1 Taking Advantage of Global Diversification: A Mutivariate-Garch Approach

We gauge the economic value of multivariate covariance estimators by assessing the risk-return performance of the resulting mean-variance efficient portfolios. A dynamic asset allocation framework is deployed, where the multivariate covariance forecasts compete against simpler nonparametric rivals widely used in the industry. A two-layer portfolio construction process for global asset allocation is developed to overcome the problem of handling large dimensional covariance structures. Based on an out-of-sample volatility timing setup the empirical results suggest that the multivariate covariance forecasts outperform their nonparametric counterparts and the proposed two-layer global asset allocation is favored over conventional portfolio selection approaches