Social Sharing
Extended Viewer
Forecasting Covariance
Jul 1, 1989
The material in this article is taken from a presentation at the annual research seminar at Pebble Beach, June 4-8, 1989. Covariance is a measure of the linear relation between two random variables. On average, if the monthly returns in 1988 from holding IBM stock were positive (negative) when the returns from holding Southwest Gas were negative (positive), then the returns on these two stocks are said to exhibit negative covariance. What is the appropriate frequency to measure covariance? Is it reasonable to assume that covariance is constant over time? Is the covariance between two stock returns for the next month predictable?