Paraules clau:: Dynare,Perturbation,Hybrid,Accuracy,Numerical methods, Approximation
Resum: Local (perturbation) methods compute solutions in one point and tend to deliver far lower accuracy levels than global solution methods. In the present paper, we develop a hybrid method that solves for some policy functions locally (using a standard perturbation method) and that solves for the other policy functions globally to satisfy certain nonlinear optimality conditions (using closed-form expressions and a numerical solver). We applied the hybrid method to solve large-scale RBC models used in the comparison analysis of Kollmann et al. (2011b). We obtained more accurate solutions than those produced by any other (either local or global) solution method participating in that comparison.