Reducing climate change impacts from the global food system through diet shifts

The study investigates how consumer dietary choices contribute unequally to greenhouse gas (GHG) emissions across and within countries and quantifies potential emission changes if global diets shift toward the EAT-Lancet planetary health diet. Contextually, the food system accounts for about one-third of global anthropogenic GHG emissions, and rising food consumption—especially animal-based diets—is linked to obesity, non-communicable diseases, and substantial emissions, while hundreds of millions still face hunger and billions cannot afford healthy diets. Extending food production to end hunger risks exacerbating climate change, suggesting that altering consumer lifestyles is necessary. Prior proposals, notably the EAT-Lancet planetary health diet, aim to improve health while keeping food-system impacts within planetary boundaries. However, evidence is limited on how different population groups would contribute differently to emissions and to mitigation under diet shifts. This study aims to fill these gaps by assessing detailed, product-level dietary emissions and their distribution by consumer expenditure groups globally, and by modeling emissions changes under universal adoption of the planetary health diet.

Existing research indicates that dietary shifts aligned with the UN Sustainable Development Goals can simultaneously address hunger (SDG 2), health (SDG 3), and climate mitigation (SDG 13). The EAT-Lancet planetary health diet provides flexible reference intake levels by food category to balance human and planetary health. Previous studies estimated environmental impacts (GHG emissions, water use) of adopting the planetary health diet at country levels, and survey-based analyses have examined income- or expenditure-specific food-related emissions within individual countries. Broader work using improved household consumption data reveals inequality in energy use and carbon emissions. Yet there has been no global assessment that disaggregates food-related emissions by specific products and detailed population groups, nor a clear understanding of how emissions from different groups would change under global dietary shifts. This study builds on that literature by combining a product-level consumption-based emissions inventory with detailed household expenditure data to quantify inequality and mitigation potentials at global scale.

The study quantifies consumption-based dietary GHG emissions (CO2, CH4, N2O) along full food supply chains, including agricultural land use and land-use change (LULUC), agricultural production, and beyond-farm processes, for 140 food products and 139 countries (baseline year 2019). Emissions are allocated to final food consumers using a physical trade-flow based consumption accounting framework built from FAOSTAT data on production, trade, and supply-chain emissions. Food loss and waste (FLW) are treated by subtracting household-level FLW (using regional factors) from national food supply to estimate net dietary intake. Quantities are converted to calorie content using FAO nutritive factors to enable aggregation into 13 food categories (grains; tubers/starchy vegetables; vegetables; fruits; dairy; red meat—beef, lamb, pork; poultry; eggs; fish; legumes; nuts; added fats; sugars) aligned with the EAT-Lancet framework. To capture within-country heterogeneity, national dietary intake by product is matched to 201 expenditure groups per country using a global household-expenditure dataset (built on the World Bank Global Consumption Database and supplemented with surveys for high-income countries). The approach assumes food consumption is proportional to food expenditure shares across 11 food items mapped to the 140 products, with constant prices across groups and identical domestic/import sourcing shares by category. Countries missing expenditure microstructure are imputed from similar neighbors. Results cover 201 groups in 139 countries, representing ~95% of the global population in 2019. Per capita dietary GHG footprints (annual) are calculated for each group, country, and region. Inequality is assessed via a GHG footprint Gini (GF-Gini) coefficient computed across the 201 groups for each country and food category. Relationships between GF-Gini and per capita GDP are explored using locally weighted regression and logarithmic regression. A planetary health diet (PHD) scenario assumes all groups in all countries attain the EAT-Lancet reference caloric intake levels by category (uniform daily adult reference at 2,500 kcal, with category composition preserved as in current national consumption). Emission changes are estimated by comparing current intake to PHD reference levels by category and applying current total emissions per calorie for each category and country. The scenario assumes unchanged production technologies, trade patterns, emission intensities, and sourcing shares; it excludes effects of land-use change from reduced production, price-induced demand shifts or affordability constraints, and inter-sectoral spillovers. Uncertainty in dietary emissions is quantified via Monte Carlo simulations reflecting uncertainties in activity data, emission factors, and process parameters.

- Global dietary emissions in 2019 are 11.4 GtCO2e (95% CI: 8.2–14.7 Gt). China (13.5%) and India (8.9%) are the largest national contributors; the top 10 contributors account for 57.3% of global dietary emissions. - Highest country-average per capita dietary footprints: Bolivia (6.1 tCO2e), followed by Luxembourg, Slovakia, Mongolia, the Netherlands, and Namibia (>5.0 tCO2e). Lowest: Haiti (0.36 tCO2e) and Yemen (0.38 tCO2e), followed by Burundi, Ghana, and Togo. - By source and calories: animal-based products contribute ~52% of emissions vs. plant-based ~48%, yet plant-based provide ~87% of calories. Red meat (5% of calories) contributes 29% of emissions; grains (51% calories) 21%; dairy (5% calories) 19%—reflecting higher emission intensities per calorie for red meat and dairy. - Regional patterns: In Australia (84%), the United States (71%), and Rest of East Asia (71%), animal-based products dominate emissions. In Indonesia, Rest of Southeast Asia, and Sub-Saharan Africa, plant-based staples (notably grains, tubers, legumes, nuts) dominate emissions due to high consumption volumes. - Within-country inequality: Per capita footprints generally rise with expenditure, mainly through higher red meat and dairy intake among wealthier groups. Inequality (GF-Gini) is highest in low-income countries and declines with higher per capita GDP, especially for animal-based products. In Sub-Saharan Africa, the top 10% by expenditure account for 40% of red meat, 39% of poultry, and 35% of dairy emissions. Western Europe shows relatively low inequality (GF-Gini <0.20 across products) compared to the United States, Australia, Canada, and Japan. - Diet shift scenario (planetary health diet): Global dietary emissions would decline by 17% (1.94 GtCO2e; 1.51–2.39 Gt) relative to 2019. Overconsuming groups (56.9% of population) would reduce emissions by 32.4% of the global total, offsetting a 15.4% increase from underconsuming groups (43.1% of population) moving to healthier intakes. - National changes: Emissions decline in 100 countries by a cumulative 2.88 GtCO2e; emissions increase in 39 countries (mainly low- and lower-middle-income in Sub-Saharan Africa and South Asia) by 0.938 GtCO2e. Largest percentage decreases: Uzbekistan (-74%), Australia (-70%), Qatar (-67%), Turkey (-65%), Tajikistan (-64%). Largest percentage increase: Iraq (+155%). Emission change correlates with per capita GDP, shifting from positive to negative with rising income, driven by animal-based emission changes. - Drivers by category: Reductions are dominated by red meat and grains. Meat/eggs/fish reductions total 2.04 GtCO2e, 94% from red meat. Largest contributors to red meat reduction: China (22%), United States (15%), Brazil (14%). Grain reductions yield 914 MtCO2e (56% in Asia). Additional reductions from sugars (240 Mt) and tubers (89 Mt). Offsetting increases arise from legumes and nuts (757 Mt), added fats (279 Mt), dairy (143 Mt), and vegetables/fruits (163 Mt). Intake of legumes and nuts increases in all regions. - Distributional effects: Per capita footprint declines concentrate in wealthier groups in high- and upper-middle-income countries (notably where red meat and dairy are dominant), while increases occur mainly among poorer groups in South and Southeast Asia and Sub-Saharan Africa due to current low red meat intake and across-the-board increases in plant-based proteins to meet reference levels.

Findings show that dietary GHG emissions are highly unequal across and within countries and strongly associated with expenditure and income levels. Shifting to the planetary health diet can simultaneously reduce global dietary emissions by 17% and alleviate underconsumption for 43.1% of the global population. The largest mitigation and equity gains come from reducing overconsumption of emission-intensive animal products—especially red meat and dairy—among affluent populations. Policy implications include targeted, context-specific measures rather than one-size-fits-all approaches: pricing reforms (e.g., taxes on environmental externalities, subsidies), ecolabeling, and increasing availability and attractiveness of lower-emission foods (e.g., menu design) to guide choices. Urban food environments and infrastructure can support healthier, lower-emission diets. Low-income countries face dual challenges of increasing nutritious food access and affordability while managing emissions. In regions like Sub-Saharan Africa, reaching PHD targets implies large increases in dairy and other nutrient-rich foods, which current domestic production systems and import capacities may not readily meet. Strategies include improving agricultural productivity and efficiency (crop/soil management, high-yielding varieties), adjusting trade policies to improve access to nutrient-rich foods, and carefully designing affordability interventions to avoid unintended distributional harms (e.g., lower farm incomes, widened wealth gaps). System-wide consequences of demand shifts include major restructuring of global agri-food production—substantial decreases in red meat, sugars, tubers, and grains supply (by calories), and large increases in legumes/nuts, added fats, and fruits/vegetables—with potential price fluctuations and spillovers to other sectors (e.g., biofuels). Policies should mitigate adverse spillovers, support scaling of low-emission supply chains (e.g., plant proteins) via incentives, and manage transitions in emission-intensive sectors (e.g., gradual crop substitution) to protect producers’ livelihoods.

This study provides a detailed global accounting of dietary GHG emissions by 140 products across 139 countries and disaggregates them by 201 household expenditure groups, revealing pronounced inequalities and identifying major product categories and consumer segments for targeted mitigation. It shows that universal adoption of the EAT-Lancet planetary health diet could reduce global dietary emissions by 17%, primarily by shifting protein sources away from red meat toward legumes and nuts, with reductions concentrated among overconsuming, wealthier populations. The results underscore the need for tailored, equity-sensitive policies that reduce overconsumption of emission-intensive foods in affluent contexts while enhancing affordability and access to nutrient-rich foods in low-income settings. Anticipated structural shifts in global food supply—large decreases in red meat (-81% by calories), sugars (-72%), tubers (-76%), and grains (-50%), and increases in legumes/nuts (+438%), added fats (+62%), and vegetables/fruits (+28%)—will require coordinated supply-side adjustments. Future research should integrate supply-side feasibility, price and affordability dynamics, and long-run system feedbacks (including land-use change, spillovers, and income effects) to design effective, just transitions in the food system.

Key limitations include: (1) Food loss and waste factors are regional and aggregated by category; differences in inedible fractions (e.g., bones, peels) may remain. (2) Household expenditure microdata are from 2011 (WBGCD), assuming unchanged group population shares and food expenditure shares; some major countries were imputed from similar peers. (3) Within-category product composition for groups is assumed equal to national composition; uniform prices across groups are assumed, ignoring quality/price variation. (4) Scenario assumptions hold production technologies, trade patterns, emission intensities, and sourcing shares constant; do not consider carbon uptake from land abandonment, non-food land uses, price-induced demand changes, affordability constraints, or cross-sector spillovers. (5) Income and expenditure levels of groups are held fixed; potential feedbacks from food-system changes to household incomes and budgets are not modeled. Uncertainty in emissions is addressed via Monte Carlo simulations, but additional data and modeling (e.g., elasticities) are needed for long-run projections.