Exploring environmental and distributional impacts of different transition pathways for healthier and sustainable diets: an economic modelling study

Exploring environmental and distributional impacts of different transition pathways for healthier and sustainable diets: an economic modelling study Marijke Kuiper, Thijs de Lange, Willem-Jan van Zeist, Hans van Meijl Summary Background The EAT–Lancet (EL) report made a convincing case that a transformative diet shift could yield substantial health benefits while helping to respect key planetary boundaries. Shifting to a more plant-based EL diet requires an unprecedented break from historic trends of rising meat consumption. By exogenously shifting diets, existing studies do not provide guidance on how to shift diets. In this study, we model specific policies that might achieve such a dietary shift and their potential economic, environmental, and distributional impacts. Methods In this economic modelling study, we used MAGNET, a global computable general equilibrium model, to explore how diets might be shifted from the business-as-usual trend of increased meat consumption between 2025 and 2050. We defined a policy bundle of three types of context-specific interventions: (1) nudging and information to shift consumer decisions, (2) adjusting fiscal policies by removing taxes on encouraged foods and subsidies on discouraged ones, and (3) introducing new price signals, such as taxes on high-emission foods and subsidies for encouraged foods. To evaluate how choice of interventions affects economic, environmental, and distributional outcomes, we analysed interactions of combined interventions and compared the policy bundle to an exogenous shift to the EL diet. Findings The EL diet recommendations cannot be reached by the policy bundle. While reducing overconsumption and underconsumption, the policy bundle left a big gap with EL diet recommendations (3 times the recommended intake for red meat, and only 80% of recommended fruit and vegetable intake, 50% for pulses and nuts). No single intervention from the policy bundle shifted all diet components in the desired direction. Decomposition of the policy bundle showed the importance of regional context. In low-income regions, taxes and subsidies accounted for the largest share in the diet shift. In other regions, nudging and information had a stronger effect than did subsidies. The exogenous EL diet scenario assumed a shift in diets beyond the range observed in empirical studies and produced a counterproductive feedback by reducing the affordability of the EL diet. GHG emissions from the primary sector (agriculture and fisheries) decreased more with the policy bundle (–4 GTon CO 2 equivalent), as GHG taxes provided incentives to reduce fossil-based inputs lacking in the consumer-focused exogenous EL diet scenario (–3 GTon CO 2 equivalent). The shift away from fossil-based inputs also led to an increase in agricultural land area (27 million ha) with the policy bundle, while the exogenous EL diet scenario resulted in a decrease in agricultural land (–35 million ha). Affordability for the average household decreased when exogenously shifting the EL diet (EL diet costs increased 26⋅1%, household income 0⋅2%), but it increased with the policy bundle as subsidies lowered EL diet costs (–4⋅4%) more than it lowered income (–2⋅0%). Affordability judged against daily wages worsened most when exogenously shifting the EL diet (27⋅7% of the workforce) and increased slightly with the policy bundle (24⋅8% of the workforce). Most of the affected workers were in encouraged primary sectors (fruit and vegetables, pulses and nuts, fish), but the negative effects extended to the non-food sectors with the exogenous EL diet. Interpretation The interventions chosen to change diets matter for the shift that can be attained and for the direction and size of the environmental, affordability, and distributional impacts. Price incentives can be a more effective, scalable, and just way to shift diets than the often preferred shift in consumer decisions, which make the EL diet more costly. Part of the trade-offs could be addressed by expanding the policy bundle with interventions to lower prices of encouraged foods (productivity increase, food loss and waste reduction) or reduce GHG emissions of food and non-food production, and with regulation of agricultural land expansion to reduce pressure on biodiversity. Increased productivity in lower-income regions could address the effect of past limited access to food and non-food technologies driving an increased reliance on imports in all scenarios. If designed well, it could improve affordability by lowering prices while increasing wages. Stimulating imports might help moderate prices of encouraged foods in case of unsurmountable local production limitations or preference for investing in manufacturing and services. These additional interventions address price and income concerns but are likely insufficient to attain a global shift to the EL diet, as food consumption is determined by more than affordability and nutritional needs. Lancet Planet Health2025; 9: 101327 Published Online November 3, 2025 https://doi.org/10.1016/ j.lanplh.2025.101327 Wageningen Social & Economic Research, (M Kuiper PhD, T de Lange MSc, W-J van Zeist PhD, Prof H van Meijl PhD), Department of Agricultural Economics and Rural Policy (T de Lange MSc, Prof H van Meijl PhD), Wageningen University & Research, Wageningen, Netherlands Correspondence to: Dr Marijke Kuiper, Wageningen Social & Economic Research, Wageningen University & Research, 6700 AB Wageningen, Netherlands [email protected] Articles www.thelancet.com/planetary-health Vol 9 October 2025 1 Funding Atkinson Center for Sustainability at Cornell university; Wageningen Social & Economic Research at Wageningen University & Research; CGIAR Research Initiative on Sustainable Healthy Diets through Food Systems Transformation (SHiFT); Horizon Europe Programme through the BrightSpace and ForestNavigator projects. Copyright © 2025 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Introduction The EAT–Lancet (EL) report 1 made a convincing case that a transformative diet shift could yield substantial health benefits while helping to respect key planetary boundaries. Shifting diets can address both global health and environmental crises of which the dire consequences are becoming increasingly apparent. The proposed transformation entails a shift towards a more plant-based diet, increased productivity of food production, and a reduction in food loss and waste. 1 Policy makers can learn from past gains in productivity and reduction of food loss and waste, but challenges such as yield gaps persist. In contrast, shifting to a more plant-based diet requires an unprecedented break from rising meat consumption as incomes rise, alongside strengthening increased fruit and vegetable consumption. This study focuses on how this break in household consumption patterns might be achieved. Most studies assume consumers to adhere to the EAT–Lancet diet, highlighting potential benefits but not specifying required interventions. 2–4 In this study, we model specific policies (nudging and information, taxes, and subsidies) that might achieve such a dietary shift compared with the business-as-usual (BAU) trend of increased meat consumption between 2025 and 2050. The interventions chosen to shift toward the EL diet affect the feasibility of the shift and its impact, which we assess in Research in context Evidence before this study The existing evidence on potential gains of the EAT–Lancet (EL) diet is dominated by empirical studies using detailed intake data to compute health and environmental gains from shifting diets. The environmental gains are generally based on lifecycle assessment data that do not take changes in production, trade, or non-food production and consumption into account. Existing studies of affordability of healthy diets use current prices or omit changes in income from a diet shift, providing only a partial assessment. Existing economic modelling studies generally impose a healthy diet as an exogenous consumption shift and are predominantly agricultural models omitting economy-wide feedback. Added value of this study Our study is, to our knowledge, the first to explicitly compare economic, environmental, and affordability impacts of exogenous consumption shifts to a policy bundle in an economywide global model. While the policy bundle comprises contextspecific interventions, they represent three generic types of policies making our findings more generally applicable: nudging and information, removing existing price distortions, and introducing new price distortions. Breaking down impacts of separate interventions and replicating food consumption from the policy bundle through an exogenous consumption shift provides new insight in how the choice of interventions matters for all three types of impacts considered in this study. Implications of all the available evidence Our findings show that the choice of intervention determines the extent to which the EL diet can be reached and thus the health gains that can be attained. Consumption shifts and price-based interventions generate different economy-wide feedback loops generating different economic, environmental and affordability outcomes. Relying on shifting consumption with nudging and information makes EL diets less affordable as it lacks targeted signals to producers to redirect production decisions provided by taxes, which can also fund subsidies to lower EL diet costs. Looking beyond primary food production is key with any largescale food system shift. Environmental gains are limited in all scenarios compared with agriculture-focused studies because of increased non-food production and consumption. Although greenhouse gas taxes are effective in reducing emissions, agricultural land use increases by biomass replacing fossil fuels and extensification of crop production. Shifting to a more plant-based diet increases the demand for agricultural labour in all scenarios making affordability of the EL diet for workers an increasing concern due to a persistent wage gap with non-agricultural sectors. Additional interventions are especially needed to improve availability of fruit and vegetables. These interventions should look beyond productivity and food loss and waste reductions to international trade in case of strong local production limitations. Any intervention needs to consider wage impacts and movement of workers from discouraged to encouraged sectors—the increasing number of workers not able to afford the EL diet are in sectors producing encouraged food. Attaining a just shift to a healthier and more sustainable food system requires not only better empirical foundations for shifting consumer decisions as food consumption is determined by more than affordability and nutritional needs. It also requires creative thinking on new interventions and sustained collective action to fundamentally change food systems. Articles 2 www.thelancet.com/planetary-health Vol 9 October 2025 terms of environment (greenhouse gas [GHG] emissions, agricultural land use), economic consequences (minimum cost of the EL diet, household income, wages), and affordability of the EL diet (relative to household food expenditures, share of workers unable to afford the EL diet). To evaluate how choice of interventions affects outcomes, we broke down the policy bundle into the contributions of interventions and compared the policy bundle to an exogenous shift to the EL and policy bundle diet. Methods Devising policy bundles to shift future diets There is no silver-bullet policy as the EL diet shift aims for healthy and sustainability goals while making the transition just. 5 The aims for health and sustainability both address externalities, such as correcting the failure of markets to account for health and environmental costs of current diets. The most effective point of intervention depends on the goal. Increasing consumption of fruit and vegetables and of pulses and nuts is most effective for reducing premature mortality 6 and diet-related health costs, implying a focus on behavioural change of domestic consumers. Using diet changes as an instrument to reduce food system GHG emissions, however, implies targeting ruminant producers. 6 Given the global impact of GHG emissions, an intervention targeting domestic producers is counterproductive if production shifts to a less efficient production abroad. Highlighting the environmental impact of food might also be more effective at changing consumption than highlighting dietary health concerns. 7 Lacking conclusive evidence on how to shift diets permanently and at population level, 7 we developed a policy bundle to steer towards the EL diet, building on frameworks for behavioural change 7,8 that grouped interventions by the extent to which they constrain behaviour. The contextspecific interventions started from macro-economic drivers, which are known to affect diet choices (figure 1). Diet changes will not be instantaneous, and economic circumstances will change over the implementation period. Any intervention thus needs to be evaluated against this PB: Policy bundle combining all interventions Enabling healthy choices Backdrop Nudging and information Assess impact of explicit policy instruments versus an assumed shift in behaviour Macro-economic developments Shift consumer behaviour towards healthy diet Align existing fiscal policies by removing taxes on foods to be consumed more Reducing prices Increasing prices Price incentives for producing and consuming the EAT–Lancet diet Assuming a shift in food consumption RTXS: Remove existing taxes fruit and vegetables, pulses and nuts, and fish NI: Shift consumption function towards more fruit and vegetables and less red meat Subsidise encouraged foods with GHG tax revenue Align existing fiscal policies by removing subsidies on foods to be consumed less Internalise part of climate change GHG_S: Repurpose primary (agriculture and fishery) GHG taxes to subsidise production of fruit and vegetables, pulses and nuts, and fish RSUB: Remove subsidies on red meat, poultry*, dairy, sugar GHG: Economy-wide tax on GHG emissions PB_E: Shift to food consumption attained with policy bundle Provides link to literature setting the EAT–Lancet diet exogenous EED: Exogenous shift of consumption to the EAT–Lancet diet BAU: Business-as-usual population and productivity change Price decreasing policies Price increasing policies Intervention theories on enabling choices Figure 1: Framework for deriving context-specific policy bundles for shifting diets in the MAGNET model Organisation of the interventions follows the distinction in intervention theories on enabling choices (yellow boxes) and interventions that steer choices by decreasing prices (green boxes) or increasing prices (blue boxes). Dashed boxes at the right side provide scenario codes used in the text and describe the quantification of the scenarios used in the MAGNET model. The policy bundle (PB) comprises all interventions. By running the interventions separately, we can break down the total impact of the policy bundle into the contributions of its components. Combinations of interventions used in figures and the main text are: align fiscal policies—combining RTXS and RSUB; GHG taxes—GHG; primary sector GHG taxes used to subsidise encouraged food sectors—GHG and GHG_S. Two scenarios shift food consumption exogenously to provide insight in the different impacts of reaching the same food consumption as in the policy bundle (PB) through an exogenous shift to policy bundle diet (PB_E), or provide a link to the existing literature by an exogenous shift to EAT–Lancet diet (EED). Details on the scenario implementation are provided in the appendix (pp 26–34). GHG=greenhouse gas. *Subsidies on poultry are kept in lower-income regions where consumption is below EAT–Lancet diet recommendation. Articles www.thelancet.com/planetary-health Vol 9 October 2025 3 backdrop. Even more so given the solid evidence on higher income driving diet changes captured by Engels’ law (decreasing food expenditure shares) and Bennet’s law (less staples and more fruit, vegetables and meat). We take population and sector-specific productivity drivers as given, assuming that a strong reduction in household income is not a desirable intervention to reduce meat consumption. The least invasive intervention is to convince consumers to shift their behaviour towards the EL diet. For example, tweaking food environments to make healthier choices easier (nudging), or providing information through foodbased dietary guidelines (information). The second group of interventions is changing price incentives in line with the EL diet—ie, increasing prices of discouraged foods and lowering prices of encouraged foods. This intervention builds on solid evidence that prices matter for production and consumption decisions, underpinning the price elasticities in applied economic models. The potential of price signals is also illustrated by a review of 1500 climate change policies, which showed that only price incentives had a consistent and strong impact on GHG emission reduction. 9 Our policy bundle consisted of three main blocs (figure 1): (1) nudging and information, to shift the food choices of consumers to the EL diet without changing economic incentives, a preferred option as it is least restrictive; (2) aligning existing fiscal policies with the EL diet, by removing taxes on encouraged foods and removing subsidies on discouraged foods; (3) introducing taxes and subsidies to represent unaccounted costs of GHG emissions, with receipts from primary (agricultural and fishery) GHG taxes used to subsidise primary sectors producing encouraged foods (fruit and vegetables, pulses and nuts, and fish). These subsidies set the net GHG tax on primary sectors to zero in line with their current exemption from GHG taxes while addressing concerns on the affordability of food 10 and aiming to reduce impacts on vulnerable households depending on primary production. Economic modelling Global computable general equilibrium (CGE) modelling can analyse economy-wide effects of diet shifts on consumption, production, trade, prices, and income. Global CGEs capture interactions across all sectors and regions, accounting for trade and market feedback while allowing for price-driven adjustments. Non-price mechanisms, such as consumer preference shifts, can also be incorporated. Market-clearing through price changes forms the core of CGE models, assuring that demand matches supply in all commodity and factor markets. It also creates interdependency between all production and consumption decisions illustrated in figure 2. To illustrate the interdependencies, consider a shift in food consumption through nudging and information that reduces demand for food. The prices in the food market will go down, leading to lower food prices. As food producers receive less for their output, they reduce their demand for production factors (land, labour, capital) to avoid losses and food production decreases. Reduced demand for factors by the food sector will lower the price of factors, which lowers the production cost of both food and non-food sectors. This reduced price moderates the reduction of factor demand by the food sector, while stimulating production of the non-food sector by lowering their production costs. The consumer demands less food, which it can purchase at a Food producer taxes and subsidies Consumer food price Consumer non-food price Food consumer taxes and subsidies Non-food producer taxes and subsidies Factor use taxes Factor market price changes such that factor demand=supply Non-food consumer taxes and subsidies Non-food market price changes such that nonfood supply=demand Food production decisions Non-food production decisions Food and non-food consumption decisions Food market price changes such that food supply=demand Food supply Producer food price Producer non-food price Producer factor prices Food demand Factor demand Non-food supply Household income from factor (land, labour, capital) supply Price adjustments Nudging and information change food consumption decisions Price incentive scenarios affecting tax and subsidies Non-food demand Figure 2: Schematic diagram of key feedback loops in the economic model Highly simplified representation of key interactions between consumption and production of different commodities in the modelling of the scenarios. Dashed lines indicate monetary flows (prices and income), solid lines indicate flows of commodities (goods, services and production factors). Feedback loops via international trade and intermediate input use of commodities (eg, use of non-food commodities in the production of food commodities) are omitted. Market equilibrium (supply=demand) is achieved by price adjustments (marked in yellow). Nudging and information change food consumption decisions (marked in green). Non-food consumption also changes as consumers decide on food and non-food consumption simultaneously, taking commodity price and income changes into account. Changed food and non-food demand modify prices and production patterns, which in turn modify demand for factors used in production and thus factor prices. Changing factor prices change the household income, which is derived from supplying production factors, closing the feedback loop to the consumer. Price incentive scenarios affect the tax and subsidies (marked in blue) that create a price wedge between market and consumer or producer prices used in decision making, affecting all food and non-food production and consumption decision directly or indirectly. Articles 4 www.thelancet.com/planetary-health Vol 9 October 2025 lower price, leaving more income to spend on non-food. This will increase demand for non-food, which may include the non-food use of the commodity (eg, as feed or fuel), moderating the reduced demand for food by consumers. The lower demand for factors, however, also lowers household income, which is based on factor payments by sectors. This decrease in income moderates the increased demand for non-food and thus the scope for the non-food sector to expand. These interrelated adjustments in the markets for food, non-food, and factors will continue until all markets are in equilibrium again with supply being equal to demand. Reducing food demand through increased consumption taxes will have a different effect. It could potentially reduce demand for non-food if the income share of food increases after the tax, setting in motion a different series of feedback loops in the global economic system. We used MAGNET, 11 a global CGE model extended to better reflect land use and bio-economy dynamics. Three key extensions addressed the EL diet’s context-specific effects, affordability, and environmental impact. The diet extension tracked food content changes in physical units, enabling regional variations in diet shifts. It relied on an endogenous variable tracing changes in the primary content of processed foods from changes in production and global trade flows. 3 Affordability was assessed by calculating the minimum cost of the EL diet relative to household food expenditures and income, and as a share of daily wages. These wages varied by five types of workers and 77 sectors providing an endogenous distribution of labour income. Market segmentation captured the persistent gap between agricultural and non-agricultural wages. 12 MAGNET was extended to capture feedback on non-food bio-based sectors, 13 which might reduce GHG emissions in the overall economy but offset environmental gains in the agricultural sector by increasing its GHG emissions and land use. Lower agricultural prices from reduced food demand can increase land use for biomass production used by non-food sectors. Similarly, reduced food spending can shift consumption to GHG-intensive non-food, partly mitigating emissions reductions. 3 The appendix (pp 10–25) shows a description of the model and data sources used. Scenario quantification The scenarios were quantified combining the MAGNET database and external data (appendix pp 26–34). The BAU scenario was a combination of population and productivity shocks. Household consumption and the country-specific EL diet recommendations by food group were based on intake data 2 and excluded food loss and waste. In the exogenous shift to EL diet (EED) scenario, EL recommendations were used to fix consumption at food group level, allowing households’ freedom to select the most preferred combination of food items in each group. The extent to which consumption could be shifted towards the EL diet using nudging and information (NI) was derived from literature. 7 Evidence was limited in terms of products (red meat, vegetables, and fruits), regional context (Northern America and Europe), population subgroup (students), interventions (mostly information), and only assessed short run impact. While we used the lowest estimates and adjust the shifts proportional to differences in intake (appendix p 27) the estimates remain optimistic for our global application to the general population. Daily red meat consumption reductions range from a 2⋅1 to 11⋅8 g across regions. Increases in fruit, vegetables, pulses and nuts range from 15⋅1 to 45⋅0 g. Alignment of fiscal policy (AP scenario) removed existing taxes on foods to be consumed more and removed existing subsidies on foods to be consumed less, as captured in the GTAP database. 14 The GHG scenario imposed a carbon price of US$100/ton CO 2 equivalent (CO 2 e) on emissions of CO 2 , CH 4 , and N 2 O, based on a previous climate-focused study using an ensemble of global economic models, including MAGNET. 15 From a polluter pays principle, agricultural sectors also need to pay for GHG emissions, being responsible for about a quarter of GHG emissions. 16 Denmark recently announced plans to start pricing GHG emissions from livestock 17 with $40/ton CO 2 e in 2030 and rising to around $100/ton CO 2 e in 2035. The latter corresponds with the imposed GHG price in this study. Primary sector GHG tax revenues were redistributed to encouraged sectors (fruit and vegetables, pulses and nuts, fish) to keep net GHG taxes on the primary sector at zero (GHG_S). The policy bundle scenario (PB) combines NI, alignment of fiscal policies, and GHG taxes with redistribution to encouraged sectors (GHG_S). We also ran a scenario in which we set the diet from the policy bundle exogenous (PB_E). The extent of the diet shift was the same as under the policy bundle scenario (PB) but achieved through different means. It shed light on the impact of how the diets were shifted, which could not be as clearly seen from a comparison to the exogenous shift the EL diet (EED) scenario, which differed in both the level of the shift and means through which it was achieved. We conducted sensitivity analyses on the policy bundle scenario for price elasticities and GHG prices, including a scenario in which GHG prices varied by regional income (appendix pp 35–41). Role of funding source The funders of this study had no role in the model set-up and scenario design, nor in the data collection, data analysis, model result interpretation, or writing of the paper. Results By 2050, the PB improved consumption across food groups and regions compared with BAU developments, but substantial gaps with the EL recommendations remained (figure 3A). Fruit and vegetable intake remains at 80% of the recommended 500 g/capita per day and pulses and nuts at 50% of 75 g/capita per day. Global average consumption of fruit and vegetables declined substantially See Online for appendix Articles www.thelancet.com/planetary-health Vol 9 October 2025 5 (–19 g/capita per day) in the BAU scenario (figure 3B) despite increasing in all but the upper middle-income regions and most so in low-income regions (15 g/capita per day; appendix pp 1–2). Figure 3 presents populationweighted averages. As population growth was highest in low-income regions, their low consumption of fruit and vegetables got a higher weight in 2050, reducing the BAU global average intake. Despite considerable progress, there was a larger number of people substantially underconsuming fruit and vegetables, making it a more pressing concern than in 2025. The PB substantially increased consumption of fruit and vegetables in low-income regions (34 g/capita per day), but the gap with EL diet recommendations remained the largest of all regions. The PB had the largest impact on consumption of red meat (pork, beef, and lamb), but consumption remained three times the EL recommendations at global level (figure 3A). NI reduced pork consumption most, while GHG taxes were most effective in reducing the more emission-intensive consumption of beef and lamb (figure 3B). The PB increased the gap with the EL diet recommendation only for dairy in low-income regions, reducing BAU consumption from 88 to 82 g/capita per day (data not shown). Decomposition of the contributions of the interventions showed that GHG taxes and aligning fiscal policies were the main drivers of reduced dairy consumption (figure 3B). With more emission-intensive production, the impact of GHG taxation in low-income countries (–6 g/capita per day; appendix p 2) was stronger than the global average (–4 g/capita per day, figure 3B). No single intervention shifted all diet components in the desired direction, and effectiveness varied by region. Taxes on agricultural emissions dominated the impact of GHG taxes on food intake, driving 98% or more of the diet shift induced by GHG taxes across regions and food groups. While at global level aligning fiscal policies had minor impact (figure 3B), it increased fish consumption in the high-income region while reducing intake for all other food groups (appendix p 1). This result was mostly driven by removal of agricultural taxes and subsidies in the EU. Although global average impact results suggested that nudging was the most effective intervention, this interpretation did not hold in all regions (appendix pp 1–2). In low-income regions, where consumers are more pricesensitive, GHG taxes and subsidies had the strongest effect across most food groups. Increasing price elasticities and varying the GHG prices in the sensitivity analyses led to minor changes in diets (appendix pp 34–35). The PB did not restrict total calorie intake, which increased slightly in lower-income regions (18 kcal/ day) compared with the BAU (appendix p 3). The EED scenario fixed total calories, addressing concerns for hunger in lowincome (199 kcal/day) and lower-middle-income regions (69 kcal/day), while targeting overweight and obesity in upper-middle-income (–101 kcal/day) and high-income regions (–118 kcal/day). Compared with the BAU, consumption shifted towards more processed foods (2–57% increase) and out-of-home consumption (30–74% increase) with the EED. The PB induced only minor changes in shares of processed foods and out-of-home consumption (appendix p 3). By explicitly targeting GHG, the PB reduced emissions most (–16 GTon CO 2 e). Despite not reaching the EL diet recommendation (figure 3A), it reduced primary sector GHG emissions (–4 Gton CO 2 e) more than the EED did (–3 Gton CO 2 e; figure 4A). GHG taxes in the PB increased costs of fertiliser and other fossil-based inputs. More extensive production and shifts to biobased substitutes for fossils increased total land area (27 million ha; figure 4B) Red meat Sugar Poultry Dairy Fruit and vegetables Pulses and nuts Fish 7·2 2·6 2·3 1·8 0·8 0·3 1·1 6·0 2·6 2·2 1·7 0·9 0·3 1·3 BAU HIC Ratio of per capita intake to EL diet recommendations by scenario (2050) Contribution of policy bundle components to change in food intake between 2025 and 2050 A B UMIC LMIC LIC World PB 5·5 1·9 2·0 0·8 1·1 0·3 1·3 0·8 1·2 0·3 1·3 4·7 1·9 2·0 BAU PB 1·2 1·7 0·6 0·5 0·5 0·5 1·0 1·0 1·7 0·6 0·5 0·6 0·6 1·1 BAU PB 2·1 1·2 0·2 0·4 0·3 0·6 0·3 1·6 1·3 0·2 0·3 0·4 0·7 0·3 BAU PB 3·5 1·8 1·2 0·7 0·7 0·4 1·0 BAU PB 2·9 1·9 1·2 0·7 0·8 0·5 1·1 Discouraged Encouraged Context-specific Beef and lamb Pork Sugar Poultry Dairy Fruit and vegetables Pulses and nuts Fish –20 –10 0 10 20 30 40 50–30 g per capita per day BAU Nudging and information (NI) Align fiscal policies (AP) GHG taxes (GHG) Primary sector GHG taxes subsidise encouraged food sectors (GHG_S) Effect of interactions between interventions Figure 3: Diet achieved with a decomposition of contribution by scenario component of global change from 2025 to 2050 (A) Ratio of per capita intake to global EL diet recommendations by scenario. Discouraged foods are marked in red and encouraged foods in green. A ratio greater than 1 implies that intake is above the EL recommendation, and a ratio lower than 1 implies that intake is below the EL recommendation. Global EL recommendations (g/capita per day): red meat 14, sugar 31, poultry 42, dairy 250, fruit and vegetables 500, pulses and nuts 75, fish 28. Model regions are grouped by World Bank income classification. All regional aggregates are population weighted averages to account for differences in country size. (B) Contribution of policy bundle components to global change in food intake between 2025 and 2050. The BAU component in panel B includes the impact of changes in population which changes regional weights from 2025 to 2050. The effect of interactions between interventions is computed as the difference between the policy bundle minus the sum over the impacts of each intervention run separately. Impact of interventions by region provided in the appendix (pp 1–2). Source: MAGNET simulations. BAU=business-as-usual. EL=EAT–Lancet. GHG=greenhouse gas. HIC=high-income countries. LMIC=lower middle-income countries. LIC=low-income countries. PB=policy bundle. UMIC=upper-middle-income countries. Articles 6 www.thelancet.com/planetary-health Vol 9 October 2025 which might increase pressure on biodiversity. 18 The use of taxes in the PB increased consumer prices, while exogenously shifting consumption decisions (EED and PB_E) did not limit non-food consumption. As a result, the reduced demand for land by restricting food consumption with the EED (125 million ha) was substantially reduced by increased use of land for non-food (90 million ha) benefitting from lower land prices, and a net contraction (–35 million ha) remained (figure 4B). Sensitivity analyses showed that a lower GHG price led to a lower reduction in GHG emissions, whereby the impact of land use was also lower, and vice versa with a higher GHG price. Varying GHG taxes by income reduced taxes for emission-intensive production in low-income regions and reduced the agricultural land expansion (appendix pp 36–38). Per-capita GHG emissions remained highly unequal (appendix p 4). Global change in greenhouse gas emissions compared to business as usual (2050, GTon CO2 equivalent) Global change in agricultural land area compared to business as usual (2050, mil. ha) Exogenous shift to EAT– Lancet diet (EED) Total : GTon CO2 equivalent Policy bundle (PB) Exogenous shift to policy bundle diet (PB_E) Nudging and information (NI) Align fiscal policies (AP) GHG taxes (GHG) Primary sector GHG taxes subsidise encouraged food sectors (GHG_S) –4 –16 –1 –1 0 –16 –16 Total : (mil. ha) –35 27 –20 –4 –14 –2 33 Exogenous shift to EAT– Lancet diet (EED) Policy bundle (PB) Exogenous shift to policy bundle diet (PB_E) Nudging and information (NI) Align fiscal policies (AP) GHG taxes (GHG) Primary sector GHG taxes subsidise encouraged food sectors (GHG_S) –18 –16 –14 –12 –10 –8 –6 –4 –2 0 2 GTon CO2 equivalent –150 –100 –50 0 50 100 Million. ha. Energy and biobased Pasture and feed Food crops Other use Services Manufacturing and mining Primary sectors Final demand A B Figure 4: GHG emissions and agricultural land use compared to the business-as-usual scenario (2050) (A) Global change in greenhouse gas emissions compared to business as usual. (B) Global change in agricultural land area compared to business as usual. Land use by activities is assigned to categories of use based on material flow balances tracing the use of the produced output. For example, land use for crops is assigned to feed based on the share of livestock sectors in total demand for crops. Source: MAGNET simulations. GHG=greenhouse gas. Articles www.thelancet.com/planetary-health Vol 9 October 2025 7 Emissions in low-income regions were 20% of those in high-income regions in the BAU scenario, further decreasing with the PB (18%) and the EED (15%) scenarios (appendix pp 4,38). The uniform tax on emissions in the PB scenario had substantial justice implications as it increased the costs of less efficient producers disproportionately located in lower income regions, which historically have less access to technology. Compared with the BAU, the PB scenario increased the reliance of low-income regions on imports of animal sourced foods (493% red meat, 33% dairy, and 3% poultry; appendix p 5). The sensitivity analyses showed that only varying the GHG taxes by income level reduced inequality in per-capita GHG emissions, so that in low-income regions they were 25% of those in highincome regions (appendix p 38), against 20% in the BAU (appendix p 4). Varying GHG taxes per capita GDP shifted the burden of GHG reduction to high-income regions with a smaller reduction in global GHG emissions and less agricultural land expansion. Although the total number of workers unable to afford the EL diet decreased in lower income regions, it increased in the upper middle-income regions. It also made the EL diet less affordable with costs exceeding household food expenditures in low-income regions (appendix pp 39–40). The EED scenario also had justice implications, as it generated reductions of 6–51% in non-primary emissions in the lower income regions (appendix p 4), reflecting a contraction of these non-targeted sectors. To address hunger and malnutrition, EED increased average consumption in lower income regions (appendix p 3), irrespective of the costs of the diet. As households are bound by their budget constraint (figure 2), they need to reduce non-food demand to accommodate the cost of the exogenously assumed diet. The EED scenario also increased dependency on imports for fruit and vegetables, pulses and nuts, and fish by up to 1747% of BAU levels (appendix p 5). Affordability of the EL diet for households depends on the cost of the EL diet and household income. Moving from high-income to low-income regions, the cost of the EL diet increased (figure 5). EL diet costs (in 2017 PPP $/capita per day) decreased in the BAU from 2025 to 2050 in high-income (2⋅9–2⋅5) and upper middle-income regions (3⋅2–2⋅8), while costs increased in lower middle-income (4⋅4–4⋅5) and low-income regions (4⋅8–5⋅2), where high population and income growth outpace productivity improvements. Despite a doubling of average household income in the BAU, the minimum cost of the EL diet exceeded average household food expenditures by 0⋅5 in low-income countries. The EED led to a peak in EL diet costs in lower income regions, driven by nearly doubling cost shares of fruit & vegetables and pulses & nuts (appendix p 6). In contrast, subsidies in the PB scenario decreased EL diet costs below 2025 reference levels. The strongest impact was in low-income regions, as emission-intensive primary production generated large GHG tax revenues redistributed as subsidies. Ruminant livestock were most affected. Contraction of emission-intensive food and non-food sectors releases production factors (land, labour, capital) and commodities used as intermediate inputs, HIC UMIC LMIC LIC NF FE ELM 58·4 26·0 2·9 82·3 35·0 2·5 86·2 30·4 2·8 80·4 36·0 2·3 81·0 36·1 2·6 18·2 6·1 3·2 34·3 10·5 2·8 33·7 11·0 3·0 33·0 10·6 2·8 34·1 10·8 2·8 9·4 1·2 4·4 24·7 5·3 4·5 20·1 8·1 6·3 23·6 5·5 4·4 24·4 5·4 4·7 2·5 –1·8 4·8 7·4 –0·5 5·2 3·3 2·9 7·1 6·9 0·2 4·7 7·3 –0·6 5·4 –10·0 10·0 30·0 Reference BAU EED PB PB_E Reference BAU EED PB PB_E Reference BAU EED PB PB_E Reference BAU EED PB PB_E 50·0 70·0 90·0 110·0 130·0 2017 PPP $/capita per day NF FE ELM Figure 5: Average household expenditure on food and non-food and minimum cost of the EL diet by scenario and region (2017 PPP US$/capita per day, 2050) Sum of the columns is the expenditure-based average household income. A negative value for FE indicates that the minimum cost of the EL diet exceeds the household food expenditures. Household food expenditures with an exogenous shift to the EL diet (EED scenario) are higher than the minimum cost of the EL diet as households’ food and non-food preferences are also taken into account in the MAGNET model when exogenously shifting to the EL diet. Model regions are grouped by World Bank income classification. Source: MAGNET simulations. BAU=business as usual. EED=exogenous shift to EL diet. EL=EAT–Lancet. ELM=minimum cost of the EL diet. FE=household food expenditure minus cost of the EL diet. HIC=high-income countries. LIC=low-income countries. LMIC=lower middle-income countries. NF=average household expenditure on non-food. PB=policy bundle. PB_E=exogenous shift to policy bundle diet. PPP=purchasing power parity. Reference=2025 reference year. UMIC=upper middle-income countries. Articles 8 www.thelancet.com/planetary-health Vol 9 October 2025 lowering the production costs for other sectors in lowincome regions. As a result, even without redistributing tax receipts as subsidies, GHG taxes lowered costs of fruit and vegetables (–5%), pulses and nuts (–5%), fish (–10%), and poultry (–3%) in low-income regions, resulting in the same EL diet costs as in the BAU (appendix p 7). Globally the main driver of increased costs with EED were production limitations increasing average costs of fruit and vegetables (67%) and pulses and nuts (163%), with stronger impacts in low-income regions (148% and 329%). Affordability decreased when using nudging and information to shift diets: globally costs of the EL diet increased more (2⋅1%) when shifting to the policy bundle diet (PB_E) and 26⋅1% with EED than the changes in household income (0% and 0⋅2%; appendix p 6). These global averages hid stronger impacts in low-income regions, where EL diet costs increased by 3⋅6% (PB_E) and 34⋅7% (EED), with household income remaining close to BAU (PB_E) or increasing by 10⋅1% (EED). Globally, affordability increased with the PB as subsidies lowered the cost of the EL diet (–4⋅4%) more than it reduced household income (–2⋅0%). Again, impacts in low-income regions were stronger (–11⋅1% decrease in cost vs –2⋅8% decrease in income). Sensitivity analyses showed minimal differences in EL diet affordability when increasing price elasticities or varying GHG prices (appendix pp 39–40). Average total household income exceeded the minimum cost of the EL diet in all regions and scenarios (figure 5) but missed differences across households. Household heterogeneity could be partly captured in MAGNET by wages of workers. The cost of the EL diet exceeded daily wages for part of the workforce in all regions and scenarios (figure 6A). Affordability of the EL diet for workers was most of a concern in EED, and least when imposing GHG taxes with a redistribution of primary GHG tax revenues. In the remainder, we defined the EL diet as unaffordable if the cost was 50% or more of a daily wage, assuming that about 50% of income is needed for non-food. 19,20 In the BAU scenario, 24⋅3% of the workforce (920 million) could not afford the EL diet, rising to 24⋅8% (940 million) with PB and 27⋅7% with EED (1048 million; appendix p 8). Geographically, these low-earning workers were mainly located in the two lower-income regions, and in terms of sectors, they were in primary production (figure 6B). Compared with the BAU scenario, the number of workers who could not afford the EL diet increased in encouraged sectors (132⋅3% with EED, 28⋅7% with PB) while decreasing in discouraged sectors (–37⋅7% with EED, –13⋅3% with PB; appendix p 8). Despite higher wages pulling more workers in encouraged food sectors, the number of workers unable to afford the EL diet thus increased, even when subsidies reduce the cost of the EL diet in the PB. The negative impact of EED on nonfood sectors was visible in an increase in workers in manufacturing and services (13⋅8%) who could not afford the EL diet, while this number slightly decreased with PB (–0⋅1%; appendix p 8). A shift to a more plant-based diet increases the demand for agricultural labour as crops are more labour-intensive than livestock. Although mostly accommodated by a shift from discouraged to encouraged primary sectors, the number of workers in primary sectors increased compared with the BAU scenario (8% with EED, 1⋅4% with PB; appendix p 9). This increase in workers in primary sectors drove an increase in the total number of workers unable to afford the EL diet compared with the BAU scenario (13⋅9% with EED, 2⋅2% with PB; appendix p 8). 0 10 20 30 40 50 60 70 80 90 100 110 120 EL diet cost as share of daily wage per worker (%) 0·45 0·65 0·85 1·05 1·25 1·45 1·65 1·85 2·05 2·250·25 Cumulative number of workers (billion) Location of workers where EL diet cost exceeds 50% of the daily wage with the policy bundle (million workers) Share of EL diet cost in daily wage by scenario (2050) Business as usual (BAU) Exogenous shift to EL diet (EED) Policy bundle (PB) Exogenous shift to policy bundle diet (PB_E) Primary sector GHG taxes subsidise encouraged food sectors (GHG_S) GHG taxes (GHG) A 30 515 Region 50% of wage Macro sectors Primary sectors 394 Total: 940 million Total: 940 million Total: 585 million 2845 940 138 217 585 59 60 164 302 Primary production Services Manufacturing and mining EL diet cost >50% daily wage EL diet cost <50% daily wage Low-income Lower middle-income Upper middle-incomeNon-food primary sector Other food sectors Discouraged food sectors Encouraged food sectors B 0 20 40 60 80 100 Proportion of workers (%) Figure 6: Workers unable to afford the EL diet and their location (regions and sectors; 2050) Not visible in the bar chart are 0⋅3 million workers in high-income regions where the EL diet costs more than 50% of a daily wage with the policy bundle. Affordability by income group is provided in the appendix p 8. Source: MAGNET simulations. EL=EAT–Lancet. GHG=greenhouse gas. Articles www.thelancet.com/planetary-health Vol 9 October 2025 9