The paper attempts to explore whether lagged variables that help predict stock returns are merely proxying for mis-measured risk. Therefore, three different ways of measuring risk are employed - one that relies on semi-parametric methods, a second based on the GARCH-M model, and the third just uses lagged squared returns.
In an application to Japanese data, four key predictor variables are identified, and are shown to have non-trivial additional forecasting power irrespective of how we measure risk. Moreover, there is no consistent relationship between expected volatility and excess returns. These results are also robust to using the international CAPM or consumption CAPM.
These findings are inconsistent with many existing models of the stock market, and they suggest either the existence of "noise" traders or the need for better models of risk.
Download is not available
