Some Practical Considerations for Applying Perturbation Methods to

Several authors have implemented computer code automating the application of perturbation methods for constructing solutions for dynamic economic models.\cite{sims01,chen99,collard,schmitt,judd98,anderson02} Such techniques facilitate the construction and use of nonlinear stochastic models for calibration and policy analysis exercises. Practitioners have begun applying these techniques in a number of contexts, including calculating welfare effects of alternative monetary policy rules. In this context, and many others, useful models will often include dozens of equations with multiple lags and leads. This paper discusses the challenges encountered in ``scaling up'' these techniques and offers several strategies for enhancing the performance and robustness of implementations of the perturbation technique.