Forecasting with a Computable General Equilibrium Model

Computable general equilibrium (hereafter CGE) models represent a rapidly emerging field in applied economic analysis. CGE • models have, with few exceptions, been used solely for policy analysis; however there is a growing demand for forecasts from these models. Let a "set of scenarios" (or information set) on the exogenous variables of a CGE model denote not only point estimates, but also measures of their precision (and, where relevant, measures of correlation among these variables). This paper gives an exposition of a method of transforming such an information set into point forecasts of the CGE model's endogenous variables and associated set of confidence intervals. This method will not usually require repeated solutions of the CGE model for use with different information sets on the exogenous variables.