AN ALTERNATIVE METHODOLOGY FOR SOLVING NONLINEAR FORWARD-LOOKING MODELS

This gaper presents a new methodology for solving nonlinear deterministic forward-looking models. Based on a relaxation algorithm described by Laffargue (1990), the methodology is theoretically founded on a general multivariate linear model. Then, a complete experimental scheme is provided in the nonlinear case, including solution time horizon selection and saddlepoint path testing strategies. The proposed experiment is mathematically robust and it does not require any expert knowledge in numerical analysis. It is especially adapted to the simulation exercises conducted on medium scale economic models.