Mean Variance Optimization of Forward Looking Systems and Worst-case Analysis

In this paper we consider expected value and mean variance optimization of a general forward--looking stochastic model. The problem is transformed into a general--nonlinear programming problem by adding extra constraints, which restrict the policy maker to commit to a certain policy. Based on this policy,and the rest of the economic structure, the agents can forecast future states except for random future disturbances. We present algorithms for computing optimal expected values based on iterative Taylor expansion and an interior point method for computing minimax robust policies. The results from both approaches are compared in order to assess the relative advantage of each approach and measure robustness against performance, and are also compared against DYNARE - a program for solving rational expectations models