet shifts adversely impact poor food avoid future greenhouse gas emissions from land-use producers and consumers? change. The Ambitious Animal Protein Reduction scenario, which spared the most land, could avoid 168 The diet shifts do not call for the world’s poor to reduce e from land-use change. billion tons of emissions of CO consumption, and they preserve an abundant role for 2 To put this reduction in perspective, global greenhouse small livestock farmers. The three shifts target popula - gas emissions in 2009 were 44 billion tons CO e. Figure tions who are currently overconsuming calories or protein, 2 ES-4 shows the global effects of the three diet shifts on or are high consumers of beef—or are projected to be by agricultural land use in 2009. 2050. They do not target undernourished or malnourished populations. Nor do they aim to eliminate the livestock All three diet shifts could contribute to a sustainable sector, which provides livelihoods to millions of poor smallholders, makes productive use of the world’s native food future, but the two shifts that reduce consumption grazing lands, and generates 40 percent of global agricul - of animal-based foods result in the largest land use and tural income. Indeed, solutions to sustainably increase greenhouse gas reductions. crop and livestock productivity are also critical to closing Our analysis of the three diet shifts, summarized in the food gap, and are covered in the of Interim Findings Figures ES-3 and ES-4, yields the following insights: Creating a Sustainable Food Future. REDUCE OVERCONSUMPTION OF CALORIES. 1. While Would reducing beef consumption result in reducing overweight and obesity is important for productive pastureland going to waste? human health, this diet shift contributed less to reducing agriculture’s resource use and environmental Reducing beef consumption is unlikely to result in impacts than the other two shifts. “wasted” pastureland for two reasons. First, beef demand is projected to nearly double between 2006 and 2050, and pasture is likely to remain the dominant source of feed. Even with increased pasture productivity, it will be diffi - cult to meet projected growth in demand without clearing more natural forests and savannas for pasture. Our beef | 10

11 Shifting Diets for a Sustainable Food Future Figure ES-4 | Reducing Animal-Based Food Consumption Results in Significant Savings in Global Agricultural Land Use millions of hectares saved relative to world reference, 2009 700 Pasture Cropland 600 500 400 300 200 100 0 Eliminate Ambitious Traditional Vegetarian Shift from Shift from Ambitious Halve Obesity Obesity & Halve Beef to Pork Beef Reduction & Overweight Mediterranean Beef to Animal Protein Overweight Reduction and Poultry Legumes M S I N M Affected I R M O I N Z 437 1,463 1,952 1,952 1,385 1,046 1,907 437 Population E L A D (millions) I I C S R O U S Reduce overconsumption Reduce overconsumption Reduce consumption P E T V of beef specifically of calories of protein by reducing consumption I L O O of animal-based foods N V E GlobAgri model. Source: Shift from Beef to Pork and Poultry The Note: scenario includes a 196 Mha decrease in pasture, but a 26 Mha increase in cropland, for an overall 170 Mha “savings.” reduction scenarios do not eliminate beef consumption, What can be done to shift people’s diets? - but just reduce it. Given the projected growth in beef con There is no silver bullet solution. To date, efforts to sumption, the proposed diet shifts are unlikely to reduce encourage more sustainable eating have largely focused beef’s global land use below today’s levels. Accordingly, on consumer education, back-of-the-package labeling, T I “reducing beef consumption” is about preventing further F and campaigns around abstinence (e.g., vegetarianism), E expansion, not creating a surplus of unused or “wasted” N M with limited success. A more holistic approach is needed E A B pastureland. Second, because native grazing land has few X that works in step with how consumers make purchasing I G M alternative uses, it is nearly always used for grazing. Even N I I Z decisions. Purchases are typically based on habit and L E L if there were large enough reductions in beef consumption E A unconscious mental processing rather than on rational, P W to reduce total pasture area, the result would not be to stop M A O R informed decisions. Furthermore, attributes like price, C E N grazing of native pasturelands but instead to free up lands A E L S taste, and quality tend to be more important than L S E S that were naturally wooded and that are wet enough to meet sustainability in purchasing decisions. Strategies that the growing demand for crops or for regenerating forests. influence these factors and engage actors in food value | 11 | April 2016 WORKING PAPER

12 Figure ES-5 | The Shift Wheel Comprises Four Strategies to Shift Consumption M S I M N R I M O I N Z E L Replicate the Inform about A D I I experience the issue C S O R S U P E T V I Disguise L Make socially O O N the change desirable V E Make socially Form habits in unacceptable new markets SHIFT CONSUMPTION Be more Meet current memorable key needs T I Deliver new Constrain F M E display compelling A N X E benefit I B M G I Enhance Enhance Z N E I L display affordability A L W E P A R M E O N C E A S S L L E S Source: Authors. The Shift Wheel comprises four complementary strategies: chains (e.g., food manufacturing companies, food service companies, supermarkets) are needed. The multinational MINIMIZE DISRUPTION. Changing food consumption businesses that are increasingly influencing consumers’ ▪ choices across the globe can play an important role in behavior typically involves changing ingrained habits. shifting consumers to more sustainable diets. This strategy seeks to minimize the disruption to consumers’ habits caused by the shift. It can include minimizing changes associated with the shift, such To help shift people’s diets, we propose a new framework based on proven private sector marketing tactics: the Shift as taste, look, texture, smell, packaging, and the product’s location within a store. Wheel (Figure ES-5). The development of the Shift Wheel was informed by a range of consumption shifts already successfully orchestrated by industry, nongovernmental Selling a compelling SELL A COMPELLING BENEFIT. ▪ organizations (NGOs), and government. These include benefit requires identifying and delivering product shifts such as from caged to free-range eggs in the United attributes (such as health or affordability) that will be Kingdom, from higher- to lower-alcohol beer in the United sufficiently motivating to the consumer to stimulate a Kingdom, and away from shark fin in China. behavior change. As plant-based proteins can be less expensive than animal-based ones, companies may have an opportunity to sell reformulated products with a greater share of plant-based ingredients at a lower price and/or an increased profit. | 12

13 Shifting Diets for a Sustainable Food Future MAXIMIZE AWARENESS. The more consumers see or What actions are needed to apply the ▪ think of a product, the greater the chance they will Shift Wheel and shift diets? consider purchasing it. Enhancing the availability and We offer four recommendations to help the food industry - display of the more sustainable food choice, and creat apply the Shift Wheel and shift diets: ing memorable advertising campaigns, can increase a product’s visibility and the chance that consumers will SET TARGETS, APPLY THE SHIFT WHEEL, LEARN FROM purchase it. ▪ THE RESULTS, AND SCALE UP SUCCESSES Companies and governments should set quantifiable What people eat is highly influ EVOLVE SOCIAL NORMS. - ▪ targets to reduce the consumption of animal-based - enced by cultural environment and social norms. In protein and beef specifically. They should use the Shift forming and educating consumers, along with efforts Wheel to drive progress toward these targets. to make the preferred food more socially desirable or the food to be shifted from less socially desirable, can ENSURE GOVERNMENT POLICIES ARE ALIGNED WITH influence or change the underlying social and cultural ▪ PROMOTING SUSTAINABLE DIET CHOICES norms that underlie people’s purchasing decisions. Governments should ensure coherence among agri - culture, health, water, and environmental policies in How can the Shift Wheel be applied to shift diets? The first relation to promoting sustainable diets. step is to analyze the landscape of animal- and plant-based food consumption in a given geography or market. Who INCREASE FUNDING FOR EFFORTS are the consumers? What are they eating? Where, when, ▪ TARGETED AT SHIFTING DIETS why, and how is this consumption occurring? The answers Governments and foundations should create funding to these questions will help identify the most promising mechanisms to support the development, testing, and intervention points. This might be a specific occasion (e.g., rollout of evidence-based strategies to shift diets. family evening meals), a product format (e.g., meatballs), a social perception (e.g., that plant-based protein is infe - CREATE A NEW INITIATIVE FOCUSED ON TESTING rior to meat), a demographic group (e.g., millennials), or ▪ AND SCALING UP STRATEGIES TO SHIFT DIETS specific outlets (e.g., school or workplace cafeterias). The A new initiative should be established to apply the next step is to design approaches to achieve the chosen Shift Wheel to specific contexts and catalyze new shift, drawing on relevant strategies from the Shift Wheel. approaches to shifting diets. Such an initiative could The final steps are to test the selected approaches and conduct pilot tests, build an evidence base, measure scale up successes. behavior change and its impacts on people and the environment, and share and scale up successes. Its goal should be to increase the share of plant-based protein in diets and reduce the consumption of beef specifically. | 13 | April 2016 WORKING PAPER

14 The World Resources Report: Box 1 | DIET MATTERS ON THE MENU Creating a Sustainable Food Future What we eat has a profound impact on our own health and - the planet’s health. In the World Resources Report’s Creat ing a Sustainable Food Future: Interim Findings (Box 1), we describe how the world food system faces a great bal - The world faces a great balancing act. How will the world ancing act: feeding the population in 2050 while advancing adequately feed nearly 10 billion people by 2050 in a manner that advances economic development and reduces pressure economic development and reducing agriculture’s pressure on the environment, while adapting to climate change? This on the environment in a changing climate. The Interim requires balancing three needs. First, the world needs to close Findings and an accompanying series of working papers a roughly 70 percent gap between the crop calories available explores a menu of solutions that could combine to achieve in 2006 and those needed by 2050 (the “food gap”). Second, this balance. the world needs agriculture to contribute to inclusive economic and social development. Third, the world needs to reduce agriculture’s impact on the environment. Based on an adjusted FAO projection of food demand and production by 2050, the world needs to close a roughly 70 This edition of the World Resources Report, Creating a percent gap between the crop calories available in 2006 , proposes a menu of solutions that Sustainable Food Future 1 Global and expected calorie demand in 2050 (Figure 1). can achieve this balance. The menu includes production- 2 population is projected to grow to 9.7 billion by 2050. focused solutions, such as sustainably increasing crop yields and livestock pasture productivity, and consumption-focused At least 3 billion more people are expected to enter the 3 solutions, such as reducing food waste, achieving replacement- global middle class by 2030, and two-thirds of the global level fertility, and reducing demand for bioenergy that makes 4 population is projected to live in cities by mid-century. dedicated use of crops and land. This working paper focuses on A wealthier, more urban global population will likely a consumption-focused solution: shifting people’s diets. - demand more food per capita—and more resource-inten 5 sive foods such as meat and dairy. Without successful Since the 1980s, the World Resources Report has provided decision makers from government, business, and civil society measures to restrain the consumption of resource-inten - with analyses and insights on major issues at the nexus of - sive foods by the world’s affluent or to reduce waste, suf development and the environment. For more information about ficiently feeding the world will require worldwide annual the World Resources Report and to access previous installments crop production in 2050 to be more than 70 percent and editions, visit www.worldresourcesreport.org. 6 higher than 2006 levels. While overall food demand—as measured by crop calo - ries (Box 2)—is projected to rise by roughly 70 percent the food gap, there would be enormous pressure to clear between 2006 and 2050, demand for animal-based foods the world’s remaining tropical forests and other natural is projected to rise even faster. Based on the latest, most ecosystems to expand croplands and pasturelands. For likely population projections, demand for meat and dairy example, to avoid further expansion of harvested area, is projected to grow by nearly 80 percent. Demand for beef the annual average increase in crop yields from 2006 to specifically—one of the most resource- and greenhouse- more than in the 2050 would need to be about one-third gas-intensive foods—is projected to grow by 95 percent 7 previous 44-year period (1962 to 2006)—a period that between 2006 and 2050. Unless curbed, the demand for 8 encompassed the Green Revolution. In addition, increases animal-based products will make it hard to achieve several in food production and the associated land-use changes of the United Nations Sustainable Development Goals, would make it even more difficult to limit global warming including those on hunger, healthy lives, management of to the internationally recognized goals of 1.5 to 2 degrees water, consumption and production, climate change, and 9 Agriculture and related Celsius above preindustrial levels. terrestrial ecosystems. land-use change accounted for nearly one-quarter of global greenhouse gas emissions in 2010. By 2050, they Efforts to feed a growing and increasingly affluent popula - could consume 70 percent of the total allowable global tion have primarily focused on increasing food production, emissions “budget” for limiting global warming to 2 rather than addressing consumption. However, if the 10 degrees. world were to rely solely on increased production to close | 14

15 Shifting Diets for a Sustainable Food Future Figure 1 | A Menu of Solutions is Required to Sustainably Close the Food Gap global annual crop production (kcal trillion) ILLUSTRATIVE Hold down consumption ? Increase production 16,300 9,500 Increase 2006 Increase Expand on to 2050 Shift diets Avoid biofuel Reduce food Achieve livestock & cropland Available food low-environmental competition for loss & waste Baseline available replacement- productivity aquaculture opportunity- food needed crops & land level fertility productivity cost lands Source: WRI analysis based on Bruinsma (2009) and Alexandratos and Bruinsma (2012). Note: Includes all crops intended for direct human consumption, animal feed, industrial uses, seeds, and biofuels. In this paper, we examine three interconnected diet shifts - The world will need to do more than increase food produc tion and the efficiency of production. Given the magnitude that can contribute to a sustainable food future: of the challenge and the environmental impacts associated Reduce overconsumption of calories. with increased production, consumption-focused solutions 1. will also be necessary. One consumption-focused solution 2. Reduce overconsumption of protein by reducing is to shift diets—the type, combination, and quantity of consumption of animal-based foods. food consumed by people. There are good reasons to shift diets aside from the need to close the food gap. One in Reduce consumption of beef specifically. 3. two people worldwide currently consumes a nutritionally imbalanced diet as a result of overconsumption, hunger, This paper is primarily tailored to businesses that can play 11 - Shifting to more nutrition or micronutrient deficiency. a role in shifting diets. This includes actors in the food ally balanced diets could profoundly affect food security, value chain, particularly food service companies and food human health, healthcare costs, natural resource use, the retailers. environmental impacts of agriculture, and animal welfare. | 15 | April 2016 WORKING PAPER

16 We do not suggest that everyone should become a vegan Box 2 | What Metric for Assessing Food or vegetarian. Nor do we aim to reduce food consumption Security and Nutritional Requirements? among undernourished or malnourished populations. We do not seek to eliminate the livestock sector, which provides livelihoods to millions of poor smallholders 12 There is no one perfect metric for assessing global food security and generates 40 percent of global agricultural income. or human nutritional requirements. FAO used economic value as of calories overconsumption Rather, we focus on reducing a metric in its 2009 estimate of a 70 percent food gap between and protein, decreasing the share of animal-based protein a 2006 and 2050. Given that food prices fluctuate, economic in diets, and reducing the consumption of beef specifically. value does not provide a consistent unit of measure over time. Another metric that has been used is food weight or volume (for For each of the three proposed diet shifts, we define the example, tons). Since this includes water, it does not provide a reliable metric of the nutritional content of food. - issues, explain why they matter, and review projected con sumption trends. We then quantify the projected effects Creating a Sustainable Food This paper—and others in the of the shifts on the land use and greenhouse gas impacts series—measures the food gap between 2006 and 2050 Future of agriculture in 2009, using the version of the GlobAgri in crop calories. Calories are consistent over time and avoid model developed for Creating a Sustainable Food Future embedded water. Measuring the calories embedded in crops (production) rather than in foods (consumption) also has the (Box 3). The greenhouse gas emissions estimates include advantage of counting not only crops intended for human - both emissions from agricultural production and emis consumption, but also animal feed, bioenergy, and other uses. sions from land-use change. Regarding land-use change However—critically for this paper—this metric does not emissions, we estimate the emissions that would occur capture the role of pastureland in supporting human food needs. from conversion of forests, savannas, and other lands 13 based on existing crop to produce the additional foods The analysis of the three diet shifts explored in this paper focuses on total calories and one important macronutrient yields, livestock efficiencies, and patterns of trade (Box (protein) because of their key role in nutritional health and the 4). We then assess the effects of the diet shifts in 2050, availability of globally consistent data. Each of the modeled including their potential to help close the food gap. diet scenarios in this paper was designed to ensure adequate amounts of calories and protein to the populations affected Next, we outline a novel approach to address the cru - by the scenarios. Two of the scenarios—vegetarian and Mediterranean-style diets—were based on realistic diets that cial question of to shift people’s diets. We consider how are high in nutrient-rich foods. strategies beyond information and education campaigns, and draw on practices employed by the consumer goods It should be noted, however, that other nutrients are needed industry that have successfully changed consumption for a balanced diet. A narrow focus on calories and protein patterns. We conclude with four recommendations for could lead to nutritionally unbalanced diets. For example, an shifting diets. “environmentally sustainable diet” that sought only to maximize calories produced per hectare could inadvertently encourage production of high-yielding, energy-dense crops, such as In our assessment, the three proposed diet shifts meet sugars and cereals, in place of lower-yielding, but nutrient- the development and environmental criteria introduced rich crops, such as fruits, vegetables, and beans. As a result, in the World Resources Report’s Interim Findings “shifting diets” should be implemented with an eye to providing (Table 1). They reduce the pressure of agriculture on not only adequate amounts of calories and protein, but also all other nutrients essential to human health. This is particularly ecosystems, climate, and water, and offer potential important given that micronutrient deficiencies—or “hidden benefits for human health. hunger”—affected more than 2 billion people in 2010–12, with the most common deficiencies including iodine, iron, zinc, and b vitamin A. Notes: a. Bruinsma (2009). b. FAO, WFP, and IFAD (2012). | 16

17 Shifting Diets for a Sustainable Food Future Box 3 | An Overview of the GlobAgri-WRR Model animal-based foods)—to “credit” that diet This paper uses the GlobAgri-WRR model and levels of food loss and waste—factors for reduced food consumption by others. that can all be modified to examine future (“GlobAgri”), which is a version of the scenarios of agricultural production and GlobAgri model developed by the Centre Patrice Dumas (CIRAD) is the principal de Coopération Internationale en Recherche food consumption. Agronomique pour le Développement architect of the GlobAgri-WRR model, To analyze the eight alternative diet scenarios (CIRAD), Princeton University, the Institut working in partnership with Tim Searchinger of Princeton University and National de la Recherche Agronomique explored in this paper, GlobAgri held all WRI. Other researchers contributing to the (INRA), and WRI. GlobAgri is a global other consumption and production factors constant. For example, the model assumes core model include Stéphane Manceron biophysical model that quantifies the greenhouse gas emissions and land-use (INRA) and Richard Waite (WRI). additional food would be supplied at the effects of agricultural production. It estimates same average crop yields, using the same average livestock production measures, emissions from agricultural production, A major strength of the GlobAgri model and with the same rates of food loss and primarily methane and nitrous oxide, and is that it incorporates other biophysical carbon dioxide from the energy used to submodels that estimate emissions or waste as in the 2009 reference scenario. produce fertilizers and pesticides or to run The model similarly assumes that the role land-use demands in specific agricultural sectors. GlobAgri therefore benefits from farm machinery. It also estimates emissions of imports and exports would remain the other researchers’ work, incorporating same; for example, if 20 percent of a crop from land-use change (Box 4). It does not include emissions from food processing, in Country A is imported from Country B, the highest levels of detail available. that percentage would remain true under Major subcomponents include a livestock transportation, retail, or cooking. model with lead developers Mario Herrero scenarios of altered demand for that crop. (CSIRO) and Petr Havlik (IIASA), with GlobAgri links food consumption decisions The GlobAgri model differs from some in each country or region (see Appendix extra contributions from Stefan Wirsenius (Chalmers University); a land-use model other global agriculture and land-use A for list of countries and regions) to the with lead developer Fabien Ramos of the models in that it does not incorporate production of the crops, meat, milk, or economic feedback effects. For example, European Commission Joint Research fish necessary to meet food demands after accounting for food loss and waste at each if people in one country were to increase Centre; a global rice model with lead developer Xiaoyuan Yan of the Chinese food consumption, the prices of those stage of the value chain from farm to fork. foods would increase, potentially Its core data for production, consumption, Institute for Soil Science; a nitrogen and yields are based on data from FAO triggering changes in food production and emissions model with lead developer Xin consumption in other countries. Economic Zhang of Princeton University; and an (2015). The model accounts for the multiple aquaculture model with lead developer models can be used to simulate these food, feed, and energy products that can feedback effects, but given the uncertainties be generated by each crop, and reflects the Mike Phillips of WorldFish and Rattanawan estimates of crop calorie content by region associated with these interactions, they Mungkung of Kasetsart University. Each of need to make a number of assumptions, these submodels had several contributors. as estimated in FAO (2015). It estimates Information on vegetarian diets was land use and greenhouse gas emissions making the results highly variable. based on information provided by Peter Furthermore, it seems inappropriate—when related to agricultural production in each of Scarborough and Paul Appleby of the the world’s countries in light of crop yields, evaluating the consequences of a resource- University of Oxford. intensive diet (for example, one high in population, diets, production methods, | 17 | April 2016 WORKING PAPER

18 How Does the GlobAgri Model Quantify the Land-Use-Related Box 4 | Greenhouse Gas Effects of Different Food Choices? 2. Land-use-change emissions are constant, meaning that production grew by The GlobAgri model estimates the amount averaged over total agricultural 200,000 tons (from 10 million tons to 10.2 of agricultural land needed to produce production and land use in the million tons). Suppose that these additional a specific quantity of food required by a Under this approach, study period. 100,000 hectares came from clearing forest given diet assuming present crop yields, land-use-change emissions per unit production systems, and trade. It assumes that contained 400 tons of CO e per hectare. 2 of food produced are generally quite In approach (2), the 100,000 hectares that any agricultural expansion triggered by low—and can even be zero if land-use of land-use change and the resulting a change in diet will come at the expense change did not occur during the study emissions of 40 million tons of CO e would of forests, savanna, or some other native 2 time frame. Some of these studies vegetation. The resulting loss of carbon be assigned to all 10.2 million tons of do this calculation just for a specific in plants and soils provides the quantity soybeans, resulting in a small quantity of crop and in a specific country. For e emissions per ton (around 4 tons of CO of greenhouse gas emissions from land 2 example, if soybean area is expanding use attributed to the diet. Conversely, per ton of soybeans produced). in Brazil, but not in the United States, under scenarios of reduced food demand, the emissions from this expanding In fact, if we change this example and GlobAgri estimates negative land-use- soybean area are allocated to all the change emissions (or avoided future assume that soybean yields grew just production of soybeans in Brazil, emissions), simulating the restoration of enough in 2010 to produce 10.2 million but none are assigned to soybean tons of soybeans on the same 5 million agricultural land to native vegetation. The production in the United States. This hectares of land (in other words, with model does not consider how economic approach would assign land-use- feedbacks might alter other demands, no expansion of the soybean area), change emissions to a European this approach would attribute no land- production systems, or yields. pork producer who imports soybeans use change (and no land-use-change from Brazil, but not one who imports emissions) to soybeans. Under approach How have other studies quantified the soybeans from the United States. (2), eating meat fed entirely by soybeans greenhouse gas effects of changes in agricultural land use? from this country has no land-use 3. Land-use-change emissions emissions cost. Informed by Schmidinger and Stehfest are attributed to marginal (2012), we identified three broad (additional) agricultural What approach (2) does not tell us is how approaches that life-cycle assessments of production. This approach— much extra land use was required for each agriculture and/or alternative diets generally employed by GlobAgri and most additional ton of soybeans in 2010. In the use to quantify the greenhouse gas effects biofuel studies—focuses on the first example above, if soybean demand associated with changes in land use: additional emissions from the had not gone up by 200,000 tons in 2010, additional land required to produce soybean area would not have expanded at 1. Land-use-change emissions are any additional amount of a crop or all. The increase of 200,000 tons required not estimated. Most conventional other food. Under this approach, land- the extra 100,000 hectares. Approach (3) life-cycle assessments of agriculture use-change emissions per unit of food therefore assigns the land-use-change account for the land area required to produced are much higher than in emissions from converting those 100,000 produce the foods being studied, but approach (2), and are never zero. hectares only to the 200,000 additional do not assign any land-use-change- tons of soybeans. In this example, related greenhouse gas emissions to A hypothetical example illustrates the therefore, each additional ton of soybeans that land. Estimates of greenhouse difference between approach (2) and is responsible for emitting 200 tons of gas emissions in such studies are approach (3). Assume that in 2009, e. Recognizing that land converted into CO 2 limited to sources from agricultural soybeans occupied 5 million hectares of agricultural production can sustain crops production (not land-use change), land in a country and produced 2 tons or livestock over many years, we follow such as methane from livestock and per hectare, for a total production of 10 the approach taken by the European Union energy used to run farm machinery. million tons. Each ton of soybeans therefore for estimating land-use-change emissions required half a hectare of cropland. for biofuels, and amortize the land-use- Now assume that in 2010, soybean area change emissions over 20 years. In our expanded by 100,000 hectares to 5.1 example, this amortization therefore assigns million hectares, while yields stayed e per year per emissions of 10 tons of CO 2 | 18

19 Shifting Diets for a Sustainable Food Future Box 4 | How Does the GlobAgri Model Quantify the Land-Use-Related Greenhouse Gas Effects of Different Food Choices? (continued) additional ton of soybeans produced. emissions. Regardless of what anyone to an increase in beef and grain prices and GlobAgri applies this “marginal” approach else does, individual dietary choices could cause a poorer person elsewhere to consume less grain or beef as prices (3) in this paper in order to gain a fuller affect demand for land “at the margin” increase. This “crowding out” of the poorer picture of the land and greenhouse gas and therefore have a significant effect on person’s food demand does reduce global greenhouse gas emissions. consequences of diet shifts, which by land-use demands relative to a world in definition happen at the margin. For any Why does GlobAgri not consider the which the poorer person’s grain demand is given yield, each additional ton of food economic effects of changes in food demanded requires a certain amount of unaffected by the wealthier person’s beef additional land, which results in a certain demand. However, it seems inappropriate, supply and demand? when evaluating the consequences of a amount of land-use-change emissions. The The GlobAgri model is designed to answer converse is also true—each ton of food no resource-intensive diet, to “credit” that the question of how much agricultural longer demanded (e.g., under scenarios diet for reduced food consumption by land would be required for a given level others, even if this does lead to potential that reduced consumption of animal-based of food demand, crop yield, and livestock overestimates of the greenhouse gas foods and as a result reduced demand for production efficiency. It does not assess crop-based feed) results in a decrease benefits of reducing consumption. By whether and by how much an increase or in agricultural land use and negative eliminating the economic feedback effects decrease in food demand by one group from the analysis, GlobAgri can more land-use-change emissions if agricultural of people leads to price changes and land reverts to native vegetation. However, transparently estimate what combinations of economic feedbacks that in turn shift other because global agricultural land is diet shifts, yield gains, and other solutions people’s demands, farmers’ yields, and/or would be necessary to achieve a sustainable expanding—as food demand growth production techniques. Such economic food future. continues to outpace yield growth—the effects are highly complex and uncertain. real-world consequences of reducing food As a result, there is limited underlying Even without considering the economic demand under the scenarios modeled in economic evidence that can be used interactions of changes in food supply this paper would be to avoid future land-use a In addition, to robustly estimate them. and demand, there are still many other change. This avoided land-use change and economic assessments do not fully associated greenhouse gas emissions are uncertainties in estimating both the type of capture the land-use “opportunity costs” lands that are likely to be converted and the what GlobAgri estimates. of diet choices. Consider, for example, quantity of carbon that conversion would a scenario in which a relatively wealthy release. All model estimates, including This approach reveals that each person’s person increases their beef consumption. those of GlobAgri, should therefore be diet matters quite a lot for agricultural land This change in demand would likely lead considered rough. use and the associated greenhouse gas Note: a. Searchinger et al. (2015) and supplement; Berry (2011). | 19 | April 2016 WORKING PAPER

20 Table 1 | Criteria Sustainable Food Future How “Shifting Diets” Performs Against the = positive = neutral/it depends PERFORMANCE COMMENT DEFINITION CRITERIA Reduces poverty and advances Careful policy choices would be needed to ensure that the diet Pover ty rural development, while still shifts make food affordable to all, and do not adversely impact poor alleviation a being cost effective livestock farmers. Gender Generates benefits for women The diet shifts would provide health benefits for women. The diet shifts would reduce the land needed for food production. Reduces pressure for agricultural expansion and Reducing consumption of animal-based food products, particularly Ecosystems intensification on existing beef and dairy, would lower the pressure to convert forests and wooded agriculture land savannas into pasturelands. b Shifting diets The diet shifts would contribute to stabilizing the climate. would reduce the need to convert forests to crop and pastureland, apply more fertilizers, and raise more livestock. Reducing overconsumption Reduces greenhouse gas of calories would reduce the need for energy for producing, processing, emissions from agriculture transporting, and storing food. Climate to levels consistent with Reducing beef consumption specifically would reduce methane emissions stabilizing the climate from enteric fermentation and manure and nitrous oxide (N O) from excreted 2 nitrogen and the chemical nitrogenous fertilizers used to produce feed for animals kept in feedlots. The diet shifts would reduce the quantity of water needed for food Reduces water consumption Water production. They would also reduce the contribution of agriculture and pollution to water pollution. Notes : a. USDA/HHS (2015). b. Hedenus et al. (2014). | 20

21 Shifting Diets for a Sustainable Food Future Unlike many other studies (Box 4), the comparison of food CONVERGING DIETS types in Figure 2 incorporates both land used for pasture Around the world, eating habits are converging toward and greenhouse gas emissions associated with changes in Western-style diets high in refined carbohydrates, added land use. Key findings from this more inclusive approach, sugars, fats, and animal-based foods. Consumption of using the GlobAgri model, include: 14 other vegetables, coarse grains, and fiber is pulses, 15 Three interconnected global trends are asso declining. - Animal-based foods are generally more resource- ▪ ciated with this convergence. First, rising incomes are intensive and environmentally impactful to produce associated with rising demand for animal-based foods, than plant-based foods. 16 Second, increasing vegetable oils, and added sugars. urbanization (also associated with rising incomes) pro - Beef and other ruminant meats (sheep and goat) are ▪ vides easy access to supermarkets, restaurants, fast food by far the most resource-intensive of foods, requiring chains, and foods that they supply, including meat, dairy, and four to six times more land and generating that many 17 Third, technological advances, processed foods and drinks. times more greenhouse gas emissions than dairy per business and economic changes, and government policies are calorie or unit of protein ultimately consumed by - transforming entire food chains, from farm to fork. Multi people. Beef and other ruminants also require more national agribusinesses, food manufacturers, retailers, than 20 times more land and generate more than 20 and food service companies increasingly influence what is times more greenhouse gas emissions than pulses per grown and what people eat, a trend that is spreading from unit of protein consumed. 18 high-income to low- and middle-income countries. Dairy’s land use and greenhouse gas emissions are ▪ These trends—combined with more sedentary lifestyles— slightly higher than those of poultry per calorie con - affect nutritional and health outcomes, including height, sumed and significantly higher than those of poultry weight, and the prevalence of noncommunicable dis - per unit of protein consumed. 19 Diet-related noncommunicable diseases include eases. hypertension, type 2 diabetes, cardiovascular diseases, Poultry and pork have similar greenhouse gas emis - 20 ▪ and certain types of cancer. sions and land use per unit of protein consumed, but poultry’s land use and emissions are higher than FAO food supply data—adjusted downward from “per pork’s per calorie consumed mainly because of the 21 capita food availability” to “per capita food consumption” high energy content of pork fat. - to account for food loss and waste during the consump tion stage of the food supply chain—indicate that the - Pulses, fruits, vegetables, and vegetable oils are gener ▪ consumption of calories and protein is already above average ally more resource-intensive to produce than sugars requirements in the majority of developed countries. Per and staple crops, but still compare very favorably to capita food consumption also is rapidly rising in emerging animal-based foods. 22 In this paper, we economies, including China and Brazil. 23 to estimate food consumption build on FAO’s projections Factoring land-use implications into agricultural ▪ levels in 2050. We find that in 2050, emerging economies greenhouse gas emissions estimates shows a fuller will likely exhibit per capita consumption levels—in terms picture of the consequences of people’s dietary choices. of calories, protein, and consumption of animal-based For all food types, the annualized emissions from land- 24 foods—comparable to today’s developed countries. use change (shown in orange in Figure 2) are far higher than emissions associated with agricultural production Why does this global convergence in diets matter? Foods (shown in light orange). For example, when considering differ vastly in terms of the quantity of land, water, and production emissions only, consumption of a million energy needed per unit of energy and protein ultimately calories of beef would generate 19 tons of CO e, while 2 consumed, and in terms of their greenhouse gas impacts the same quantity of pulses would generate 0.4 tons of 25 (Figure 2). Although the data in Figure 2 are global e, a savings of 18.6 tons of CO CO e. But when factoring 2 2 26 —masking means for current agricultural production e in land use, emissions would fall from 201 tons of CO 2 variations among locations, production systems, and e (for pulses), a savings of 194 (for beef) to 7 tons of CO 2 farm management practices (Box 5)—they enable general e or more than 10 times the amount when tons of CO 2 comparisons across food types. considering only production-related emissions. | 21 | April 2016 WORKING PAPER

22 PR OTEIN CONSUMED CAL ORIE CONSUMED Foods Differ Vastly in Resource Use and Environmental Impacts Figure 2 | ON PR OTEIN CONSUMED ORIES CONSUMED OCAL PER T PER MILLION KIL ANIMAL-B -BASED PLANT ASED ASED -BASED PLANT ANIMAL-B 3 3 1, 000 m t CO e t CO e 1, 000 m ha ha 2 2 120 3,000 250 180 15 10 LAND USE (ha) LAND USE (ha) Pasture Pasture Cropland Cropland 3 3 TER CONSUMPTION (1,000 m FRESHWA ) ) FRESHWA TER CONSUMPTION (1,000 m ater Rainw ater Rainw 100 2,500 150 ation Irrig ation Irrig 12 8 200 GHG EMISSIONS (t CO e) e) GHG EMISSIONS (t CO 2 2 hang e Land-use c e hang Land-use c Agricultural production Agricultural production 2,000 80 120 9 6 150 60 1,500 90 6 4 100 40 1,000 60 3 50 2 500 20 30 0 0 Pulses Rice Maize Roots & Wheat Beef y Wheat Dair Poultr Pork Eggs y Fish Rice Rapeseed y Beef Pulses Ro ots & Sug Fish ar Maize Eggs Pork Poultry Dair So ybean Oil Fruits & Sunflow er med) (far Tubers & Mustard Tubers med) (far Seed Oil Vegetabl es Seed Oil Sources: GlobAgri model (land use and greenhouse gas emissions), authors’ calculations from Mekonnen and Hoekstra (2011, 2012) (freshwater consumption), and Waite et al. (2014) (farmed fish freshwater consumption). Data presented are global means. Entries are ordered left to right by amount of total land use. Indicators for animal-based foods include resource use to produce feed, including pasture. Notes: Tons of har vested products were converted to quantities of calories and protein using the global average edible calorie and protein contents of food types as reported in FAO (2015). “Fish” includes all aquatic animal products. Freshwater use for farmed fish products is shown as rainwater and irrigation combined. Land use and greenhouse gas emissions estimates are based on a marginal analysis (i.e., additional agricultural land use and emissions per additional million calories or ton of protein consumed). Based on the approach taken by the European Union for estimating emissions from land-use change for biofuels, land-use change impacts are amortized over a period of 20 years and then shown as annual impacts. Land use and greenhouse gas emissions estimates for beef production are based on dedicated beef production, not beef that is a coproduct of dairy. Dairy figures are lower in GlobAgri than some other models because GlobAgri assumes that beef produced by dairy systems displaces beef produced by dedicated beef-production systems. | 22

23 Shifting Diets for a Sustainable Food Future CAL ORIE CONSUMED OTEIN CONSUMED PR Figure 2 | Foods Differ Vastly in Resource Use and Environmental Impacts (continued) PER T ON PR OCAL OTEIN CONSUMED PER MILLION KIL ORIES CONSUMED PLANT ASED ANIMAL-B ASED -BASED ANIMAL-B -BASED PLANT 3 3 1, 000 m e t CO t CO e ha 1, 000 m ha 2 2 3,000 120 250 180 15 10 LAND USE (ha) LAND USE (ha) Pasture Pasture Cropland Cropland 3 3 ) TER CONSUMPTION (1,000 m FRESHWA FRESHWA TER CONSUMPTION (1,000 m ) ater Rainw Rainw ater 100 2,500 150 ation Irrig Irrig ation 12 8 200 GHG EMISSIONS (t CO e) GHG EMISSIONS (t CO e) 2 2 hang e Land-use c hang Land-use c e Agricultural production Agricultural production 80 2,000 120 9 150 6 60 1,500 90 6 4 100 40 1,000 60 3 2 50 500 20 30 0 0 Dair Pork Eggs Rice Maize Roots & y Wheat Wheat y Beef Fish Pulses Poultr Rapeseed Rice Poultry Dair y Beef Pulses Ro ots & Sug Fish ar Maize Eggs Pork So ybean Oil Fruits & Sunflow er (far med) Tubers & Mustard med) Tubers (far es Vegetabl Seed Oil Seed Oil Sources: GlobAgri model (land use and greenhouse gas emissions), authors’ calculations from Mekonnen and Hoekstra (2011, 2012) (freshwater consumption), and Waite et al. (2014) (farmed fish freshwater consumption). Notes: Data presented are global means. Entries are ordered left to right by amount of total land use. Indicators for animal-based foods include resource use to produce feed, including pasture. Tons of har vested products were converted to quantities of calories and protein using the global average edible calorie and protein contents of food types as reported in FAO (2015). “Fish” includes all aquatic animal products. Freshwater use for farmed fish products is shown as rainwater and irrigation combined. Land use and greenhouse gas emissions estimates are based on a marginal analysis (i.e., additional agricultural land use and emissions per additional million calories or ton of protein consumed). Based on the approach taken by the European Union for estimating emissions from land-use change for biofuels, land-use change impacts are amortized over a period of 20 years and then shown as annual impacts. Land use and greenhouse gas emissions estimates for beef production are based on dedicated beef production, not beef that is a coproduct of dairy. Dairy figures are lower in GlobAgri than some other models because GlobAgri assumes that beef produced by dairy systems displaces beef produced by dedicated beef-production systems. | 23 | April 2016 WORKING PAPER

24 The global convergence toward Western-style diets, high Box 5 | Improving Agricultural Productivity in animal-based foods, has enormous implications for the for a Sustainable Food Future resource needs and environmental impacts of agriculture. 27 is a case in point. The average diet of the United States Figure 3 compares the average daily diets of the world It is not just the type of food consumed that determines and the United States in 2009 based on the number of the environmental and resource use impacts of agriculture, - calories of each food type consumed, as well as the asso but also the way that food is produced. The GlobAgri ciated land-use needs and greenhouse gas emissions. In model incorporates the wide diversity of crop, livestock, the United States, the average daily diet contained nearly and aquaculture systems in all of the world’s regions. For instance, it includes estimates for the quantity of nitrogen 500 more calories than the average world diet, including used by type of crop and region. It also incorporates nearly 400 additional animal-based calories. As a result, regional estimates of the greenhouse gas emissions the agricultural land use and greenhouse gas emissions as - generated from the energy used to run farm machinery sociated with the average daily US diet were almost double and to produce and apply pesticides. 28 those associated with the average daily world diet. Animal-based foods accounted for nearly 85 percent of the World agricultural productivity has increased over time, as advances in farming technology and practices have production-related greenhouse gas emissions and nearly boosted crop and pasture yields in some places. While 90 percent of agricultural land use associated with the those advances have generally increased chemical, water, 29 Shifting the diets of people who current - average US diet. and energy inputs, there have also been substantial ly consume high amounts of calories and animal-based a improvements in input efficiency in recent decades. foods, and those projected to by 2050, could therefore Overall, modern livestock systems—particularly beef and dairy—generate fewer greenhouse gas emissions and use significantly reduce agriculture’s impact on resources and less land than traditional production systems, especially the environment. b when coupled with gains in crop yield for animal feeds. As noted above, factoring land-use implications into As discussed in other Creating a Sustainable Food c agricultural greenhouse gas emissions estimates shows crop and livestock productivity can installments, Future a fuller picture of the consequences of people’s dietary be significantly increased without increasing inputs or shifting to the most concentrated feedlot systems. For choices. For example, if an additional person eating the example, Brazil’s National Plan for Low Carbon Emissions average daily US diet were added to the world population in Agriculture aims to increase input efficiency, boost in 2009, the one-time emissions resulting from convert - productivity, and reduce agricultural greenhouse gas ing a hectare of land to agriculture to feed that person emissions and other environmental impacts by intensifying 30 would be about 300 tons CO e. This amount is equal to production on degraded pasturelands and through 2 integrated crop-livestock-forest systems, no-till 17 times the average US per person energy-related carbon d 31 farming, and other practices. dioxide emissions in 2009. In other words, the emissions from clearing additional land to feed an additional person Still, although there are tradeoffs among production systems, eating the US diet are equal to 17 years of that person’s and agricultural productivity will likely continue to increase, energy-related CO emissions. reducing overconsumption is likely to generate environmental, 2 resource-use, health, and other benefits, regardless of the production systems employed. This is especially true for If diets around the world continue to converge, more animal-based foods, which are resource intensive. people will be consuming more food that is more resource intensive and more environmentally impactful to produce, Notes: posing significant challenges to a sustainable food future. a. For a global over view, see Garnett et al. (2015). For progress in 32 nitrogen use efficiency in some developed countries in recent years, see below that We examine three interconnected diet shifts Lassaletta et al. (2014). For progress in water use efficiency in the US, can address these challenges: see Schaible and Aillery (2012). b. Herrero et al. (2013). 1. Reduce overconsumption of calories. c. See, for example, installments on improving crop breeding (Searchinger et al. 2014), improving land and water management (Winterbottom Reduce overconsumption of protein by reducing 2. et al. 2013), limiting crop expansion to lands with low environmental opportunity costs (Hanson and Searchinger 2015), reducing greenhouse consumption of animal-based foods. gas emissions and water use from rice production (Adhya et al. 2014), improving productivity of aquaculture (Waite et al. 2014), and increasing Reduce consumption of beef specifically. 3. productivity of pasture and grazing lands (Searchinger et al. 2013). d. Marques (2013). | 24

25 Shifting Diets for a Sustainable Food Future Figure 3 | Land Use and Greenhouse Gas Emissions DIET SHIFT 1: REDUCE Associated with the Average US Diet Were OVERCONSUMPTION OF CALORIES Nearly Twice the World Average The first diet shift aims to reduce overconsumption of per capita values, 2009 calories. Overconsumption of calories occurs when dietary calorie intake exceeds estimated energy requirements for - an active and healthy life. Unnecessary calorie consump Plant-Based Foods Other Animal-Based Foods Dairy Beef tion results in unnecessary use of inputs (e.g., land, water, energy) and unnecessary environmental impacts related WORLD (REFERENCE) US (REFERENCE) to the production of the excess calories. This diet shift targets countries and populations with high calorie intake now and those with high projected calorie intake by 2050. DAILY FOOD It would help reduce the number of obese and overweight 2,433 2,904 CONSUMPTION people, and could result in significant potential savings in (KCAL) healthcare costs. What is the issue with 0.96 0.49 overconsumption of calories? The number of obese and overweight people is high and growing. In 2013, 2.1 billion people were overweight or AGRICULTURAL 33 —more than two and a half times the number of obese LAND USE 34 (HECTARES) chronically undernourished people in the world. Glob - ally, 37 percent of adults over the age of 20 were over - 35 weight in 2013 and 12 percent were obese. 1.5 According to FAO, the global average daily energy require - 1.4 36 although ment for an adult is 2,353 calories per day, 1.2 individual energy requirements depend on age, sex, height, weight, level of physical activity, and pregnancy 0.8 0.9 GHG EMISSIONS 37 In 2009, however, per capita calorie con or lactation. - FROM AGRICULTURAL 0.6 sumption exceeded this average requirement in regions PRODUCTION 38 E) (TONS CO 2 containing half of the world’s population (Figure 4). 0.3 When people persistently overconsume calories, they can become overweight or obese. 0 A number of efforts have sought to explain the global rise - in the number of people who are obese or overweight. Fac 20 tors identified include increased consumption of energy- 15.2 dense foods that are high in fat, decreased physical activity 15 as a result of increasingly sedentary work, and changing GHG EMISSIONS modes of personal transportation, which are all associ - 10 FROM LAND-USE 39 These factors are ated with increasing urbanization. 7.6 CHANGE E) (TONS CO compounded by increased access to low-cost convenience 2 5 and processed foods and sugar-sweetened beverages, increased dining out, persuasive marketing by food and 0 beverage companies, and government subsidies for food 40 GlobAgri model. Source: production that reduce the cost of food to consumers. Note: “US” data are for United States and Canada. Land-use change emissions are amortized over a period of 20 years and then shown as annual impacts. Calculations assume global average efficiencies (calories produced per hectare or per ton of CO e 2 emitted) for all food types. “Other animal-based foods” includes pork, poultry, eggs, fish (aquatic animals), sheep, and goat. 25 | | April 2016 WORKING PAPER

26 Figure 4 | Average Per Capita Calorie Consumption Exceeds Average Daily Energy Requirements in Regions Containing Half of the World’s Population kcal/capita/day, 2009 Asia China Latin America US & Middle East & India (ex. China & India) Canada (ex. Brazil) North Africa 3,500 European Former Brazil OECD (other) Sub-Saharan Union Soviet Union Africa 3,000 2,500 Average daily energy requirement 2,000 1,500 1,000 500 0 0 2 7 1 6 3 4 5 Population (billions) Source: GlobAgri model with source data from FAO (2015) and FAO (2011a). Width of bars is proportional to each region’s population. Average daily energy requirement of 2,353 kcal/capita/day is given in FAO (2014). Individuals’ energy requirements vary depending on age, sex, height, weight, pregnancy/lactation, and level of physical activity. type 2 diabetes alone (for which obesity is a risk factor) Why does it matter? accounted for 13 percent of healthcare costs in China in The high and rising incidence of overconsumption 46 2010. of calories matters because it leads to “unnecessary” consumption of calories and the associated land, water, - Obesity and overweight can negatively impact productiv energy, chemicals, and other inputs that go into producing ity. Obese and overweight people have higher absenteeism the calories. rates than healthy people, and are at risk of loss of future 47 Direct costs (of obesity- income from premature death. Obesity and overweight also take a toll on human health. related medical treatment) and indirect costs (related Obesity is a risk factor for several noncommunicable to lost productivity) accounted for EUR 32.8 billion in - diseases, including hypertension, type 2 diabetes, cardio 48 Europe in 2002. In the United States alone, a 2006 - vascular diseases, and certain types of cancer (endome medical expenditure panel survey and a 2008 national 41 Obesity also increases the risk of trial, breast, and colon). - health and wellness survey estimated that the cost of obe 42 premature death. sity among full-time workers was $73.1 billion per year as a result of decreased productivity, sick days, and general The health impacts of obesity and overweight drive up 49 In another study, the economic cost medical expenses. healthcare costs. According to one OECD study, obese of lost productivity from obesity-related early mortality people on average incur 25 percent higher healthcare costs was estimated at $49 billion per year in the United States 43 than a person of normal weight. In the United States, 50 A 2009 World Economic Forum survey of and Canada. the healthcare costs of obesity accounted for 12 percent of business executives identified noncommunicable diseases, 44 the growth in health spending between 1987 and 2001. such as those associated with diet, as a leading threat to In absolute terms, US obesity-related healthcare costs 51 global economic growth. 45 were approximately $147 billion in 2008. Spending on | 26

27 Shifting Diets for a Sustainable Food Future Combining the costs of healthcare, lost productivity, and from around 29 percent to 37 percent for men and from 54 30 percent to 38 percent in women. investments in obesity prevention and mitigation, the The number of McKinsey Global Institute estimated the worldwide eco - obese or overweight children and adolescents also grew nomic impact of obesity in 2012 to be around $2.0 trillion, in developed countries, with nearly one in four now obese or 2.8 percent of global gross domestic product (GDP). or overweight. In Tonga, Samoa, and Kuwait, more than half of all adults were obese in 2013. More than half of the This economic impact was roughly equivalent to the cost 52 world’s obese individuals lived in 10 countries in 2013: the of armed conflict or smoking. United States, China, India, Russia, Brazil, Mexico, Egypt, 55 Germany, Pakistan, and Indonesia. What are the trends? The global trend over the past five decades has been Emerging economies like China and Brazil have shown toward greater per capita availability of calories, meaning growth in caloric availability over several decades, climb - the calories available per person at the consumption stage ing above the world average and surpassing some devel - without taking into account food wasted during the con - oped countries. Looking out to 2050, caloric availability sumption stage. Figure 5 shows historical trends between is projected to continue rising, though more strongly in 1961 and 2011, and projections to 2050, for a range of 56 China than in Brazil. Once considered a high-income 53 countries and regions. country problem, the numbers of obese or overweight people is now rising in low- and middle-income countries, The number of people who are obese or overweight is particularly in urban areas—although obesity is on the rise rising globally and reaching epidemic proportions in some 57 In in rural areas and among poor populations as well. countries. Between 1980 and 2013, the number of adults China, the obesity rate tripled from 1991 to 2006. The rate 2 worldwide with a BMI of 25 kg/m or greater increased Figure 5 | Per Capita Calorie Availability is on the Rise kcal/capita/day 4,000 U a es St t d ite n n o i n U opean Eur 3,500 a z i l Br i n a h C W O RL D a i r g e N i 3,000 a ndonesi I Jap n a ndi I a a h t E opi i 2,500 2,000 1,500 1,000 2040 2020 2050 1961 2030 2010 2000 1990 1980 1970 Sources: FAO (2015) for historical data 1961–2011, Alexandratos and Bruinsma (2012) for 2050 projection, linear interpolation from 2012–2050. | 27 | April 2016 WORKING PAPER

28 70 of increase in overweight adults in China has been one of In 2009, the at the consumption stage are excluded. the most rapid in the world, faster even than in the United average daily per capita food energy consumption in 79 58 In China, roughly 350 million people—more States. countries and territories was lower than average energy - requirements; the majority of these countries and territo than a quarter of the adult population—are overweight or 71 ries were in sub-Saharan Africa and Asia. obese. This is twice as many as those who are undernour - 59 In Brazil, obesity rates tripled among men and ished. 60 almost doubled among women from 1973 to 2003. In What would be the land and greenhouse addition, rates of obesity in Brazil have increased dispro - gas benefits of reducing overconsumption 61 portionately among low-income groups. of calories? Obesity can increase even in countries that continue to have Reducing overconsumption of calories could reduce high levels of child stunting from insufficient nutrition. agricultural resource use and environmental impacts. We In Egypt, South Africa, and Mexico, adult obesity rates of used the GlobAgri model to determine the effects of reduc - more than 30 percent coexist with child stunting rates of 30 ing overconsumption of calories on agricultural land use 62 percent, 23 percent, and 15 percent, respectively. and greenhouse gas emissions in 2009. By modifying food consumption data for 2009 (the “reference” scenario), Obesity rates typically grow with a country’s wealth until we conducted two types of analysis. First, we quantified 63 annual incomes reach roughly $5,000 per person. After the per person effects of applying two calorie-reduction that, other factors drive further rises in obesity. In China, scenarios (described below) to the consumption pattern of for example, over the period 1989–2006, low socioeco - a high-consuming country—the United States (Figure 6). nomic status was associated with women being over - Second, we quantified the global effects of the two calorie- weight, while high socioeconomic status remained a risk reduction scenarios across all of the world’s regions (Table 64 72 factor for men being overweight. Globally, the rate of When applied at the global level, the two calorie- 2). increase of overweight and obesity slowed between 2003 reduction scenarios altered the diets of the 680 million and 2013 relative to the previous decade, especially in people who were obese and the 1.4 billion people who 73 developed countries, offering hope that obesity rates may were overweight in 2009. 65 have peaked in some countries. One study in the United States found evidence that, among certain populations, ELIMINATE OBESITY AND HALVE OVERWEIGHT SCENARIO. ▪ further income increases (above approximately $30,000 - Relative to the reference scenario, obesity is elimi per person) were associated with declining obesity rates, nated and the number of people overweight is halved. possibly due to the ability to afford healthier foods and Informed by FAO (2004) and Hall et al. (2011a), this 66 increase physical activities. scenario assumes that an obese person on average consumes 500 more calories per day than a person Although the global trend is toward overconsumption of eating the average daily energy requirement, and that calories, many people remain hungry; an estimated 795 each overweight person on average consumes 250 million people in 2014–16 were chronically undernour - more calories per day than the average daily energy 67 74 Sub-Saharan Africa had the highest prevalence of ished. requirement of people with sedentary lifestyles. chronic hunger at 23 percent. The highest absolute number of undernourished people—more than half a billion—was HALVE OBESITY AND OVERWEIGHT SCENARIO. Relative to ▪ 68 For example, Ethiopia’s per capita concentrated in Asia. the reference scenario, the number of people that are - calorie availability in 2011 was barely more than 2,000 calo obese and overweight are both halved, using the same ries. Because this number is below the average daily energy assumptions as the Eliminate Obesity and Halve requirement—and because “availability” figures include Overweight scenario. food loss and waste that are not ultimately consumed—this low level suggests significant levels of hunger in Ethiopia. In both scenarios, calorie consumption was reduced across Data suggest that 36 percent of Ethiopia’s population was all food types by equal proportions—that is, if in a given 69 undernourished in 2010–12. Looking out to 2050, Ethio - region cereal consumption was reduced by 3 percent, pia is predicted to have more than 2,700 calories available the consumption of all other food types (such as sugars, per person per day. This exceeds the average daily energy vegetable oils, and animal-based foods) was also reduced requirement, but not by much when food loss and waste by 3 percent. At the global level, calorie consumption was | 28

29 Shifting Diets for a Sustainable Food Future Figure 6 | Reducing Overconsumption of Calories Reduces the Agricultural Land Use and Greenhouse Gas Emissions Associated with the Average US Diet by 4 to 6 Percent per capita values, 2009 Plant-Based Foods Other Animal-Based Foods Dairy Beef US (ELIMINATE OBESITY US (HALVE OBESITY US (REFERENCE) & HALVE OVERWEIGHT) & OVERWEIGHT) DAILY FOOD CONSUMPTION 2,796 2,726 2,904 (KCAL) 0.93 0.96 0.90 AGRICULTURAL LAND USE (HECTARES) 1.5 1.4 1.3 1.3 1.2 0.9 GHG EMISSIONS FROM AGRICULTURAL 0.6 PRODUCTION (TONS CO E) 2 0.3 0 20 15.2 14.7 14.3 15 GHG EMISSIONS 10 FROM LAND-USE CHANGE (TONS CO E) 2 5 0 Source: GlobAgri model. “US” data are for United States and Canada. Land-use change emissions are amortized over a period of 20 years and then shown as annual impacts. Calculations assume global average Note: efficiencies (calories produced per hectare or per ton of CO e emitted) for all food types. “Other animal-based foods” includes pork, poultry, eggs, fish (aquatic animals), sheep, and goat. 2 | 29 | April 2016 WORKING PAPER

30 decrease by 90 million hectares in the Eliminate Obesity and Halve Obesity reduced by 3 percent under the scenario and by 2 percent under the Halve Overweight scenario and by 140 million hectares in and Overweight scenario. scenario. Eliminate Obesity and Halve Overweight Halve Obesity and Overweight the Greenhouse gas emissions from agricultural production The effects of the two scenarios on the agricultural land would decrease by 2 percent in the Halve Obesity and scenario and 3 percent in the Eliminate use and greenhouse gas emissions associated with the Overweight average diet of a high-consuming country—the United Obesity and Halve Overweight scenario. In addition, the avoided future emissions from land-use change— States—are shown in Figure 6. The US (Reference) bars show the agricultural land use and greenhouse gas assuming diet changes were sustained over time—would be 19.9 billion tons CO Halve Obesity and emissions associated with the average daily US diet in e in the 2 2009. The other two bars show how the two scenarios scenario and 34.6 billion tons CO Overweight e in the 2 75 scenario. Eliminate Obesity and Halve Overweight To would reduce per capita calorie consumption—as well as the associated greenhouse gas emissions—by 6 percent put this reduction in perspective, global greenhouse gas 76 emissions in 2009 were 44 billion tons CO e. relative to reference under the Eliminate Obesity and 2 Halve Overweight scenario and by 4 percent under the scenario. Although not quantified here, reducing overconsumption Halve Obesity and Overweight of calories would likely generate significant human health benefits by contributing to a reduction in the incidence of Table 2 shows the global-level effects of the two scenarios on agricultural land use and greenhouse gas emissions overweight and obesity in the population. in 2009. Relative to the 2009 reference, land use would Global Effects of Reducing Overconsumption of Calories on Table 2 | Agricultural Land Use and Greenhouse Gas Emissions in 2009 REDUCTION IN GHG AVOIDED FUTURE GHG EMISSIONS FROM REDUCTION IN AGRICULTURAL EMISSIONS FROM A SCENARIO B AGRICULTURAL (MILLION HA) AGRICULTURAL LAND USE LAND-USE PRODUCTION C (MILLION TONS CO (MILLION TONS CO E) CHANGE E) 2 2 ELIMINATE OBESITY AND 84 Pastureland HALVE OVERWEIGHT 54 Cropland 194 34,564 APPLIED TO 1,385 M PEOPLE TOTAL 138 HALVE OBESITY AND Pastureland 56 OVERWEIGHT Cropland 36 19,908 126 APPLIED TO 1,046 M PEOPLE TOTAL 92 GlobAgri model. Source: Notes: a. Reference scenario included a world population of 6.8 billion, agricultural land use of 5 billion hectares (3.4 billion hectares of pastureland and 1.6 billion hectares of cropland), and 6.9 billion tons of greenhouse gas emissions from agricultural production. b. “Cropland” includes land for aquaculture farms. c. These estimates assume that the diet changes are sustained over time. If other improvements to the food system (e.g., yield gains) allowed the world to avoid future land-use change, these scenarios would allow some existing agricultural lands to revert to native vegetation and sequester the equivalent amount of carbon. | 30

31 Shifting Diets for a Sustainable Food Future What is the issue with DIET SHIFT 2: REDUCE overconsumption of protein? OVERCONSUMPTION OF PROTEIN Many people—especially in rich countries—consume more BY REDUCING CONSUMPTION protein than they need. Furthermore, the share of animal- OF ANIMAL-BASED FOODS based protein—including meats, dairy, fish, and eggs—is 77 Animal-based protein sources are growing in diets. The second diet shift aims to reduce overconsumption of generally more environmentally impactful and resource protein by reducing consumption of animal-based foods and 78 intensive to produce than plant-based sources (Figure 2). increasing the proportion of plant-based protein in diets. Overconsumption of protein occurs when consumption exceeds In much of the world, protein consumption exceeds estimated dietary requirements. As with overconsumption - estimated dietary requirements. The average daily pro of calories, overconsumption of protein results in 79 tein requirement for adults is around 50 grams per day, unnecessary use of inputs and unnecessary environmental although individual requirements vary, as they do for impacts related to the production of the excess protein. 80 energy. However, in 2009, global average per capita protein consumption was approximately 68 grams per day—or 36 - This diet shift targets countries and populations that cur 81 percent higher than the average daily adult requirement. rently overconsume protein and consume high amounts In the world’s wealthiest regions, protein consumption was of animal-based protein—or are projected to by 2050. It higher still (Figure 7). In more than 90 percent of the world’s does not target undernourished or malnourished people, countries and territories, average daily per capita protein nor does it seek to eliminate animal-based food consump - 82 consumption exceeded estimated requirements in 2009. tion, recognizing that livestock production is an important source of livelihood and income. Protein Consumption Exceeds Average Estimated Daily Requirements in Figure 7 | All the World’s Regions, and is Highest in Developed Countries g/capita/day, 2009 Animal-based protein Sub-Saharan OECD (other) Former US & Canada Brazil Plant-based protein Africa Soviet Union 100 India European Middle East & China Latin America Asia (ex. Brazil) Union North Africa (ex. China & India) 90 80 70 60 Average daily 50 protein requirement 40 30 20 10 0 4 5 6 0 1 2 3 7 Population (billions) GlobAgri model with source data from FAO (2015) and FAO (2011a). Width of bars is proportional to each region’s population. Average daily protein requirement of 50 g/day is based Source: on an average adult body weight of 62 kg (Walpole et al. 2012) and recommended protein intake of 0.8 g/kg body weight/day (Paul 1989). Individuals’ energy requirements vary depending on age, gender, height, weight, pregnancy/lactation, and level of physical activity. | 31 | April 2016 WORKING PAPER

32 In addition to overall overconsumption of protein, the animal-based protein in overconsuming populations without risk that diets will be deficient in protein. dietary share of animal-based protein relative to plant- based protein is growing globally (Figure 8). Meat and protein are often considered more desirable than plant- However, reducing consumption of animal-based foods based food sources, although misconceptions about the - should not be a goal for people who are underconsum ing. Animal-based foods provide a concentrated source of importance of meat and protein in diets are common (Box some vitamins and minerals that are particularly valuable 6). As incomes rise and people move to urban areas, people typically shift from low-cost plant-based foods to higher- to young children in developing countries whose diet is 83 85 - Studies have demonstrated large ben cost diets more heavy in animal-based food sources. otherwise poor. efits from modest increases in meat in the diets of the poor - Changes across food value chains—including high invest 86 ment in the animal-food sector and feed crops, increases in sub-Saharan Africa. - in livestock productivity, improvements in milk pasteuri zation and cold chains, and a drop in prices of animal- Given that consumption of animal-based foods will likely - based foods relative to plant-based foods—have also con continue to grow in developing countries, this diet shift 84 preserves an abundant role for the world’s small livestock tributed to the rise in animal-based food consumption. 87 farmers. In one survey of low-income countries, nearly Global average per capita availability of animal-based protein two-thirds of rural households kept livestock. Another survey of 13 low-income countries in Asia, Latin America, and Africa grew faster (59 percent) than that of plant-based protein (14 percent) over the period 1961 to 2009 (Figure 8). Given found that livestock provided 10–20 percent of the average income of rural households in each of the lowest three of five that the average per capita consumption of protein already 88 income categories. greatly exceeds estimated dietary requirements in the world’s For these reasons, the analysis in this paper focuses on reductions in animal-based protein con wealthiest regions, it is possible to reduce consumption of - 89 sumption only in regions with high levels of consumption. Figure 8 | Per Capita Availability of Animal-Based Protein has Grown Faster than Plant-Based Protein over the Past 50 Years g/capita/day, 2009 80 70 60 Animal-Based Protein 50 40 30 Plant-Based Protein 20 10 0 2009 1961 1965 1970 1975 1980 1985 1990 1995 2000 2005 FAO (2015). Source: | 32

33 Shifting Diets for a Sustainable Food Future Debunking Protein and Meat Myths Box 6 | Protein is an essential macronutrient for building, maintaining, and repairing the human body’s tissues. Nine of the 20 amino acids that are used to make protein cannot be produced by the human body and must be obtained from food. However, several myths overstate the importance of protein in diets, especially from animal-based sources. Plant-based foods MYTH: Animal-based foods MYTH: More protein is better MYTH: ▪ ▪ ▪ are better sources of protein than need to be combined in More protein is not necessarily better, plant-based foods single meals to meet protein unless an individual is malnourished nutritional needs This myth stems from the fact that or undernourished. Although the word animal-based foods provide a complete It is not necessary to ensure that plant- proteios “protein” comes from the Greek , based protein sources are combined source of the essential amino acids meaning “of prime importance,” protein is that humans need, while plant-based to produce a complete set of essential no more important than the other nutrients foods—with the exception of a few such amino acids at every meal sitting. required for good health. While protein is Separate consumption of amino acids as soy and quinoa—lack some amino an essential part of a healthy diet, people during different meals still ensures the acids. However, plant-based foods can be may not need as much as they think. For f nutritional benefits of complementarity. readily combined to provide the full set instance, the average American adult con - of essential amino acids, as with rice and Once consumed, both animal- and plant- sumed 66 percent more protein per day based proteins are broken down during beans or peanut butter and bread. And in 2012 than the average estimated daily while meat also contains high levels of digestion into separate amino acids. requirement, but 21 percent of adults still essential micronutrients, including iron, These amino acids are then stored for considered themselves deficient in protein g a later use by the body. A and B vitamins, and zinc, a diverse in a 2014 sur vey. The World Health plant-based diet can also provide an Organization suggests that only 10–15 d adequate supply of micronutrients. percent of the daily calorie requirement b However, people following a strictly A balanced needs to come from protein. plant-based diet must take care to get plant-based diet can easily meet this need. enough Vitamin B12, which only occurs Overconsumption of protein is linked to naturally in animal-based foods but is some health problems, including kidney e available in supplements. stones and the deterioration of kidney c function in patients with renal disease. Notes: a. USDA (2014), French (2015). b. WHO (2003). c. WHO, FAO, and UNU (2007). d. Craig and Mangels (2009). e. Antony (2012). f. Young and Pellett (1994). g. Tufts University Health & Nutrition Letter (2012). Modern livestock systems that concentrate animals for all Why does it matter? or part of their lives can increase production efficiencies, Overconsumption of protein and the increasing share of 91 but with tradeoffs for other sustainability objectives. animal-based protein in diets pose a significant threat to They tend to concentrate manure, which can lead to odor achieving a sustainable food future. The production of 92 They can also give rise and water pollution problems. - animal-based foods accounted for more than three-quar 93 The use of antibiotics to to animal welfare concerns. ters of global agricultural land use and around two-thirds prevent infections in concentrated livestock production of agriculture’s production-related greenhouse gas emis - systems also raises indirect human health concerns. Stud - sions in 2009, while only contributing 37 percent of total ies have linked the use of antibiotics in livestock produc - 90 protein consumed by people in that year. Furthermore, tion to rising antimicrobial resistance—a serious health because many animal-based foods (e.g., pork and poultry) 94 threat to people. rely on crop-based feed, increased demand for these foods will widen the food gap relative to increased demand for plant-based foods. | 33 | April 2016 WORKING PAPER

34 - In terms of the direct impacts of animal-based food con Are Processed Foods Relevant Box 7 | sumption on human health, there is conflicting evidence to a Sustainable Food Future? about whether high consumption of some animal-based foods is linked to noncommunicable diseases. Some stud - ies have linked red meat consumption with cardiovascular 95 disease, type 2 diabetes, and colorectal cancer. Red meat Food processing occurs across a spectrum, running from unprocessed and minimally processed foods (e.g., peeled or also has been associated with increases in total mortality 96 frozen vegetables, fresh milk, white rice) through moderately by 10–44 percent, cardiovascular disease mortality by processed foods with added flavors (e.g., salted peanut butter, 98 97 and cancer mortality by 10–32 percent. 18–28 percent, sweetened yogurt, whole-grain breads), to highly processed - Other research has highlighted the importance of distin foods whose original food sources are unrecognizable (e.g., guishing between unprocessed red meats (e.g., beef, veal, soups, potato chips, chicken nuggets, fish fingers, crackers, a Globally, processed foods frozen pizza, soft drinks, candy). pork, lamb) and processed meats (e.g., bacon, bologna, b are a growing portion of people’s diets at all income levels. sausages, salami) when linking health outcomes with meat 99 The International Agency for Research on consumption. Food processing can contribute to a sustainable food future. Cancer, for example, has classified processed meat (Box In countries with limited food processing, for instance, losses 7) as “carcinogenic to humans,” while listing red meat as in food storage and retailing tend to be high. Processing can 100 - Generally, it is difficult to dis “probably carcinogenic.” be an important means of reducing food losses, preser ving c food, and contributing to a safe and abundant food supply. tinguish the effects of consuming an individual food from In countries that consume high amounts of animal-based the effects of the rest of the diet on human health, and of foods, food processors can play a critical role in holding course correlation is not necessarily causation. down demand for those foods by reformulating products high in animal-based ingredients to contain a larger share of plant-based ingredients, or by introducing vegetarian What are the trends? substitutes (as discussed later in this paper). Globally, per capita consumption of animal-based protein 101 has been rising since 1961. Figure 9 shows how per However, excessive consumption of processed foods may also contribute to adverse health impacts, such as capita animal-based protein availability has changed over d This those arising from obesity and poor dietary quality. time for a range of countries and regions. Looking for - is because some highly processed foods contain higher ward, we project a 79 percent increase in total consump - levels of sugars, sodium, and fats; lower levels of fiber; tion of animal-based foods (measured in calories) between and are of overall lower nutritional value than unprocessed 102 e 2006 and 2050. or minimally processed foods. This paper focuses on greenhouse gas emissions and Countries with high average per capita availability of land use at the farm level, and not in food processing, animal-based protein in 2011 were typically high-income because the bulk of emissions and nearly all of the land use countries, such as the United States, Western European demands occur on the farm. The majority of greenhouse gas countries, and Japan. As shown in Figure 7, in the United emissions from farm to fork—in the case of animal-based States, Canada, and the European Union, consumption of products, between 60 and 90 percent—tend to occur during f Processing, retail, and cooking can agricultural production. animal-based protein alone (after adjusting availability make up a large share of emissions in wealthy countries figures downward for food loss and waste during the con - and for some highly processed food products, such as sumption stage) exceeded estimated daily requirements g tomato ketchup. Still, because greenhouse gas emissions 103 In some high- for protein from all sources in 2009. related to processing, transportation, retail, and household consuming countries, availability has plateaued and even consumption primarily result from energy use, these - declined. Possible factors for peaking or declining avail emissions can best be addressed through mitigation in the energy sector rather than the agriculture sector. ability include market saturation, slowing income growth, 104 moral and ethical concerns, and health concerns. Notes: The variation in current levels of animal-based protein a. Poti et al. (2015). consumption among developed countries suggests that b. Popkin (2014). c. Floros et al. (2010). lowering per capita consumption is possible. For example, d. Poti et al. (2015). per capita meat consumption in the United Kingdom is e. Moubarac et al. (2013). 105 one-third less than in the United States. f. Arcand et al. (2012), Foster et al. (2006). g. Andersson et al. (1998). | 34

35 Shifting Diets for a Sustainable Food Future Per Capita Availability of Animal-Based Protein is on the Rise Figure 9 | g/capita/day 80 n U d es t a St ite 70 opean n o i Eur U n Jap a n 60 Br a z i l 50 a C h i n 40 O RL D W a I ndonesi 30 a r i N g e i 20 I ndi a 10 E i opi a h t 0 1990 1970 1961 2030 2020 2010 2000 2040 1980 2050 FAO (2015) for historical data 1961–2011, authors’ calculations based on Alexandratos and Bruinsma (2012) for 2050 projection, linear interpolation from 2012–2050. Sources: Growth in animal-based protein consumption has been Countries with rapidly rising per capita availability of driven by rising demand for poultry, which has increased animal-based protein as of 2011 have typically experienced increased per capita income, urbanization, and access to at around three times the rate of population growth over 106 108 each of the past five decades. These include China and This trend will likely supermarkets and restaurants. Brazil. India is an exception as its per capita availability continue, with total global poultry consumption projected of animal-based protein has remained relatively low even to grow by nearly 130 percent (i.e., more than doubling) between 2006 and 2050, outpacing growth in all other as incomes have risen. India’s lower consumption level 109 animal product sectors. likely results from cultural and religious factors. Within developing countries and emerging economies, per capita Low-income countries typically have low per capita consumption of animal-based foods tends to be highest in urban areas. For example, in China in 2011, per capita availability of animal-based protein. For example, in Ethiopia, animal-based protein availability has stayed animal-based food consumption was nearly twice as high 107 below 10 grams per person per day for decades, even as in urban areas as in rural areas. the world average grew to more than 30 grams per person Strong growth in per capita animal-based protein avail by 2011 (Figure 9). Solutions to sustainably increase - ability is projected to continue to 2050 in China and livestock and fish production are discussed in other papers 110 Creating a Sustainable Food Future series. in the Indonesia, with slower growth projected in Brazil. | 35 | April 2016 WORKING PAPER

36 AMBITIOUS ANIMAL PROTEIN REDUCTION SCENARIO. Looking forward, our projections (based on FAO projec - ▪ This scenario modified average per capita animal- tions adjusted upward to ensure adequate caloric avail - based protein consumption levels in regions where ability) estimate an increase in per capita availability of average daily per capita consumption (of all foods, animal-based foods in sub-Saharan Africa of 34 percent both plant- and animal-based) was above 60 grams between 2006 and 2050 (measured by calories). However, of protein and 2,500 calories in 2009—indicating even with this 34 percent increase, animal-based protein 113 114 were home to 1.9 overconsumption. These regions availability across sub-Saharan Africa would still only be billion people in 2009. In each overconsuming region, 13 grams per person per day in 2050—the equivalent of protein consumption was reduced to exactly 60 grams just one-and-a-half cups of whole milk and less than half per capita per day by reducing consumption of only of world average animal-based protein availability in 2011. 115 Overall, this scenario reduced animal-based foods. Because FAO projects that more than 2 billion people in animal-based protein consumption in 2009 by about sub-Saharan Africa will continue to consume low amounts half in the United States, Canada, and Brazil; by about of animal-based foods in 2050, our overall food demand one-quarter in the European Union; and by 17 percent estimates for 2050 (leading to the 70 percent food gap) are 116 Because this scenario reduced animal- globally. arguably conservative. Studies that assume animal-based based food consumption in regions overconsuming food consumption will rise in ways that match the global protein, and did not increase consumption of any patterns for increases in income project greater growth in 111 other foods, overall world calorie consumption was animal-based food consumption than the FAO. 117 reduced by 2.4 percent. What would be the land and greenhouse This sce TRADITIONAL MEDITERRANEAN DIET SCENARIO. - ▪ gas effects of reducing overconsumption nario modified the diets of the European Union, the of protein by reducing consumption of United States, and Canada—regions that consumed animal-based foods? more than 2,500 calories and 40 grams of animal- based protein per person per day in 2009. We modi - - Reducing overconsumption of protein by reducing con fied diets to reflect the food consumption patterns in sumption of animal-based foods could reduce agricultural Spain and Greece in 1980 as given in FAO (2015). We resource use and environmental impacts. Using the - chose the year 1980 to more closely mimic what peo GlobAgri model, we ran three scenarios of diet shifts away ple probably think is the “traditional Mediterranean from animal-based foods. For each scenario, we examined diet,” before the rise in obesity in these countries. the per person effects in one high-consuming country (the The “traditional Mediterranean diet” promoted and United States) and then aggregate effects across high- studied by health experts is defined as high in fruits, 112 consuming segments of the global population. vegetables, pulses, whole grains, fish, and poultry; but 118 It is low in red meats, sugars, and whole-fat dairy. Ambitious Animal Protein Reduction sce We applied the - true that the Spanish/Greek diet in 1980 (compared nario (see below) to all regions overconsuming both protein to the European, US, and Canadian diets in 2009) and calories, home to nearly 2 billion people or 28 percent of contained smaller shares of sugars, dairy, and most world population in 2009. The scenario, the most ambitious meats (including beef), as well as larger shares of fish, of the three, was designed to be an “upper bound” for what pulses, fruits, and vegetables. However, the Spanish/ might be achieved by altering the diets of a vast popula - Greek diet in 1980 also contained larger shares of eggs Traditional Mediterranean tion. Two other scenarios—the and sheep and goat meat than the European, US, and Diet scenarios—were based on actual Vegetarian Diet and 119 Canadian diets in 2009, meaning that the overall diet patterns in Mediterranean countries in 1980 and proportion of animal-based foods consumed did not among UK vegetarians in the 1990s. We applied these two change much under this scenario. In each region more “realistic” scenarios over a much smaller number of modified, we modified the diets of half of the popula - people (only 440 million) for two reasons. First, for the latter 120 or around 440 million people in all. We held tion, two scenarios we only altered diets of regions and countries - overall calorie consumption constant from 2009 refer consuming high amounts of animal-based protein in 2009— ence levels to isolate the effect of the diet shift. namely Europe, the United States, and Canada. Second, noting that a 100 percent shift of a region’s population to a Mediterranean or vegetarian diet was unlikely, we only altered the diets of half of those regions’ populations. | 36

37 Shifting Diets for a Sustainable Food Future 123 VEGETARIAN DIET SCENARIO. As in the Traditional The effects of eliminating meat consumption ▪ Mediterranean Diet - scenario, consumption was modi dwarfed the smaller effects from increased consump - 124 fied in the European Union, the United States, and tion of plant-based foods. Canada only, shifting half of the population (around 440 million people) to the actual average vegetarian Both the Ambitious Animal Protein Reduction sce - ▪ diet as observed in the EPIC-Oxford cohort study, nario and the Vegetarian Diet scenario reduced the conducted in the United Kingdom between 1993 and per person land use and greenhouse gas emissions 121 1999. For purposes of this scenario, a vegetarian diet associated with the US diet to around those associated may contain eggs or dairy, but no meat. Overall calorie with the world average diet. In fact, the Ambitious consumption was held constant from 2009 reference scenario when applied to Animal Protein Reduction levels to isolate the effect of the diet shift away from the United States was strikingly similar to the 2009 animal-based foods. world average diet. We first examine the effects of applying the three scenar - The global effects of applying the three scenarios to food - ios to the average consumption pattern of one high-con - consumption in 2009 are shown in Table 3 and summa suming country—the United States. The results are shown rized below: in Figure 10. The “US (Reference)” bars show the average US daily diet in 2009, and the associated agricultural land Ambitious Animal Protein Reduction scenario The ▪ use and greenhouse gas emissions. The “World (Refer - delivered the greatest land use and greenhouse gas ence)” bars show the average world daily diet, and the reduction benefits, because it delivered large per associated land use and emissions, for comparison’s sake. person benefits (Figure 10) and was also applied The findings from applying these three scenarios to the across the diets of 1.9 billion people. Total agricultural 122 average US daily diet in 2009 are summarized below: land use declined by 13 percent (nearly 650 million hectares)—equivalent to an area of land roughly The land use and greenhouse gas impacts of the US twice the size of India, or greater than the entire area ▪ diet are overwhelmingly driven by consumption of of land converted to agriculture between 1961 and 125 - animal-based foods. Almost 90 percent of the agri Pastureland declined by 15 percent (about 2006. cultural land used to produce the average US diet 500 million hectares) and cropland declined by about stemmed from animal-based food production (shown 9 percent (130 million hectares) thanks to the reduced in red, orange, and yellow in Figure 10), includ - need for animal feed. Greenhouse gas emissions ing both cropland and pasture. Similarly, almost 85 from agricultural production declined by 10 percent. percent of the greenhouse gas emissions associated Sparing this large land area from agricultural use with producing the food for the average US diet were could also avoid future greenhouse gas emissions related to animal-based foods. from land-use change. Assuming diet changes were sustained over time, this scenario would avoid 168 The scenario Ambitious Animal Protein Reduction e. To put this reduction billion tons of emissions of CO 2 ▪ - reduced per person agricultural land use and produc in perspective, global greenhouse gas emissions in 126 tion-related greenhouse gas emissions by 40 to 45 2009 were 44 billion tons CO e. 2 percent relative to reference. The Traditional Mediterranean Diet scenario, when ▪ Shifting to a Traditional Mediterranean Diet had a applied to half of the population of high-consuming ▪ modest effect, reducing per person agricultural land regions, had only minor impacts on land use and use and production-related greenhouse gas emissions greenhouse gas emissions. Total agricultural land use by just over 10 percent relative to reference. and production-related greenhouse gas emissions 127 As noted above, declined by less than 0.5 percent. Vegetarian Diet scenario reduced per person ag - The this minor impact is due to the fact that overall ▪ ricultural land use and production-related greenhouse animal-based food consumption did not actually gas emissions by around 50 percent relative to refer - drop by much in this scenario relative to reference. ence, with emissions shrinking by more than one-half. In particular, consumption of ruminant meats (i.e., These reductions were achieved even with a four-fold beef, sheep, and goat) in Spain and Greece in 1980 increase in the consumption of fruits and vegetables. was lower than in the United States and Canada in | 37 | April 2016 WORKING PAPER

38 Figure 10 | Reducing Consumption of Animal-Based Foods Reduces the Agricultural Land Use and Greenhouse Gas Emissions Associated with the Average US Diet by up to Half per capita values, 2009 Plant-Based Foods Dairy Other Animal-Based Foods Beef US (TRADITIONAL US (AMBITIOUS ANIMAL WORLD (REFERENCE) US (VEGETARIAN) US (REFERENCE) MEDITERRANEAN) PROTEIN REDUCTION) DAILY FOOD CONSUMPTION 2,520 2,904 2,433 2,904 2,904 (KCAL) 0.96 0.49 0.53 0.50 0.85 AGRICULTURAL LAND USE (HECTARES) 1.5 1.4 1.2 1.2 0.8 0.9 GHG EMISSIONS 0.8 FROM AGRICULTURAL 0.6 0.6 PRODUCTION E) (TONS CO 2 0.3 0 20 15.2 15 13.5 GHG EMISSIONS 10 FROM LAND-USE 8.4 7.9 7.6 CHANGE E) (TONS CO 2 5 0 Source: GlobAgri model. e emitted) for all food types. All “US” data are for United States and Canada. Calculations assume global average efficiencies (calories produced per hectare or per ton of CO Note: 2 The vegetarian diet scenario, which uses data from Scarborough et al. (2014), includes small amounts of meat, as “vegetarians” were self-reported. | 38

39 Shifting Diets for a Sustainable Food Future by 4 percent. If the diet changes were sustained over 2009, but actually higher than in the European Union in 128 2009. time, this scenario also would avoid 37 billion tons of As a result, the modest land use and greenhouse e from future land-use change—an emissions of CO gas reductions from this scenario when applied to the 2 amount approaching total greenhouse gas emissions US diet (Figure 10) were mostly canceled out when the 129 It is important to note, however, that the in 2009. scenario was applied more broadly to include a region with a lower level of ruminant meat consumption. Vegetarian Diet much smaller global benefits of the Ambitious Ani scenario, relative to the benefits of the - scenario, which shifted half of Vegetarian Diet The mal Protein Reduction scenario, were due primarily ▪ the population of high-consuming regions to veg - Vegetarian Diet to the lower level of ambition of the etarian diets, reduced land use and greenhouse gas scenario. The Vegetarian Diet scenario was applied to emissions by a greater amount than the Traditional a smaller number of people (only 440 million versus scenario, but a lesser amount 1.9 billion) for the reasons discussed above. However, Mediterranean Diet - Vegetarian than the Ambitious Animal Protein Reduction sce per person environmental benefits of the Diet Ambitious Animal Protein Reduction - and the nario. Total agricultural land use declined by 150 mil scenarios were similar (Figure 10). lion hectares, and greenhouse gas emissions declined Table 3 | Global Effects of Reducing Overconsumption of Protein by Reducing Consumption of Animal-Based Foods on Agricultural Land Use and Greenhouse Gas Emissions in 2009 AVOIDED FUTURE GHG REDUCTION IN GHG EMISSIONS FROM EMISSIONS FROM REDUCTION IN AGRICULTURAL A SCENARIO B LAND USE (MILLION HA) AGRICULTURAL PRODUCTION AGRICULTURAL LAND-USE C CHANGE (MILLION TONS CO E) E) (MILLION TONS CO 2 2 AMBITIOUS ANIMAL 508 Pastureland PROTEIN REDUCTION 133 Cropland APPLIED TO 1,907 M PEOPLE 715 168,206 TOTAL 641 TRADITIONAL Pastureland 14 MEDITERRANEAN DIET 4 Cropland -4,066 10 APPLIED TO 437 M PEOPLE TOTAL 18 113 Pastureland VEGETARIAN DIET 37 Cropland APPLIED TO 437 M PEOPLE 36,532 287 TOTAL 150 GlobAgri model. Source: Notes: a. Reference scenario included a world population of 6.8 billion, agricultural land use of 5 billion hectares (3.4 billion hectares of pastureland and 1.6 billion hectares of cropland), and 6.9 billion tons of greenhouse gas emissions from agricultural production. b. “Cropland” includes land for aquaculture farms. c. These estimates assume that the diet changes are sustained over time. If other improvements to the food system (e.g., yield gains) allowed the world to avoid future land-use change, these scenarios would allow some existing agricultural lands to revert to native vegetation and sequester the equivalent amount of carbon. | 39 | April 2016 WORKING PAPER

40 greenhouse gas emissions associated with land-use Taken together, this analysis suggests that reducing overconsumption of protein by reducing consumption of change—at the global level, provided that a large number animal-based foods could make a significant contribution of people shift their diets. Large reductions in land use to a sustainable food future. Moreover, these findings are - resulting from a reduction in animal-based food consump tion could free up enough land to meet future growth in in line with those of other researchers (Box 8). Benefits animal-based food consumption for those who are cur include deep per person savings in land use and green - - rently low consumers—without net agricultural expansion. house gas emissions among high-consuming populations, and dramatic reductions in agricultural land use—and Box 8 | Summary of Findings from Previous Studies them with the maize, wheat, and soybeans Several publications have used models and more than twice as many hectares of to estimate the effects of reducing land are used for grazing by livestock as that would otherwise have been fed to d are used for the production of all crops consumption of animal-based foods In reality, people are more likely livestock. c a e to switch to pulses, fruits, and vegetables. on resources and the environment. combined. In particular, models that account for all The land and emissions savings of these alternative foods, while still significant, sources of animal feed—including both The GlobAgri model builds on previous human-edible grains and human-inedible are less than the savings that would be studies and addresses some of their achieved if people ate animal feeds (Figure limitations. Some studies have analyzed grasses—have found that large reductions in animal-based protein consumption could 2). Still other studies are based not on one only broad and likely unrealistic dietary shifts. For example, Stehfest et al. (2009) consistent model, but on average results more than offset the increased land-use from multiple studies (often by multiple analyzed a scenario in which poultry demands arising from projected growth in food demand to 2050, even if some consumption was reduced to 44 percent authors) of the efficiency of various food products. Such an approach can introduce decreases in animal-based foods were of projected 2050 levels and beef to 52 inconsistencies due to different underlying replaced with increases in plant-based percent, with shifts to pulses replacing f GlobAgri, in methods and assumptions. the animal-based protein. Other studies foods. These reductions in agricultural contrast, applies a consistent methodology land use result from the significant share overestimate the potential land gains from g to all foods across the world. of crops and land that goes to livestock reducing animal-based food consumption by assuming that if people ate fewer production; one-quarter to one-third of all b animal-based foods, they would replace crops were used for livestock feed in 2006, Found that a US diet based on animal products required 3–4 times as much land and 2–4 times as much Eshel et al. (2009) nitrogen fertilizer as a vegetarian alternative. Modeled a “healthy diet” scenario, based on recommendations by the Har vard Medical School for Public Health, Stehfest et al. (2009) that included reducing consumption of beef, poultry/eggs, and pork to 52 percent, 44 percent, and 35 percent of global projected consumption levels (respectively) in 2050. The scenario freed up enough existing agricultural land to allow substantial reforestation and sequestering of carbon, and reduced greenhouse gas mitigation costs by more than 50 percent for the period 2005–50. Examined the effects of shifting to “healthy diets” that reduce consumption of sugar, oil, meat, and dairy while Bajzelj et al. (2014) increasing consumption of fruits and vegetables. Found that shifting to “healthy diets” reduced global cropland demand by 5 percent, pastureland demand by 25 percent, greenhouse gas emissions by 41 percent, and irrigation water demand by 3 percent relative to 2050 baseline projections. Found that reducing ruminant meat and dairy consumption—in addition to improving agricultural productivity Hedenus et al. (2014), and efficiency, and reducing greenhouse gas emissions from fossil fuels and deforestation—is a necessary Bryngelsson et al. (2016) strategy to meet European Union and global emissions targets to limit global warming to 2 degrees Celsius. | 40

41 Shifting Diets for a Sustainable Food Future Box 8 | Summary of Findings From Previous Studies (continued) Scarborough et al. Analyzed the greenhouse gas impacts of UK diets and found that relative to an average UK meat-eater diet, (2014) vegetarian diets produce one-third fewer greenhouse gases and vegan diets produce one-half fewer greenhouse gases. Tilman and Clark (2014) Predicted that global-average per capita dietary greenhouse gas emissions would increase by nearly one-third between 2009 and 2050 as incomes rose. Estimated that, relative to the projected 2050 global-average diet, per capita dietary greenhouse gas emissions would be 30 percent, 45 percent, and 55 percent lower under Mediterranean, pescetarian (vegetarian diet with fish), and vegetarian diets respectively. Tyszler et al. (2014) Modeled a diet for the Netherlands that both met nutritional requirements and reduced environmental impacts by reducing consumption of meat, cheese, and milk to 30 percent, 40 percent, and 84 percent (respectively) relative to the average Dutch diet, while raising consumption of fruits, vegetables, nuts, and seeds. The modeled diet provided a 38 percent reduction in greenhouse gas emissions and a 40 percent reduction in land use relative to the average Dutch diet. Westhoek et al. Predicted that halving consumption of meat, dairy, and eggs in the EU would reduce nitrogen emissions by 40 (2014, 2015) percent and greenhouse gas emissions by 25–40 percent. Also predicted a 23 percent reduction in domestic cropland needed to feed each EU citizen. Tom et al. (2015) Found that shifting from the current US diet to one that reduced overall caloric intake and also followed increased energy use by 38 percent, blue water footprint by 10 percent, and US dietary guidelines actually greenhouse gas emissions by 6 percent. However, the scenario modeled included not only a 25 percent in dairy consumption—leading to an overall 13 decrease in meat consumption but also a 78 percent increase increase percent in animal-based food consumption. Notes: a. Because each of the studies discussed in this box uses a different approach, and because some include greenhouse gas emissions from food processing and retail and not merely production-related emissions at the farm, their results are not directly comparable to each other or to GlobAgri results. b. Foley et al. (2011) calculates that one-third of all crops are used for animal feed while data used for Alexandratos and Bruinsma (2012) suggest that figure is 25 percent calculated on a caloric basis. c. FAO (2011b), FAO (2015). d. Foley et al. (2011). e. Stehfest et al. (2009), Eshel et al. (2009), Tyszler et al. (2014). f. For example, Tom et al. (2015) conducted a meta-analysis of data from life-cycle analyses of energy use, water use, and greenhouse gas emissions of more than 100 food types in industrialized countries, but note that “various climates, transport modes and distances, food-related technology, and production methods are reflected among the data compiled,” not to mention the fact that different studies used different boundaries (e.g., “farm to farm gate” versus “farm to fork”). Because of this, results across the studies averaged might reflect not only true differences in environmental performance among food types but also differences in methods and assumptions among underlying studies. g. This methodological consistency enables GlobAgri to analyze issues regarding diets, livestock production systems, emissions, trade, transformations between finished and raw agricultural products, and losses and wastes with great detail. | 41 | April 2016 WORKING PAPER

42 - underscores the importance of reducing beef consump DIET SHIFT 3: SHIFT FROM tion specifically for two reasons. First, demand for beef is BEEF SPECIFICALLY expected to nearly double between 2006 and 2050. Second, cattle have one of the lowest energy conversion efficiencies of The third diet shift focuses on reducing beef consumption all animal-based foods, leading to very high resource use and or shifting consumption to other animal- and plant-based environmental impacts per unit of beef produced. foods. It targets countries and populations that are high consumers of beef, relative to the world average over the Beef is not the primary source of animal-based protein past 50 years, or are projected to be high consumers by in most regions today, but consumption varies widely by 2050. This diet shift focuses on reducing rather than elim - region. World average beef-based protein consumption inating beef consumption; it recognizes that some beef was 3.2 grams of protein per capita per day in 2009, while consumption supports the livelihoods of cattle-dependent in the United States, Canada, and Latin America (includ - pastoralists, makes use of the productive capacity of native ing Brazil), people consumed more than twice that amount grazing lands, or is an offshoot of dairy production. 130 (Figure 11). The term “beef” includes cattle, bison, African buffalo, Total demand for beef, however, is projected to increase water buffalo, yak, and the four-horned and spiral-horned 131 by 95 percent between 2006 and 2050. This growth will antelopes. This shift focuses on cattle because they are the in turn drive increased production. The global population - most significant beef source in terms of quantity con of cattle is projected to increase from 1.5 billion to 2.6 sumed by people. Cattle are widely consumed around the 132 billion head between 2000 and 2050. While traditional world, except in countries such as India and Nepal, where pastoralists, in general, use dry, native grazing lands they are considered sacred by most Hindus. with great efficiency, they manage only a fraction of the world’s cattle. Without significant increases in productiv - What is the issue with beef? 133 there is a risk that growing ity on remaining pastureland, - The previous diet shift called for a reduction in overcon demand for beef, if left unchecked, will drive further expan - sumption of protein, of which beef is a source. This diet shift sion of pastureland into natural forests and savannas. Beef Consumption Varies Widely by Region Figure 11 | g protein/capita/day, 2009 Middle East & European India Brazil Latin America China North Africa Union (ex. Brazil) 15 US & Former Asia Sub-Saharan Africa OECD (other) Canada Soviet Union (ex. China & India) 10 5 World average 0 2 3 4 5 1 7 6 0 Population (billions) Source: GlobAgri model with source data from FAO (2015) and FAO (2011a). Width of bars is proportional to each region’s population. World average per capita consumption was 3.2 g of beef-based protein/capita/day. | 42

43 Shifting Diets for a Sustainable Food Future Beef is Inefficient in Creating Human-Edible Calories and Protein Figure 12 | Percent or “Units of Edible Output Per 100 Units of Feed Input” 25 20 Protein Calories 15 10 5 0 Farmed Sheep Beef Egg Poultry Pork Milk (buffalo) Milk (cattle) Farmed Shrimp Finfish Sources: Terrestrial animal products: Wirsenius et al. (2010), Wirsenius (2000). Finfish and shrimp: WRI analysis based on USDA (2013), NRC (2011), Tacon and Metian (2008), Wirsenius (2000), and FAO (1989). “Edible output” refers to the calorie and protein content of bone-free carcass. “Feed input” includes both human-edible feeds (e.g., grains) and human-inedible feeds Notes: (e.g., grasses, crop residues). However, in a world where native grasslands are nearly A near-doubling in beef production would have high all used, and further expansion of pastureland would environmental impacts, because, as Figure 12 shows, beef convert natural forests and savannas, limiting the “feed is a particularly inefficient animal product. Wirsenius et in” parameter to human-edible animal feed ignores the al. (2010) estimate that only 1 percent of gross cattle feed - large environmental impacts associated with land conver energy is converted into human-edible calories. In terms sion into pastureland. Given the expected growth in beef of protein, the conversion efficiency from “protein in” demand, these environmental impacts are very relevant. A to “protein out” is a mere 4 percent. In contrast, by this more complete way to measure conversion efficiencies across estimate, milk, pork, poultry, farmed finfish and shrimp, livestock products is to count all “feed in” at each stage of and eggs convert animal feed to edible food at 6 to 13 times 134 production and then compare “energy or protein out” versus the efficiency of beef. While sheep and goat are also “energy or protein in,” as in Figure 12. This more inclusive highly inefficient—with similar conversion efficiencies to - approach results in lower conversion efficiencies than typi beef—they are consumed in smaller quantities globally. cally assumed for ruminants such as cattle, sheep, and goat. Beef represented 12 percent of global animal-based protein consumption in 2009 versus only 2 percent for sheep and 135 goat combined. Why does it matter? Beef consumption has by far the greatest impact on Beef’s conversion efficiency is lower than often assumed. resource use and the environment of all commonly con - Measures of the conversion efficiency of inputs to human- sumed foods (Figure 2), stemming from its low efficiency in edible outputs for livestock typically compare quantities converting feed inputs to human edible calories and protein of “feed in” with quantities of “food out.” Calculations of (Figure 12). According to one US study, beef production “feed in” usually include only human-edible feeds, leaving required 28 times more land per calorie consumed than the out feeds that are not edible by people, such as grasses, 137 One-quarter of the average of other livestock categories. crop residues, and food processing wastes, even though Earth’s landmass, excluding Antarctica, is used as pasture - - they constitute roughly 80 percent of livestock feed glob 138 Beef production consumes two to four times more land. ally as measured by digestible energy, and an even greater freshwater than other livestock categories, and up to 7.5 136 share when measured by weight. times more freshwater than plant-based foods, per unit of | 43 | April 2016 WORKING PAPER

44 139 Ruminants such as cattle were responsible for 47 percent of protein delivered. Overall, beef accounts for one-third of production-related greenhouse gas emissions from agricul - the global water footprint of farm animal production, more 140 ture in 2010, without taking land-use impacts into account than any other animal category. 141 (Figure 13). Because total greenhouse gas emissions from Beef also has a disproportionate impact on climate change. agricultural production represented 13 percent of global Cattle production generates more greenhouse gas emissions greenhouse gas emissions in 2010, ruminants contributed - per unit of human-edible output than every other com about 6 percent of total global greenhouse gas emissions 142 in 2010, before accounting for land-use change. monly eaten animal-based food. Greenhouse gas emissions from cattle production originate from five main sources: What are the trends? Methane from the ruminant digestive process (known ▪ Figure 14 shows changes in per capita beef availability as “enteric fermentation”) over time for a range of countries and regions. By 2050, FAO projects that on a per capita basis, global availability Methane from manure management ▪ of beef will approach that of the European Union in 2011. Per capita demand growth will be especially strong in Nitrous oxide from excreted nitrogen in manure ▪ China, more than doubling between 2011 and 2050. In Brazil and the United States, two of the world’s top beef Nitrous oxide from the chemical nitrogenous ▪ consumers as of 2011, per capita consumption is projected fertilizers used to produce feed for cattle 143 to slightly rise and to decline, respectively. Carbon dioxide and nitrous oxide from deforestation ▪ and conversion of grassland into pastureland. Ruminants Contributed Nearly Half of Global Greenhouse Gas Emissions from Figure 13 | Agricultural Production in 2010 e 100% = 49.1 Gt CO 2 Manure management 12% Other % 7 Rice a LULUCF 11% e) 2 % 0 1 Ruminant enteric Agricultural 13% fermentation production 35% AGRICULTURAL b Energy % 7 1 PRODUCTION Total GHG Emissions (GT CO Energy 64% (Non- c agricultural) 12% 20% Ruminant wastes on pastures Soil fertilization 100% = 6.4 Gt CO e 2 Source: WRI analysis based on UNEP (2012), FAO (2012a), EIA (2012), IEA (2012), and Houghton (2008) with adjustments. Figures may not equal 100 percent due to rounding. Notes: a. LULUCF = Land Use, Land-Use Change, and Forestry. b. Includes emissions from on-farm energy consumption as well as from manufacturing of farm tractors, irrigation pumps, other machinery, and key inputs such as fertilizer. It excludes emissions from the transport of food. c. Excludes emissions from agricultural energy sources described above. | 44

45 Shifting Diets for a Sustainable Food Future Per Capita Beef Availability is Projected to Rise to 2050 Figure 14 | g/capita/day 25 20 Br a z i l 15 a t es d U n ite St 10 o Eur opean U n i n O D W RL 5 Jap n a C h i n a E opi i h a t ndonesi a I a i r e g i N a ndi I 0 1980 1961 1990 2000 1970 2020 2030 2040 2050 2010 FAO (2015) for historical data 1961–2011, authors’ calculations based on Alexandratos and Bruinsma (2012) for 2050 projection, linear interpolation from 2012–2050. Sources: 47 - In the United States and Europe, per capita beef consump in 2008. In Brazil, Argentina, and other parts of Latin America, beef has become a cultural staple because of tion has already receded from historical highs. United 148 abundant grazing land. Nevertheless, Latin America has States per capita annual beef consumption has declined 144 begun to adopt modern chicken and pork production. It is Reasons postulated for 27 percent since the 1970s. declining beef consumption include health concerns, an plausible that a combination of health concerns, increased availability of other livestock products, and public cam - increase in women in the workforce (beef takes longer than chicken to prepare), increased availability of low-cost paigns could help reduce beef consumption in Latin ready-to-cook chicken products, and more families eating America. 145 in restaurants where other meat choices are available. In In China, per capita beef availability is still only half of the Europe, per capita beef availability declined by 29 percent between 1991 and 2011, and is expected to remain rela - world average, but it is growing and is expected to con - 146 tively stagnant to 2050. - tinue to grow. In India, growing demand for dairy prod ucts, together with higher prices, is spurring an expansion In Brazil, per capita beef availability has increased steadily in the cattle population, even while beef consumption 149 over the past decades, and is now more than three times remains quite low. the world average, having surpassed the United States | 45 | April 2016 WORKING PAPER

46 We first show the per person effects of applying the three What would be the land and greenhouse gas scenarios to the average consumption pattern of one high- benefits of reducing beef consumption? consuming country—the United States—in Figure 15. The Reducing beef consumption could reduce agricultural “US (Reference)” bars show the average US daily diet in resource use and environmental impacts. Using the - 2009, and the associated agricultural land use and green GlobAgri model, we ran three scenarios to examine the house gas emissions. The other bars show how caloric effects of reducing beef consumption on agricultural consumption, land use, and greenhouse gas emissions land use and greenhouse gas emissions. For each changed under the beef reduction scenarios. scenario, we examined the per person effects in one 156 high-consuming country (the United States) and the Our findings under these scenarios included: aggregate effects across high-consuming segments of 150 the global population. - Nearly half of the agricultural land use and green ▪ house gas emissions associated with supplying the AMBITIOUS BEEF REDUCTION SCENARIO. Beef average American diet stemmed from beef alone (red ▪ consumption levels were modified in regions where portion of “US (Reference)” bar). daily per capita beef consumption in 2009 was above the world average of 3.2 grams of protein, and —which cut Ambitious Beef Reduction scenario The ▪ where per capita average calorie consumption was US beef consumption by more than 70 percent— 151 above 2,500. In each of these regions—together required one-third less agricultural land (mostly home to 1.5 billion people in 2009—per capita beef driven by a reduction in pastureland by nearly half), 152 consumption was reduced to the world average. This and resulted in a 35 percent drop in greenhouse gas scenario led to a 20 percent reduction in beef in Latin emissions from agricultural production. America (excluding Brazil); 40 percent in Europe; and more than 70 percent in Brazil, Canada, and Shift from Beef to Pork and Poultry and Shift The ▪ the United States. Globally, the scenario led to a 30 scenarios both reduced from Beef to Legumes 153 percent reduction in beef consumption. agricultural land use and greenhouse gas emissions by about 15 percent. The results of the two scenarios SHIFT FROM BEEF TO PORK AND POULTRY SCENARIO. were similar because the one-third reduction in beef ▪ In countries such as the United States, recent reduc - consumption had a far greater effect on land use and tions in beef consumption have been accompanied by greenhouse gas emissions than the corresponding increases in pork and poultry consumption, suggest - changes associated with increased consumption of ing that consumers are substituting pork and poultry pork, poultry, or legumes. for beef. Beef consumption levels were modified in regions where daily per capita beef consumption The global effects of applying the three beef reduction in 2009 was above the world average, regardless of scenarios to food consumption in 2009 are shown in Table caloric consumption level—together home to nearly 2 4 and summarized below. 154 billion people. The scenario reduced per capita beef 155 consumption by 33 percent in these regions and Under all three beef reduction scenarios, world ▪ fully replaced the reduced beef consumption with pork pastureland declined by around 200 to 300 million and poultry. Overall calorie consumption remained hectares in 2009, representing 6 to 9 percent of all unchanged relative to the 2009 reference. pastureland and 4 to 6 percent of total agricultural land. To put this change into perspective, the reduc - Consumption SHIFT FROM BEEF TO LEGUMES SCENARIO. tion in pastureland from reducing beef consumption ▪ levels were modified in the same regions as the above is similar to the entire global expansion in pastureland 157 - scenario. The scenario reduced per capita beef consump between 1961 and 2009 (270 million hectares). tion by 33 percent in these regions and fully replaced the reduced beef consumption with pulses and soy. This Ambitious Beef Cropland slightly decreased under the ▪ scenario could represent reformulation of beef-based Reduction and Shift from Beef to Legumes scenarios, products (e.g., meatballs that are two-thirds beef and reflecting the relatively small amount of feed crops - one-third plant-based proteins). Overall calorie consump dedicated to global beef production in 2009. Although tion remained unchanged relative to the 2009 reference. total agricultural land decreased, cropland slightly | 46

47 Shifting Diets for a Sustainable Food Future Reducing Beef Consumption Reduces the Agricultural Land Use and Greenhouse Gas Figure 15 | Emissions Associated with the Average US Diet by up to One-Third per capita values, 2009 Plant-Based Foods Dairy Other Animal-Based Foods Beef US (SHIFT FROM BEEF US (AMBITIOUS US (SHIFT FROM US (REFERENCE) BEEF REDUCTION) TO PORK AND POULTRY) BEEF TO LEGUMES) DAILY FOOD CONSUMPTION 2,904 2,904 2,834 2,904 (KCAL) 0.82 0.96 0.64 0.83 AGRICULTURAL LAND USE (HECTARES) 1.5 1.4 1.2 1.1 1.2 0.9 0.9 GHG EMISSIONS FROM AGRICULTURAL 0.6 PRODUCTION E) (TONS CO 2 0.3 0 20 15.2 15 13.2 13.0 GHG EMISSIONS 10.2 10 FROM LAND-USE CHANGE E) (TONS CO 2 5 0 Source: GlobAgri model. “US” data are for United States and Canada. Land-use change emissions are amortized over a period of 20 years and then shown as annual impacts. Calculations assume global average Note: efficiencies (calories produced per hectare or per ton of CO e emitted) for all food types. 2 | 47 | April 2016 WORKING PAPER

48 Table 4 | Global Effects of Reducing Beef Consumption on Agricultural Land Use and Greenhouse Gas Emissions in 2009 AVOIDED FUTURE GHG REDUCTION IN GHG EMISSIONS FROM REDUCTION IN AGRICULTURAL EMISSIONS FROM A SCENARIO B LAND USE (MILLION HA) AGRICULTURAL LAND-USE AGRICULTURAL PRODUCTION C E) (MILLION TONS CO CHANGE (MILLION TONS CO E) 2 2 AMBITIOUS BEEF REDUCTION 291 Pastureland APPLIED TO 1,463 M PEOPLE 15 Cropland 98,298 418 TOTAL 307 SHIFT FROM BEEF TO Pastureland 196 PORK AND POULTRY -26 Cropland APPLIED TO 1,952 M PEOPLE 51,116 238 TOTAL 170 SHIFT FROM BEEF 211 Pastureland TO LEGUMES 7 Cropland APPLIED TO 1,952 M PEOPLE 299 66,396 TOTAL 218 GlobAgri model. Source: Notes: Figures may not total correctly due to rounding. a. Reference scenario included a world population of 6.8 billion, agricultural land use of 5 billion hectares (3.4 billion hectares of pastureland and 1.6 billion hectares of cropland), and 6.9 billion tons of greenhouse gas emissions from agricultural production. b. “Cropland” includes land for aquaculture farms. c. These estimates assume that the diet changes are sustained over time. If other improvements to the food system (e.g., yield gains) allowed the world to avoid future land-use change, these scenarios would allow some existing agricultural lands to revert to native vegetation and sequester the equivalent amount of carbon. For some people, shifting from beef to other meats increased under the Shift from Beef to Pork and Poultry or to legumes is more likely than just reducing beef scenario, due to the increase in crop-based consumption. In the United States and Europe, per capita feeds needed for pork and poultry production more beef availability (suggesting consumption) has already than offsetting those no longer necessary for beef declined by more than 25 percent from historical highs, production. 159 while availability of pork and chicken has increased. Each scenario resulted in reductions in greenhouse gas The GlobAgri results suggest that a reduction in global emissions from agricultural production, from 4 to 6 beef consumption on the order of 30 percent—even if percent relative to reference. In addition, the avoided the reduced beef consumption were replaced with other meats—could alleviate pressure to further expand global future emissions from land-use change—assuming the diet changes were sustained over time—ranged from 51 to pastureland, and that a portion of “spared” pastureland could probably be used to accommodate cropland e. To put this reduction in perspective, 98 billion tons CO 2 expansion, while relieving agricultural pressure on forests global greenhouse gas emissions in 2009 were 44 billion 158 tons CO e. and savannas. 2 | 48

49 Shifting Diets for a Sustainable Food Future because the number of people affected by each diet shift EFFECTS OF THE DIET SHIFTS IN 2050 will likely increase by mid-century. Table 5 shows the The GlobAgri results for the three diet shifts showed number of people likely to be affected by the diet scenarios significant effects on agricultural land use and greenhouse in 2050, when applying the scenario “rules” to a larger gas emissions when applied to food consumption in 2009. 161 Ambitious Animal The and more affluent population. We do not model the effects of the diet shifts in 2050 in scenario, for example, affected nearly Protein Reduction this paper, but we draw several conclusions using the but absent changes in demand 2 billion people in 2009, GlobAgri 2009 results and FAO’s food production and would affect more than 9 billion people in 2050 when 160 consumption projections for 2050. nearly all regions are projected to consume in excess of 60 grams of protein per day. The resulting global land - The effects of the three diet shifts on land use and agri use and greenhouse gas effects in 2050 could therefore be cultural greenhouse gas emissions relative to “business as greater than they were in 2009, depending on the level of usual” are likely to be even greater in 2050 than in 2009 improvements in agricultural production efficiency. Effects of Diet Shifts in 2009 Versus 2050 Table 5 | CHANGE IN CROP CALORIE # PEOPLE # PEOPLE AFFECTED, AFFECTED, PRODUCTION, 2009 SCENARIO WHO AFFECTED 2009 2050 (PERCENT RELATIVE (MILLIONS) TO REFERENCE) (MILLIONS) DIET SHIFT 1: Reduce overconsumption of calories Eliminate Obesity and Total obese population and half of 2,078 1,385 -3.3 overweight population Halve Overweight Half of obese and overweight Halve Obesity and Overweight 1,046 1,569 -2.1 populations Reduce overconsumption of protein by reducing consumption of animal-based foods DIET SHIFT 2: Regions consuming more than Ambitious Animal 1,907 -8.5 9,444 60 g of protein and 2,500 kcal per Protein Reduction person per day Half of population of regions consuming -0.6 437 1,638 more than 40 g of animal-based protein Traditional Mediterranean Diet and 2,500 (total) kcal per person per day Half of population of regions consuming more than 40 g of animal-based protein Vegetarian Diet -4.5 437 1,638 and 2,500 (total) kcal per person per day DIET SHIFT 3: Reduce consumption of beef specifically Regions consuming beef above 2009 world average (3.2 g beef-based protein 1,463 4,019 -1.1 Ambitious Beef Reduction per person per day) and 2,500 (total) kcal per person per day Regions consuming beef above 2009 world +1.2 1,952 4,299 Shift from Beef to Pork and Poultry average Regions consuming beef above 2009 world Shift from Beef to Legumes -0.6 1,952 4,299 average Source: GlobAgri model. | 49 | April 2016 WORKING PAPER

50 All three diet shifts would likely help close the expected Move beyond information and food gap (measured in crop calories) in 2050 (Figure 1). education campaigns The GlobAgri estimates of “crop calorie production” under Typical strategies to shift diets rely on nutrition labeling the three diet shifts in 2009 (Table 5) suggest that they or public health campaigns about the benefits of different could help close the gap. Nearly all scenarios resulted in a food types or diets. Public health campaigns range from drop in crop calorie needs in 2009 relative to the reference advocating for abstinence (e.g., vegetarianism or Meat Ambitious Animal Protein Reduction - sce level—with the Free Mondays); recommending balanced diets (e.g., the nario causing the largest drop at 8.5 percent—meaning that UK “eatwell” plate, Chinese Pagoda, US ChooseMyPlate, they could also be expected to reduce crop calorie demand Canadian Food Rainbow); promoting fruits and veg - (and therefore help close the crop calorie gap) in 2050. etables; and warning against excessive consumption of However, the extent to which the shifts could help close particular food types. the gap will depend on improvements in the efficiency of 162 agricultural production systems between now and 2050. There is limited evidence, however, that consumers make regular use of information labels and education campaigns scenario, in Ambitious Animal Protein Reduction The 165 when buying food. A recent review of the influence of particular, goes the furthest in incorporating all three nutritional labeling, for example, found information to diet shifts, as it reduces consumption of calories, animal- have only a modest impact at best on purchasing behav - based protein, and beef. Based on FAO’s assumption 166 In addition, a review of the effectiveness of educa - ior. that 25 percent of all crops (measured by calories) will be 163 - tion campaigns to increase fruit and vegetable consump dedicated to animal feed in 2050, applying this scenario 167 tion in Europe has reported a small impact. to projected consumption patterns in 2050 could reduce the food gap by 30 percent—significantly reducing the British Medical Journal in 2011 Analysis published in the challenge of sustainably feeding nearly 10 billion people 164 - found a similar pattern within the restaurant environ Other scenarios could also reduce the by mid-century. ment. Calorie and nutritional information about food food gap, but Table 5 suggests they would do so to a lesser served at fast-food chains in New York City resulted in extent. no change in average calories bought, and only one in six 168 people said they actually used the information. - The lim SHIFTING STRATEGIES ited role of information alone is underscored by the fact that 44 percent of male doctors and 55 percent of nurses FOR SHIFTING DIETS surveyed in the United States are overweight, even though Overall, our analysis and many others show that what they have ready access to information and education on and how much people eat has a major impact on food 169 the links between diet and health. security, resource use, and the environmental impacts of agriculture. Of the three diet shifts examined in this paper, In light of how consumers shop, the limited effectiveness the two shifts that reduced consumption of animal-based of information and education strategies is not surprising. foods resulted in the largest potential contributions to a Consumers are bombarded with messages every day from sustainable food future. However, looking out to 2050, multiple sources and, as a result, the information is likely the current trend of rising consumption of animal-based 170 Much of con to be screened out or quickly forgotten. - foods will likely continue, absent significant actions to sumer purchasing behavior is highly routinized, especially shift demand. in a shopping behavior context, and product evaluation 171 Few people notice information and even fewer is rare. Changing people’s consumption behavior is no easy task. remember and respond to it. Shoppers tend to make Food choices are influenced by a variety of interacting 172 purchases quickly and automatically, as if on autopilot, factors, including price and taste of the food; age, gender, and repeat these habitual behaviors even if they report an health, income, geography, social identity, and culture 173 What ends up in the shopping intention to do otherwise. of the consumer; and exposure to a variety of external cart is usually based on habit and unconscious mental factors, such as marketing, media, and ease of access processing rather than on rational, informed decisions. to supermarkets and restaurants. What can be done to influence people’s food choices on a large enough scale to contribute to a sustainable food future? | 50

51 Shifting Diets for a Sustainable Food Future Interventions to change food consumption behavior, Engage the food industry, especially therefore, need to affect not only consumers’ rational, major food retailers informed decisions but also their automatic or uncon - Until now, efforts to shift diets have primarily been led by scious decisions. This insight suggests that interventions - governments and nongovernmental organizations. How must go beyond information and education campaigns— ever, consumers make the majority of their food choices designed to help rational consumers make better choices— in stores and restaurants; influencing these choices will by altering consumers’ choices and the contexts and ways 174 require the engagement of the food industry, particularly in which those choices are presented. the retail and food service sectors. What kinds of behavior change interventions might be Global food consumption patterns are converging as the employed to shift consumers’ habits? Table 6 is adapted food industry consolidates and gives rise to large-scale from a UK government study that examined successful food processors, wholesale food companies, supermarkets interventions to reduce smoking, increase purchases of and other retail store chains, and restaurant chains. In energy-efficient products, and achieve other social goals. developed countries like the United States and countries It categorizes behavior change interventions along a in Western Europe, these changes have been taking place continuum from a high degree of intervention (e.g., bans for more than a century, but they have also occurred—at a on certain products) to a low degree of intervention (e.g., comparatively rapid pace—in Latin America, Asia, Eastern altering the way choices are presented) in individuals’ 176 Europe, and parts of Africa since the 1980s. lives, and provides examples relevant to the diet shifts 175 discussed in this paper. It suggests that a wide variety of possible interventions exist beyond information and education. ehavior f o xamples E D w iets nterventions I hange C S B hifting f o Types Table 6 | ith HIGH DEGREE of inter vention (eliminate choice) LOW DEGREE of inter vention (guide or enable choice) Persuasion Eliminate or Provision of Use of social Non-fiscal Changes Fiscal Fiscal Changes to information the default norms to physical incentives restrict choice disincentives incentives and policy environment disincentives TYPE OF INTERVENTION Put healthy Marketing “30-day diet Ban trans Nutritional Tax on fat, Provide Provide salad Formulate plant-based sugar, or meat information challenge” fats, remove foods at campaign labeling, as default eye level in meat from alternative where about peers’ dietary side dish in restaurant to be successful guidelines dietary restaurant buffets, or employee choices end of aisles menu cheaper than or checkout earns a day off conventional lines in animal-based food supermarkets SHIFTING DIETS EXAMPLE Adapted from House of Lords (2011). Source: | 51 | April 2016 WORKING PAPER

52 Supermarkets are a case in point. Supermarkets accounted SHIFT WHEEL: A FRAMEWORK for 70 to 80 percent of food retail sales in the United FOR SHIFTING CONSUMPTION 177 States and France in 2000. They are also playing an increasingly important role in developing countries. From The world needs a new approach to shifting diets—one 1980 to 2000, supermarkets grew their share of food retail that uses strategies that work in step with how people sales from an estimated 5–20 percent to 50–60 percent purchase and target the places where they make their in East Asia, Latin America, urban China, South Africa, purchases. The food industry, governments, NGOs, 178 This expansion continued through and Central Europe. research organizations, and others will all need to play the first decade of the 2000s; supermarket sales grew at a a role in identifying and influencing the key factors that 40 percent compound annual growth rate in China, India, prompt people to choose animal-based food products over 179 - and Vietnam between 2001 and 2009. New supermar plant-based foods. kets typically open in urban areas with concentrations of affluent consumers before diffusing to middle- and However, there is limited publicly available data on lower-income consumers and expanding from urban to food consumption behavior at the point of purchase. As 180 rural areas. Supermarkets increase consumers’ access a result, there is a major knowledge gap on what really to foods more common in developed countries like meat, drives people’s choices and what alternative strategies dairy products, temperate fruits and vegetables, and could be used to reduce high consumption of animal- 181 processed foods and drinks. based food products, especially beef. People are also increasingly choosing to dine out—in To help address this knowledge gap and design more restaurants, cafeterias, and other food service facilities. In ooked across the field of fast- effective strategies, we l 185 the United States, expenditures on “food away from home” moving consumer goods —not just food—at a number as a share of total food expenditures grew from 25 percent of specific consumption shifts that have been successfully 182 In China, out-of-home in 1954 to 50 percent in 2013. orchestrated by industry, NGOs, and government. Notable food consumption grew by more than 100-fold between examples include the shifts from incandescent to long-life 1978 and 2008, as people increasingly eat food from street light bulbs, from caged to free-range eggs in the United 183 The stalls, traditional restaurants, and fast-food outlets. Kingdom, from big box to compact washing powder, from drivers of this trend toward increased dining out include a higher- to lower-alcohol beer in Europe, from butter to larger share of women in the workplace, higher incomes, plant-based spreads, from trans fats to healthier fats, smaller households, more affordable and convenient and a shift away from shark fin in China. While the shifts fast-food outlets, and increases in advertising by large examined primarily drew on experience from developed 184 - Given that these drivers are broadly rel restaurants. - countries, the resulting insights are likely to also be rel evant across the globe, restaurants and other food service evant to developing countries, given that food purchasing facilities will likely capture an increasing amount of global decisions around the globe are increasingly occurring in food sales in coming decades. supermarkets and food service facilities such as restau - rants and cafeterias. This global consolidation of the food industry means that large-scale actors in the food industry should be an We analyzed these shifts by reviewing published literature important focus of initiatives to shift dietary habits. The and market data reports, commissioning sales research, next section focuses on strategies that can be deployed and consulting marketing strategy professionals and by supermarkets, food service companies, and food academic behavior specialists. The resulting insights— manufacturers. gleaned from recurring themes across the different examples of consumption shifts—informed the develop - ment of the “Shift Wheel” framework (Figure 16). The Shift Wheel comprises four complementary strategies to shift consumption: (1) minimize disruption; (2) sell a compelling benefit; (3) maximize awareness and display; and (4) evolve social norms. Each Shift Wheel strategy is described below, along with examples and case studies from the food and beverage sector of consumption shifts. | 52

53 Shifting Diets for a Sustainable Food Future Figure 16 | The Shift Wheel Comprises Four Strategies to Shift Consumption M S I M N R I M O I N Z E L Replicate the Inform about A D I I experience the issue C S O R S U P E T V I Disguise L Make socially O O N the change desirable V E Make socially Form habits in unacceptable new markets SHIFT CONSUMPTION Be more Meet current memorable key needs T I Constrain Deliver new F M E compelling display A N X E benefit I B M G I Enhance Enhance Z N E I L display affordability A L W E P A R M E O N C E A S S L L E S Authors. Source: chicken, minced/ground beef and tuna products to Minimize Disruption replicate the familiar texture of the meat alternative as Changing food consumption behavior is challenging closely as possible. New manufactured animal product because it requires breaking current habits and investing alternatives are pursuing this approach in particular time and effort in establishing new ones. A change in taste, (Box 9). look, texture, smell, packaging, and even in-store location can be a major barrier to changing a consumer’s food Other products are replicating packaging formats and buying decision. Therefore, an effective route to change is product placement. For example, several brands of minimizing how perceivable any differences are, making soy milk have launched packaging that looks similar them less disruptive to the consumer. Approaches that to that of fresh cow’s milk and, rather than being minimize disruption include: stored at room temperature near long-life ultra- high temperature processed (UHT) milk, are being This approach seeks to REPLICATE THE EXPERIENCE. ▪ placed in retailers’ chillers alongside fresh milk. replicate the familiar taste, texture, look, and in- Similarly, on restaurant menus, meat, egg, and dairy store position of the “conventional” animal product. alternatives can be placed alongside their animal Brands such as Quorn (a meat substitute made from protein counterparts rather than in special vegetarian mycoprotein) have, over the years, evolved their sections. | 53 | April 2016 WORKING PAPER

54 Animal Product Replicates Box 9 | Companies and research organizations are developing and improving animal product M substitutes to minimize the disruption from S I N M I R M O shifting from conventional meat, eggs, I N Z E Replicate the L Inform about and dairy. Their approach to replicating D A I I experience the issue S C animal products usually takes one of two R O U S forms, either the “meat” is constructed from P E T V I manipulating plant or fungal material, or it Disguise O L Make socially N O is grown in the lab from animal stem cells. the change desirable V E These “meat” companies include Quorn, Beyond Meat, Impossible Foods, and Make socially Form habits in Hampton Creek. The ingredients in Beyond unacceptable new markets Meat include soy protein, pea protein, and carrot fiber. Impossible Foods is a startup SHIFT WHEEL that is developing faux meat based on the heme of legumes and nitrogen fixing Be more Meet current a plants. Heme, a molecule also found in memorable key needs the hemoglobin of animal blood, gives the T faux meat an animal-based meat flavor. I F Hampton Creek uses Canadian yellow peas E Deliver new Constrain M N to create an eggless mayonnaise alternative A E display compelling X B called “Just Mayo,” and a similar approach I benefit M G N I to create egg- and dairy-free cookie dough I Z Enhance L E Enhance L and powdered scrambled faux eggs. A display E W affordability P The company is working on plant-based A M R O E C alternatives to ice cream, ranch dressing, N A E S L and other animal-based foods. Its goal is S L E S to produce plant-based alternatives that not only have a lower environmental impact and reduce animal welfare concerns, but are more affordable and healthier than the conventional animal-based products, b without compromising on taste or texture. Maastricht University leads a lab working eggless mayonnaise alternative below that And in 2015, Oregon State University c on cultured meat. The objective is to of conventional mayonnaise, affordability is researchers patented a new strain of red successfully create real meat without especially an issue for cultured meat since marine algae that is high in protein and f the environmental impacts generated by “cell culture is one of the most expensive tastes like bacon. d conventional sources, by har vesting and resource-intensive techniques in g animal stem cells and growing them in a The lab at Maastricht Animal product replicates offer a promising modern biology.” petri dish. In 2013, the first public tasting contribution toward reducing livestock University predicts cultured meat will be of this cultured meat showed success in able to compete with real meats in 10 to 20 consumption. However, several challenges h replicating the texture and density of real Finally, replicated animal products years. remain, including taste, affordability, e meat, although the flavor seemed bland. may face an issue of trust with consumers safety, and trust. While Hampton Creek has been able to lower the price of its yet to be convinced about their merits (e.g., Notes: nutritional and health benefits). a. Wall Street Journal (2014). b. Henry (2015). c. Jha (2013). d. Hanlon (2012). e. Maastricht University (2014). f. Oregon State University (2015). g. Bello (2013). h. Bello (2013). | 54

55 Shifting Diets for a Sustainable Food Future A number of products have DISGUISE THE CHANGE. Sell a Compelling Benefit ▪ blended in new ingredients within current formats to Selling a compelling benefit involves marketing a product help disguise the shift toward plant-based ingredients. - attribute that is known to stimulate consumers’ purchas For example, the “Lurpack” brand of butter has - ing decisions. Not all food consumption shifts are dis released a number of variants, such as “Lurpack guisable; some will be apparent to consumers. In these Lighter,” which has around 30 percent vegetable fat M - situations, one approach is to explore whether the alterna S I N M blended into the butter. These inclusions are listed I R M O tive product has attributes that are appealing enough to I N Z E in the ingredients label, but the marketing leads with L D A incentivize target consumers to change their purchasing I I S C messaging about its buttery taste and spread-ability R O behavior. Critical to selling a compelling benefit is defining U S (as a result of the vegetable fat inclusion). Within P E T - and communicating attributes that are sufficiently moti V I O L the meat category, a German company “Hackplus” N O vating to stimulate behavior change with the majority of V launched a minced/ground product consisting of E consumers—rather than factors that motivate only a small 70 percent meat and 30 percent plant protein. It is niche of consumers, such as the environment (Box 10). marketed as a product “for those who opt for a healthy Approaches to selling a compelling benefit include: diet but do not want to give up meat altogether” and it has 30 percent less fat and 30 percent less cholesterol This approach leverages a MEET CURRENT KEY NEEDS. ▪ than traditional ground meat. benefit already perceived by consumers. The UK egg industry, for example, has built upon and reinforced Small, imperceptible steps are another approach the consumer perception that eggs from free-range to disguising change. Sometimes referred to as chickens taste better than those from cage-reared T “stealth changes,” this approach has been used by I - chickens. Brands such as “Happy Eggs,” with their ta F E food companies to steadily cut sodium and sugar M N - gline “happy hens lay tasty eggs,” demonstrate this ap A E levels in food. For example, salt levels in UK bread X B proach. Although free-range eggs are 30–50 percent I M G have fallen by an average of 20 percent over the past N I more expensive than conventional eggs, this quality I Z L E decade through actions by manufacturers. The change L A E association has helped capture around 45 percent of W P has been gradual over time and therefore largely A 187 M the UK market. A similar example from the UK is R O E C unnoticeable to consumers. A 2013 study estimated N A E the re-positioning of the fish “Pilchards” to “Cornish S L S L that this change has resulted in approximately 2,400 E S Sardines.” The fish was re-named with a view to en - 186 strokes and heart attacks being prevented each year. hancing its perceived taste benefits since sardines are a favored Mediterranean meal compared to that of the Getting consumers to FORM HABITS IN NEW MARKETS. ▪ poorly perceived “Pilchard.” Since this repositioning purchase healthy and more sustainable products is in the late 1990s, catches of this fish in Cornwall have less disruptive if they have yet to form buying habits. increased from 6 tons per year in the early 1990s to This approach is especially relevant to countries - 2,000 tons in 2008 as fishers, processors, and retail where consumption of animal-based protein and 188 ers have worked together to rebuild demand. beef is rapidly rising or is projected to do so by 2050. Introducing programs that limit consumers shifting DELIVER A COMPELLING NEW BENEFIT. This approach ▪ into buying more animal-based food products in entails creating a new benefit for consumers that is not geographies or social groups without a prior history or currently provided or advertised by existing options. unformed buying norms can be an effective strategy. For example, Birds Eye repositioned their pollock- based fish fingers (fish sticks), which are a more sustainable alternative to cod fish fingers, as healthier “Omega 3 Fish Fingers” and, in doing so, helped shift a large proportion of sales to more sustainable 189 pollock. This approach makes the ENHANCE AFFORDABILITY. ▪ desired alternative more affordable to the consumer. Price is an influential factor in food purchases and is frequently cited by consumers as a key determinant of | 55 | April 2016 WORKING PAPER

56 the United States—have established taxes on foods choice (Box 10). At a national level, it is estimated that 192 - income explains 65 to 70 percent of the variation in However, the “fat tax” in Den deemed unhealthy. mark was abolished after a year in 2013 in large part the average proportion of protein from animal-based 190 food sources among countries. because consumers were able to cross the border into It is likely that the Germany and purchase the same products without a falling price of chicken, relative to the price of beef, 193 - Political will to enact such taxes is often low, be has played a role in the rise of per capita chicken con - tax. cause of resistance from the food industry and because sumption in the United States (and the decline in per 191 of fears that such taxes are regressive—that they could capita beef consumption) since 1970. 194 impact poor consumers more than the rich. Taxing certain foods (e.g., those high in fat, salt, or Because plant-based proteins can be cheaper than sugar) in order to make other foods comparatively 195 more affordable has been pioneered by a few nations. animal-based ones, companies can sell reformulated products with a greater share of plant-based ingredi - - Although little is yet known about the real-world effec ents at a lower price point and/or an increased profit. tiveness of food taxes, there is reason to believe they - Indeed, part of Hampton Creek’s business model (Box can influence consumer decisions in the right circum - stances (Box 11). Since around 2010, several coun 9) is to sell plant-based alternatives that are not only - indistinguishable from, but also cheaper than, conven tries—including Barbados, Chile, Denmark, Finland, France, Hungary, Mexico, and local governments in tional animal-based products. Box 10 | The Myth of the Green Consumer Numerous consumer attitude sur veys have purchasing decisions falls significantly research often disregards other factors. for all but a small minority. Figure 17 When product attributes such as price reported that a large majority of consumers are interested in buying brands with an demonstrates this for the United Kingdom. and quality are thrown into the mix, the environmental benefit. However, this relative importance of the environment to FIGURE 17 | PRICE AND QUALITY ARE MORE IMPORTANT TO UK CONSUMERS THAN ETHICAL OR ENVIRONMENTAL CONSIDERATIONS FACTORS INFLUENCING CONSUMER PRODUCT CHOICE, PERCENTAGE OF SHOPPER RESPONSES 41 22 25 Price 17 6 42 Promotions 13 16 33 Quality or performance Taste or smell 12 11 28 Healthy option 9 11 28 Use by date 8 4 36 Familiar 7 6 34 Most important Brand 7 28 Rated within top 2 Ease of using 18 Rated within top 5 Ethical or eco-friendly 4 16 0 60 40 20 100 80 : DEFRA (2014b). Source | 56

57 Shifting Diets for a Sustainable Food Future Do Food Taxes Influence Consumption? Box 11 | Modeling studies and limited real-world Studies on food taxes highlight important not, it is likely that people will switch from caveats. First, taxes imposed by a country experience provide good reasons to believe beef to chicken (rather than consuming that food taxes imposed at the retail level less food overall). Fourth, tax rates will at the agricultural production level, such as could change purchasing choices. Reviews likely have to be high to substantially a beef production tax, are unlikely to work because production can shift to another reduce consumption; one sur vey found of the limited efforts either to tax certain c that demand elasticities for meats were types of unhealthy foods or to subsidize Second, as the Denmark “fat country. often around 1, implying that roughly a 10 tax” experience suggests, taxes imposed healthier foods indicate a significant a effect on consumption in some cases. over broader regions are likely to be more percent tax would be needed to achieve just d effective than those imposed in a single a 10 percent reduction in consumption. For example, Mexico’s 10 percent tax on Such taxes could have unfair distributional country when consumers have the ability sugar-sweetened beverages, enacted in consequences unless they are rebated to shop abroad. Third, taxes will be more January 2014, led to a 12 percent decline effective when the desired substitutes are through subsidies or reduced taxes on other in soft drink purchases (relative to expected purchases without the tax) by December of untaxed, taxed less, or subsidized. For necessities. b that year. example, if beef is taxed but chicken is Notes: a. Thow et al. (2014), Nordström and Thunström (2009), Hawkes (2012), Thow et al. (2010), Jensen and Smed (2013), Colchero et al. (2016). b. Colchero et al. (2016). c. Dumortier et al. (2012). d. Gallet (2010). - to displaying their products by providing greater mar Maximize Awareness and Display gins to retailers or running promotional campaigns, - The more chances consumers have to see and buy a prod such as offering discounts or engaging celebrity chefs uct, the greater the chance they will consider purchasing to feature their products. it. For example, people tend to consume more of items 196 they see first in a buffet. Similarly, the more prominent In some cases, undesired food CONSTRAIN DISPLAY. ▪ a product is in a consumer’s memory, the greater the - choices have been curtailed by limiting product distri likelihood they will purchase it. Approaches to maximize bution and display. Public food procurement policies awareness and display include: in schools, hospitals, prisons, and government offices have been used to influence consumption habits of Improving the physical availability ENHANCE DISPLAY. ▪ large parts of the population. The complete removal or visibility of a product—by putting it in a place that or “choice editing” from stores is possible, but it is consumers easily access—can lead to higher sales of a sensitive; 46 percent of British shoppers are in favor desired food choice. A school cafeteria in the US state of more choice editing for ethical reasons but 26 per - of Minnesota found that students waiting to pay faced cent object. With regard to choice editing for health an array of grain-based snacks, chips, granola bars, reasons, 73 percent of British shoppers were found to and desserts by the cash register that led to impulse 198 be against it. Some countries also are experimenting purchases. Rather than remove these packaged food with limiting marketing of undesirable foods. Chile products, which would have reduced total sales, the passed a law in 2012 that limits children’s exposure cafeteria replaced them with fruits. As a result, fruit (through marketing and sales) to foods that are high sales increased, snack food sales decreased, and total in calories, salt, sugar, and fat—although these provi - 197 revenue did not significantly decrease. In a retail en - 199 sions have yet to be implemented. - vironment, food manufacturers can encourage retail ers to increase the amount and quality of space given | 57 | April 2016 WORKING PAPER

58 MAKE SOCIALLY UNACCEPTABLE. BE MORE MEMORABLE. Consumers have been shown A number of campaigns ▪ ▪ have helped make a specific food socially unaccept to shop quickly, allowing little time to evaluate the - able to consumers. For example, in 2008 the celebrity brands they purchase. The majority screen out infor - chefs Hugh Fearnley-Whittingstall and Jamie Oliver mation about new products and instead seek out prod - both launched high-profile TV programs and cam ucts and brands that are in their current repertoire - paigns to highlight the issues associated with buy of choices. Products can disrupt these predetermined - ing non-free-range chicken. During the campaign, choices by being more noticeable in a purchasing situ - sales of free-range poultry reportedly increased by 35 ation (the two previous approaches), or by being more percent relative to the previous year, while sales of thought of in a purchasing situation. Creating memo - 203 caged birds fell by 7 percent. In another example, rable advertising campaigns and building consumers’ WildAid launched a campaign to draw attention to the memory associations with the desired food can, over devastating impacts of shark fishing, helping to reduce - time, increase the probability that it will be remem 200 - consumption of shark fins in China (Box 12). It is im bered and purchased. Coca-Cola, for example, is associated in many consumers’ minds with the color portant to note, however, that the long-term impact of these campaigns is unknown. red, its distinctive bottle shape, its logo script, and its 201 ability to refresh on a hot day. In the United States, In contrast, another MAKE SOCIALLY DESIRABLE. agricultural commodity marketing programs have ▪ been responsible for several memorable advertising approach is to make the preferred food type socially desirable. For example, in 2012 celebrity chef Delia campaigns, such as “Got Milk?” and “Beef: It’s What’s for Dinner.” Developing memorable marketing pro Smith helped increase UK sales of gammon (ham) - grams for plant-based foods could play an important nearly three-fold relative to the previous year after featuring a recipe for gammon on the television. role in shifting purchasing behavior. The chef’s influence over food sales has been called the “Delia effect,” a term coined when sales of Evolve social and cultural norms cranberries quadrupled the day after she used them Research has shown that the cultural environment and 204 on television. - social norms of the group a person belongs to can influ ence what and how much that person eats. A study in INFORM ABOUT THE ISSUE. As noted earlier, information the Journal of the Academy of Nutrition and Dietetics, ▪ and education campaigns, such as nutrition labeling for example, reported that people eat more when others or public health campaigns, have been the leading around them are eating more, and choose food types based policy strategy to shift consumption in recent decades, on what they perceive will help them fit in with a given but a variety of evidence shows that information and 202 group and gain social approval. Adapting the underlying 205 education alone is insufficient to lead to action. social and cultural norms is difficult, but offers another Still, information and education can be a valuable potential strategy for change. Approaches to evolve social ingredient in a broader effort to shift consumption, and cultural norms include: as evidenced by its role in reducing consumption of shark fin in China (Box 12) and trans fats across several countries (Box 13). In many cases, information can lead to indirect or multiplier effects, by raising the profile of an issue, prompting product reformula - tion (in the case of labeling), or forming the basis of food and nutrition policy and programs (e.g., national 206 dietary guidelines). | 58

59 Shifting Diets for a Sustainable Food Future Box 12 | Reducing the Unsustainable Consumption of Shark Fin in China Shark fin soup originated as a tradition in the Sung Dynasty (AD 960–1279) and M became an essential royal banquet dish S I N M I R M O in the Ming Dynasty (circa 1368–1644). I N Z E L Replicate the Inform about It became a popular status symbol among D A I I S experience C the issue China’s emerging middle class in recent R O U S years, and was regularly eaten at formal P E T V I occasions such as business receptions and O L Disguise Make socially N O weddings. The demand for shark fin has the change V desirable E led to as many as an estimated 73 million a sharks killed annually from 1996–2000 and the near extinction of the 14 most-caught Make socially b Form habits in shark species in the shark fin trade. China unacceptable new markets and Hong Kong account for 94 percent of c the demand for shark fins. SHIFT WHEEL In 2006, the conser vation organization Be more Meet current WildAid began a series of public ser vice memorable key needs announcements on the devastating T effects of shark fishing. The campaign I F featured high-profile celebrities like E Constrain Deliver new M N former professional basketball player Yao A E compelling display X B Ming, various Olympic athletes, CEOs, I benefit M G N I and famous actors and screenwriters, I Z Enhance L E Enhance L all publicly declaring their opposition to A E display W P affordability shark fin soup and challenging its social A M R O E C acceptability. N A E S L S L E S Building on the campaign, several well- established businessmen petitioned the National People’s Congress with the support of 30 members of Congress, to ban shark fin at government banquets. In July 2012, China’s State Council, the It is still early, but the Chinese Ministry of After the ban, studies found that shark national administrative body, issued a ban Commerce reported a 70 percent decline in fins contain “dangerously high levels e on ser ving shark fin at official government shark fin sales during the Spring Festival of mercury and other heavy metals,” f receptions. Originally given one to three In addition, prices of decreasing their traditional health appeal. period of 2012–13. years to come into effect, the ban was fins fell by 20–30 percent in major fishing In addition, a series of TV reports in 2013 enforced within half a year because it markets in Asia after the first six months of revealed multiple restaurants and markets g coincided with a crackdown on extravagant the ban. selling fake shark fins, furthering consumer d spending by government officials. doubts about consuming shark fin soup. Notes: a. Bakalar (2006). b. Shark Savers (2015). c. Save Our Seas Foundation (2015). d. Flannery (2014) and China Daily (2012). e. Evans (2014). f. Ministry of Commerce People’s Republic of China (2013). g. Videl (2014). | 59 | April 2016 WORKING PAPER

60 Shifting from Trans Fatty Acids to Healthier Fats Box 13 | Trans fatty acids (TFAs) are unsaturated fats found in industrially produced partially M hydrogenated vegetable oils, and also S I N M I R M naturally (at low levels) in meat and dairy. O I N Z E L Replicate the Inform about Consumption of industrially produced TFAs D A I I S experience C the issue has been associated with an increased risk R O U S of heart disease, infertility, Alzheimer’s P E T V I disease, diabetes, and some cancers. By O L Disguise Make socially N O the 1990s, studies had made the TFA- the change desirable V E health link clear, leading to public health campaigns to remove TFAs from diets. In 2009, the World Health Organization called Make socially Form habits in for the elimination of industrially produced unacceptable new markets TFAs from the global food supply. However, while some food companies have made SHIFT WHEEL large efforts to reformulate their products to remove TFAs, others in the food industry Be more Meet current have been resistant to removing TFAs memorable key needs because they are cheap; semisolid at room T temperature, making them easy to use in I F baked products; and have a long shelf life. E Deliver new Constrain M N A E compelling display X B Several types of public and private policies I benefit M G N I have aimed to reduce the amount of TFAs I Z Enhance L E Enhance L in the food supply, ranging from voluntary A E display W P affordability TFA limits agreed to by companies to A M R O E C mandatory labeling and/or bans enforced N A E S L by national and subnational governments. S L E S As of 2012, these types of policies were in place in more than 10 countries across North and South America, Western Europe, and Asia. All types of policies have been effective in decreasing the amount of TFAs in food products. In general, as the food industry has reduced the amount of TFAs, it The use of industrially produced, States found a 58 percent reduction in has increased levels of the healthier mono- partially hydrogenated vegetable oils TFA in blood plasma following mandatory and polyunsaturated fatty acids and total fat labeling. (Still, labeling policy was only became common following earlier public levels have remained relatively constant. one influence in the United States, because health campaigns in the 1960s aimed at decreasing the use of animal fats in foods. 20 percent of the country’s population Not surprisingly, bans have been the most also lives in areas covered by a TFA ban.) And just as those earlier campaigns led effective type of policy in reducing the However, labeling has its limitations as to a rise in consumption of an unhealthy amount of TFAs in the food supply. For an instrument of change. Higher-income, alternative, there are concerns that the example, Denmark introduced a ban on anti-TFA movement could lead to a rise higher-educated segments of the population industrially produced TFAs in 2003, and in consumption of palm oil—which is are most likely to shift consumption in by 2006 they had been virtually eliminated cheap and abundant, but high in saturated response to labeling. If cheaper, high-TFA from the country’s food supply. products are still available, price-conscious fatty acids, and associated with tropical consumers might not be swayed by deforestation. The rise in TFA consumption It also appears that mandatory labeling laws labeling. In addition, in low- and middle- ser ves as a cautionary tale for those have helped to drive product reformulation income countries, the main source of TFAs seeking to catalyze consumption shifts: it and consumer choice, as consumers comes from street vendors, not processed will be necessary to not only identify the increasingly demand low-TFA foods. For “undesirable” foods but also to encourage food sold by retailers, limiting the potential instance, a study conducted in the United reach of labeling. shifts toward the “desirable” alternative(s). Source: Summarized from Downs et al. (2013). | 60

61 Shifting Diets for a Sustainable Food Future Shifting to Lower-Alcohol Beer in the United Kingdom Box 14 | In 2011, the UK government challenged the beverage industry to remove 1 billion M M units of alcohol from the nation’s diet S S I I N N M M I I R R M M by the end of 2015. Keen to do this in a O O I I N N Z Z E E L L Replicate the Inform about way that maintained their sales volume, D D A A I I I I S S experience C C the issue manufacturers sought to shift consumers to R R O O U U S S lower-alcohol drinks. P P E E T T V V I I O O L L Disguise Make socially N N O O Low-alcohol beer had already been the change desirable V V E E available in the United Kingdom for many years but was an unpopular choice. Key barriers to its consumption included: Make socially Form habits in unacceptable new markets It did not sell a compelling benefit— ▪ consumers were not interested in a SHIFT WHEEL low-alcohol benefit. Be more Meet current It was disruptive to consumers— memorable ▪ key needs alcohol is a key contributor to taste and T T removing it changed the taste of beer. I I F F E E Deliver new Constrain M M N N Awareness and display were limited— A A E E ▪ display compelling X X B B low-alcohol beers were displayed in I I benefit M M G G N N I I the low-traffic areas of stores and rarely I I Z Z Enhance L L E E Enhance L L featured on highly visible display-ends. A A E E display W W P P affordability A A M M R R O O E E C C In 2012, Molson Coors launched a range N N A A E E S S L L of beers called Carling Zest that only S S L L E E S S contained 2.8 percent alcohol (versus the usual 4.8 percent). To disguise the lack of alcohol taste, the beer was launched with lemon, lime, and ginger flavors and was promoted as “light refreshment.” Around the same time, the UK government announced Following the launch of Carling Zest, other products—a largely imperceptible shift a 50 percent duty (tax) reduction on beers brands, such as Foster’s Radler, have to consumers. For example, Stella Artois, of 2.8 percent alcohol by volume or less followed suit in launching similar offerings. Budweiser, and Beck’s all reduced the and a 25 percent increase for those over alcohol content of their popular lagers from Low-alcohol beers have experienced a d The company kept the price 7.5 percent. 5 percent to 4.8 percent. significant growth over the previous years, of Carling Zest the same as its standard rising from 0.2 percent of the UK retail and b (higher-alcohol) Carling lager and invested wholesale beer and cider market in 2010 Between 2011 and 2013, the number of in advertising the new brand and securing units of alcohol consumed in the United to a predicted 5 percent of the market by c promotional displays in stores. A number Kingdom was reduced by 1.9 billion, 2024. of retailers also featured Carling Zest in the already exceeding the goal set for 2015. main beer aisle. Taking a complementary approach, Of this reduction, an estimated 1.3 billion manufacturers have also slightly reduced units were due to reductions in the alcohol the alcohol content of their best-selling content of beverages, especially beer, which contributed 1.2 billion units to the e reduction. Notes: a. Leicester (2011). b. Leicester (2011). c. Brown (2014). d. Donnelly (2015). e. Health Improvement Analytical Team (2014). | 61 | April 2016 WORKING PAPER

62 Box 15 | Reducing the Barriers to Consumption of a Novel Food Source During World War II During World War II, American citizens faced a potential meat shortage, as meat M S was rationed to soldiers and allies for the I N M I R M O war effort. Needing to find a substitute I N Z E L Replicate the Inform about protein source, the government sought to D A I I S experience C the issue incorporate organ meats into citizens’ diets. R O U S P The Department of Defense established E T V I the Committee on Food Habits in 1940, O L Disguise Make socially N O led by anthropologist Margaret Mead and the change desirable V E consisting of psychologists, sociologists, anthropologists, food scientists, dieticians, and home economists. The committee’s Make socially Form habits in role was to field-test and identify optimal unacceptable new markets ways to shift dietary habits toward a novel food source. Since organ meats were SHIFT WHEEL not a preferred option of consumers, the barriers to consumption were high. Simply Be more Meet current educating consumers about organ meats’ memorable key needs nutritional benefits would not have been T effective. Thus, the committee advised that I F the barriers to consumption must first be E Deliver new Constrain M N addressed and removed. Major barriers, and A E compelling display X B the strategies to overcome them, included: I benefit M G N I I Z Enhance L E Enhance L People Perception of appropriateness: A E display W P affordability do not perceive the new food as appropriate A M R O E C for them to consume. Studies found the N A E S L S following strategies to be effective in L E S changing that perception: (1) using relatable role models within audience’s social group to set examples; (2) exposing people at a young age, as this leads to easier adoption later in life; and (3) aligning the new food accessible in markets), familiar, and exactly In order to increase familiarity, Unfamiliarity: with a compelling cause, such as national as expected (i.e., in look, taste, and feel). studies suggested (1) increasing the security or patriotism during the war effort. availability of the new food (e.g., in butcher Marketers of these novel foods would need to communicate these traits to consumers. shops), as this increases the perception of Studies found the Unusual taste: acceptability; and (2) offering a variety of following strategies to be effective in the type of food being introduced, as studies World War II ended before the meat increasing taste tolerance to a new food: found that focusing on just one type of shortages became critical, and the (1) preparing the new food in a familiar organ meat led to low adoption, while variety government did not have to implement manner, adapting popular recipes to helped increase acceptance and adoption. many of the committee’s recommendations. incorporate the new food; (2) presenting However, the lessons learned from the the food with a familiar appearance; and committee’s findings can be applied to The committee found that an accepted food is (3) introducing the new food as a side dish a food that is SAFE: that is, it is selected (i.e., current efforts to encourage people to shift to highly palatable existing foods, thereby their dietary habits. needs to taste good), available (i.e., easily increasing favorability. Source: Summarized from Wansink (2002). | 62

63 Shifting Diets for a Sustainable Food Future In nearly all the case studies reviewed, a shift in consump - RECOMMENDATIONS tion behavior was achieved by using multiple strategies We conclude by outlining four recommendations to from the Shift Wheel, including minimizing disruptions help shift diets. These recommendations focus on to consumers, marketing product attributes important to increasing the share of plant-based protein in diets, while - consumers, maximizing awareness and availability of pre reducing consumption of animal-based protein and beef ferred products, and evolving social norms around food M S specifically. Our analysis demonstrates that these two diet I N M consumption. Shifts also typically involved groups across I R M O I shifts in particular, if implemented at a wide scale, could N Z E a range of sectors, including manufacturers, retailers, L D A I make the most significant contribution to a sustainable I S C nongovernmental organizations, and government agencies R O U food future in terms of closing the food gap and reducing S P working in concert to bring about change. Furthermore, in E T V agriculture’s resource use and environmental impacts. I O L many of the cases, the shifts created a financial benefit for N O V the companies involved. Box 14, which outlines the shift E toward lower-alcohol beer in the United Kingdom, shows 1. Set targets, apply the Shift Wheel, learn how collaboration and coordinated action among different from the results, and scale up successes groups using multiple strategies were critical in driving Governments, food retailers, food service companies, WHO: the change. Box 15, which highlights the findings of the companies with office cafeterias, food manufacturers, Committee on Food Habits established in the United NGOs, research organizations, and other private sector States during World War II, shows how the government organizations. intended to use multiple strategies to encourage citizens to incorporate protein-rich organ meats into their diets while 207 WHAT AND WHY: Governments and companies in the food T - It is strik domestic meat was rationed for the war effort. I F value chain, and companies that provide food services, E ing how many of the insights on dietary change gleaned in M N A E should set quantifiable targets to reduce the consumption the 1940s are still relevant today. X B I M of animal-based protein and beef specifically. They should G N I I Z experiment with using the Shift Wheel to drive progress L E Given the significant benefits of shifting diets, how L A E W P toward these targets. The growing market dominance of might the Shift Wheel be applied to achieve this end? A M R O E large food manufacturers, retailers, and service companies C The first step would be to analyze the landscape of N A E S L S makes them especially well placed to influence consumer L E animal- and plant-based food consumption in a given S choices. Increasing the share of plant-based protein in geography or market. Who are the consumers? What food sales (relative to animal-based protein) can reduce are they eating? Where, when, why, and how is this costs, as animal-based proteins can be more expensive consumption occurring? This analysis would help identify 208 Shifting to more sustain than plant-based proteins. - the most promising intervention points. These could be a able food consumption choices can also help businesses specific occasion (e.g., family evening meals); a particular deliver on their sustainability commitments, including product format (e.g., meatballs); a social perception (e.g., those regarding water, climate change, and deforestation. that plant-based protein is inferior to meat); certain Governments should also use the Shift Wheel to drive demographic groups (e.g., millennials); or specific outlets sustainable food choices in government-managed facilities (e.g., school or workplace cafeterias). The next step would such as schools, prisons, and hospitals. be to design approaches to achieve the chosen shift, drawing on relevant strategies from the Shift Wheel. The HOW: Targets to shift diets can take a number of forms, final steps would be testing the selected approaches and including reductions in the use of beef and animal-based scaling up successes. protein as well as increases in plant-based food sources. Additionally, food service companies should make veg - etarian options—or options low in animal-based ingredi - ents—more prominent on menus and store shelves, and devote increased advertising resources to them. NGOs and research organizations should help build communi - t ties of practice with the private sector and others to pilo test the Shift Wheel. The results should be measured and evaluated, and the lessons widely disseminated in order to accelerate transfer and scaling up of successes within | 63 | April 2016 WORKING PAPER

64 and across countries. Potential allies for such initiatives 3. Increase funding for efforts targeted at - include healthcare providers, the environment commu shifting diets - nity, and animal welfare groups. For example, Kaiser Per Philanthropic community, government, research WHO: manente, a US healthcare provider, has created a 30-day grant organizations, consumer data agencies, and challenge for its customers to eat a plant-based diet and retailers. 209 see if it has a positive impact on their health. WHAT AND WHY: The funding community should increase 2. Ensure government policies are aligned with support for research and actions to shift diets, especially those that go beyond information and education cam - promoting sustainable food choices paigns. As this paper has shown, diet shifts can deliver WHO: Government agencies at all levels from city to significant environmental, health, resource use, and food national scale. security benefits, serving multiple objectives. Yet the amount of funding currently focused on shifting diets is A broad range of government policies WHAT AND WHY: tiny relative to the amounts focused on increasing the effi - already influence diet choices. Diet choices, in turn, affect ciency of food production. There is no dedicated funding multiple policy goals, including public health, agricultural mechanism for investing in new ideas for shifting diets, production, rural development, climate change mitiga - even though it holds significant promise for closing the tion, biodiversity protection, and food and water security. food gap, reducing climate change, and contributing to the Environmental sustainability has been discussed in the Sustainable Development Goals. context of nutrition recommendations and policies in several countries in the past decade, including Australia, Governments and foundations should create fund - HOW: Brazil, Germany, the Netherlands, Sweden, and the United ing mechanisms to support the development, testing, 210 States. However, government policies are not always and rollout of evidence-based strategies to shift diets. aligned and can work at cross-purposes. One example is This could include funds for NGOs and research orga - that of government support for meat and dairy producers. nizations. Market research agencies, food retailers, and Bailey et al. (2014) noted that livestock subsidies in OECD service companies should also provide resources in kind, countries amounted to $53 billion in 2013, and pork such as access to data on food consumption behavior, and subsidies in China exceeded $22 billion in 2012. partnering to test and evaluate the application of the Shift Wheel in retail stores and restaurants. Governments should ensure coherence among HOW: health, agriculture, water, and environmental policies in relation to promoting sustainable diets. As a first step, 4. Create a new initiative focused on testing governments should establish a multidisciplinary cross- and scaling up strategies to shift diets agency task force to identify policies and regulations that WHO: Philanthropic community, business, and NGOs. influence diet choices; assess whether they are aligned with promoting healthy, sustainable diets; and recom - WHAT AND WHY: To our knowledge, there is no initiative mend changes to ensure alignment. Key agencies to dedicated exclusively to convening across disciplines, - involve include agriculture, health, environment, educa developing research, piloting actions to shift diets at the - tion, forests, water, and the lead agency for implement point of purchase, and scaling up the results. To date, ing the United Nations Sustainable Development Goals. much of the discussion on shifting diets has happened in Agriculture production subsidies should be an important proverbial silos and is not underpinned by data on current focus given their size and influence on what types of food food consumption patterns or on the efficacy and cost- farmers produce, although subsidy reform is politically effectiveness of different interventions to effect dietary difficult. As discussed above, several countries have also changes. Nor is it focused on delivering scalable results. recently established taxes on “unhealthy” foods high in fat, Moreover, potentially influential actors have been miss - - salt, and/or sugar. Others are experimenting with regula ing from the conversation—including research agencies, tions to clearly label unhealthy foods and/or limit the marketing strategists, advertising agencies, and important marketing of those foods. Governments should evaluate actors within food supply chains, such as manufacturers, the effectiveness of these taxes and regulations and scale retailers, restaurant chains, and celebrity chefs. up approaches that prove effective. | 64

65 Shifting Diets for a Sustainable Food Future A new initiative should be established to convene HOW: CALL TO ACTION marketing and consumer behavior change experts In a world that is on a course to demand more than 70 and others involved in food value chains, catalyze new percent more food, nearly 80 percent more animal-based approaches to shifting diets, conduct pilot tests, build an foods, and 95 percent more beef between 2006 and 2050, evidence base, and share and scale up successes. Its goal much of the action on how to sustainably feed the world should not be to turn everyone into a vegan or vegetarian, by mid-century has focused on boosting agricultural - but rather to promote diets that encourage greater con production and productivity. However, it will be extremely - sumption of plant-based foods, while reducing consump difficult, if not impossible, to meet the challenge through tion of animal-based protein and beef specifically. The productivity gains alone, given that global yields would organization should prioritize countries that are already need to rise one-third more quickly than they did during consuming high amounts of animal-based protein and the Green Revolution. Therefore, it will also be critical to beef, or are on their way to becoming high consumers. shift food consumption patterns in the coming decades. Results should be measured and evaluated to examine the extent of behavior change; existence of unintended con - The three diet shifts recommended in this paper, which sequences (and ways to mitigate them); and the impact of aim to reduce overconsumption of food—especially the behavior change on key economic, social (e.g., health), resource-intensive foods—can (a) close the food gap by and environmental indicators. Over time, this initiative nearly one-third; (b) significantly reduce agriculture’s should apply its knowledge on shifting diets to other pressure on ecosystems, freshwater, and the climate; and consumption-based challenges, such as transportation, (c) contribute to several Sustainable Development Goals. housing, and energy use. “Shifting diets” is therefore an essential item on the menu for a sustainable food future. | 65 | April 2016 WORKING PAPER

66 APPENDIX A. GLOBAGRI REGIONS Table A1 shows how the world’s countries and territories East and North Africa, OECD (other countries), sub- are grouped into 11 regions in GlobAgri: Asia (except Saharan Africa, and the United States and Canada. Brazil, China and India), Brazil, China, European Union, Former China, and India are countries that are counted as their Soviet Union, India, Latin America (except Brazil), Middle own regions. Table A1 | Countries, Territories, and Regions in the GlobAgri Model GLOBAGRI REGION COUNTRY OR TERRITORY COUNTRY OR TERRITORY GLOBAGRI REGION Middle East and North Africa Brunei Darussalam Asia (ex. China and India) Afghanistan OECD, other countries Bulgaria Albania EU 27 Middle East and North Africa Sub-Saharan Africa Burkina Faso Algeria Burundi Sub-Saharan Africa American Samoa Asia (ex. China and India) Asia (ex. China and India) OECD, other countries Cambodia Andorra Cameroon Sub-Saharan Africa Angola Sub-Saharan Africa US & Canada Canada Latin America (ex. Brazil) Anguilla Not classified Cape Verde Sub-Saharan Africa Antarctica Antigua and Barbuda Latin America (ex. Brazil) Cayman Islands Latin America (ex. Brazil) Central African Republic Argentina Latin America (ex. Brazil) Sub-Saharan Africa Sub-Saharan Africa Chad Armenia Former Soviet Union Aruba EU 27 Channel Islands Latin America (ex. Brazil) Australia Latin America (ex. Brazil) Chile OECD, other countries EU 27 China China Austria Christmas Island Azerbaijan Former Soviet Union Not classified Not classified Latin America (ex. Brazil) Bahamas Cocos (Keeling) Islands Bahrain Colombia Latin America (ex. Brazil) Middle East and North Africa Comoros Asia (ex. China and India) Sub-Saharan Africa Bangladesh Sub-Saharan Africa Congo Latin America (ex. Brazil) Barbados Cook Islands Belarus Asia (ex. China and India) Former Soviet Union Costa Rica EU 27 Belgium Latin America (ex. Brazil) Belgium-Luxembourg EU 27 Côte d’Ivoire Sub-Saharan Africa Belize Latin America (ex. Brazil) OECD, other countries Croatia Benin Sub-Saharan Africa Cuba Latin America (ex. Brazil) EU 27 Bermuda OECD, other countries Cyprus Asia (ex. China and India) Bhutan EU 27 Czech Republic Bolivia (Plurinational State of) Latin America (ex. Brazil) Czechoslovakia EU 27 OECD, other countries Bosnia and Herzegovina Democratic People’s Republic of Korea Asia (ex. China and India) Botswana Sub-Saharan Africa Sub-Saharan Africa Democratic Republic of the Congo Brazil Denmark Brazil EU 27 Not classified British Indian Ocean Territory Sub-Saharan Africa Djibouti Dominica British Virgin Islands Latin America (ex. Brazil) Latin America (ex. Brazil) | 66

67 Shifting Diets for a Sustainable Food Future Countries, Territories, and Regions in the GlobAgri Model (continued) Table A1 | COUNTRY OR TERRITORY GLOBAGRI REGION COUNTRY OR TERRITORY GLOBAGRI REGION Middle East and North Africa Latin America (ex. Brazil) Iraq Dominican Republic Ecuador Latin America (ex. Brazil) Ireland EU 27 Isle of Man Egypt Middle East and North Africa EU 27 El Salvador Latin America (ex. Brazil) Israel Middle East and North Africa Sub-Saharan Africa Equatorial Guinea EU 27 Italy Eritrea Latin America (ex. Brazil) Jamaica Sub-Saharan Africa OECD, other countries Japan Estonia EU 27 Sub-Saharan Africa Ethiopia Jordan Middle East and North Africa Former Soviet Union Ethiopia PDR Kazakhstan Sub-Saharan Africa Kenya Sub-Saharan Africa Falkland Islands (Malvinas) Latin America (ex. Brazil) Kiribati Asia (ex. China and India) Faroe Islands OECD, other countries Asia (ex. China and India) Kuwait Fiji Middle East and North Africa EU 27 Kyrgyzstan Former Soviet Union Finland Lao People’s Democratic Republic EU 27 France Asia (ex. China and India) Latin America (ex. Brazil) Latvia EU 27 French Guiana Lebanon Asia (ex. China and India) French Polynesia Middle East and North Africa Sub-Saharan Africa Lesotho French Southern and Antarctic Not classified Territories Liberia Sub-Saharan Africa Gabon Sub-Saharan Africa Middle East and North Africa Libya Gambia Sub-Saharan Africa OECD, other countries Liechtenstein Georgia Former Soviet Union EU 27 Lithuania Germany EU 27 EU 27 Luxembourg Ghana Sub-Saharan Africa Sub-Saharan Africa Madagascar OECD, other countries Gibraltar Malawi Sub-Saharan Africa Greece EU 27 Asia (ex. China and India) Malaysia Greenland OECD, other countries Asia (ex. China and India) Maldives Grenada Latin America (ex. Brazil) Mali Sub-Saharan Africa Guadeloupe Latin America (ex. Brazil) Malta EU 27 Guam Asia (ex. China and India) Asia (ex. China and India) Marshall Islands Guatemala Latin America (ex. Brazil) Latin America (ex. Brazil) Martinique Sub-Saharan Africa Guinea Mauritania Sub-Saharan Africa Sub-Saharan Africa Guinea-Bissau Mauritius Sub-Saharan Africa Latin America (ex. Brazil) Guyana Mayotte Sub-Saharan Africa Latin America (ex. Brazil) Haiti Mexico Latin America (ex. Brazil) OECD, other countries Holy See Micronesia (Federated States of) Asia (ex. China and India) Honduras Latin America (ex. Brazil) Monaco OECD, other countries EU 27 Hungary Asia (ex. China and India) Mongolia OECD, other countries Iceland Montenegro OECD, other countries India India Latin America (ex. Brazil) Montserrat Asia (ex. China and India) Indonesia Morocco Middle East and North Africa Middle East and North Africa Iran (Islamic Republic of) | 67 | April 2016 WORKING PAPER

68 Countries, Territories, and Regions in the GlobAgri Model (continued) Table A1 | GLOBAGRI REGION COUNTRY OR TERRITORY COUNTRY OR TERRITORY GLOBAGRI REGION OECD, other countries Sub-Saharan Africa Mozambique Saint Pierre and Miquelon Myanmar Asia (ex. China and India) Latin America (ex. Brazil) Saint Vincent and the Grenadines Samoa Namibia Sub-Saharan Africa Asia (ex. China and India) Nauru OECD, other countries San Marino Asia (ex. China and India) São Tomé and Príncipe Asia (ex. China and India) Sub-Saharan Africa Nepal Netherlands Saudi Arabia EU 27 Middle East and North Africa Netherlands Antilles Latin America (ex. Brazil) Sub-Saharan Africa Senegal New Caledonia Asia (ex. China and India) Serbia OECD, other countries OECD, other countries New Zealand OECD, other countries Serbia and Montenegro Nicaragua Sub-Saharan Africa Latin America (ex. Brazil) Seychelles Sub-Saharan Africa Niger Sub-Saharan Africa Sierra Leone Singapore Asia (ex. China and India) Nigeria Sub-Saharan Africa Asia (ex. China and India) Slovakia EU 27 Niue Asia (ex. China and India) Slovenia EU 27 Norfolk Island Solomon Islands Asia (ex. China and India) Northern Mariana Islands Asia (ex. China and India) Sub-Saharan Africa Somalia OECD, other countries Norway Occupied Palestinian Territory South Africa Sub-Saharan Africa Middle East and North Africa Oman Middle East and North Africa Spain EU 27 Pacific Islands Trust Territory Not classified Asia (ex. China and India) Sri Lanka Pakistan Asia (ex. China and India) Sub-Saharan Africa Sudan Palau Asia (ex. China and India) Latin America (ex. Brazil) Suriname Panama Latin America (ex. Brazil) Svalbard and Jan Mayen Islands OECD, other countries Asia (ex. China and India) Papua New Guinea Sub-Saharan Africa Swaziland Paraguay Latin America (ex. Brazil) Sweden EU 27 Peru Latin America (ex. Brazil) OECD, other countries Switzerland Philippines Asia (ex. China and India) Middle East and North Africa Syrian Arab Republic Not classified Pitcairn Islands Tajikistan Former Soviet Union EU 27 Poland Asia (ex. China and India) Thailand Portugal EU 27 The former Yugoslav Republic of OECD, other countries Macedonia Puerto Rico Latin America (ex. Brazil) Asia (ex. China and India) Timor-Leste Qatar Middle East and North Africa Togo Sub-Saharan Africa Asia (ex. China and India) Republic of Korea Tokelau Asia (ex. China and India) Republic of Moldova Former Soviet Union Asia (ex. China and India) Tonga Sub-Saharan Africa Réunion Trinidad and Tobago Latin America (ex. Brazil) Romania EU 27 Tunisia Middle East and North Africa Russian Federation Former Soviet Union Middle East and North Africa Turkey Sub-Saharan Africa Rwanda Turkmenistan Former Soviet Union Sub-Saharan Africa Saint Helena Turks and Caicos Islands Latin America (ex. Brazil) Saint Kitts and Nevis Latin America (ex. Brazil) Latin America (ex. Brazil) Saint Lucia | 68

69 Shifting Diets for a Sustainable Food Future Countries, Territories, and Regions Table A1 | APPENDIX B. DIET SHIFT SCENARIOS in the GlobAgri Model (continued) This appendix contains further detail on the data sources, calculation methods, and assumptions behind each of the diet shift scenarios in this paper. Table B1 also contains fur - COUNTRY OR TERRITORY GLOBAGRI REGION ther detail on the caloric composition of the reference diets Tuvalu Asia (ex. China and India) and scenarios presented in Figures ES-3, 3, 6, 10, and 15. Sub-Saharan Africa Uganda Former Soviet Union Ukraine DIET SHIFT 1: REDUCE Middle East and North Africa United Arab Emirates OVERCONSUMPTION OF CALORIES EU 27 United Kingdom Scenario: Eliminate Obesity and Halve Overweight Sub-Saharan Africa United Republic of Tanzania Use country-level obesity and overweight prevalence ▪ United States of America US & Canada data (mean percentage values) from Ng et al. (2014a) Latin America (ex. Brazil) United States Virgin Islands and data on percentage of population ages 0–14 from Unspecified Not classified World Bank (2015). Latin America (ex. Brazil) Uruguay Assume even distribution of women vs. men and ▪ USSR Former Soviet Union girls vs. boys in each country. Uzbekistan Former Soviet Union Apply region-level or nearby country data for ▪ Asia (ex. China and India) Vanuatu countries with no data in Ng et al. (2014a) and World Latin America (ex. Brazil) Venezuela (Bolivarian Republic of) Bank (2015). Viet Nam Asia (ex. China and India) Scale 2013 obesity/overweight prevalence values ▪ Wake Island Not classified in Ng et al. (2014a) to 2009 population data in Wallis and Futuna Islands Asia (ex. China and India) GlobAgri model to estimate the number of obese and overweight adults and children in 2009. (Although Middle East and North Africa Western Sahara “adult” in Ng et al. 2014a is defined as age 20 or older, Yemen Middle East and North Africa and in World Bank 2015 is defined as age 15 or older, Sub-Saharan Africa Yemen Dem for convenience, these calculations assume that the Yugoslav SFR OECD, other countries definitions of “adult” are equivalent.) Zambia Sub-Saharan Africa Assume that it takes an extra 500 kcal/day (sustained Zimbabwe Sub-Saharan Africa ▪ over time) for a person to become obese and extra 250 kcal/day (sustained over time) to become overweight, as informed by FAO (2004) and Hall et al. (2011a). Assume a 100 percent reduction in obesity and a ▪ 50 percent reduction in overweight relative to 2009 reference. Calculate the number of “avoided excess calories” ▪ consumed each year in each country (multiplying the reduction in the obese population * 500 kcal/day * 365 - days/year, and multiplying the reduction in the over weight population * 250 kcal/day * 365 days/year). Divide the “avoided excess calories” by the number of ▪ calories actually consumed in each country in 2009 to determine a “calorie adjustment factor” for each country. Lower the number of calories actually consumed in ▪ 2009 by each country’s calorie adjustment factor, applying the factor equally across all food categories in each country. | 69 | April 2016 WORKING PAPER

70 - For 2050, assume the number of obese and over If “adjusted calorie consumption” is below 2,500 kcal/ ▪ ▪ weight people will increase by 50 percent relative to capita/day, raise the “animal product adjustment factor” until per capita calorie consumption is back to 2009 (from approximately 2.1 billion to approximately 2,500. 3.1 billion). The implied compounded annual rate of growth in obesity and overweight of about 1 percent per Finally, in Brazil, US & Canada, Latin America [ex. ▪ year is slower than historical growth rates and takes - Brazil], Middle East and North Africa, and the Euro into account the fact that countries with high obesity pean Union, multiply 2009 consumption of each ani - 211 levels have started to see growth rates taper off. mal food group (aquatic animal products, beef, dairy, eggs, pork meat, poultry meat, small ruminant meat) Scenario: Halve Obesity and Halve Overweight by each region’s “animal product adjustment factor” (ranges from 0.11 to 0.93) to determine adjusted con - Same as the Eliminate Obesity and Halve Overweight ▪ sumption of each product in 2009. scenario, but assume a 50 percent reduction in obesity and a 50 percent reduction in overweight relative to For 2050, these criteria would lead to all world ▪ 2009 reference. regions being affected by this scenario, except the former Soviet Union. DIET SHIFT 2: REDUCE OVERCONSUMPTION Scenario: Traditional Mediterranean Diet OF PROTEIN BY REDUCING CONSUMPTION OF ANIMAL-BASED FOODS Use a weighted average of the Spanish and Greek ▪ national diets from 1980 as the “Mediterranean 1980” Scenario: Ambitious Animal Protein Reduction diet (as given in the “food supply” column of FAO Use 2009 region-level consumption data 2015), mapped to GlobAgri food categories. (The ▪ from GlobAgri. Mediterranean 1980 diet has a per capita food supply (availability) of just more than 3,000 kcal/capita/day, Only modify diets in regions where daily per capita ▪ suggesting low levels of both hunger and obesity in consumption (of all foods) is above 2,500 kcal (a those countries in 1980.) consumption level above FAO’s average daily energy requirement of 2,353 kcal/capita/day) and 60 g of Only modify diets in regions where average daily per ▪ protein (a level well above the average estimated daily capita consumption was above 2,500 calories and 40 - requirement of 50 g protein/capita/day). This require grams of animal-based protein (in 2009, this includes ment eliminates Asia (ex. China and India), the former EU and US & Canada). Soviet Union, China, India, other OECD, and sub-Sa - Scale consumed calories to 2009 reference levels in haran Africa. (2,500 kcal and 60 g protein per capita ▪ the regions where consumption levels were modified. daily thresholds were also used in Bajzelj et al. 2014). For the global analysis, assume that 50 percent of In all other regions (Brazil, US & Canada, Latin ▪ ▪ people in those regions shift their consumption to the America [ex. Brazil], Middle East and North Africa, “Mediterranean 1980” diet. European Union), calculate the 2009 “excess protein For 2050, these criteria would lead to the following consumption” above 60 g/day. ▪ regions being affected by this scenario: Brazil, Subtract “excess protein consumption” from actual ▪ US & Canada, China, Latin America (ex. Brazil), 2009 animal protein consumption to determine other OECD, and European Union. “adjusted animal protein consumption.” Divide “adjusted animal protein consumption” by ▪ Scenario: Vegetarian Diet actual 2009 animal protein consumption to determine Use the UK 1993–99 vegetarian diet as reported “animal product adjustment factor.” ▪ in Scarborough et al. (2014), mapped to GlobAgri Apply the “animal product adjustment factor” to 2009 ▪ food categories. (This vegetarian diet includes small animal product calorie consumption to determine the amounts of meat, as “vegetarians” in the study were downward adjustment in calories of animal products. self-reported.) - Subtract the “adjustment in calories of animal prod ▪ ucts” from actual 2009 total calorie consumption to determine “adjusted calorie consumption.” | 70

71 Shifting Diets for a Sustainable Food Future Only modify diets in regions where average daily per Scenario: Shift from Beef to Pork and Poultry ▪ capita consumption was above 2,500 calories and 40 Use 2009 region-level consumption data ▪ grams of animal-based protein (same regions as in the from GlobAgri. Traditional Mediterranean Diet scenario). Only modify diets in regions where daily beef ▪ Scale consumed calories to 2009 reference levels in ▪ consumption is above world average. This includes the regions where consumption levels were modified. all regions modified in the Ambitious Beef Reduction For the global analysis, assume that 50 percent of scenario, plus the former Soviet Union and other ▪ people in those regions shift their consumption to the OECD. Because this scenario does not alter calorie “UK vegetarian” diet. consumption, it is fine to alter the diets of the regions that consumed below 2,500 kcal/capita/day (former For 2050, the regions affected by this scenario would ▪ Soviet Union and other OECD). Traditional be the same as those affected by the scenario. Mediterranean Diet In these regions, reduce per capita beef consumption ▪ by 33 percent. (Justification: FAO 2015 data shows that per capita beef consumption has dropped from DIET SHIFT 3: SHIFT FROM BEEF peak levels by 27 percent in Japan, 32 percent in the SPECIFICALLY EU, and 40 percent in the US.) Scenario: Ambitious Beef Reduction Replace that reduced beef consumption with increases ▪ Use 2009 region-level consumption data in consumption of pork and poultry (proportionately ▪ from GlobAgri. to what people in each region consumed in 2009). Only modify diets in regions where daily beef Overall calorie consumption remains unchanged, as ▪ ▪ consumption is above 2009 world average (3.2 g people just shift from one type of meat calories to two protein/capita/day) and where kcal consumption is other types. above 2,500 kcal/capita/day. This eliminates Asia For 2050, these criteria would lead to the following (ex. China and India), China, Former Soviet Union, ▪ regions being affected by this scenario: Brazil, US and India, Middle East and North Africa, sub-Saharan Canada, China, former Soviet Union, Latin America Africa, and other OECD. (ex. Brazil), Middle East and North Africa, other In all other regions (Brazil, US & Canada, Latin OECD, and European Union. ▪ America [ex. Brazil], European Union), reduce beef consumption to world average (3.2 g protein/ Scenario: Shift from Beef to Legumes capita day). World average divided by actual beef Use 2009 region-level consumption data consumption gives the “beef adjustment factor” ▪ from GlobAgri. for each region. - As in the sce Shift from Beef to Pork and Poultry Multiply the “beef adjustment factor” (which ranges ▪ ▪ nario, only modify diets in regions where daily beef from 0.25 to 0.60) by 2009 beef consumption in consumption is above world average. In these regions, each region. Due to regional differences in the ratio reduce beef consumption by 33 percent. of beef-based calories to beef-based protein, adjusted beef consumption varied between 25 and 35 kcal/ Replace that reduced beef consumption with (equal) ▪ capita/day in regions where the scenario was applied. increases in consumption of pulses and soybeans (simulating, for instance, ground beef being replaced Verify that applying the “beef adjustment factor” ▪ with a mixture that is two-thirds beef, one-third plant does not reduce overall calorie consumption below proteins). 2,500 kcal/capita/day, or protein consumption below 60 g protein/capita/day, in any of the regions Overall calorie consumption remains unchanged, ▪ where applied. as people just shift from one type of calories to two others. For 2050, these criteria would lead to the following ▪ regions being affected by this scenario: Brazil, US For 2050, the regions affected by this scenario would ▪ and Canada, China, Latin America (ex. Brazil), be the same as those affected by the Shift from Beef to Middle East and North Africa, other OECD, and . scenario Pork and Poultry European Union. | 71 | April 2016 WORKING PAPER

72 Table B1 | Detailed Composition of Reference Diets and Scenarios per capita daily food consumption (kcal), 2009 REDUCE OVERCONSUMPTION OF REDUCE REDUCE CONSUMPTION PROTEIN BY REDUCING CONSUMPTION OVERCONSUMPTION OF BEEF SPECIFICALLY OF CALORIES OF ANIMAL-BASED FOODS US (Beef to US (Ambitious US (Halve US (Eliminate US World US (Beef to US US (Ambitious US (Traditional Obesity & Halve Food type Obesity & Animal Protein Pork and Beef Reduction) (Reference) Mediterranean) (Reference) (Vegetarian) Legumes) Overweight) Poultry) Overweight) Reduction) Beef 94 36 88 90 46 80 3 25 63 63 306 Dairy 287 294 149 119 432 306 306 306 306 45 45 31 43 44 22 60 78 45 45 Eggs 21 Fish 21 26 20 21 10 35 2 21 21 Pork 113 106 109 55 129 4 113 125 113 100 67 Poultry 155 159 80 46 0 165 183 165 165 ANIMAL-BASED FOODS 3 Sheep and goat 3 3 2 3 2 3 3 3 10 15 Fibers 13 13 14 11 0 14 14 14 14 Fruits and vegetables 138 149 130 133 138 275 596 138 138 138 10 Maize 131 134 140 146 36 140 140 140 140 Oil (other oilcrops) 29 38 27 28 29 271 64 29 29 29 63 Other cereals 105 91 93 97 97 156 97 97 97 27 54 Other oilcrops 54 37 51 52 54 54 0 54 Other plant products 54 51 52 54 59 89 54 54 54 28 66 48 62 63 66 Other products 4 66 66 66 37 4 Palm oil 1 1 1 50 0 1 1 1 1 Pulses 40 58 38 39 40 64 164 40 40 56 Rapeseed and 0 45 42 43 45 45 20 45 45 0 mustard seed Rapeseed and 56 56 29 52 54 56 0 0 56 56 mustard seed oil PLANT-BASED FOODS 74 268 251 258 Rice 493 111 268 268 268 268 Roots and tubers 63 118 59 61 63 155 230 63 63 63 Soybean oil 395 371 380 395 69 0 395 395 395 71 1 1 1 1 0 94 1 1 17 Soybeans 13 Sugar plants and 253 194 237 243 253 281 191 253 253 253 products Sunflower seed 0 5 5 5 5 0 5 5 5 0 Sunflower seed oil 6 27 6 6 6 117 7 6 6 6 690 Wheat 406 416 432 442 639 432 432 432 432 716 Animal-based foods 747 367 702 719 363 681 522 678 747 2,156 2,188 2,156 2,065 2,025 2,077 Plant-based foods 2,223 2,382 2,156 2,156 TOTALS 2,520 2,904 2,904 2,834 2,904 2,796 2,904 GRAND TOTAL 2,904 2,433 2,726 GlobAgri model with source data from FAO (2015) and FAO (2011a). Source: All “US” data are for United States and Canada. The vegetarian diet scenario, which uses data from Scarborough et al. (2014), includes small amounts of meat, as “vegetarians” were self-reported. Note: | 72

73 Shifting Diets for a Sustainable Food Future REFERENCES Beattie, G., L. McGuire, and L. Sale. 2010. “Do we actually look at the carbon Adhya, T. K., B. Linquist, T. Searchinger, R. Wassmann, and X. Yan. 2014. footprint of a product in the initial few seconds? An experimental analysis “Wetting and Drying: Reducing Greenhouse Gas Emissions and Saving of unconscious eye movements.” International Journal of Environmental, Water from Rice Production.” Working Paper, Installment 8 of Creating a Cultural, Economic & Social Sustainability 6: 47–65. . Washington, DC: World Resources Institute. Sustainable Food Future Behan, D. F., and S. H. Cox. 2010. “Obesity and its Relation to Mortality and Ajani, U. A., P. A. Lotufo, J. M. Gaziano, I. M. Lee, A. Spelsberg, J. E. Buring, Morbidity Costs.” Society of Actuaries . Accessible at: . World agriculture towards Alexandratos, N., and J. Bruinsma. 2012. Bello, D. 2013. “Cultured Beef: Do We Really Need a $380,000 Burger Grown . Rome: Food and Agriculture Organization of 2030/2050: The 2012 revision in Petri Dishes?” Scientific American , August 5, 2013. Accessible at: . American Diabetes Association. 2008. “Economic costs of diabetes in the 596–615. Diabetes Care 31: U.S. in 2007.” Berry, S. 2011. “Biofuel Policy and Empirical Inputs to GTAP Models.” Report to the California Air Resources Board (CARB). Sacramento, CA: Anand, S. S., C. Hawkes, R. J. de Souza, A. Mente, M. Dehghan, R. Nugent, CARB. M. A. Zulyniak, T. Weis, A. M. Bernstein, R. M. Krauss, D. Kromhout, D. J. A. Jenkins, V. Malik, M. A. Martinez-Gonzalez, D. Mozaffarian, S. Yusuf, W. Binnie, M. A., K. Barlow, V. Johnson, and C. Harrison. 2014. “Red meats: C. Willett, and B. M. Popkin. 2015. “Food Consumption and its Impact on 98: 445–451. Time for a paradigm shift in dietary advice.” Meat Science Cardiovascular Disease: Importance of Solutions Focused on the Globalized Food System.” Journal of the American College of Cardiology 66: 1590–1614. Birds Eye. 2015. “Leading the way in sustainable fisheries development.” Accessible at: . assessment (LCA) of tomato ketchup: a case study.” Journal of Cleaner Production 6: 277–288. Bloom, D. E., E. T. Cafiero, E. Jané-Llopis, S. Abrahams-Gessel, L. R. Bloom, S. Fathima, A. B. Feigl, T. Gaziano, M. Mowafi, A. Pandya, K. Prettner, L. Antony, A. C. 2012. “Megaloblastic anemias.” In R. Hoffman, E. J. Benz Jr., L. The Global Rosenberg, B. Seligman, A. Z. Stein, and C. Weinstein. 2011. Hematology: Basic E. Silberstein, H. Heslop, J. Weitz, and J. Anastasi (eds.). Geneva: World Economic Economic Burden of Noncommunicable Diseases. Principles and Practice. London: Churchill Livingstone. Forum. Arcand, Y., D. Maxime, and R. Zareifard. 2012. “Life cycle assessment of BLS (US Bureau of Labor Statistics, US Department of Labor). 2015. processed food.” In J. I. Boye and Y. Arcand (eds.) Green Technologies in “Average Retail Food and Energy Prices, U. S. and Midwest Region.” New York: Springer US. Food Production and Processing. Washington, DC: BLS. Accessible at: < http://www.bls.gov/regions/mid- atlantic/data/AverageRetailFoodAndEnergyPrices_USandMidwest_Table. Armstrong, B., and R. Doll. 1975. “Environmental factors and cancer htm>. incidence and mortality in different countries, with special reference to dietary practices.” International Journal of Cancer 15: 617–631. Bouvard, V., D. Loomis, K. Z. Guyton, Y. Grosse, F. El Ghissassi, L. Benbrahim-Tallaa, N. Guha, H. Mattock, and K. Straif. 2015. “Carcinogenicity Sustainable Diets: Helping Shoppers. Arnold, H., and T. Pickard. 2013. of consumption of red and processed meat.” Lancet Oncology , published London: IGD (The Institute of Grocery Distribution). online October 26, 2015. Paper prepared on behalf of the International Agency for Research on Cancer Monograph Working Group. Accessible at: Bailey, R., A. Froggatt, and L. Wellesley. 2014. “Livestock—Climate Change’s . Forgotten Sector: Global Public Opinion on Meat and Dairy Consumption.” London: Chatham House, Royal Institute of International Affairs. Brinsden, H. C., F. J. He, K. H. Jenner, and G. A. MacGregor. 2013. “Sur veys of the salt content in UK Bread: progress made and further reductions Bailey, R., and D. R. Harper. 2015. “Reviewing Inter ventions for Healthy and 3: e002936. possible.” BMJ Open Sustainable Diets.” London: Chatham House, Royal Institute of International Affairs. Brown, N. 2014. “Half of Brits are trying to drink less—but booze sales are up.” The Grocer, October 31, 2014. Accessible at: . 4: 924–929. Nature Climate Change climate mitigation.” Bruinsma, J. 2009. The Resource Outlook to 2050: By how much do land, Bakalar, N. 2006. “38 Million Sharks Killed for Fins Annually, Experts Rome: Food and Agriculture water and crop yields need to increase by 2050? Estimate.” National Geographic, October 12, 2006. Accessible at: . | 73 | April 2016 WORKING PAPER

74 Bryngelsson, D., S. Wirsenius, F. Hedenus, and U. Sonesson. 2016. “How Davis, S. J., and K. Caldeira. 2010. “Consumption-based accounting of CO 2 emissions.” can the EU climate targets be met? A combined analysis of technological and Proceedings of the National Academy of Sciences of the United 152–164. 107 (12): 5687–5692. States of America Food Policy 59: demand-side changes in food and agriculture.” DEFRA (Department for Environment, Food, and Rural Affairs). 2014a. Buckland, G., A. Agudo, N. Travier, J. M. Huerta, L. Cirera, and M. J. Tormo. “National Statistics: United Kingdom Egg Statistics, Quarter 2, 2014.” 2011. “Adherence to the Mediterranean diet reduces mortality in the Spanish cohort of the European Prospective Investigation into Cancer and Nutrition London: DEFRA. 106 (10): 1581–91. British Journal of Nutrition (EPIC-Spain).” DEFRA (Department for Environment, Food, and Rural Affairs). 2014b. “Food Statistics Pocketbook 2014.” London: DEFRA. CAIT Climate Data Explorer. 2015. “CAIT Climate Data Explorer.” Washington, DC: World Resources Institute. Available online at: . Delgado, C., M. Rosegrant, H. Steinfeld, S. Ehui, and C. Courbois. 1999. “Livestock to 2020: The next food revolution.” IFPRI Food, Agriculture, and the Environment Discussion Paper 28. Washington, DC: IFPRI (International Campbell, T. C., C. Junshi, T. Burn, B. Parpia, Q. Yinsheng, C. Chumming, Food Policy Research Institute). and C. Geissler. 1992. “China: from diseases of poverty to diseases of Ecology of affluence. Policy implications of the epidemiological transition.” Del Gobbo, L. C., S. Khatibzadeh, F. Imamura, R. Micha, P. Shi, M. Smith, Food and Nutrition 27: 133–144. S. S. Myers, and D. Mozaffarian. 2015. “Assessing global dietary habits: Capacci, S., M. Mazzocchi, B. Shankar, J. B. Macias, W. Verbeke, F. J. Perez- a comparison of national estimates from the FAO and the Global Dietary Cueto, A. Koziol-Kozakowska, B. Piorecka, B. Niedzwiedzka, D. D’Addesa, Database.” American Journal of Clinical Nutrition 101: 1038–1046. A. Saba, A. Turrini, J. Aschemann-Witzel, T. Bech-Larsen, M. Strand, J. Wills, and W. B. Traill. 2012. “Policies to promote health eating in Europe: a Di Maso, M., R. Talamini, C. Bosetti, M. Montella, A. Zucchetto, M. Libra, Nutrition Review structured review of policies and their effectiveness.” E. Negri, F. Levi, C. La Vecchia, S. Franceschi, D. Serraino, and J. Polesel. 70 (3): 2013. “Red meat and cancer risk in a network of case-control studies 188–200. 24 (12): 3107–3112. Annals of Oncology focusing on cooking practices.” Caraher, M., and G. Cowburn. 2005. “Taxing food: implications for public health nutrition.” Dobbs, R., C. Sawers, F. Thompson, J. Manyika, J. Woetzel, P. Child, S. 8 (8): 1242–1249. Public Health Nutrition Overcoming Obesity: An initial economic McKenna, and A. Spatharou. 2014. CDC (Centers for Disease Control and Prevention). 2015. “Nutrition for analysis. McKinsey Global Institute. Everyone: Protein.” Atlanta, GA: CDC. Accessible at: . , The Telegraph Donnelly, L. 2015. “Britain turns its back on strong beer.” March 8, 2015. Accessible at: . Cecchini, M., F. Sassi, J. A. Lauer, Y. Y. Lee, V. Guajardo-Barron, and D. Chisholm. 2010. “Tackling of unhealthy diets, physical inactivity, and obesity: health effects and cost-effectiveness.” Downs, S. M., A. M. Thow, and S. R. Leeder. 2013. “The effectiveness of Lancet 376: 1775–84. policies for reducing dietary trans fat: a systematic review of the evidence.” . 2012. “Shark fins off official receptions’ menus soon.” July 3, China Daily 91: 262–269. Bulletin of the World Health Organization 2012. Accessible at: < http://www.chinadaily.com.cn/xinhua/2012-07-03/ content_6342306.html>. Dumanovsky, T. 2011. “Changes in energy content of lunchtime purchases from fast food restaurants after introduction of calorie labelling: cross Colchero, M. A., B. M. Popkin, J. A. Rivera, and S. W. Ng. 2016. “Beverage sectional customer sur veys.” 343: d4464. BMJ purchases from stores in Mexico under the excise tax on sugar sweetened 2016: 352:h6704. BMJ beverages: obser vational study.” Dumortier, J., D. J. Hayes, M. Carriquiry, F. Dong, X. Du, A. Elobeid, J. F. Fabiosa, P. A. Martin, and K. Mulik. 2012. “The effects of potential changes Cornish Sardine Management Association. 2015. “Historical background.” in United States beef production on global grazing systems and greenhouse Accessible at: . Dwyer, T., and B. S. Hetzel. 1980. “A comparison of trends of coronary heart Cor valán, C., M. Reyes, M. L. Garmendia, and R. Uauy. 2013. “Structural disease mortality in Australia, USA, and England and Wales with reference responses to the obesity and noncommunicable diseases epidemic: the to three major risk factors—hypertension, cigarette smoking and diet.” 14: 79–87. Obesity Reviews Chilean Law of Food Labeling and Advertising.” International Journal of Epidemiology 9: 65–71. Craig, W. J., and A. R. Mangels. 2009. “Position of the American Dietetic Economist. 2012. “A Fat Chance: The Danish government rescinds its Journal of the American Dietetic Association unwieldy fat tax.” November 17, 2012. Accessible at: . | 74

75 Shifting Diets for a Sustainable Food Future 2014. “Chubby little emperors: Why China is under- and FAO (Food and Agriculture Organization of the United Nations). 2015. Economist. “FAOSTAT.” Rome: FAO. over-nourished at the same time.” June 14, 2014. Accessible at: . FAO, IFAD, and WFP (Food and Agriculture Organization of the United Nations, International Fund for Agricultural Development, and World Food Ecorys. 2014. Food taxes and their impact on competitiveness in the agri- Programme). 2015. The State of Food Insecurity in the World 2015. Meeting food sector. Rotterdam: Ecorys. the 2015 international hunger targets: taking stock of uneven progress. Rome: FAO. Annual Energy EIA (United States Energy Information Administration). 2012. FAO, WFP, and IFAD (Food and Agriculture Organization of the United Washington, DC: EIA. Outlook 2012: With Projections to 2035. Nations, World Food Programme, and International Fund for Agricultural EIA (United States Energy Information Administration). 2015. “U.S. Energy- Development). 2012. The State of Food Insecurity in the World 2012. Economic growth is necessary but not sufficient to accelerate reduction of Related Carbon Dioxide Emissions, 2014.” Washington, DC: EIA. Accessible hunger and malnutrition. Rome: FAO. at: . FAO, WHO, and UNU (Food and Agriculture Organization of the United Eshel, G., P. Martin, and E. Bowen. 2009. “Land use and reactive nitrogen 14: 1–14. discharge: effects of dietary choices.” Nations, World Health Organization, and United Nations University). 1985. Earth Interactions Energy and protein requirements. Geneva: WHO. Eshel, G., A. Shepon, T. Makov, and R. Milo. 2014. “Land, irrigation water, greenhouse gas, and reactive nitrogen burdens of meat, eggs, and dairy Finkelstein, E. A., J. G. Trogdon, J. W. Cohen, and W. Dietz. 2009. “Annual medical spending attributable to obesity: Payer- and ser vice-specific production in the United States.” Proceedings of the National Academy of 111 (33): 11996–12001. estimates.” Health Affairs 28 (5): w822–w831. Sciences of the United States of America Estruch, R, E. Ros, J. Salas-Salvado, M. I. Covas, D. Corella, and F. Aros. Finkelstein, E. A., M. C. DiBonaventura, S. M. Burgess, and B. C. Hale. 2010. “The Costs of Obesity in the Workplace.” Journal of Occupational and 2013. “Primary prevention of cardiovascular disease with a Mediterranean 52 (10): 971–976. 368 (14): 1279–90. New England Journal of Medicine diet.” Environmental Medicine Flannery, R. 2014. “Chinese Billionaire Guo Guangchang to be Feted as Evans, M. 2014. “Shark fin soup sales plunge in China.” April 10, Al Jazeera, 2014. Accessible at: . at: . Yield FAO (Food and Agriculture Organization of the United Nations). 1989. and Nutritional Value of the Commercially More Important Fish Species. Floros, J. D., R. Newsome, and W. Fisher. 2010. “Feeding the World Rome: FAO. Today and Tomorrow: The Importance of Food Science and Technology.” Comprehensive Reviews in Food Science and Food Safety FAO (Food and Agriculture Organization of the United Nations). 2004 . The 9: 572–599. world food summit five years later: Mobilizing the political will and resources Rome: FAO. to banish world hunger. Foley, J. A., N. Ramankutty, K. A. Brauman, E. S. Cassidy, J. S. Gerber, M. Johnston, N. D. Mueller, C. O’Connell, D. K. Ray, P. C. West, C. Balzer, E. M. Bennett, S. R. Carpenter, J. Hill, C. Monfreda, S. Polasky, J. Rockstrom, FAO (Food and Agriculture Organization of the United Nations). 2011a. Rome: Global food losses and food waste—extent, causes, and prevention. J. Sheehan, S. Siebert, D. Tilman, and D. P. M. Zaks. 2011. “Solutions for a 478: 337–342. cultivated planet.” FAO. Nature Foresight. 2011. Synthesis Report C6: Raising the Limits of Sustainable FAO (Food and Agriculture Organization of the United Nations). 2011b. The Production. London: Government Office for Science. State of the World’s Land and Water Resources for Food and Agriculture. Rome: FAO. Foster, C., K. Green, M. Bleda, P. Dewick, B. Evans, A. Flynn, and J. Mylan. 2006. Environmental Impacts of Food Production and Consumption: FAO (Food and Agriculture Organization of the United Nations). 2012a. A Report to the Department for Environment, Food and Rural Affairs. 1990–2005. Rome: FAO. Global forest land-use change Manchester Business School. London: Defra. FAO (Food and Agriculture Organization of the United Nations). 2012b. FAO Statistical Yearbook 2012 . World Food and “Feeding the world.” In Fox, N., and K. Ward. 2007. “Health, Ethics and Environment: A Qualitative Study of Vegetarian Motivations.” Appetite 50 (2–3): 422–29. Agriculture. Rome: FAO. FAO (Food and Agriculture Organization of the United Nations). 2013. Acta Veterinaria Fraser, D. 2008. “Understanding Animal Welfare.” The State of Food and Agriculture 2013. Rome: FAO. 50 (Suppl. 1): S1. Scandinavica FAO (Food and Agriculture Organization of the United Nations). 2014. “Food Security Indicators.” October 15, 2014 revision. Rome: FAO. | 75 | April 2016 WORKING PAPER

76 Hammer, P., E. Riebe, and R. Kennedy. 2009. “How clutter affects advertising French, S. 2015. “Marketing Trends in Protein: Are You Capitalizing on the 49 (2): 159–163. Journal of Advertising Research Opportunity?” Presentation at the Protein Trends & Technologies Seminar, effectiveness.” Oak Brook, IL, USA, May 5, 2015. Harleysville, PA, USA: Natural Marketing Han, K., A. M. Trinkoff, C. L. Storr, and J. Geiger-Brown. 2011. “Job Stress Institute. JONA (Journal of Nursing and Work Schedules in Relation to Nurse Obesity.” 41 (11): 488–495. Administration) Fry, J., and W. Finley. 2005. “The prevalence and costs of obesity in the EU.” 64: 359–362. Proceedings of the Nutritional Society Hanlon, M. 2012. “Fake meat: is science fiction on the verge of becoming Guardian Fung, T. T., K. M. Rexrode, C. S. Mantzoros, J. E. Manson, W. C. Willett, fact?” , June 22, 2012. Accessible at: . from coronary heart disease and stroke in women.” 119 (8): Circulation Hanson, C., and T. Searchinger. 2015. “Ensuring Crop Expansion is Limited 1093–1100. to Lands with Low Environmental Opportunity Costs.” Working Paper. Installment 10 of Washington, DC: Creating a Sustainable Food Future. Gallet, C. A. 2010. “Meat meets meta: a quantitative review of the price 92: World Resources Institute. elasticity of meat.” American Journal of Agricultural Economics 258–272. Hawkes, C. 2008. “Agro-food industry growth and obesity in China: what Garnett, T. 2014a. “What is a sustainable healthy diet?” Discussion paper. role for regulating food advertising and promotion and nutrition labelling?” Obesity Reviews Oxford, UK: Food Climate Research Network. 9 (suppl.1): 151–161. Garnett, T. 2014b. “Changing what we eat: A call for research & action Hawkes, C. 2012. “Food taxes: what type of evidence is available to inform on widespread adoption of sustainable healthy eating.” Oxford, UK: Food policy development?” Nutrition Bulletin 37: 51–56. Climate Research Network. HHS and CDC (U.S. Department of Health & Human Ser vices, Centers for Disease Control and Prevention). 2013. Antibiotic resistance threats in the Garnett, T., and A. Wilkes. 2014. Appetite for Change: Social, economic and Atlanta, GA: HHS and CDC. United States. environmental transformations in China’s food system. Oxford, UK: Food Climate Research Network. Health Improvement Analytical Team. 2014. “Responsibility Deal: Garnett, T., E. Röös, and D. Little. 2015. “Lean, green, mean, obscene...? Monitoring the number of units of alcohol sold—Second interim report, What is efficiency? And is it sustainable?” Oxford, UK: Food Climate 2013 data.” London: Health Improvement Analytical Team, Department of Health. Accessible at: . Godfray, H. C. J., J. R. Beddington, I. R. Crute, L. Haddad, D. Lawrence, J. Hedenus, F., S. Wirsenius, and D. J. Johansson. 2014. “The importance of Muir, J. Pretty, S. Robinson, S. M. Thomas, and C. Toulmin. 2010. “Food Science Security: The Challenge of Feeding 9 Billion People.” 327 (5967): reduced meat and dairy consumption for meeting stringent climate change targets.” 124: 79–91. Climate Change 812–18. Gortmaker, S., B. A. Swinburn, D. Levy, R. Carter, P. L. Mabry, D. T. The Washington Henry, S. 2015. “Can this company do better than the egg?” Finegood, T. Huang, T. Marsh, and M. L. Moodie. 2011. “Changing the future March 2, 2015. Accessible at: . Grecu, A. M., and K. W. Rotthoff. 2015. “Economic Growth and Obesity: 22 (7): Applied Economics Letters Findings of an Obesity Kuznets Cur ve.” Herrero, M., P. Havlík, H. Valin, A. Notenbaert, M. C. Rufino, P. K. Thornton, 539–543. M. Blümmel, F. Weiss, D. Grace, and M. Obersteiner. “Biomass use, production, feed efficiencies, and greenhouse gas emissions from global Proceedings of the National Academy of Sciences of the Guo, L. B., and R. M. Gifford. 2002. “Soil carbon stocks and land use livestock systems.” 8 (4): 345–360. 110 (52): 20888–20893. change: a meta analysis.” Global Change Biology United States of America Haley, M. M. 2001. “Changing Consumer Demand for Meat: The U.S. Hickman, M. 2011. “The campaign that changed the eating habits of a Example, 1970–2000.” In A. Regmi (ed.). Independent , January October 23, 2011. Accessible at: . Economic Research Ser vice. Hoekstra, A. Y., A. K. Chapagain, M. M. Aldaya, and M. M. Mekonnen. Hall, K., G. Sacks, D. Chandramohan, C. Chow, Y. Wang, S. Gortmaker, and The Water Footprint Assessment Manual: Setting the Global Standard. 2011. B. Swinburn. 2011a. “Quantification of the effect of energy imbalance on bodyweight.” The Lancet 378: 826–837. London: Earthscan. Review of Business Hall, S., A. Delaporte, M. J. Phillips, M. Beveridge, and M. O’Keefe. 2011b. Hojjat, T. A. 2015. “The Economic Analysis of Obesity.” Penang, and Finance Studies 6 (1): 81–98. Blue Frontiers: Managing the environmental costs of aquaculture. Malaysia: WorldFish Center. | 76

77 Shifting Diets for a Sustainable Food Future Larsson, S. C., and N. Orsini. 2013. “Red meat and processed meat Houghton, R. A. 2008. “Carbon flux to the atmosphere from land-use changes: 1850–2005.” In: TRENDS: A compendium of data on global American Journal of consumption and all-cause mortality: a meta-analysis.” Oak Ridge, TN: Carbon Dioxide Information Analysis Center, Oak change. Epidemiology 179: 282–289. Ridge National Laboratory. Lassaletta, L., G. Billen, B. Grizzetti, J. Anglade, and J. Garnier. 2014. Behaviour Change. Science and Technology Select “50 year trends in nitrogen use efficiency of world cropping systems: the House of Lords. 2011. nd Committee, 2 relationship between yield and nitrogen input to cropland.” Environmental Report of Session 2010–12. London: Authority of the House of Lords. Research Letters 9: 105011. IEA (International Energy Agency). 2012. World Energy Outlook 2012. Paris: Leicester, A. 2011. “Small beer? assessing the Government’s alcohol IEA Publications. policies.” London: The Institute for Fiscal Studies. Accessible at: . Jar vis, L. S. 1986. Livestock Development in Latin America. Washington, DC: Le Mouël, C., A. Forslund, P. Marty, S. Manceron, E. Marajo-Petitzon, M.-A. World Bank. Caillaud, and B. Schmitt. 2015. “Le système agricole et alimentaire de la Jensen, J. D., and S. Smed. 2013. “The Danish tax on saturated fat—Short- région Afrique du Nord—Moyen-Orient à l’horizon 2050 : projections de tendance et analyse de sensibilité.” Final Pluriagri report: INRA-DEPE & run effects on consumption, substitution patterns, and consumer prices of fats.” INRA-SAE2. Food Policy 42: 18–31. Lipinski, B., C. Hanson, J. Lomax, L. Kitinoja, R. Waite, and T. Searchinger. Jha, A. 2013. “Google’s Sergey Brin bankrolled world’s first synthetic beef hamburger.” Guardian 2013. “Reducing Food Loss and Waste.” Working Paper, Installment 2 of , August 5, 2013. Accessible at: . Institute. Ji, M. F., and W. Wood. 2007. “Purchase and consumption habits: Not Maastricht University. 2014. “First-ever public tasting of lab- necessarily what you intend.” Journal of Consumer Psychology grown cultured beef burger.” July 3, 2014. Accessible at: . Jones-Smith, J. C., P. Gordon-Larsen, A. Siddiqi, and B. M. Popkin. 2011. Marques, J. M. 2013. “National Plan for Low Carbon Emissions in “Emerging disparities in overweight by educational attainment in Chinese adults (1989–2006).” Agriculture—ABC Plan in Brazil.” Presentation. Brasilia: Ministry of International Journal of Obesity 36: 866–875. Agriculture, Livestock and Food Supply, Department of Agricultural Production and Sustainability. Accessible at: . Economics to Improve Meal Selection.” Choices: The Magazine of Food, Farm, and Resource Issues 24 (3): 1–7. Martinez-Gonzalez, M. A., M. Garcia-Lopez, M. Bes-Rastrollo, E. Toledo, E. H. Martinez-Lapiscina, and M. Delgado-Rodriguez. 2011. “Mediterranean The Plant-Based Diet: a healthier way to eat. Kaiser Permanente. 2013. diet and the incidence of cardiovascular disease: a Spanish cohort.” Oakland, CA: Kaiser Permanente. 21 (4): 237–44. Nutrition, Metabolism and Cardiovascular Diseases Kearney, J. 2010. “Food consumption trends and drivers.” P hilosophical McDonald, C., B. Sharp, and A. S. C. Ehrenberg. 2002. “Impulse Purchasing 365: 2793–2807. Transactions of the Royal Society B Marketing Learning 9. London: The R&D Initiative, South Bank Patterns.” Keats, S., and S. Wiggins. 2014. “Future diets: implications for agriculture University. and food prices.” London: Overseas Development Institute. Mekonnen, M. M., and A. Y. Hoekstra. 2011. “The green, blue and grey water The Emerging Middle Class in Developing Countries. Paris: Kharas, H. 2010. Hydrology and Earth System footprint of crops and derived crop products.” Sciences 15: 1577–1600. Organisation for Economic Co-operation and Development. Mekonnen, M. M., and A. Y. Hoekstra. 2012. “A Global Assessment of the Khoury, C. K., A. D. Bjorkman, H. Dempewolf, J. Ramirez-Villegas, L. Water Footprint of Farm Animal Products.” Ecosystems 15: 401–415. Guarino, A. Jar vis, L. H. Rieseberg, and P. C. Struik. 2014. “Increasing homogeneity in global food supplies and the implications for food security.” 111 (11): 4001–4006. PNAS 2013. “Regular Press Ministry of Commerce People’s Republic of China. Conference of the Ministry of Commerce on February 20, 2013.” February 20, 2013. Accessible at: . Antibiotic Use in Food Animals: Perspective, Policy, and Potential.” Health Reports 127 (1): 4–22. Menu Labeling: does providing nutrition Larson, N. and M. Story. 2009. information at the point of purchase affect consumer behavior? Princeton, NJ: Robert Wood Johnson Foundation. | 77 | April 2016 WORKING PAPER

78 NRC (National Research Council). 2011. U.S. Beef Demand Mintert, J. R., G. T. Tonsor, and T. C. Schroeder. 2009. Nutrient Requirements of Fish and Shrimp. Manhattan, Washington, DC: The National Academies Press. Drivers and Enhancement Opportunities: A Research Summary. KS: Kansas State University Agricultural Experiment Station and Cooperative Extension Ser vice. Accessible at: . 169 (3): 339–46. American Journal of Epidemiology Monteiro, C. A., W. L. Conde, and B. M. Popkin. 2007. “Income-Specific OECD (Organisation for Economic Co-operation and Development). 2010. Trends in Obesity in Brazil: 1975–2003.” American Journal of Public Health 97 (10): 1808–1812. Paris: OECD. Healthy Choices. OECD (Organisation for Economic Co-operation and Development). 2011. Moubarac, J.-C., A. P. B. Martins, R. M. Claro, R. B. Levy, G. Cannon, and C. A. Monteiro. 2013. “Consumption of ultra-processed foods and likely impact Environmental Outlook to 2050: Climate Change. (pre-release version) Paris: 16 (12): Public Health Nutrition on human health: evidence from Canada.” OECD. 1–9. OECD (Organisation for Economic Co-operation and Development). 2014. Paris: OECD. Obesity Update. Neumann, C. G., M. W. Demment, A. Maretzki, N. Drorbaugh, and K. A. Galvin. 2010. “The livestock revolution and animal source food consumption: Oregon State University. 2015. “OSU researchers discover the unicorn— Benefits, risks and challenges in urban and rural settings of developing seaweed that tastes like bacon!” Cor vallis, OR: Oregon State University. countries.” In H. Steinfeld, H. A. Mooney, and F. Schneider (eds.). Accessible at: . Washington, DC: Island Press. Responses. Pan, A., Q. Sun, A. M. Bernstein, M. B. Schulze, J. E. Manson, M. J. Ng, M., T. Fleming, M. Robinson, B. Thompson, N. Graetz, C. Margono, E. C. Mullany, S. Biryukov, C. Abbafati, S. F. Abera, J. P. Abraham, N. M. E. Stampfer, W. C. Willett, and F. B. Hu. 2012. “Red meat consumption and Abu-Rmeileh, T. Achoki, F. S. Albuhairan, Z. A. Alemu, R. Alfonso, M. K. mortality: results from 2 prospective cohort studies.” American Journal of 172 (7): 555–563. Ali, R. Ali, N. A. Guzman, W. Ammar, P. Anwari, A. Banerjee, S. Barquera, S. Clinical Nutrition Basu, D. A. Bennett, Z. Bhutta, J. Blore, N. Cabral, I. C. Nonato, J. Chang, R. Paul, G. L. 1989. “Dietary protein requirements of physically active Chowdhury, K. J. Cour ville, M. H. Criqui, D. K. Cundiff, K. C. Dabhadkar, L. individuals.” Dandona, A. Davis, A. Dayama, S. D. Dharmaratne, E. L. Ding, A. M. Durrani, 8 (3): 154–176. Sports Medicine A. Eseghamati, F. Farzadfar, D. F. J. Fay, V. L. Feigin, A. Flaxman, M. H. Forouzanfar, A. Goto, M. A. Green, R. Gupta, N. Hafezi-Nejad, G. J. Hankey, Pica-Ciamarra, U., L. Tasciotti, J. Otte, and A. Zezza. 2011. “Livestock Assets, H. C. Harewood, R. Havmoeller, S. Hay, L. Hernandez, A. Husseini, B. T. Livestock Income and Rural Households: Cross-country Evidence from Isdisov, N. Ikeda, F. Islami, E. Jahangir, S. K. Jassal, S. H. Jee, M. Jeffreys, J. Household Sur veys.” Joint Paper of the World Bank, FAO, AU-IBAR, and ILRI. Washington, DC: World Bank. B. Jonas, E. K. Kabagambe, S. E. A. H. Khalifa, A. P. Kenge, Y. S. Khader, Y. Khang, D. Kim, R. W. Kimokoti, J. M. Kinge, Y. Kokubo, S. Kosen, G. Kwan, Pingali, P. 2007. “Westernization of Asian diets and the transformation of T. Lai, M. Leinsalu, Y. Li, X. Liang, S. Liu, G. Logroscino, R. P. Lotufo, Y. food systems: implications for research and policy.” Food Policy Lu, J. Ma, N. K. Mainoo, G. A. Mensah, T. R. Merriman, A. H. Mokdad, J. 32 (3): 281–298. Moschandreas, M. Naghavi, A. Naheed, D. Nand, K. M. V. Narayan, E. L. Nelson, M. L. Neuhouser, M. I. Nisar, T. Ohkubo, S. O. Oti, A. Pedroza, D. Popkin, B. M. 2002. “An over view on the nutrition transition and its health Prabhakarn, N. Roy, U. Sampson, H. Seo, S. G. Sepanlou, K. Shibuya, R. implications: the Bellagio meeting.” Public Health Nutrition 5 (1A): 93–103. Shiri, I. Shiue, G. M. Singh, J. A. Singh, V. Skirbekk, N. J. C. Stapelberg, L. Sturua, B. L. Sykes, M. Tobias, B. X. Tran, L. Trasande, H. Toyoshima, Popkin, B. M. 2003. “The Nutrition Transition in the Developing World.” S. van de Vijver, T. J. Vasankari, J. L. Veerman, G. Velasquez-Melendez, Development Policy Review V. V. Vlassov, S. E. Vollset, T. Vos, C. Wang, X. Wang, E. Weiderpass, A. 21 (5–6): 581–597. Werdecker, J. L. Wright, Y. C. Yang, H. Yatsuya, J. Yoon, Y. Zhao, M. Zhou, S. Zhu, A. D. Lopez, C. J. L. Murray, and E. Gakidou. 2014a. “Global, regional, Popkin, B. M., and P. Gordon-Larsen. 2004. “The nutrition transition: worldwide obesity dynamics and their determinants.” and national prevalence of overweight and obesity in children and adults International Journal of 28: S2–S9. during 1980–2013: A systematic analysis for the Global Burden of Disease Obesity 384 (9945): 766–781. The Lancet Study 2013.” Popkin, B. M., L. S. Adair, and S. W. Ng. 2012. “NOW AND THEN: The Global Nutrition Transition: The Pandemic of Obesity in Developing Countries.” Ng, S. W., M. M. Slining, and B. M. Popkin. 2014b. “Turning point for US Nutrition Reviews 70 (1): 3–21. diets? Recessionary effects or behavioral shifts in foods purchased and consumed.” American Journal of Clinical Nutrition 99 (3): 609–616. Popkin, B. M. 2014. “Nutrition, agriculture and the global food system in low and middle income countries.” Nordström, J., and L. Thunström. 2009. “The impact of tax reforms designed Food Policy 47: 91–96. to encourage healthier grain consumption.” Journal of Health Economics 28 (3): 622–634. | 78

79 Shifting Diets for a Sustainable Food Future Scarmeas, N., Y. Stern, M. X. Tang, R. Mayeux, and J. A. Luchsinger. 2006. Poti, J. M., M. A. Mendez, S. W. Ng, and B. M. Popkin. 2015. “Is the degree “Mediterranean diet and risk for Alzheimer’s disease.” of food processing and convenience linked with the nutritional quality of Annals of Neurology American Journal of Clinical Nutrition 59 (6): 912–21. foods purchased by US households?” 101: 1251–62. Schaible, G., and M. Aillery. 2012. “Water Conser vation in Irrigated Agriculture: Trends and Challenges in the Face of Emerging Demands.” Reardon, T., C. P. Timmer, C. B. Barrett, and J. Berdegué. 2003. “The rise of supermarkets in Africa, Asia, and Latin America.” American Journal of Washington, DC: U.S. Department of Agriculture, Economic Research Ser vice. 85 (5): 1140–1146. Agricultural Economics Reardon, T., S. Henson, and J. Berdegué. 2007. “‘Proactive fast-tracking’ implications of land Schmidinger, K., and E. Stehfest. 2012. “Including CO 2 diffusion of supermarkets in developing countries: implications for market The occupation in LCAs—method and examples for livestock products.” International Journal of Life Cycle Assessment. DOI 10.1007/s11367-012- institutions and trade.” 7 (4): 399–431. Journal of Economic Geography 0434-7. Reardon, T., and C. P. Timmer. 2012. “The economics of the food system 4 (1): 225–264. Annual Review of Resource Economics revolution.” Searchinger, T., C. Hanson, J. Ranganathan, B. Lipinski, R. Waite, R. Creating a Sustainable Winterbottom, A. Dinshaw, and R. Heimlich. 2013. Reardon, T., K. Z. Chen, B. Minten, L. Adriano, T. A. Dao, J. Wang and S. D. Food Future: Interim Findings of the 2013–14 World Resources Report. Annals of the Gupta (2014). “The quiet revolution in Asia’s rice value chains.” Washington, DC: World Resources Institute. 1331: 106–118. New York Academy of Sciences Searchinger, T., C. Hanson, and J-M. Lacape. 2014. “Crop Breeding: Reisch, L. A., W. Gwozdz, G. Barba, S. De Henauw, N. Lascorz, and I. Pigeot. Creating Renewing the Global Commitment.” Working Paper, Installment 7 of Washington, DC: World Resources Institute. a Sustainable Food Future. 2013. “Experimental evidence on the impact of food advertising on children’s knowledge about and preferences for healthful food.” Journal of Obesity . Accessible at: . Searchinger, T., R. Edwards, D. Mulligan, R. Heimlich, and R. Plevin. 2015. Science 347: “Do biofuel policies seek to cut emissions by cutting food?” 1420–1422. Robinson, E., J. Thomas, P. Aveyard, and S. Higgs. (2014). “What Everyone Else Is Eating: A Systematic Review and Meta-Analysis of the Effect of Informational Eating Norms on Eating Behavior.” Journal of the Academy of Semba, R. D., M. Shardell, F. A. Sakr Ashour, R. Moaddel, I. Trehan, K. Nutrition and Dietetics 114 (3): 414–429. M. Maleta, M. I. Ordiz, K. Kraemer, M. A. Khadeer, L. Ferrucci, and M. J. Manary. 2016. “Child Stunting is Associated with Low Circulating Essential Amino Acids.” EBioMedicine (in press). Accessible at: . Egeberg, A. Tjonneland, L. Nailler, M. Boutron-Tuault, R. Clavel-Chapelo, V. Krogh, D. Palli, S. Panico, R. Tumino, F. Ricceri, M. M. Bergmann, H. Boeing, Shark Savers. 2015. “The impact of the shark fin trade.” n.d. Accessible at: K. Li, R. Kaaks, K. Khaw, N. J. Wareham, F. L. Crow, T. J. Key, A. Naska, A. . C. L. Parr, G. Skeie, P. Jakszyn, M. Sanchez, J. M. Huerta, M. L. Redondo, A. Barricarte, P. Amiano, I. Drake, E. Sonestedt, G. Hallmans, I. Johansson, Oxford, UK: How Brands Grow: What Marketers Don’t Know. Sharp, B. 2010. V. Fedirko, I. Romieux, P. Ferrari, T. Norat, A. C. Vergnaud, E. Riboli, and Oxford University Press. J. Linseisen. 2013. “Meat consumption and mortality: results from the Cancer Causes European Prospective Investigation into cancer and nutrition.” Sinha, R., A. J. Cross, B. I. Graubard, M. F. Leitzmann, and A. Schatzkin. 24: 685–693. and Control 2009. “Meat intake and mortality: a prospective study of over half a million Archives of Internal Medicine 169: 562–571. Romaguera, D., T. Norat, T. Mouw, A. M. May, C. Bamia, and N. Slimani. people.” 2009. “Adherence to the Mediterranean diet is associated with lower Guardian , abdominal adiposity in European men and women.” Smith, D. 2008. “Food suppliers poised to cash in on Delia effect.” Journal of Nutrition 139 February 10, 2008. Accessible at: . Rosegrant, M. W., and P. K. Thornton. 2008. “Do higher meat and milk prices Id21 insights 72: 4. Smith, P., M. Bustamante, H. Ahammad, H. Clark, H. Dong, E. A. Elsiddig, adversely affect poor people?” H. Haberl, R. Harper, J. House, M. Jafari, O. Masera, C. Mbow, N. H. Ravindranath, C. W. Rice, C. Robledo Abad, A. Romanovskaya, F. Sperling, Save Our Seas Foundation. 2015. “How many sharks are caught each year?” and F. Tubiello. 2014. “Agriculture, Forestry and Other Land Use (AFOLU).” n.d. Accessible at: . In O. Edenhofer, R. Pichs-Madruga, Y. Sokona, E. Farahani, S. Kadner, K. Seyboth, A. Adler, I. Baum, S. Brunner, P. Eickemeier, B. Kriemann, J. Scarborough, P., P. N. Appleby, A. Mizdrak, A. D. M. Briggs, R. C. Travis, Savolainen, S. Schlömer, C. von Stechow, T. Zwickel and J.C. Minx (eds.). Climate Change 2014: Mitigation of Climate Change. Contribution of K. E. Bradbury, and T. J. Key. 2014. “Dietary greenhouse gas emissions of Climatic Change Working Group III to the Fifth Assessment Report of the Intergovernmental meat-eaters, fish-eaters, vegetarians and vegans in the UK.” 125: 179–192. New York and Cambridge, UK: Cambridge Panel on Climate Change. University Press. | 79 | April 2016 WORKING PAPER

80 Tulchinsky, T. H. 2010. “Micronutrient deficiency conditions: global health Spencer, E. A., P. N. Appleby, G. K. Davey, and T. J. Key. 2002. “Diet and 32: 243–255. Public Health Reviews issues.” body mass index in 38,000 EPIC-Oxford meat-eaters, fish-eaters, vegetarians 27: 728–734. International Journal of Obesity and vegans.” Tyszler, M., G. F. H. Kramer, and H. Blonk. 2014. “Just eating healthier is not Stehfest, E., L. Bouwman, D. P. van Vuuren, M. G. J. den Elzen, B. Eickhout, enough: studying the environmental impact of different diet scenarios for the th and P. Kabat. 2009. “Climate benefits of changing diet.” 95: Netherlands by Linear Programming.” Paper presented at the 9 Climatic Change International Conference LCA of Food, San Francisco, CA, October 8–10, 2014. 83–102. Steinfeld, H., P. Gerber, T. Wassenaar, V. Castel, M. Rosales, and C. de Haan. UNDESA (United Nations, Department of Economic and Social Affairs, World Urbanization Prospects: The 2014 2006. “Livestock’s Long Shadow: Environmental Issues and Options.” Rome: Population Division). 2014. Revision. New York: United Nations. Food and Agriculture Organization of the United Nations. UNDESA (United Nations, Department of Economic and Social Affairs, Stummer, R. 2003. “Who are you calling pilchard? It’s ‘Cornish sardine’ World Population Prospects: The 2015 Revision. to you...” Independent , August 17, 2003. Accessible at: . 2012. UNEP (United Nations Environment Programme). The Emissions Gap Swinburn, B. A., G. Sacks, K. D. Hall, K. McPherson, D. T. Finegood, M. L. Nairobi: UNEP. Report 2012. Moodie, and S. L. Gortmaker. 2011. “The global obesity pandemic: shaped The Lancet UNEP (United Nations Environment Programme). 2013. The Emissions Gap by global drivers and local environments.” 378 (9793): 804–814. Nairobi: UNEP. Report 2013. Tacon, A. G. J., and M. Metian. 2008. “Global Over view on the Use of Fish Meal and Fish Oil in Industrially Compounded Aquafeeds: Trends and Future UNFCCC (United Nations Framework Convention on Climate Change). 2015. Aquaculture “Adoption of the Paris Agreement.” Bonn, Germany: UNFCCC. Prospects.” 285: 146–158. Nudge: Improving Decisions about Thaler, R. H., and C. R. Sunstein. 2008. What We Eat in USDA (United States Department of Agriculture). 2014. New Haven, CT: Yale University Press. Washington, DC: USDA. America, NHANES 2011–2012. Health, Wealth, and Happiness. Thow, A. M., S. Jan, S. Leeder, and B. Swinburn. 2010. “The effect of fiscal National Nutrient USDA (United States Department of Agriculture). 2015. Database for Standard Reference Release 28. Washington, DC: USDA. policy on diet, obesity and chronic disease: a systematic review.” Bulletin of the World Health Organization 88 (8): 609–614. Accessible at: < http://ndb.nal.usda.gov/ndb/foods>. USDA/ERS (United States Department of Agriculture, Economic Research Thow, A. M., S. Downs, and S. Jan. 2014. “A systematic review of the Ser vice). 2014a. “Food Expenditures.” Accessible at: . Nutrition Reviews recent evidence.” 72: 561–565. Tilman, D., C. Balzer, J. Hill, and B. Beford. 2011. “Global food demand and USDA/ERS (United States Department of Agriculture, Economic Research the sustainable intensification of agriculture.” Proceedings of the National Ser vice). 2014b. “Food-Away-From-Home.” Accessible at: . Tilman, D., and M. Clark. 2014. “Global diets link environmental USDA/ERS (United States Department of Agriculture, Economic Research sustainability and human health.” Nature 515: 518–522. Ser vice). 2015a. “Fruit and Vegetable Prices.” Accessible at: . Tom, M. S., P. S. Fischbeck, and C. T. Hendrickson. 2015. “Energy use, blue water footprint, and greenhouse gas emissions for current food consumption Environment Systems and USDA/ERS (United States Department of Agriculture, Economic Research patterns and dietary recommendations in the US.” Ser vice). 2015b. “Meat Price Spreads.” Accessible at: . Tootelian, D. H., and K. Ross. 2000. “Products labels: what information USDA/FAS (United States Department of Agriculture and Foreign Agricultural do consumers want and will they believe it?” Journal of Food Products Ser vice). 2014. Livestock and Poultry: World Markets and Trade . Washington Marketing 61 (1): 25–38. DC: USDA. Tufts University Health & Nutrition Letter. 2012. “Are You Getting Enough USDA/HHS (United States Department of Agriculture and United States Protein? For most Americans, the answer is an emphatic ‘Yes!’ but emerging research suggests older adults may need more high-quality protein Department of Health and Human Ser vices). 2015. “Scientific Report of the throughout the day.” September, 2012. Accessible at: . | 80

81 Shifting Diets for a Sustainable Food Future Variyam, J. N. 2005. “Nutrition Labeling in the Food-Away-From-Home Westhoek, H., J. P. Lesschen, A. Leip, T. Rood, S. Wagner, A. De Marco, D. Sector: An Economic Assessment.” Washington, DC: U. S. Department of Murphy-Bokern, C. Pallière, C. M. Howard, O. Oenema, and M. A. Sutton. Agriculture, Economic Research Ser vice. Nitrogen on the Table: The influence of food choices on nitrogen 2015. emissions and the European environment. (European Nitrogen Assessment Videl, J. 2014. “This could be the year we start to save, not slaughter, the Special Report on Nitrogen and Food.) Edinburgh, UK: Centre for Ecology & , January 11, 2014. Accessible at: . WHO (World Health Organization). 2003. “WHO/FAO release independent Vyth, E. L., I. H. M. Steenhuis, A. J. C. Roodenburg, J. Brug, and J. C. Expert Report on diet and chronic disease.” Geneva: WHO. Accessible at: Seidell. 2010. “Front-of-pack nutrition label stimulates healthier product . development: A quantitative analysis.” International Journal of Behavioral Nutrition and Physical Activity 7: 65. WHO (World Health Organization). 2012. “Fact sheet N°311: obesity and overweight.” Geneva: WHO. Accessible at: . Kaushik, R. Mungkung, S. Nawapakpilai, and M. Phillips. 2014. “Improving Productivity and Environmental Performance of Aquaculture.” Working Paper, WHO, FAO, and UNU (World Health Organization, Food and Agriculture Installment 5 of Creating a Sustainable Food Future. Washington, DC: World Organization of the United Nations, and United Nations University). Resources Institute. Protein and Amino Acid Requirements in Human Nutrition. 2007. Geneva: WHO. Accessible at: . 2014. Accessible at: . Winter, P. L., and J. R. Rossiter. 1988. “Pattern-Matching Purchase Behavior and Stochastic Brand Choice: A Low Involvement Product Category Walpole, S. C., D. Prieto-Merino, P. Edwards, J. Cleland, G. Stevens, and Model.” Working Paper. Los Angeles, CA: Graduate School of Management, I. Roberts. 2012. “The weight of nations: an estimation of adult human University of California, Los Angeles. BMC Public Health biomass.” 12: 439. Winterbottom, R., C. Reij, D. Garrity, J. Glover, D. Hellums, M. McGahuey, Wansink, B. 2002. “Changing Eating Habits on the Home Front: Lost Lessons and S. Scherr. 2013. “Improving Land and Water Management.” Working 21 (1): Journal of Public Policy & Marketing from World War II Research.” Paper, Installment 4 of Washington, DC: Creating a Sustainable Food Future. 90–99. World Resources Institute. Wansink, B., and A. S. Hanks. 2013. “Slim by design: ser ving healthy foods Wirsenius, S. 2000. Human Use of Land and Organic Materials: Modeling (10): PloS ONE 8 first in buffet lines improves overall meal selection.” the Turnover of Biomass in the Global Food System. PhD thesis, Chalmers e77055. University of Technology and Göteborg University. Water Footprint Network. 2016. “Frequently asked questions.” The Wirsenius, S., C. Azar, and G. Berndes. 2010. “How much land is needed Hague, The Netherlands: Water Footprint Network. Accessible at: < http:// for global food production under scenarios of dietary changes and livestock waterfootprint.org/en/water-footprint/frequently-asked-questions/#CP2>. productivity increases in 2030?” Agricultural Systems 103 (9): 621–38. Insights Weeks, J. 2012. “The Meat of the Issue.” , October 26, 2012. Journal of Wood, W., and D. T. Neal. 2009. “The habitual consumer.” Accessible at: . 19 (4): 579–592. Consumer Psychology Wellesley, L., C. Happer, and A. Froggatt. 2015. “Changing Climate, World Bank. 2015. World Development Indicators. Washington, DC: Changing Diets: Pathways to Lower Meat Consumption.” London: World Bank. Chatham House. Young, V., and P. Pellet. 1994. “Plant proteins in relation to human protein Westhoek, H., T. Rood, M. van de Berg, J. Janse, D. Nijdam, M. Reudink, and amino acid nutrition.” The American Journal of Clinical Nutrition 59 (5): The Protein Puzzle: The consumption and production and E. Stehfest. 2011. 12035–12125. The Hague: Netherlands of meat, dairy and fish in the European Union. Environmental Assessment Agency. Zhai, F. Y., S. F. Du, Z. H. Wang, J. G. Zhang, W. W. Du, and B. M. Popkin. 2014. “Dynamics of the Chinese diet and the role of urbanicity, 1991–2011.” Westhoek, H., J. P. Lesschen, T. Rood, S. Wagner, A. De Marco, D. Murphy- Obesity Reviews 15 (Suppl. 1), 16–26. Bokern, A. Leip, H. van Grinsven, M. A. Sutton, and O. Oenema. 2014. “Food choices, health and environment: Effects of cutting Europe’s meat and dairy 16: 196–205. Global Environmental Change intake.” | 81 | April 2016 WORKING PAPER

82 in the 2009–10 National Health and Nutrition Examination sur vey was ENDNOTES estimated at 2,081 kcal/capita/day, far lower than the 3,652 kcal/capita/ 1. Authors’ calculations from Searchinger et al. (2013), adjusted upward to day (food availability) given in FAO (2015) for the United States in reflect the latest United Nations estimate of 9.7 billion people as given in 2009. Even when the FAO food availability figure is adjusted downward UNDESA (2015). This crop calorie gap, which we estimate at 71 percent, for food waste, the corresponding estimate of food consumption is sometimes referred to as the “food gap” in this paper. This paper, and derived from FAO data in GlobAgri is still around 2,900 kcal/capita/ Creating a Sustainable Food Future series, rely on data from others in the day. Several factors could explain the discrepancy between FAO and the FAO Food Balance Sheets (FAO 2015) and an FAO projection of food natural sur vey estimates. In some contexts, people may underestimate demand and production by 2050 by long-time experts Jelle Bruinsma their actual calorie consumption in national sur veys. Another possible and Nikos Alexandratos (Alexandratos and Bruinsma 2012). Searchinger explanation is that there is even more food waste than estimated by FAO et al. (2013) and this paper’s authors adjusted the FAO 2050 projection (2011a). If waste figures are higher than estimated by FAO (2011a), our of food demand upward in two ways: (1) to ensure 3,000 calories per calculation of the land use and greenhouse gas consequences of diets person per day are available in all the world’s regions by 2050, and (2) to in each country would still be accurate (so long as FAO food availability reflect the latest United Nations mid-range population estimate for 2050. estimates are accurate). This error would just mean that more of the environmental burden of supplying food results from waste of that food Two possible ways to quantify human calorie requirements are calories along the supply chain. from crop production or calories from all food available directly for human consumption. Measuring food directly available to people omits Ongoing global efforts to produce better estimates of food consumption, calories in animal feed, but counts calories in animal products. Each and of food losses and waste, should in the future permit refinement of approach has its merits. The estimated food gap between 2006 and 2050 the dietary estimates in this paper. by either measure is similar, ranging from 71 percent for the needed UNDESA (2015). 9.7 billion people in 2050 reflects the medium-growth 2. increase in crop production to 68 percent for the needed increase in food scenario. calories available for direct human consumption (Searchinger et al. 2013, adjusted by the authors of this paper). 3. “Middle class” is defined by OECD as having per capita income of $3,650 to $36,500 per year or $10 to $100 per day in purchasing power Using the food balance sheets in FAO (2015) to estimate diets requires parity terms. “Middle class” data from Kharas (2010). several assumptions. For example, in nearly all countries, food balance sheets suggest more food available per person than people actually 4. UNDESA (2014). eat in part because “available food” includes food that people waste in their homes or dining out, and ultimately do not consume. To estimate Foresight (2011). 5. consumption, the GlobAgri model subtracted waste estimated from these food balance sheets by region based on FAO (2011a). The GlobAgri Searchinger et al. (2013). 6. consumption estimates compare quite favorably with our own estimates using data from Lipinski et al. (2013) and FAO (2011a), as well as the - Authors’ calculations, adjusted upward from FAO projections in Alexan 7. European Union consumption estimates using a different food waste dratos and Bruinsma (2012). See endnote 1 for more on adjustments. dataset reported in Westhoek et al. (2015). As widely acknowledged, 8. Searchinger et al. (2013). these waste estimates are rough. In addition, our analysis determined that some of the wastes estimated in FAO (2011a) are already reflected in 9. UNFCCC (2015). processing wastes that the food balance sheets use to compute available food from raw products. Our analysis adjusted for these processing 10. Searchinger et al. (2013). wastes. Garnett (2014a), Swinburn et al. (2011), FAO (2013), Tulchinsky (2010). 11. Strengths of these FAO data sources include the inclusion of nearly all of the world’s countries, relatively comparable methods across countries, Steinfeld et al. (2006). 12. and open access to data. Furthermore, food availability estimates are When diet shifts reduce agricultural land use, the resulting negative 13. ultimately derived from production and trade data. Use of FAO estimates emissions from land-use change represent one-time gains in land-based of food availability to estimate actual diets (consumption) is therefore carbon sequestration (or, alternatively, avoided future emissions from necessary to link food consumption estimates in a consistent way to food land-use change). production estimates (how many crops and animal products are actually produced, and where in the world that production occurs), which in turn 14. Pulses are annual leguminous crops har vested for dry grain, including is necessary to estimate the land use and greenhouse gas emissions beans, peas, and lentils. generated to supply the food produced for human consumption. In short, there is currently no alternative to using FAO data to make these Keats and Wiggins (2014), Khoury et al. (2014), Tilman and Clark 15. estimates. (2014), Popkin et al. (2012). However, FAO food balance sheets result in estimates of actual dietary 16. Delgado et al. (1999), Popkin (2003), Popkin and Gordon-Larsen (2004), intake in many countries that are inconsistent with separate estimates Kearney (2010). Indeed, Popkin (2003) notes that preferences for dietary of actual calorie consumption in those countries, typically undertaken fats and sugars may be an innate human trait. through national diet sur veys. For example, Del Gobbo et al. (2015) note that mean total energy intake (consumption) in the United States Khoury et al. (2014), Pingali (2007). 17. | 82

83 Shifting Diets for a Sustainable Food Future Freshwater availability on earth is determined by annual precipitation 18. Anand et al. (2015), Reardon et al. (2014). above land. One part of the precipitation evaporates and the other part 19. Popkin and Gordon-Larsen (2004). runs off to the ocean through aquifers and rivers. Both the evaporative flow and the runoff flow can be made productive for human purposes. 20. USDA and HHS (2010), WHO (2012). The evaporative flow can be used for crop growth or left for maintaining natural ecosystems; the green water footprint measures which part of - In this paper, we use the term “per capita [calorie or protein] availabil 21. the total evaporative flow is actually appropriated for human purposes. reaching the consumer , as defined in ity” to mean the quantity of food The runoff flow—the water flowing in aquifers and rivers—can be used the FAO Food Balance Sheets (FAO 2015). We use the term “per capita for all sorts of purposes, including irrigation, washing, processing, and , when actually consumed consumption” to mean the quantity of food cooling. The blue water footprint measures the volume of groundwater accounting for food waste at the consumption stage of the value chain. and surface water consumed. “Consumption” quantities (which exclude all food loss and waste) are therefore lower than “availability” quantities. Data on “per capita Since freshwater availability on earth is limited, it is important to know consumption” are from the GlobAgri model, using source data from FAO how it is allocated over various purposes, to inform discussions around (2015) on “per capita availability” and FAO (2011a) on food loss and use of water for maintaining natural ecosystems versus production of waste. Because historical rates of food loss and waste are unknown, food or energy, or around the use of water for basic needs versus pro - graphs showing trends from 1961 display “availability” instead of duction of luxury goods. Water footprint estimates, when overlaid with “consumption.” maps of water stress, can also identify “hotspots” where water footprint reduction is most urgent. 22. FAO (2015). 26. Data are from the most recent years possible. Most data are from 2009. Alexandratos and Bruinsma (2012). 23. Data on aquaculture production are from 2008 (as reported in Hall et al. 2011b and Waite et al. 2014), and data on water use efficiency are from 24. - While the FAO data paint a broad picture of food availability and con 1996–2005 (as reported in Mekonnen and Hoekstra 2011, 2012). sumption at the national level, food consumption sur veys, such as the China Health and Nutrition Sur vey, reveal differences in diets consumed 27. The analysis in this section—and similar sections analyzing the per by different population groups within countries (FAO 2015). In particular, capita effects of diet shifts later in the paper—uses actual average per diets vary between rural and urban areas and between high- and low-in - capita food consumption for the “United States and Canada” region in come groups. In China, for example, adults in urban areas consumed an 2009. Because the United States’ population was 90 percent of the total average of 400 calories from animal-based foods per day in 2011, while population of this region in 2009, and because consumption patterns those in rural areas only consumed 220 calories, and urban consumers across the US and Canada are quite similar, we present these findings as ate 40 percent more processed food per capita than rural consumers for “the United States” for simplicity’s sake. (Zhai et al. 2014). Given this variation in diets, inter ventions to shift diets will need to be carefully targeted in terms of countries and segments of GlobAgri model. More precisely, the per person land use and greenhouse 28. the population within countries. gas effects of each diet, as modeled in GlobAgri and shown in Figures 3, 6, 10, and 15, are the marginal effects of adding one additional person Land use and greenhouse gas emissions are estimated by GlobAgri. 25. to the world population in 2009. This is why the per person land-use Water use estimates are from authors’ calculations using data from change emissions are higher than the agricultural production emissions; Mekonnen and Hoekstra (2011, 2012). The following additional informa - because yields and trade patterns are held constant, GlobAgri estimates tion about the water use estimates are summarized from Hoekstra et al. the annual emissions that would result from converting the additional (2011) and Water Footprint Network (2016): land (roughly 0.5 hectares for the average world diet and roughly 1 hectare for the average US diet) from natural ecosystems to agricultural The water use estimates are divided into “blue” and “green” water production. footprints. “Blue water footprint” represents the volume of surface and groundwater consumed as a result of the production of a crop or 29. GlobAgri model. animal-based food (i.e., irrigation). “Water consumption” refers to the volume of freshwater used and then evaporated or incorporated into a 30. GlobAgri model. Note that land-use change emissions in Figure 3 are product. It also includes water abstracted from surface or groundwater amortized over a period of 20 years and then shown as annual impacts. in a watershed and returned to another watershed or the sea (but not The annual per person land-related agriculture emissions from consum - to the watershed from which it was withdrawn). “Green water footprint” ing the average world diet (around 8 tons CO e in 2009 as shown in 2 represents the volume of rainwater consumed during the production of a Figure 3), when multiplied by the world population of 6.8 billion, do - crop or animal-based food, and is equal to the total rainwater evapo not equate to the annual estimates of global land-use change emissions transpiration (from fields and plantations) plus the water incorporated (around 5 billion tons CO e globally per year as given in Smith et al. 2 into the har vested crop. In the case of grazing land, Mekonnen and 2014). (Multiplying 8 tons CO e per person by the world population 2 Hoekstra (2012) only calculate the evapotranspiration for the portion would lead to an estimate of more than 50 billion tons CO e globally, 2 of grass consumed by animals (versus all of the water evapotranspired about 10 times higher than actual land-use change emissions.) This is from the entire surface area). This narrower scope helps to explain why because GlobAgri estimates land-use change at the margin, and only 81 green water use in Figure 2 does not more closely track total land use million people (not 6.8 billion people) were added to the world popula - as calculated by GlobAgri (especially for cattle, which rely heavily on tion in 2009. Given steady growth in crop and livestock yields, there grasses for feed). would be no land-use change emissions if increases in food demand were fully met by agricultural productivity increases and people’s diets | 83 | April 2016 WORKING PAPER

84 did not change. Land-use change emissions occur when food demand threshold in Figure 4, in 2010–12, 12 percent of China’s population growth cannot be fully met by yield gains—as is currently the case. Each was undernourished, as were 6 percent in Nigeria, and 11 percent in individual’s consumption affects this quantity of expansion and emis - Indonesia, according to FAO, IFAD, and WFP (2015). This underscores the importance of properly targeting diet shifts at “overconsuming” sions, and the GlobAgri model attempts to estimate by how much. For segments of the population within a country or region more on calculations of land-use change emissions, see Box 4. Authors’ calculations. Total US energy-related emissions of 5,386 million 39. WHO (2012). 31. (EIA 2015), when divided by a US population of 306.8 million, tons CO 2 40. Gortmaker et al. (2011), Spencer et al. (2002), Campbell et al. (1992). equal per capita emissions of 17.6 tons CO e in 2009. Land-use-change 2 emissions of 300 tons CO e are therefore equal to roughly 17 times 2 USDA and HHS (2010), WHO (2012). 41. average US per capita energy-related CO emissions in 2009. Energy- 2 emissions are those stemming from the burning of fossil related CO 2 USDA and HHS (2010). 42. - fuels. These estimates differ in that the dietary land-use-change emis sions include the global consequences of diets, while the energy-related 43. OECD (2010). emissions calculate only those emissions from energy use within the US. Factoring in a portion of energy emissions associated with imported 44. American Diabetes Association (2008). products increases those US energy emissions somewhat. For example, Finkelstein et al. (2009). 45. Davis and Caldeira (2010) estimate that US consumption-based CO 2 emissions (defined as the amount of emissions associated with the (2014). Economist 46. consumption of goods and ser vices in a country, after accounting for imports and exports) were 22 tons per capita per year in 2004. 47. Bloom et al. (2011), Hojjat (2015). 32. The three diet shifts are interconnected because they are not mutually 48. Fry and Finley (2005). exclusive. Figures 6, 10, and 15, which show the effects of the three diet shifts on caloric consumption in the United States, make this point Finkelstein et al. (2010). 49. clear. The two scenarios that reduce overconsumption of calories (Figure 6) also reduce animal-based food consumption, including beef. The Behan and Cox (2010). 50. Ambitious Beef Ambitious Animal Protein Reduction (Figure 10) and 51. Bloom et al. (2011). (Figure 15) scenarios also reduce calories in all affected Reduction regions. In addition, although overall calorie consumption was held 52. Dobbs et al. (2014). and Traditional Mediterranean Diet constant in the Vegetarian Diet scenarios (relative to the reference levels) to isolate the effects of the 53. We chose the countries and regions in Figure 5 because they have high shifts away from resource-intensive foods (Figure 10), in practice a populations, are home to more than half of the world’s people, and cover shift to a Mediterranean or vegetarian diet could also reduce calorie a wide range of geographies and stages of economic development. The consumption (further reducing the associated agricultural land use and nine countries and regions shown in Figures 5, 9, and 14 include seven greenhouse gas emissions). of the ten most populous countries projected for 2050 (medium fertility scenario), plus Japan, which was the 11th most populous country in 33. The World Health Organization defines “overweight” as having a body 2015. The population of the European Union—the only region included mass index (BMI) greater than or equal to 25 and “obese” as having a in these figures—was 505 million in 2015. All countries and regions BMI greater than or equal to 30. BMI is an index of weight-for-height shown will have a population of at least 100 million in 2050 under that is commonly used to classify overweight and obesity in adults. It is UNDESA’s medium fertility scenario. All told, these countries and regions defined as a person’s weight in kilograms divided by the square of his 2 - were home to 60 percent of the world’s population in 2015 and are pro height in meters (kg/m ) (WHO 2012). jected to contain 53 percent of the world’s population in 2050 (Authors’ calculations from UNDESA 2015). Ng et al. (2014a). 34. 54. All statistics in this paragraph are from Ng et al. (2014a). Ng et al. (2014a). 35. FAO (2014). 55. 36. Ng et al. (2014a). Countries are listed in order of number of obese individuals. 37. FAO, WHO, and UNU (1985). 56. Alexandratos and Bruinsma (2012). 38. GlobAgri model with source data from FAO (2015) and FAO (2011a). Popkin et al. (2012). Although median levels of consumption would give the most accurate 57. picture of an “average” person’s consumption in a given country or 58. Popkin (2002). region, data presented in Figures 4, 7, and 11 are means, because means are the only globally available averages. Of course, countries 59. (2014). Economist exceeding the 2,353 calorie threshold on an average basis will likely have a percentage of their populations below the threshold. For 60. Cecchini et al. (2010). instance, although China, Nigeria, and Indonesia all lie above the 61. Monteiro et al. (2007). | 84

85 Shifting Diets for a Sustainable Food Future There is no perfect way to calculate greenhouse gas emissions from land 62. FAO, WFP, and IFAD (2012). 75. use attributable to agricultural demand. Our emissions estimates here are 63. Gortmaker et al. (2011). based on the GlobAgri model. For each individual crop or animal-based food, the model estimates the additional amount of land that would be 64. Jones-Smith et al. (2011). used to produce an additional quantity of that product as shown either in a region, a set of regions, or the world. It also estimates the amount Within a population, the degree to which calorie availability has peaked 65. of carbon this agricultural land would otherwise store. When forests and varies based on a number of factors, such as socioeconomic status and savannas are converted to annual cropland, the amount of carbon stored race/ethnicity. See Ng et al. (2014b) for a discussion relevant to the in vegetation is nearly eliminated, and while the soil carbon numbers United States. vary, a general estimate of a loss of 25 percent of the carbon in the top meter of soil is a reasonable estimate based on meta-analyses (Search - 66. Grecu and Rotthoff (2015). inger et al. 2015; Guo and Gifford 2002). When lands are converted from natural vegetation to agriculture, the bulk of carbon loss and therefore 67. FAO, IFAD, and WFP (2015). emissions occur quickly (although soils may continue to lose carbon for 68. FAO, IFAD, and WFP (2015). many years), and are one-time emissions. But based on the approach taken by the European Union for estimating emissions from land-use 69. FAO, IFAD, and WFP (2015). change for biofuels, we show one-twentieth of these emissions as the - annual emissions. This approach assigns one-twentieth of these emis See notes 1 and 21 for more on adjusting availability for consumption 70. sions to each year of crop production for twenty years. Although losses loss and waste. will not occur indefinitely, this approach recognizes the time value of reducing greenhouse gas emissions earlier rather than later, and it also GlobAgri model with source data from FAO (2015) and FAO (2011a). 71. provides a way of combining emissions from land-use change with those from food production into one level of total emissions. For most of the The first analysis of the per person effects of the diet scenarios involves 72. diet scenarios analyzed in this paper, there is a reduction in crop and quantifying the additional (marginal) agricultural land use and green - pasture demands and therefore a reduction in total land-use demands. house gas emissions required to add one average US resident to the The GlobAgri model estimates the amount of carbon that these lands world population in 2009 and then assessing how these would change would sequester over time by regrowing native vegetation and, in the under the diet scenarios. The second analysis of the global effects of case of abandoned cropland, rebuilding soil carbon. Because this carbon the diet scenarios involves quantifying actual global agricultural land regrowth occurs over longer periods of time, and because it is hard to use and greenhouse gas emissions in 2009, and then assessing how imagine a sudden diet shift by millions of people in a single year, we de - these would change under alternative diet scenarios applied across all cided it was not plausible to allocate these emissions over only a 20-year overconsuming populations. period. Furthermore, because global agricultural land is expanding—as For full descriptions of the data sources, calculations, and assumptions 73. food demand growth continues to outpace yield growth—the real-world consequences of reducing food demand under the scenarios modeled in underlying each scenario, and further detail on the caloric composition of this paper would be to avoid future land-use change. Therefore, in Tables the reference diets and scenarios, see Appendix B. Under all scenarios, 2, 3, and 4, we display these emissions as avoided future emissions agricultural yields (of crops and livestock) and food consumption (by from land-use change. For more on the calculations of greenhouse gas non-affected populations) were held constant. emissions from land-use change, see Box 4. FAO has estimated that consumption of 2,700 to 3,000 kcal/person/day 74. CAIT Climate Data Explorer (2015). will lead to obesity by people with sedentary lifestyles (FAO 2004). Using 76. the mid-point of 2,850 kcal, and assuming that an acceptable diet would 77. “Fish” is defined in this paper as all aquatic animals, including finfish, consist of 2,350 kcal/person/day, this estimate implies that the elimina - crustaceans, and mollusks. Other (less-commonly-consumed) animal- tion of obesity would reduce consumption by 500 kcal/person/day. This based protein sources include animal fats and offal (FAO 2015). estimate is also generally consistent with the estimate of the excess calorie consumption for extremely obese US adults—those with a BMI As noted above, the global-level data shown in Figure 2 mask variations 78. over 35—of roughly 500 kcal/person/day (Hall et al. 2011a). The latter among locations, production systems, and farm management practices maintain obese estimate represents the increased calorie consumption to (Box 5). conditions for US adults, and is actually more than double the increased obese. As Hall et al. (2011a) become calorie consumption necessary to 79. Similarly to other developed countries, the US government (CDC 2015) explain, the estimate represents a revised view upward compared to the lists the estimated daily requirement for protein as 56 grams per day for traditional view of only 200 kcal/person/day, which did not account for an adult man and 46 grams per day for an adult woman, or an average of the greater calorie intake required to maintain the larger body size of the 51 grams of protein per day. Paul (1989) estimates the average protein overweight or obese. The 500 kcal/day assumes that all obese children requirement at 0.8 g per kg of body weight per day. Since the average have a similar overconsumption. We assume half this amount for leading adult in the world weighed 62 kg in 2005 (Walpole et al. 2012), applying to and sustaining being overweight. With this assumption, we do not the rule of 0.8 g/kg/day would yield an estimated global average protein intend to imply that reducing calorie consumption is all that is needed requirement of 49.6 grams per day. Other international estimates are to reduce obesity in the global population; the focus of this paper is on lower still; for instance, FAO, WHO, and UNU (1985) estimate an average the potential for “shifting diets” to contribute to closing the food gap and requirement of 0.75 g/kg/day. Furthermore, these estimates are conser - - thus here we focus on overconsumption of calories (instead of comple vative to ensure that they cover individual variations within a population mentary approaches such as increasing physical activity). group; for example, the estimated protein requirement of 0.8 g per kg of | 85 | April 2016 WORKING PAPER

86 96. body weight per day given in Paul (1989) includes 0.35 g/kg/day as a Larsson and Orsini (2013), Rohrmann et al. (2013), Pan et al. (2012). safety margin. 97. Larsson and Orsini (2013). 80. FAO, WHO, and UNU (1985). Factors include age, sex, height, weight, Di Maso et al. (2013), Pan et al (2012). 98. level of physical activity, and pregnancy and lactation. 99. Binnie et al. (2014). GlobAgri model with source data from FAO (2015) and FAO (2011a). 81. Bouvard et al. (2015). “Processed meat” refers to meat that has been 100. 82. GlobAgri model with source data from FAO (2015) and FAO (2011a). Of transformed through salting, curing, fermentation, smoking, or other course, countries exceeding the threshold of consumption of 50 grams processes to enhance flavor or improve preser vation. Most processed - of protein per capita per day will likely have a percentage of their popula meats contain pork or beef, but might also contain other red meats, tions below the threshold. For example, Semba et al. (2016) found that poultry, offal (e.g., liver), or meat byproducts such as blood. in rural villages in southern Malawi, chronically malnourished young children were low in all essential amino acids, and more than 60 percent 101. FAO (2015). of these children were stunted. See the discussion around Figure 4. 102. - Authors’ calculations, adjusted upward from FAO projections in Alexan 83. Keats and Wiggins (2014), FAO (2015), Tilman and Clark (2014), Weeks dratos and Bruinsma (2012). See endnote 1 for more on adjustments. (2012). GlobAgri model with source data from FAO (2015) and FAO (2011a). 103. Popkin et al. (2012), Delgado et al. (1999). 84. 104. Weeks (2012), Fox and Ward (2007). 85. Godfray et al. (2010). 105. Foresight (2011). Neumann et al. (2010). 86. Delgado et al. (1999), Khoury et al. (2014). 106. 87. Godfray et al. (2010), Steinfeld et al. (2006). Zhai et al. (2014). Zhai et al. (2014) note that animal-based food con 107. - Pica-Ciamarra et al. (2011). 88. sumption is now rising in rural China as well; between 1991 and 2011, per capita animal-based food consumption in urban China remained 89. However, it is true that the possible effects of the diet shifts in this paper relatively unchanged, while it grew by about 30 percent in rural areas. would not necessarily be limited to livestock farmers. For example, given that livestock in overconsuming countries are fed largely on grains, 108. FAO (2012b). reducing consumption in those countries would lead to surplus grains and lower prices for grains globally. This could help poor consumers Authors’ calculations based on Alexandratos and Bruinsma (2012) with 109. in developing countries, but could hurt poor farmers. The GlobAgri WRI adjustments. See endnote 1 for more on adjustments. model did not estimate economic effects of the various diet scenarios in this paper, but such effects would need to be carefully monitored and 110. See Searchinger et al. (2013) and Waite et al. (2014). managed to avoid the diet shifts harming poor farmers. Tilman et al. (2011). 111. 90. GlobAgri model. “Agricultural land for animal-based food production” includes pastureland plus cropland used for growing feeds. For full descriptions of the data sources, calculations, and assumptions 112. underlying each scenario, and further detail on the caloric composition See Garnett et al. (2015) for an in-depth discussion on the various defi - 91. - of the reference diets and scenarios, see Appendix B. Under all sce nitions of “environmental efficiency” in animal-based food production, - narios, agricultural yields (of crops and livestock) and food consump tradeoffs, and implications for sustainability. tion (by nonaffected populations) were held constant. The analysis for the United States also included Canada. Mekonnen and Hoekstra (2012). 92. These minimum consumption levels give a buffer between world aver - 113. 93. Precise impacts depend on how animal welfare is defined (Fraser 2008), age daily energy requirements (2,353 kcal/capita/day) and average daily but generally increasing the number of animals confined in intensive, protein requirements (50 grams/capita/day), and are also equal to the cramped, industrial-style farm production systems, often with high levels minimum consumption levels used in Bajzelj et al. (2014). of ammonia, raises welfare concerns. Of course, production systems can be improved, but improving the conditions in which animals are kept 114. Overconsuming countries and regions in 2009 included Brazil, the can also create tradeoffs for resource use and environmental impacts, by United States and Canada, Latin America (ex. Brazil), the Middle East increasing feed requirements, greenhouse gas emissions, and land use and North Africa, and the European Union. relative to more intensive systems (Westhoek et al. 2011). 115. One way to picture this scenario is to remove the parts of the red bars Landers et al. (2012). Although data are limited, the quantity of 94. above the “60 grams of protein line” in Figure 4. We also ensured that antibiotics used in animal food production in the United States likely this reduction in protein consumption did not cause total calorie intake exceeds the quantity used to treat humans (HHS and CDC 2013). to drop below 2,500 calories per day. In regions where calorie intake did drop below 2,500, we adjusted animal-based protein consumption Dwyer and Hetzel (1980), Armstrong and Doll (1975), Sinha et al. 95. back upward until calorie intake was exactly 2,500. (2009), Larsson and Orsini (2013). | 86

87 Shifting Diets for a Sustainable Food Future CAIT Climate Data Explorer (2015). Ambitious Animal Protein Reduction Authors’ calculations. The 116. scenario 126. led global animal protein consumption to fall from 61.8 million tons in 127. Land-use change emissions were slightly positive under this scenario, 2009 to 51.0 million tons, a reduction of approximately 17 percent. This even though there was a small net reduction in agricultural land use. 17 percent figure is for the entire world, so it includes regions whose - This result is most likely due to the fact that the soil carbon in the addi diets were not altered. tional land that went into production under this scenario (e.g., for more pulses and fish) was slightly more than the carbon in the land taken out This global caloric reduction of 2.4 percent was greater than the caloric 117. reduction under the Halve Obesity and Overweight scenario (2.1 per of production (e.g., for beef), as estimated by GlobAgri. - cent) but less than the reduction under the Eliminate Obesity and Halve 128. FAO (2015). Overweight scenario (3.1 percent). 129. Global greenhouse gas emissions in 2009 were 44 billion tons (CAIT Anand et al. (2015), Buckland et al. (2011), Estruch et al. (2013), Fung 118. Climate Data Explorer). et al. (2009), Martinez-Gonzalez et al. (2011), Nunez-Cordoba et al. (2009), Romaguera et al. (2009), Scarmeas et al. (2006). The intention 130. GlobAgri model with source data from FAO (2015) and FAO (2011a). here is not to advocate that the whole world shift to a Mediterranean diet as eaten in the Mediterranean region, but to explore the effects of - 131. Authors’ calculations, adjusted upward from FAO projections in Alexan a commonly studied “healthy diet” on agriculture’s resource use and dratos and Bruinsma (2012). See endnote 1 for more on adjustments. environmental impacts. A “Mediterranean-style diet” could be adapted to all regional diets (see, for example, examples for adaptation in East 132. Rosegrant and Thornton (2008). Asia, South Asia, Middle East, Africa, North and South America, and Europe in Anand et al. 2015, Supplementary Table 1). Increasing pastureland productivity is another item on the menu for a 133. sustainable food future (Figure 1) and is addressed in Searchinger et al. 119. FAO (2015). (2013). - The effect of switching half of these regions’ populations to a Mediter 120. See Waite et al. (2014) for a discussion of the conversion efficiency and 134. ranean (or vegetarian) diet would also be equivalent to that of switch - environmental performance of aquaculture, including farmed finfish, ing the regions’ entire populations halfway toward the alternative diet. crustaceans, and mollusks. Farmed mollusks (e.g., clams, mussels, Regardless of the interpretation, we felt that a 50 percent switch was scallops, and oysters) and filter-feeding carps are even more efficient more plausible than a 100 percent switch when modifying the diets of than the other animal products shown in Figure 8 because they obtain entire regions. all their food from plankton and dead and decaying organic matter suspended in the surrounding water—meaning there is no “food-out/ 121. Scarborough et al. (2014). These data were the best and most recent terrestrial feed-in” ratio. representation of an actual (not stylized) vegetarian diet. We converted the raw consumption data from Scarborough et al. (2014) to GlobAgri Authors’ calculations from FAO (2015). 135. food categories to be able to compare the environmental effects of this diet to the others analyzed in this paper. 136. Wirsenius et al. (2010). 122. All results are from GlobAgri. Eshel et al. (2014). 137. 123. The vegetarian respondents in Scarborough et al. (2014) were self- 138. FAO (2011b). identified, and a small percentage (less than 1 percent) reported Authors’ calculations from Mekonnen and Hoekstra (2012) and average 139. eating some level of meat—which explains the small amount of beef protein content of animal-based foods in FAO (2015). consumption reported in the “Vegetarian” bars of Figure 10. Most of the “other animal products” category shown in the “Vegetarian” bars of Mekonnen and Hoekstra (2012). 140. Figure 10 is composed of eggs. Authors’ analysis based on UNEP (2012), FAO (2012a), EIA (2012), IEA 141. - This insight also suggests that any concerns about micronutrient defi 124. (2012), and Houghton (2008) with adjustments. Ambitious Animal Protein ciency in the scenarios analyzed (e.g., in the scenario, which reduces US animal product consumption by Reduction 142. Authors’ calculations (0.47 * 0.13 = 0.06). about half and overall calorie consumption by almost 400 kcal/person/ day relative to 2009 reference) could be readily addressed by adding in 143. Alexandratos and Bruinsma (2012). appropriate plant-based products to maintain a balanced diet, without greatly affecting overall land use and greenhouse gas emissions. For 144. Pan et al. (2012). example, Bajzelj et al. (2014) set minimums of three portions of veg - 145. Mintert et al. (2009). etables per day (136 kcal/capita/day) and two portions of fruit per day (119 kcal/capita/day) in their “healthy diets” scenarios—adjusting fruit Authors’ calculations from FAO (2015) and Alexandratos and Bruinsma 146. and vegetable consumption upward to meet these minimums would add (2012). a relatively small amount of land use and greenhouse gas emissions to the scenario results. FAO (2015). 147. 125. FAO (2015). Jar vis (1986). 148. | 87 | April 2016 WORKING PAPER

88 Alexandratos and Bruinsma (2012). 163. 149. USDA/FAS (2014). 164. Authors’ calculations. The 71 percent crop calorie gap shown in Figure 150. For full descriptions of the data sources, calculations, and assumptions 1 is equal to approximately 6,800 trillion calories. Alexandratos and underlying each scenario, and further detail on the caloric composition Bruinsma (2012) assume that 25 percent of all crops (measured by of the reference diets and scenarios, see Appendix B. Under all calories) will be dedicated to animal feed in 2050, or 16,300 * (0.25) = scenarios, agricultural yields (of crops and livestock) and food 4,075 trillion calories of feed crops. In a hypothetical extreme scenario consumption (by nonaffected populations) were held constant. where animal product consumption was completely eliminated by 2050, 151. Regions affected by this scenario included Brazil, the European Union, demand for animal feed would disappear and crop calorie needs in Latin America (ex. Brazil), and the US and Canada. Animal Pro 2050 would drop by 4,075 trillion calories. The Ambitious - scenario, when applied to projected 2050 consumption tein Reduction 152. One way to picture this scenario is that the parts of the bars above the patterns, would result in roughly half of animal-based protein being “world average” line in Figure 11 were removed. The scenario was - removed from the human diet relative to FAO’s baseline projection (au designed to ensure that reductions in beef consumption did not cause thors’ calculations). Therefore, under that scenario, crop calorie needs per capita daily calorie consumption to drop below 2,500 calories or in 2050 would drop by half of 4,075 trillion calories, or 2,038 trillion total protein consumption to drop below 60 grams. In regions where calories—equal to 30 percent of the 6,800 trillion calorie food gap. consumption did drop below 2,500 calories, beef consumption was adjusted back upward so calorie consumption was exactly 2,500. Beattie et al. (2010), Tootelian and Ross (2000). 165. 153. Authors’ calculations. The Ambitious Beef Reduction scenario led global Larson and Story (2009). However, studies have shown that labeling 166. beef-based protein consumption to fall from 7.9 million tons to 5.5 can provide an incentive to food companies to reformulate products to million tons, a reduction of approximately 30 percent. This 30 percent make them healthier (Vyth et al. 2010, Variyam 2005). figure is for the entire world, so it includes regions whose diets were 167. Capacci et al. (2012). not altered. Dumanovsky (2011). 168. Regions affected by this scenario included Brazil, the European Union, 154. the former Soviet Union, Latin America (ex. Brazil), US and Canada, 169. Ajani et al. (2004), Han et al. (2011). and other OECD. They included all regions affected by the Ambitious Beef Reduction scenario but also others that consumed less than 2,500 170. Hammer et al. (2009). calories per capita per day in 2009, so affected nearly 2 billion people in all. 171. Winter and Rossiter (1988), McDonald et al. (2002). 155. The 33 percent level was chosen following the obser vation that FAO Wood and Neal (2009). 172. (2015) data show that in the US, EU, and Japan, per capita consumption has already declined by 27–40 percent from peak levels. We assumed Ji and Wood (2007). 173. that similar reductions could be achieved in other high-consuming regions, and that further reductions in Northern America and Europe Thaler and Sunstein (2008), Garnett (2014b), Bailey and Harper (2015), 174. were plausible. Wellesley et al. (2015). 156. All results are from GlobAgri model. House of Lords (2011). 175. 157. FAO (2015). Reardon and Timmer (2012). 176. 158. CAIT Climate Data Explorer (2015). 177. Reardon et al. (2003). 159. FAO (2015). Reardon et al. (2003), Reardon and Timmer (2012). 178. 160. The final World Resources Report will use GlobAgri to quantify 179. Reardon and Timmer (2012). the environmental effects of the diet shifts on 2050 baseline food 180. Reardon et al. (2007). consumption. Pingali (2007). 181. Authors’ calculations. See Appendix B for more on assumptions related 161. to the diet scenarios in 2050. The 2050 population projections are from USDA/ERS (2014a). 182. UNDESA (2015). 183. Hawkes (2008) in Garnett and Wilkes (2014). As an example, the one scenario that required more crop calories 162. Shift from Beef to Pork and Poultry relative to reference in 2009—the USDA/ERS (2014b). 184. scenario, which involved a switch from a predominantly grass-fed meat to two predominantly crop-fed meats—could switch to a “gap narrowing” scenario in 2050 if beef production becomes more crop-fed and pork and poultry production become more efficient in the use of crop-based feeds. | 88

89 Shifting Diets for a Sustainable Food Future Fast-moving consumer goods are products that are sold quickly and at Wansink and Hanks (2013). 185. 196. relatively low cost (including foods and beverages). 197. Just and Wansink (2009). 186. Brinsden et al. (2013). Arnold and Pickard (2013). 198. DEFRA (2014a). 187. 199. Cor valán et al. (2013). Stummer (2003), Cornish Sardine Management Association (2015). 188. 200. Sharp (2010), Chapter 12. Birds Eye (2015). 189. 201. Sharp (2010), Chapter 12. 190. Keats and Wiggins (2014). 202. Robinson et al. (2014). 191. Haley (2001). 203. Hickman (2011). 192. Ecorys (2014), Thow et al. (2014), Nordström and L. Thunström (2009), 204. Smith (2008). Hawkes (2012), Thow et al. (2010), Jensen and Smed (2013), Colchero et al. (2016). 205. House of Lords (2011), Reisch et al. (2013). Beattie et al. (2010), Tootelian and Ross (2000), Capacci et al. (2012), Dumanovsky (2011). 193. Economist (2012). 206. USDA/HHS (2015). Caraher and Cowburn (2005). 194. 207. Wansink (2002). For example, based on average US retail prices in 2013, the price per 195. gram of protein ranged from 0.9 cents for dried lentils, 1.1 cents for 208. See note 200. wheat flour, 1.2 cents for dried black beans, and 2.3 cents for dried white rice, to 2.7 cents for eggs, 2.9 cents for milk, 3.1 cents for fresh 209. Kaiser Permanente (2013). whole chicken, and 4.4 cents for ground beef. Authors’ calculations based on USDA/ERS (2015a), USDA/ERS (2015b), BLS (2015), and USDA/HHS (2015). 210. USDA (2015). OECD (2014), Ng et al. (2014a). 211. | 89 | April 2016 WORKING PAPER

90 ABOUT THE AUTHORS ACKNOWLEDGMENTS Janet Ranganathan (Vice President for Science & Research, WRI) - The authors would like to acknowledge the following individuals for their valu able guidance and critical reviews: Neal Barnard (Physicians Committee for Contact: [email protected] Responsible Medicine), T. Colin Campbell (Cornell University), Christopher Delgado (WRI), Shenggen Fan (IFPRI), Tara Garnett (Food Climate Research (Senior Fellow, WRI) Daniel Vennard - Network, University of Oxford), Hilary Green (Nestlé), Michael Hamm (Depart Richard Waite (Associate, WRI) ment of Food Science and Human Nutrition, Michigan State University), Craig Hanson (WRI), Susan Levin (Physicians Committee for Responsible Medicine), Contact: [email protected] Jacqueline Macalister (IKEA), Nicky Martin (Compass Group), Carlos Nobre (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior), Joanne (Associate, WRI) Brian Lipinski Lupton (Texas A&M University), Charles McNeill (UNDP), Miriam Nelson (Friedman School of Nutrition Science and Policy, Tufts University), Barry Popkin (Nutrition Transition Research Program, University of North Carolina AUTHORS AND GLOBAGRI-WRR Chapel Hill), Matthew Prescott (Humane Society of the United States), Anne Roulin (Nestlé), Ryan Sarsfield (WRI), Tim Thomas (IFPRI), Ben Welle (WRI), MODEL AUTHORS Laura Malaguzzi Valeri (WRI), Klaus von Grebmer (IFPRI), Sivan Yosef (IFPRI), Patrice Dumas (Researcher, Centre de Coopération Internationale en Deborah Zabarenko (WRI), and Li Zhou (Chinese Academy of Social Sciences). Recherche Agronomique pour le Développement, CIRAD) Thanks to the following individuals for their valuable assistance and Tim Searchinger (Senior Fellow, WRI; Research Scholar, Princeton University) contributions: Francis Gassert (WRI), Janice Ho (WRI), Chuck Kent (WRI), Colin McCormick (WRI), Allison Meyer (WRI), Aaryaman Singhal (WRI), Caroline Contact: [email protected] Vexler (WRI), William Hua Wen (WRI), and Lauren Zelin (WRI). In particular, we thank Peter Scarborough and Paul Appleby of the University of Oxford for providing data on vegetarian diets. GLOBAGRI-WRR MODEL AUTHORS We also thank Emily Schabacker for style editing, and Bob Livernash and Agneta Forslund (Institut national de la recherche agronomique, INRA) editing and proofreading. In addition, we thank Carni Hyacinth Billings for copy Klirs and Julie Moretti for publication layout and design. (INRA) Hervé Guyomard For this working paper, WRI is grateful for the generous financial support of Stéphane Manceron (INRA) the Norwegian Ministry of Foreign Affairs, the Netherlands Ministry of Foreign (INRA) Elodie Marajo-Petitzon - Affairs, the United Nations Development Programme, the United Nations Envi ronment Programme, and the World Bank. (INRA) Chantal Le Mouël This working paper represents the views of the authors alone. It does not necessarily (IIASA) Petr Havlik represent the views of CIRAD, INRA, or the World Resources Report’s funders. (Commonwealth Scientific and Mario Herrero Industrial Research Organisation, CSIRO) ABOUT WRI Xin Zhang (Princeton University) World Resources Institute is a global research organization that turns big ideas into action at the nexus of environment, economic opportunity and Stefan Wirsenius (Chalmers University of Technology) human well-being. (European Commission Joint Research Centre) Fabien Ramos Our Challenge Natural resources are at the foundation of economic opportunity and human Xiaoyuan Yan (Chinese Institute for Social Science) well-being. But today, we are depleting Earth’s resources at rates that are not sustainable, endangering economies and people’s lives. People depend on Michael Phillips (WorldFish) clean water, fertile land, healthy forests, and a stable climate. Livable cities and clean energy are essential for a sustainable planet. We must address Rattanawan Mungkung (Kasetsart University) these urgent, global challenges this decade. Our Vision This paper uses the GlobAgri-WRR model developed by CIRAD, Princeton - We envision an equitable and prosperous planet driven by the wise manage - University, INRA, and WRI. A separate version of GlobAgri (GlobAgri-PLU ment of natural resources. We aspire to create a world where the actions of RIAGRI), which has many differences but shares some common databases, government, business, and communities combine to eliminate poverty and was used for Le Mouël et al. (2015). sustain the natural environment for all people. Copyright 2016 World Resources Institute. This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of the license, visit http://creativecommons.org/licenses/by/4.0/ www.WRI.org 10 G Street, NE | Washington, DC 20002 |

Related documents