On stochastic simulation of forward-looking models

The solution of a nonlinear macroeconometric model with expectations of future-dated variables generally has to be approximated by numerical simulation. A brief review of deterministic, and stochastic dynamic simulations of a backward-looking model is followed by a conceptual presentation of methods for dynamic simulation of a forward-looking (rational expectations) model. I distinguish between uncertainty faced by rational agents and by the modeller, and suggest different ways of simulating random variables in the model. Simulations of simple linear and nonlinear univariate time-series models illustrate the methods. © Kluwer Academic Publishers 2004.