Model Averaging in Predictive Regressions

This paper considers forecast combination in a predictive regression. We construct the point forecast by combining predictions from all possible linear regression models given a set of potentially relevant predictors. We derive the asymptotic risk of least squares averaging estimators in a local asymptotic framework. We then develop a frequentist model averaging criterion, an asymptotically unbiased estimator of the asymptotic risk, to select forecast weights. Monte Carlo simulations show that our averaging estimator compares favorably with alternative methods such as weighted AIC, weighted BIC, Mallows model averaging, and jackknife model averaging. The proposed method is applied to stock return predictions.