The Use of Outside Information in Econometric Forecasting

Abstract

This paper examines the potential value of using outside information such as monthly data to update or modify quarterly econometric forecasts. A best linear unbiased updating procedure is introduced and applied to a small experimental macroeconometric model patterned after the Michigan Quarterly Econometric Model. It is found that accurate outside information has the potential for substantial improvement in the accuracy of short-term forecasts. Based on a common definition of dynamic accuracy this gain appears to be concentrated in near-term forecasts and is much less pronounced as the forecast horizon is extended.