Executive summary It is no longer feasible to look at agricultural livelihoods, food, health and the management of natural resources in isolation. The 2030 Agenda for Sustainable Development stresses the urgent need to take concerted action and pursue policies directed at transformational change. It calls for a new agricultural approach achieving multiple benefits to ensure sufficient, safe and nutritious food through a stable multifunctional landscape, while respecting human rights. In India, agri-food systems are under increasing pressure to meet nutrition, health and poverty eradication targets while reversing the depletion of water tables, soil degradation, deforestation, land degradation and threats to agrobiodiversity. In addition, climate change impacts related to rising temperatures, increasing frequency of extreme weather events, shifts in precipitation and hydrology, will expose the country to increased vulnerability and threats. Systemic challenges require systemic solutions. Ending poverty and achieving zero hunger, while ensuring inclusive growth and sustainably managing the planet’s ecosystems in the context of climate change and biodiversity loss, will only be possible through holistic and integrated approaches. This report summarizes the findings of the foresight study “AgroEco2050” (2019-2022) jointly carried out by Rythu Sadhikara Samstha, Department of Agriculture of the Government of Andhra Pradesh (RySS), the French Agricultural Research Centre for International Development (CIRAD) and the Food and Agriculture Organization of the United Nations (FAO). The AgroEco2050 foresight process was based on the collective expertise and vision of a multistakeholder group which worked together in India from 2020 to 2022, as well as a unique statistical overview of past structural transformations from the 1970s to 2019 in India (in terms of land use, population, employment, sectoral economic growth, productivity…). The group quantified some dimensions of two rather comprehensive scenarios for Andhra Pradesh (AP), a State of South India with 53 million inhabitants in 2020 and 9.3 million farmers. One scenario focuses on intensification of the industrial agriculture model which is currently dominant worldwide, and the other on a full agroecological transition (AE) through natural farming (NF). The model Agribiom-India and its interface developed and customized for this participative foresight exercise (B. Dorin, Cirad) was used interactively by the group to test the coherence and adjust the assumptions made for both scenarios. Natural farming is growing in India. In Andhra Pradesh, agroecology-based natural farming is practiced through the Andhra Pradesh Community-managed Natural Farming programme (APCNF), supported since 2016 by the Government of Andhra Pradesh (GoAP) to promote rapid inclusive economic growth, farmer’s happiness and healthy food. The programme supports farmers’ transition from conventional, chemical-based agriculture to natural farming. The number of farmers practicing NF partially or fully has gone up from around 40,000 in 2016 to 700,000 in 2020-21—an increase of 17 times in 4 years, with a vision to reach all farmers in the state by 2031 (an estimated six million farmer households). If it succeeds, Andhra Pradesh would lead the establishment of a new food systems approach within the subcontinent that would address the ecological, financial, nutritional and social challenges in an integrated manner. At the world level, APCNF could also become the first example of massive scaling-up of agroecology through a single programme. In this regard and to further sustain its vision and work, in 2018 the GoAP requested CIRAD and FAO to partner in conducting a foresight exercise on agriculture by 2050, with a scenario showing the potential impacts of a conversion of its million farmers to Natural Farming. Foresight is a general term to describe an action-oriented, scientific study of the future. It is a systematic, collective, medium-to-long-term vision-building process aimed at enabling present-day decisions and mobilizing joint actions. It suggests possible, probable and preferred futures; but foresight work never makes very specific predictions or forecasts, unlike, say, the weather bureau. The scenarios built in foresight exercise, and in this particular study, are “ideal types” and are not precise forecasts. They don’t pretend to be predictions of the future. The purpose is to present the main storylines or elements of the scenarios. The ideal type is an abstract model which, when used as a standard of comparison, unveils aspects of the real world in a clearer, more systematic way, to help understand the potential implications of a particular pathway that societies may take. It is a “constructed ideal” used to present an approximation of reality by selecting and accentuating certain elements. Each and every element mentioned in the ideal type may not be found in the real world. This scenario building exercise had three main objectives: (1) to provide an image of what a complete transition of Andhra Pradesh to agroecology could look like, especially to better understand the conditions of such a transformation; (2) to compare this ideal type of transformation with that of industrial agriculture (IA) which has guided public policies, industries, agricultural production and food since the 1960s in India; (3) to contribute to national and international debates and researches on the future of food and agriculture including agroecology, with the overall aim to transition towards sustainable food systems, leaving no one behind. Both our scenarios foresee an Andhra population of almost 60 million inhabitants in 2050 (compared to almost 53 million in 2019, and 33 million in 1980), with more than 35 million adults in working age (the “labour force”), considered here between 20 and 64 years old (as opposed to 15-64 years found in the literature for decades, without updating the steady and desired increases in education and study levels). In the AE scenario, about 93% of the 20-64 year old adults would be employed (against less than 70% in 2019), including 10 million in agriculture and allied activities which would represent 30% of the workforce (against 42% in 2019 and 60% in 1980). That would be a net increase of 700,000 farmers compared to 2019, or as many farmers as in the early 1970s. In the AE scenario, all farmers would practice NF in 2050. In doing so, they would regenerate degraded soils and ecosystems with complex multi-crop-livestock landscapes including rich soil microbiomes and trees. Without touching current forest or shrub areas, they would even cultivate 2 million hectares more than in 2019, totalling more than 8 million hectares in 2050, reversing past trends of desertification and increase in fallow areas as NF would enable regeneration of dry and degraded lands. This full-employment scenario largely based on agriculture and allied activities, coupled with larger cultivated lands and the low-input but high-output practices of NF all over the year, would lead to a high growth of the Gross Value Added (GVA) in agriculture (6% per annum over 2019-2050, against 4% over 1980-2019), as well an all-sectors economic growth of 6.5% p.a. (against 5.8% over 1980-2019). With this AE scenario, AP would then be embarked in what we call a “Farmer-Developing path” (FD), where farmers are more numerous but where their income gap with nonfarmers is narrowing, unlike in the past decades (1980-2019) which were deeply marked by a “jobless growth” and a growing agrarian crisis. By contrast, in the IA scenario, unemployment among the 20-64-year olds would remain at the disturbing level of 30% in 2019. Indeed, in this scenario, the automation of human activities and energy consumption would accelerate, in particular in agriculture where the number of farmers would be almost halved, from 9.3 million adults in 2019 to 5 million in 2050, or from 42% of the workforce to 20%. These remaining farmers would all practice industrial farming. They would specialize in few products (paddy, palm oil, cotton, silk, cow milk, chicken, aquaculture, fruits and vegetable…) which they would produce with capital- and input-intensive techniques (GMOs, synthetic fertilizers, pesticides or antibiotics, hydroponic greenhouses, robots, precision agriculture, artificial intelligence…) largely through export-oriented contract farming with large Indian or foreign agribusiness multinationals. For this form of agriculture to be competitive, only best and city-centric land would be exploited. Fallow areas would then continue to increase as in the past, from 2.4 million hectares in 2019 to nearly 3 million in 2050, which, combined with increasing urbanization, would reduce the cultivated area to 5.5 million hectares in 2050 (6.2 million in 2019). All in all, with less land and farmers, but also higher costs of production, the agricultural GVA would increase only by 3.5% p.a. on average over 2019-2050 (4% over 1980-2019), which, coupled with a rather high unemployment rate (hence less demand), would lead to an overall GVA growth of 6% p.a. over 2019-2050. This last assumption is lower than in the AE scenario (6.5%) but remains optimistic compared to past trends (5.8% over 1980-2019), with the strong assumption that the factors of production (soil, water, air, biodiversity, human health, etc.) would not deteriorate further in this scenario, although many believe otherwise. With all these assumptions about the future in 2050 in an industrial agriculture scenario, AP would then be embarked in a “Farmer-Excluding” path (FE), where farmers are fewer in number but with an ever-growing income gap with nonfarmers, as in the past decades (1980-2019) marked by a deep agrarian crisis and farmer protests, but also ever-increasing agricultural and food subsidies to mitigate negative impacts as much as possible. In both scenarios, the average income of farmers would be multiplied by about 5.5 compared to 2019 (all in constant 2011-12 rupees), to reach about 3000 INR per day and per farmer (or nearly 1.1 million rupees/year). Farmers would then earn almost the same, but for different reasons: in the IA scenario, farmers would produce more calories per hectare (almost 44,000 kcal/day against 36,000 in the AE scenario) and each on larger area (1.11 ha/farmer on average against 0.83 ha in the AE scenario, and 0.67 ha in 2019), but at higher costs and lower nutritive quality: in the AE scenario, each farmer would earn 10.3 paise per kilocalorie produced, while it would be 6.1 paise in the IA scenario (moreover without deducting capital costs for machineries and others, much higher in the IA scenario). However, a farmer-nonfarmer income gap (i.e., difference in average incomes between farmers and nonfarmers) would remain in both scenarios. It would be less than in 2019, but would still represent 47% of the average income of IA farmers in 2050, and 22% the average income of AE farmers (against 62% in 2019). In OECD countries, historically this average income inequality between farmers and nonfarmers was narrowed by a drastic reduction in the number of farmers, allowing those remaining in farming to run increasingly large and robotic industrial farms with various direct and indirect public supports. In our Indian AE scenario, there is no dream of such capital-, energy- and land-intensive agriculture for very few farmers; the vision (also evidenced in the past through abundant scientific literature) is that production of food and environmental services is more efficient when farmers operate at small scale. But while small farm size makes agriculture more efficient and productive per hectare, it also prevents individual farmers from increasing their income through farm consolidation and robotisation as has happened in OECD countries. To make up for this, farmers could then be remunerated for the environmental services (currently unpaid) which they provide when following agroecological approaches. This would help close the remaining income gap between farmers and nonfarmers in 2050. With AE, these environmental services to local and global societies would be numerous, such as water saving and filtering, storage of soil organic carbon and mitigation of climate change, protection of pollinators and biological control agents, resilience to biotic and abiotic shocks thanks to highly biodiverse agro-ecosystems. Payments for Ecosystem Services (PES) could be granted to each farmer (whether cultivator or labourer) after evaluating and monitoring the extent to which her/his village or region practices agroecology providing multiple ecosystem and health services. In our AE scenario, these PES would amount to about 5.7% of Andhra's GDP in 2050 to completely close the average income gap between farmers and nonfarmers. By contrast, the cost of policies to reduce income inequalities would be much higher in the IA scenario than in the AE scenario (and they would not boost environmental services, as they do in the AE scenario). In the IA scenario, policies to reduce income inequalities would deploy instruments adapted to its logic: (i) price support and input subsidies (to credit, insurance, power, irrigation, chemicals, genetics, robotization, etc.) as today but on a higher level (6.8% of the total GVA in Andhra Pradesh in 2050, against close to 2.5% in the late 2010s in India) to really close the farm-nonfarm labour productivity gap, and (ii) Universal Basic Income (UBI) for the unemployed, which, if it amounted to only 25% of the average income of nonfarmers, would then represent 11.4% of the Andhra GDP in 2050. In the quasi-full employment AE scenario, this percentage would be reduced to 1.9%. Overall, such policies to reduce inequality would cost more than 18% of GDP in the IA scenario, while it would cost less than 8% in the AE scenario, with in addition a much higher efficiency on various fronts (inequality, environment, health...). Last but not least, we show that after combining the anticipated cultivated areas and annual yields of each scenario, the total food production of 2050 expressed in kilocalories per inhabitant would be significantly higher in the AE scenario than in the IA scenario (5000 kcal/capita/day against 4050, and 3660 kcal in 2019). As previously mentioned, this AE food production would also be much more balanced and healthier than today and the IA scenario. Overall, our two scenarios illustrate two possible but radically different visions of agricultural science and productivity, of societal goals and choices, with their own trade-offs and necessary transformations in both. Compared to the current techno-centric and capital-intensive industrial agriculture and food that the IA scenario would amplify, our expert group was predominantly in favour of the AE scenario, because this AE scenario would be : (a) at least as productive in terms of useful biomass per surface unit; (b) more resilient to economic, climate and biotic shocks; (c) more labour intensive than capital intensive; (d) more profitable for farmer households since their input costs would be cut, their diverse, tasty, nutritious and healthy foods would be better priced on local and international markets, and their coproduction of environmental goods or services would be paid for their local or global values, such as safe water, biodiversity reservoirs, soil fertility, nutrient recycling, pollination, fight against diseases or floods, mitigation of, and adaptation to climate change. These multiple benefits (or positive impacts) of the AE scenario would also enable India to achieve multiple Sustainable Development Goals of the 2030 Agenda to respond to pressing economic, social and ecological challenges and global commitments.