Linear model averaging by minimizing mean-squared forecast error unbiased estimator

摘要

This paper presents a new ordinary least squares model averaging method which is proposed to be a preferable alternative to Mallows Model Averaging (MMA), Bayesian Model Averaging (BMA) and naive simple forecast average. The method is developed to deal with possibly non-nested models and selects forecast weights by minimizing the unbiased estimator of mean-squared forecast error (MSFE). Proposed method also yields forecast confidence intervals with given significance level what is not possible when applying other model averaging methods. In addition out-of-sample simulation and empirical testing proves the supremacy of MSFE model averaging over existing combination approaches.