The Economic Value of Exploiting Time-Varying Return Moments

This paper evaluates out-of-sample dynamic portfolio performance to examine the economic value of exploiting time variation in the risk premium and in the volatility of stock returns to a multiperiod investor. We find that ignoring time variation in these return moments leads to significant utility costs. The time-varying risk premium plays a more important role than time-varying volatility in forming portfolio weights and in improving portfolio performance because the conditional expected return is a more persistent process than conditional volatility. Strategies using the dividend yield as a state variable perform poorly out of sample, although the good in-sample performance of these strategies has been documented in the literature. Unlike prior studies, this paper finds that the utility cost of myopic behavior is small. ∗School of Banking and Finance, Australian School of Business, University of New South Wales, Sydney, Australia, Tel: +61 2 9385 5952, Email: [email protected]. I am grateful to Pierluigi Balduzzi, David Chapman, Alan Marcus, and Zhijie Xiao, for their helpful suggestions. I also thank Joachim Inkmann, Edward Kane, Van Nguyen, Jun Qian, Piet Sercu, and seminar participants at Boston College and the 2008 meetings of the European Finance Association for useful comments. I am fully responsible for all the weaknesses of the paper.