摘要: This study investigates the forecasting performance of GARCH(1,1) model by adding an effective covariate. Based on assumption that many volatility predictors are available to help forecast a target variable, this shows how construct covariate from these and plug it into model. presents method building such contains maximum possible amount predictor information for volatility. The loading constructed proposed is simply eigenvector matrix. enjoys advantages easy implementation interpretation. Simulations empirical analysis verify performs better than other methods volatility, results quite robust misspecification. Specifically, reduces mean square error 30% S&P 500 Index. also useful in improving several GARCH-family models value-at-risk. Copyright © 2013 John Wiley & Sons, Ltd.