The Ukraine war and rising commodity prices: Implications for developing countries (2024)

• •

  The Ukraine War severely disrupted global markets for food, fuel, and fertilizer.

• •

  The macroeconomic effects of rising prices depend on countries' unique economic structures.

• •

  Socioeconomic impacts reflect countries' and individuals' consumption and employment patterns.

• •

  Household consumption and poverty are most affected by higher fuel and food prices.

• •

  Deteriorating undernourishment and diet quality are driven more by rising food prices.

2.1. Economywide models and data

Economywide models are used to assess the impacts of global price shocks on production, workers, and populations in 19 developing countries. More specifically, we use the International Food Policy Research Institute's Rural Investment and Policy Analysis (RIAPA) modeling system. At the core of RIAPA are Computable General Equilibrium (CGE) models, which capture all producers (sectors) and consumers (households) in an economy and track how they interact with each other in markets for products and factors (i.e., land, labor, and capital) to determine prices, incomes, and expenditures (see Diao and Thurlow 2012). RIAPA's main database is called a Social Accounting Matrix (SAM), which combines and reconciles data from a range of farm, firm, and household surveys, as well as data on domestic production, international trade, and government revenues and expenditures.

RIAPA captures many of the country-specific factors driving the impacts of the global crisis. Each economy is separated into detailed subsectors. Half of these are in agriculture, agro-processing, and food services, allowing us to track food supply chains and price shocks from farmers to consumers, and vice-versa. The models also separate fertilizers and fuels and track their use by consumers and as production inputs (e.g., fertilizers for farming or fuels for transport). Information on the flow of goods and services between sectors and to final users is drawn from supply-and-use tables embodied within the SAMs. Producers in the models maximize profits by (i) raising prices, subject to demand constraints; (ii) favoring cheaper inputs, including labor, land, and capital; and (iii) deciding whether to supply domestic or export markets.

RIAPA also separates countries into population groups using national household surveys. Workers are divided across rural and urban areas and by education levels, and households are grouped based on their rural-urban location, farm-nonfarm status, and per capita consumption quintile. This allows RIAPA to track which workers are most affected by the global crisis and which households are most affected by changes in wages, employment, and revenues from farm and nonfarm enterprises. Households in the models maximize welfare (consumption) by demanding relatively cheaper products, including switching between imports and domestic products.

The model and country data determine price transmission from global to local markets. The share of imports (exports) in total supply (demand) is a key factor. For example, countries that import fuels will experience larger domestic price increases when world prices rise than will countries that produce their own fuels. However, countries that export fuels may also experience rising domestic prices, as more domestic output is redirected to export markets. Price transmission also depends on the ability of producers and consumers to switch between domestic and foreign products/markets. This is represented in the models by product-specific trade substitution elasticities, which are drawn from the Global Trade Analysis Project (Dimaranan 2006).² Elasticities are usually higher when domestic and foreign are more hom*ogenous, such as wheat and petroleum. Consumer responses to changes in real incomes are also based on elasticities, which are estimated using national household surveys or drawn from the literature.

Finally, we make assumptions in the models to better reflect the short-run constraints that countries face in responding to global price shocks. First, we assume that central banks are not positioned to accommodate the price shocks through changes in foreign currency reserves. Second, farmers cannot reallocate crop land, but, instead, receive windfall revenues from higher prices based on pre-crisis land allocations. Similarly, producers of fertilizers and fuels cannot increase output, but do receive windfall gains. Third, we allow underemployment amongst less-educated workers to increase when economies contract, which is consistent with labor-markets being demand-driven in the short-run. Finally, we focus on the impacts of the global shocks without accounting for actual or potential policy responses to mitigate adverse impacts. These assumptions capture the rapid onset of the global crisis, reflect structural constraints, and are consistent with our focus on measuring short-run impacts at the peak of the global price increases.

In summary, our models are calibrated to detailed databases, built using official national accounts and nationally representative household surveys. The models provide a consistent representation of each country's unique production, trade, and consumption patterns, as well as the generation and distribution of incomes. The models use behavioral functions, grounded in economic theory, to represent producer and consumer behavior while respecting all macroeconomic identities. These functions are calibrated using elasticities estimated from household surveys or global databases. These models have been used previously to estimate the welfare implications of rising world food and fuel prices (e.g., Arndt et al., 2008); to evaluate the impacts of farmers' fertilizer use (Arndt et al., 2016); and to decompose historical drivers of poverty (Arndt et al., 2012). The models are therefore well-suited to evaluating the impacts of the recent food, fuel, and fertilizer crisis.

2.2. Baseline food and fuel supply patterns

The impact of rising world commodity prices in the models depends on countries' unique production, trade, and consumption patterns. Although maize, wheat, and edible oils are consumed in all the countries included in our analysis, the reliance on imports varies by crop and by country. African countries are broadly self-sufficient in maize, the continent's major staple food (Van Ittersum et al., 2016). However, most of the 14 African countries included in our analysis rely on imported wheat grains since this crop is not widely grown. Households in Africa also consume relatively few foods derived from wheat. Exceptions include Egypt, where bread is the most important staple, and Ethiopia, where wheat makes up a larger share than maize in the food basket and is produced at scale domestically.

World rice prices declined slightly between mid-2021 and April 2022, which will have benefited consumers in the five Asian countries included in our analysis, since rice is the major staple food (Adjao and Staatz 2015). Most of these Asian countries are also large producers of rice and are not overly reliant on imports. Rice is also more widely consumed than wheat in many African countries, including the Democratic Republic of the Congo (DRC), Ghana, and Senegal. Although rice is an important crop in Ghana and Senegal, most domestic supply still comes from imports. Falling world prices are therefore more likely to benefit consumers in these two countries.

In most countries, the share of edible oil imports – mostly palm oil – in domestic supply is smaller than wheat imports but larger than maize imports. On the other hand, imported edible oil products are close substitutes for domestically produced and consumed edible oils. Many countries grow various oilseeds and some export oilseeds, including Ethiopia, Tanzania, and Uganda. Therefore, the net effect of global food price movements is not immediately evident for many countries.

Turning to fuels, almost all oil products (i.e., crude oil and processed petroleum) are imported in the studied countries. Exceptions include Nigeria, which is one of Africa's largest oil exporters, as well as Egypt and Ghana, both of which export natural gas, crude oil, or refined petroleum. Other countries, such as DRC, Niger, and Uganda, produce crude oil at a small scale, but still mainly rely on imports. While crude oil producers in exporting countries will benefit from rising global oil prices, the impact of higher oil and petroleum prices on other sectors of their economies is ambiguous. Moreover, the impact of higher oil prices on households is harder to assess as direct consumption of oil products is limited. Instead, oil products are primarily used as an intermediate input into the production of other goods and services. For example, input use accounts for around 70–90 percent of total demand for oil products, with significant demand from the transport sector. Fuel price increases, therefore, indirectly affect the prices of all marketed goods and services in the economy.

The RIAPA model tracks the flow of domestic and imported inputs between sectors and estimates the net effect on final product prices. Impacts on households also depend on the importance of the affected commodities in their consumption baskets. Shares of cereals and edible oils in total food expenditure vary significantly across countries (i.e., 15–35 percent). Root crops are another important staple food in most countries, allowing consumers to switch to these nontraded foods when cereals prices rise. The RIAPA models measure incomes and food and nonfood expenditures for different population groups, and link to survey-based microsimulation models to track changes in consumption patterns of individual households. Unpacking populations is crucial since household spending patterns vary across population subgroups. Cereals and edible oils, for example, are often important in the consumption baskets of poorer or rural households, and so these households are expected to be adversely affected by higher prices for these products.

2.3. Baseline fertilizer use and farm productivity

Unless existing fertilizer subsidy programs are designed to cushion the effect of price shocks, rising fertilizer prices will likely cause some farmers to reduce their use of this input, leading to lower agricultural production and higher food prices. The magnitude of this decline depends on: (i) the amount of fertilizer currently used to grow crops (i.e., fertilizer adoption and application rates); (ii) the responsiveness of fertilizer demand to changes in prices (i.e., the price elasticity of demand for fertilizer); and (iii) the expected productivity losses for farmers who lower their fertilizer application rates. Our analysis captures both the more immediate impact of higher fertilizer prices on farm production costs, and the medium-term effects of higher prices on fertilizer use, and hence farm productivity.

Information on fertilizer use by crop was calculated using national farm or household survey data for most countries, while, for the few countries that lacked such data, we relied on global data (Ludemann et al., 2022) and local expert assessments. For Kenya and Nigeria, we used the International Fertilizer Development Center's (IFDC) Fertilizer Use by Crop (FUBC) estimates (see Africa Fertilizer, 2022). Together, these data confirm that fertilizer adoption rates vary widely across countries.³ Farmers in Asian countries, such as Bangladesh and Cambodia, tend to have higher adoption rates than their African counterparts. Adoption rates also vary among African countries, with relatively high adoption in Ethiopia and very low adoption in Uganda. Finally, adoption rates vary across crops within a country. In the countries included in our analysis, adoption rates tend to be higher for major grain crops and export crops than for root crops, reflecting either their higher commercial value or the effects of fertilizer subsidy programs, which typically target key staples and export crops. In Ethiopia, for example, fertilizer adoption rates for maize, wheat, and teff are around 90 percent, whereas the adoption rate for sorghum and millet is only around 30 percent. Variations in fertilizer use across countries and crops imply that the impacts of higher fertilizer prices will also vary across countries.

In addition to the direct impact of higher fertilizer prices on production costs, we model an additional impact channel where changes in fertilizer quantities affect productivity (or crop yields) directly. Since the precise quantities of fertilizer applied to crops and yield responses to changes in fertilization rates are difficult to estimate, especially for smaller crops or on plots that are intercropped, we adopt a conservative set of assumptions to estimate potential yield losses associated with the rise in fertilizer prices. First, we assume a price elasticity of farmers’ demand for fertilizer of −0.15, implying that a doubling of fertilizer prices leads to a 15 percent decline in fertilizer use. Despite a dearth of evidence on fertilizer own-price elasticities in developing countries, our assumed elasticity falls within the range estimated by Li et al. (2011) for South and East Asia, including wheat (−0.17), rice (−0.16), and maize (−0.11). Second, we assume crop yields on plots that are fertilized are 20 percent higher than on unfertilized plots. Our assumption is broadly consistent with the literature, which generally finds higher yields for farmers using fertilizer (see Hemming et al., 2018, in which authors found fertilizer and seed subsidies are associated with greater use of inputs, higher agricultural yields, and increased farmer incomes). Given crop-specific estimates of the share of land that is fertilized, we can estimate the change in fertilizer use and yields on fertilized plots as well as the average productivity change in each agricultural subsector under the assumption that yields on unfertilized plots are unaffected.

A key variable in this equation is the fertilizer price change. While in the initial simulation shock the international (real) price for fertilizer is doubled for all countries (Fig. 1), the change in the domestic price of fertilizer differs by country depending on whether fertilizer is also produced domestically. Domestically produced fertilizer accounts for a small amount of fertilizer supply in most countries. Zambia, for instance, relies entirely on imported fertilizer, and hence the domestic price of fertilizer also doubles. In Egypt, on the other hand, the country exports one-third of domestically produced fertilizer and imports about six percent of high-quality fertilizer supply, resulting in domestic fertilizer prices rising by only 27 percent, on average.

2.4. Model simulations

Several scenarios are designed to simulate the effects of higher world prices (see Fig. 1) and productivity losses from reduced fertilizer use (as discussed above). These are: (i) food price shocks: rising import and export prices for maize, wheat, and edible oils, and declining import and export prices for rice; (ii) fuel price shocks: rising import and export prices for crude oil and oil products; and (iii) fertilizer price shocks: rising fertilizer import prices combined with lower crop yields due to the resulting reduction in fertilizer use. The combined effect of all three impact channels is also simulated. Simulation results apply to the 2022 growing season, and so the results should be interpreted as reflecting the period after the initial direct and indirect (spillover) effects across sectors and households have occurred, but before governments introduce any mitigative policies, or the private sector adjusts its investments in response to the crisis. Impacts on economies and food systems are discussed first, before turning to impacts on poverty and food security.

4.1. Consumption spending and poverty

Real consumption spending, which includes the value of home consumption, falls in all 19 countries. The percentage decline in consumption is larger than that of GDP. In the modeling approach, households, rather than governments and investors, are viewed as likely to be more directly impacted. This approach is taken because many households are adversely affected by both rising prices and falling incomes. Food expenditure also accounts for a much larger share of household consumption than food production does in national GDP. As a result, rising food prices are relatively more important in explaining declines in consumption than they are in explaining GDP losses. Ultimately, while the global crisis does not substantially reduce national incomes (i.e., GDP), it does have an uneven impact on the macroeconomic distribution of national income, with consumption taking the brunt of the impact.

Falling household consumption leads to an increase in poverty. Our poverty assessment uses survey-based microsimulation models linked sequentially to economywide models (see Arndt et al., 2012). Each country model contains 15 representative household groups that separate the total population in rural farm, rural nonfarm, and urban households, each across expenditure quintiles. Every survey household is mapped to one of the models’ representative households. Estimated changes in consumption spending are transferred from the household in the model down to the household in the survey, where its poverty status is recomputed using the US$1.90-a-day international poverty line (Ferreira et al., 2015). Panel A in Fig. 3 indicates that the global price shocks increase national poverty headcount rates in all 19 countries. There is wide variation across countries, with poverty rates rising by as much as 7.6 percentage points in Myanmar, but by less than one percentage point in Ghana, Niger, Nigeria, Uganda, and Zambia.

The rise in national poverty equates to an additional 27.2 million people falling below the international poverty line across the 19 countries (see Table 1), with 72 percent of these people living in rural areas (see Panel A in Fig. 4). This reflects higher rural population shares and higher initial rural poverty rates. The increase in world fuel and fertilizer prices are the largest drivers of this increase in poverty, which is consistent with Fig. 2, which showed these impact channels driving the decline in national GDP or incomes. There is some variation across rural and urban areas, with rural poverty driven more by the fertilizer shocks, which reduces farm incomes, and urban poverty driven more by fuel shocks, which disproportionately affect the costs of commodities and services that urban households consume more intensively.

Particularly after some time to adjust, smallholder farmers, who make up much of the rural population in Africa and Asia, may benefit from higher food prices, since this has the potential to raise farm incomes (see Headey and Martin 2016) and stimulate production with spillover implications for rural wages. In the analysis here, the net effect is typically negative for rural households. The models take into account the limited scope for adjustment of the farm sector in the time frame considered, nonfarm income sources of rural households, and the larger share of food in rural consumption baskets than farm incomes in total income.

4.2. Undernourishment and diet deprivation

Undernourishment also becomes more pervasive. A person is deemed undernourished when they consume fewer calories than what is required for a healthy life (FAO 2022). Calorie availability is assessed at household-level and requirements are adjusted for the age and gender composition of the household. With falling household consumption and rising poverty, the prevalence of undernourishment rises in all 19 countries, with increases in the range of 0.3–4.4 percentage points (see Panel B in Fig. 3). In total, 22.3 million more people become undernourished due to the global price crisis (see Table 1). As with poverty, most people who become undernourished (66 percent) live in rural areas (see Panel B in Fig. 4). Rising food prices are the most important driver of rising undernourishment – far more so than they were in explaining the increase in poverty. This is because rising prices for wheat and edible oils have a more direct and adverse impact on the cost and affordability of calorie heavy foods than they do on the broader consumer basket that determines a household's poverty status. Higher fertilizer prices are again an important driver of rising undernourishment. Our findings suggest that it is the concentration of world price increases across a narrow set of commodities important for people living near the calorie threshold, such as wheat and fertilizers, that amplifies the impacts of this crisis on food security indicators.

Our estimated changes in undernourishment may be somewhat overstated, because the demand system included the model – a Linear Expenditure System (LES) – incorporates minimum subsistence consumption quantities, which, in this context, limits substitution away from food items that become very expensive after the world price shock (i.e., wheat and edible oils). That said, declines in calorie availability is a common feature of negative economic shocks. For example, using longitudinal data from 52 countries, Headey and Ruel (2022) associate a ten percent decline in per capita income with a 15–18 percent increase in moderate or severe wasting in children under five.

Diet quality also deteriorates for many households. Diet quality is another important indicator of households' food and nutrition security status. The diet quality indicator used in this analysis, the Reference Diet Deprivation (ReDD) index (Pauw et al., 2021), is based on whether a household's food consumption meets the recommended calorie intakes across a diverse set of six food groups. Specifically, ReDD is a composite indicator of the incidence, breadth, and depth of diet deprivation within a population group or subgroup. The reference calorie intakes are defined by the healthy reference diet developed by the EAT-Lancet Commission (Willett et al., 2019). The food groups include staples (cereals and roots), fruits, vegetables, dairy, protein foods (animal-sourced and plant-based) and added fats (which includes edible oils). As with the poverty assessment, a survey-based microsimulation tool measures changes in the number of people who are diet deprived.

Prior to the crisis, few households accessed a healthy diet, with most failing to consume enough calories from food groups such as vegetables, dairy, and protein foods. Many households also consumed more than the recommended number of calories from staple foods. Households may therefore be deemed to have poor quality diets even if they are not undernourished in terms of energy intake. In Ghana, for example, our ReDD results reveal that the average person suffers four deprivations out of the six required food groups. Diets might deteriorate for several reasons. First, a decline in disposable incomes makes healthy diets less affordable. Second, an increase in the relative cost of food negatively affects the affordability of a healthy diet. Finally, relative changes in the average prices of different food groups may cause changes in food consumption patterns. As consumers shift consumption away from more expensive food groups and toward relatively cheaper ones, the rates of diet deprivation can change within the respective food groups. The net effect on diet quality depends on the magnitude of these consumption shifts relative to the calorie thresholds.

As discussed above, household disposable incomes decline across all 19 countries. Model results indicates that the cost of a healthy diet also increases in real terms in most countries, with the largest increase in the Philippines (4.7 percent). Exceptions include Bangladesh, Mali, Myanmar, and Nepal, where real diet costs decline slightly (i.e., food prices decline relative to nonfood prices). The largest decline is in Myanmar, where the real cost of a healthy diet falls by 2.4 percent. The cost of the “added fats” food group increases in all countries due to rising edible oil import prices. This is an important driver of the increase in the overall cost of the healthy reference diet. There is no readily observable consistency across countries in terms of movements in the cost of other food groups.

The combined effect of declining disposable incomes and rising diet costs (in most countries) causes diet quality to deteriorate. Relative food price changes, in turn, further encourage shifts in food consumption patterns, often exacerbating the quality of diets. Panel C in Fig. 3 reports the changing share of the population in each country who have diets that are worsening, i.e., the population that becomes deprived in at least one additional food group because of the crisis. Across the 19 studies, an additional 50.4 million people experience worsening diets, with 70 percent living in rural areas. Rising global food prices are overwhelmingly the driver of deteriorating diets, explaining 76 percent of the increase in the population with worsening diets. The increase in world fuel prices is a relatively minor driver of falling diet quality, mainly because the effect of this shock is diffused across the entire economy, including non-food sectors. In general, though, the nature of the 2022 global crisis makes it particularly detrimental for food security, especially diet quality.

The welfare metrics presented above for poverty, undernourishment, and diet quality are all threshold measures. The focus is on the increase in the number of people who fail to meet expenditure, calorie consumption, and/or dietary diversity thresholds. It merits mentioning that the welfare of people already poor, undernourished, and/or consuming low-quality diets declines as well, potentially dramatically as these people often live near survival thresholds. For example, Headey and Ruel (2022) find that food price inflation drives up wasting rates among children under five years old.

¹Five Asian countries (Bangladesh, Cambodia, Myanmar, Nepal, Philippines) and 14 African countries (Democratic Republic of Congo, Egypt, Ethiopia, Ghana, Kenya, Mali, Malawi, Niger, Nigeria, Rwanda, Senegal, Tanzania, Uganda, Zambia).

²The models use constant elasticity of substitution (CES) and transformation (CET) functions that allow imperfect substitution between domestic, imported and exported products/markets. Tables S1–S19 in the Supplementary Materials summarize base-year production and trade patterns for the 19 countries, and Table S20 provides the elasticity values used in the CES and CET functions. Details on model closures are also provided in the Supplementary Materials.

³Fertilizer Use by Crop (FUBC) estimates for each country are shown in Table S21 in the Supplementary Materials.

^{Appendix A}Supplementary data to this article can be found online at https://doi.org/10.1016/j.gfs.2023.100680.

Multimedia component 1:

Adjao R.T., Staatz J.M. Asian rice economy changes and implications for Sub-Saharan Africa. Global Food Security. 2015;5:50–55. [Google Scholar]

Arndt C., Benfica R., Maximiano N., Nucifora A. Higher fuel and food prices: impacts and responses for Mozambique. Agricult. Econom. 2008;39(2008):497–511. [Google Scholar]

Arndt C., Hussain M.A., Jones E.S., Nhate V., Tarp F., Thurlow J. Explaining the evolution of poverty: the case of Mozambique. American J. Agricult. Econom. 2012;94(4):854–872. [Google Scholar]

Arndt C., Pauw K., Thurlow J. The economywide impacts and risks of Malawi's farm input subsidy program. American J. Agricult. Econom. 2016;98(3) 2016: 982-980. [Google Scholar]

Bentley A.R., Donovan J., Sonder K., Baudron F., Lewis J.M., Voss R., Rutsaert P., Poole N., Kamoun S., Saunders D.G.O., Hodson D., Hughes D.P., Negra C., Ibba M.I., Snapp S., Sida T.S., Jaleta M., Tesfaye K., Becker-Reshef I., Govaerts B. Near- to long-term measures to stabilize global wheat supplies and food security. Nat. Food. 2022;3:483–486. [PubMed] [Google Scholar]

Diao X., Thurlow J. In: Strategies and Priorities for African Agriculture: Economywide Perspectives from Country Studies. Diao X., Thurlow J., Benin S., Fan S., editors. International Food Policy Research Institute; Washington D.C., USA: 2012. A recursive dynamic computable general Equilibrium model. [Google Scholar]

Dimaranan B., editor. Global Trade, Assistance, and Production: the GTAP 6 Data Base. Center for Global Trade Analysis; Indiana: 2006. Purdue University. [Google Scholar]

FAO . The Food and Agriculture Organization; Rome, Italy: 2022. The State of Food Security and Nutrition in the World 2022. (United Nations) [Google Scholar]

Ferreira F.H.G., Chen S., Dabalen A., Dikhanov Y., Hamadeh N., Jolliffe D., Narayan A., Prydz E.B., Revenga A., Sangraula P., Serajuddin U., Yoshida N. Policy Research Working Paper No. 7432; Washington, D.C.: 2015. A Global Count of the Extreme Poor in 2012: Data Issues, Methodology, and Initial Results. (World Bank) [Google Scholar]

Headey D., Martin W. The impact of food prices on poverty and food security. Ann. Rev. Res. Econom. 2016;8(1):329–351. [Google Scholar]

Headey D., Ruel M. Economic shocks predict increases in child wasting prevalence. Nat. Commun. 2022;13:2157. [PMC free article] [PubMed] [Google Scholar]

Hemming D.J., Chirwa E.W., Dorward A., Ruffhead H.J., Hill R., Osborn J., Langer L., Harman L., Asaoka H., Coffey C., Phillips D. Agricultural input subsidies for improving productivity, farm income, consumer welfare and wider growth in low- and lower-middle-income countries: a systematic review. Campbell Syst. Rev. 2018;14(1):1–153. [Google Scholar]

Li X.Y., Timsina J., Dixon J. Economics of nitrogen use in cereal systems in South and East Asia. J. Agricult. Sci. Technol. 2011;5:1. [Google Scholar]

Ludemann C.I., Gruere A., Hefer P., Dobermann A. Global data on fertilizer use by crop and by country. Scientific Data. 2022;9(501):1–8. [PMC free article] [PubMed] [Google Scholar]

Pauw K., Ecker O., Thurlow J., Comstock A. International Food Policy Research Institute (IFPRI); Washington, DC: 2021. Costing Healthy Diets and Measuring Deprivation: New Indicators and Modeling Approaches. IFPRI Discussion Paper 02073. [Google Scholar]

Thurlow, J., B. Holtemeyer, K. Pauw and J. Randriamamonjy. Forthcoming. Measuring Agri-Food Systems: New Indicators and Estimates. IFPRI Discussion Paper, Washington, D.C.: International Food Policy Research Institute (IFPRI).

Van Ittersum M.K., van Bussel L.G.J., Wolf J., et al. Can Sub-Saharan Africa feed itself? Proceed. National Acad. Sci. 2016;113(52):14964–14969. [PMC free article] [PubMed] [Google Scholar]

WFP . World Food Program; Rome, Italy: 2022. Projected Increase in Acute Food Insecurity Due to War in Ukraine. [Google Scholar]

Willett W., Rockström W., Loken B., Springmann M., Lang T., Vermeulen S., Others Food in the Anthropocene: the EAT–Lancet Commission on healthy diets from sustainable food systems. The Lancet Commissions. 2019;393(10970):447–492. [PubMed] [Google Scholar]