Optimal control techniques for large, nonlinear, rational expectations models

Numerical results are presented for the use of penalty function methods applied to large, nonlinear, rational expectations models. The use of these methods to solve the basic two-point boundary value problem for generating model-consistent expectations is considered. An experiment with three models of the UK economy is presented, and the results are compared with first-order iterative techniques. The penalty function approach is found to be feasible but consistently less efficient. Three ways of applying penalty function methods for the optimal control of a consistent expectations model are considered. These three techniques can generate different solutions which can be interpreted as the solutions to three different dynamic games played between policy makers and economic agents. Results on the application of optimal control to derive inflation-unemployment trade offs for three macroeconomic models are presented