Prior information in forecasting with econometric models

summary: in this paper we suggested on the basis of the kalman filter theory systematic procedures to incorporate prior information in the forecasting process. according to our proposals a forecasting exercise involves the following three steps. step 1. whenever a new observation becomes available a correction for the measurement error is made by utilizing the information contained in the latest available forecast for this variable. the corrected value is used as model input for lagged variables and eventually for updating parameter estimates. step 2. parameter estimates are updated advantageously by using sequential estimators. the time series of the parameter estimates indicate the validity of the parameter specification. step 3. all available prior information about the endogenous model variables is cast into an anticipations model which together with the structural model yields the optimal forecast.;