| Abstract: | In this article, a multivariate component model for conditional asset return covariance is developed as an extension to the univariate volatility component model of Engle & Lee (1999). The conditional covariance now is decomposed into a long-run (trend) component and a short-run (transitory) component. Through the decomposition, relationships like the long-run correlation and volatility copersistence can be studied solely upon examining the long-run trend of the conditional covariance. The decomposition also has important implications in studying portfolio hedging problems such as the multi-period minimum-variance hedging for long-term portfolio management. The empirical study in this article focuses on estimating the covariance component structure between the S&P 500 cash and futures markets and their contemporary and long-run correlation relationship and the volatility copersistence relationship. © John Wiley & Sons, Inc. Jrl Fut Mark 19: 877–894, 1999 |
