A System of Time-Varying Models for Predictive Regressions: Theory and Application

This thesis proposes a system of time-varying coefficients models for predictive regressions in order to address some of the econometric issues associated with forecasting stock returns. Based on the proposed time-varying models, we find that the widely used financial predictors, such as dividend-price ratio and dividend yield, show significant power for predicting future stock returns, both in-sample and out-of-sample.