A physically-based model of long-term food demand

•Our new food demand model includes local income inequality and income-based food wastage for the first time.•Global crop and grass demand could increase from 5.5 to 10.9Gton during 2011–2050 (SSP-2).•Consumption inequality within regions is of the same magnitude as the inequality between regions.•Global undernourishment drops from 700 million people in 2015 to 270 in 2050, but only if supply increases to meet demand.•Overall efficiency can be improved much more by targeting animal-centric diets rather than food waste. Reducing hunger while staying within planetary boundaries of pollution, land use and fresh water use is one of the most urgent sustainable development goals. It is imperative to understand future food demand, the agricultural system, and the interactions with other natural and human systems. Studying such interactions in the long-term future is often done with Integrated Assessment Modelling. In this paper we develop a new food demand model to make projections several decades ahead, having 46 detailed food categories and population segmented by income and urban vs rural. The core of our model is a set of relationships between income and dietary patterns, with differences between regions and income inequalities within a region. Hereby we take a different, more long-term-oriented approach than elasticity-based macro-economic models (Computable General Equilibrium (CGE) and Partial Equilibrium (PE) models). The physical and detailed nature of our model allows for fine-grained scenario exploration. We first apply the model to the newly developed Shared Socio-economic Pathways (SSP) scenarios, and then to additional sustainable development scenarios of food waste reduction and dietary change. We conclude that total demand for crops and grass could increase roughly 35–165% between 2010 and 2100, that this future demand growth can be tempered more effectively by replacing animal products than by reducing food waste, and that income-based consumption inequality persists and is a contributing factor to our estimate that 270 million people could still be undernourished in 2050.