The paper addresses the question of whether the use of macroeconomic factors in covariance forecasting models improves asset allocation. Mounting evidence suggests that both the first and second moments of asset returns are time-varying and predictable. However, the economic value of accounting for this time-variation in asset allocation is still an open question. In this paper, I forecast the covariance matrix of monthly realized returns between stocks, bonds and commodities using a Vector AutoRegression (VAR) model with macroeconomic factors as exogenous variables. I find that a volatility timing strategy involving monthly re-balancing of a portfolio based on these VAR forecasts produces economic gains over portfolios based on alternative covariance estimates.

pdf