Use of multivariate methods to summarise the results from an evolutionary policy optimisation model (APolO): A case study for Scottish agriculture

摘要

We describe how multivariate statistical methods can be used to explore the output from a new type of hierarchical policy optimisation model which has been applied to formulation of agri-environment policy for Scotland. The model (APolO) generates output in four classes of variable for a population of Pareto-optimal policy solutions: Policy objective values; policy instrument values; total gross margin from farming; land use activity levels. When policy decisions are summarised as a trade-off betweenonly two conflicting objectives, the trade-off can be visualised as a curve or tradeoff front allowing relatively easy selection of the policy formulation which best balances the objectives of the policy maker. When multiple objectives are in play, thedomination rules of Pareto-optimality mean that no clear pattern may be visible among a set of equally optimal policy formulations. In this situation multivariate statistical methods which reveal the underlying covariance (or correlation) structure amongthe output variables from the policy model may be useful in simplifying the visualisation and interpretation of the model's output.