Predictive ability of three different estimates of “cay” to excess stock returns - A comparative study Germany & U.S -

The results of (Lettau, M.; Ludvison, S.,(2001)) show that Cay-LL has a significant predictive power both in the in-sample and the out-of-sample forecast of excess return. Our study departs from Lettau, M.; Ludvison, S.,(2001) in adding and comparing other two estimates of “cay” namely “Cay-Ols’ and “Cay-Dls” besides “Cay-LL” for forecasting excess return in both Germany and U.S over the period 1969:2 to 2005:1. Using quarterly data for both Germany and U.S over the period 1969:2 to 2005:1. We find that Cay-Ols proved to have the strongest in-sample forecast and out-of-sample forecast of the nested models of excess stock returns over the treasury bill rate in the U.S. We also find that the three different methods of estimating cay, Cay-Ols, Cay-Dls and Cay-LL, do not have any significant effect in either the in-sample forecast or the out-of-sample forecast of nested models in Germany. Finally analyzing the out-of sample forecast of non-nested models, using the Diebold Mariano(DM) test, we find that for the case of U.S, Cay-ols, Cay-Dls or Cay-LL proved to have equal predictive accuracy. On the other hand for the case of Germany, neither Cay-Ols nor Cay-Dls have equal predictive accuracy when compared to Cay-LL.