Abstract: Food security is an issue of risk. If climate change is not responded to with diet, technology and/or policy changes, it may lead to reduced food security for the world population, in particular the poorer part which in longer periods may not afford to purchase food in sufficient quantity and quality. In order to improve the situation, certain policy changes may be required.  In some cases are policy recommendations relatively obvious, while in other cases a deeper insight in the stochastic dynamics of food supply and storage is required to assess the consequences of policy proposals. The relatively obvious part is that farmers need be responsive in periods of low total production, so that sufficient supply restores quickly. Moreover, trade should allow local shortages to be covered. Many national policies with the goal of self-sufficiency aim in the opposite direction with stable prices and production and relatively less flexibility in production. The stochastic dynamics of food supply can be analysed in more detail with a dynamic stochastic general equilibrium model (DSGE). Although agriculture by nature is about taking decisions under uncertainty, quantitative stochastic dynamic models for policy analysis in agriculture have not yet emerged. The contribution in MACSUR is a formalization of a class of DSGE-s based on representation of biological processes managed with regard to outcomes due to uncertain nature. No Label

Abstract: This report provides an overview of how grasslands are represented in six different farmscale  models represented in MACSUR. A survey was conducted, followed by a workshop in  which modellers discussed the results of the survey, and identified research challenges and  knowledge gaps. The workshop was attended by grassland as well as livestock specialists.  The investigated models differed largely with respect to how grasslands were represented,  e.g. as regards weather and management factors accounted for, spatial and temporal  resolution, and output variables. All models had grassland modules that simulate DM yield  and herbage N content (or crude protein (CP) content = N content x 6.25). Many models  also simulate P content, whereas only one simulate K content. About half of the model  simulate herbage energy value and/or herbage fibre content and fibre and/or dry matter  digestibility. Critical input data required from grassland models to simulate ruminant  productivity and GHG emissions at farm scale was identified by the workshop participants.  The different types of input data required were ranked in order of importance as regards  their influence on important system outputs. For simulation of ruminant productivity and  GHG emissions, herbage DM yield was ranked as the most important input variable from  grassland models, followed by CP content together with at least one variable describing  herbage fibre characteristics. These findings suggest that work on improving the ability of  the current grassland models with respect to simulation of fibre/energy should be  prioritized in farm-scale modelling aiming at quantifying livestock production and GHG  emissions under different management regimes and climate conditions. More work is also needed on model evaluation, a task that has not been prioritized yet for some models.