Abstract
In this paper we study the economic significance of simple time series models of stock return predictability. We investigate the practical usefulness of recent findings on time series return predictability of stock returns and their volatility for dynamic tactical asset allocation decisions. We introduce a mean variance investor with an investment horizon of one year who takes investment decisions daily. When stock returns follow a random walk this investor holds constant proportions of a stock market index and a risk free asset. Using past data and knowledge of some well known return predictability results (i.e. predictability based on calendar anomalies and predictability from economic variables like dividend yields and short term interest rates), we evaluate whether, how and to what extent these predictability results might affect his investment decisions. For this investor we also investigate the practical usefulness of knowledge about the predictability over time of market volatility. The design we choose is as follows. We give the predictability results the benefit of the doubt and assume that all the estimates and models are correct and indeed accurately describe the true return generating process. We then analyze analytically, numerically and by Monte Carlo simulation the effect of investment decisions--conditional on these predictability results--on the return distribution of his portfolio. We also introduce transactions costs in this setting; these influence investment choices. Our main findings are that small transactions costs substantially reduce potential benefits of trading on calendar anomalies. Generally, however trading remains profitable under the assumption that stock market returns are partially predictable from economic variables like dividend yields and interest rates. This holds for relatively large transactions costs. Trading on volatility predictability is not profitable, except in the case of negligible transactions costs.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have