Predictability of Stock Return and Volatility: A Factor Based Approach

Using factor based approaches, we investigate a return and volatility forecasting procedure that exploits all the available information by still keeping the econometric framework at considerable size. Our findings demonstrate that factor based approaches provide substantial gains when predicting the sign of the excess returns and state of the volatility separately as well as jointly. A striking result is that the performance of this procedure increases especially after 1990, where the existing evidence suggests lower predictive power of many econometric forecasting procedures. For evaluating the economic significance of the factor based approach, we also simulate a mean-variance investor that uses return and volatility forecasts to determine optimal portfolio weights. In line with the predictive performance, under moderate transaction costs, a mean-variance investor would be willing to pay several hundreds of basis points per annum for switching from passive and dynamic strategies based on benchmark models to dynamic strategies that employ factor based approaches after 1990.