Forecasting using autocorrelated errors and multicollinear predictor variables

Multicollinearity and autocorrelated errors problems are well known in regression analysis and in some economic forecasting problems. In the literature, these problems have been examined separately from each other. In this article they are considered together for a given multiple linear regression model, and an estimation procedure for regression coefficients is introduced. The method presented is also demonstrated on a sales forecasting problem. (C) 1998 Elsevier Science Ltd. All rights reserved.