Table 2.

Main food demand models. Source: authors' own elaboration from literature.

model	description
World Food Model
FAO (Alexandratos 2009).	multi-country partial equilibrium model with 52 separate commodities in each country. Simulations are based on assumptions of mean growth of per capita income and population. In this model, food for direct human consumption is projected in per capita terms using the base year data for this variable. In addition, biofuels enter as demand of food for other uses. Income demand elasticities are constant, and taken from FAO estimates, and complemented with the SWOPSIM model of the United States Department of Agriculture and the MTM model of OECD.
The World Grains Model
The World Bank (Mitchell et al. 1997).	non-spatial, partial-equilibrium trade model used to forecast commodity projections. It is also based on constant income demand elasticities.
The IMPACT model
The International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT); International Food Policy Research Institute.	multi-country/region model where markets are linked through trade. It uses estimated income demand elasticities and covers 36 countries and regions, and 16 commodities, including all cereals, soya beans, roots and tubers, meats, milk, eggs, oils, oilcakes, and meals. Demand is a function of prices, income, and population growth. The IMPACT income demand parameters are based on average aggregate income elasticities for each country, given the income level and distribution of population between urban and rural areas as they evolve over the projection.
FAPRI model
Food and Agricultural Policy Research Institute.	multi-market, partial-equilibrium model of world agriculture, food and biofuel markets. It is based on estimated constant elasticities from the FAPRI elasticity database, and the simulations are based on exogenous changes on growth, population and exchange rates.
SWOPSIM Model
The Static World Policy Simulation; US Department for Agriculture.	comparative statics, multi-product, multi-region partial equilibrium with 20 agricultural commodities. Income demand elasticities are based on estimated elasticities from a large number of sources at the SWOPSIM database.
Ministerial Trade Mandate (MTM) Model OECD	comparative–static model with several country or regional sub-models aimed at analysing policy changes in the medium term. In this model, also exogenous constant elasticities are imposed.
other models	description
GIDD model
The Global Income Distribution Dynamics (GIDD) model; (Bussolo et al. 2008b).	model that links simulations from a CGE model to household surveys in order to generate changes and predictions on income distribution. While it has the advantage of considering income dynamics, the demand side of the CGE model is based on a single representative household in each country that maximizes an extended linear expenditure system (ELES). The system captures various substitution possibilities across commodities and shifts in demand towards commodities with higher income elasticities over time. However, changes on income distribution do not feed back to shaping aggregate demand elasticities and the potential bias from aggregation, in addition to linear Engel curves, still remain.
ENVISAGE model
The Environmental Impact and Sustainability Applied General Equilibrium Model (van der Mensbrugghe 2008).	dynamic general equilibrium model calibrated using the GTAP database with 2004 as base year with a carbon emission database. The model can run with the 113 countries/regions and 57 commodity groups of the GTAP database. There is a single representative household that consumes goods and services and saves, and the model is designed with several different consumer demand specifications including the CDE (see Hertel 1997), the LES/ELES (see van der Mensbrugghe 2008) and the AIDADS (see Rimmer & Powell 1996).