KRYLOV METHODS FOR SOLVING MODELS WITH FORWARD-LOOKING VARIABLES

The simulation of large macroeconometric models containing forward-loo king variables can become impractical when using exact Newton methods. The difficulties generally arise from the use of direct methods for t he solution of the linear system in the Newton step. In such cases, no nstationary iterative methods, also called Krylov methods, provide an interesting alternative. In this paper we apply such methods to simula te a real world econometric model. Our numerical experiments confirm t he interesting features of these techniques: low computational complex ity and storage requirements. We also discuss a block preconditioner s uitable for the particular class of models solved. (C) 1998 Elsevier S cience B.V. All rights reserved.